nhieu.dcct@gmail.

com
Foundations of Engineering Mechanics
Series Editors: V.I. Babitsky, J. Wittenburg

[email protected]
For further volumes:
http://www.springer.com/series/3582
[email protected]
Sergey M. Aleynikov†

Spatial Contact Problems

in Geotechnics

Boundary-Element Method

With 295 Figures

[email protected]

123
Series Editors:
V.I. Babitsky J. Wittenburg
University Loughborough Universität Karlsruhe
Department of Mechanical Engineering Fakultät Maschinenbau
Loughborough LE11 3TU, Leicestershire Institut für Technische Mechanik
United Kingdom Kaiserstrasse 12
76128 Karlsruhe
Germany
Author:
Sergey M. Aleynikov†
Voronezh State Architecture and Civil
Engineering University
20-Letya Oktyabrya Street, 84
394006, Voronezh
Russia
e-mail: [email protected]

Originally published in Russian as “Boundary Element Method in Contact Problems

for Elastic Spatial-and-Nonhomogeneous Bases”, in 2000 by Publishing House of Civil
Engineering Universities Association, Moscow, Russia, ISBN 5-93093-053-8, 601 p.

ISSN 1612-1384 e-ISSN 1860-6237

ISBN 978-3-540-25138-5 e-ISBN 978-3-540-44776-4
DOI 10.1007/b11479
Springer Heidelberg Dordrecht London New York
Library of Congress Control Number: 2009934850

© Springer-Verlag Berlin Heidelberg 2011

This work is subject to copyright. All rights are reserved, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting,
reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publication
or parts thereof is permitted only under the provisions of the German Copyright Law of September 9,
1965, in its current version, and permission for use must always be obtained from Springer. Violations
are liable to prosecution under the German Copyright Law.
The use of general descriptive names, registered names, trademarks, etc. in this publication does not
imply, even in the absence of a specific statement, that such names are exempt from the relevant protective
laws and regulations and therefore free for general use.

Cover design: deblik, Berlin

[email protected]
Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com)

Obituary

The author of this monograph – Dr. Sergey M. Aleynikov, a well-known specialist

in the field of numerical simulation of soil-foundation contact interaction, passed
away on August 4, 2009.
Sergey M. Aleynikov graduated from the Department of Applied Mathematics
and Mechanics of the Voronezh State University, then he took a Postgraduate Course
at the Heat and Gas Transfer Institute of the USA Academy of Science in the city
of Minsk, the capital of Belorussia. After defending the PhD thesis in 1983, he
started working at the High Mathematics Chair at the Voronezh State University of
Architecture and Civil Engineering, and finally, he was the Head of this Chair for
five last years.
At the university he began, in close collaboration with Dr. S.V. Ikonin, actively
developing the Boundary Integral Approach (BEM approach) for solving non-
classical problem analyses of foundation engineering interaction. In co-authorship
with Dr. A. A. Sedaev, he generalized the method of duel grids for numerical solu-
tion of elastic Hertzian contact problems and suggested the method of generation of
random (or irregular) dual grids. On the base of N.K. Snitko’s ideas, Dr. Aleynikov
proposed the integrated approach to the definition of dominant function for base-
ment soil with a depth variable modulus of elasticity.
He became a recognized specialist in the area of modern method analyses of
foundation engineering interaction. More than once he went abroad with lectures
and for conducting mutual research in Denmark, Spain, Poland, Canada, Germany,
Belgium, the Netherlands and Croatia.
In 2007 S. M. Aleynikov made the translation of the book ‘Boundary Elements.
Theory and Applications’ “written by G.T. Katzikadelis”.
His professional competence was highly appreciated and widely recognized. He
was a full member of Russian Transport Academy, a member of the panel of Rus-
sian Society for Soil Mechanics and Geotechnical Engineering, a member of the
International Society for Soil Mechanics and Geotechnical Engineering (ISSMGE),
and the European Technical Committee ERTC7 “Numerical Methods in Geotech-
nical Engineering”. Dr. S.M. Aleynikov a big was a member of the Expert Council
[email protected]
of the Highest Certification Commission of Education and Science Ministry of the
Russian Federation.

v
vi Obituary

This survey monograph by S. M. Aleynikov was published in Russia in 2001 and

won a high appraisal of the Russian academic community in the field. There is no
doubt its English translation will be useful for Western readers as well.
The image of this indefatigable and original person will be held in our remem-
brance forever.

Voronezh, Russia Professor Igor S. Surovtsev

(VSUAC)

[email protected]
Foreword

The theory of elasticity occupies a prominent position in the development of geome-

chanics and in particular the study of interaction between structural elements and
geomaterials. The general area of contact between geomaterials and structural ele-
ments is referred to as soil–structure interaction and solutions based on the theory of
elasticity have been successfully applied for the study of structural foundations, lay-
ered soil systems, earth-retaining structures and tunnels. The subject matter related
to soil–structure interaction also forms an important component in work related to
the mathematical theory of contact problems starting with the pre-eminent work of
Boussinesq and Hertz. The mathematical theory of elastostatic contact problems in
particular attracted the attention of the earlier Russian school of eminent elasticians,
including Galin, Ufliand, Muskhelishvili, Shtaerman, Koronev, Popov, and others,
with special emphasis on the application of elastic contact problems to structural
foundations made by Gorbunov-Posadov and colleagues. These contributions were
less well known in the English literature in the middle of the last century and a sys-
tematic exposition of the contributions of the Russian researchers to soil–structure
interaction was documented in the treatise by Selvadurai and to contact mechan-
ics documented in the comprehensive volume by Gladwell. Since the publication
of these expository volumes in the 1980s, a number of authoritative volumes have
appeared in the area of elastostatic contact problems, where both classical and non-
classical contact mechanics problems were discussed; a critical examination of the
influences of frictional and unilateral contact problems have found applications in
the treatment of traditional interface mechanics problems as well as new develop-
ments in materials science and advanced materials modelling.
The present volume is a welcome addition to the literature on elastostatic contact
problems with special reference to geomechanics. The volume commences with a
systematic exposition of the fundamental solutions of Kelvin, Boussinesq, Cerruti
and Mindlin, which is the underpinning of many interesting applications of elasto-
statics to contact problems in geomechanics. An aspect of spatial non-homogeneity,
which is relevant to accreted materials that attain increases in the stiffness with
depth due to gravitational effects, is also included in the presentation. The volume
[email protected]
proceeds to the presentation of the traditional contact problems in elastostatics that
include axisymmetric and torsional indentation problems. A chapter in the volume
is devoted to the numerical implementation of the contact problem by appeal to

vii
viii Foreword

boundary elements. This is a useful development that complements the analyti-

cal aspects and contains sufficient depth to enable the reader to appreciate the
contributions that numerical schemes can make to the formulation and solution
of contact problems in geomechanics. The treatment of the class of contact prob-
lems is not restricted to linear responses; non-linear effects arising from non-linear
deformability of the geomaterial is also presented, with the approaches encom-
passing both finite difference and boundary element methods. The conventional
analytical approaches to the formulation and solution of contact problems in geome-
chanics are by necessity restricted to simplified geometries. The author has ventured
to include approaches that can be used for the study of complex foundation shapes
associated with pile foundations and other interactions between piles and founda-
tion bases. Finally, the volume culminates with the study of the mechanics of con-
tact between a structure and a poroelastic material saturated with an incompressible
fluid. The integral equations governing this class of poroelastic contact problem are
summarized and numerical techniques for their solution are presented.
The volume is a very useful contribution to the literature in geomechanics of
contact problems as applied to practical problems involving the interaction of geo-
materials and supporting soils. It contains a balance of analytical and computational
approaches and this makes it a volume that will be of benefit to the researcher and
practitioner alike.

William Scott Professor and James McGill Professor A.P.S. Selvadurai FRSC
McGill University, Montreal, QC, Canada

[email protected]
Preface

The studies of contact interaction in the mechanics of deformable solids have been
carried out since late 19th century, starting from the works of Winkler (1867), Hertz
(1881), and Boussinesq (1885). These studies have been further developed by spe-
cialists in the mechanics of deformable solids as well as in structural mechanics,
bases and foundations. Thousands of papers on this topic have been published, most
of their authors using simplifying assumptions of theoretical modeling on a flat or
axially symmetrical stressed state of a base under a punch (a foundation model). It
is seen from the detailed analysis of references found in literature that mathematical
modeling of essentially spatial contact interaction is in its early stage.
The existing methods for the calculation of complex-shaped foundations are, as
a rule, based on a bed coefficient hypothesis. This results in the introduction of
empirical coefficients into the calculation methods, thus restricting the range of their
application. In the recent years more attention is paid to finite-element approach
to mathematical modeling of spatial contact interaction of foundations with bases.
However, in such studies the dimensionality of the algebraic analogue of the con-
tact problem sharply increases and the problem must be restricted to a number of
partial problems – for example, by imposing restrictions to shape and size of both
the foundations themselves and the soil massifs around the foundations, by consid-
ering loads in assumption of existence of symmetry axes or planes in the calculation
scheme etc. Such studies are rather rare and lack proper consideration of loads of
general spatial type (horizontal, vertical forces and moments) and the possibility of
their combined action. And extremely rare are studies where the complex shape of
various foundations, applied in industrial and civil engineering, is fully taken into
account and theoretically based calculations are made.
Creation of new progressive foundation structures and solution of current prob-
lems of geotechnical engineering result in more complicated problems to be solved
and in the increasing accuracy of the calculation results. The mathematical descrip-
tion of the problems has become so complicated that traditional methods are no
longer suitable for their solution. The lack of reliable mathematical methods to a
certain extent retards elaboration and implementation of new foundation structures
[email protected]
in engineering. Hence, the development of boundary element method (BEM), a rel-
atively new trend in structural mechanics, based on boundary integral equations,
seems to be quite promising from the point of view of both theory and application as

ix
x Preface

an efficient tool for solving 3-D problems. The BEM advantages over other methods
of numeric modeling consist in lowering the problem dimensionality (not the whole
calculation domain is subject to discretization, but only the boundary surface), in
the possibility of a detailed analysis of separate stressed areas, in the simplified data
preparation stage etc. This determines the broad application of BEM for solving var-
ious problems of structural mechanics, especially the unlimited domains. Simultane-
ously, by the present time numerical implementation of BEM to the spatial problems
of structural mechanics in the field of interaction of foundations and bases has not
been sufficiently elaborated yet and appropriate boundary element algorithms and
software are still unavailable. Therefore, there is an urgent need to develop efficient
numeric approaches using the BEM to solve spatial contact problems of interaction
of complex-shaped volumetric punches with deformed bases.
The present book is devoted to one of the BEM application areas – numerical
modeling of contact interaction of rigid foundation structures with soil. The main
attention is paid to the specific features of stress-strained states of elastic bases at
spatial conditions. Contrary to the finite element method, special literature for the
BEM application in mechanics of spatial contact interactions between bases and
foundations is at present unavailable. In recent publications, devoted to the calcula-
tion of bases and foundations, BEM is merely mentioned. On the other hand, well-
known books, describing theory and application of BEM, do not appropriately cover
the issues of creating calculation models and numerical algorithms for analyzing
spatial contact interaction of foundation structures with soil bases.
The whole material is set in six chapters. The first chapter presents some intro-
ductive data while reviewing spatial contact models in geotechnics. Classical funda-
mental solutions for the spatial theory of elasticity obtained by Boussinesq, Cerruti,
Mindlin are quoted as well as their generalizations, suitable for calculating construc-
tions on elastic nonclassical bases. The properties of the influence functions are ana-
lyzed, required for characterizing elastic bases with nonhomogeneous deformation
properties (connected half-spaces, elastic layers of constant and variable thickness).
In the same chapter a numerical-and-analytical procedure is developed for
construction of fundamental solutions of spatial elasticity theory for multilayer
bases without restrictions on the layer thickness and elastic parameters. Using
the two-dimensional Fourier transform, the formulae have been derived, enabling
three-dimensional contact problems for complex-shaped structures deepened into
spatially nonhomogeneous (layered) soils to be solved in the framework of the
BEM numerical algorithm. The final part of the first chapter contains the results on
the formulation of influence functions for elastic bases with variable deformation
properties. The Boussinesq problem is solved for an elastic half-space when the
deformation modulus increases with depth according to a most general law. Proper
relations, enabling adequate description of the experimental data, are considered.
An efficient numerical-and-analytical procedure is developed for construction of the
influence functions, taking into account the soil deformation modulus variation with
[email protected]
depth. All the theoretical results for the influence functions were obtained within a
unique approach enabling all the main types of nonhomogeneities of natural soil
bases to be taken into account.
Preface xi

The second chapter is devoted to the mathematical formulation of mixed prob-

lems of the elasticity theory for a half-space and to the numerical-and-analytical
methods of their solution. The results obtained in this chapter on developing the
mathematical means are the reference data for BEM-based numerical modeling of
the spatial contact interaction. The boundary integral equations of the spatial contact
problem are written for the case when the calculation scheme is accepted in the form
of variously deepened punches undergoing the action of the spatial system of forces.
It is shown how to reduce the initial integral equation system of the contact problem
with respect to the contact stress function and the punch displacement parameters
to the appropriate finite-dimensional algebraic analogue. Much attention is paid to
calculating the matrix coefficients of the resolving system of algebraic equations. A
numerical-and-analytical procedure is given for integrating Mindlin’s fundamental
solutions over flat triangular and quadrangular boundary elements, arbitrary oriented
in the half-space. For convenience, to apply the developed approach in practical
calculations, the boundary integral equations of the spatial contact problems for a
number of essential special cases are presented. The contact problems at axial load-
ing and torsion of absolutely rigid rotation bodies deepened into the half-space, are
considered. Boundary-element formulations of the contact problems for complex-
shaped punches with flat and smooth bases (shallow foundations), situated on spa-
tially nonhomogeneous bases of the semi-infinite elastic massif type are presented.
The third chapter deals with practical implementation of the developed numer-
ical algorithms and substantiation of the reliability of the numerical solutions. It
presents the general characteristics and structure of the Rostwerk software pack-
age for investigating three-dimensional stress-strained states of elastic bases corre-
sponding to the interaction of foundation structures with soil under force factors of
general kind. Procedures for creating input databases are described in detail. Algo-
rithms and modules for automatic formation of boundary element grids in plane and
in space are presented. An original algorithm for triangulation of flat single- and
multiply connected domains, bounded by straight line segments or circle arcs, is
described. An algorithm of generation (according to the given triangulation) of dual
polygonal boundary element grids of Dirichlet cell type is considered. The created
object library of boundary element modules, partitioned into boundary elements,
enabling spatial discretization of complex-shaped surfaces of foundation structures,
is described. Specific features of solving the systems of linear algebraic equations
with asymmetric and close-packed matrices, arising in boundary element analysis,
are considered. For solving such systems by direct (Gauss type) methods a spe-
cial scaling procedure is applied, improving the conditioning of matrices for the
finite-dimensional algebraic analogue of a contact problem. The data about the reli-
ability of the numerical solutions are presented. The BEM accuracy and efficiency
are demonstrated by the examples of the solved test problems for flat punches of cir-
cular, annular and polygonal shapes. Boundary-element solutions for spatial contact
problems concerning a rigid spherical inclusion and a cylindrical deepened punch
[email protected]
in an elastic half-space are obtained. The final part of the chapter gives the results
for numerical-and-analytical solution of the spatial contact problem on impressing
a deepened conical punch into an elastic half-space. The method of determination
xii Preface

of the deformation modulus from tests for deepened conical indenters with different
angles by static loading is substantiated theoretically.
In the fourth chapter the results of the boundary-element solutions of spatial
contact problems for complex-shaped punches, located on the surfaces of elastic
nonclassical bases, are analyzed. The problems under consideration correspond to
the modeling of contact interaction of shallow foundations with elastic nonhomo-
geneous bases. Contact pressure fields under punches of various shape under an
eccentric load (a contact problem on a strongly inclined punch) are obtained. The
influence of non-uniform (over the area) compressibility as well as depth-dependent
nonhomogeneity of the base deformational properties on the formation and devel-
opment of detachment zones, settlements and slopes of punches with the increase
of the absolute values of overturning moments is shown. An algorithm to calculate
the boundaries of the section core for rigid complex-shaped foundation plates from
the stress values is described. Some optimization problems are solved for load and
shape parameter control in order to provide uniform settlement of rigid foundation.
As an example for the application of the developed boundary element method, a
contact problem is solved and the elastic base stress-strained state is determined
for a rigid strip foundation of variable width. In the same chapter a spatial contact
model of the base is built taking into account nonlinear elastic soil properties. A
procedure for the model parameter characterization based on the direct punch test
data is considered. Finally, the chapter contains the studies of contact problems of
bending of orthotropic plates situated on elastic nonclassical bases, performed by
BEM combined with finite difference method.
In the fifth chapter BEM is applied to calculate contact interaction of foundation
structures with soil, taking into account the deepening factor. The need for spatially-
based calculation of bases of deepened foundations is explained. The principles for
foundation structure calculations from the base deformations are briefly reviewed as
well as the existing problem formulations and solution methods for spatial problems
of contact interaction of deepened foundation structures with soil bases. Solutions of
spatial contact problems for deepened monolithic-type foundation structures most
widely used in the recent years are also considered, namely for (1) pyramidal piles;
(2) foundations made of short vertical or inclined bored piles with caps; (3) bored
pile foundations with support extensions; (4) slot foundations with the longitudinal
cross-section of various shape. Heterogeneous stress-strained states of the base are
taken into account as well as the formation of cavities between the soil and the foun-
dation structures. The effect of the foundation shape on its displacement and slope
at various spatial loading is estimated quantitatively. Numerous examples show the
results of the boundary-element modeling to be in good agreement with the exper-
imental measurements performed for spatial foundation structures, in most cases
BEM results being closer to the experiment than those obtained by other known
calculation methods.
Finally, the sixth chapter presents solutions of spatial problems of applied geome-
[email protected]
chanics related to variation of pore pressure in the soil. The influence of the pore
pressure decline on the soil settlement and cracking as well as the induced seis-
micity and other environmental hazards due to pumping out gas and oil deposits or
Preface xiii

intense removal of underground water at industrial or civil engineering is discussed.

The methods for numerical modelling of soil mass deformations due to the reduc-
tion of the pore pressure are described. The approach is based on the application of
integral representations for displacements in a half-space saturated with liquid (or
gas) according to the theory of linear pore-elasticity (filtration consolidation). Spa-
tial deformation of the earth surface due to operating horizontal gas-and-oil wells or
water drains is studied with the account of the run-off mode. Finally, the results for
boundary-element solutions of the spatial contract interaction of structures with the
soil at reduced pore pressure are presented.
The studies, presented in the book, are of applied character and have been
initially oriented at geotechnical objects in industrial and civil engineering. The
boundary element methods developed are suitable for wide applications to calculate
the spatial deformation of soil bases. They provide high reliability and efficiency
of design solutions for foundation structures. Moreover, the boundary element
approach presented here can be helpful for solving other spatially-based problems
of mechanics and mathematical physics.
The book summarizes the studies performed in the recent years in Voronezh
State University of Architecture and Civil Engineering. The author is grateful to
Prof. Viktor N. Nikolaevskiy for his all-round support as well as to Dr. Sergey
V. Ikonin and Dr. Alexandr A. Sedaev for fruitful communications and helpful dis-
cussions which have enabled the book to be made more substantial.
The preparation of the book for publication was essentially supported by
the Dean of Geotechnical Faculty in Varaždin, University of Zagreb (Croatia),
Prof. Mladen Kranjčec and the Vice Dean of the faculty Dr. Božo Soldo. The trans-
lation from Russian would not have been possible without the key professional con-
tribution of Dr. Yuriy Azhniuk from Institute of Electron Physics, Ukr. Nat Acad Sci
(Uzhhorod, Ukraine). The author is truly indebted to all of them.
The author hopes that the work presented in the book can be a helpful study
for numerical experiments in geotechnical engineering and will be grateful to the
readers for their comments.

Voronezh, Russia Sergey M. Aleynikov

[email protected]
[email protected]
Contents

1 Spatial Contact Models of Elastic Bases . . . . . . . . . . . . . . . . 1

1.1 Fundamental Solutions of Static Problems of Spatial
Theory of Elasticity . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Concentrated Forces in an Elastic Body . . . . . . . . . . 1
1.1.2 Green’s Displacement Tensor . . . . . . . . . . . . . . . . 2
1.1.3 Kelvin’s Tensor of Influence . . . . . . . . . . . . . . . . 3
1.2 Elastic Homogeneous Isotropic Half-Space . . . . . . . . . . . . 5
1.2.1 Mindlin’s Solution . . . . . . . . . . . . . . . . . . . . . 5
1.2.2 Boussinesq and Cerruti Solutions . . . . . . . . . . . . . . 6
1.3 Coupled Half-Spaces . . . . . . . . . . . . . . . . . . . . . . . . 8
1.4 Elastic Layered Bases . . . . . . . . . . . . . . . . . . . . . . . . 12
1.4.1 Constant-Width Elastic Layer . . . . . . . . . . . . . . . . 12
1.4.2 Variable-Thickness Elastic Layer . . . . . . . . . . . . . . 17
1.4.3 Multilayer Elastic Half-Space . . . . . . . . . . . . . . . . 25
1.5 Elastic Bases with the Deformation Modulus, Variable with Depth 55
1.5.1 Variation of Deformation Modulus with Depth . . . . . . . 55
1.5.2 Normal Concentrated Force Acting on the
Half-Space Surface . . . . . . . . . . . . . . . . . . . . . 58
1.5.3 Settlement of a Nonhomogeneous Half-Space Surface . . . 63
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
2 Static Analysis of Contact Problems for an Elastic Half-Space . . . . 91
2.1 Boundary Integral Equations of the Contact Problem
for an Absolutely Rigid Punch, Deepened into an Elastic
Half-Space, Under a Spatial Load System . . . . . . . . . . . . . 91
2.2 Finite-Measure Analogue of the Contact Problem Using
Direct Boundary-Element Method . . . . . . . . . . . . . . . . . 96
2.3 Numerical-and-Analytical Method of Integration
of Fundamental Mindlin’s Solutions . . . . . . . . . . . . . . . . 101
2.4 Punch in the Shape of a Rotation Body, Deepened
[email protected]
into an Elastic Half-Space . . . . . . . . . . . . . . . . . . . . . . 108
2.4.1 Axisymmetric Contact Problem . . . . . . . . . . . . . . . 110
2.4.2 Torsion of an Axisymmetric Punch in an Elastic
Half-Space . . . . . . . . . . . . . . . . . . . . . . . . . . 115

xv
xvi Contents

2.5 Contact Problems for Rigid Punches Located on the

Elastic Base Surface . . . . . . . . . . . . . . . . . . . . . . . . . 119
2.5.1 Indentation of a Punch with a Flat Smooth Base
into an Elastic Half-Space . . . . . . . . . . . . . . . . . . 121
2.5.2 Torsion of an Elastic Half-Space by a Rigid Punch . . . . . 126
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
3 Computer Implementation of Boundary-Element Algorithms . . . . 135
3.1 Software for Solving Spatial Problems of Contact
of Foundations with Soil Bases . . . . . . . . . . . . . . . . . . . 136
3.2 Specific Features of Numerical Solutions of Linear
Algebraic Equation Systems with Non-symmetrical
Matrices, Arising in Boundary-Element Analysis . . . . . . . . . 146
3.3 Effective Discretization of 2-D Domains of Complex
Shape at Numerical Solving of Spatial Contact Problems
of Theory of Elasticity . . . . . . . . . . . . . . . . . . . . . . . 150
3.3.1 Algorithm of Triangulation in the Boundary-
Element Method . . . . . . . . . . . . . . . . . . . . . . . 151
3.3.2 Dual Grids and Their Application in Boundary-
Element Method . . . . . . . . . . . . . . . . . . . . . . . 162
3.4 Automated Construction of Spatial Grids of Boundary
Elements on the Surfaces of Contact of Deepened
Foundation Structures with Soil . . . . . . . . . . . . . . . . . . . 174
3.5 Test Examples of Numerical Modeling of Spatial Problems
of Contact Interaction . . . . . . . . . . . . . . . . . . . . . . . . 192
3.5.1 Contact Problems for Flat Punches with a Smooth Base . . 192
3.5.2 Contact Problems with the Account of the
Deepening Factor for Axisymmetric Punches,
Interacting with an Elastic Half-Space . . . . . . . . . . . 218
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
4 Contact Interaction of Shallow Foundations
with Nonhomogeneous Bases . . . . . . . . . . . . . . . . . . . . . . 251
4.1 Spatial Contact Problems for Rigid Flat-Bottom Punches . . . . . 253
4.2 Contact Problems for Rigid Rectangular Punches, Resting
on Elastic Nonhomogeneous Bases . . . . . . . . . . . . . . . . . 278
4.2.1 Contact Interaction at Central Loading . . . . . . . . . . . 282
4.2.2 Contact Interaction at Off-Centre Loading with the
Account of Unilateral Constraints . . . . . . . . . . . . . 295
4.3 Control of the Parameters of Loading and Shape to Provide
a Uniform Settlement of Rigid Foundation Plates . . . . . . . . . 300
4.3.1 Formulation of the Problem and Its Numerical
Implementation . . . . . . . . . . . . . . . . . . . . . . . 301
[email protected]
4.3.2 External Load Control . . . . . . . . . . . . . . . . . . . . 303
4.3.3 Shape Parameter Control . . . . . . . . . . . . . . . . . . 307
Contents xvii

4.4 Spatial Stress-Strained State of the Base of a Rigid Strip

Variable-Width Foundation . . . . . . . . . . . . . . . . . . . . . 311
4.4.1 Contact Problem for a Variable-Width Strip Foundation . . 312
4.4.2 Stress-Strained State of a Strip Foundation Base . . . . . . 315
4.4.3 Contact Pressure Distribution in the Area of the
Strip Foundation Width Variation . . . . . . . . . . . . . . 317
4.5 Calculation of the Section Kernel Boundary for Rigid
Foundation Plates . . . . . . . . . . . . . . . . . . . . . . . . . . 323
4.6 Numerical Algorithms of Solving Boundary Integral
Equations in Spatial Contact Problems for a Nonlinearly
Deformable Base . . . . . . . . . . . . . . . . . . . . . . . . . . 334
4.6.1 Spatial Contact Model for a Nonlinearly
Deformable Base . . . . . . . . . . . . . . . . . . . . . . 335
4.6.2 System of Nonlinear Contact Equations of the
Contact Problem for Absolutely Rigid Punches of
a Complex Shape with a Flat Base . . . . . . . . . . . . . 337
4.6.3 Iterative Processes of Solving a Finite-Measure
Analogue of the Spatial Contact Problem for a
Nonlinearly Deformable Base . . . . . . . . . . . . . . . . 339
4.6.4 Contact Problem for a Round Punch on a
Nonlinearly Deformable Base . . . . . . . . . . . . . . . . 341
4.6.5 Estimation of Nonlinear Deformation Effects from
Punch Test Results . . . . . . . . . . . . . . . . . . . . . 348
4.7 Contact Problem for Orthotropic Foundation Plates
with the Account of the Specific Features of Spatially
Nonhomogeneous Base Deformation . . . . . . . . . . . . . . . . 351
4.7.1 Static Calculations of Foundation Plates on Elastic Bases . 352
4.7.2 System of Integro-Differential Equations of
Bending of a Plate, Resting on an Elastic Base . . . . . . . 358
4.7.3 Calculation of Rectangular Orthotropic Plates
Based on Combining Finite-Difference and
Boundary-Element Methods . . . . . . . . . . . . . . . . 361
4.7.4 Examples of Numerical Modelling of the Contact
Interaction of Plates with Elastic Bases . . . . . . . . . . . 364
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 372
5 Calculation of Bases for Rigid Complex-Shaped Deepened
Foundations According to the Second Limiting State in a
Three-Dimensional Formulation . . . . . . . . . . . . . . . . . . . . 385
5.1 General Information on the Calculation of Bases for
Foundation Structures from the Deformations . . . . . . . . . . . 390
5.2 Spatial Problems for Calculation of Foundation Bases with
[email protected]
the Account of the Depth Factor . . . . . . . . . . . . . . . . . . 396
5.3 Calculation of Bases for Pyramidal Piles Under Vertical,
Horizontal, and Momental Loads . . . . . . . . . . . . . . . . . . 415
xviii Contents

5.3.1 Existing Approaches to the Calculation of Piles

with a Variable Cross-Section . . . . . . . . . . . . . . . . 416
5.3.2 Calculation for the Vertical Load . . . . . . . . . . . . . . 420
5.3.3 Calculation for the Action of a Horizontal Load . . . . . . 420
5.3.4 Calculation for the Action of an Inclined Load . . . . . . . 422
5.3.5 Calculation for the Combined Action of an Inclined
Force and a Moment . . . . . . . . . . . . . . . . . . . . 423
5.4 Interaction of Bases and Rigid Bored Foundations with
Vertical and Inclined Piles . . . . . . . . . . . . . . . . . . . . . 424
5.4.1 Structure, Design, and Specific Features of
Calculation of Rigid Pile Foundations with Short
Piles and a Pile Raft . . . . . . . . . . . . . . . . . . . . . 425
5.4.2 Vertical Cylindrical Piles Under an Inclined Load . . . . . 428
5.4.3 Foundations with Inclined Piles and a Rectangular
Pile Raft . . . . . . . . . . . . . . . . . . . . . . . . . . . 435
5.5 Spatial Contact Problem for a Bored Pile Foundation
with a Widening . . . . . . . . . . . . . . . . . . . . . . . . . . . 438
5.5.1 Production and Structures of Bored Pile
Foundations with a Support Widening . . . . . . . . . . . 439
5.5.2 Engineering Methods for Calculation of Bored Pile
Foundation Bases from the Base Deformation . . . . . . . 441
5.5.3 Calculation of Deformations of the Base of a Bored
Pile Foundation with a Spheroconical Widening
Under a Central Loading (Axisymmetric Contact Problem) 443
5.5.4 Calculation of Displacements and Slopes of a
Bored Pile Foundation Under an Inclined Load . . . . . . 448
5.6 Calculation of Contact Interaction of Bases with Slotted
Foundations of Industrial and Civil Buildings . . . . . . . . . . . 454
5.6.1 Slotted Foundations of Various Structural Shapes . . . . . 454
5.6.2 Calculation of Slotted Foundations Based on the
Base Deformation . . . . . . . . . . . . . . . . . . . . . . 457
5.6.3 Contact Stress on the Lateral Surface of a Slotted
Foundation . . . . . . . . . . . . . . . . . . . . . . . . . 476
5.6.4 Slotted Foundations with Lateral Widenings . . . . . . . . 491
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493
6 Spatial Contact Problems for Porous Elastic Bases . . . . . . . . . . 505
6.1 Soil Mass Deformation Due to the Pore Pressure Decline . . . . . 509
6.1.1 Integral Representation of Displacements
in a Porous Elastic Medium . . . . . . . . . . . . . . . . . 509
6.1.2 Dilatation Relations . . . . . . . . . . . . . . . . . . . . . 512
6.2 Distribution of Pressure in a Layer in Case of Functioning
[email protected]
Horizontal Wells . . . . . . . . . . . . . . . . . . . . . . . . . . 514
6.2.1 Distributed Sources of Predetermined Intensity . . . . . . 514
6.2.2 Account of the Finite Radius of the Well . . . . . . . . . . 516
Contents xix

6.3 Contact Problems for Foundation Structures at a Reduced

Pore Pressure in the Soil . . . . . . . . . . . . . . . . . . . . . . 517
6.3.1 Integral Equations of a Spatial Contact Problem . . . . . . 517
6.3.2 Finite-Dimensional Algebraic Analogue of the
Integral Equation System . . . . . . . . . . . . . . . . . . 519
6.3.3 Numerical Algorithm of Solution of the Contact Problem . 520
6.3.4 Contact Problem for Shallow Foundations . . . . . . . . . 522
6.4 Examples of Numerical Calculations . . . . . . . . . . . . . . . . 524
6.4.1 Spatial Deformation of the Land Surface . . . . . . . . . . 526
6.4.2 Surface Deformations of the Layer . . . . . . . . . . . . . 532
6.4.3 Settlements and Slopes of Rigid Foundation Plates . . . . 533
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 534
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 537
Appendix A Fundamental Solutions of Spatial Theory of
Elasticity for a Homogeneous Isotropic Half-Space . . . . . . . . . . 543
Appendix B Numerical Schemes for Surface Integral Calculations . . 555
Appendix C Round Punch on an Elastic Layer of Variable
Thickness at Central and Off-Centre Load . . . . . . . . . . . . . . 569
Appendix D Foundation Under a Tower-Type Structure on a
Wedge Base . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 581
Appendix E Finite-Difference Equations of Cylindrical Bend
of Orthotropic Slabs Located on an Elastic Foundation . . . . . . . 591
Appendix F Calculation of the Base for a Pyramidal Pile
Under Vertical Load According to the “Instructions Manual
for Design of Foundations Made of Pyramidal Piles” . . . . . . . . . 603
Appendix G Isolines of Contact Stress on a Lateral Surface of
a Slotted Foundation . . . . . . . . . . . . . . . . . . . . . . . . . . . 613
Appendix H Numeric Schemes of Volume Integration . . . . . . . . . 629
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 637

[email protected]
[email protected]
Chapter 1
Spatial Contact Models of Elastic Bases

Abstract The first chapter presents some introductive data while reviewing spatial
contact models in geotechnics. Classical fundamental solutions for the spatial the-
ory of elasticity obtained by Boussinesq, Cerruti, Mindlin are quoted as well as their
generalizations, suitable for calculating constructions on elastic nonclassical bases.
The properties of the influence functions are analyzed, required for characterizing
elastic bases with nonhomogeneous deformation properties (connected half-spaces,
elastic layers of constant and variable thickness). A numerical-and-analytical pro-
cedure is developed for construction of fundamental solutions of spatial elasticity
theory for multilayer bases without restrictions on the layer thickness and elastic
parameters. Using the two-dimensional Fourier transformation, the formulae have
been derived, enabling three-dimensional contact problems for complex-shaped
structures deepened into spatially nonhomogeneous (layered) soils to be solved in
the framework of the boundary-element method numerical algorithm. The final part
of the first chapter contains the results on the formulation of influence functions
for elastic bases with variable deformation properties. The Boussinesq problem is
solved for an elastic half-space when the deformation modulus increases with depth
according to a most general law. Proper relations, enabling adequate description of
the experimental data, are considered. An efficient numerical-and-analytical proce-
dure is developed for construction of the influence functions, taking into account
the soil deformation modulus variation with depth. All the theoretical results for the
influence functions were obtained within a unique approach enabling all the main
types of nonhomogeneities of natural soil bases to be taken into account.

1.1 Fundamental Solutions of Static Problems of Spatial Theory

of Elasticity

1.1.1 Concentrated Forces in an Elastic Body

[email protected]
One of the most important problems of mathematical theory of elasticity is construc-
tion of a solution of the problem of determination of displacements and stresses in
an elastic body, loaded by a unit concentrated force. In case such solutions exist in

S.M. Aleynikov, Spatial Contact Problems in Geotechnics, Foundations of 1

Engineering Mechanics, DOI 10.1007/b11479_1,  C Springer-Verlag Berlin Heidelberg 2011
2 1 Spatial Contact Models of Elastic Bases

linear theory of elasticity, the solutions for the same body at arbitrary loading can
be obtained rather easily (in quadratures).
A unit concentrated force, acting in a point K(ξ ,η,ζ ) (the “source point”) of an
elastic body V in the direction of OXk axis, is treated [92] as a singular distribution
of mass forces in the form

Ai = δ(N − K)δik (1.1)

where δ ik is the Kronecker symbol, δ(N − K) = δ(x1 − ξ )δ(x2 − η)δ(x3 − ζ ), δ(t) is

the Dirac delta function [151].
It follows from Eq. (1.1) that
 
Ai dV = δ(N − K)δik dV(N) = δik , (1.2)
V V

i.e. the components of the concentrated force of Eq. (1.1) in the direction of the
OXi axis are nonzero, and in all points N(x1 ,x2 ,x3 ) = K(ξ ,η,ζ ) the distribution of
Eq. (1.1) is zero. In the point K(ξ ,η,ζ ) itself the mass force value is infinite, and
according to Eq. (1.2), the integral over the volume of the elastic body V is unit.
In the point of application of the concentrated force infinite displacements arise
what is inconvenient for using the corresponding solutions in theory of elasticity.
However, in all applications of the idealized solution under consideration integration
of products of the infinite solutions and elementary volumes of areas is performed,
finally resulting in finite values of displacements and stresses in the elastic body.
Note that the idealized condition of the concentrated force action corresponds to
the problem of a load, distributed over a small area, degenerating into a point, under
a condition of the value of the main traction vector remaining unchanged. As noted
in [95], introduction of concentrated forces is justified by essential advantages for
solving boundary problems in the integral formulation.

1.1.2 Green’s Displacement Tensor

Let a unit concentrated force, acting in an elastic body V, result in displacements
Ui (j) (N,K) where i = 1, 2, 3 is an index, indicating the displacement vector compo-
nent number, j = 1, 2, 3 is an index, indicating the concentrated force direction, N(x1 ,
x2 , x3 ) is the observation point, K(ξ ,η,ζ ) is the concentrated force application point
(Fig. 1.1). The mathematical problem of determination of the displacements
Ui (j) (N,K) is reduced to the integration of a system of three non-uniform second-
order partial differential equations of elliptical type (fundamental Lamé equations)
[email protected]
(k) (k)
μ Ui + (λ + μ)Uj, ji + δik δ(N − K) = 0(i,j,k = 1,2,3) (1.3)
1.1 Fundamental Solutions of Static Problems of Spatial Theory of Elasticity 3

Fig. 1.1 Components of the

displacement vector due to
the action of a unit
concentrated force Ai , applied
in the point K(ξ ,η,ζ ) in an
elastic body V

with given boundary conditions. In Eq. (1.3) λ and μ are the Lamé elastic
constants,  is a three-dimensional Laplace operator in a Cartesian coordinate
system OX1 X2 X3 , comma means differentiation; as usual, summation over the
repeated indices is assumed. It is shown in [92] that the displacement matrix

⎛ ⎞
(1) (2) (3)
U1 U1 U1
⎜ ⎟
⎜ (1) (2) (3) ⎟
⎜ U2 U2 U2 ⎟
⎜ ⎟
⎝ (1) (2) (3) ⎠
U3 U3 U3

forms a symmetrical second-rank tensor, i.e. Ui (K,K  ) = Uj(i) (K  ,K) (the Maxwell
(j)

theorem for work reciprocity for concentrated forces). Consequently, for the dis-
placement functions Ui (j) , called the fundamental solutions of static theory of elas-
(j)
ticity (Green’s displacement functions), tensor notations are convenient. Ui =
(i)
Uj = Uij = Uji .

1.1.3 Kelvin’s Tensor of Influence

As noted above, by using the Green’s displacement functions, the solutions of
boundary problems of theory of elasticity are obtained in quadratures. Therefore,
a key point of solving the problems of theory of elasticity is of the Green’s function
determination. However, in the three-dimensional case the problem of the Green’s
function determination for bounded domains meets strong difficulties [92, 151].
Now this function has been obtained for a class of elastic bodies, rather limited in
shape. The most easily obtained [92] is the expression for the displacement tensor
in an unlimited domain (fundamental Kelvin’s solution for an elastic space):
[email protected]

B
Uij (N,K) = A δij − R ,ij
R
4 1 Spatial Contact Models of Elastic Bases

where R ≡ R(N,K) = [(x1 − ξ )2 + (x2 − η)2 + (x3 − ζ )2 ]1/2 is the distance between
the point K(ξ ,η,ζ ), to which the load is applied, and a point N(x1 , x2 , x3 ) of the space
(the observation point),
λ+μ 2(λ + 2μ)
A= ,B= .
8π μ(λ + 2μ) λ + 2μ
In an extended form Eq. (1.4) is given by
 

1 3 − 4ν ∂ 1
Uij (K, N) = · δij − zj =
16π G(1 − ν) R ∂xi R

1 1 ∂ 2R 1 3 − 4ν z i zj
= δij R − · = · δij + 3 ,
8π G 2(1 − ν) ∂xi ∂xj 16π G(1 − ν) R R
zi = xi − ζi ,ζ1 = ξ ,ζ2 = η,ζ3 = ζ ; i,j = 1,2,3.

As one can see, the components of the fundamental Kelvin’s solution have a sin-
gularity of the order of 1/R in the concentrated force application point K. According
to Sect. 1.1.1, the fundamental Kelvin’s solution (1.4) corresponds to the displace-
ments arising in the point N(x1 , x2 , x3 ) of an infinite body due to a unit concentrated
force, applied to the point K(ξ ,η,ζ ) in the direction of the OXj axis. This solution
can be treated as the Green’s function of the influence function for an infinite elastic
medium. Using the Hooke law, one can easily obtain stresses, corresponding to the
Kelvin displacements
−1 
σjki (N, K) = (1 − 2ν)(R,k δij + R,j δjk − R,i δjk ) + 3R,i R,j R,k .
8π (1 − ν)R2

The fundamental Kelvin’s solution for the displacements Eq. (1.5) and stresses
Eq. (1.6) enables one to obtain the solution of any spatial problem of theory of elas-
ticity for a homogeneous isotropic linearly deformable medium. The fundamental
Kelvin’s solution is most successfully applied for solving spatial problems of the-
ory of elasticity using the boundary integral equation method for finite-size bodies
embedded in an infinite space with the same elastic properties [29].
The solution of contact (mixed) problems for elastic spatially nonhomogeneous
half-spaces, being the most popular geomechanical models of soil bases, on the
surface of which uniform boundary conditions for the stresses are given, is most
effectively accomplished by using special fundamental solutions. In the subsequent
sections of this chapter such fundamental solutions of spatial theory of elasticity are
considered when unit concentrated forces act inside or on the surface of elastic lay-
ers and half-spaces, whose mechanical characteristics take into account the natural
conditions of soil masses. The main advantages in using such fundamental solutions
(influence functions, calculated or contact models of bases) consist in the boundary
[email protected]
conditions on the free surface and on the surfaces of separation of different elastic
layers being automatically fulfilled as well as variation of the deformation properties
with depth, anisotropy, etc. being taken into account.
1.2 Elastic Homogeneous Isotropic Half-Space 5

1.2 Elastic Homogeneous Isotropic Half-Space

1.2.1 Mindlin’s Solution

Consider a unit concentrated force, acting inside a homogeneous half-space x3 ≥ 0,
whose boundary plane x3 = 0 is load-free. This problem was first solved by Mindlin
[77–79] who used the method of superposition of special partial solutions for the
infinite space. Later the Mindlin displacement tensor was obtained using different
approaches: integral Fourier transformation method [92], reflection method [72],
Green’s function method for the harmonic Dirichlet problem [136]. The Mindlin’s
solution is obtained in an explicit form [29] by summation of 18 deformation ker-
nels, following from the Kelvin’s solution (six for each of the three force compo-
nents).
The most compact of the known forms of expressions for the Mindlin displace-
ment Ui (j) and stress σmq (i) tensors is as follows:

Uij (K, N) = [16π G(1 − ν)]−1 {(3 − 4ν)δij R−1 −3 (1) (1)
1 + zi zj R1 + (8ν − 12ν + 5)×
2

δij R−1 −3
(2) (2) (2) (2)
2 + (3−4ν)[zi zj − 2ξ3 (zi δj3 + zj δi3 )(1−δi3 δj3 )]R2 +

+2ξ3 (z3 − ξ3 )(1 − 2δi3 )(R22 δij − 3zi zj )R−5

(2) (2) (2)
2 −
(2) (2) (2)
−4(1 − ν)(1 − 2ν)[z3 δij (1 − δi3 ) − zi δj3 + zj δi3 +

+zi zj (1 − δi3 )(1 − δj3 )(R2 + z3 )−1 ]R−1 −1

(2) (2) (2) (2)
2 (R2 + z3 ) },
(1.7)

σmq (K, N) = [8π (1 − ν)]−1 {(1 − 2ν)(zi δmq − zq δim − zm δiq )R−3
(i) (1) (1) (1)
1

−3zi zm zq R−5 −3
(1) (1) (1) (2) (2) (2)
1 − (1 − 2ν)(zi δmq − zq δim − zm δiq )R2

−2(1 − 2ν)ξ3 [3δmq δi3 + (δm3 δqi + δq3 δmi )(1 − 2δm3 δq3 δi3 )]R−3
2

−3(3 − 4ν)zm zq (zi − 2ξ3 δi3 )R−5

(2) (2) (2) (2)
2 + 6ξ3 (1 − 2δi3 )[ξ3 zi δmq
(2) (2) (2) (2) (2) (2) (2)
−(z3 − ξ3 )(zq δim + zm δiq ) + (1 − 2ν)(zi zq δ3m + zi zm δ3q
(1.8)
−z3 zi δmq )]R−5 −7
(2) (2) (2) (2) (2) (2)
2 + 30(z3 − ξ3 )ξ3 zi zm zq (1 − 2δi3 )R2
  
+4(1 − ν)(1 − 2ν) (δmq − δm3 δq3 ) zi R−2
(2) (2) −1
2 − δi3 (R2 + z3 )
(2) (2) (2)
−(1 − δq3 )(1 − δm3 ) [(1 − δi3 )(zi δmq + zq δim + zm δiq )
−δi3 zm zq (2R2 + z3 )R−2 −2
(2) (2) (2) (2) (2) (2) (2)
2 − (1 − δi3 )zi zm zq (3R2 + z3 )R2
  
[email protected]
×(R2 + z(2) −1 (R + z(2)
−2
R−1
3 ) 2 3 2

where G is the shear modulus and ν is the Poisson ratio of the soil,
6 1 Spatial Contact Models of Elastic Bases

(1) (2) (1) (2) (1)

z1 = ζ1 − ξ1 = z1 ; z2 = ζ2 − ξ2 = z2 ; z3 = ζ3 − ξ3 ;
(2) (1)2 (1)2 (1)2
z3 = ζ3 + ξ3 ; R1 = (z1 + z2 + z3 )0.5 ;
(2)2 (2)2 (2)2
R2 = (z1 + z2 + z3 )0.5 .
In Eqs. (1.7) and (1.8) R1 is the distance between the observation point N and the
loading point K (point of application of the concentrated force), R2 is the distance
between the point N and a point K̃(ξ , η, ζ ), being a mirror reflection image of the
point K with respect to a free surface x3 = 0 of the half-space (Fig. 1.2).

Fig. 1.2 Geometrical scheme of the Mindlin problem of action of concentrated forces in an elastic
half-space

For a given direction of the unit concentrated force the first terms for each compo-
nent of the Mindlin displacement and stress tensors are the corresponding Kelvin’s
solutions. In Eqs. (1.7) and (1.8) all terms of the tensor components, except the
first ones, contain the coordinates of the imaginary loading point K̃, what, as shown
by Mindlin, enables the condition of vanishing of stresses at the boundary surface
x3 = 0 to be fulfilled. As one can see, at large distances R = R1 ≈R2 from the load-
ing point the displacement Ui (j) and stress σmq (i) functions decrease as 1/R and 1/R2 ,
respectively. Consequently, Eqs. (1.7) and (1.8) determine the stress-strained state
near the point of application of the concentrated force with respect to the points of
the elastic body, located at a large distance from the point K (R → ∞) where the
half-space can be nominally considered as if they were fixed [54].

1.2.2 Boussinesq and Cerruti Solutions

[email protected]
Consider a particular case of the Mindlin’s solution when unit concentrated forces
act in the point K(ξ ,η,ζ ), located on the elastic half-space surface (Fig. 1.3). With
a limiting transition in Eqs. (1.7) and (1.8) at ζ→0, assuming R1 = R2 = R =
1.2 Elastic Homogeneous Isotropic Half-Space 7

Fig. 1.3 Calculation scheme

to the problems of
Boussinesq and Cerruti


(x1 − ξ )2 + (x2 − η)2 + x32 , one obtains an expression for the fundamental tensors
of displacements and stresses in the following form:

Uij (K, N) = (4π G)−1 {(2ν − 1)δij R−1 + zi zj R−3 − (1 − 2ν)[Bij zi zj R−1 (R + z3 )−2
+(Aij z3 − zi δ3j + zj δij )R−1 (R + z3 )−1 ]},
(1.9)

(i) 3  zi zm zl 1 − 2ν zi (R + z3 ) − R2 δi3 zm zl (2R + z3 )
σml =
(K, N) + A ml + Diml 3
2π R 5 3 R (R + z3 )
3 R (R + z3 )2


zi zm zl (3R + z3 ) − (zi δml + zl δim + zm δil )R (R + z3 )
2
+Ciml
R3 (R + z3 )3
 (1.10)
where R = z1 + z2 + z3 ; z1 = x1 − ξ1 ; z2 = x2 − ξ2 ; z3 = x3 ; Aim = δim − δi3 δ3m ;
2 2 2

Bim = (1 − δi3 )(1 − δ3m );Diml = Bml δi3 ; Ciml = Bim (1 − δ3 l ); i,j,l,k,m = 1,3.
As follows from the quoted expressions, the obtained equations combine the
known fundamental Boussinesq solutions [40, 50, 54, 92] (for a concentrated force,
acting normally to the half-space surface A1 = A2 = 0, A3 = 1) and Cerruti solu-
tions (for tangential forces, acting on the half-space surface, A1 = 1(0); A2 = 0(1);
A3 = 0).
Equations (1.9) and (1.10) for the fields of deformations and stresses caused by
concentrated normal and tangential forces, in principle, enable one to solve the prob-
lems of a stress-strained state of an elastic half-space, on the surface of which given,
in general case, inclined, loads act. The resulting displacements and stresses from
the action of loads, distributed over an arbitrary finite domain on the half-space
surface, are obtained by integration, using the superposition principle [54, 92]. In
a closed form, determination of the stress-strained state in a half-space under sur-
face loads has been performed only for the domains of the most simple geomet-
rical shape: circular and ring-shaped [40, 54, 146], rectangular [68], elliptical [71,
[email protected]
110, 141].
The difficulties of application of analytical methods for the loads being dis-
tributed over domains of complex geometry can be overcome using various numeri-
cal methods of integration of the Boussinesq and Cerruti fundamental solutions. We
8 1 Spatial Contact Models of Elastic Bases

have proposed [8, 9] an effective version of the numerical method of calculation of

displacements and stresses in the case of surface loads of arbitrary form, distributed
over the areas of a given geometry. The possibilities of this approach were tested
for the example of analysis of the stress-strained state of the elastic half-space in
case its square domain being loaded by a uniformly distributed load or a load, being
varied linearly or parabolically. In particular, the range of relative depths, for which
the angular point method [140] is applicable independently of the loading type, is
determined. Using the developed approach, for practical geotechnical purposes the
surface deformations of various elastic bases were determined for circular, elliptical
and ring-shaped (concentric and eccentric) loading domains [9, 12].
The expressions for the Mindlin, Boussinesq, and Cerruti fundamental solutions
in the extended form are given in Appendix A. From the literature the most impor-
tant are fundamental solutions of spatial theory of elasticity for a homogeneous
half-space with a fixed boundary [128, 135], for a nonhomogeneous half-space with
the Poisson ratio variable in depth [24], for anisotropic half-spaces [36, 60, 64, 117,
123, 145].

1.3 Coupled Half-Spaces

Consider a full system of expressions for the components of the Green’s displace-
ment tensor, obtained from the solution of the problem of the action of concen-
trated forces within contacting half-spaces with different deformational character-
istics. This system is convenient for application in the boundary-element analysis.
No displacement of layers along the boundary is assumed for both vertical and hori-
zontal directions, i.e. the case of two coupled half-spaces with different elastic con-
stants E1 , ν1 and E2 , ν2 is considered. The forces ate applied in the point K(ξ , η, ζ )
(Fig. 1.4). The origin of coordinates is set on the elastic half-space separation sur-
face; the OZ axis is directed towards the first half-space normally to the common

[email protected]

Fig. 1.4 Unit concentrated forces in coupled half-spaces

1.3 Coupled Half-Spaces 9

half-space boundary, which is the OXY plane. The problem under consideration was
solved by Plevako [100], using the representation of the displacement vector compo-
nents in terms of Galerkin vector and three arbitrary biharmonical functions which
are unambiguously determined due to the boundary conditions of the half-space
coupling being satisfied. Equality of all three displacements as well as of normal
and two tangential stresses are taken as such conditions. In view of the Plevako fun-
damental solution being sufficiently universal, here we quote the expressions for the
displacement tensor components for each of the half-spaces:
Half-space No. 1

1 x12 3 − 4ν1 4(1 − ν1 ) − 2γ
U (1) (K, N) = − + +
16π G1 (1 − ν1 ) R2 3 R2 R1
 
1 3x12 α + 2(1 − ν1 )(3 − 4ν1 )β + (1 − 2ν1 )β − 2γ + 3 − 4ν1
+2βζ z − 5 − ×
R13 R1 R1 + z 1
   
x12 1 x12
× 1− + (3 − 4ν1 )β − 3 ,
R1 (R1 + z1 ) R1 R1
(1.11)

x1 y1 1 (3 − 4ν1 )β 6βζ z
V (1) (K, N) = − − 5 +
16π G1 (1 − ν1 ) R32 R13
R1

(1.12)
α + 2(1 − ν1 )(3 − 4ν1 )β + (1 − 2ν1 )β − 2γ + 3 − 4ν1
+ ,
R1 (R1 + z1 )2
x1 −z2 (3 − 4ν1 )βz2 6βζ zz1
W (1) (K, N) = + + −
16π G1 (1 − ν1 ) R2 3 R13 R51

(1.13)
α − β − 4(1 − ν1 )(1 − 2ν1 )β − 1
− ,
R1 (R1 + z1 )
1 y21 3 − 4ν1 4(1 − ν1 ) − 2γ
U (2) (K, N) = − + +
16π G1 (1 − ν1 ) R32 R2 R1
 
1 3y21 α + 2(1 − ν1 )(3 − 4ν1 )β + (1 − 2ν1 )β − 2γ + 3 − 4ν1
+2βζ z − 5 − ×
R32 R2 R1 + z 1
   
y21 1 y21
× 1− + (3 − 4ν1 )β − 3 ,
R1 (R1 + z1 ) R1 R1
(1.14)

−x1 y1 1 (3 − 4ν1 )β 6βζ z
V (2) (K, N) =
[email protected] − − 5 +
16π G1 (1 − ν1 ) R2 3 R13 R1

(1.15)
α + 2(1 − ν1 )(3 − 4ν1 )β + (1 − 2ν1 )β − 2γ + 3 − 4ν1
+ ,
R1 (R1 + z1 )2
10 1 Spatial Contact Models of Elastic Bases


−y1 −z2 (3 − 4ν1 )βz2 6βζ zz1
W (2) (K, N) = + + −
16π G1 (1 − ν1 ) R32 R32 R51

(1.16)
α − β − 4(1 − ν1 )(1 − 2ν1 )β − 1
− ,
R1 (R1 + z1 )

−x1 −z2 (3 − 4ν1 )βz2 6βzζ z1
U (3) (K, N) = + − −
16π G1 (1 − ν1 ) R32 R31 R51

(1.17)
β + 4(1 − ν1 )(1 − 2ν1 )β − α + 1
− ,
R1 (R1 + z1 )

−y1 −z2
(3 − 4ν1 )βz2 6βzζ z1
V (3) (K, N) = + − −
16π G1 (1 − ν1 ) R32 R31 R51

(1.18)
β + 4ν1 (1 − ν1 )(1 − 2ν1 )β − α + 1
− ,
R1 (R1 + z1 )

−1 3 − 4ν1 z2 4(1 − ν1 )(1 − 2ν1 )β + α − 1
W (3) (K, N) = + 23 −
16π G1 (1 − ν1 ) R2 R2 R1

β[(3 − 4ν1 )z21 + 2ζ z] 6βzζ z21
− + .
R31 R51
(1.19)

Half-space No. 2

 
1 2γ 2z(α − β − 1) 2R 2 − z 2
U (1) (K, N) = + 1 − x12 2
16π G2 (1 − ν1 ) R2 R2 (R2 − z2 ) R2 (R2 − z2 )
 
(3 − 4ν2 )α + [1 − 2(ν1 − ν2 )] (β + 1) − 2γ 2
x1
+ 1−
R2 − z2 R2 (R2 − z2 )
 
1 x2
− (β + 1) − 13 ,
R2 R2
(1.20)


x1 y 1 z(2R2 − z2 )
=
V (1) (K, N)
[email protected] −2(α − β − 1) 3
16π G2 (1 − ν1 ) R2 (R2 − z2 )
 (1.21)
(3 − 4ν2 )α + [1 − 2(ν1 − ν2 )](β + 1) − 2γ (β + 1)
− + ,
R2 (R2 − z2 )2 R32
1.3 Coupled Half-Spaces 11

x1 (3 − 4ν2 )α − [3 − 2(ν1 + ν2 )](β + 1)
W (1) (K, N) =
16π G2 (1 − ν1 ) R2 (R2 − z2 )
 (1.22)
(β + 1)ζ + z(2α − β − 1)
− ,
R32  
1 2γ 2z(α − β − 1) 2R2 − z2
U (K, N) =
(2) + 1 − x1 2
2
16π G2 (1 − ν1 ) R2 R2 (R2 − z2 ) R2 (R2 − z2 )
 
(3 − 4ν2 )α + [1 − 2(ν1 − ν2 )](β + 1) − 2γ x12
+ 1−
R2 − z 2 R2 (R2 − z2 )
 
1 x2
− (β + 1) − 13 ,
R2 R2
(1.23)

x1 y 1 z(2R2 − z2 )
V (2) (K, N) = −2(α − β − 1) 3
16π G2 (1 − ν1 ) R2 (R2 − z2 )2
 (1.24)
(3 − 4ν2 )α + [1 − 2(ν1 − ν2 )](β + 1) − 2γ (β + 1)
− + ,
R1 (R1 + z2 )2 R32

x1 (3 − 4ν2 )α − [3 − 2(ν1 + ν2 )](β + 1)
W (2) (K, N) =
16π G2 (1 − ν1 ) R2 (R2 − z2 )
 (1.25)
(β + 1)ζ + z(2α − β − 1)
− ,
R32

−x1 [3 − 2(ν1 + ν2 )](β + 1) − (3 − 4ν2 )α
U (K, N) =
(3)
16π G2 (1 − ν1 ) R2 (R2 − z2 )
 (1.26)
2αz − z1 (β + 1)
− ,
R32

−y1 [3 − 2(ν1 + ν2 )](β + 1) − (3 − 4ν2 )α
V (3) (K, N) =
16π G2 (1 − ν1 ) R2 (R2 − z2 )
 (1.27)
2αz − z1 (β + 1)
− ,
R32

−1 2(ν2 − ν1 )](β + 1) + (3 − 4ν2 )α
W (K, N) =
(3)
16π G2 (1 − ν1 ) R2
 (1.28)
[2αz − z1 (β + 1)]z2
+ ,
R32

[email protected] 
where x1 = x − ξ ; y1 = y − η; z1 = z − ζ ; z2 = z + ζ; R1 = x12 + y21 + z21 ;

R2 = x12 + y21 + z22 ;
12 1 Spatial Contact Models of Elastic Bases

ξ , η, ζ are the coordinates of the point K of application of the unit concentrated

forces, N(x, y, z) – is a field point (observation point),

G2 E2 (1 + ν1 ) 8 k(1 − ν1 )[k(1 − ν1 ) + (1 − ν2 )]
k= = ;α= ;
G1 E1 (1 + ν2 ) (k + 3 − 4ν2 )[k(3 − 4ν1 ) + 1]
k−1 k
β= ; γ = 4(1 − ν1 ) .
k(3 − 4ν1 ) + 1 k+1
The necessary execution of limiting transitions is checked directly.
The Kelvin’s solution for an infinite homogeneous elastic body follows from
Eqs. (1.11)–(1.28) if the deformation parameters of the two half-spaces coincide
(E1 = E2 , ν1 = ν2 ). One also arrives at the same result at ζ→∞, since the terms,
containing R1 , disappear, and deals with a fundamental solution for an infinite elastic
medium with the deformation parameters E1 , ν1 .
If one implies ζ = 0 and E2 = 0, then the Plevako solution is transformed into a
combined Boussinesq-Cerruti solution for the half-space No. 1.
At E2 = 0 one arrives at the Mindlin’s solution for a concentrated force, acting
inside an elastic half-space No. 1.
At ζ = 0 the concentrated force acts on the boundary of the coupled half-spaces.
The Plevako solution coincides with the Vasilyev solution [149].
If the lower half-space is absolutely rigid (E2 = ∞), one obtains a solution of
the Shandru problem [128, 135] of a concentrated force, acting in a homogeneous
half-space with a restrained boundary.
It is seen that the Plevako solution (1.11)–(1.28) generalized all main funda-
mental solutions for an elastic half-space. Meanwhile, this solution is also of an
independent interest. The application of the Mindlin’s solution is known to be jus-
tified for spatial problems of theory of elasticity in case of relatively small depth of
structures in the soil base. The Kelvin’s solution is advisable to be applied at large
distances from the base surface. The Plevako equations are useful for geotechnical
calculations in the case when a deep foundation structure is located near the bound-
ary of two layers of a thick base, the mechanical characteristics of these layers being
different.

1.4 Elastic Layered Bases

1.4.1 Constant-Width Elastic Layer

An infinite elastic medium, restricted by parallel plane, is called a spatial elastic
layer [92]. Denote the layer thickness as H and introduce a Cartesian coordinate
system with an origin on the upper boundary of the layer x3 = 0 (Fig. 1.5). If an infi-
nite elastic layer is resting on an incompressible half-space, such calculation scheme
[email protected]
corresponds to a popular model of a finite-thickness elastic base, substantiated by
many authors and included into the engineering regulations for the determination
of the stress-strained state in soil bases [144]. This base model is applicable in the
1.4 Elastic Layered Bases 13

Fig. 1.5 Concentrated force,

acting on the free surface of
an elastic layer with an
absolutely rigid base as an
underlayer

cases of rocks or low-compressible soils lying at a certain depth. If there is no rigid

underlying half-space, then a nominal compressible layer is introduced into consid-
eration, its thickness being set from long-term observations of settlements of large-
scale foundations [143]. It is noted in a number of papers that a constant-thickness
elastic layer as a calculation model for a soil base smoothes some disadvantages,
inherent to the elastic half-space. When the elastic layer model is used, the settle-
ments are smaller and their decay with the distance from the load application point
is faster than for the elastic half-space. In such model the calculated forces in the
structures, resting on elastic bases, are reduced. It should be noted that typically
the calculations of the soil bases themselves and foundation structures, interacting
with them, are reduced to the determination of normal displacements of the base
surface points as well as an a priori unknown distribution of contact pressures over
the areas of the structure contact with the base. In general, such problems result
in integral or integro-differential equations. A very important characteristic, intro-
duced into these equations, is the fundamental solution for the problem of action of a
normal concentrated force on the layer boundary. In particular, for the deformation-
and-strength calculation of shallow foundation structures (typically rigid or flexible
plates and beams) it is sufficient to know only the dependence of the base surface
vertical displacements on the normal load or, in other words, to know the calculation
model (in our case – the contact model) of the base. The equations for the normal
displacements of the layer surface points x3 = 0 due to a unit normal force applied
in an arbitrary point of this surface are given below in the explicit form. Suitable for
practical application contact models of nonhomogeneous bases whose properties
vary with depth, are considered.
Elastic layer, nonhomogeneous with depth. Consider an isotropic layer S = { |x1 ,
x2 | < ∞, 0 ≤ x ≤ H } with a one-dimensional nonhomogeneity of elastic properties
with depth. Let the shear modulus G of the layer material be constant and the
Poisson ratio ν(x3 ) be an arbitrary differentiable function of coordinate along the
layer thickness. In this case the modulus of elasticity of the layer material E(x3 ) =
2G[1+ν(x3 )] will be a positive function of the coordinate x3 .
The problem of action of a unit concentrated force, normal to a layer with no
tangential stress (σα3 (x1 , x2 , 0) = 0), is solved by Borodachev [25] in the assumption
of absence of friction forces at the contact of a layer with an absolutely rigid base
[email protected]
(U3 (x1 , x2 , H) = 0, σα3 (x1 , x2 , 0) = 0, α = 1, 2). The influence function for the
base model under consideration with sufficiently general nonuniformity of elastic
properties can be presented in the form
14 1 Spatial Contact Models of Elastic Bases

∞
1
W (3)
(K, N) = W(x,y,ξ ,η) = W(x − ξ , y − η) = W(r) = Q(k)J0 (rk)dk
4π G
0
 (1.29)
where r = (x1 − ξ )2 + (y − η)2 ; Jo (t) is the Bessel function of the first kind of
the zeroth order;

4 sinh2 (Hk)
Q(k) = ;
k(2L + E22 + E12 L2 )

H
E1 (k) = exp ( − Hk); E2 (k) = exp (Hk); L = γ (s)ds;
0

H H
L1 (k) = γ (s) exp ( − 2ks)ds; L2 = γ (s) exp (2ks)ds;
0 0

γ (s) = [1 − ν(s)]−1 .
The construction of the influence function (1.29) for such character of nonuni-
formity of elastic properties with the layer thickness was enabled by introduction
of stress functions, satisfying second-order partial differential equations with con-
stant coefficients and subsequent application of theory of two-dimensional integral
Fourier transformation [141]. As an example, Borodachev has analyzed in detail the
law of variation of elastic properties of the layer material, when

x3
γ (x3 ) = α1 + α2 cos (nπ ), (1.30)
H
n = 1, 2, 3,. . ., α 1 and α 2 are the model parameters. Figure 1.6 shows the examples
of Poisson ratio dependences in accordance with Eq. (1.30). In case of variation of
deformation properties of the elastic layer with depth according to Eq. (1.30), the
influence function is given by

W(r) = W0 (r)[1 − ψ(R)] (1.31)

where
   
∞ k ∞ k
ψ (R) = T1 (R) − αT2 (R) ; T1 = X J0 (k) dk; T2 = Y J0 (k) dk;
0 R 0 R
X (k) = 1 + [1 − cosh (k)] [k + sinh (k)]−1 ;
[cosh (k) − 1] {k + [1 − θ (k)] sinh (k)}
Y (k) = ;
[k + sinh (k)] {k + [1 + aθ (k)] sinh (k)}
 −1
[email protected]
θ (k) = k2 k2 + n2 π 2 ; a = a2 /a1 ; R = r/2H;

1 − ν(0) 1
W0 (r) = · is the influence function (the Boussinesq problem solu-
2π G r
tion) for a homogeneous elastic half-space with a shear modulus G and Poisson ratio
1.4 Elastic Layered Bases 15

Fig. 1.6 Variation of the Poisson ratio across the finite-thickness nonhomogeneous elastic layer;
ν(0) = 0.3

ν = ν(0). In a particular case, when in Eq. (1.31) H→∞ (in this case the elastic non-
homogeneous layer 0 ≤ x ≤ H is transformed into a half-space x3 ≥ 0), the influence
function for an elastic half-space with a variable Poisson ratio [24, 43] is obtained:

∞ ∞
∞ 1 J0 (rk)dk
W (r) = , (k) = γ (s) exp ( − 2ks)ds.
4π G (k)k
0 0
Elastic homogeneous layer. In the case of a homogeneous layer material, when
ν (x3 ) = ν = const, the function Q(k) from the integral representation of Eq. (1.29)
is given by

4(1 − ν) sinh2 (Hk)

Qo (k) = , (1.32)
2Hk + sinh (2Hk)

and a well-known [37] formula for the vertical displacements of the elastic layer
surface is valid:

∞
2(1 − ν 2 ) sinh2 (Hk)
W H (r) = J0 (rk)dk. (1.33)
πE 2Hk + sinh (2Hk)
0

One should note that Eq. (1.33), similarly to Eq. (1.31), corresponds to the condi-
[email protected]
tion of zero tangential stress on the lower surface of the layer, i.e. when there are no
obstacles for horizontal displacements. According to the Egorov solution [37], the
function Q(k) for a layer, rigidly restrained (U = V= W = 0) on the lower surface
x3 = H, is given by
16 1 Spatial Contact Models of Elastic Bases

[(3 − 4ν) sinh 2α − 2α](1 − ν)

Q1 (k) = , α = kH, (1.34)
(3 − 4ν) cosh2 α + α 2 + (1 − 2ν)2

and the corresponding fundamental solution is written in the form

∞
(1 − ν 2 ) (3 − 4ν) sinh 2α − 2α
W (r) =
E
J0 (rk)dk. (1.35)
2π E (3 − 4ν) cosh2 α + α 2 + (1 − 2ν)2
0

A check for the asymptotic behaviour of the integrals in Eqs. (1.33) and (1.35) at
H→∞, using a known equality

∞
1
J0 (rα)dα = ,
r
0
results, as one should expect, to the classical Boussinesq solution W0 (r) = (1 −
ν 2 )/π Er. In other words, with the layer thickness increase the type of boundary
conditions on its lower boundary will not affect the settlement values of the free
boundary.
It is noted in [37] that the improper integrals in Eqs. (1.33) and (1.35) are not
expressed in a finite form in terms of elementary functions. On the other hand, direct
numerical integration (e.g., using the Simpson rule) is noted to be ineffective in this
case. Following [18], Egorov suggested the following approximation, applicable for
practical purposes:

2 sinh2 α  4
= Bi e−Ai α (1.36)
2α + sinh 2α
i=0

where A0 = 0, A1 = 0.8, A2 = 1.4, A3 = 0, A4 = 0.6, B0 = 1, B1 = 0.426,

B2 = −6.051, B3 = 7.395, B4 = −2.770.
As a result, the improper integral (1.33) can be expressed in terms of elementary
functions by means of the equation [49]

∞
e−α a J0 (bα)dα = (a2 + b2 )−1/2 .
0
Thus, the fundamental solution for an elastic layer with a slippage on the lower
boundary is presented in the following explicit algebraic form:

[email protected]
∞
1 − ν2  1 − ν2 
4 4
Bi
W1 (r) = Bi e−Ai kH J0 (rk)dk =  . (1.37)
πE πE A 2 · H 2 + r 2
i=0 0 i=0 i
1.4 Elastic Layered Bases 17

One should note that, contrary to Eq. (1.33), the solution (1.35) requires approx-
imation of the integrand function for each given value of the Poisson ratio ν. There-
fore, for the value of ν = 0.3, the most widely used in soil mechanics, Egorov
[37] proposed a rather successful approximation of the fundamental solution (1.35),
corresponding to the condition of rigid constraint of the elastic layer on the lower
boundary x3 = H

25  r 2
1 − ν2 −
W2 (r) = W1 − · 0.2086 · e 22 H (1.38)
π HE
It is quite evident that in case of necessity similar approximations of the displace-
ment function W2 (r) can be constructed for any other value of the ν coefficient.
In the studies where the elastic base model is used in the form of a finite-thickness
layer with constrained lower boundary (e.g. [34, 110]), the calculations are per-
formed, based on the Burmister solution [31]. It can be easily shown that this solu-
tion exactly coincides with the corresponding Egorov solution (1.35). However, it
should be noted that the Burmister solution was obtained 2 years earlier and has an
advantage, additionally giving the distribution of vertical compressive stress in the
layer.
Finally note that the influence functions from the action of concentrated forces
within the finite-thickness layer are obtained below (Sect. 1.4.3) as a particular case
of solution of the spatial problem for a to-layer base.

1.4.2 Variable-Thickness Elastic Layer

A spatial elastic wedge is used as the model of a base with a linearly varied depth
(variable-thickness elastic layer). The necessity of taking into account the com-
pressed layer thickness variation in the soil base calculation scheme often arises in
practice when the underlying dense layer (e.g. rock) is inclined. Related issues are
also problems of calculation of foundation structures, located near slopes (the base
model is also an elastic spatial wedge). The wedge angle for such bases is obtuse (or
at least right) and both faces outside the contact domain are stress-free. Both vari-
ants of the calculation scheme of the base as a variable-thickness layer have been
studied rather incompletely. The solutions for most of the contact problems for an
elastic wedge have been obtained, as a rule, in a flat formulation [7, 30, 41, 73, 116,
122, 130, 131, 133, 153].
In cylindrical coordinates r, φ, z where the z axis is directed along the wedge rib,
consider normal P and directed along the r axis tangential Q concentrated forces,
acting on the φ = α face of the spatial elastic wedge in the points r = x, z = ± y
(Fig. 1.7a). The boundary conditions on this wedge face can be written in the fol-
[email protected]
lowing form:

σϕ = Pδ(r − x)δ(|z| − y), τrϕ = −Qδ(r − x)δ(|z| − y), τϕ z = 0 (1.39 )

18 1 Spatial Contact Models of Elastic Bases

(a) (b)

Fig. 1.7 Coordinate systems for a variable-thickness elastic layer: (a) polar, (b) Cartesian

On the face ν = 0 one of the following conditions is supposed to be fulfilled:

(a) σϕ = τrϕ = τϕ z = 0 (zero stress),

(b) v = τrϕ = τϕ z = 0 (sliding fixation), (1.39 )
(c) u = v = w = 0 (rigid fixation).

In [69] exact Green’s formulae are derived (in the form of Neumann series over
the powers of (1–2 ν) where ν is the Poisson ratio) for displacements and stresses
in a variable-thickness elastic layer. A method for solving the first boundary prob-
lem for the spatial wedge is used, consisting in its reduction to the Gilbert bound-
ary problem, generalized according to Vekua [94, 148]. Using the known formulae
[148], the stresses and displacements were expressed in terms of three harmonic
functions, determined in the form of Fourier and Kontorovich-Lebedev integrals in
the complex plane. A transition is made from the boundary conditions (1.39) to
the Gilbert boundary problem, generalized according to Vekua [94], in which the
functional equations with shift are reduced to the second-order Fredholm integral
equations with respect to auxiliary functions n (μ), n = 1, 2, 3, 4, being contained
later in the expressions for the stresses and displacements in the wedge (0 < μ < ∞;
transformation parameter βx > 0)

∞ Qfn(μ,α)
φ (μ) − (1 − 2ν) Ln (μ,y) φn (y) dy = P + ×
0 2 (1 − ν) (1 − 2ν)
πμ Q π μ ∞
× cosh K (β
[email protected]
iμ x) − cosh hn (t,α) Kit (β x) × (1.40)
2 2 (1 − ν) 2 0
sinh (π t) dt
×
cosh (π t) − cosh (π μ)
1.4 Elastic Layered Bases 19

∞
where Kit = exp [ − x cosh (t)] cos (τ t)dt is the modified second-order Bessel
0
function (Macdonald function);

π μ π y ∞
Ln (μ,y) = 2 cosh sinh Wn (y,α) ×
2 2 0
sin π t gn (t,α) dt
×
(cosh π t + cosh π μ) (cosh π t + cosh π y)
cosh αμ ∓ cos α
(a) W1,2 (μ,α) = ± g1,2 (μ,a) =
cosh αμ ± μ sin α
⎧ ⎫
⎨ coth αμ/2 ⎬ sin2 α
= ,
⎩ tanh αμ/2 ⎭ cosh αμ ∓ cos 2α

μ 2 (1 − ν) (1 − 2ν) sin α
f1,2 (μ,α) = ±
W1,2 (μ,α) cosh αμ ∓ cos α

(1 − 2ν) sinh αμ ∓ μ sin α

h1,2 (μ,α) =
cosh αμ ∓ cos α

(b) W3 (μ,α) = W1 (μ,2α),g3 (μ,α) = g1 (μ,2α),f3 (μ,α) = f1 (μ,2α),

h3 (μ,α) = h1 (μ,2α);

2 k sinh 2αμ − 2μ sin 2α

(c) W4 (μ,α) = , k = 3 − 4ν,
2 k cosh 2αμ + 2μ2 − 2μ2 cos 2α + k2 + 1
  
g4 (μ,α) = −g2 (μ,2α) + sin2 a g5 (μ) 2g6 (μ) − μg1 (μ) +

+g8 (μ) 2g1 (μ) − μg6 (μ) / cosh αμ − 2 (1 − ν) sin α g5 (μ) ×
× (sin 3α − sin 2α cosh 2αμ) − g8 (μ) cos 2α sinh 2αμ)} /g9 (μ)
g5 (μ) = k sinh 2αμ cos 2α − μ sin 2α,
g6 (μ) = cosh 2αμ cos 2α − cosh 3αμ − cosh αμ cos 4α,
g7 (μ) = sinh 2αμ sin 2α + sinh αμ sin 4α, g8 (μ) = (k cosh 2αμ − 1) sin 2α,
 
g9 (μ) = g25 (μ) + g28 (μ) sinh2 2αμ + cos2 2α
μ 4 (1 − ν) (1 − 2ν) μ sin2 α
f4 (μ,α) = +
W4 (μ,α) k sin 2αμ − μ sin 2α
[email protected]
k (1 − 2ν) (cosh 2αμ − 1) − 2μ2 sin2 α
h4 (μ,α) =
k sinh 2αμ − μ sin 2α
(1.41)
20 1 Spatial Contact Models of Elastic Bases

It should be noted that the kernel of the integral equation (1.40) depends on the
boundary conditions at φ = 0, and the right-hand part depends as well on the type
of the load on the other face of the wedge.
In [94, 148] Eq. (1.40) was obtained for the problem (b) and a normal load,
symmetrical with respect to z, applied to the face φ = α; it is proven that if this load
is such that the right-hand side of Eq. (1.40) belongs to the space L2 (0,∞), then the
method of successive approximations is applicable for solving such an equation.
At Q = 0, P=0 the right-hand part of Eq. (1.40) does not belong to L2 (0,∞),
but lies within the space of continuous and restricted on the semiaxis functions
CM (0,∞). In [69] it is shown that in the case of the problem (b), the corresponding
integral operator in the left-hand part of Eq. (1.40) for all α∈[0, 2π ] is the operator
of compression in the space CM (0,∞) at least at ν > 0.053. For the problem (c) the
calculations have shown that at α = π k/12, k = 1, 2,. . .,24, ν = 0.25, 0.30; 0.35
the solution of the integral equation (1.40) can be also constructed by the method of
successive approximations. Besides, a detailed analysis, based on numerical integra-
tion, has shown that for the problem (a), e.g. for the case of α = π /2 the solution of
Eq. (1.40) at n=2 can be presented as a Neumann series, if the condition ν > 0.116
is fulfilled; and in the case α = 3π /4 already for any ν∈[0, 1/2]. At α→0 the method
of successive approximations is applicable at ν>0.092. For a fixed angle α∈[0, 2π]
in the problems (a) and (c) the corresponding Neumann series, as a rule, converge
practically at any practically significant value of the Poisson ratio.
Using the solution of the integral equation (1.40) in the form of the Neumann
series and denoting the right-hand part of Eq. (1.40) as Gn (μ, βx), one can present
the expression for the normal displacements on the wedge face φ = α (z ≥ 0) as
follows:

∞ ∞
4 1−ν πμ
v= 3 · sinh W(μ,α,β x)Kiμ (βr) cos βy cos βzdβdμ; (1.42)
π G 2
0 0
μ μ
(a) W(μ,α,β x) = W1 (μ,α)B1 {G1 (y,βx)} − W2 (μ,α)B2 {G2 (y,βx)};

μ
(b) (n = 3), (c)(n = 4) W(μ,α,β x) = 2Wn (μ,α)Bn {Gn (y,βx)};
μ '
∞
μ
Bn = (1 − 2ν)m (An )m ,
m=0

μ ∞
An {Gn (y,β x)} = Ln (μ,y)Gn (y,β x)dy.
0
The explicit form of the kernel of the integral equation obtained on the base
of Eq. (1.42), enabled different analytical methods to be applied for its solution
[41]. It is the uniform convergence of functional series over the powers of (1–2ν) in
[email protected]
CM (0,∞) that is important to be used in Eq. (1.42). In [70] a spatial contact problem
of a punch in the shape of an elliptical paraboloid being indented into a face of
an elastic wedge (without the account of friction). For solving the corresponding
1.4 Elastic Layered Bases 21

integral equation the asymptotic method of “high λ” is applied [153], enabling at

α = π /2 the contact domain and the punch settlement to be determined (however,
with a limited accuracy). Note that here the dimensionless parameter λ characterizes
the relative distance of the contact domain from the wedge rib.
For the case (a) at α = π one can obtain the following expressions from Eq.
(1.42) after integration [69]:

   
P(1 − ν) 1 1 Q(1 − 2ν)(r − x) 1 1
v= + + + 2 ,
2π G R+ R− 4π G R2+ R−

R± = (r − x)2 + (z ± y)2 ,
what, with the account of parity of the problem with respect to z, coincides with the
known solutions of Boussinesq and Cerruti problems for a half-space.
If the right-hand side of Eq. (1.42) for the displacements is multiplied by cosz,
integrated over z from 0 to ∞, and then a limiting transition at →0 is performed,
then, according to [69], one arrives at the corresponding equations of the flat prob-
lem for a wedge, given in [7] for a normal load.
Note that the possibility of obtaining a solution of a problem of action of a con-
centrated force in a spatial wedge at ν=1/2 in the form of an expansion into series
over the powers of (1–2 ν) was first formulated by Uflyand [147]. The problems
for concentrated forces in a wedge are known to be solved exactly at ν = 1/2 (the
deformation is not accompanied by the volume variation), when fragmentation of
boundary conditions occurs to determine three harmonic functions in the Papkovic-
Neuber representation [5, 51, 147, 156]. In [147], while the first boundary-value
problem is being solved using the integral transformation method, it is partitioned
into a symmetrical one and a skew-symmetrical one with respect to the angular
coordinate, and the quadratures, being contained in the explicit solution, are then
obtained using residues. In [156] the Kantorovich-Lebedev method of integral trans-
formations is extended to the case of an elastic incompressible wedge with a right
dihedral angle (an elastic quarter-space). Using a limiting transition, the problem
of a concentrated force on the elastic wedge rib is solved, and then the solution is
extended to the case of any dihedral angle value. In a recent paper [51] a dual inte-
gral transformation is applied – the Fourier transformation along the wedge rib and
the Kantorovich-Lebedev transformation along the radial coordinate in the section
plane perpendicular to the wedge rib. Problems for a wedge, loaded by a concen-
trated internal force or a force, concentrated in a point at the wedge rib, are solved in
images. For a particular case, when the force acts within the wedge cross-section and
is directed normally to the radius-vector of its application point, the image reversals
are reduced to one-dimensional integrals. Note that at ν = 1/2 Eqs. (1.42) give the
Green’s functions for an incompressible wedge from [5, 6]. At ν = 1/2 approximate
solutions of the first boundary-value problem for a spatial wedge were constructed
[email protected]
only for a quarter-space either asymptotically, or numerically [27, 28, 52, 55, 61].
Equation (1.42), though with difficulties, can be used for solving spatial con-
tact problems of theory of elasticity for a wedge (including the account of friction),
22 1 Spatial Contact Models of Elastic Bases

arising at deformation-and-strength calculations of foundation structures, resting on

non-uniformly compressible bases. However, in spite of all advantages of the fun-
damental solution of Eqs. (1.42) for a spatial wedge, described in [6, 69], its prac-
tical application cannot be admitted without reservation as convenient and efficient.
The main difficulty here consists in the necessity of a preliminary solution of inte-
gral equations using expansions into functional series whose convergence at certain
parameters of the problem can be rather slow. In particular, the authors of [69] have
drawn attention to a considerable slowdown of convergence of the Neumann series
(1.42) for α = π/2. The developed method is rather cumbersome and complicated,
it requires high mathematical culture of the calculators, and there is no guarantee of
the solution convergence at some combinations of the parameters α and ν. Besides,
one should add that the fundamental solution of Eq. (1.42) has been obtained only
for a normal load, symmetrical with respect to z. As follows from the analysis of the
spatial contact problem [70], using of the fundamental solution, proposed in [69],
can be finally accomplished only on the base of asymptotic methods (e.g., the “high
λ method”), which are efficient at sufficient distance of the contact domain from the
wedge rib, i.e. the solutions being obtained are mostly of theoretical importance. We
can hardly believe that the expression of Eq. (1.42) can be directly applied for prac-
tical solution of spatial problems with contact domains of complicated geometry. A
more simple and convenient from the practical point of view is the use of an approx-
imate solution of the mixed spatial problem of a vertical concentrated force acting
on a horizontal face of an elastic wedge proposed by Fedorovskiy and Onopa [39]
(Fig. 1.7b). While constructing the solution, the authors have invoked the “fictitious
force” method (or the source method) in combination with the collocation method
[105]. Displacements in the spatial wedge are determined in the form of superpo-
sition of two solutions for a half-space with a stress-free boundary: the Boussinesq
solution and a solution, corresponding to the location of fictitious forces in an imag-
inary point, symmetrical to the point of application of the real concentrated force
with respect to the restrained face of the elastic layer. The second solution is built
on the base of the choice of intensities of the fictitious forces (the sources) to provide
automatic fulfillment of boundary conditions on the wedge faces.
According to [39], the influence function, corresponding to the action of a unit
concentrated load normally to the elastic wedge free surface, is given by

 
1 
4
1−ν 1 ak
W (3)
(K, N) = ω(x,y,ξ ,η) = −  , (1.43)
2π G R R1
k=1 (R/R1 )2 + (kb)2
 
R= (x − ξ )2 + (y − η)2 , R1 = (x + ξ )2 + (y − η)2

where b, ak , k = 1,4 are coefficients, depending on α, ν, ξ , η are the coordinates

[email protected]
of the point of application of the concentrated force, x, y are the coordinates of the
field (the observation point), α is the angle of inclination of a rigid underlying soil,
ν is the Poisson ratio, G = E/2(1+ν) is the shear modulus.
1.4 Elastic Layered Bases 23

The coefficients b, ak , k = 1,4 in Eq. (1.43) for various values of the wedge
angle α and the Poisson ratio ν are proposed to be found by the least-square method
[39]. The values of the coefficients b, ak , k=1,4, used in our calculations, are listed
in Table 1.1. References [39, 115] give more detailed tables of these coefficients
for the values 5◦ ≤ α ≤ 89◦ and 0 ≤ ν ≤ 0.5. The tables do not contain the angle
α = 90◦ , since in this case the equation system is degenerate. As in the case of
the constant-thickness layer, Eq. (1.43) contains only four additive terms to the
Boussinesq solution. Using of simple algebraic functions in Eq. (1.43) results in a
considerable simplification of the calculation methods and shortage of the calcula-
tion scope for the studies of the contact interaction processes.
As an example, Fig. 1.8 shows isolines of dimensionless settlements W/Wo
(Wo = P(1–ν)d/2π G is the characteristic displacement measure) of the free sur-
face of the elastic wedge (a = 30◦ ) and a half-space with the same deformation
characteristics (ν = 0.25) under a vertical concentrated force P in the point K with
coordinates ξ = d, η = 0. Due to the presence of a rigid underlying soil base, the
settlements of the surface of the variable-thickness layer will always be smaller than
for the corresponding values for the half-space. As seen from the comparison of the
calculation data (Fig. 1.8a, b), besides the quantitative differences, the fundamental
solution of Eq. (1.43) differs from the classical Boussinesq solution (ak = 0, k = 1,4)
in a number of qualitative features. First, the domain of definition of the influence
function (1.43) is not the plane, but the half-plane x ≥ 0. Second, the fundamental
solution of Eq. (1.43), having the following structure:

W (3) (K, N) = ω(x − ξ ,x + ξ ,y − η) = ω(x,ξ ,y − η)

is not a difference solution with respect to the variables x, ξ (i.e. in the direction of
the layered base depth increase), and, consequently, is nonlinear and anisotropic.
In a small vicinity of the point K(ξ , η, ζ ) of the concentrated force application, the
influence functions for the half-space and the variable-thickness layer behave quali-
tatively similar, this being revealed in an unlimited growth of settlements according
to hyperbolic law. However, with the increase of the distance form the point K(ξ , η)
the difference of the two solutions becomes more essential. As is clearly seen from
the settlement isolines, plotted in Fig. 1.8b, the solution of Eq. (1.43), due to its
approximate character, has a noticeable defect, resulting in the formation of an area
of negative (opposite to the force direction) displacements, arising between the elas-
tic wedge rib and the concentrated force application point. Note that the authors of
[39] did not discuss this defect (probably, they have not found it). The discussed
character of deformation of the horizontal surface of the variable-thickness elastic
wedge is quantitatively rather small and diminishes to zero rather soon with the
increase of the distance between the concentrated force application point and the
wedge rib. It is quite clear that in case of high accuracy of solution of contact prob-
[email protected]
lems in the direct vicinity of the wedge rib being required, the situation can be
improved only by increase of the number of additional terms in the approximate
representation of Eq. (1.43).
24 1 Spatial Contact Models of Elastic Bases

Table 1.1 Values of ak and b coefficients at various ν and α

α ν a1 a2 a3 a4 B

5◦ 0 −0.1859 −4.179 −4.415 −1.410 0.0194

0.25 −0.2084 −0.3373 3.587 −2.765 0.0194
0.35 −0.2371 −2.584 9.778 −6.037 0.0194
0.50 −0.2084 −10.09 26.03 −14.75 0.0194
15◦ 0 −0.1013 −1.751 5.778 −3.129 0.0405
0.25 −0.0877 −1.77 6.412 −3.755 0.0405
0.35 −0.0889 −1.966 7.389 −4.553 0.0405
0.50 −0.1008 −2.589 10.34 −6.928 0.0405
20◦ 0 0.0211 −0.7522 4.894 −3.208 0.0697
0.25 −0.0326 −0.1323 4.086 −2.984 0.0697
0.35 −0.0645 0.2019 3.911 −3.127 0.0697
0.50 0.1405 0.9806 3.894 −3.849 0.0697
25◦ 0 −0.020 −0.1491 4.187 −3.096 0.0979
0.25 −0.0799 0.6939 2.76 −2.458 0.0979
0.35 −0.1181 1.212 2.120 −2.312 0.0979
0.50 −0.2112 2.505 0.8496 −2.279 0.0979
30◦ 0 −0.1261 3.619 −2.193 0.6272 0.175
0.25 −0.0916 1.019 2.277 −2.313 0.125
0.35 0.1324 1.636 1.368 −1.992 0.125
0.50 0.2345 3.220 −0.6396 −1.504 0.125
40◦ 0 −0.0918 2.183 0.7424 −2.08 0.2249
0.25 −0.0684 0.7601 3.108 −2.977 0.1749
0.35 −0.1063 1.394 2.106 −2.594 0.1749
0.50 −0.2040 3.050 −0.1037 −2.015 0.1749
45◦ 0 −0.1033 3.971 −3.079 −0.0604 0.2974
0.25 −0.1224 2.935 −1.224 −0.7867 0.2474
0.35 −0.0803 0.9443 3.195 −3.308 0.1974
0.50 −0.1687 2.445 1.412 −3.053 0.1974
60◦ 0 0.0555 −1.580 8.086 −5.799 0.2531
0.25 −0.0596 1.341 4.130 −5.023 0.3531
0.35 −0.010 2.585 0.9658 −2.980 0.3531
0.50 −0.1309 2.378 2.644 −4.627 0.3031
75◦ 0 0.0596 −1.527 6.112 −3.550 0.2880
0.25 0.0416 −1.179 5.223 −3.012 0.2880
0.35 0.0369 −1.166 5.702 −3.607 0.2880
0.5 0.0335 −1.432 8.570 −6.614 0.2880
80◦ 0 0.0509 −1.253 4.590 −2.123 0.2946
0.25 0.0338 −0.9102 3.591 −1.457 0.2946
0.35 0.0299 −0.909 4.009 −1.966 0.2946
0.50 −0.0325 −1.287 6.968 −4.941 0.2946
89◦ 0 0.0215 −0.4724 1.099 0.9120 0.2999
0.25 0.0206 −0.4447 1.020 0.8807 0.2999
0.35 0.0237 −0.5434 1.548 0.3678 0.2999
0.5 0.0375 −0.9932 3.905 −1.785 0.2999

[email protected]
1.4 Elastic Layered Bases 25

(a) (b)

Fig. 1.8 Isolines of dimensionless settlements of the surface of (a) an elastic half-space, (b) of
variable-thickness elastic layer (α = 30◦ ) under a normal concentrated force, ν = 0.25

The experience of numerous calculations of spatial problems of contact inter-

action has shown that, in spite of its approximate character, the influence function
(1.43) enables the non-uniform compressibility of the base to be taken into account
with sufficient accuracy, what is important for rational design of foundations at com-
plicated engineering-and-geological conditions.

1.4.3 Multilayer Elastic Half-Space

Models of elastic layered media with plane-parallel layers rather adequately reflect
mechanical properties of many objects, widely spread in practice. In the field of
geotechnics these are first of all road and airfield surfaces, soil bases with natural
layering. The effect of the base lamination should be estimated while designing such
important objects as high-rise buildings, funnels, barrages, dams, mines.
The main advantage of the models of multilayer media is the possibility of a
stepwise variation of the soil base mechanical properties at the transition from one
layer to another to be described. A number of studies are devoted to the issues of
theory and calculation of multilayer bases. Among the most significant studies in
this field of mechanics of nonhomogeneous media one should mention Refs. [20,
58, 59, 82, 89, 90, 91, 106, 137, 152, 110]. The overwhelming majority of studies
regarding the problems of static of elastic multilayer bases deal with the issues of the
medium deformation by surface loads only. The most convenient methods of solu-
tion of these problems of theory of elasticity for multilayer bases are such methods
when the conditions of joint deformation of layers finally result in the problem of
finding auxiliary functions, determining the stress-strained state of the upper layer.
[email protected]
The parameters of the stress-strained state for other layers are found from special
recurrent relations. It is quite important that the order of the resolving system of
linear algebraic equations does not depend on the number of layers in the base.
26 1 Spatial Contact Models of Elastic Bases

In spite of a considerable success in developing methods for solving the problems

of theory of elasticity for multilayer bases under a surface load, the number of pub-
lications devoted to the case of loading inside the layers or at interlayer boundaries,
is quite limited [32, 33, 38, 129, 132, 134]. The investigation of the stress-strained
state of multilayer media due to internal forces is the basis for working out a gen-
eral approach to solving the boundary problems of theory of elasticity for bases of
complex structure, containing cavities, inclusions, various defects and gaps between
the layers. In geotechnics, the field of application of such studies is rather broad:
essentially deepened foundation structures, wells, mine working, etc. The classi-
cal method of superposition of singularities, once having been applied by Mindlin
for a homogeneous isotropic half-space, is inapplicable for more complicated
systems, which are layered media. As a rule, in such cases integral transformations
are required.
Pioneer studies by Shekhter and Prikhodchenko [129, 132] present a general
method for the exact solution of the problem of the distribution of stresses and dis-
placements in an elastic layer under a vertical force and a horizontal force acting
inside it, at the conditions of a flat problem of theory of elasticity, and under a ver-
tical force in the case of an axisymmetric spatial problem. Later the same authors
have also solved the spatial problem for a layer under a horizontal force acting inside
it [134]. The essence of the method used consists in the following: on the stress-
strained state due to the action of a concentrated force within the space (the Kelvin’s
solution) a continuous trained solution is superimposed, which is chosen in such a
way that the boundary conditions at the upper and lower faces of the layer be satis-
fied. However, the proposed method has not been applied for multilayer bases (with
the account of satisfaction of the conjugation conditions at the boundaries of each
layer). Besides, no numerical results have been obtained even for a one-layer sys-
tem, since the representation of the displacement vector components in terms of the
Galerkin vector contains unlimited functions. This is an essential inconvenience for
the numerical application of the methods, since inadmissibly large numbers appear
at high and low values of the Hankel transformation parameters. Later, an axisym-
metric problem of action of a concentrated vertical force inside an elastic half-space,
consisting of a finite number of plane-parallel layers, was considered by Yegorov,
Barvashov, and Fedorovskiy [38]. These authors seek for the solution in each layer,
using the biharmonic Love function, presented using the Hankel transformation.
The stress function is partitioned into two terms: the stress function, corresponding
to the Kelvin’s solution for the action of a concentrated force in the space, and a
stress function, determined from the layer contact conditions and satisfying the zero
stress conditions for the free surface. An approximate solution of the resolving sys-
tem of linear algebraic equations is found in the form of a matrix power series, each
term of which induces a correction due to the decrease of the approximation dis-
crepancies by “reflection” from the layer boundaries. By means of a transition from
the Hankel transformations to the Fourier transformation, the approach proposed
[email protected]
in [38] can be extended to the spatial non-axisymmetric problem. As one should
expect, the calculations have shown the satisfactory accuracy of approximation to
be obtained only for the points close to the point of application of the concentrated
load. However, besides the same difficulties as for the application of the Shekhter
1.4 Elastic Layered Bases 27

method, a necessity of representation of the approximate solution using the partial

sums of the matrix series, is added. To our knowledge, no solutions of practical
problems regarding calculation of deepened foundation structures using the method
proposed by Yegorov, Barvashov, and Fedorovskiy, are available. Therefore, there
are no data available, enabling the fundamental solutions for layer bases of [38] to
be considered suitable for broad practical application.
A considerable success in solving the spatial problem of the action of concen-
trated forces directed vertically and horizontally within a layered half-space, was
achieved in [32]. The solution for a two-layer system (an elastic layer, resting
on an elastic half-space) was obtained in a closed form as Hankel integrals from
exponential decay functions. A general solution of the Navier’s equations of elastic
equilibrium for non-axisymmetric problems was used in the form of Muki [83] by
means of expansion into Fourier series over the angular coordinate and the Hankel
transformation over the radial coordinate. A special numerical method was devel-
oped for the evaluation of improper integrals, including an exponential approxi-
mation of high accuracy for the reciprocal values of common denominators of the
integrands using the least-square method in the integral formulation. The fundamen-
tal solutions for the two-layer system, presented in [32], were thoroughly tested in
the limiting cases, using the Mindlin’s solution. The proposed closed form of the
approximate solution enables the results to be obtained with high accuracy and can
be extended to another, more complicated multilayer bases. Meanwhile, the effi-
ciency of practical application of the approach is rather limited, since it requires
labour-consuming approximation of the reciprocal values of the integrand function
denominators at each specific combination of the elastic parameters of the layers.
Note, that the fundamental solution for the two-layer system, presented in [32], was
later repeated by Davies and Bannerjee [33] for the forces inside the base, acting
on the contact surface of separation of the layer and the half-space. These authors
presented the fundamental solution in an improved (for the purpose of numerical
integration) form due to the analytical evaluation of slowly decreasing terms in the
integrand expressions.
Below the closed solutions of spatial problems of theory of elasticity for multi-
layer media, obtained on the basis of the two-dimensional Fourier transformation,
are presented. The elaborated formulae can be conveniently used as fundamental
singular solutions for layered media in the framework of the numerical algorithm
of the boundary-element method. For the sake of better visual understanding of
the results, a three-layer system is considered, the two-layer one being considered
as a particular case. The first case is most often met in the geotechnical practice,
when the foundation base contains an internal layer of strongly or weakly com-
pressible soil. Another typical case in subterranean geomechanics is a three-layer
elastically deformable system “elastic layer + elastic oil(gas)-containing layer +
finite-thickness elastic layer (or elastic half-space)”. If two adjacent layers of the
base have the same elastic parameters, then the three-layer system becomes a two-
[email protected]
layer one and one of the interlayer boundaries becomes fictitious. One arrives at the
problem of concentrated forces acting within the upper or the lower layer. In the
two-layer system the assignment of the fictitious boundary enables the problem of
concentrated forces, acting within a finite-thickness layer, to be solved. In case of
28 1 Spatial Contact Models of Elastic Bases

necessity, the formulations proposed for the two- and three-layer systems, can be
without any principal difficulties generalized, in a more cumbersome form, for any
finite number of layers. The most convenient for visualization results are obtained
for the number of layers, not exceeding five, what corresponds to the geotechnical
conditions in practice.
Three-layer base. The calculation scheme for a three-layer deformable system is
presented in Fig. 1.9. Let us consider each layer to be solid, and in case of deforma-
tion the adjacent layers to be deformed together. The component layers have their
own independent thicknesses hk , elastic parameters Ek , ν and are related to a local
orthogonal Cartesian coordinate system Xk , Yk , Zk (k = 1, 2, 3) with the origin on the
upper boundary of the layer. Let us find the displacements, stresses and dilatation
functions due to unit concentrated forces on the upper boundaries of each layer. This
will enable the deformation of the daylight surface as well of the surfaces, bounding
each layer, to be calculated based on the integral representations. It is convenient to
characterize the stressed state at each of the nine concentrated forces (three per each
layer) by a separate component of the matrix
( x (
(T y (
( 1 T1 Q1 (
( x (
( T y (
( 2 T2 Q2 ( (1.44)
( x (
( T T3 Q3 (
y
3

[email protected]

Fig. 1.9 Concentrated forces on the separation surfaces in a three-layer base

1.4 Elastic Layered Bases 29

where Tk x and Tk y are tangential and Qk are normal concentrated forces on the upper
boundary of the k-th layer.
Take the advantage of the Lamé equations of spatial theory of elasticity [71],
performed for each layer
⎧
⎪ 2G(1 − ν) ∂εv
⎪
⎪ + GU = 0,
⎪
⎪ 1 − 2ν ∂x
⎪
⎪
⎨ 2G(1 − ν) ∂ε
v
+ GV = 0, (1.45)
⎪
⎪ 1 − 2ν ∂y
⎪
⎪
⎪
⎪ 2G(1 − ν) ∂εv
⎪
⎩ + GW = 0
1 − 2ν ∂z
where G and ν are the elasticity parameters,

∂U ∂V ∂W ∂2 ∂2 ∂2
εv = + + ;  = 2 + 2 + 2.
∂x ∂y ∂z ∂x ∂y ∂z
Now apply the two-dimensional Fourier transformation

 ∞

f̄ (ξ ,η) = F f = f (x,y)ei(ξ x+ηy) dxdy (direct), (1.46)
−∞
 ∞
1
f (x,y) = f̄ (ξ ,η)e−i(ξ x+ηy) dξ dη (inverse) (1.47)
4π 2
−∞

to Eqs. (1.45). Taking into account that

∂ m+n f
F = ( − iξ )m ( − iη)n f̄ ,
∂xm ∂yn

one obtains instead of Eqs. (1.45) the following system of ordinary differential equa-
tions:
⎧  
⎪ 2G (1 − ν) ∂ W̄ 2
2 Ū + G d Ū = 0,
⎪
⎪ −ξ 2 Ū − ξ η V̄ − iξ − G p
⎪
⎪ 1 − 2ν dz dz2
⎪
⎪
⎪
⎨  
2G (1 − ν) ∂ W̄ d2 V̄
−ξ ηŪ − η2 V̄ + iη − G p2 V̄ + G 2 = 0, (1.48)
⎪
⎪ 1 − 2ν dz dz
⎪
⎪
⎪
⎪  
⎪ 2G (1 − ν) d
⎪ ∂ W̄ d2 W̄
⎩ −iξ Ū − iηV̄ + − G p2 W̄ + G 2 = 0
1 − 2ν dz dz dz
[email protected]
where p2 = ξ 2 + η2 .
If now in the image space a transition is performed to new functions S and T,
related to Ū and V̄ as follows:
30 1 Spatial Contact Models of Elastic Bases

S = iξ Ū + iηV̄, T = iηŪ − iξ V̄, (1.49)

then a new system of differential equations with respect to the functions S, T, and
W̄ is obtained:

⎧    2 
⎪
⎪ 2G( 1 − ν) 2 dW̄ d S
⎪
⎪ p −S +G − p S = 0,
2
⎪
⎪ 1 − 2ν dz dz2
⎪
⎪    2 
⎨
2G( 1 − ν) d2 W̄ dS d W̄
− +G − p W̄ = 0,
2 (1.50)
⎪
⎪ 1 − 2ν dz2 dz dz2
⎪
⎪
⎪
⎪
⎪
⎪ d2 T
⎩ − p2 T = 0.
dz2

Note that as a result of the transition to the system (1.50), the third equation
becomes independent of the first and the second one. The common solution of the
system (1.50) is easily obtained in a standard way and is given by

L  
pW̄ = α (3 − 4ν) sinh pz − pz cosh pz + 2 (1 − ν)
4 (1 − ν) 2

β 2 (1 − ν) cosh pz − pz sinh pz
 
+2 (1 − ν) γ (1 − 2ν) sinh pz − pz cosh pz − δ pz sinh pz ,
L  
S=− α pz sinh pz + 2 (1 − ν) β (1 − 2ν) sinh pz + pz cosh pz
4 (1 − ν)2
 
+2 (1 − ν) γ 2 (1 − ν) cosh pz + pz sinh pz + δ (3 − 4ν) sin pz + pz cosh pz ,

1
T = L γ̄ cosh pz + δ̃ sinh pz
1−ν
(1.51)
where L = 2( 1 − ν 2 )/E and the integration functions α, β, γ , δ, γ̄ , δ̄ of the
parameters ξ and η of the integral Fourier transformation are determined from the
boundary conditions, given on the surfaces, bounding each layer.
Since the boundary conditions are formulated in terms of characteristics of the
stress-strained state, the transforms of the displacements and stress tensor compo-
nents should be expressed in terms of S, T, and W̄ functions.
Application of the direct Fourier transformation to the Hooke’s law

 
∂Ui ∂Uj 2Gν
σij = G
[email protected] + + εv δij
∂xj ∂xi 1 − 2ν
enables the transforms of stresses and dilatation function to be obtained in the fol-
lowing form:
1.4 Elastic Layered Bases 31

2Gν
σ̄x = 2G( − iξ Ū) + ε̄v ,
1 − 2ν
2Gν
σ̄y = 2G( − iηV̄) + ε̄v ,
1 − 2ν
dW̄ 2Gν
σ̄z = 2G + ε̄v ,
dz 1 − 2ν
 
dV̄ (1.52)
τ̄yz = G −iηW̄ + ,
dz
 
dŪ
τ̄xz = G −iξ W̄ + ,
dz
 
τ̄xy = G −iηŪ − iξ V̄ ,
dW̄
ε̄v = −iξ Ū − iηV̄ + .
dz
Using Eq. (1.49) in combination with Eq. (1.52), for each layer one will have
in the Fourier transforms the sought expressions of the stress-strained state compo-
nents, using the S, T, and W̄ functions.

1 1
Ū = − (iξ S + iηT), V̄ = − 2 (iηS − iξ T),
p2 p
2  

ξ ξη ν dW̄
σ̄x = −2G 2 S + 2 T − −S ,
p p 1 − 2ν dz
2  

η ξη ν dW̄
σ̄y = −2G 2 S − 2 T − −S , (1.53)
p p 1 − 2ν dz
 
1 − ν dW̄ ν
σ̄z = 2G − S ,
1 − 2ν dz 1 − 2ν
 
iη dS iξ dT
τ̄yz = −G iηW̄ + 2 − 2 ,
 p dz p dz 
ξ dS η dT
τ̄xz = −G ξ W̄ + 2 + 2 ,
p dz p dz
 
ξ 2 − η2 2ξ η
τ̄xy = G T− 2 S ,
p2 p

dW̄
ε̄v = − S.
dz
[email protected]
Thus, the boundary problem under consideration is reduced to the determina-
tion of functional coefficients α, β, γ , δ, γ̄ , δ̄, for each elastic layer. Using these
coefficients, based on the conditions of continuity of displacements and stresses on
the layer surface, the auxiliary functions S and T become unambiguous as well as
32 1 Spatial Contact Models of Elastic Bases

the transforms of the displacements and stresses and the dilatation function. After
rather cumbersome intermediate transformation, the further required (additionally
to Eq. (1.51)) relations in a detailed notation are given by

1
σ̄z = α cosh pz − pz sinh pz + β (sinh pz − pz cosh pz) −
2 (1 − ν)
(1.54)
1 
−γ pz sinh pz − δ (1 − 2ν) sinh pz + pz cosh pz ,
2 (1 − ν)
1  α 
τ = − iξ̄ τ̄xz + iητ̄yz = pz cosh pz − (1 − 2ν) sinh pz +
p 2 (1 − ν)

(1.55)
pz sinh pz
+βpz sinh pz + γ (sinh pz + pz cosh pz) + δ cosh pz + ,
2 (1 − ν)
1 
τ̃ = − iητ̄xz − iξ̄ τ̄yz = γ̃ (1 − ν) sinh pz + δ̃ cosh pz, (1.56)
p
(1 + ν) (1 − 2ν)  
ε̄ν = α + 2γ (1 − ν) cosh pz + [δ + 2β (1 − ν)] sin pz+
E (1 − ν)

δ
+ pz cosh pz .
1 − 2ν
(1.57)

Now consider the formulation of the boundary conditions, corresponding to the

common deformations in the three-layer system (Fig. 1.9) with the account of the
action of concentrated forces of the form of Eq. (1.44). Let a uniform normal pres-
sure q = const act on the upper boundary (z = 0) of one of the layers, being dis-
tributed over a circle x2 + y2 ≤ a2 :
*
q, x2 + y2 ≤ a2
σz =
0, x2 + y2 > a2
where q = Q/πa2 , Q is the main traction vector.
If now the known transition equation [19]

 ∞ ∞
±i(ξ x+ηy)
f (r) e dxdy = 2π r · f (r)J0 (rρ) dr
−∞ 0
is used, as well as the property of the Bessel function [63]

1
1
[email protected]
x J0 (ax)dx = J1 (a),
a
0
one obtains
1.4 Elastic Layered Bases 33

a
2Q J1 (ap)
σ̄z = 2 r J0 (rp)dr = 2Q
a ap
0

Here Jm (x) i the Bessel function of the first kind of the order m, r = x2 + y2 .
Making, as usual, the load action radius a tend to zero and keeping constant the
main traction vector Q, with the account of the limiting relation

J1 (ρ) 1
lim = ,
ρ →0 ρ 2
finally for a normal concentrated force one obtains

σ̄z = Q.

Similarly, for the action of tangential concentrated forces Tx and Ty , directed

along the x and y axes and applied to the coordinate origin, accordingly, one arrives
at

τ̄xz = T x , τ̄yz = T y . (1.58)

Thus, the boundary conditions, expressing the absence of stress on the free sur-
face of the first layer, the continuity of the U, V, and W displacements, the balance of
normal σ z and tangential τ xz , τ yz stresses on the common boundary of two adjacent
layers as well as the equality of the displacements on the lower boundary of the last
layer to zero

U(x,y,h3 ) = V(x,y,h3 ) = W(x,y,h3 ) = 0,

for the Fourier transforms will take the form:

1st layer (at z = 0):

y
σ̄z = Q1 , τ̄xz = T1x , τ̄yz = T1 ; (1.59)

2nd layer:
⎧ + +
⎨ Ū2 |z=0 = Ū1 +z=h1 S2 |z=0 = S1 +z=h1
+ or +
⎩ V̄2 |z=0 = V̄1 +z=h1 T2 |z=0 = T1 +z=h ,
1

+
W̄2 |z=0 = W̄1 +z=h1 ,
[email protected]
+ +
+
σ̄z(2) +z=0 + Q2 = σ̄z(1) +z=h1 ,
34 1 Spatial Contact Models of Elastic Bases
+
(1) +
(2)
τ̄xz |z=0 + T2x = τ̄xz z=h1 ,

y +
(1) +
(2)
τ̄yz |z=0 + T2 = τ̄yz z=h1 ; (1.60)

3rd layer:
* + +
Ū3 |z=0 = Ū3 +z=h2 S3 |z=0 = S2 +z=h2
+ or +
V̄3 |z=0 = V̄2 +z=h 2 T3 |z=0 = T2 +z=h , 2

+
W̄3 |z=0 = W̄2 +z=h2 ,

+ +
+
σ̄z(3) +z=0 + Q3 = σ̄z(2) +z=h2 ,

+
(2) +
(3)
τ̄xz |z=0 + T3x = τ̄xz z=h2 ,

y +
(2) +
τ̄yz
(3)
|z=0 + T3 = τ̄yz z=h2 ,

* + +
Ū3 +z=h3 = 0 S3 +z=h3 = 0
+ or +
V̄3 +z=h3 = 0 T3 +z=h3 = 0,
+
W̄3 +z=h3 = 0. (1.61)

On the upper boundary of each layer (z = 0), as follows from Eqs. (1.51), (1.53),
((1.54), (1.55), (1.56), (1.57), the following conditions are fulfilled:

α = σ̄z |z=0 ,

E
β= pW̄ |z=0 ,
2( 1 − ν 2 )

E E
γ =− (iξ Ū + iηV̄) |z=0 = − S |z=0 ,
2( 1 − ν )
2 2( 1 − ν 2 )

1
δ = τ |z=0 = − (iξ τ̄xz + iητ̄yz ) |z=0 , (1.62)
p
E E
γ̃ =
[email protected] (iηŪ − iξ V̄) |z=0 = T |z=0 ,
2( 1 − ν 2 ) 2( 1 − ν 2 )

1
δ̃ = τ̃ |z=0 = (iητ̄xz − iξ τ̄yz ) |z=0 .
p
1.4 Elastic Layered Bases 35

Therefore, by using Eqs. (1.51), (1.52), (1.54), (1.55), (1.56), (1.57) as well as
relations

1 1
τ̄yz = (iξ τ̄ + iητ ), τ̄xz = (iξ τ − iητ̄ ) ,
p p
the boundary conditions (1.59)−(1.61) in a detailed notation will be given by

α1 = Q1 , (1.63)

i y
δ1 = − (T1x ξ + T1 η), (1.64)
p
i x y
δ̃1 = (T η − T1 ξ ), (1.65)
p 1
L1  
γ2 = α 1 p 1 sinh p1 + 2(1 − ν 1 )β 1 (1 − 2ν 1 ) sinh p 1 + p1 cosh p1
4L2 (1 − ν1 )2

+2(1 − ν1 )γ1 2(1 − ν1 ) cosh p1 + p1 sinh p1
,
+δ1 (3 − 4ν1 ) sinh p1 + p1 cosh p1 ,
(1.66)

L1 1
γ̃2 = (γ̃1 cosh p1 + δ̃1 sinh p1 ), (1.67)
L2 1 − ν1
L1
β2 = {α1 [(3 − 4ν1 ) sinh p1 − p1 cosh p1 ]+
4L2 (1 − ν1 )2
 (1.68)
+2(1 − ν1 )β1 2(1 − ν1 ) cosh p1 − p1 sinh p1 +

+2 (1 − ν1 ) γ1 (1 − 2ν1 ) sinh p1 − p1 cosh p1 − δ1 p1 sinh p1 } ,
1
α2 + Q2 = α1 [ cosh p1 − p1 sinh p1 ] + β1 ( sinh p1 − p1 cosh p1 )
2(1 − ν1 )
δ1
−γ1 p1 sinh p1 − [(1 − 2ν1 ) sinh p1 + p1 cosh p1 ],
2(1 − ν1 )
(1.69)

i y 1
δ2 = (ξ T2x + ηT2 ) + {α1 [ − (1 − 2ν1 ) sinh p1 + p1 cosh p1 ]+
p 2(1 − ν1 )
(1.70)
+2(1 − ν1 )β1 p1 sinh p1 + 2(1 − ν1 )γ1 [ sinh p1 + p1 cosh p1 ]+
[email protected]
+δ [p sinh p + 2(1 − ν ) cosh p ]},
1 1 1 1 1

i y
δ̃2 = − (ηT2x − ξ T2 ) + (1 − ν1 )γ̃1 sinh p1 + δ̃1 cosh p1 , (1.71)
p
36 1 Spatial Contact Models of Elastic Bases

L2  
γ3 = α 2 p 2 sinh p2 + 2(1 − ν 2 )β2 (1 − 2ν 2 ) sinh p 2 + p2 cosh p2
4L3 (1 − ν2 )2

+2 (1 − ν2 ) γ2 2 (1 − ν2 ) cosh p2 + p2 sinh p2
,
+δ2 (3 − 4ν2 ) sinh p2 + p2 cosh p2 ,
(1.72)
L2 1
γ̃3 = (γ̃2 cosh p2 + δ̃2 sinh p2 ), (1.73)
L3 1 − ν2
L2
β3 = {α2 [(3 − 4ν2 ) sinh p2 − p2 cosh p2 ]
4L3 (1 − ν2 )2
 (1.74)
+2(1 − ν2 )β2 2(1 − ν2 ) cosh p2 − p2 sinh p2

+2 (1 − ν2 ) γ2 (1 − 2ν2 ) sinh p2 − p2 cosh p2 − δ2 p2 sinh p2 } ,
1
α3 + Q3 = α2 [ cosh p2 − p2 sinh p2 ] + β2 ( sinh p2 − p2 cosh p2 )
2(1 − ν2 )
δ2
−γ2 p2 sinh p2 − [(1 − 2ν2 ) sinh p2 + p2 cosh p2 ],
2(1 − ν2 )
(1.75)
i y 1
δ3 = (ξ T3x + ηT3 ) + {α2 [ − (1 − 2ν2 ) sinh p2 + p2 cosh p2 ]+
p 2(1 − ν2 )
(1.76)
+2(1 − ν2 )β2 p2 sinh p2 + 2(1 − ν2 )γ2 [ sinh p2 + p2 cosh p2 ]+
+δ2 [p2 sinh p2 + 2(1 − ν2 ) cosh p2 ]},
i y
δ̃3 = − (ηT3x − ξ T3 ) + (1 − ν2 )γ̃2 sinh p2 + δ̃2 cosh p2 , (1.77)
p
α3 p3 sinh p3 + 2(1 − ν3 )β3 [(1 − 2ν3 ) sinh p3 + p3 cosh p3 ] + 2(1 − ν3 )
γ3 [2(1 − ν3 ) cosh p3 + p3 sinh p3 ] + δ3 [(3 − 4ν3 ) sinh p3 + p3 cosh p3 ] = 0,
(1.78)

1
γ̃3 cosh p3 + δ̃3 sinh p3 = 0, (1.79)
1 − ν3
α3 [(3 − 4ν3 ) sinh p3 − p3 cosh p3 ] + 2(1 − ν3 )β3 [2(1 − ν3 ) cosh p3 − p3 sinh p3 ]+
+2(1 − ν3 )γ3 [(1 − 2ν3 ) sinh p3 − p3 cosh p3 ] − δ3 p3 sinh p3 = 0
(1.80)
where pk = hk p (k = 1, 2, 3).
Thus, for each of the nine concentrated forces under consideration in Eq. (1.44),
a system of 15 linear algebraic equations (1.63)−(1.80) to determine 15 functions
k , δk , δ̃k , γ m , β m , γ̃k (k = 2,3; m = 1, 2, 3) of the parameters ξ , η (α 1 , δ 1 , δ̃1
α[email protected]
being always given on the upper boundary of the 1st layer) is obtained, its solution
determining the transforms of the stress-strained state and the dilatation function for
each layer.
1.4 Elastic Layered Bases 37

Note some features of the system of Eqs. (1.63)−(1.80) whose account is impor-
tant for the numerical implementation of the method proposed. The unknowns γ̃m , δ̃k
(m = 1, 2, 3; k = 2, 3) are included only in Eqs. (1.67), (1.71), (1.73), (1.77), (1.79)
and can be found independently of α k , β m , γ m , δk (k = 2,3; m = 1, 2, 3). When a
normal concentrated force acting on the upper surface of any layer is considered,
then the system of Eqs. (1.67), (1.71), (1.73), (1.77), (1.79) is uniform and, conse-
quently, has a trivial solution. In this case the functions Tk = 0, and all equations for
the transforms of the displacements and stresses are considerably simplified. Simul-
taneously, when each tangential concentrated force is considered, the system is non-
uniform. However, its solution by the substitution method can be easily obtained in
an analytical form, what is taken into account while developing the computation
algorithm. The solution of the system of the rest ten equations to determine α k , β m ,
γ m , δ k (k = 2,3, m = 1, 2, 3) is also presented in an analytical form and used in the
calculations by sequential substitutions without application of standard routines of
matrix algebra, requiring much computation time.
When a concentrated force, acting solely on the free surface of a multilayer sys-
tem, is considered, a rather effective way to speed up the computation process is the
method of compliance functions [111, 112], depending only on the integral trans-
formation parameters, thicknesses and moduli of elasticity of the base layers, and
do not depend on the loads being applied. The compliance functions are found prior
to the boundary problem solution, rather simple recurrent relations existing for their
determination. When the compliance functions are used, there is no need to deal
with any auxiliary systems of high-order linear algebraic equations, what is essen-
tially important for large number of layers. Note that the compliance functions were
first considered for solving contact problems for a one-layer base (an elastic layer
on a non-deformable half-space) [3, 15, 18, 66, 154], and then for two-layer bases
as well [109, 150]. Later Petrishin, Privarnikov, Shevlyakov, and Naumov described
the properties of the compliance functions for essentially multilayer bases both in
the case of smooth and bonded layers [87, 96, 97].
If the account of concentrated forces acting on interlayer boundaries within the
base is required, the compliance functions are inapplicable. Therefore, all the cal-
culations performed here were carried out using a common method, applicable in
case the base being loaded both on the free surface and on the internal surfaces of
contacting layers.
After the functions α k , β m , γ m , δ k , γ̃m , δ̃k (k = 2,3, m = 1, 2, 3) having been
found, it is possible to determine the stress-strained state and the dilatation functions
for each layer of the soil base in the Fourier transforms based on the following
expressions:
1 1
(iξ S + iηT),V̄ = − 2 (iηS − iξ T),W̄ = W̄(p),
Ū = −
p2 p
 2 
ξ ξη ν
[email protected]
σ̄x = −2G 2 S + 2 T − ε̄v , (1.81)
p p 1 − 2ν
 2 
η ξη ν
σ̄y = −2G 2 S − 2 T − ε̄v ,
p p 1 − 2ν
38 1 Spatial Contact Models of Elastic Bases
 
1 − ν dW̄ ν
σ̄z = 2G − S ,
1 − 2ν dz 1 − 2ν

i i
τ̄yz = (τ η + τ̃ ξ ),τ̄xz = − (τ̃ η − τ ξ ),
p p

 
ξ 2 − η2 2ξ η dW̄
τ̄xy = G 2
T − 2 S ,ε̄v = − S.
p p dz
The functions S, T and W̄, τ , τ̃ , being contained in Eqs. (1.81), have been deter-
mined above in Eqs. (1.51), (1.55), and (1.56).
Application of the inverse Fourier transformation (1.47) to the transforms (1.81)
in combination with the Eq. [19]

 ∞ ∞
f (p)e±i(ξ x+ηy) dξ dη = 2π pf (p)J0 (pr)dp
−∞ 0
 ∞ 2π x ∞ 2
iξ f (p)e±i(ξ x+ηy) dξ dη = ∓ p f (p)J1 (pr)dp
−∞ r 0

 ∞ x2 ∞
(iξ )2 f (p)e±i(ξ x+ηy) dξ dη = −2π 2 p3 f (p)J0 (pr)dp+
−∞ r 0

y2 − x2 ∞ 3 J1 (pr)
+ p f (p) dp ,
r2 0 pr

 ∞ y2 ∞
(iη)2 f (p)e±i(ξ x+ηy) dξ dη = −2π 2 p3 f (p)J0 (pr)dp+
−∞ r 0

x2 − y2 ∞ 3 J1 (pr)
+ p f (p) dp
r2 0 pr

 ∞ 2π xy ∞ 3
(iξ )(iη)f (p)e±i(ξ x+ηy) dξ dη = − 2 p f (p)J0 (pr)dp−
−∞ r 0

∞ 3 J1 (pr)
−2 p f (p) dp ,
0 pr

 ∞ ±i(ξ x3 ∞
3
(iξ ) f (p)e x+ηy) dξ dη = ±2π 3 p4 f (p)J1 (pr)dp−
−∞ r 0

x(3y2 − x2 ) ∞ 4
[email protected] prJ0 (pr) − 2J1 (pr)
− p f (p) dp ,
r3 0 (pr)2
1.4 Elastic Layered Bases 39

 ∞ yx2 ∞ 4
(iξ )2 (iη)f (p)e±i(ξ x+ηy) dξ dη = ±2π p f (p)J1 (pr)dp−
−∞ r3 0

y(y2 − 3x2 ) ∞ 4 prJ0 (pr) − 2J1 (pr)
− p f (p) dp ,
r3 0 (pr)2
2 ∞
 ∞ xy  4
(iη)2 (iξ )f (p)e±i(ξ x+η y) dξ dη = ±2π p f (p)J1 (pr)dp−
−∞ r3 0

x(x2 − 3y2 ) ∞ 4 prJ0 (pr) − 2J1 (pr) 
− 3
p f (p) 2
dp , r = x2 + y2
r 0 (pr)

enables the true values of the stress-strained state characteristics and dilatation func-
tions to be calculated for each layer. Here we present in an extended form the for-
mulae, corresponding to Eq. (1.81), obtained using the Hankel integrals.
The displacement and stress fields, when a normal concentrated force Qi is act-
ing on the upper boundary of the i-th layer (i = 1, 2, 3), are determined from the
following expressions:

∞
Qi x
U (i) (x,y,z) = − S(i) (p)J1 (pr)dp, (1.82)
2π r
0
∞
Qi y
V (i) (x,y,z) = − S(i) (p)J1 (pr)dp, (1.83)
2π r
0
∞
Qi
W (i) (x,y,z) = pW̄ (i) (p)J0 (pr)dp, (1.84)
2π
0
∞
Qi
σz(i) (x,y,z) = pσ̄z(i) (p)J0 (pr)dp, (1.85)
2π
 0

(i) Qi Gi x2 ∞ (i) y2 − x2 ∞ (i) J1 (pr)
σx (x,y,z) = pS (p)J0 (pr)dp+ + pS (p) dp +
π r2 0 r2 0 pr
Qi Gi νi ∞ (i)
+ · pε̄v (p)J0 (pr)dp,
π 1 − 2νi 0
(1.86)
2 ∞ 
(i) Qi Ei y  (i) x2 − y2 ∞ (i) J1 (pr)
σy (x,y,z) = pS (p)J 0 (pr)dp + pS (p) dp +
2π (1 + νi ) r2 0 r2 0 pr
[email protected]
Qi Gi νi ∞ (i)
+ pε̄v (p)J0 (pr)dp,
π (1 − 2νi ) 0
(1.87)
40 1 Spatial Contact Models of Elastic Bases

∞
Qi y
τyz
(i)
(x,y,z) = · pτ (i) (p)J1 (pr)dp, (1.88)
2π r
0

∞
Qi x
τxz
(i)
(x,y,z) = · pτ (i) (p)J1 (pr)dp, (1.89)
2π r
0
⎡∞ ⎤
 ∞
Q G xy J (pr)
· 2 ⎣ pS(i) (p)J0 (pr)dp − 2 pS(i) (p) dp⎦ ,
i i 1
(i)
τxy (x,y,z) = (1.90)
π r pr
0 0

∞
Qi
εv(i) (x,y,z) = pε̄v(i) (p)J0 (pr)dp. (1.91)
2π
0

If a tangential concentrated force T x is acting in the positive direction of the xi

axis on the upper boundary of the i-th layer, the following equations are obtained to
determine the stress-strained state in an elastic base:

x2 ∞ (i)
Tix y2 − x2 ∞ (i) J1 (pr)
U (i) (x,y,z) = 2
S (p)J0 (pr)dp + S (p) dp+
r 0 2π r2 0 pr
 (1.92)
y2 ∞ (i) x2 − y2 ∞ (i) J1 (pr)
+ 2 T (p)J0 (pr)dp + T (p) dp ,
r 0 r2 0 pr
* 
Ti
x
xy ∞ (i) ∞ (i) J1 (pr)
V (i) (x,y,z) = S (p)J 0 (pr)dp − 2 S (p) dp−
2π r2 0 0 pr
1 (1.93)
∞ (i) ∞ (i) J1 (pr)
− T (p)J0 (pr)dp + 2 T (p) dp ,
0 0 pr

∞
Tx x
W (x,y,z) = i ·
(i)
pW̄(p)J1 (pr)dp, (1.94)
2π r
0

Gi Tix x3 ∞ x(3y2 − x2 ) ∞ prJ0 (pr) − 2J1 (pr)
σx(i) = pS(i) (p)J1 (pr)dp− pS(i) (p) dp+
π r3 0 r3 0 (pr)2
xy2 ∞ x(x2 − 3y2 ) ∞ prJ0 (pr) − 2J1 (pr)
+ pT (i) (p)J1 (pr)dp − pT (i) (p) dp+
r3 0 r3 0 (pr)2

[email protected]
νi x ∞ (i)
+ · pε̄v (p)J1 (pr)dp ,
1 − 2νi r 0
(1.95)
1.4 Elastic Layered Bases 41

(i) Gi Tix xy2 ∞ (i)
σy (x,y,z) = pS (p)J1 (pr)dp−
π r3 0
x(x2 − 3y2 ) ∞ (i) prJ0 (pr) − 2J1 (pr) xy2 ∞ (i)
− 3
pS (p) 2
dp − 3 pT (p)J1 (pr)dp+
r 0 (pr) r 0
x(x2 − 3y2 ) ∞ (i) prJ0 (pr) − 2J1 (pr)
+ pT (p) dp+
r3 0 (pr)2

νi x ∞ (i)
+ · pε̄v (p)J1 (pr)dp ,
1 − 2νi r 0
(1.96)

∞
Tx x
σz(i) (x,y,z) = i · pσ̄z(i) (p)J1 (pr)dp, (1.97)
2π r
0

Tix  xy ∞
J1 (pr) ∞ (i)
(i)
τyz (x,y,z) = 2 dp −
pτ (i) (p) pτ (p)J0 (pr)dp−
2π r2 0 pr 0
1 (1.98)
∞ ∞ J1 (pr)
− pτ̃ (p)J0 (pr)dp + 2 pτ̃ (p) dp ,
0 0 pr

(i) T x x2 ∞ (i) y2 − x2 ∞ (i) J1 (pr)
τxz (x,y,z) = − i pτ (p)J0 (pr)dp + pτ (p) dp−
2π r2 0 r2 0 pr

y2 ∞ (i) x2 − y2 ∞ (i) J1 (pr)
− 2 τ̃ (p)J0 (pr)dp − pτ̃ (p) dp ,
r 0 r2 0 pr
(1.99)

(i) Gi Tix y(y2 − 3x2 ) ∞ (i) prJ0 (pr) − 2J1 (pr)
τxy (x,y,z) = 2 pS (p) dp−
2π r3 0 (pr)2

yx2 ∞ (i) y3 ∞ (i)

−2 pS (p)J 1 (pr)dp − pT (p)J1 (pr)dp+ (1.100)
r3 0 r3 0

x(3y2 − x2 ) ∞ (i) prJ0 (pr) − 2J1 (pr)
+ pT (p) dp ,
r3 0 (pr)2
∞
Tx x
εv(i) (x,y,z) = i · pε̄v(i) (p)J1 (pr)dp. (1.101)
2π r
0
[email protected]
In the case of the tangential concentrated force Ti y , acting in the positive direction
of the yi axis on the upper boundary of the i-th layer, one should replace x with y
and, vice versa, y with x, in the corresponding equations for the case of the Ti x force.
42 1 Spatial Contact Models of Elastic Bases

Two-layer base. The calculation scheme for a two-layer deformable system is

presented in Fig. 1.10. The dimensionality of the functional system of equations,
corresponding to the boundary conditions of the displacement and stress continuity,
is reduced due to the presence of only one internal contact layer and, consequently,
obtaining their solutions becomes less labour-consuming. Here we present the form
of the reduced system in case concentrated forces acting only on the interlayer sur-
face, having implied Q1 = 0, T1 h = T1 y = 0, what at once results in

α1 = δ1 = δ̃1 = 0, (1.102)
L1  
γ2 = β1 (1 − 2ν1 ) sinh p1 + p1 cosh p1
2L2 (1 − ν1 ) (1.103)
,
+γ1 2(1 − ν1 ) cosh p1 + p1 sinh p1 ,
L1  
β2 = β1 2(1 − ν1 ) cosh p1 − p1 sinh p1
2L2 (1 − ν1 ) (1.104)
,
+γ1 (1 − 2ν1 ) sinh p1 − p1 cosh p1 ,
α2 + Q2 = β1 ( sinh p1 − p1 cosh p1 ) − γ1 p1 sinh p1 , (1.105)
i y 
δ2 = (ξ T2x + ηT2 ) + β1 p1 sinh p1 + γ1 (sinh p1 + p1 cosh p1 ) , (1.106)
p

α2 p2 sinh p2 + 2(1 − ν2 )β2 (1 − 2ν2 ) sinh p2 + p2 cosh p2

+2(1 − ν2 )γ2 2(1 − ν2 ) cosh p2 + p2 sinh p2 (1.107)

+δ2 (3 − 4ν2 ) sinh p2 + p2 cosh p2 = 0,
 
α2 (3 − 4ν2 ) sinh p2 − p2 cosh p2 + 2(1 − ν2 )β2 2(1 − ν2 ) cosh p2 − p2 sinh p2
 ,
+2(1 − ν2 )γ2 (1 − 2ν2 ) sinh p2 − p2 cosh p2 − δ2 p2 sinh p2 = 0
(1.108)

1
γ̃2 cosh p2 + δ̃2 sinh p2 = 0, (1.109)
1 − ν2
L1
γ̃2 = γ̃1 cosh p1 , (1.110)
L2
i y
δ̃2 = − (ηT2x − ξ T2 ) + (1 − ν1 )γ̃1 sinh p1 (1.111)
p

where

pk = hk p (k = 1,2) .

[email protected]
The functional system of Eqs. (1.103−1.111) should be solved by the method of
sequential elimination using the same scheme, as in the case of the three-layer base,
considering separately the vertical Q2 and the horizontal forces T2 h , T2 y .
1.4 Elastic Layered Bases 43

Fig. 1.10 Concentrated

forces in a finite-thickness
elastic layer

In a particular case, when the deformation parameters of both layers are equal
(ν 1 = ν 2 = ν, E1 = E2 = E), one arrives at the simplest case for the two-layer base
under concentrated forces acting on a fictitious surface z = ζ separating the homo-
geneous layers (Fig. 1.10). Such calculation scheme corresponds to the problem of
action of concentrated forces within a homogeneous finite-thickness layer and is of
separate interest. Since while solving problems of concentrated forces, acting inside
a multilayer half-space, according to the approach applied, a fictitious separation
surface should be introduced at the level of the force application point, this problem
has important methodological aspects and deserves a more detailed consideration.
Concentrated forces, acting in a finite-thickness elastic layer. At equal deforma-
tion parameters of the layers and L1 = L2 = L = 2(1−ν 2 )/E, the system of Eqs.
(1.103–1.111) is written in the form

1  
γ2 = β1 (1 − 2ν) sinh p1 + p1 cosh p1
2(1 − ν) (1.112)
,
+γ1 2(1 − ν) cosh p1 + p1 sinh p1 ,
1  
β2 = β1 2(1 − ν) cosh p1 − p1 sinh p1
2(1 − ν) (1.113)
,
+γ1 (1 − 2ν) sinh p1 − p1 cosh p1 ,
α2 + Q2 = β1 ( sinh p1 − p1 cosh p1 ) − γ1 p1 sinh p1 , (1.114)

i y 
δ2 = (ξ T2x + ηT2 ) + β1 p1 sinh p1 + γ1 (sinh p1 + p1 cosh p1 ) , (1.115)
p

α2 p2 sinh p2 + 2(1 − ν)β2 (1 − 2ν) sinh p2 + p2 cosh p2

+2(1 − ν)γ2 2(1 − ν) cosh p2 + p2 sinh p2 , (1.116)

+δ2 (3 − 4ν) sinh p2 + p2 cosh p2 = 0
 
α2 (3 − 4ν) sinh p2 − p2 cosh p2 + 2(1 − ν)β2 2(1 − ν) cosh p2 − p2 sinh p2 +

+2(1 − ν)γ2 (1 − 2ν) sinh p2 − p2 cosh p2 − δ2 p2 sinh p2 = 0,
[email protected] (1.117)

1
γ̃2 cosh p2 + δ̃2 sinh p2 = 0, (1.118)
1 − ν2
44 1 Spatial Contact Models of Elastic Bases

γ̃2 = γ̃1 cosh p1 , (1.119)

i y
δ̃2 = − (ηT2x − ξ T2 ) + (1 − ν)γ̃1 sinh p1 (1.120)
p
where pk = hk p (k = 1, 2).
Further on we consider the most important from the practical point of view case
when only the vertical force Q2 is acting. By assuming T h = T y = 0 in the system
of Eqs. (1.112–1.120) and by denoting

e1 = exp ( − 2p1 ), e2 = exp ( − 2p2 ), ck = (1 + ek )/2, sk = (1 − ek )/2,

ε = exp ( − p1 ), n1 = 2(1 − ν), n2 = 1 − 2ν, n3 = 3 − 4ν,

one arrives at a relatively simple linear algebraic system with respect to the values
β 1 , γ 1 , β 2 , γ 2 , α2 , δ 2 :

β1 (n2 s1 + p1 c1 ) + γ1 (n1 c1 + p1 s1 ) − n1 εγ2 = 0, (1.121)

β1 (n1 c1 − p1 s1 ) + γ1 (n2 s1 − p1 c1 ) − n1 εβ2 = 0, (1.122)

β1 (s1 − p1 c1 ) − γ1 p1 s1 − εα2 = εQ2 , (1.123)

β1 p1 s1 + γ1 (s1 + p1 c1 ) − εδ2 = 0, (1.124)

α2 p2 s2 +n1 β2 (n2 s2 + p2 c2 )+n1 γ2 (n1 c2 + p2 s2 )+δ2 (n3 s2 + p2 c2 ) = 0, (1.125)

α2 (n3 s2 − p2 c2 )+n1 β2 (n1 c2 − p2 s2 )+n1 γ2 (n2 s2 − p2 c2 )−δ2 p2 s2 = 0. (1.126)

The solution of the system of Eqs. (1.121–1.126) is presented in the explicit form:

1 −α ζ̄ Nβ1
e β1 =
, (1.127)
2μ1 D
1 −α ζ̄ Nγ1
γ1 = e , (1.128)
2μ1 D
1 Nβ
β2 = 2
· 2, (1.129)
8μ1 D
1 Nγ
γ2 = · 2, (1.130)
4μ21 D
1 Nδ
[email protected]
δ2 = −
4μ1 D
· 2, (1.131)
1 Nα
α2 = − · 2 (1.132)
2μ1 D
1.4 Elastic Layered Bases 45

where the following notations are used:

   
D (α) = 2 μ5 + 2α 2 e−2α + μ3 1 + e−4α ,
    
Nβ1 α,ζ̄ = μ3 2μ1 + α ζ̄ + 2μ1 μ3 + α (1 − 2α) ζ̄ + 2α (α − 2μ1 ) e−2α +

+ −2μ1 μ3 + α (1 + 2α) ζ̄ − 2α (α + 2μ1 ) e−2α(1−ζ̄ ) +
 
+μ3 −2μ1 + α ζ̄ e−2α(2−ζ̄ ) ,
    
Nγ1 α,ζ̄ = −μ3 μ2 + α ζ̄ + μ2 μ3 − α (1 + 2α) ζ̄ + 2α (α − μ2 ) e−2α +

+ μ2 μ3 + α (1 − 2α) ζ̄ + 2α (α + μ2 ) e−2α (1−ζ̄ ) +
 
+μ3 −μ2 + α ζ̄ e−2α(2−ζ̄ ) ,
    
Nβ2 α,ζ̄ = 4α 2α 2 ζ̄ 2 + 2 2νμ4 − α 2 − 2 ζ̄ − μ5 e−2α +
 
+μ23 1 − e−4α +

+μ3 2α 2 ζ̄ 2 + 2μ3 α ζ̄ + μ5 e−2α ζ̄ +

+ −2α 2 ζ̄ 2 + 2α (2α + μ3 ) ζ̄ − 2αμ3 − 2α 2 − μ23 e−2α(1−ζ̄ ) +

+μ3 −2α 2 ζ̄ 2 + 2μ3 α ζ̄ − μ5 e−2α(2−ζ̄ ) +

+ 2α 2 ζ̄ 2 − 2α (2α − μ3 ) ζ̄ − 2αμ3 + 2α 2 + μ23 e−2α(1+ζ̄ ) ,
  
Nγ2 α,ζ̄ = 2μ3 −2α 2 ζ̄ + 2μ1 μ2 + α 2 e−2α +

+μ3 α 2 ζ̄ 2 − 2μ1 μ2 e−2α ζ̄ +
 2
+α 2 1 − ζ̄ e−2α(1−ζ̄ ) +

+μ3 α 2 ζ̄ 2 − 2μ1 μ2 e−2α (2−ζ̄ ) +
 2
+α 2 1 − ζ̄ e−2α(1+ζ̄ ) ,
    
Nδ2 α,ζ̄ = 4α 2α 2 ζ̄ 2 + 2 2νμ4 − α 2 − 2 ζ̄ + μ2 e−2α +
 
+μ2 μ3 1 − e−4α +

+μ3 2α 2 ζ̄ 2 + 2μ2 α ζ̄ − μ2 e−2α ζ̄ +

+ −2α 2 ζ̄ 2 + 2α (2α + μ2 ) ζ̄ − 2α (α + μ2 ) − μ2 μ3 e−2α (1−ζ̄ ) +

+μ3 −2α 2 ζ̄ 2 + 2μ2 α ζ̄ + μ2 e−2α(2−ζ̄ ) +

+ 2α 2 ζ̄ 2 − 2α (2α − μ2 ) ζ̄ + 2α (α − μ2 ) + μ2 μ3 e−2α(1+ζ̄ ) ,
      
Nα2 α,ζ̄ = 2 α 2 μ3 − 2μ2 ζ̄ + μ1 μ6 e−2α + μ1 μ3 1 + e−4α +
[email protected]

+μ3 α 2 ζ̄ 2 + 2μ1 α ζ̄ + μ1 e−2α ζ̄ +
46 1 Spatial Contact Models of Elastic Bases


+ α 2 ζ̄ 2 − 2α (α + μ1 ) ζ̄ + α (α + 2μ1 ) + μ1 μ3 e−2α(1−ζ̄ ) +
2 2  −2α 2−ζ̄
+μ3 α ζ̄ − 2μ1 α ζ̄ + μ1 e ( ) +

+ α 2 ζ̄ 2 − 2α (α − μ1 ) ζ̄ + α (α − 2μ1 ) + μ1 μ3 e−2α(1+ζ̄ ) ,
μ1 = 1 − ν, μ2 = 1 − 2ν, μ3 = 3 − 4ν, μ4 = 3 − 2ν,
μ5 = 8ν 2 − 12ν + 5,

2
μ6 = 16ν 2 − 20ν + 7 are dimensionless parameters, α = ρH, ζ̄ = ζ H, ρ = r/H
are dimensionless variables.
The dependence of the functional coefficients β 1 , γ 1 , β 2 , γ 2 , α 2 , δ 2 on the
dimensionless parameter α of the Hankel transformation for the fixed values of
ν = 0.3, ζ̄ = 0.25 is illustrated by Fig. 1.11a. As follows from the plotted data, all
the dependences, with the exception of α2 , have a non-monotonous character of the
initial part with distinct extrema. However, with the increase of the transformation
parameter all the dependences rather soon (at α ≈ 25) achieve their limiting values.

(a)

(b)

[email protected]
Fig. 1.11 Dependence of functional coefficients on the transformation parameter α (a) for a
constant-thickness elastic layer and (b) for a half-space; ν = 0.3, ζ/H = 0.2
1.4 Elastic Layered Bases 47

When the elastic layer is of infinite thickness, one arrives at the classical base
model in the form of a homogeneous half-space with the fundamental singular
Mindlin’s solution. The limiting transition at H → in Eqs. (1.84)–(1.89) results
in

1 −α ζ̄
β1 (α,ζ̄ ) = e (2μ1 + α ζ̄ ),
2μ1
1 −α ζ̄
γ1 (α,ζ̄ ) = − e (μ2 + α ζ̄ ),
2μ1
1  
2 ζ̄ 2 + 2μ α ζ̄ + μ )e−2α ζ̄ ,
β2 (α,ζ̄ ) = μ 3 + (2α 3 5
8μ21
1
γ2 (α,ζ̄ ) = (α 2 ζ̄ 2 − 2μ1 μ2 )e−2α ζ̄ ,
4μ21
1  
δ2 (α,ζ̄ ) = − μ2 + (2α 2 ζ̄ 2 + 2μ2 α ζ̄ − μ2 )e−2α ζ̄ ,
4μ1
1  
α2 (α,ζ̄ ) = − μ1 + (α 2 ζ̄ 2 + 2μ1 α ζ̄ + μ1 )e−2α ζ̄ .
2μ1

The dependences of the functional coefficients β 1 , γ 1 , β 2 , γ 2 , α 2 , δ 2 on the

transformation parameter α, corresponding to the Mindlin’s solution, for the same ν
and ζ̄ values as for the case of Fig. 1.11a, are plotted in Fig. 1.11b. It is seen that in
this case β 1 , β 2 , and α 2 are already strictly monotonous. The extrema in the plots
of δ 2 , γ 1 , and γ 2 are rather weak and, as shown by the calculations, soon disappear
with the increase of the ζ̄ parameter, i.e. with the increase of the distance between
the force application point from the half-space surface.
The obtained expressions for the functional coefficients β 1 , γ 1 , β 2 , γ 2 , α 2 , δ 2
together with Eqs. (1.82)–(1.91) enable the stress-strained state to be calculated at
any point of the elastic layer. We investigate the settlements of the free layer surface
in case a vertical concentrated force acting inside it at different depth of its appli-
cation. Since the free surface (z = 0) belongs to the first fictitious layer, its vertical
displacements are fully characterized by the functional coefficient β 1 The depen-
dences of the latter coefficient for a finite-thickness layer and a half-space at differ-
ent depths of the concentrated force application point are illustrated by Fig. 1.12.
As should be expected, for both cases the areas, encompassed with the calculated
curves, decrease with ζ̄ , tending to zero in the limit. At α→∞ the character of the
β 1 function decrease is monotonous in all cases, determined by the most slowly
decreasing exponent e−α ζ̄ . The initial part of the β 1 curves for the finite-thickness
layer and the half-space is qualitatively different. In the first case the β 1 depen-
dences, starting from the zero point, increase, possess a characteristic maximum,
[email protected]
decreasing with ζ̄ increase. In the case of the elastic half-space the curves under
consideration exhibit strictly monotonous exponential decay from the unit value. In
fact, the β 1 (α, ζ̄ , ν) function determines the character of free surface settlements,
48 1 Spatial Contact Models of Elastic Bases

(a)

(b)

Fig. 1.12 Dependence of functional coefficient β 1 on the transformation parameter α: (a) for a
constant-thickness elastic layer and (b) for a half-space at various values of ζ /H; ν = 0.3

whose values, according to Eq. (1.84), can be presented in the form

∞
(1 − ν 2 )
W(p,ζ̄ ,z = 0) = Q2 β1 (α,ζ̄ ,ν)J0 (αρ)dα. (1.133)
π EH
0

The results of the numerical calculations according to Eq. (1.133) are illustrated
by the plots (Figs. 1.13, 1.14, and 1.15) of dimensionless settlements W̄ = W/W∗
[email protected]
where W∗ = Q2 /π EH is the chosen displacement measure. The largest settlement
values are observed when the force is acting on the base free surface (ζ = 0). The
functional coefficient β 1 (α,0,ν) for this case is given by
1.4 Elastic Layered Bases 49

Fig. 1.13 Elastic base free surface settlements due to a vertical concentrated force

Fig. 1.14 Elastic base free surface settlements due to a vertical concentrated force acting inside
elastic bases

μ3 (1 − e−4α ) − 4αe−2α 1 μ3 sinh (2α) − 2α

β1 = −2α −4α
= ,(H < ∞),
2(μ5 + 2α )e
2
[email protected] + μ3 (1 + e ) 2 μ3 cosh2 α + α 2 + μ22
β1 = 1, (H → ∞).
50 1 Spatial Contact Models of Elastic Bases

Fig. 1.15 Elastic base free surface settlements due to a vertical concentrated force acting inside a
finite-thickness elastic layer at different Poisson ratio values

At such values of β 1 Eq. (1.133) produces the Yegorov (1.35) and Boussinesq
W0 (r) = (1−ν 2 )/π Er solutions, respectively. At ρ→0, i.e. near the vertical force
application point, the settlement values for both contact models behave in a similar
way, unlimitedly increasing according to a hyperbolic law (Fig. 1.13). Due to the
infinite half-space depth, the Boussinesq solution always gives the largest settlement
values with the lowest decay rate at ρ→∞.
Action of the concentrated force inside the layer (ζ̄ =0) results in finite surface
displacements (Figs. 1.13, 1.14, and 1.15). The smaller is the depth of the force
application point, the larger will be the settlement cone (Fig. 1.13). As one should
expect, the deeper is the elastic layer, the larger are the free surface settlements
for a fixed ζ̄ . Such comparison of the settlements is performed in Fig. 1.14 for
the layer and the half-space in the layer depth scale. The calculation data for the
settlements of a finite-thickness elastic layer surface at ζ̄ = 0.25 and ν = 0.2 and
0.45 are shown in Fig. 1.15. These data enable the effect of compressibility on the
base free surface deformation to be estimated. In particular, with the increase of the
Poisson ratio the settlements decrease, simultaneously their decay rate with ρ→∞
increases.
The determination of other characteristics of the stress-strained state in the layer
does not experience any difficulties either. For example, the calculation formula to
determine the horizontal (radial) displacements of the layer surface can be given by

∞
(1 − ν 2 )
Ur (r,ζ̄ ,z = 0) = Q2 γ1 (α,ζ̄ ,ν)J1 (αρ)dα. (1.134)
π EH
[email protected] 0

The results of the calculations, performed in accordance with Eq. (1.134), are
shown in Fig. 1.16. The horizontal displacement curves for ζ̄ =0 start from the
1.4 Elastic Layered Bases 51

(a)

(b)

Fig. 1.16 Horizontal displacements of elastic base free surfaces at different depths of a vertical
concentrated force application

zero value, exhibit a characteristic maximum of the absolute values, and then
monotonously decay at ρ→∞. As shown by the calculations, horizontal displace-
ments near the concentrated force action line (below the point of the characteristic
maximum of the absolute value) are practically independent of the layer thickness
and are determined solely by the depth of the concentrated force application point.
[email protected]
Simultaneously, with the increase of the distance from the concentrated force line of
action, the difference in the models of the elastic bases (layers of different thickness)
is revealed quite significantly.
52 1 Spatial Contact Models of Elastic Bases

In order to obtain equations, determining the stresses and displacements in a mul-

tilayer base loaded by a tangential concentrated force T, one should at first find the
functional coefficients from the system of Eqs. (1.112)–(1.120) at unit right-hand
parts, and then multiply the obtain results by iξ T/p or by – iηT/p in accordance
with the separation of Eqs. (1.112)–(1.120) into independent subsystems, and add
the obtained so5s in order to determine finally the transforms of the sought values.
This procedure does not meet any principal difficulties, but results in quite cumber-
some formulations and, therefore, is not presented here. A more careful attention is
required to calculate the Hankel integrals from the oscillating functions contained
in the final equations for the stresses and displacements, obtained by application of
the inverse Fourier transformations to the transforms found.
Quite popular is a direct method of calculation of improper integrals with weights
in the form of the Bessel functions [124]. In order to obtain results of the required
accuracy, for each value of the oscillation parameter the integration interval is par-
titioned by the Bessel function zeroes into the segments of constant sign of the
integrand; on each of the segments the Gauss-Legendre quadrature formula with fif-
teen nodes is applied. The calculation error is checked by the absolute values of the
integration results on each segment. It is quite clear that achievement of the required
calculation accuracy at high values of the oscillation parameter by the direct method
under consideration is related to large computation time.
The difficulties arising at the calculation of improper integrals from oscillating
functions (too large computation time or too low calculation accuracy), can be over-
come by improvement of the known quadrature formulae. A rather efficient method
of calculation of the Hankel integrals for solving axisymmetric problems, torsion
problems, and some three-dimensional boundary problems of theory of elasticity
was proposed by Godes and Privarnikov [45]. Similarly to the Filon method for the
calculation of sine- and cosine-Fourier transformations, these authors have devel-
oped special quadrature formulae with a given absolute error, independent of the
oscillation parameter. The experience of numerical calculations ha shown the appli-
cation of the quadrature formulae proposed in [45] to result in the accuracy not
worse than the Gauss method with six nodes, when the integration interval was par-
titioned for each value of the oscillation parameter into such number of intervals
that each of them should contain not more than one extremum point of the inte-
grand function. The time of calculation of integrals using the Gauss method linearly
increased with the oscillation parameter, while for the special quadrature formulae
it was independent of this parameter. Note that an inconvenience of application of
the Godes and Privarnikov quadrature formulae consists in the necessity of a correct
preliminary choice of a finite integration interval, large enough for the remainder
to be negligible. A method of the Hankel integral calculation based on the same
idea was later proposed by Minoru et al. [80]. However, this method implies the
use of the Lommel function and the hypergeometrical function what encumbers its
computer implementation.
[email protected]
For mass calculations of improper Hankel integrals, arising in the framework
of the boundary-element method, we used an effective numerical-and-analytical
method, including presentation of the integrals in terms of two components
1.4 Elastic Layered Bases 53

∞ A   ∞
N(α) N(α) D(α) 1
Jm (αρ)dα = 1− Jm (αρ)dα + N(α)Jm (αρ)dα
D(α) D(α) D∗ D∗
0 0 0
(1.135)
where D∗ is the limiting value of the integrand denominator at α→∞, A is the
finite upper limit when at α > A one can with high accuracy imply that D(α) ≈ D∗ ,
m = 0,1. The first term in Eq. (1.135) is rather effectively estimated on the basis
of the adaptive numerical Romberg integration scheme similarly to the way it was
done in our earlier paper [11]. The second term is expressed in a closed form, using
the formula [49]

∞   
n −qα n −m ∂n ( q2 + ρ 2 − q)m
α e Jm (αρ)dα = ( − 1) ρ  , q > 0.
∂qn q2 + ρ 2
0

The upper limit value A is determined for each multilayer system in accordance
with the law of the asymptotic behaviour of the integrand denominator. For example,
for a concentrated force in an elastic layer the estimation of the approximation of
D(α) to D∗ has enabled one to assume A = 10 with a relative error of  = 10−6 for
all Poisson ratio values (Table 1.2).
The proposed approach to the calculation of the two-layer system deformation
can be easily generalized for the case of existence of elastic constraints between the

Table 1.2 D(α)/ D∗ ratio at different values of the transformation parameter α

α Poisson ratio ν
0 0.1 0.2 0.3 0.4 0.5

0.5 2.484227 2.374806 2.279106 2.206273 2.175909 2.238974

1.0 1.649880 1.630448 1.623633 1.637851 1.691125 1.830327
1.5 1.317797 1.323414 1.338315 1.368690 1.427803 1.550136
2.0 1.159071 1.167712 1.182160 1.206285 1.248381 1.330017
2.5 1.078655 1.084944 1.094499 1.109500 1.134612 1.181970
3.0 1.038014 1.041725 1.047148 1.055420 1.068986 1.094199
3.5 1.017935 1.019908 1.022732 1.026972 1.033845 1.046507
4.0 1.008275 1.009259 1.010650 1.012718 1.016045 1.022141
4.5 1.003743 1.004213 1.004871 1.005844 1.007401 1.010243
5.0 1.001665 1.001882 1.002184 1.002629 1.003339 1.004631
5.5 1.000729 1.000827 1.000963 1.001162 1.001479 1.002054
6.0 1.000315 1.000359 1.000418 1.000506 1.000645 1.000897
6.5 1.000135 1.000154 1.000180 1.000218 1.000278 1.000386
7.0 1.000057 1.000065 1.000076 1.000093 1.000118 1.000165
7.5 1.000024 1.000027 1.000032 1.000039 1.000050 1.000069
8.0 1.000010 1.000011 1.000013 1.000016 1.000021 1.000029
8.5 1.000004 1.000005 1.000006 1.000007 1.000009 1.000012
[email protected]
9.0 1.000002 1.000002 1.000002 1.000003 1.000004 1.000005
9.5 1.000001 1.000001 1.000001 1.000001 1.000001 1.000002
10.0 1.000000 1.000000 1.000000 1.000000 1.000001 1.000001
54 1 Spatial Contact Models of Elastic Bases

layers in the following way [44]. The k-th and the (k+1)-th layers are assumed to be
connected by elastic constraints with a coefficient μk , i.e. the stresses σ z , τ yz , τ xz
and vertical displacements W on their common boundary coincide, and

(k+1) (k) 1 − νk (k)

uj (x,y,0) − uj (x,y,hk ) = μk τjz (x,y,hk )
Gk
where ν k is the Poisson ratio of the k-th layer, j = x,y. At μk = 0 the case of the half-
space cohesion is obtained, at μk →∞ the layers can slide without friction along
their common boundary. Thus, the boundary conditions of conjugation of the con-
tacting layers in the matrix form are given by
⎛ (k+1) ⎞ ⎛ ⎞ ⎛ σ (k) ⎞
σz 1 0 0 000 z
⎜ (k+1) ⎟ ⎜ ⎟⎜ ⎟
⎜τ ⎟ ⎜0 1 ⎜ τ (k) ⎟
⎜ xz ⎟ ⎜ 0 0 0 0⎟
⎟⎜⎜ xz ⎟
⎜ ⎟ ⎜ ⎟ ⎜ (k) ⎟
⎜ (k+1) ⎟ ⎜0 0 ⎟
⎜ τyz ⎟ ⎜ 1 0 0 0⎟
⎟⎜ τyz ⎟
⎜ ⎟ ⎜ ⎟⎜⎜ ⎟ (1.136)
⎜ (k+1) ⎟ ⎜ 0 ψk (k) ⎟
⎜ ux ⎟ ⎜ 0 1 0 0⎟
⎟⎜ x ⎟
⎜ u
⎜ ⎟ ⎜ ⎟⎜ ⎟
⎜ (k+1) ⎟ ⎜0 0 ⎟
⎜ uy ⎟ ⎝ ψk 0 1 0⎟
⎠⎜ u
(k) ⎟
⎝ ⎠ ⎝ y ⎠
(k+1) 0 0 0 001 (k)
uz uz

where Ψ k = (1−ν k )μk /Gk . The limiting cases of such conditions of the simultane-
ous layer deformation are full cohesion and smooth contact.
Application of the contact conditions of a general type (1.136) enables one to
model the phenomena of interaction at the layer boundaries, encountered in prac-
tice. For example for a soil with an underlayer of a rigid base of a rock type, it is
quite natural to assume the condition of full cohesion (μk = 0) at moderate loads.
A smooth contact (zero tangential stress, μk = ∞) is characteristic, e.g., for the
boundary between the thawed frozen soil when there is a water-saturated interlayer,
eliminating friction between the layers. It is quite clear that this case is characterized
by minimal mathematical difficulties. Intermediate cases (0<μk <∞) can be accom-
plished when artificial bases are constructed using modern geosynthetic materials,
accepting considerable tensile forces even at small elongation: geogrids, geotex-
tiles (fabrics and grids), applied for the separation and binding of layers, drainage
in ground and hydrotechnical engineering, reinforcement of asphalt concrete road
pavement, etc.
Thus, using the two-dimensional Fourier transformation, a numerical-and-
analytical method of construction of fundamental solutions of spatial problems of
theory of elasticity for multilayer bases without any restrictions on the layer thick-
nesses and elastic parameters has been developed. The method includes the effective
procedure of calculation of improper Hankel integrals with exponentially decay-
[email protected]
ing kernels. The obtained results enable three-dimensional contact problems for
complex-shaped structures, deepened into spatially nonhomogeneous (layered) soils
to be solved within the numerical algorithm of the boundary-element method.
1.5 Elastic Bases with the Deformation Modulus, Variable with Depth 55

Based on the solutions, developed for the layered linearly deformable half-space,
numerical modelling of spatial contact interaction of soil bases with drop caissons
of bridge piers was performed [10]. Attempts to solve such problems had been made
earlier in a number of studies where the interaction of the foundation with a mul-
tilayer base was modelled by Winkler constraints. Such solutions do not take into
account the distributive ability of the soil base and interaction of the layers with each
other: the rigidity of the constraints on the contact surface is determined by a rough
indirect method. The noted shortcomings were overcome when the model of elastic
layered media, more adequate to the nature, was applied. To estimate the possibil-
ities and advantages of the proposed method of calculation of drop caissons based
on the base deformations, the examples of deepened foundation structures, penetrat-
ing through the layers of sand loam, dense fine sand, and low-plasticity clay, were
considered. The proposed fundamental solutions are shown to give more reliable
predictions of interaction of the drop caissons with the layered bases in comparison
with the methods, using the Winkler model of the soil.

1.5 Elastic Bases with the Deformation Modulus,

Variable with Depth
In the literature in mechanics of deformable solids a number of dependences are
quoted for the account of spatial nonuniformity of mechanical properties of vari-
ous type [4, 13, 23, 24, 26, 46, 58, 59, 67, 74, 75, 88, 93, 98, 101–104, 107, 108,
118, 120, 137–139]. An important geotechnical problem consists in the determi-
nation of the stress-strained state in an elastic nonhomogeneous half-space whose
modulus of deformation (or shear modulus, at a constant Poisson ratio) is varied
with depth [2, 16, 21, 22, 35, 56, 42, 57, 62, 65, 76, 81, 84–86, 113, 114, 119, 121,
125–127, 142, 155, 157]. This problem is most successfully solved on the basis of
development of contact models of the soil base of natural bedding. A review of pub-
lications on the problem of application of theory of elasticity of nonhomogeneous
media for solving contact problems for a half-space, when the elastic parameters are
functions of the spatial parameters, was recently performed by Selvadurai [126]. As
follows from a number of studies, the contact model of the elastic base in the form
of a nonhomogeneous half-space with a deformation modulus, regularly increasing
with depth, describes the deformational properties of real bases more exactly than
widely applied base models in the form of a homogeneous elastic half-space of a
finite-thickness layer and, hence, it corresponds better to the mechanism of contact
interaction of soil bases and foundation structures.

1.5.1 Variation of Deformation Modulus with Depth

[email protected]
As a result of processing of the experimental data, obtained at static tests, performed
on the base surface, an average value of the deformation modulus E of the punch
56 1 Spatial Contact Models of Elastic Bases

base active area is determined, its value being usually smaller than for the deeper
layers of the soil. Simultaneously, the available experimental data on the soil mass
deformability give the evidence for a regular increase of the deformation modulus
with depth and show a noticeable increase of the soil deformability at open surfaces
in underground workings. Thus, the average constant value of E, obtained from
the measurements of displacements of test site surface, based on the solutions of
theory of elasticity for a homogeneous half-space, is rather approximate and does
not enable the real properties of soil bases to be characterized with the required
reliability. Besides, its value will evidently be lower than the real one since the
major part of the punch base corresponds to the loss of strength area.
In the simplest case, the dependence of the deformation modulus on depth is
determined, using flexible punches, from the measurement of the test site surface
in the direction of the load application in 15–20 points. In this case the solution of
the problem of theory of elasticity regarding a homogeneous elastic half-space or a
homogeneous limited-thickness layer under a uniform load [121].
A rather pronounced increase of the deformation modulus with deepening into
the mass soil arises due to the pressure of the upper layers of the soil onto the lower
ones, the stressed state due to the surface load as well as natural bedding of the mass
soil itself and is experimentally confirmed in a number of studies. The form of the
E(z) dependence is determined for specific engineering-and-geological conditions
and is specific for soils of various type. Empirical parameters, characterizing the
base nonhomogeneity, can be obtained at standard soil tests in a well at various
depth or from field experiments using punches of various bearing area [35].
In the literature, the main attention was paid to the E(z) dependences, enabling
the exact solutions of contact problems to be obtained for punches of canonical
shape. The account of the experimental data for the deformational characteristics
at the natural bedding of the soil mass by the E(z) dependence itself was not suffi-
ciently correct. The most known is [4, 16, 21, 22, 62, 81, 84, 108, 113, 119, 120] the
model of a nonhomogeneous half-space with the deformation modulus, increasing
according to the power law

E(z) = E0 · zn ; (1.137)

E0 , n − const.

Evidently, in this case the results of calculation, using the corresponding influ-
ence functions, are hardly suitable for practice since the deformation modulus tends
to zero at the daylight surface. References [14, 35, 56, 57, 67, 76, 85, 102, 142]
employed more realistic dependences (Fig. 1.17), taking into account the nonzero
deformation modulus value E(0) = E0 =0 on the half-space surface

[email protected]
E(z) = E0 + En · zn , (1.138)

E(z) = E0 (1 + k · z)b , (1.139)

1.5 Elastic Bases with the Deformation Modulus, Variable with Depth 57

Fig. 1.17 Various laws of deformation modulus increase with depth

E(z) = E0 · ekz (1.140)

where En , n, b, and k are empirical parameters, characterizing the variation of the

soil properties with depth.
Though for Eqs. (1.138), (1.139), and (1.140) the exception E(0) = 0 does not
hold (like it was for Eq. (1.137)), still these dependences (1.138), (1.139), and
(1.140) are not free from disadvantages. The use of the influence functions, found
for Eqs. (1.138) and (1.139), is limited due to the “rigid” relationship of the n and
b parameters with the Poisson ratio value for the soil. The dependences (1.138),
(1.139), and (1.140) have one more common shortcoming revealed in an unlimited
growth of the deformation modulus with the base depth increase. This fact con-
tradicts with the field observation data [125], which indicate a characteristic sta-
bilization of the deformation modulus with the increase of the compressed base
thickness (depth), when, with the increase of natural stress and the decrease of the
soil jointing, soil and rock properties are converging to those of a homogeneous lin-
early deformable medium. Therefore, increasingly important for practical purposes
becomes the development of influence functions (solution of the Boussinesq prob-
lem), in case dependences of a more general form being used [86, 114, 125, 126,
[email protected]
155, 157]

E = E(z,E0 ,E∞ ,n,k,a,b,...), (1.141)

58 1 Spatial Contact Models of Elastic Bases

taking into account the realistic variation of the deformation modulus with depth
due to the intrinsic weight of the soil, history of the layer formation, distribution of
water, etc.
Below we consider contact models of a nonhomogeneous elastic half-space
whose deformation modulus increases with depth according to the dependences
of a rather general form. It is admissible to assume the Poisson ratio for soils to
be constant since its value varies insignificantly and makes no essential effect on
the stress-strained state characteristics. Taking into account the known relationship
between the elastic constants G = E/2(1+ν), at such consideration the shear modu-
lus G = G(z) will also be variable with depth.

1.5.2 Normal Concentrated Force Acting on the Half-Space

Surface

Let a concentrated force P act orthogonally to the plane z = 0 of a nonhomogeneous

half-space along the z axis, directed inside the half-space (Fig. 1.18). Let the origin
of a cylindrical (r,φ,z) and a spherical (R,Θ,Ψ ) coordinate systems be put in the
point O of the force application. Due to the axial symmetry of the problem, similarly
to the case of a homogeneous half-space, let the following distribution of radial
displacements be assumed:

D · cos θ
uR = (1.142)
R

where θ is the angle, counted from the z axis, 0 ≤ θ ≤π/2. For the radial deforma-
tions  R and stresses σ R

[email protected]
Fig. 1.18 Calculation scheme to the problem of a concentrated force acting on a nonhomogeneous
elastic half-space
1.5 Elastic Bases with the Deformation Modulus, Variable with Depth 59

D · cos θ D · cos θ
εR = − , σR = E(z) · εR = −E(z) . (1.143)
R2 R2

Using the formulae for a transition to the cylindrical coordinates, one obtains

σz = σR cos2 θ . (1.144)

Since (See Fig. 1.18)

z 
cos θ = ,R = z2 + r2 ,
R
then
D·z
σR = −E(z) · , (1.145)
R3
D · z3
σz = −E(z) · . (1.146)
R5
The factor D, contained in Eqs. (1.143), (1.145), and (1.146), can be determined
from the equilibrium equation of a hemisphere, cut out in the vicinity of the point
O of application of the force P, by collecting the pressures from the surface of this
hemisphere
 
P+ σR · cos θ dS = 0,
S
or, in an extended notation,

2π π/2
P+ dϕ σR · cos θ · R2 · sin θ · dθ = 0.
0 0
Taking into account Eq. (1.145), for the constant D one can obtain

P
D= . (1.147)
1
2π E(Rt)t2 dt
0

At E = E0 = const, Eq. (1.147) takes the form of the known Boussinesq solution
[17]:

3P
[email protected] = 2π E0 .
Using Eqs. (1.146) and (1.147) with the account of transformation of stress into
the cylindrical coordinate system, the vertical stress σ z is given by
60 1 Spatial Contact Models of Elastic Bases

P · E(z) z3
σz = − · 5. (1.148)
1 R
2π E(Rt)t2 dt
0

Hence, as should be expected, at E(z) = E0 = const the known Bousinesq formula

for compressive stress in a homogeneous half-space is obtained:

3P z3
σz = − · . (1.149)
2π R5
In order to determine the spatial stress-strained state of the nonhomogeneous
elastic base, at first the vertical deformation should be found, using the Hooke’s
law:

1 
εz = σz − ν(σx + σy ) , (1.150)
E(z)

or, in case of an axisymmetric stress-strained state,

1 
εz = σz − ν(σr + σϕ ) . (1.151)
E(z)

Since the distributions of the σ r and σ φ stresses in the nonhomogeneous half-

space are unknown, we take advantage of the semi-inverse method, assuming the
hypotheses for the distribution of σ r and σ φ (or σ x and σ y ). Following [142], we
make the simplest assumption of proportionality of the normal stresses:

σr − σz , σϕ − σz . (1.152)

Then

γ P·γ z3 P·γ z3
εz = σz = − · = − · (1.153)
E(z) 1 R5 2π · e(R) R5
2π E(Rt)t2 dt
0

where γ is an indefinite coefficient and a notation is introduced:

1
e(R) = E(Rt)t2 dt. (1.154)
0
[email protected]
The function e(R) characterizes the degree of the elastic base nonhomogeneity
due to the deformation modulus variation with depth.
Now, using the expression for the vertical deformation
1.5 Elastic Bases with the Deformation Modulus, Variable with Depth 61

dw
εz = ,
dz
one can easily find settlements (vertical displacements) depending on the depth z:

∞
P·γ ς 3 dς
w(r,z) = . (1.155)
2π e(R) · R5
z

The proportionality coefficient γ can be easily determined from the compari-

son (at any depth) of w values, obtained from Eq. (1.155) and for the case of the
homogeneous half-space (E(z) = E0 = const). Since on the surface z = 0 of the
homogeneous half-space


(1 − ν 2 ) · P 1
w0 = · ,r= x2 + y2 , (1.156)
π · E0 r

and the integral value

∞
z3 dz 2 1
= · , (1.157)
R5 3 r
0

and taking into account that

1
E0
e(R) = E0 t2 dt = , (1.158)
3
0

one obtains for the indefinite factor γ = 1−ν 2 . Thus, Eq. (1.155) is obtained on the
basis of the hypothesis

σr = σϕ = 0.5ν · σz (1.159)

which is in agreement with the assumption of Eq. (1.152) used earlier [142]. As
seen from Eq. (1.159), according to the assumed hypothesis, vertical stresses in the
nonhomogeneous half-space are distributed in the horizontal plane in accordance
with the Poisson’s law in equal ratios both in radial and tangential directions. Let us
find other components of the stress-strained state.
In a cylindrical coordinate system tangential stress τ rz is given by [17]

[email protected] 1
τrz = σR sin 2θ .
2
Since in our case
62 1 Spatial Contact Models of Elastic Bases

z r 1 rz 
cos θ = , sin θ = , sin 2θ = 2 , R = z2 + r2 ,
R R 2 R
then

P E(z) z2 · r
τrz = − · · . (1.160)
2π e(R) R5

For the normal stresses σ r and σ φ (or σ x and σ y ), according to Eqs. (1.148) and
(1.159), one obtains

ν P E(z) z3
σr = σϕ = σz = − · · . (1.161)
2 4π e(R) R5

The radial displacement ur is found, using the dependence

ur
εϕ =
r
where

1 
εϕ = σϕ − ν(σr + σz ) .
E(z)

In our case, with the account of Eq. (1.161),

ν(1 + ν)P rz3

ur = · . (1.162)
4π · e(R) R5

At R→∞, as should be expected, ur →0.

Finally, we calculate the stressed state dilatation in the half-space due to the force
P:

(3) 1−ν  (1 − 2ν)(1 + ν)P z3

εV = σz + σϕ + σr = − · 5. (1.163)
E(z) 2π · e(R) R

Now, if necessary, one can easily obtain the six components of stress and three
components of displacement in the Cartesian coordinate system. If the cylindrical
[email protected]
coordinate system is converted to a Cartesian coordinate system, the stress com-
ponents σ r , σ φ , τ rz will be converted to the components σ x , σ y , τ xy , τ yz , τ zx as
follows:
1.5 Elastic Bases with the Deformation Modulus, Variable with Depth 63
⎧
⎪
⎪ σx = σr · cos2 ϕ + σϕ · sin2 ϕ,
⎪
⎪
⎪
⎪
⎪
⎪ σ = σr · sin2 ϕ + σϕ · cos2 ϕ,
⎨ y
τxy = (σr − σϕ ) · sin2 ϕ/2, (1.164)
⎪
⎪
⎪
⎪
⎪
⎪ τzx = τrz · cos ϕ,
⎪
⎪
⎩
τyz = τrz · sin ϕ.

The displacements u, v in the x and y directions, respectively, are expressed in

terms of the radial displacement ur :

u = ur · cos ϕ, v = ur · sin ϕ. (1.165)

The vertical stresses σ x and displacements w are not converted and are found,
using Eqs. (1.148) and (1.155) at γ = 1−ν 2 . For Eqs. (1.164) and (1.165) one
should use the known transformation formulae

x = r · cos ϕ, y = r · sin ϕ;

y2 x2 − y2 2xy
cos 2ϕ = 1 − 2 · 2
= 2
, sin 2ϕ = 2 .
r r r
As one can see, the consideration of the proposed nonhomogeneous base model
results in formulae whose structure is similar to those of classical theory of elastic-
ity. Simultaneously, one should expect that the predictions of the nonhomogeneous
base model, using Eqs. (1.155), (1.160)–(1.166), will be closer to the characteris-
tics of real soil bases. In particular, the experimental data show that the distributive
properties (the ability of distribution of a vertical load in the horizontal direction) of
a homogeneous elastic base are exaggerated [47, 48]. As shown in the subsequent
section, the application of the nonhomogeneous base calculation models consid-
ered here, results in a reduction of the base distributive ability and, consequently,
a decrease of the calculation forces for design of shallow foundation structures of
beam or plate type on elastic bases.

1.5.3 Settlement of a Nonhomogeneous Half-Space Surface

For solution of practical geotechnical problems the most interesting is calculation
of vertical displacements (settlements) of the soil base surface. For the model of
[email protected]
a nonhomogeneous elastic half-space under consideration, when the modulus of
elasticity is a given function of depth E = E(z), according to Eqs. (1.154) and (1.155)
one obtains
64 1 Spatial Contact Models of Elastic Bases

∞
(1 − ν 2 ) · P z3 dz
w |z=0 = w(r) = . (1.166)
2π e(R) · R5
0

1
Here e(R) = E(Rt)t2 dt is the function of the nonhomogeneity degree.
0
First consider the cases of linear and square nonlinearities (Fig. 1.17), proposed
to be taken into account by Snitko [142]. He gave some estimations of the nonho-
mogeneous half-space surface settlements when the deformation modulus increases
with depth according to a linear or a parabolic law. Even rather rough estimations for
such cases of nonhomogeneity show much lower distributive ability of the soil than
for the homogeneous half-space model, what is important to be taken into account
at the design and calculation of foundation structures. However, no formulae were
given in [142] for the influence functions which enable the mixed (contact) problems
of theory of elasticity to be reduced to boundary integral equations and numerical
methods to be applied for solving spatial problems of soil mechanics, first of all, for
the calculation of contact stresses and settlements for foundation structures of given
geometry. Note also that the knowledge of the influence functions in the explicit
form is important since it enables not only more effective numerical algorithms to
be constructed, but also the applicability of an influence function to be evaluated
qualitatively in view of the account of the nonhomogeneity characteristics.
Linear law of the deformation modulus increase. Let the soil deformation modu-
lus vary with depth according to a linear law

E1 − E0  z
E(z) = E0 + α · z = E0 + · z = E0 1 + B (1.167)
h h
where E0 is the deformation modulus value at the soil daylight surface, E∞ is its
value at the depth z = h, B = (E1 –E0 ) / E0
In the case under consideration.

1 1
E0 αR
e(R) = e(Rt)t dt =
2
(E0 + αRt) · t2 dt = + , (1.168)
3 4
0 0

and the integral (1.166) after rather cumbersome transformations can be calculated
in the finite form, and the following expression for the settlement of the daylight
surface is obtained:

2(1 − ν 2 )P ∞ z3 dz (1 − ν 2 ) · P 1
w|z=0 = w(r) =   = · +
π 4 π · E0 r
[email protected] E0 + αR · R
0 5
(1.169)
 

(1 − ν ) · P B 1
2 B·r B ·r
2 2 B·r
+ · 2 − + 1 − 2 2 ln
π · E0 h A 2 A·h A ·h B·r+A·h
1.5 Elastic Bases with the Deformation Modulus, Variable with Depth 65

where for the sake of simplicity a notation A = 4/3 is introduced.

If dimensionless settlements

w ∗ (1 − ν 2 ) · P
V= ,w = (1.170)
w∗ π · E0 · h
and distance
3 y
r x 2 2
r̃ = = + (1.171)
h h h
are introduced, then Eq. (1.169) can be written in a more convenient form
 

1 B 1 B · r̃ B2 · r̃2 B · r̃
V= + 2 − + 1− ln . (1.172)
r̃ A 2 A A2 B · r̃ + A

It is easily seen that the first terms in Eqs. (1.169) and (1.172) are the Boussinesq
solutions for a homogeneous elastic half-space, to which the obtained solution is
reduced at B→0.
Parabolic law of the deformation modulus increase. Dependence of the deforma-
tion modulus on depth is given by
 
z2
E(z) = E0 + β · z = E0 2
1 + B (1.173)
h2

where the notation B = (E1 –E0 )/E0 is still used for the deformation modulus relative
variation with depth. Then consecutively calculate

1 1
E0 βR2
e(R) = E(Rt)t dt =
2
(E0 + βR2 t2 ) · t2 dt = + , (1.174)
3 5
0 0
∞
3(1 − ν 2 )P z3 dz
w|z=0 = w(r) =   . (1.175)
2π 3
0 E0 + βR2 · R5
5

The latter improper integral after rather cumbersome transformations is reduced

to a set of elementary functions, and Eq. (1.175) for the settlement of a nonhomo-
geneous elastic half-space in the dimensionless form is given by

1 3C   √
V= + C · r̃ − (1 + C2 · r̃2 ) · arccot(C · r̃) ,C = 0.6B. (1.176)
r̃ 2
[email protected]
Comparative calculations of the settlements of a nonhomogeneous half-space
surface for the cases of linear and square increase of the deformation modulus with
66 1 Spatial Contact Models of Elastic Bases

Fig. 1.19 Settlements of the surface of elastic bases with linear (n = 1) and parabolic (n = 2)
nonuniformities due to a vertical concentrated force

depth are presented in Fig. 1.19. The same figure shows the settlements plotted ver-
sus the distance from the point of application of the concentrated force P for a homo-
geneous half-space (the Boussinesq solution). As one can see, with the increase of
the B parameter, characterizing the degree of the soil base nonhomogeneity with
depth, higher concentration of stresses occurs and the distributive ability of the soil
is revealed to a lower extent. Simultaneously, with the increase of the B parameter,
characterizing the rate of the deformation modulus increase with depth, the differ-
ence in the soil surface settlements for the linear and square variation law becomes
less significant. At B>20 the surface deflections resemble the Dirac δ-function and
the nonhomogeneous base contact models (1.172) and (1.176) approach the Winkler
type model.
Note that the direct application of Eq. (1.176) is rather complicated for calcula-
tions. This is related to the specific features of computer arithmetic in case of mul-
tiplication of factors of essentially different orders of magnitude. The experience of
numerical calculations has shown that with the increase of the B parameter, direct
calculations using Eq. (1.176) at Cr̃ > 5, due to the loss of accuracy, result in nega-
tive deflections. A detailed analysis has shown that even double precision in using
standard computer mathematical procedures cannot overcome the errors arising at
[email protected]
the calculation of a term, containing an inverse trigonometric function. The prob-
lem could be solved only by application of a special subroutine for the calculation
of y = arctanx function for x>1, using a power series [49]
1.5 Elastic Bases with the Deformation Modulus, Variable with Depth 67

∞
π 1 1 1 π  (−1)k
arctan x = − + 3 − 5 +... = − , x > 1. (1.177)
2 x 3x 5x 2 (2 k + 1) x2 k+1
k=0

Application of not more than 50 terms of the series (1.177) provided the cor-
rect calculation of settlements, using Eq. (1.176) for all Cr̃>1 with the guaranteed
accuracy.
Note the character of the asymptotic behaviour of the solutions found at r→0
and r→∞.
Using the known expansions of elementary functions, contained in Eqs. (1.172)
and (1.176) into power series, one obtains

1 − ν2 1
w≈ · + O(r−1 ), r → 0 (1.178)
π E0 r

where O is the Landau order symbol. Consequently, at variable modulus of elastic-

ity, the settlements due to a normal concentrated force have the same singularity in
the point r = 0, as in the case of a homogeneous half-space with the modulus of
elasticity E0 . In other words, near the concentrated force application point the sur-
face settlements do not undergo any influence of variation of the base mechanical
properties with depth.
Let us estimate the character of the displacement decay with the increasing
distance from the concentrated force application point. We take the advantage of
asymptotic expressions, following from Eqs. (1.172) and (1.176) at r→∞:

1 4 1
V≈ (n = 1), r̃ >> · ; (1.179)
2r̃ 3 B
1 1 5 1
V≈ · 3 (n = 2), r̃2 >> · . (1.180)
3B r̃ 3 B

As seen from Eqs. (1.179) and (1.180), with the increase of the degree of nonuni-
formity of mechanical properties (i.e. with the increase of the n parameter) an
increase of the rate of the settlement decrease with the distance from the concen-
trated force application point is observed. For the linear law of the deformation
modulus increase (n = 1) the surface settlements far from the concentrated force
application point are exactly twice smaller than for the homogeneous half-space.
Simultaneously, it follows from Eq. (1.180) that at the square law of the deformation
modulus increase (n = 2) this decrease rate also increases with the nonhomogeneity
parameter B.
General power laws of increase of the deformation modulus. In Refs. [35, 56]
a base model is considered with the deformation modulus, varied according to the
[email protected]
law

E(z) = E0 + En · zn , n ≥ 0 (1.181)
68 1 Spatial Contact Models of Elastic Bases

where E0 is the deformation modulus under the foundation bottom. The book [35]
also suggests possible methods for determination of the empirical parameters in Eq.
(1.181) using punch test data or well measurements.
It is difficult to find the soil base displacements in a closed analytical form
according to Eq. (1.166) in the case of the law of Eq. (1.181). Barvashov [14]
obtained an expression for the influence function of integral type, corresponding
to Eq. (1.181) in a simplifying assumption of the soil base being presented by a
parallel combination of a homogeneous half-space with the deformation modulus
E0 and a half-space, for which the deformation modulus varies with depth accord-
ing to the law En ·zn . In order to obtain a more simple and explicit formula for the
calculation of the nonhomogeneous base settlements, it is convenient to apply the
following engineering approach [35].
For the nonhomogeneous soil base model, when E(z) = En ·zn , in [4] the vertical
displacements of points in the half-space, loaded normally by a concentrated force,
applied to the surface, were obtained in the form

∞

(1 + ν) · P  (1 − ν)(m + n + 1)m z2 z m−1
wn = A m + · (1.182)
En · Rn+1 (m + n)(n + 1) R2 R
m=0

where
 
m+n+2
 c0 am 
2
R = x2 + y2 + z2 , Am =  , m = 0,1,2, . . . ,
√ m+n+3
2 π
2

πq πq p2 − q2
a0 = cos , a1 = q · sin , am = am−2 , m ≥ 2;
2 2 m(m − 1)
p = m−1, if m is odd, p = m−2, if m is even;

q= (1 + n)(1 − nν/(1 − ν)), 0 < n < 1;

   
2n+3 n+3+q n+3−q
C0 =  · ,
4π (n + 2) 2 2
G(h) is a gamma function.
The series of Eq. (1.182) converges everywhere, except R = 0. At n = 0, i.e. in
the case of a half-space with a constant deformation modulus, the known Boussinesq
solution is obtained:

[email protected]
P(1 + ν) z2 2(1 − ν)
w0 = · + .
2π E0 R3 R
1.5 Elastic Bases with the Deformation Modulus, Variable with Depth 69

Then the solution (1.182) is used to find the base settlements at E(z)=E0 +En ·zn
in the following way [35]. The nonhomogeneous half-space is nominally partitioned
into a homogeneous half-space with the deformation modulus E and a nonhomoge-
neous half-space with the deformation modulus En ·zn . If the fraction of the entire
force, accepted by the homogeneous half-space is denoted as λ, then the fraction of
the entire force, accepted by the half-space with the deformation modulus, increas-
ing as En ·zn , will be (1−λ). At such consideration, the displacements of any point
of the nominally homogeneous half-space will be given by

w = λ · w0 . (1.183)

The displacement of the same point due to the force under consideration, corre-
sponding to the nominal half-space with the variable deformation modulus, should
then be assumed

w = (1 − λ)wn . (1.184)

Since the components, nominally comprising the half-space, should be presented

as functioning together, one should equalize the right-hand parts of Eqs. (1.183) and
(1.184), i.e. assume

λw0 = (1 − λ)wn .

After simple transformations the unknown factor is found:

⎡ ⎤
⎢ (En /E0 )(z2 Rn−2 + 2Rn (1 − ν)) ⎥
⎢ ⎥
λ = 1/ ⎢
 + 1⎥ , (1.185)
⎣ '∞ (1 − ν)(m + n + 1)m z2 z m−1 ⎦
2π Am + 2 ·
m=0 (m + n)(n + 1) R R

and then from Eq. (1.183) the sought displacement for the half-space with the defor-
mation modulus, varied according to the power law (1.181), is determined:

z2 2(1 − ν)
P(1 + ν) 3 +
R R
w=  2 n−2  . (1.186)
En z R + 2R (1 − ν)
n

 + 2π E0
'∞ (1 − ν)(m + n + 1)m z2 z m−1
Am + 2 ·
m=0 (m + n)(n + 1) R R
[email protected]
For the vertical displacements of points on the half-space surface (z = 0) Eq.
(1.186) is considerably simplified and at n > 0 is given by
70 1 Spatial Contact Models of Elastic Bases

P(1 − ν 2 ) 1
w= ·
(1.187)
π E0 r 2(1 − ν) En n
· ·r +1
β E0

Where
 
√ n+1
C0 π (1 − ν) 
2 πq
β=   · q · sin , r = x2 + y2 .
n+2 2
(n + 1)
2
According to Klein and Durayev [56], the settlements of a half-space with a
deformation modulus, increasing with depth according to the law of Eq. (1.181),
can be found, using the following formula

P(1 − ν 2 ) 1
w= ·
. (1.188)
π rE0 2(1 − ν )(1 + n)
2 En n
· ·r +1
(3 + n)(1 − ν − nν) E0

This formula is obtained, based on an assumption of a relationship between the

Poisson ratio ν and the power index n being fulfilled:

(2 + n) · ν = 1. (1.189)
In case this condition being not obeyed, Eq. (1.188) becomes theoretically unrea-
sonable. Comparison of the displacements, calculated according to Eqs. (1.187) and
(1.188), performed in [35] as well as our calculations show that the settlement val-
ues are sufficiently close, but not identical at (2+n). ν = 1. Difference between the
settlements, calculated according to these formulae depending on ν at (2+n). ν=1,
can be quite essential, e.g.
at ν = 0.27; n = 0.4–16%,
at ν= 0.42; n = 1.3–29%.
Therefore, Eq. (1.187) is more suitable for practical application, since it is not
related to the requirement of the condition of Eq. (1.189) to be fulfilled. The compar-
ative calculations performed for the nonuniformity model of Eq. (1.181) according
to Eqs. (1.172), (1.176), (1.187), and (1.188) at n = 1, 2 have shown that the most
decisive effect on the settlements is made by the deformation modulus variation rate
with depth: at fixed values of B, n, and ν the difference in the calculated settlement
values did not exceed 1.5%. Therefore, when the deformation modulus variation
law (1.181) is used, for significant simplification of the calculations the application
of Eq. (1.188) is justified, since it will fully meet the accuracy requirements for
practical application.
Plevako [102] has emphasized physical irreality of a medium with the defor-
[email protected]
mation modulus varying according to the power law (1.137) and studied the
stress-strained state of a half-space with the elasticity modulus as a function of
z-coordinate given by
1.5 Elastic Bases with the Deformation Modulus, Variable with Depth 71

E(z) = E0 (1 + k · z)b , (1.190)

and a constant Poisson ratio. The conditions, at which the common solution of the
flat, axisymmetric, and spatial problems of theory of elasticity can be expressed in
terms of harmonic functions, were found. The corresponding values of the Poisson
ratio ν which determine the particular cases of nonuniformity, enabling the common
solution of equations of theory of elasticity to be obtained, are listed in Table 1.3 for
all possible b≤10 (according to [102]).

Table 1.3 Values of ν for different magnitudes of b

b 2 3 4 5 6 7 8 8 9 10 10
ν 1/4 0 1/6 1/6 1/8 1/15 0 1/10 1/16 1/56 1/10

Note that for a soil medium, as follows from the table, the most appropriate is
the value ν = 1/4 which is “rigidly” connected to the value of b = 2.
The results, obtained in [102], were, in particular, applied to solve the problem
of the stress-strained state of the half-space with the nonuniformity function of Eq.
(1.190) under a concentrated force applied normally to the boundary surface. In the
final form the formula for the vertical displacements (b = 2, ν = 1/4) is given by

∞
kP 2λ2 ς + λ (ς + 3) + 2
w(r,z) = e−λςλ · J0 (λρ)dλ (1.191)
4π G0 (1 + ς ) 2λ2 + 6λ + 3
0

where

ρ = kr, ς = kz, r= (x − ξ )2 + (y − η)2 ,

J0 (t) is the Bessel function of the first kind of the zeroth order.
At z = 0 Eq. (1.191) produces the desired expression to determine the displace-
ments of the points of the homogeneous half-space surface

∞
kP 3λ + 2
w|z=0 = w(r,0) = λ· J0 (λp)dλ, (1.192)
4π G0 2λ2 + 6λ + 3
0
∞
3P kP 9/2 + 7λ
w(r,0) = − J0 (λρ)dλ. (1.192 )
8π G0 r 4π G0 2λ2 + 6λ + 3
0


The first term of Eq. (1.192 ) corresponds to the homogeneous half-space with
[email protected]
the same Poisson ratio ν and shear modulus G0 = E0 /2(1+ν).
In order to calculate the nonhomogeneous half-space surface settlements, in
[102] Eq. (1.192), using the known formulae [49, 53]
72 1 Spatial Contact Models of Elastic Bases

∞
1
J0 (λρ)dλ = , ρ > 0, (1.193)
ρ
0

∞
J0 (λρ) π
dλ = T0 (γρ), T0 (γρ) = H0 (γρ) − N0 (γρ), |arg γ | < π , (1.194)
λ+γ 2
0

is given by

6    

3P π 7 √ 7 √
w|z=0 = 1− ρ + 3 T0 (γ1 ρ) + − 3 T0 (γ2 ρ)
8π G0 r 3 4 4
(1.195)
1 √ 1 √
where γ1 = (3 + 3), γ2 = (3 − 3), and the Struve H0 (γ ρ) and Neumann
2 2
N0 (γ ρ) functions are calculated with the required degree of accuracy by summation
of the corresponding series.
Note that the approach to the solution of problems of theory of elasticity of non-
homogeneous media using harmonic functions has a limited practical value since the
solution of the corresponding equations encounters strong difficulties even when the
nonuniformity functions are specified; besides, it is possible only for certain values
of the elasticity parameters. Furthermore, the approach to the determination of the
surface deformation, used in [102], is not only restricted by the choice of the nonuni-
formity function, but also related to an inconvenience of carrying out engineering
calculations due to the necessity of special functions to be calculated.
According to the method proposed here, at any dependence of the deformation
modulus E = E(z), the calculation of the nonhomogeneous half-space surface set-
tlements can be reduced, according to Eqs. (1.157) and (1.166), to the calculation of
vertical displacements as follows:

⎡ ⎤
∞
(1 − ν 2 ) · P ⎣ 1 z3 ⎦
w(r) = · − (b,kR) · 5 dz (1.196)
π · E0 r R
0

where

1 3e(R̂)/E0 − 1
(b,kR) = · ; R̂ = kR. (1.197)
2 e(R̂)/E0
[email protected]
The function of the degree of nonuniformity e(R̂) for the deformation modulus
increase law (1.190) after transformations, using integration by parts, is given by
1.5 Elastic Bases with the Deformation Modulus, Variable with Depth 73

1
e(R̂) = E0 (1 + kRt)b · t2 dt =
0
 
E0 2R̂(1 + R̂) b+1 2(1 + R̂)b+3 2
= R̂2 (1 + R̂)b+1 − + − .
R̂3 (b + 1) b+2 (b + 2)(b + 3) (b + 2)(b + 3)
(1.198)

Note that the limiting values of the function  = 1 (corresponding to the law
(1.190)) are obtained, using Eq. (1.198):

3
lim 1 (b,kR) = 0, lim 1 (b,kR) = . (1.199)
kR→0 kR→∞ 2
Both limits in Eq. (1.199) have fixed values, independent of the nonuniformity
parameter b. The dependences of the 1 function on the reduced distance kR at
various values of b are illustrated by Fig. 1.20. As seen from the figure, at b > 1
the 1 function tends rather fast to its limiting value. On the contrary, at b ≤ 1 the
character of 1 variation is sufficiently smooth.

Fig. 1.20 Weight function for the power law of the deformation modulus variation with depth, Eq.
(1.180)

Figure 1.21 shows the vertical displacements of the nonhomogeneous base sur-
face at b = 2, ν = 1/4 and various values of the parameter c = kh (the value of c =
0 corresponds to the homogeneous half-space, h is the chosen linear measure). As
follows from the data presented, the surface settlements, obtained using the equa-
tions of theory of elasticity (1.195) slightly differ from the values obtained using the
approximate calculation model of Eq. (1.196): not more than by 1.5% at c = 1, and
[email protected]
by 3.2% at c = 5, what is almost unresolved graphically. Consequently, it is reason-
able to consider the proposed numerical-and-analytical approach to the calculation
of nonhomogeneous base settlements to be applicable for wide practical application
74 1 Spatial Contact Models of Elastic Bases

Fig. 1.21 Settlements of the surface of elastic bases with the power-law nonuniformity, Eq.
(1.180), due to a vertical concentrated force

of Eq. (1.190) at various values of ν and b. Being simple in realization and simul-
taneously possessing sufficient accuracy of calculation, the approach is also helpful
from the practical point of view, in particular, due to the possibility of strict account
of experimental data regarding the growth of the soil deformation modulus with
depth.
Though the considered laws of the deformation modulus variation (1.138) and
(1.139) of power type, contrary to the purely theoretical one of Eq. (1.137), result
in finite values of E on the soil base surface, still they produce an unlimited growth
of E(z) with depth. This circumstance contradicts the realistic concept of the defor-
mation modulus stabilization with the increase of z. The data of the experimental
studies show that in most cases the elastic nonuniformity is localized near the soil
base surface, being the result of the processes of weathering, thawing, chemical
effects, and a number of other geoecological phenomena.
Consider the deformation of nonhomogeneous base surfaces for the dependences
E(z), obtained by approximation of the experimental data and corresponding to the
real growth of the deformation modulus with depth from the minimal value E0 at
the base daylight surface to a certain limiting finite value E∞ at a considerable
depth (Fig. 1.17). Variation of the deformation properties between these two limiting
values of the deformation modulus can occur with different rate and is determined
in accordance with an a priori substantiated empirical dependence.
[email protected]
For empirical dependences of the deformation modulus growth with depth, as
a rule, no exact solutions of the problem of action of a concentrated force can be
obtained. Therefore, it becomes necessary to develop efficient, sufficiently universal
1.5 Elastic Bases with the Deformation Modulus, Variable with Depth 75

numerical algorithms of representation of influence functions, which do not require

large computer resources. Below, with an example of several dependences, adequate
to the experimental data, which correctly correspond to the soil densification with
depth, the possibilities of the elaborated numerical-and-analytical method of repre-
sentation of influence functions are illustrated.
Exponential laws of the deformation modulus increase
 
E(z) = E0 · 1 + α(1 − ekz ) . (1.200)

Here α = (E∞ − E0 )/E0 , k is an empirical parameter, determining the defor-

mation modulus increase rate, 0 ≤ k ≤ ∞. Note that in this case the deformation
modulus increases at large depths by factor of not higher than (1+α) in comparison
with its daylight surface value.
While constructing the influence function for an elastic half-space with defor-
mational nonuniformity of Eq. (1.200), first one calculates e(R), using Eq. (1.154),
obtaining

−kR  

α+1 e 2 2
e(R)/E0 = +α 1+ + 3 3 (e−kR − 1) . (1.201)
3 kR kR k R

Substitution of the obtained expression into Eq. (1.166) results in the following
expression for the nonhomogeneous base surface settlement

w|z=0 = w(r) =
(1 − ν 2 )P ∞ z3 dz
= 6 −kR  
 .
2π E0 0 α + 1 e 2 2 −kR
+α 1+ + 3 3 (e − 1) · R 5
3 kR kR k R
(1.202)

Direct application of the obtained expression is hardly efficient due to a consider-

able loss of accuracy in the integrand calculations and the necessity of application of
special adaptive integration schemes, requiring considerable computer resources in
order to achieve the required accuracy at the variation of the r parameter in a semi-
infinite interval 0 < r < ∞. The experience of numerical calculations has enabled an
optimal (from the point of view of labour consumption) method, enabling the sought
integral value to be obtained with a high degree of accuracy. The efficiency of this
method is achieved due to the selection of the main part of the integrand expression
and calculation of the remaining improper integral using the quadrature formulae of
[email protected]
the highest accuracy degree.
Taking into account Eq. (1.157), one can present Eq. (1.202) in the following
form:
76 1 Spatial Contact Models of Elastic Bases
⎛ ⎞
∞
(1 − ν 2 ) · P 1 z3 ⎠
w(r) = ⎝ − 2 (α,kR) · dz (1.203)
π · E0 r R5
0

where the weight function 2 (α, kR) is given by

α 3 3 
3 k R + α e−kR (k2 R2 + 2kR) + 2(e−kR − 1)
2 (α,kR) = · 3 . (1.204)
2 1+α 3 3
k R + α e−kR (k2 R2 + 2kR) + 2(e−kR − 1)
3

As one can see, at such formulation of the displacement function, the first term
in Eq. (1.203) is the classical Boussinesq solution for the homogeneous elastic half-
space, and the second (integral) term determines the effect of the soil deformation
properties nonuniformity with depth.
The integral

∞
z3
2 (α,kR) · dz (1.205)
R5
0

is calculated numerically, the knowledge of the characteristic features of behaviour

of the weight function 2 (α,kR) being used. It directly follows from Eq. (1.204)
that

3 α
lim 2 (α,kR) = 0, lim 2 (α,kR) = · .
kR→0 kR→∞ 2 1+α
The function 2 (α,kR) is plotted in Fig. 1.22 versus a dimensionless variable kR
for various values of the α parameter. With the increase of R̂= kR from 0 to ∞ the
function 2 (α,kR), being limited, monotonously increases from 0, asymptotically
approaching its limiting value in the infinity. Thus, the convergence of the integral
(1.205) increases at small R and is of the same rate as that for the integral (1.157) at
moderate and high R.
In order to estimate the improper integral of Eq. (1.205) with a guaranteed accu-
racy and minimal computation time, we convert it to a definite integral on the stan-
dard interval [−1,1], using a transformation

z−a
q= ,
z+a
i.e. we assume in Eq. (1.205)

[email protected] 1+q
z=a .
1−q
Then
1.5 Elastic Bases with the Deformation Modulus, Variable with Depth 77

Fig. 1.22 Weight functions for various laws of elastic modulus variation with depth

 
1+q
∞ ∞ 1 G a 1
z3 1−q
(α,kR) dz = G(z)dz = 2a dq = Ĝ(q)dq. (1.206)
R5 (1 − q2 )
0 0 −1 −1

The obtained integral is calculated, using the known Gauss-Legendre quadrature

formula of the highest degree of accuracy [1]

1 
m
Ĝ(q)dq ≈ Ĝ(qi ) · ci
−1 i=1

where qi and ci are the quadrature nodes and weights. The working formula to cal-
culate the integral (1.206) under consideration finally takes the form

∞ 
m
G(z)dz ≈ G(zi ) · pi (1.207)
0 i=1

where

2aci 1 + qi
pi = ; zi = a ; i = 1, . . . ,m.
(1 − qi ) 2 1 − qi
Equation (1.207) is exact, when G(z) is a linear combination of poles 1/(z+α)i ,
= 2,3,. . .,2m+1. In other cases, when G(z) is not an exact combination of poles,
[email protected]
or when G(z) contains poles of the power higher than 2m+1, as noted in [99], Eq.
(1.207) gives the best regards by the optimal choice of the (free) transformation
parameter a value. The calculations regarding the numerical evaluation of the inte-
78 1 Spatial Contact Models of Elastic Bases

gral (1.157), possessing an exact value, have shown that application of Eq. (1.157)
gives the results, practically coinciding with the exact ones, at m = 10 and a = s · r
where s = 1,3. Using of higher (m>10) number of terms of the quadrature formula
Eq. (1.207) does not result in a noticeable increase of the calculation accuracy.
Calculations of the half-space surface settlements, performed using Eq. (1.207)
with various values of the nonuniformity parameters, are shown in Fig. 1.23. As one
can see, the main effect on the decrease of the distributive ability of the base is made
by the deformation modulus difference value α = (E∞ -E0 )/E0 . With the increase of
the k parameter, the rate of variation of the deformation properties, the settlement
curves are transformed, getting steeper what also indicates the decrease of the base
distributive ability.

Fig. 1.23 Settlements of the surface of elastic bases with the exponential nonuniformity of Eq.
(1.200) type due to a vertical concentrated force

20 . Sapegin, Nikitina, and Shvedova [121] have experimentally justified the for-
mula
1
E(z) = (1.208)
Q1 + Q2 e−kz
where Q1 and Q2 are parameters, characterizing a given soil base. For further con-
sideration, the dependence of Eq. (1.208) can be conveniently presented in one of
the following forms:
[email protected] 1
E(z) =   , (1.209)
1 1 1
+ − e−kz
E∞ E0 E∞
1.5 Elastic Bases with the Deformation Modulus, Variable with Depth 79

1+α
E(z) = E0 (1.210)
1 + αe−kR
where, similarly to the above case, α = (E∞ − E0 )/E0 .
Development of the experimental data, obtained at specific engineering-and-
geological conditions, using Eq. (1.208), results in [121]

104
E(z) = MPa, (1.211)
0.445 + 1.555 · e−4.3z
or, in the notations, used above,

k = 4.3m−1 , E0 = 0.5 · 104 MPa, E∞ = 2.247 · 104 MPa, α = 3.494.

According to Eq. (1.154),

1 1
t2 dt
e(kR) = E(Rt)t dt = E0 (1 + α)
2
. (1.212)
1 + αe−kRt
0 0

The latter integral can be evaluated only numerically. The dependences e(kR)/E0
for different α are plotted in Fig. 1.24. Note that the numerical values of the inte-
gral converge very fast for all values of the kR≥0 and α parameters. The use of a
maximum of 10 nodes in the Gauss-Legendre quadrature formula has proven to be
sufficient in order to provide the calculation accuracy ε = 10−5 .

Fig. 1.24 Dimensionless nonuniformity function for the exponential law of deformation modulus
variation, Eq. (1.210)
[email protected]
For a known e(R) function, the nonhomogeneous base settlements are determined
using Eq. (1.166), which, similarly to the above case, is presented in a form, conve-
80 1 Spatial Contact Models of Elastic Bases

nient for numerical calculations:

⎛ ⎞
∞
(1 − ν 2 ) · P ⎝ 1 z3 ⎠
w(r) = − 3 (α,kR) · 5 dz , (1.213)
π · E0 r R
0
1 3e(R̂)/E0 − 1
3 (α,kR) = · ; R̂ = kR. (1.214)
2 e(R̂)/E0

Here, similarly to the above case, the limiting values are the following:

3 α
lim 3 (α,kR) = 0, lim 3 (α,kR) = · .
kR→0 kR→∞ 2 1+α
Comparison of the weight functions for the exponential laws of the deformation
modulus variation shows (Fig. 1.22) that the difference in the values of 2 and 3
occurs at kR≤10, increasing, as one should expect, with the nonuniformity param-
eter α. Using the weight function 3 in Eq. (1.213), nonhomogeneous half-space
surface settlements were calculated (Fig. 1.25). The distributive ability of the base
decreases both with the deformation modulus difference α = (E∞ − E0 )/E0 , and
with the increase of the deformation properties variation rate k. However, the set-
tlement values are much less sensitive to the nonuniformity parameters than in the
above case of the nonuniformity of exponential type (Fig. 1.23).

Fig. 1.25 Settlements of the surface of an elastic half-space with the exponential nonuniformity
of Eq. (1.210) type due to a vertical concentrated force
[email protected]
Inverse trigonometric law of the deformation modulus increase. Consider the
formulae for calculating the nonhomogeneous base surface settlements when the
1.5 Elastic Bases with the Deformation Modulus, Variable with Depth 81

deformation modulus experimental data are processed according to the following

dependence:

2
E(z) = E0 · 1 + α arctan(kz) = E0 · (1 + α̂ · arctanẑ) (1.215)
π

2
where α̂ = α ; ẑ = kz; α = (E∞ − E0 )/E0 and, still, α = (E∞ − E0 )/E0 ; k
π
is an empirical parameter, determining the deformation modulus increase rate with
depth, 0 ≤ k < ∞.
The required calculations sequentially result in:
 

E0 1 1
e(R) = 1 + α · arctan(kR) − + 3 3 ln (1 + k2 R2 ) , (1.216)
3 2kR 2 k R
w|z=0 = w(r) =
3(1 − ν 2 )P ∞ z3 dz
 
.
2π E0 0 1 + α arctan(kR) − 1 + 1
ln (1 + k R ) · R
2 2 5
2kR 2 k3 R3
(1.217)

Similarly to the above case, we present the latter expression in the form, conve-
nient for numerical calculations:
⎛ ⎞
∞
(1 − ν 2 ) · P ⎝ 1 z3 ⎠
w(r) = −  4 (α,kR) · 5 dz (1.218)
π · E0 r R
0

where

3α 2 k3 R3 arctan(kR) − k2 R2 + ln (1 + k2 R2 )
4 (α,kR) = · 3 3 .
2 2 k R + α 2 k3 R3 arctan(kR) − k2 R2 + ln (1 + k2 R2 )

Accordingly, the limiting transitions lead to

3 π
lim 4 (α,kR) = 0, lim 4 (α,kR) = · .
kR→0 kR→∞ 2 1 + απ
The 4 (α,kR) dependences are plotted in Fig. 1.22, and the plots of vertical dis-
placements of the surface of a soil base with spatial nonhomogeneity of Eq. (1.215)-
type are shown in a dimensionless form in Fig. 1.26. As seen from the calculation
results, the displacements according to Eq. (1.218) are in a qualitative agreement
[email protected]
with the predictions of the above considered contact model (1.203), and the quanti-
tative differences are related to the character of the deformation modulus variation
with the base depth.
82 1 Spatial Contact Models of Elastic Bases

Fig. 1.26 Settlements of the surface of an elastic half-space with the inverse trigonometric nonuni-
formity of Eq. (1.215) type due to a vertical concentrated force

Thus, a numerical-and-analytical method for calculation of settlements of a non-

homogeneous half-space settlements is developed for different laws of the defor-
mation modulus increase from the value E0 = 0 at the daylight surface to the final
limit E∞ at considerable depths. Improper integrals in the calculation formulae are
determined on the basis of an efficient procedure of numerical integration, optimal
from the point of view of computation time due to the application of quadratures
of the highest degree of accuracy. In the particular cases of linear and parabolic
laws of the deformation modulus increase with depth, analytical expressions for the
influence functions are obtained. The analysis of a large series of the calculations
performed has shown the soil base surface settlements to be essentially affected by
the nonuniformity of their elastic properties. The proposed influence functions are
applicable for the description of the soil functioning in the course of condensation,
setting much lower distributive ability of the nonhomogeneous base than for the
homogeneous half-space model. The empirical parameters of the contact models
under consideration, characterizing the base nonhomogeneity, can be determined
from processing the data of standard soil tests in a well at different depths, or from
punch tests, using punches of different bearing area.
At the end of this chapter we should note that the developed method of con-
struction of influence function for bases with the deformation modulus, increasing
with depth, results (in the framework of the boundary-element method) to effec-
[email protected]
tive numerical algorithms of solution of spatial contact problems of soil mechanics
for shallow foundations. As shown below (Sect. 4.2), one can take into account the
effect of the soil nonhomogeneity degree on the contact interaction of bases and
foundations, especially with big dimensions.
References 83

References
1. Abramowitz M, Stegun I A (1972) Handbook of mathematical functions with formulas,
graphs, and mathematical tables, 10th edn. Dover, New York
2. Aizikovich S M, Aleksandrov V M (1984) Axially symmetrical problem of impression
of a round punch into an elastic half-space, nonhomogeneous in depth. Mekh Tverd Tela
(issue 2): 73–82 (in Russian)
3. Aleksandrov A Ya (1963) On the solution of certain contact problems of the theory of elas-
ticity. J Appl Math Mech 27: 1490–1494
4. Aleksandrov A Ya, Solovyov Yu I (1978) Spatial problems of theory of elas-ticity (applica-
tion of methods of theory of complex-variable functions). Nauka, Moscow (in Russian)
5. Aleksandrov V M, Pozharskii D A (1989) Action of a strip-shaped stamp on an elastic
incompressible three-dimensional wedge. Intern Appl Mech 25: 749–756
6. Alexandrov V M, Pozharskii D A (2001) Three-dimensional contact problems. Kluwer,
Amsterdam
7. Alexandrov V M, Romalis B L (1986) Contact problems in machine building. Mashinos-
troyenie, Moscow (in Russian)
8. Aleynikov S M, Ikonin S V (1988) Numerical integration of singular solutions of theory
of elasticity applied to the calculation of soil bases. In: Applied problems of me-chanics of
solid media. Voronezh State University, Voronezh, pp. 132–136 (in Russian)
9. Aleynikov S M, Ikonin S V (1990) Three-dimensional deformation of the surface of a
tapered elastic layer. Soil Mech Found Eng 27: 218–222
10. Aleynikov S M, Ikonin S V (1996) Numerical modeling of contact interac-tion of bases with
dredging wells of bridge piers. In: Geotechnical engineering for transporta-tion infrastruc-
ture. Proc 12th Europe Conf Soil Mech Found Eng (Amsterdam, 7–10 June 1999), vol. 3.
Theory and practice, planning and design, construction and maintenance. Balkema, Rotter-
dam, pp. 1719–1724
11. Aleynikov S M, Nikolaevskiy V N, Ramazanov T K (1994) Migration the stresses for two-
level tectonic models. In: Dynamic processes in geophysical medium. Nauka, Moscow, pp.
11–21 (in Russian)
12. Aleynikov S M, Sedaev A A (1990) Calculation of ring-shaped foundations at complicated
geotechnical conditions. In: Calculation and design of bases and foundations at complex
geotechnical conditions. VPI, Voronezh, pp. 19–26 (in Russian)
13. Awojobi A O (1976) On the hyperbolic variation of elastic modulus in a non-homogeneous
medium. Intern J Solids Struct 12: 81–92
14. Barvashov V A (1976) Composite models of an earth foundation bed. Soil Mech Found Eng
13: 61–65
15. Belenkiy M Ya (1952) Mixed problem of theory of elasticity for an infinitely long strip. Prikl
Mat Mekh 16: 283–292 (in Russian)
16. Belik G I, Rvachev V L (1962) On the main integral equation of contact problem of theory
of elasticity for a half-space whose elasticity modulus is a power function of depth. Dokl
AN UkrSSR (issue 8): 1041–1044 (in Russian)
17. Bezukhov N I (1968) Fundamentals of theory of elasticity, plasticity and creep. Vysshaya
Shkola, Moscow (in Russian)
18. Birman S Ye (1953) On the settlement of a rigid punch on an elastic layer resting on an
incompressible base. Dokl AN SSSR 93(issue 5): 791–794 (in Russian)
19. Bochner S (1959) Lectures on Fourier integrals. Princeton University Press, Princeton, NJ
20. Bolotin V V, Novichkov Yu N (1980) Mechanics of multilayer structures. Mashinostroyenie,
Moscow (in Russian)
21. Booker J R, Balaam N P, Davis E H (1985) The behaviour of an elastic non-homogeneous
[email protected]
half-space. Part 1. Line and point loads. Intern J Num Anal Method Geomech 9: 353–367
22. Booker J R, Balaam N P, Davis E H (1985) The behaviour of an elastic non-homogeneous
half-space. Part 2. Circular and strip footings Intern J Num Anal Method Ge-omech 9:
369–381
84 1 Spatial Contact Models of Elastic Bases

23. Borodachev A N (1984) Pressure of an elliptical punch on a nonhomogene-ous elastic half-

space. Dokl AN UkrSSR Ser A (issue 7): 30–33 (in Russian)
24. Borodachev A N (1985) Kernel matrix for an elastic half-space with a vari-able Poisson
ratio. Dokl AN UkrSSR Ser A (issue 12): 21–23 (in Russian)
25. Borodachev A N (1988) Elastic equilibrium in a layer inhomogeneous with depth. Intern
Appl Mech 24: 753–758
26. Borodachev A N, Dudinskii V I (1986) Contact problem for an elastic half-space with a
variable Poisson ratio. Mech Solids 21(issue 1): 86–91
27. Bosakov S V (1988) Action of a concentrated force on an elastic quarter-space. Teor i Prikl
Mekh. Vysheyshaya Shkola, Minsk 15: 100–108 (in Russian)
28. Bosakov S V, Fomicheva N M (1986) Displacements of a spatial wedge boundary due to
a normally concentrated force at arbitrary Poisson ratio. Vopr Stroit i Arkhit. Vysheyshaya
Shkola, Minsk 15: 53–56 (in Russian)
29. Brebbia C, Telles J, Wrobel L (1984) Boundary element techniques. Springer-Verlag, Berlin
30. Bronshtein M I (1970) On the calculation of a base underlaid by an inclined incompressible
layer and loaded by a concentrated force or a punch. In: Bases, foundations and underground
structures. Tr NIIOSP 60: 40–82 (in Russian)
31. Burmister D M (1956) Stress and displacement characteristics of a two-layer rigid base soil
system: influence diagrams and practical applications. Proc Highway Res 35: 773–814
32. Chan K S, Karasudhi P, Lee S L (1974) Force at a point in the interior of a layered elastic
half-space. Intern J Solids Struct 10: 1179–1199
33. Davies T G, Banerjee P K (1978) The displacement field due to a point load at the inter-face
of a two layer elastic half-space. Geotechnique 28: 43–56
34. Dempsey J P, Li H (1989) A rigid rectangular footing on an elastic layer. Geotechnique 9:
147–152
35. Durayev A Ye (1991) Calculation of structures on a soil base with a defor-mation modulud
increasing with depth. Mordovian University, Saransk (in Russian)
36. Dymek F, Dymek J (1988) Funkcje Greena gla polprzestrzeni sprezystej transwersalnie
izotropowej. Gornictwo 12: 211–229
37. Egorov K E (1958) On the deformation of a finite-thickness base. In: Bases and foundations.
Soil Mech Tr NIIOSP 34: 5–33 (in Russian)
38. Egorov K E, Barvashov V A, Fedorovskii V G (1973) On the application of theory of elas-
ticity to the calculation of bases under structures. In: Proc 8th Intern Conf Soil Mech Found
Eng, Stroyizdat, Moscow, pp. 72–83 (in Russian)
39. Fedorovskii V G, Onopa I A (1985) Surface settlements of a wedge-shape bed under a
concentrated load. Soil Mech Found Eng 22: 79–84
40. Florin V A (1959) Fundamentals of soil mechanics. Stroyizdat, Mos-cow/Leningrad (in Rus-
sian)
41. Galin L A (ed) (1976) Development of theory of contact problems in the USSR, Nauka,
Moscow (in Russian)
42. Gibson R E (1967) Some results concerning displacements and stresses on a non-
homogeneous elastic half-space. Geotechnique 17: 58–67
43. Gibson R E, Sills G C (1969) On the loaded elastic half-space with a depth varying Poisson’s
ratio. ZAMP 20: 691–695
44. Godes Yu Ya (1987) Functions of compliance of a multilayer base with elas-tic constraints
between the layers. In: Nonlinear problems of hydroaeromechanics and theory of elasticity.
Dnipropetrovsk State University, Dnipropetrovsk, pp. 92–97 (in Russian)
45. Godes Yu Ya, Privarnikov A K (1983) Quadrature formulae for Hankel inte-grals.
Dnipropetrovsk State University, Dnipropetrovsk, VINITI No. 327-84Dep (in Russian)
[email protected]
46. Golecki J J, Knops R J (1969) Introduction to a linear elasto-statics with variable Poisson’s
ratio. Acad Gorn-Hutn w Krakowie 30: 81–92
47. Gorbunov-Posadov M I, Malikova T A, Solomin V I (1984) Calculation of structures on an
elastic base. Stroyizdat, Moscow (in Russian)
References 85

48. Gorlov A M, Serebryaniy R V (1968) Automated calculation of rectangular plates on an

elastic base. Stroyizdat, Moscow (in Russian)
49. Gradshtein I S, Ryzhyk I M (1980) Tables of integrals, series and products. Academic Press,
New York
50. Hahn H G (1985) Elastizitatstheorie. Teubner, Stuttgart
51. Hanson M T, Xu Y, Keer L M (1994) Stress analysis for a tree-dimensional incompressible
wedge under body force or surface loading. Q J Mech Appl Math 47: 141–158
52. Hetenyi M A (1970) A general solution for the elastic quarter space. Trans ASME Ser E J
Appl Mech 37: 70–76
53. Janke E, Emde F, Lösch F (1966) Tafeln Höherer Funk-tionen. Teubner, Stuttgart
54. Johnson K L (1985) Contact mechanics. Cambridge University Press, Cam-bridge
55. Keer L M, Lee J C, Mura T (1984) Hetenyi’s elastic quarter space problem revisited. Intern
J Solids Struct 20: 513–524
56. Klein G K, Durayev A Ye (1971) Account of increase off the soil deforma-tion modulus
with depth at the calculation of beams on a solid base. Gidroteh Stroit (issue 7): 19–21 (in
Russian)
57. Kogan B M, Zinchenko V D (1960) Stressed state of a nonhomogeneous layer, resting on an
elastic half-space. Izv Vuzov Stroit Arkhit (issue 3): 8–18 (in Russian)
58. Kolchin G B, Faverman E A (1972) Theory of elasticity of nonhomogeneous bodies. Bibli-
ographical index. Stiinţa, Chişinău (in Russian)
59. Kolchin G B, Faverman E A (1977) Theory of elasticity of nonhomogeneous bodies. Bibli-
ographical index (1970–1973). Stiinţa, Chişinău (in Russian)
60. Koning H (1957) Stress distribution in a homogeneous, anisotropic, elastic semi-infinite
solid. In: Proc 4th Intern Conf Soil Mech Found Eng, Butterworths, London, pp. 335–338
61. Kopasenko V Yu, Kebedev V K (1977) Mixed problem of theory of elastic-ity for a quarter-
space. Izv SKNC VSH. Nat Sci (issue 3) 30–31 (in Russian)
62. Korenev B G (1957) Punch, resting on an elastic half-space whose deforma-tion modulus is
a power function of depth. Dokl AN SSSR 112: 823–826 (in Russian)
63. Korenev B G (1971) Introduction into Bessel function theory. Nauka, Mos-cow (in Russian)
64. Kupradze V D, Gegeliya T G, Basheleyshvili M O, Burchuladze T V (1979) Three-
dimensional problems of the mathematical theory of elasticity and thermoelasticity. North-
Holland, Amsterdam
65. Kushner S G (1990) Calculation of settlements of bases under buildings and structures. Budi-
velnyk, Kyiv (in Russian)
66. Lebedev N N, Ufliand Ia S (1958) Axisymmetric contact problem for an elastic layer. J Appl
Math Mech 22: 442–450
67. Lomakin V A (1976) Theory of elasticity of nonhomogeneous bodies. Mos-cow State Uni-
versity, Moscow (in Russian)
68. Love A E H (1959) A treatise on the mathematical theory of elasticity, 4th edn. Cambridge
University Press, Cambridge
69. Lubyagin I A, Pozharskii D A, Chebakov M I (1991) Generalization of Boussinesq and
Cerruti problems for the case of an elastic spatial wedge. Dokl AN SSSR 321(issue 1): 58–
62 (in Russian)
70. Lubyagin I A, Pozharskii D A, Chebakov M I (1992) Embedding of a punch in the form of
an elliptic paraboloid into an elastic spatial wedge. J Appl Math Mech 56: 244–252
71. Lur’e A I (1964) Three-dimensional problems in the theory of elas-ticity. Interscience, New
York
72. Lur’e A I (1970) Theory of elasticity. Nauka, Moscow (in Russian)
[email protected]
73. Lur’e A I, Brachkovskii B S (1941) Solution of a flat problem of theory of elasticity for a
wedge. Works Leningrad Polytechnic Institute 3: 158–165 (in Russian)
74. Makovenko S Ya (1988) The influence tensor for an elastic medium with Poisson’s ratio
varying in one direction. J Appl Math Mech 52: 263–266
86 1 Spatial Contact Models of Elastic Bases

75. Martynenko M D, Dashkevich A A (1985) Some problems of bending of round plates and
slabs on a nonhomogeneous half-space. In: Boundary problems and automa-tion of their
solution. KhAI, Kharkiv, pp. 168–170 (in Russian)
76. Martynenko M D, Dashkevich A A (1989) Bending of slans on a nonho-mogeneous base
of exponential type with the account of their thickness deformability. Probl Mashinostr 32:
57–60 (in Russian)
77. Mindlin R D (1936) Force at a point in the interior of a semi-infinite solid. Physics 7: 195–
202
78. Mindlin R D (1953) Force in the interior of a semi infinite solid. In: Proc First Midwestern
Conf Solid Mech Univ Illinois, Urbana, pp. 111–118
79. Mindlin R D, Cheng D H (1950) Nuclei of strain in the semi-infinite solid. J Appl Phys 21:
926–931
80. Minoru H, Koji K, Jun K (1984) A numerical method of Hankel transform for axisymmetric
problems of elasticity. Bull JSME 27: 1333–1338
81. Mossakovskii V I (1958) Pressure of a circular die on an elastic half-space, whose modulus
of elasticity is an exponential function of depth. J Appl Math Mech 22: 168–171
82. Mozharovskii V V, Starzhinskii V Ye (1988) Applied mechanics of layered bodies of com-
posites: Flat contact problems. Nauka i tekhnika, Minsk (in Russian)
83. Muki R (1960) Asymmetric problems of the theory of elasticity for a semi-infinite solid and
a thick plate. In: Sneddon I N and Hill R (eds) Progress in solid mechanics. North-Holland,
Amsterdam, vol. 1, pp. 399–439
84. Muravskii G (1997) Green functions for a compressible linearly non-homogeneous half-
space. Archive Appl Mech 67: 521–534
85. Muravskii G (1997) Time-harmonic problem for a non-homogeneous half-space with expo-
nentially varying shear modulus. Intern J Solids Struct 34: 3119–3139
86. Muravskii G (1997) Time-harmonic problem for a non-homogeneous half-space with shear
modulus limited at infinite depth. Eur J Mech A/Solids 16: 277–294
87. Naumov Yu A, Shevlyakov Yu A (1967) On the bending of round plates on a multilayer
base. Inzh Zh Mekh Tverd Tela (issue 1): 154–167 (in Russian)
88. Naumov Yu A, Shevlyakov Yu A, Chistyak V I (1970) On solving the fun-damental prob-
lems of the theory of elasticity for a layer with an arbitrary nonhomogeneity along its thick-
ness. Intern Appl Mech 6: 406–711
89. Nikishin V S, Shapiro G S (1970) Spatial problems of theory of elasticity for multilayer
media. VC AN SSSR, Moscow (in Russian)
90. Nikishin V S, Shapiro G S (1973) Problems of theory of elasticity for multi-layer media.
Naula, Moscow (in Russian)
91. Novotný B, Hanuška A (1983) Teória vrstev-natého polpriestoru [Theory of layered half-
space]. Veda, Bratislava (in Slo-vak)
92. Nowacki W (1970) Teoria sprezystosci. PWN, Warszawa
93. Olszak W (ed) (1958) Non-homogeneity in elasticity and plasticity. In: Proc IUTAM Symp
(Warsaw, September 2–9 1958). Pergamon Press, Lon-don
94. Orlyuk Ye I (1979) Functional equations of the spatial problem of theory of elasticity for a
wedge and their solution. Dokl AN USSR Ser A (issue 3): 194–198 (in Russian)
95. Parton V Z, Perlin P I (1981) Methods of mathematical theory of elasticity. Nauka, Moscow
(in Russian)
96. Petrishin V I (1965) On the torsion of a multilayer base. Prikl Mekh 1(issue 6): 127–129 (in
Russian)
97. Petrishin V I, Privarnikov A K (1965) Main boundary problems of theory of elasticity for
multilayer bases. Prikl Mekh 1(issue 4): 58–66 (in Russian)
[email protected]
98. Piskunov V G, Prisyazhnyuk V K (1985) Calculation of nonhomogeneous plates on a non-
homogeneous half-space. Stroit Mekh Raschet Sooruzh (issue 1): 25–28 (in Russian)
99. Pissanetzky S (1983) An infinite element and a formula for numerical quad-rature over an
infinite interval. Intern J Num Method Eng 19: 913–927
References 87

100. Plevako V P (1969) A point force inside a pair of cohering half-spaces. Soil Mech Found
Eng 6: 165–169
101. Plevako V P (1971) On the theory of elasticity of inhomogeneous media. J Appl Math Mech
35: 806–813
102. Plevako V P (1972) On the possibility of using harmonic functions for solv-ing problems of
the theory of elasticity of nonhomogeneous media. J Appl Math Mech 36: 834–842
103. Plevako V P (1972) The state of stress of a nonhomogeneous layer resting on an elastic
half-space. Intern Appl Mech 8: 397–403
104. Plevako V P (1973) The deformation of a nonhomogeneous half-space un-der the action of
a surface load. Intern Appl Mech 9: 593–598
105. Pobedrya B Ye (1981) Numerical methods in theory of elasticity and plas-ticity. Moscow
State University, Moscow (in Russian)
106. Pobedrya B Ye (1984) Mechanics of composite materials. Moscow State University,
Moscow (in Russian)
107. Popov G Ya (1982) Concentration of elastic stress near punches, cuts, fine inclusions and
supports. Nauka, Moscow (in Russian)
108. Popov G Ya (1982) Contact problems for a linearly deformable base. Vyshcha Shkola,
Kyiv/Odesa (in Russian)
109. Potelezhko V P, Filippov A P (1967) The contact problem for a slab lying on an elastic base.
Intern Appl Mech 3(issue 1): 49–51
110. Poulos H G, Davis E H (1974) Elastic solutions for soil and rock mechan-ics. Wiley, New
York
111. Privarnikov A K (1973) Spatial deformation of a multilayer base. In: Stabil-ity and strength
of structural elements. Dnipropetrovsk State University, Dnipropetrovsk, pp. 27–45 (in
Russian)
112. Privarnikov A K, Godes Yu A (1986) On the solution of the first boundary problem for an
elastic multilayer base. In: Stability and strength of structural elements. Dni-propetrovsk
State University, Dnipropetrovsk, pp. 6–28 (in Russian)
113. Rakov A K, Rvachev V L (1961) Contact problem of theory of elasticity for a half-space
with an elasticity module being a power function of depth. Dokl AN UkrSSR (is-sue 3):
286–290 (in Russian)
114. Rao C R A (1970) On the integration of the axisymmetric stress equations of motion for
nonhomogeneous elastic media. Arch Mech Stosow 22: 63–73
115. Recomendations on the calculation of settlements and slopes for rectangular foundations on
a wedge-shaped base (1985) Gersevanov NIIOSP, Moscow (in Russian)
116. Reut V V, Tikhonenko L Ia (1980) Bending of wedge-like plates with elas-tically-fastened
or reinforced edges. J Appl Math Mech 44: 104–110
117. Romanchik V S, Knyazeva L P (1983) Stress-strained state of a cubic ani-sotropic space.
Teoret i Prikl Mekh. Vysheyshaya Shkola, Minsk 10: 41–47 (in Russian)
118. Rostovtsev N A (1964) On the theory of elasticity of a nonhomoqeneous medium. J Appl
Math Mech 28: 745–757
119. Rostovtsev N A, Khranevskaia I E (1971) The solution of the boussinesq problem for a half-
space whose modulus of elasticity is a power function of the depth. J Appl Math Mech 35:
1000–1009 (in Russian)
120. Rvachev V L, Protsenko V S (1977) Contact problems of theory of elastic-ity for nonclassi-
cal domains. Naukova Dumka, Kyiv (in Russian)
121. Sapegin D D, Nikitin A A, Shvedova T K (1976) Study of deformational properties of rock
bases with the account of their deformability in depth. Izv VNIIG Ve-deneyeva 111: 67–73
(in Russian)
122. Sarkisyan V S, Avetikyan V K (1987) On the solution of a contact problem for an elastic
[email protected]
wedge reinforced by a finite and a semifinite stringers on its boundary. In: Me-chanics.
Yerevan State University, Yerevan (issue 6), pp. 65–70 (in Russian)
123. Schclar N A (1994) Anisotropic using boundary elements. Computational Mechanics Publi-
cations, Southampton
88 1 Spatial Contact Models of Elastic Bases

124. Selvadurai A P S (1979) Elastic analysis of soil-foundations interaction. El-sevier,

Amsterdam
125. Selvadurai A P S (1995) On the indentation of a non-homogeneous elastic geomaterial:
Analytical and computational estimates. In: Pande & Pitrusczak (eds) Numerical Models in
Geomechanics-NUMOG V: Proc. 5th Intern Symp (Davos, Switzer-land). Balkema, Rotter-
dam, pp. 381–389
126. Selvadurai A P S (1996) The settlement of a rigid circular foundation rest-ing on a half-space
exhibiting a near surface elastic non-homogeneity. Intern J Num Anal Method Geomech 20:
251–364
127. Selvadurai A P S, Lan Q (1998) Axisymmetric mixed boundary value prob-lems for an
elastic halfspace with a periodic non-homogeneity. Intern J Solids Struct 35: 1813–1826
128. Shandru N (1964) On internal point loads in an elastic half-space with fixed edges. Bull
Math Soc Sci Math Phys RPR 5: 205–224
129. Shekhter O Ya (1965) A horizontal force in a three-dimensional layer on a rigid base. Soil
Mech Found Eng 2: 11–14
130. Shekhter O Ya (1965) A point force applied inside an elastic wedge. Soil Mech Found Eng
2: 334–337
131. Shekhter O Ya (1970) On the solutions of problems for an infinite elastic wedge with loads
inside. In: Bases, foundations and underground structures. Tr NIIOSP 59: 22–30 (in Russian)
132. Shekhter O Ya, Prikhodchenko O E (1964) Stress and displacement distri-butions in an elas-
tic layer acted on by internal point forces. Soil Mech Found Eng 1: 275–279
133. Shekhter O Ya, Bronshtein M I (1970) Some practical applications of prob-lems for an infi-
nite elastic wedge. In: Bases, foundations and underground structures. Tr NIIOSP 60: 12–40
(in Russian)
134. Shekhter O Ya, Prikhodchenko O E (1966) Distribution of stresses and dis-placements in an
elastic wedge with a free surface, resting on a rigid base, with concentrated forces acting
inside. In: Bases and foundations. Tr NIIOSP 56: 12–35 (in Russian)
135. Sheremet V D (1982) Green’s displacement tensor for an elastic half-space with a fixed
boundary plane. In: Numerical analysis in the problems of mechanics. Stiinta, Chisinau 70:
139–143 (in Russian)
136. Sheremet V D (1983) A new solution and generalization of the Mindlin’s problem for an
elastic half-space with a fixed boundary plane. In: Mathematical studies: Numerical analysis
in the problems of mechanics. Stiinta, Chisinau 75: 134–140 (in Russian)
137. Shevlyakov Yu A (1977) Matrix algorithms in theory of elasticity of non-homogeneous
media. Vyshcha Shkola, Kyiv/Odesa (in Russian)
138. Shevlyakov Yu A, Naumov Yu A, Chistyak V I (1968) On calculating in-homogeneous foun-
dations. Intern Appl Mech 4(issue 9): 42–45
139. Shirinkulov T Sh (1972) Calculation of engineering structures on a nonho-mogeneous elastic
base. FAN, Tashkent (in Russian)
140. Shvets V B, Ginzburg L L, Goldshtein V M et al. (1987) Handbook on me-chanics and
dynamics of soils. Budivelnyk, Kyiv (in Russian)
141. Sneddon I (1995) Fourier transforms. Dover, New York
142. Snitko N K (1980) On the action of a concentrated force on a nonhomoge-neous elastic
half-space. Stroit Mekh Raschet Sooruzh (issue 2): 76–78 (in Russian)
143. Solomin V I, Shmatkov S B (1986) Methods of calculation and optimal de-sign of ferrocon-
crete foundation structures. Stroyizdat, Moscow (in Russian)
144. Sorochan E A, Trofimenkov Yu G (ed.) (1985) Bases, foundations and un-derground struc-
tures: Designer’s handbook. Stroyizdat, Moscow (in Russian)
[email protected]
145. Sveklo V A (1964) Boussinesq type problems for the anisotropic half-space. J Appl Math
Mech 28: 1099–1105
146. Timoshenko S, Wojnowski-Krieger S (1959) Theory of plates and shells. McGraw Hill, New
York
References 89

147. Ufliand Ia S (1972) Some spatial problems of theory of elasticity for a wedge. In: Mechan-
ics of a solid medium and related problems of analysis. Nauka, Moscow, pp. 549–554 (in
Russian)
148. Ulitko A F (1979) Method of vector eigenfunctions in spatial problems of theory of elasticity.
Naukova Dumka, Kyiv (in Russian)
149. Vasilyev Yu N (1965) Problem of action of a concentrated force applied to the boundary of
coupled half-spaces. Stroit Mekh Raschet Sooruzh (issue 4): 15–16 (in Russian)
150. Vilkov I M (1963) Flat contact problem for a two-layer base under a sym-metrical load
acting on a rigid punch. Izv AN SSSR Mekh Mashinostr (issue 4): 172–174 (in Russian)
151. Vladinirov V S (1967) Equations of mathematical physics. Nauka, Moscow (in Russian)
152. Vlasov V Z, Leontiev N N (1960) Beams, plates and shells on an elastic base. Phys Math
Liter Publ, Moscow (in Russian)
153. Vorovich I I, Aleksandrov V M, Babeshko V A (1972) Nonclassical mixed problems of
theory of elasticity. Nauka, Moscow (in Russian)
154. Vorovich I I, Ustinov Yu A (1959) Pressure of a die on an elastic layer of finite thickness. J
Appl Math Mech 23: 637–650
155. Vrettos C (1991) Time-harmonic Boussinesq problem for a continuously non-homogeneous
soil. Earthquake Eng Struct Dynam 20: 961–977
156. Yefimov A B, Yefimov D G (1966) Concentrated effects on an elastic in-compressible
wedge. Izv AN SSSR Mekh Tverd Tela (issue 6): 89–92 (in Russian)
157. Yevdokimov P D, Sapegin D D (1964) Strength, resistance to shear, and de-formability of
bases of buildings on rock. Energiya, Moscow (in Russian)

[email protected]
[email protected]
Chapter 2
Static Analysis of Contact Problems
for an Elastic Half-Space

Abstract The second chapter is devoted to the mathematical formulation of mixed

problems of the elasticity theory for a half-space and to the numerical-and-analytical
methods of their solution. The results obtained in this chapter on developing the
mathematical means are the reference data for BEM-based numerical modeling of
the spatial contact interaction. The integrated boundary equations of the spatial con-
tact problem are written for the case when the calculation scheme is accepted in
the form of variously deepened punches undergoing the action of the spatial sys-
tem of forces. It is shown how to reduce the initial integral equation system of the
contact problem with respect to the contact stress function and the punch displace-
ment parameters to the appropriate finite-dimensional algebraic analogue. Much
attention is paid to calculating the matrix coefficients of the resolving system of
algebraic equations. A numerical-and-analytical procedure is given for integrating
Mindlin’s fundamental solutions over flat triangular and quadrangular boundary ele-
ments, arbitrary oriented in the half-space. For convenience, to apply the developed
approach in practical calculations, the boundary integral equations of the spatial
contact problems for a number of essential special cases are presented. The contact
problems at axial loading and torsion of absolutely rigid rotation bodies deepened
into the half-space, are considered. Boundary-element formulations of the contact
problems for complex-shaped punches with flat and smooth bases (shallow foun-
dations), situated on spatially nonhomogeneous bases of the semi-infinite elastic
massif type are presented.

2.1 Boundary Integral Equations of the Contact Problem

for an Absolutely Rigid Punch, Deepened into an Elastic
Half-Space, Under a Spatial Load System
Consider an elastic homogeneous half-space z ≥ 0, containing a cavity Swith a
[email protected]
boundary  from the side of the surface z = 0. The mechanical properties of the
half-space are determined by the elastic modulus E and Poisson’s ratio ν. We
assume the surface z = 0, being the boundary of the half-space, to be free from any

S.M. Aleynikov, Spatial Contact Problems in Geotechnics, Foundations of 91

Engineering Mechanics, DOI 10.1007/b11479_2,  C Springer-Verlag Berlin Heidelberg 2011
92 2 Static Analysis of Contact Problems for an Elastic Half-Space

Fig. 2.1 Calculation scheme

for the contact problem of a
volumetric punch, deepened
into an elastic half-space

load. In the cavity S an absolutely rigid volumetric punch is deepened, subject to a

static load, reduced to a resultant force P = {P1 , P2 , P3 } and a resultant moment
M = {M1 , M2 , M3 } where Pi , Mi (i = 1, 2, 3) are the projections of the corre-
sponding vectors onto the axes of the Cartesian coordinate system OX1 X2 X3 (OXYZ)
(Fig. 2.1). The contact problem of spatial theory of elasticity for the deepened punch
consists in the determination of contact stress on the surface of interaction of the
elastic medium with the punch as well as the determination of the parameters of its
displacement as a rigid solid. We assume the punch to be welded with the elastic
half-space, i.e. on the contact surface of the punch and the base the displacements
coincide (the second-order boundary conditions according to Galin [14] are ful-
filled). In order to derive the main equations of the contact problem we follow a
rather demonstrative method, first considered by Kovneristov [20] and later applied
by Shishov [29] while solving the problems in an axisymmetric arrangement. The
method suggests the involvement of Betti’s theorem of reciprocity [24] that requires
the concept of basic and auxiliary states of an elastic body to be introduced.
As a basic state we consider an elastic half-space with a cavity S, in each point of
whose surface the displacements ui (N) are given and the stresses Pj (N) are acting,
being a distributed reaction from the side of the punch (Fig. 2.1). The stress-strained
state of the base in the initial contact problem will be equivalent to the basic state
introduced into the consideration.
In order to build auxiliary states consider a solid elastic half-space, loaded in a
point K (ξ,η,ζ) by unit concentrated forces

Aj = δ(x − ξ , y − η, z − ζ ), i = 1,2,3 (2.1)

directed along the coordinate axes, respectively. The load point K (ξ,η,ζ) is chosen
[email protected]
outside the domain bounded by the surface . Virtually remove the elastic body
S from the half-space, the surface  of the body S being identical to that of the
rigid punch. In order to keep the half-space, weakened by the cavity, in equilibrium,
2.1 Boundary Integral Equations of the Contact Problem for an Absolutely Rigid Punch 93

Fig. 2.2 Elastic half-space,

weakened by a cavity,
corresponding to the
deepened punch shape

forces σij (N, K) and displacements Uij (N, K), being the fundamental Mindlin’s solu-
tions, should be distributed over the surface  (Fig. 2.2).
We take the advantage of the Betti’s theorem of reciprocity [24], linking the
solution of two different problems for the same domain of an elastic body: the work
of the system of forces of the basic state on the displacements of the auxiliary state
is equal to the work, performed by the system of forces of the auxiliary state on the
displacements of the basic state. The equations of reciprocity of the works for the
basic and the auxiliary states considered in this contact problem, are given by


p1 (N)U1i (K, N) + p2 (N)U2i (K, N) + p3 (N)U3i (K, N) d =
 
(2.2)
= [σ1i (K, N)u1 (N) + σ2i (K, N)u2 (N) + σ3i (K, N)u3 (N)] d + ui (K),

i = 1,2,3.
Equation (2.2) gives the integral representation of displacements at any point
(outside the punch) of the elastic half-space and is known as Somigliana identity for
the displacements [10, 24]. This equation could be immediately used formally as an
initial one. Note that Eq. (2.2) lacks the integrity over the half-space surface, since
the absence of stress at the free surface of the elastic half-space in the basic state
had been initially assumed, and the fundamental Mindlin equation was obtained
under the same condition. The Somigliana equation explains the main advantage
of the boundary integral equation method (and the boundary-element method as a
method of its numerical implementation), consisting in the fact the displacement
vector components (and, consequently, the stresses) are determined solely by the
boundary data at the punch surface. In other words, if the displacement values ui
[email protected]
and forces pj at the  boundary are known, then using the Somigliana identity (2.2)
one can always find the displacements and, consequently, deformations and stresses
at any internal point K (ξ, η, ζ) of the elastic half-space.
94 2 Static Analysis of Contact Problems for an Elastic Half-Space

The deformations of the half-space can be determined using Eq. (2.2) in a con-
ventional way after differentiating and using the Cauchy relations

∂u1 ∂u2 ∂u3

ε11 = (K), ε22 = (K), ε33 = (K),
∂ξ ∂η ∂ζ
   
1 ∂u1 ∂u2 1 ∂u2 ∂u3
ε12 = ε21 = (K) + (K) , ε23 = ε32 = (K) + (K) ,
2 ∂η ∂ξ 2 ∂ζ ∂η
 
1 ∂u1 ∂u3
ε13 = ε31 = (K) + (K) ,
2 ∂ζ ∂ξ
which afterwards enables the stress tensor components in the elastic half-space for
the basic state to be determined using Hooke equations

ν
σk (K) = 2G θ (K) + εk (K) , k = 1,2,3,
1 − 2ν

τ12 = 2Gε12 (K), τ23 = 2Gε23 (K), τ13 = 2Gε13 (K)

where G = E/2(1+ν) is the shear modulus, θ (K) = ε11 +ε22 +ε33 is dilatation. The
obtained equations are cumbersome and, therefore, not given here in the extended
form.
In order to obtain the equations of the contact problem, we direct the point
K (ξ,η,ζ) of application of the unit concentrated forces toward the deepened punch
surface, i.e. perform a limiting transition from the internal point to the boundary
one. The limiting transition results in a system of three boundary integral equations
⎡ ⎤
 3
1 ⎣
ui (K) = (pj (N)Uji (K, N) − uj (N)σji (K, N))⎦ d, (2.3)
2
 j=1

i, j = 1, 2, 3; K(ξ ,η,ζ ) ∈ , N(x1 , x2 , x3 ) ∈ .

The factor 1/2 on the left-hand side of Eq. (2.3) arises due to the fact the unit
forces in the point K (ξ,η,ζ) in the auxiliary state are divided by the surface  in
two parts: one acts at the half-space with the cavity, the other acts at the punch-
shaped elastic body being removed. The singularity of the equations consists in an
unlimited increase of the integrands at N→K. It will be shown below (Sect. 2.3) that
all the integrals, containing functions Uij (N, K) with a weak singularity (of the 1/R
type), can be calculated for flat integration surfaces numerically-and-analytically
with any degree of accuracy. Integrals, containing functions σij (N, K) with a strong
[email protected]
singularity (of the 1/R2 type), require special calculation in the sense of the Cauchy
principal value. Below it will be shown that for the contact problems considered
here the integrals, containing cores with strong singularities, can be excluded out of
2.1 Boundary Integral Equations of the Contact Problem for an Absolutely Rigid Punch 95

consideration after the account of displacements of the punch as a rigid solid as well
as the application of equilibrium equations for the auxiliary state.
The displacement of the punch as a rigid solid enables the following relation to
be written for the points on the contact surface [21]:

ui (K) = i − εijk ζj ψk (2.4)

where i are translational displacements of the punch, ψ k are small rotations of

the punch with respect to the coordinate axes, i, j, k = 1, 2, 3, K(ξ ,η,ζ ) ∈ , ζ1 =
ξ , ζ2 = η, ζ3 = ζ .
The boundary integral equations (2.3) of the contact problem for the deepened
punch with the account of Eq. (2.4) take the following form (summation over the
repeated indices is assumed):
  
1 
i − εijk ζj ψk = pj Uji d − j σji d + ψk εjlk ζl σji d, (2.5)
2
  

i, j,k,l =1, 2, 3.
The three obtained equations (2.5) can be essentially simplified by using the
equilibrium equations for the elastic body S in the shape of the deepened punch for
the auxiliary states from the action of the unit concentrated forces on the surface :
 
1 1
σji d = δji , εjlk ζl σji d = εijk ζj , i, j, k, l = 1, 2, 3. (2.6)
2 2
 

With the account of Eq. (2.6) the boundary integral equations of the spatial con-
tact problem for the rigid punch deepened into an elastic half-space, are given by

⎡ ⎤
 3
 
⎣ pj (N)Uji (K, N) ⎦ d = i − εijk ζj ψk , i,j,k,l = 1,2,3. (2.7)
 j=1

Equation (2.7) assert that the displacement of any point on the punch contact
surface is numerically equal to the work of contact forces pj (N) in the basic state on
the displacements Uji (K, N) of the auxiliary state.
Rigid displacements i and rotations ψ k of the punch (i, k = 1, 2, 3) are also
unknown, and to determine them one should invoke six equations of the punch equi-
librium (in the basic state):
 
Pi = pi (N)d, Mi = εijk xj pk (N)d, (i, j, k = 1,3). (2.8)
[email protected]
 

Thus the solution of the spatial contact problem for an absolutely rigid punch
of arbitrary shape, deepened into an elastic half-space, under an external static load
96 2 Static Analysis of Contact Problems for an Elastic Half-Space

of a general type is determined by a system of nine integral equations (2.7), (2.8)

which for convenience hereinafter can be presented in the following extended form:

⎧ 
⎪ 
⎪
⎪ p1 (N)U11 (K, N) + p2 (N)U12 (K, N) + p3 (N)U13 (K, N) d = 1 + ηψ3 − ζ ψ2 ,
⎪
⎪
⎪
⎪
⎪
⎪  
⎪
⎪ 
⎨
p1 (N)U21 (K, N) + p2 (N)U22 (K, N) + p3 (N)U23 (K, N) d = 2 + ζ ψ1 − ξ ψ3 ,
⎪
⎪
⎪
⎪
⎪  
⎪
⎪ 
⎪
⎪ p1 (N)U31 (K, N) + p2 (N)U32 (K, N) + p3 (N)U33 (K, N) d = 3 + ξ ψ2 − ηψ1 ,
⎪
⎪
⎩ 
(2.9)

⎧   
⎪
⎪ p1 (N)d = P1 , p2 (N)d = P2 , p3 (N)d = P3
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪    
⎪
⎪ 
⎪
⎪ p3 (N)x2 − p2 (N)x3 d = M1 ,
⎪
⎪
⎨ 
 (2.10)
⎪ 
⎪
⎪ p1 (N)x3 − p3 (N)x1 d = M2 ,
⎪
⎪
⎪
⎪
⎪
⎪  
⎪
⎪ 
⎪
⎪ p2 (N)x1 − p1 (N)x2 d = M3 .
⎪
⎪
⎪
⎩


Having solved the system of Eqs. (2.9), (2.10), one can determine three func-
tions of contact stresses pi and six parameters i , ψ i ψi (i = 1, 2, 3) of the punch
displacement as a rigid solid, i.e. the stress-strained state at the contact surface  is
determined.

2.2 Finite-Measure Analogue of the Contact Problem Using

Direct Boundary-Element Method

Analytical solutions of the system of integral equations (2.7), (2.8) formulated in

Sect. 2.1 for deepened punches of any particular geometrical shape have not been
obtained yet even for the simplest loading schemes. The main difficulty here, as has
been noted by many authors, concerns the integration of the fundamental Mindlin’s
solution.
For the numerical solution of the spatial contact problem, formulated in the most
[email protected]
general way, we use the boundary-element method in its direct formulation [7, 10,
34] when the unknown function values on the boundary have the physical sense
of contact pressures and play the role of source densities determining the stress-
2.2 Finite-Measure Analogue of the Contact Problem 97

strained state inside the stressed domain. The method application to an essentially
spatial contact problem under consideration is reduced to the following stages:

(1) discretization of the boundary surface G by means of a finite element ensemble;

(2) determination for the boundary elements of a finite set of nodes with respect
to which the collocation method is applied, enabling the node values of the
unknowns to be related based on a finite-measure analogue of the initial integral
equations;
(3) formation of the resolving system of algebraic equations whose coefficients
are calculated by analytical and/or numerical integration over each boundary
element;
(4) direct or iterative solution of the resolving system of algebraic equations;
(5) finding of the stress-strained state in the given internal points of the stressed
medium with invoking the schemes of numerical integration of various orders.

Hereinafter we restrict ourselves to the discretization of the contact surface 

with a set of boundary elements of polygonal (as a rule, triangular, and/or quadran-
gular) shape (Fig. 2.3). The overwhelming majority of volumetric deepened punches
in the problems of civil engineering (first of all, for geotechnical purposes, see Sects.
3.3 and 3.4) are restricted by fragments of planes or second-order (conical, cylin-
drical, or spherical) surfaces. A moderate number of flat boundary elements enables
the punch boundary of practically any geometrical shape to be approximated with a
required accuracy. Therefore, application of non-flat boundary elements in the con-
tact problems of geotechnics is hardly appropriate. Here we note once again that,
since in this approach one uses the fundamental Mindlin’s solution for the problem
of the concentrated force inside the elastic half-space (automatically satisfying the
boundary conditions on the stress-free base surface), only the contact surface of the
punch and the base can be discretized.

[email protected]
Fig. 2.3 Discretization of the
contact surface of the punch
and the elastic base using the
boundary elements
98 2 Static Analysis of Contact Problems for an Elastic Half-Space

After the boundary-element discretization of the boundary  the integral equa-

tion system of the contact problem is given by
⎧ 
⎪ 'm
⎪
⎪ i = pj (N)Uji (Kf , N)d  + ζk ψj εijk ;
⎪
⎪
⎪
⎪ q=1
⎪
⎪  q
⎨  
' m 'm
(2.11)
⎪
⎪ P i = p i (N)d ; M i = εijk pj (N)xk d ;
⎪
⎪
⎪
⎪
q=1
q
q=1
q
⎪
⎪
⎪
⎩
i,j,k = 1,3, f ,q = 1,2,...m

where m is the number of the boundary elements on the punch contact sur-
face,  q is the surface of the q-th boundary element, Kf are the collocation
points (the finite-element gravity centres), ζk are the coordinates of the point Kf
(ζ1 = ξ , ζ2 = η, ζ3 = ζ ).
The system of Eqs. (2.11) is the consequence of the system of Eqs. (2.7), (2.8)
where the calculation of 2-D integrals over the surface G is substituted by the sum
of integrals over the flat surfaces of the introduced boundary elements  q , q =
1, 2. . .m.
Within each boundary element one should assume that the contact forces pi vary
according to a pre-given law. As a rule, polynomial (constant, linear, quadratic,
or higher-order) approximation is applied [7, 10, 17, 34, 36]. Application of the
boundary-element method to the solution of spatial static problems for finite-size
bodies (local strength problems) shows that quite satisfactory results are achieved
already at application of piecewise constant or piecewise linear approximation of
the unknown densities, in particular, of the stress function. Note that in [16], based
on the analysis of the literature, a hypothetic idea is suggested to choose the order of
approximation for each boundary element by a unit higher than the order of approx-
imation of the functions to be found. Though this statement has not been proved
strictly (it is only confirmed by calculations for the flat and the axisymmetric cases),
a conclusion is made that it is appropriate to combine flat boundary elements and
constancy of the sought function, second-order elements and linear variation of the
sought functions etc. Violation of this correspondence is not justified since it does
not result in a guaranteed increase of the accuracy of the approximate solution. Thus,
at further application of flat boundary elements the approximation of piecewise con-
stant variation of the sought function of contact stress will be to a certain extent
justified. Then, taking into account that stress pi in theory of elasticity is presented
by a derivative of displacement ui , the application of constant stress on a boundary
element corresponds to a linear variation of displacement in the plane of each finite
element. This is in agreement with the linear distribution of displacements of the
boundary surface of an absolutely rigid punch and is an additional argument for the
[email protected]
piecewise constant approximation of the contact stress function in the proposed ver-
sion of the numerical boundary-element method with application of flat boundary
elements.
2.2 Finite-Measure Analogue of the Contact Problem 99

For m collocation nodes, in which the condition of fulfillment of boundary inte-

gral equations is set, we choose points, uniformly distributed over the discretized
punch surface. It is quite natural to obtain the first 3m equations of the algebraic
system of the boundary-element method by locating the unit forces of the auxiliary
state in the gravity centres of the boundary elements. Then, in accordance with the
approximation applied, the system of integral boundary equations of the spatial con-
tact problem for the absolutely rigid punch deepened into a half-space together with
the integral equilibrium equations can be given in the following discrete form:
⎧ 
⎪ 'm
⎪ 
⎪ i = p (N ) Uji (Kf , Nq )d  + ζk ψj εijk ;
⎪
⎪
j q
⎪
⎪ q=1
⎪
⎨ q

' m 'm (2.12)

⎪
⎪ Pi = pj (Nq )sq ; Mi = pj (Nq )εijk xk sq ;
⎪
⎪
⎪
⎪ q=1 q=1
⎪
⎪
⎩
i, j, k = 1,3, q, f = 1,2,...m

(q)
where pj (Nq ) = pj are the averaged values of contact stresses in the j-th direc-
tion within the q-th boundary element, Nq ∈ q ,  q is the surface of the q-th
boundary element, Sq = mes(q ) is the surface of the q-th boundary element.
In an extended notation the equation system (2.12) is given by
⎡
 
'
m
⎢
⎣p1 (Nq ) U1j (Kf , Nq )d  + p2 (Nq ) U2j (Kf , Nq )d +
q=1
q q
⎤ ⎧  + ζ ψ − η ψ , j = 1,
 ⎪
⎪ 1 f 2 f 3
⎨
⎥
+p3 (Nq ) U3j (Kf , Nq )d  ⎦ = 2 + ξf ψ3 − ζf ψ1 , j = 2,
⎪
⎪
q ⎩
3 + ηf ψ1 − ξf ψ2 , j = 3, f = 1,...,m,
(2.13)


m 
m 
m
p1 (Nq )sq = P1 , p2 (Nq )sq = P2 , p3 (Nq )sq = P3 , (2.14a, b, c)
q=1 q=1 q=1


m

p3 (Nq )yq − p2 (Nq )zq sq = M1 , (2.14d)
q=1


m

p1 (Nq )zq − p3 (Nq )xq sq = M2 , (2.14e)
q=1
[email protected]

m

p2 (Nq )xq − p1 (Nq )yq sq = M3 (2.14f)
q=1
100 2 Static Analysis of Contact Problems for an Elastic Half-Space

and enables the resolving system of linear algebraic equations of the boundary-
element method to be set in a matrix form

A·Z = B (2.15)
 
D3m×3m C3m×6
where A = is a square block matrix of the order of (3m + 6),
T6×3m 0
⎛ (fq) (fq) (fq)
⎞
δ11 δ12 δ13
 ⎜ ⎟
(fq) (fq) ⎜ (fq) (fq) (fq) ⎟
D3m×3m = δij is the influence matrix, δij = ⎜ δ21 δ22 δ23 ⎟ , f , q =
⎝ ⎠
(fq) (fq) (fq)
δ31 δ32 δ33
1,m;
⎛ ⎞
1 0 0 0 z1 −y1
⎜0 1 0 −z1 0 x1 ⎟
⎜ ⎟
⎜0 0 1 y1 −x1 0 ⎟
⎜ ⎟
C3m×6 = −⎜
⎜ ... ... ... ... ... ... ⎟
⎟,
⎜1 0 0 0 zm −ym ⎟
⎜ ⎟
⎝0 1 0 −zm 0 xm ⎠
0 0 1 ym −xm 0
⎛
s1 0 0 | s2 0 0 |...
⎜ 0 s 0 | 0 s 0 |...
⎜ 1 2
⎜ 0 0 s | 0 0 s |...
T6×3m =⎜
⎜
1 2
⎜ 0 −z 1 s 1 y1 s 1 | 0 −z 2 s 2 y2 s 2 |...
⎝ z1 s1 0 −x1 s1 | z2 s2 0 −x2 s2 |...
−y1 s1 x1 s1 0 | −y2 s2 x2 s2 0 |...
⎞
... | sm 0 0
... | 0 sm 0 ⎟
⎟
... | 0 0 sm ⎟ ⎟,
... | 0 −zm sm ym sm ⎟ ⎟
... | zm sm 0 −xm sm ⎠
... | −ym sm xm sm 0
Z and B are column vectors of the size (3m + 6):

Z = (p1 (N1 ), p2 (N1 ), p3 (N1 )..p1 (Nm ), p2 (Nm ), p3 (Nm ); 1 ,2 ,3 ,ψ1 ,ψ2 ,ψ3 )T ,
B = (0,0,0,...;P1 , P2 , P3 , M1 , M2 , M3 )T .
The dimensionality of the system of Eq. (2.15) equals (3m+6)× (3m+6) where
m is the total number of the boundary elements used for the approximation of the
contact surface of the punch and the elastic base. The vector of the unknowns Z
[email protected] (k)
includes 3m components of contact stresses pi (Nk ) = pi as well as six parameters
i , ψ i of the punch displacement as a rigid solid (i = 1, 2, 3; k= 1, 2,...m). In a
general case, the block D3m×3m of the matrix A is non-symmetrical and completely
filled. This block is characterized by the diagonal predominance of coefficients.
2.3 Numerical-and-Analytical Method of Integration of Fundamental Mindlin’s Solutions 101

Application of a conventional Gauss elimination method to solve the linear algebraic

equation system (2.15), as shown by a vast experience of calculations performed,
results in the numerical solution accuracy and stability, sufficient for the practical
purposes. The details of efficient implementation of algorithms of solutions of linear
algebraic equation systems of the boundary-element method, appropriate for the
specific features of Eq. (2.15), are given below in Sect. 3 (Sect. 3.4).
The coefficients of the main block D3m×3m of the matrix A are surface integrals
of the fundamental Mindlin’s solution

(fq)
δij = Uij (Kf , N)d (N) (2.16)
q

The analytical calculation of these integrals over flat triangular or quadrangular

domains, arbitrarily oriented in an elastic half-space, seems impossible. In practice
we have carried out the efficient integration by means of an original numerical-and-
analytical approach. It can be assumed that formation of the matrix coefficients of
Eq. (2.15) is the key point of the whole boundary-element method since it requires
both regular and improper integrals to be calculated with high precision and simul-
taneously in an optimal way from the point of view of the computation time. These
issues need to be considered in more detail what is performed in the following
subsection.

2.3 Numerical-and-Analytical Method of Integration

of Fundamental Mindlin’s Solutions

Formation of the matrix of coefficients of the resolving system (2.15) of linear alge-
braic equations of the boundary-element method is reduced to the calculations of
surface integrals of the fundamental Mindlin’s solution for the displacements

δij = Uij (Kf , N)d (N) (2.17)
q

The domains of integration of Eq. (2.17) are the simplest flat polygons (triangles
and quadrangles), arbitrarily oriented in the half-space.
The difficulties in the calculation of the integrals of Eq. (2.17) consist in the fact
that when the double integrals are reduced to iterated integrals, the primitives cannot
be found; besides, near the collocation points (when they belong to the integration
domain ) the integrands become unlimited. In the last case direct application
of standard procedures of numerical integration does not lead to the desired results
since for 2-D improper integrals it is very difficult to reveal the specific features
[email protected]
in the vicinity of the point K(ξ ,η,ζ ) of application of unit concentrated force by a
finite number of summands of the cubature formulae. The experience of numerical
calculations has shown that this requires a quite considerable increase of the num-
ber of integration points and, simultaneously, their concentration near the integrand
102 2 Static Analysis of Contact Problems for an Elastic Half-Space

singularities (adaptive numerical integration [34]) in order to obtain the result of

the desired accuracy. As a result, sufficient accuracy of the approximate values of
improper surface integrals requires too much computer time.
Consider the calculation of surface integrals of U(K,N) Mindlin displacements
based on the complementary possibilities provided by analytical and numerical inte-
gration methods with direct account of the integrand structure.
In Mindlin equations (1.7) for the displacements the singular terms are funda-
mental Kelvin’s solutions for the whole space, other terms have no singularities
(since they correspond to an imaginary point K̃(ξ ,η, − ζ ) of unit concentrated force
application). Hence, similarly to [9, 10], it is natural to present Mindlin formulae in
the form
Uij = (Uij )K + (Uij )C (2.18)

where superscripts K and C correspond to the singular Kelvin’s solution and the
auxiliary (regular) solution, respectively. As shown by the experience of numerous
calculations, analytical determination of improper integrals of the Kelvin functions
(containing only R1 powers, at K ∈ ΔΓ ) and numerical integration of complete
Mindlin’s solutions at K ∈ / ΔΓ has appeared an efficient (both in accuracy and in
speed) combined method of calculation of improper surface integrals in the spatial
problems of elasticity theory for a half-space.
Numerical integration was performed using the cubature formulae of various
order with the highest accuracy degree. In each separate case the choice of the
number of nodes of the cubature formulae was performed on the base of empiric
criteria obtained from an extended series of numerical experiments, including the
dependences on the discretization degree and the contact surface shape. A common
feature of the obtained regularities was an increase of the order of quadratures with
the decrease of the distance from the point K to the integration domain. The detailed
data on the numerical integration procedures are presented in Appendix B.
In order to determine the improper surface integrals with a weak (integrable)
singularity in the centre of gravity of the boundary element one can apply analyti-
cal transformations. As mentioned above, the singularities in the integrand expres-
sions are determined by the summands of the fundamental Kelvin’s solutions for an
unbounded elastic space [7]:

1 3 − 4ν zi zj
(Uij )K (K, N) = · δij + 3 (2.19)
16π G(1 − ν) R R

where zi = ζi − ξi , R = z21 + z22 + z23 , N(ζ1 ,ζ2 ,ζ3 ) is a point in the integration
domain, K(ξ 1 , ξ 2 , ξ 3 ) is the point of application of a unit force (source), ξi are
global Cartesian coordinates. Note that the Kelvin’s solution is a special case of the
Mindlin’s solution and can be obtained from it at R1 = R, R2 → R.
[email protected]
From the tensor notation of Eq. (2.19) it follows that the problem is reduced to
the exact calculation of the following integrals with a singularity in the centre of
gravity of a flat boundary element
2.3 Numerical-and-Analytical Method of Integration of Fundamental Mindlin’s Solutions 103

   
d z1 z3 z2 z3 z21
I1 = , I2 = d, I3 = d, I4 = d, (2.20)
R1 R31 R31 R31
   
  
z22 z23 z1 z2
I5 = d, I6 = d, I7 = d.
R31 R31 R31
  
If boundary elements on the surface of the half-space x3 = 0 are used, an addi-
tional pair of surface integrals should be included into consideration:
 
z1 z2
I8 = 2
d, I9 = d. (2.21)
R1 R21
 

We connect the point of application of the unit concentrated force K(ξ ,η,ζ ) ∈ 
with the vertices of the boundary element  j (Fig. 2.4) within which this point is
located. As a result, the flat integration domain will be divided into m̄ additional
triangular subelements  jk where k= 1, 2,..,m̄. Here m̄ = 3 for a triangular bound-
ary element, m̄ = 4 for a quadrangular one. It is clear that such an additional mesh,
being internal for each boundary element, does not lead to any changes in the gen-
eral approximating grid on the contact surface. Then, each of the improper integrals
considered in Eqs. (2.20) and (2.21) should be substituted by a sum
 β γ m̄   α β γ

zα1 z2 z3 z1 z2 z3
d = d. (2.22)
Rδ1 k=1  k
Rδ1
j j

Here α, β, γ , δ are integer powers, determined in accordance with Eqs. (2.18)

and (2.19).
Separate terms in the sum of Eq. (2.22) will be reduced to a set of elementary
functions by performing calculations according to a uniform procedure. The apices

Fig. 2.4 Representation of

[email protected]
triangular and quadrangular
elements using triangular
subelements with a common
vertex
104 2 Static Analysis of Contact Problems for an Elastic Half-Space

Fig. 2.5 Geometrical

representations for integration
over a flat triangular domain
with a singularity of the
integrand in one of the
vertices

of a currently considered fragment  jk anti-clockwise are put into correspondence

with the points M1 (X1 , Y1 , Z1 ), M2 (X2 , Y2 , Z2 ), M3 (X3 , Y3 , Z3 ) in such a way that
the first of them be simultaneously the point of application of the unit concentrated
force (the boundary-element centre of gravity). In the plane of the triangle M1 M2 M3
we introduce a polar coordinate system with a pole in the point M1 , the polar axis
being normal to the M2 M3 side (Fig. 2.5). We will show that the singularities of the
integrands for the integrals under consideration in the point M1 , due to the introduc-
tion of the +above polar
+ coordinate systems, annihilate.
Denote +M1 M ∗ + = |r̄∗ | = D, M1 M 3 = r̄13 , M1 M 2 = r̄12 . Then

 
R̄1 = R1 cos ϕ · ēρ + sin ϕ · ē⊥

where ēρ = |r̄r̄∗∗ | = {A1 , A2 , A3 } ; ē⊥ = |r̄r̄23

23 |
= {B1 ,B2 ,B3 } .
After the transition to the polar coordinates (R1 , ϕ) I1 can be readily calculated:

 ϕ2 D/cos ϕ ϕ2

d dϕ
I1 = = dϕ dR1 = D =
R1 cos ϕ
j k ϕ1 0 ϕ1
⎛ ϕ1 ⎞
(2.23)
[email protected]
  ϕ2
1 + sin ϕ2 1 − sin ϕ1 1 + tan 1 − tan
=
D
· ln ·
⎜ 2 2 ⎟
2 1 − sin ϕ2 1 + sin ϕ1
= D ln ⎝ ϕ2 · ϕ1 ⎠ .
1 − tan 1 + tan
2 2
2.3 Numerical-and-Analytical Method of Integration of Fundamental Mindlin’s Solutions 105

The components of the vector R̄1 are written as

z1 = R1 (A1 · cos ϕ + B1 · sin ϕ) ,

z2 = R1 (A2 · cos ϕ + B2 · sin ϕ) , (2.24)
z3 = R1 (A3 · cos ϕ + B3 · sin ϕ)

or, since x = R1 cos ϕ, y = R1 sin ϕ are the components of the R̄1 vector in the
plane of the triangle M1 M2 M3 (in the Cartesian coordinate system OX  Y  , formed
by eρ ,e⊥ vectors), then
z1 = A1 · x + B1 · y ,
z2 = A2 · x + B2 · y , (2.25)
z3 = A3 · x + B3 · y .

Using Eqs. (2.24), (2.25), the sought integrals can be given by

Iq = Qq · J1 + Sq · J2 + Tq · J3 , q = 2,3,...,7;
I8 = A1 · J4 + B1 · J5 , I9 = A2 · J4 + B2 · J5
where

Q2 = A1 · A3 , S2 = B1 · B3 , T2 = A3 · B1 + A1 · B3 ;
Q3 = A2 · A3 , S3 = B2 · B3 , T3 = A3 · B2 + A2 · B3 ;
Q4 = A21 , S4 = B21 , T4 = 2A1 · B1 ;
Q5 = A22 , S5 = B22 , T5 = 2A2 · B2 ;
Q6 = A23 , S6 = B23 , T6 = 2A3 · B3 ;
Q7 = A1 · A2 , S7 = B1 · B2 , T7 = A2 · B1 + A1 · B2 ;
    2     2   
x y xy
J1 = 3
d, J2 = 3
d, J3 = d,
R1 R1 R31
j k j k j k
 
x y
J4 = d, J5 = d.
R21 R21
j k j k

Now the integrals J1 , J2 ,. . ., J5 after the transition to the polar coordinates are
obtained in quadratures

J1 = D (sin ϕ2 − sin ϕ1 ) , J2 = I1 − J1 , J3 = −D (cos ϕ2 − cos ϕ1 ) ,

 + +  + +
+ ϕ1 + + cos ϕ1 +
J4 = D A1 (ϕ2 − ϕ1 ) + B1 ln + cos
cos ϕ2 + , J5 = D A2 (ϕ2 − ϕ1 ) + B2 ln + cos ϕ2 + .

(2.26)
[email protected]
Considering the coordinates Xi , Yi , Zi , (i= 1, 2, 3) of the apices of the triangular
subelement  jk to be known, for the sake of completeness we give the formulae
to determine the Ai , Bi (i= 1, 2, 3), and D values:
106 2 Static Analysis of Contact Problems for an Elastic Half-Space

X∗ −X1 Y∗ −Y1 Z∗ −Z1

A1 = D , A2 = D , A3 = D ,
X3 −X2 Y3 −Y2 Z3 −Z2
B1 = B2 =
, B3 =, ,
R̂ R̂ R̂

R̂ = (X3 − X2 )2 + (Y3 − Y2 )2 + (Z3 − Z2 )2 ,

D = (X∗ − X1 )2 + (Y∗ − Y1 )2 + (Z∗ − Z1 )2 ,
X∗ = X2 + (X2 − X3 ) · t, Y∗ = Y2 + (Y2 − Y3 ) · t, Z∗ = Z2 + (Z2 − Z3 ) · t,
t = ((X2 − X1 ) (X3 − X2 ) + (Y2 − Y1 ) (Y3 − Y2 ) + (Z2 − Z1 ) (Z3 − Z2 )) /R̂2 .

The presented expressions Eqs. (2.23), (2.26) for the improper integrals over a
flat triangular domain with a singularity in one of its apices enable the accuracy of
calculation of the diagonal coefficients of the canonical equation matrix to be essen-
tially increased at a simultaneous decrease of the computation time (in comparison
with only numerical integration being used).
In the analytical integration we mostly followed the approach suggested by Cruse
[12] who was among the first to obtain analytical expressions for diagonal coeffi-
cients of the influence matrix for a flat triangular boundary element. In addition to
[12], we have also obtained analytical expressions for special (limiting) integrals,
arising at the application of boundary elements on the half-space surface. As one
should expect, the final equations (2.23), (2.26) with the accuracy to identity trans-
formations, are in agreement with the results of [12]. Besides, note that the expres-
sions obtained here have the advantages of giving directly the formulae with the
known coordinates of the boundary-element nodes what is convenient for practical
applications.
In a series of numerical comparison experiments it has been found that the pre-
sented analytical-and-numerical integration method appeared comparable in effi-
ciency with the known methods (in view of speed at the given calculation accuracy).
In our case the natural increase of the computation time for the integration of the
Mindlin’s solution is caused by the fact the latter being more complicated than the
Kelvin’s solution (the presence of additional eighteen deformation cores) and is to a
great extent justified by the condition of vanishing of stress at the half-space surface
being automatically satisfied.
In view of the comparison performed we would give a brief account of other
existing approaches to the formation of influence matrices in the direct boundary-
element method for spatial problems of theory of elasticity [5, 6, 35–38]. In all of
the known studies the results are obtained in closed form only in the case when the
external normal does not change its direction, i.e. for flat boundary elements being
used.
A rather descriptive numerical-and-analytical method of calculation of the matrix
of an algebraic analogue of the system of boundary integral equations was suggested
[email protected]
by Yakimchuk and Kvitka [38]. In the boundary-element plane a local polar coordi-
nate system was introduced. The surface integrals were reduced to iterated integrals
for which the integration over the polar radius was performed analytically using a
2.3 Numerical-and-Analytical Method of Integration of Fundamental Mindlin’s Solutions 107

software for computation of indefinite integrals (Analytic language). Numerical inte-

gration over the angular variable, using the quadrature formulae, is recommended.
The procedure of formation of the influence matrix, suggested in [38] which its
authors call semianalytical, does not seem to have visible advantages and has not
been further developed or spread for the solution of spatial problems of theory of
elasticity in the studies performed by other groups. Since we have performed effi-
cient numerical integration of the fundamental Mindlin’s solutions over the optimal
quadrature formulae without a transition to consideration of iterated integrals, the
semianalytical method, proposed in [38], in case being applied to a half-space, will
only create additional difficulties and will obviously be inefficient.
Expressions of a rather cumbersome structure at the analytical calculation of
integrals from the Kelvin’s solution with density functions in the form of algebraic
polynomials are given in [36]. The formulae have a sufficiently general form and are
applicable both in the cases the collocation point (pole) belonging to the integration
domain (K ∈ Δ) and being located outside it (K ∈ / Δ). Parallel translation and
rotation of the Cartesian coordinate system axes are used for the transition to the
plane of the boundary element . Later, in [37], Roytfarb et al. have also obtained
in a closed analytical form the expressions for the coefficients of the Kelvin influ-
ence matrix in a special case (with respect to [37]) of piecewise constant approxi-
mation of the sought densities and using a local polar coordinate system linked to
a side of a polygon, arbitrarily oriented in space. In spite of the obvious efficiency
of the methods developed in [36, 37], they possess certain inconveniences in the
practical application of the obtained results for solution of problems for an elastic
half-space. Namely, the formulae for the primitives contain a great number of tran-
scendental functions, the reliable calculation of which is known to require double-
precision computations and, hence, additional computation time. In the case under
consideration, when part of the terms (Uij )c in the Mindlin’s solution is always sub-
ject to numerical integration, the use of analytical transformation to obtain all the
coefficients of the Kelvin influence matrix (the total number of integrals is 3m×3m
where m is the number of the boundary elements on the contact surface) is abso-
lutely unjustified. As has been shown by intentional numerical experiments, this
increases the computation time by factor of 1.5/2 without a noticeable increase of
the calculation accuracy.
In [5, 6, 35], the earlier approach of [36] is developed, using the method of analyt-
ical integration of the Mindlin’s solutions over triangular flat elements. The practi-
cal applicability of the proposed method was restricted by the presence of primitives
only for the cases when the flat integration domain was parallel to the axes of the
global coordinate system, in which the expressions for the integrands were written.
Such approach leaves beyond consideration a great class of problems, important for
applications, when at the approximation of contact surfaces boundary elements with
different angles of inclination with respect to the coordinate axes arise (Sect. 3.4).
Unfortunately, until now we failed to obtain primitives in double integrals from the
[email protected]
additional terms of the Mindlin’s solution for flat, arbitrarily oriented boundary ele-
ments even using such modern powerful software for analytical transformations as
Matcad, Maple, Mathematica, Derive etc.
108 2 Static Analysis of Contact Problems for an Elastic Half-Space

2.4 Punch in the Shape of a Rotation Body, Deepened

into an Elastic Half-Space
If the contact surface of a punch is a rotation surface and its loading and the bound-
ary conditions possess axial symmetry, the spatial problem of theory of elasticity is
essentially simplified. In this case it is quite natural to use more precise and efficient
procedures of numerical solution, taking into account the symmetry of the problem.
This will save the computation resources and revise the calculation formulae, finally
resulting in a more rational design solutions.
In a cylindrical coordinate system (r, ϕ, z), for which the Oz axis is combined
with the punch axis, all the parameters of the stress-strained state are independent
of the angular coordinate φ and, due to such azimuthal symmetry, the contact prob-
lem becomes two-dimensional. The simplest axisymmetric punches are a sphere, a
cylinder, a cone (including a frustum of a cone). More complicated axisymmetric
structures are presented in Sect. 3.4.
Contact problems with an axial symmetry for punches, deepened into an elastic
half-space, can be divided in two groups. The first one corresponds to the forced

(a) (b)

(c)

[email protected]
Fig. 2.6 Axially symmetrical contact problem for a rigid punch, deepened into an elastic half-
space: (a) calculation scheme; (b) contact stress in the horizontal plane; (c) cyclic (over the angular
coordinate) discretization of the contact surface and representation of the contact stress vector on
the boundary elements
2.4 Punch in the Shape of a Rotation Body, Deepened into an Elastic Half-Space 109

loading of the punch along the symmetry axis and is characterized by action of
tangential and normal stresses on the contact surface as well as radial and axial dis-
placements in the stressed base (Fig. 2.6a). The second group corresponds to the
punch torsion under a torque, collinear to the symmetry axis, and is characterized
by the action of a pair of tangential contact stresses and solely tangential displace-
ments in the elastic half-space, i.e. the stress distribution, inversely symmetrical with
respect to the axis (Fig. 2.7a).
For axisymmetric problems, being one of the most important classes of spatial
problems of theory of elasticity, there are efficient methods of solution; a great
number of forms of general solution are known [31]. The problem is eventually
two-dimensional, for its solution well developed means of theory of analytic and
p-analytic functions can be used [3, 26]. According to [31], the studies of axisym-
metric stress-strained state of bodies of finite size is one of the most extensively
developed fields of theory of elasticity, the better results have been achieved only
for the flat problem. Nevertheless, still no analytical solutions of contact problems
with axial symmetry for punches of even the simplest shape, deepened into an elas-
tic half-space, are available.
Below we present the integral equations for contact problems with axial symme-
try and construct efficient boundary-element algorithms of their solution, suitable

(a) (b)

(c)

[email protected]
Fig. 2.7 Torsion of an elastic half-space by a rigid deepened punch in the shape of a rotation body:
(a) calculation scheme; (b) stress-strained state of the horizontal plane; (c) cyclic (over the angular
coordinate) discretization of the contact surface and representation of the contact stress vector on
the boundary elements
110 2 Static Analysis of Contact Problems for an Elastic Half-Space

for punches in the shape of rotation bodies without any restrictions on the merid-
ional cross-section shape.

2.4.1 Axisymmetric Contact Problem

At central loading by an axial force P3 = Pz a deepened absolutely rigid punch in
the shape of a rotation body will be displaced only vertically. The stressed state of
an elastic half-space is characterized by radial pr and vertical pz components of the
contact stress vector (there is no tangential stress due to the axial symmetry) which
depend only on the vertical coordinate (Fig. 2.6a, b).
The system of equations of the spatial contact problem for a deepened axisym-
metric punch, written using the theorem of reciprocity of works for the basic and
the auxiliary states (see Sect. 2.1), is given by
⎧    
⎪ pr (N)Ur (K, N) + pz (N)W (K, N) d − 3 z (K, N)d = 0,
(1) (1)
⎪
⎪ p(1)
⎪
⎨
 
   
⎪
⎪ 1
⎪
⎪ pr (N)Ur (K, N) + pz (N)W (K, N) d − 3
(3) (3)
z (K, N)d = 3
p(3)
⎩ 2
 
(2.27)
where 3 is vertical displacement of the punch, pr (N), pz (N) are the projections of
the contact stress vector in the point N on the cylindrical coordinate system axes,
Ur(k) (K, N), W (k) (K, N) are displacements of points of the elastic half-space, deter-
mined from the following formulae
6 (k)
Ur (K, N) = U1 k (K, N) · cos ϕ + U2 k (K, N) · sin ϕ,
(2.28)
W (k) (K, N) = U3 k (K, N), k = 1,3

Uij (K, N), i, j = 1,3 is the fundamental Mindlin’s solution, written in the global
Cartesian coordinate system,  is the punch contact surface, points N ∈  and
K∈Γ.
The obtained integral equations (2.27) are essentially simplified if one takes
into account the equilibrium equations of an elastic body in the shape of the deep-
ened punch in the auxiliary state under the action of unit concentrated forces (see
Sect. 2.1):
 
1
p(1)
z (K, N)d = 0, z (K, N)d =
p(3) . (2.29)
2
 
[email protected]
After substitution of Eq. (2.29) into Eq. (2.27), one obtains integral equations
of the axisymmetric contact problem for a rigid punch, deepened into an elastic
half-space in the shape of a rotation body, under an axial load:
2.4 Punch in the Shape of a Rotation Body, Deepened into an Elastic Half-Space 111

⎧   
⎪
⎪ pr (N)Ur(1) (K, N) + pz (N)W (1) (K, N) d = 0,
⎪
⎪
⎨

   (2.30)
⎪
⎪
⎪
⎪ p (N)U (3)
(K, N) + p (N)W (3)
(K, N) d = 3 ,
⎩ r r z


Boundary integral equations for the axisymmetric problem of Eq. (2.30) can be
as well obtained in a formal way using the general equations of the spatial contact
problem (2.9) in a special case when the punch does not undergo any rotations
(ψ1 = ψ2 = ψ3 = 0) and displacements across the symmetry axis (1 = 2 = 0).
Then the integral equation system (2.9) is given by
⎧  
⎪
⎪ p1 (N)U11 (K, N) + p2 (N)U12 (K, N) + p3 (N)U13 (K, N) d = 0,
⎪
⎪
⎪
⎪
⎪
⎪ 
⎪
⎪ 
⎨ 
p1 (N)U21 (K, N) + p2 (N)U22 (K, N) + p3 (N)U23 (K, N) d = 0, (2.31)
⎪
⎪
⎪
⎪ 

⎪
⎪ 
⎪
⎪
⎪
⎪ p1 (N)U31 (K, N) + p2 (N)U32 (K, N) + p3 (N)U33 (K, N) d = 3 .
⎩


In the plane, orthogonal to the punch symmetry axis, for each contact point S
we introduce a local coordinate system whose axes are directed tangentially (t)
and normally (n) to the cross-section contour (Fig. 2.6b). Then, using the formu-
lae for transformation of the displacement and stress vector components at the axis
rotation
6 6
p1 = pr · cos ϕ − pt · sin ϕ, U1 = Ur · cos ϕ − Ut · sin ϕ,
(2.32)
p2 = pr · sin ϕ − pt · cos ϕ, U2 = Ur · sin ϕ − Ut · cos ϕ,

the system (2.31) can be presented in projections on the cylindrical coordinate sys-
tem axes
⎧   
⎪
⎪ (1) (1)
pr (N)Ur(1) (K, N) + pt (N)Ut (K, N) + p3 (N)U3 (K, N) d = 0,
⎪
⎪
⎪
⎪
⎪
⎪ 
⎪
⎪   
⎨ (2) (2)
pr (N)Ur(2) (K, N) + pt (N)Ut (K, N) + p3 (N)U3 (K, N) d = 0,
⎪
⎪ 
⎪
⎪
⎪  
⎪
⎪

⎪
⎪ p (N)U (3)
(K, N) + p (N)U
(3)
(K, N) + p (N)U
(3)
(K, N) d = 3 .
⎪
[email protected]
⎩ r r t t 3 3

(2.33)
112 2 Static Analysis of Contact Problems for an Elastic Half-Space

Due to the symmetry at the axial loading the tangential components of the contact
stress vector will be zero (pt = 0) what considerably simplifies the system (2.33).
Taking into account the fact that the first and the second equations of Eq. (2.33)
are linearly dependent (i.e. one of them is a consequence of the other), the inte-
gral equation system of the axisymmetric contact problem takes the above form of
Eq. (2.30).
Equation (2.30) should be complemented with an integral equation of equilib-
rium

pz (N) d = P3 (2.34)


where P3 is the resultant of the external forces, applied to the punch in the direction
of the z axis.
Note that other five integral equations of equilibrium in the system (2.10) are
fulfilled identically since at the axial loading P1 = P2 = M1 = M2 = M3 = 0,
pr and pz are independent of the angular coordinate ϕ, and after the transition form
double integrals to iterated integrals each of the terms will contain zero factors

2π 2π
cos ϕdϕ = 0, sin ϕdϕ = 0.
0 0
Thus, the axisymmetric problem of theory of elasticity, consisting in the deter-
mination of the contact forces pr , pz and vertical displacements 3 , is reduced to
the solution of the integral equation system (2.30) under the integral condition of
Eq. (2.34) being fulfilled. Having found the solution, one can easily calculate the
normal p(n) and tangential p(τ) contact stresses based on the known relations
6
p(n) = pr · cos θ + pz · sin θ ,
(2.35)
p(τ ) = −pr · sin θ + pz · cos θ

where θ is the angle between the external normal to the contact surface and the
horizontal plane (Fig. 2.6a).
For an approximate solution of the axisymmetric contact problem under consid-
eration we use the direct boundary-element method combined with the piecewise
constant approximation of the contact stress function what will enable the integral
equations (2.30) and (2.34) to be reduced to a system of linear algebraic equations.
The most convenient way is to divide the contact surface of the punch in the
shape of a rotation body into flat boundary elements whose nodes are formed by
intersection of the “geographical” system of coordinate lines. For this purpose we
build Q planes, containing the symmetry axis, turned by equal angles ϕ = 2π/Q.
[email protected]
As a result, on the punch surface Q meridional zones will be formed. Then we
build M = M +1 horizontal planes, not necessarily equidistant. Consequently, the
surface of the deepened part of the punch will be divided into M×Q boundary ele-
ments, among which there will be Q triangular and (M–1)×Q quadrangular elements
2.4 Punch in the Shape of a Rotation Body, Deepened into an Elastic Half-Space 113

(Fig. 2.6c). We can say that we use cyclic discretization of the rotation surface over
the angular coordinate since at a rotation around the symmetry axis by an angle,
multiple of ϕ = 2π/Q, there will be a coincidence of all the boundary-element
nodes with their initial positions. Note that the variation of the distance between the
horizontal planes enables one, taking into account the curvature of the punch gener-
atrix, to perform discretization uniformly with the required vertical condensation.
After the discretization of the surface of contact between the deepened part of the
punch and the elastic half-space, the formed system of linear algebraic equations of
the boundary-element method is given by
⎧  
⎪
⎪ '
N
⎪
⎪ [p (N ) d + p (N ) W (1) (Ki , N)d] = 0,
⎪
⎪
r n z n
⎪
⎪ n=1
⎪
⎪ n n
⎪
⎨' N
 
[pr (Nn ) Ur(3) (Ki , N)d + pz (Nn ) W (3) (Ki , N)d] = 3, (2.36)
⎪
⎪
⎪
⎪
n=1
n n
⎪
⎪
⎪
⎪
⎪
⎪ '
N
⎪
⎩ pz (Nn )sn = P3 , i = 1,2,...,N.
n=1

Here the following notations are used: N = M×Q is the total number of the bound
elements on the punch contact surface; pr (Nn ), pz (Nn ) are the averaged values of the
radial and vertical contact stress, respectively, within the n-th boundary element; Ki
are the collocation points (the gravity centres of the boundary elements); Sn is the
area of the n-th boundary element.
We further reduce the system (2.36), assuming the above discretization of the
contact surface between the punch and the base to be regular, cyclic over the angular
coordinate. The latter condition enables one to increase essentially the dimension-
ality of the algebraic analogue in comparison with the system of Eqs. (2.10) and
(2.9) for the general spatial contact problem. If the punch generatrix was divided
by the horizontal planes into M sections, and over the angular coordinate into Q
meridional zones (being determined by equal dihedral angles ϕ = 2π/Q), then the
system (2.36) with the account of the cyclicity requirement is given by

⎧ ⎡ ⎤
⎪
⎪  
⎪
⎪ '
M ⎢ '
Q
(1) '
Q
⎥
⎪
⎪ ⎣pr (Nm ) Ur (Ki , N)d + pz (Nm ) W (1) (Ki , N)d ⎦ = 0,
⎪
⎪
⎪
⎪
m=1 q=1
M(q−1)+m
q=1
M(q−1)+m
⎪
⎪
⎪
⎪ ⎡ ⎤
⎨
 
'
M ⎢ '
Q
(3) '
Q
⎥ ,
⎪
⎪ ⎣pr (Nm ) Ur (Ki , N)d + pz (Nm ) W (3) (Ki , N)d ⎦ = 3
⎪
⎪
⎪
⎪ m=1 q=1 q=1
⎪
⎪ M(q−1)+m M(q−1)+m
⎪
⎪
⎪
⎪
⎪
⎪ '
M
⎩ pz (Nm )sm = P3 /Q, i = 1,2,...,M.
[email protected]
m=1
(2.37)

Here M = N/Q is the number of boundary elements in a single meridional zone

of the punch contact surface.
114 2 Static Analysis of Contact Problems for an Elastic Half-Space

For the sake of convenient realization of numerical algorithms the algebraic ana-
logue Eq. (2.37) of the integral equation system of the axisymmetric contact prob-
lem is presented in the matrix form:

A·Z = B (2.38)
 
D2 M×2 M C2 M×1
where A = is a square block matrix of the order (2M + 1),
T1×2 M 0
⎛ ⎞
(1) (1) (1) (1) (1) (1)
a11 b11 a21 b21 ... aM1 bM1
⎜ ⎟
⎜ (3) (3) (3) (3) ⎟
⎜ a11 b11 a21 b(3) a (3)
b M1 ⎟
⎜ 21 ... M1 ⎟
⎜ ⎟
⎜ ... ... ⎟
D2M×2M = ⎜ ... ... ... ... ⎟ is the influence matrix,
⎜ ⎟
⎜ ⎟
⎜ (1) (1) (1) (1) (1) (1) ⎟
⎜ a1M b1M a2M b2M ... aMM bMM ⎟
⎝ ⎠
(3) (3) (3) (3) (3) (3)
a1M b1M a2M b2M ... aMM bMM
⎛ ⎞
0
⎜1⎟
⎜ ⎟
⎜0⎟
⎜ ⎟
⎜ ⎟
C2 M×1 = − ⎜ 1 ⎟ , T1×2 M = (0,s1 ;0,s2 ;...;0,sM ) ;
⎜ .. ⎟
⎜. ⎟
⎜ ⎟
⎝0⎠
1
Z and B are column vectors of the size (2M + 1):

Z = (pr (N1 ), pz (N1 ); pr (N2 ), pz (N2 );...pr (NM ),pz (NM );3 )T ,

B = (0,0;0,0;...;P3 )T ;

Si (i = 1, 2,. . .M) are the areas of flat triangles or quadrangles, dividing the merid-
ional zone into the boundary elements whose numbering is determined in accor-
dance with the vertical coordinate variation;

(k) '
Q
aim = Ur(k) (Ki , N)d =
q=1
M(q−1)+m
 
'
Q '
Q
= cos ϕq U1 k (Ki , N)d + sin ϕq U2 k (Ki , N)d;
q=1 q=1
M(q−1)+m M(q−1)+m
[email protected]
 
Q ' '
Q
b(k)
im = W (k) (Ki , N)d = U3 k (Ki , N)d
q=1 q=1
M(q−1)+m M(q−1)+m
2.4 Punch in the Shape of a Rotation Body, Deepened into an Elastic Half-Space 115

are the coefficients of the influence matrix D2M×2M , determined by the numerical-
and-analytical integration method, described above in Sect. 2.3, ϕq = (2q−1)π/Q,
k = 1, 3; 4, m = 1, 2,. . ., M.
It is seen that the dimensionality of the algebraic analogue of the boundary
contact problem equals (2M + 1) with respect to the values of 3 and pr (Nm ),
pz (Nm ) (m = 1, 2,. . .M) where M is the boundary element number along the punch
generatrix.

2.4.2 Torsion of an Axisymmetric Punch in an Elastic Half-Space

Consider a punch in the shape of a rotation body, deepened into an elastic half-space,
under the action of a sole torque M3 =0 (P1 = P2 = P3 = M1 = M2 = 0). Then
the punch displacement will be determined only by the angle ψ 3 =0 (ψ 1 = ψ 2 =
1 = 2 = 3 = 0), characterizing the punch rotation around the Oz axis
(Fig. 2.7a). The equation system (2.9) in the special case under consideration is
given by
⎧ 
⎪ 
⎪
⎪ p1 (N)U11 (K, N) + p2 (N)U12 (K, N) + p3 (N)U13 (K, N) d = η · ψ3 ,
⎪
⎪
⎪
⎪
⎪
⎪ 
⎪
⎪ 
⎨ 
p1 (N)U21 (K, N) + p2 (N)U22 (K, N) + p3 (N)U23 (K, N) d = −ξ · ψ3 ,
⎪
⎪
⎪
⎪ 
⎪
⎪ 
⎪
⎪ 
⎪
⎪ p1 (N)U31 (K, N) + p2 (N)U32 (K, N) + p3 (N)U33 (K, N) d = 0.
⎪
⎩

(2.39)

Similarly to the case of the axisymmetric problem (see Sect. 2.4.1), using the
formulae (2.32) for the transformation of the vector components at the axis rotation,
the system (2.39) is presented in projections onto the cylindrical coordinate system
axes
⎧   
⎪
⎪ (1)
+
(1)
+ (1)
d = η · ψ3 ,
⎪
⎪ pr (N)U r (K, N) pt (N)Ut (K, N) pz (N)Uz (K, N)
⎪
⎪
⎪
⎪ 
⎪
⎪  
⎪
⎨ 
(2)
pr (N)Ur(2) (K, N) + pt (N)Ut (K, N) + pz (N)Uz(2) (K, N) d = −ξ · ψ3 ,
⎪
⎪
⎪
⎪ 
⎪
⎪   
⎪
⎪
⎪
⎪
(3)
pr (N)Ur(3) (K, N) + pt (N)Ut (K, N) + pz (N)Uz(3) (K, N) d = 0.
⎪
⎩

[email protected] (2.40)

Since, according to the problem formulation, the punch does not undergo axial
forces, there will arise no vertical stress on the contact surface, i.e. pz = 0. Besides,
116 2 Static Analysis of Contact Problems for an Elastic Half-Space

the system (2.40) will be additionally simplified due to the fact that its first and sec-
ond equations are linearly dependent (one is the consequence of the other). As a
result, the system of boundary integral equations for the contact problem of an elas-
tic half-space torsion due to the axial rotation of a deepened punch will be given by

⎧   
⎪
⎪ (1)
+
(1)
d = η · ψ3 ,
⎪
⎪ p r (N)Ur (K, N) pt (N)Ut (K, N)
⎪
⎨ 
   (2.41)
⎪
⎪
⎪
⎪ p (N)U (3)
(K, N) + p (N)U
(3)
(K, N) d = 0,
⎪
⎩
r r t t


where
⎧
⎨ Ur(k) (K, N) = U1 k (K, N) · cos ϕ + U2 k (K, N) · sin ϕ,
(2.42)
⎩ U (k) (K, N) = −U (K, N) · sin ϕ + U (K, N) · cos ϕ,k = 1,3.
t 1k 2k

The equation system (2.41) becomes closed if it is complemented by the integral

equilibrium equation


x1 p2 (N) − x2 p1 (N) d = M3 . (2.43)


Evidently, the other five equilibrium equations of the system (2.10) will be iden-
tically fulfilled due to the symmetry of the problem and independence of the con-
tact stress on the angular coordinate, similarly to the case of the axial loading
(Sect. 2.4.1).
The equilibrium equation (2.43), similarly to the boundary integral equations
(2.41), can be written in terms of the radial and tangential projections of the stress
vector. Taking into account that x1 = rcosϕ, x2 = rsinϕ, and p1 and p2 are expressed
in terms of pr and pt using Eq. (2.32), the equilibrium equation (2.43) is given by

pt · rd = M3 (2.44)


where r = x12 + x22 = r(z) is the radial coordinate of the contact surface points.
Thus, the inverse symmetrical problem of theory of elasticity, consisting in the
determination of the contact forces pr and pt and rotation angles ψ3 , is reduced
to the solution of the integral equation system (2.41), the integral condition of Eq.
(2.44) being fulfilled. Note that the found solution will simultaneously determine
the normal p(n) = pr and tangential p(τ) = pt contact stresses (Fig. 2.7c).
[email protected]
The approximate solution of the inverse symmetrical contact problem will be
obtained similarly to the above considered (Sect. 2.4.1) axisymmetric problem,
using the direct boundary-element method in combination with the piecewise
2.4 Punch in the Shape of a Rotation Body, Deepened into an Elastic Half-Space 117

constant approximation of the contact stress function. Assuming the discretization

with the cyclic symmetry to be performed (Fig. 2.7c) and omitting cumbersome
intermediate calculations, the integral equations (2.41) and (2.44) can be readily
reduced to the following system of linear algebraic equations of the direct boundary-
element method:

⎧ ⎡ ⎤
⎪
⎪  
⎪
⎪ ' ⎢
M '
Q
(1) '
Q
(1) ⎥
⎪
⎪ ⎣pr (Nm ) Ur (Ki , N)d + pt (Nm ) Ut (Ki , N)d ⎦ = ηi · ψ3 ,
⎪
⎪ m=1
⎪
⎪
q=1
M(q−1)+m
q=1
M(q−1)+m
⎪
⎪
⎪
⎪ ⎡ ⎤
⎨
 
'
M ⎢ '
Q
(3) '
Q
(3) ⎥
⎪
⎪ ⎣pr (Nm ) Ur (Ki , N)d + pt (Nm ) Ut (Ki , N)d ⎦ = 0,
⎪
⎪
⎪
⎪ m=1 q=1 q=1
⎪
⎪ M(q−1)+m M(q−1)+m
⎪
⎪
⎪
⎪
⎪
⎪ '
M
⎩ pt (Nm )rm sm = M3 /Q, i = 1,2,...,M.
m=1
(2.45)

Here pr (Nm ) and pt (Nm ) are the averaged values of the radial and tangential con-
tact stress, respectively, within the m-th boundary element, the rest of notations
being the same as those in Sect. 2.4.1.
In the matrix form the algebraic analogue (2.45) of the integral equation system
is given by

A·Z = B (2.46)
 
F2 M×2 M G2 M×1
where A = is a square block matrix of the order (2M + 1),
H1×2 M 0
⎛ (1) (1)
⎞
g11 h11 g(1) (1) (1) (1)
21 h21 ... gM1 hM1
⎜ ⎟
⎜ (3) (3) (3) ⎟
⎜ g11 h11 g(3) h (3)
... g (3)
h M1 ⎟
⎜ 21 21 M1 ⎟
⎜ ⎟
F2 M×2 M = ⎜ ... ... ... ... ... ... ⎟ is the influence matrix,
⎜ ⎟
⎜ (1) (1) (1) (1) (1) (1) ⎟
⎜ g1 M h1 M g2 M h2 M ... gMM hMM ⎟
⎝ ⎠
(3) (3) (3) (3) (3) (3)
g1 M h1 M g2 M h2 M ... gMM hMM
⎞ ⎛
η1
⎜0 ⎟
⎜ ⎟
⎜ η2 ⎟
⎜ ⎟
⎜ ⎟
G2 M×1 = − ⎜ 0 ⎟ , H1×2 M = (0,r1 s1 ;0,r2 s2 ;...;0,rM sM ) ;
⎜ .. ⎟
[email protected]
⎜. ⎟
⎜ ⎟
⎝ ηM ⎠
0
118 2 Static Analysis of Contact Problems for an Elastic Half-Space

Z and B are column vectors of the size (2M + 1):

Z = (pr (N1 ), pt (N1 ); pr (N2 ),pt (N2 );...pr (NM ),pt (NM );ψ3 )T ,

B = (0,0;0,0;...;M3 )T ;


(k) '
Q
gim = Ur(k) (Ki , N) d =
q=1
M(q−1)+m
 
'
Q '
Q
= cos ϕq U1 k (Ki , N) d + sin ϕq U2 k (Ki , N) d;
q=1 q=1
M(q−1)+m M(q−1)+m

(k) '
Q
(k)
him = Ut (Ki , N) d =
q=1
M(q−1)+m
 
'
Q '
Q
=− sin ϕq U1 k (Ki , N) d + cos ϕq U2 k (Ki , N) d
q=1 q=1
M(q−1)+m M(q−1)+m

are the coefficients of the influence matrix F2M×2M , ϕq = (2q−1)π/Q, k = 1, 3;

i, m = 1, 2,. . .,M.
Evidently, the dimensionality and the structure of the matrix representation of
the algebraic analogue of the boundary contact problem in terms of ψ3 and pr (Nm ),
pz (Nm ) (m = 1, 2,. . .M) values are the same as in the above axisymmetric case.
Thus, we have considered the integral equations of the contact problems for deep-
ened punches with axial symmetry and have constructed an efficient method of their
solution ob the base of direct boundary-element formulation. The results of the for-
mulations presented in Sects. 2.4.1 and 2.4.2 enable numerical solutions of the class
of contact problems of theory of elasticity, important for practical application, to
be effectively constructed. Attention should be paid to the following main advan-
tages of the elaborated algorithms, increasing the efficiency of application of the
boundary-element method in engineering.
As noted above, spatial contact problems of theory of elasticity with axial sym-
metry are two-dimensional, since, due to the independence of the geometrical shape
of the punch and the boundary conditions on the angular coordinate, the character-
istics of the stress-strained state of the base will be determined only by virtue of the
radial r and vertical z coordinates. The application of the boundary-element method
with a special cyclic discretization of the contact surface additionally reduces the
geometrical dimensionality of the problem: the contact stresses are to be determined
only for the points of the broken line, approximating the punch generatrix. Conse-
[email protected]
quently the application of the boundary-element method, enabling the possibility of
further reduction of dimensionality, reduces the axisymmetric contact problem to a
one-dimensional one.
2.5 Contact Problems for Rigid Punches Located on the Elastic Base Surface 119

The time, required for the solution of the contact problems with axial sym-
metry, will be essentially shortened in comparison with the problems in gen-
eral spatial formulation, due to the two reasons. First, the dimensionality of the
resolving systems of algebraic equations is reduced, and for these systems, in
case Gauss elimination method being used, the solution time is proportional to
N3 (N is the number of the equations). Besides, due to the angular (cyclic) sym-
metry the time for the computation of the influence matrix coefficients will be
reduced by factor of 3Q/2. In practice, in our calculations of contact problems
with axial symmetry, based on the proposed reduced formulation, with the number
of the boundary elements of about 400, the total computation time was reduced in
average by factor of 20 in comparison with that required in case of application of
the spatial scheme of the most general way. It should be also mentioned that by
increasing the number of the boundary elements along the angular coordinate one
can increase the accuracy of the numerical solution of contact problems with axial
symmetry without increasing the dimensionality of the system of resolving algebraic
equations, increasing only the computation time for the formation of the influence
matrix, i.e. without extending the computer RAM resources.
Finally we note that in a great many of studies, devoted to the solution of axisym-
metric problems of theory of elasticity for the finite-size bodies, using the Kelvin’s
solution, the implementation of the boundary-element method implies a procedure
of analytical integration over the angular coordinate (See, e.g., the references in
[7, 10, 17]). This results in complete elliptical integrals of the first and second order,
which, in turn, for the sake of convenience of further numerical calculations, are
presented in the form of an expansion over polynomials [1]. A number of authors
note that in the axisymmetric case the integral cores have a rather complicated form
and the related calculations are cumbersome. In [29], devoted to the axisymmetric
contact problem for a rigid deepened punch, integration of the Mindlin’s solution
over the angular coordinate is performed numerically. Evidently, such an algorithm
can be efficient only for the punches of cylindrical shape when the radial coordi-
nate of the contact surface points remains constant. The approach developed here
seems more convenient since it enables the solutions for both axisymmetric prob-
lems and problems of general spatial formulation to be obtained, based on the same
computation algorithm of formation of influence matrices, without loss in accuracy.

2.5 Contact Problems for Rigid Punches Located on the Elastic

Base Surface
Considering a spatial contact problem for a rigid punch and an elastic base, we
analyze a limiting case when the punch is not deepened, i.e. is located on the elas-
tic base surface (Fig. 2.8). We also assume the punch bottom to be flat; then the
[email protected]
contact domain F will be a part of the half-space surface. Hence, in the boundary
integral equations of the general spatial contact problem (2.9), (2.10) one should
imply z = ζ = 0, and the fundamental Mindlin’s solution to be transformed into a
120 2 Static Analysis of Contact Problems for an Elastic Half-Space

Fig. 2.8 Calculation scheme

for the spatial contact
problem for a non-deepened
rigid punch with a flat
bottom, located at the surface
of the elastic base

combined Boussinesq-Cerruti solution. The account of the above statements enables

the boundary integral equations of the spatial problem of theory of elasticity for a
flat rigid punch, contacting an elastic half-space on its surface, to be written in the
following form:

⎧   
⎪
⎪ (1) (K, N) + p (N)V (1) (K, N) + p (N)W (1) (K, N) ds =  + η · ψ ,
⎪
⎪ p1 (N)U 2 3 1 3
⎪
⎪
⎪
⎪ F
⎪
⎪  
⎪
⎨ 
p1 (N)U (2) (K, N) + p2 (N)V (2) (K, N) + p3 (N)W (2) (K, N) ds = 2 − ξ · ψ3 ,
⎪
⎪ F
⎪
⎪
⎪
⎪   
⎪
⎪
⎪
⎪ p (N)U (3) (K, N) + p (N)V (3) (K, N) + p (N)W (3) (K, N) ds =  − η · ψ + ξ · ψ .
⎪
⎩
1 2 3 3 1 2
F
(2.47)

To make the system (2.47) closed we take into account six static equilibrium
equations whose form is more simple than in Eqs. (2.10):
⎧   
⎪
⎪ = = p3 (N)ds = P3
⎪
⎪ p1 (N)ds P 1 , p 2 (N)ds P 2 ,
⎪
⎨ F F F
  
⎪
⎪ 
⎪
⎪ p3 (N)x2 ds = M1 , p3 (N)x1 ds= − M2 , p2 (N)x1 − p1 (N)x2 ds = M3 .
⎪
⎩
F F F
(2.48)

In Eqs. (2.47), (2.48) the following notations are assumed: p1 (N) = px (x, y),
p2 (N) = py (x, y), p3 (N) = pz (x, y) are the sought contact stress functions, acting
in a flat domain F; U(k) , V(k) , and W(k) (k = 1, 2, 3) are the components of the
[email protected]
combined Boussinesq-Cerruti solution for the displacements of the half-space sur-
face due to unit concentrated forces, acting in the direction of the coordinate axes
xk (k = 1, 2, 3).
2.5 Contact Problems for Rigid Punches Located on the Elastic Base Surface 121

In comparison with the initial system of Eqs. (2.9), (2.10) for the general spa-
tial contact problem for a deepened punch, the integral equation system (2.47) and
(2.48) is simpler. First, the contact pressure functions have to be determined in a flat
domain and, hence, both the discretization of the contact domain and the interpola-
tion of the discrete numerical results at the analysis of the contact problem solution
will be easier than for a curved surface. Second, the integration procedure at the
formation of the influence matrix of the resolving system of algebraic equations
of the boundary-element method will require less time than in the case of the full
Mindlin’s solution. Nevertheless, the total computation time for the solution of the
spatial contact problem for a non-deepened punch will be of the same order as for
the solution of the contact problem in the most general spatial formulation.
Then we consider two important special cases of spatial loading of non-deepened
punches, for which the integral equation system (2.47), (2.48) is essentially simpli-
fied, following separately the force balance at contact deformation in the vertical
and horizontal planes, respectively. The first case corresponds to the punch inden-
tation by a vertical force and pull-out torques, acting with respect to the coordinate
axes in the punch base plane; in the contact domain only vertical (normal) stress
exists. In the second case, the punch, linked to the half-space, undergoes a torque;
only horizontal (tangential) stresses act on the contact surface.

2.5.1 Indentation of a Punch with a Flat Smooth Base

into an Elastic Half-Space

Let the punch base be smooth, i.e. in the contact domain F tangential stress is zero.
In this case the external load system will not include horizontal forces P1 , P2 and
torque M3 leading to the tangential stress in the contact domain. The punch will
be indented into the base by a vertical force P3 and torques M1 , M2 , rotating the
punch around the axes Ox1 and Ox2 . Besides, since no friction between the punch
bottom and the base surface is assumed, then only one of the three displacement vec-
tor components will be varied, namely the vertical one. Thus, in the system (2.47),
(2.48) one should imply 1 = 2 = 0, ψ3 = 0, P1 = P2 = M3 = 0, p1 =
p2 = 0. Then the interaction of the punch with the base will be characterized by
the function of vertical stress (contact pressure) p3 = p(x, y) and vertical displace-
ment of the points of the flat punch bottom W = 3 −ηψ1 + ξψ2 . The above
assumptions will reduce the integral equation system of the spatial contact problem
for a smooth punch with a flat base to a single equation


p3 (N)W (3) (K, N) = 3 − η · ψ1 + ξ · ψ2 . (2.49)
F
[email protected]
In the equilibrium equation system (2.48) three conditions, corresponding to the
horizontal force balance, are fulfilled identically, and the other three are given by
122 2 Static Analysis of Contact Problems for an Elastic Half-Space
⎧
⎪
⎪ p3 (N)ds = P3 ,
⎪
⎪
⎪
⎨ F
  (2.50)
⎪
⎪
⎪
⎪ p (N)x ds = M , − p3 (N)x1 ds = M2 .
⎪
⎩
3 2 1
F F

The system (2.49), (2.50) can be written in the form, corresponding to the nota-
tions, established in the literature

(ξ , η)ω(x, y, ξ , η)dξ dη = Wc + ψx · (x − xc ) + ψy · (y − yc ), (2.51)
p
 
p(x, y)dxdy = P, p(x, y)ydxdy = P · yc − Mx ,
F F
 (2.52)
p(x, y)xdxdy = P · xc + My
F

where

F is the domain of the punch contact with the elastic base;

p(x, y) = p3 (x, y, 0) is the sought contact pressure function,
W(x,y) = Wc + ψx · (x − xc ) + ψy · (y − yc ) is the displacement of the point
N(x,y) of the contact surface of the punch and the elastic base,
Wc is the vertical displacement of the punch centre (xc , yc ) (point of application
of the external forces and torques),
ψ x and ψ y are the punch slopes with respect  to the coordinate axes,
ω(x, y, ξ , η) = W (3) (K, N) = (1 − ν 2 )/π E (x − ξ )2 + (y − η)2 is the influence
function (Boussinesq solution),
P, Mx , My are the external vertical force and tilting moments.

Equation (2.51) expresses the fact that the punch displacement W(N) is numeri-
cally equal to the sum of works from the contact forces p (x, y) in the basic state on
the vertical displacements of the base surface

(1 − ν 2 ) 1
ω (x, y, ξ , η) = · , R= (x − ξ )2 + (y − η)2
πE R
of the auxiliary state, resulting from the action of a unit vertical concentrated force.
Thus, the spatial contact problem on the indentation of a rigid punch with a
smooth flat bottom into an elastic half-space is reduced to the finding from a 2-D
[email protected]
integral equation (2.51) of the parameters Wc , ψ x and ψ y , determining the punch
position, and the contact pressure function p(ξ ,η) over its bottom, satisfying the
equilibrium conditions (2.52).
2.5 Contact Problems for Rigid Punches Located on the Elastic Base Surface 123

While characterizing the integral equation (2.51), note that, in spite of its simple
form, the general studies of the corresponding mixed problem of theory of elastic-
ity are far from being finished [4, 15, 30]. A closed solution of the integral equa-
tion (2.51), except for its axisymmetric case, has been found only for elliptical and
ring-shaped contact domains. Construction of approximate solutions of the integral
equation (2.51) for contact domains of a rather general shape is a serious computa-
tional problem. Our overview of the main solution methods for the spatial contact
problem for flat complex-shaped punches, interacting with elastic, spatially nonho-
mogeneous bases, is presented in Sect. 4.1.
We obtain a finite-measure algebraic analogue of the integral equation system
(2.51), (2.52) of the contact problem under consideration using the boundary ele-
ment method. For this purpose first we approximate the contact surface by boundary
elements of, in general, polygonal shape. This can be done using practically any of
the known considerable amount of mesh generators for arbitrary flat domains, used
in finite-difference, finite- and boundary-element methods and having their own spe-
cific features. The proposed and applied here algorithms of automatic mesh of flat
surfaces of a rather general shape are considered in Sect. 3.3. They are rather eco-
nomical (require small computation time in comparison with the numerical solution
of the problem, for which the mesh is built) and are capable of controlling the mesh
geometry using easily treated “free” parameters (condensation degree, maximal and
minimal boundary element size etc.).
In the simplest version of the boundary-element method the points of appli-
cation of the unit concentrated forces are located in the gravity centres of the
boundary elements and a piecewise contact approximation of the contact pres-
sures p(ξ , η) = const within each boundary element is assumed. Then, instead of
Eqs. (2.51) and (2.52), a linear algebraic equation system with (m + 3) unknowns is
obtained:
⎧
⎪ p1 δi1 + p2 δi2 + ... + pm δim − Wc − ψx · (xc − xi ) − ψy · (yc − yi ) = 0, i = 1, m,
⎪
⎪
⎪
⎪
⎨ p1 s1 + p2 s2 + ... + pm sm = P,
⎪
⎪ p1 s1 x1 + p2 s2 x2 + ... + pm sm xm = P · xc + My ,
⎪
⎪
⎪
⎩
p1 s1 y1 + p2 s2 y2 + ... + pm sm ym = P · yc − Mx ,
(2.53)
Here pi (ξ i , ηi ) are the sought contact stresses for the boundary elements
(i= 1, . . ., m),

δij = ω (x, y, ξ , η) dξ dη (2.54)
Fj

the vertical displacement of the base surface in the point (xi , yi ) coinciding with
[email protected]
the gravity centre of the i-th element, due to a unit load, uniformly distributed over
the domain Fj of the j-th element, Si (i = 1, 2,. . .m) are the areas of the flat bound-
ary elements whose combination approximates the contact domain F.
124 2 Static Analysis of Contact Problems for an Elastic Half-Space

For numerical solution of the system (2.53) one can take the advantage of its
matrix form

A·Z = B (2.55)

where
 
Dm×m Cm×3
A= – is a block matrix of the size (m+3)×(m+3),
(T3×m
( 0
D = (δij ( is the influence matrix, i, j = 1,. . .m; C and T are rectangular
matrices

⎛ ⎞
1 x1 − xc y1 − yc ⎛ ⎞
⎜ ⎟ s1 s2 ... sm
⎜ 1 x2 − xc y2 − yc ⎟ ⎜ ⎟
C = −⎜ ⎟ , T = ⎜ s1 · x1 s2 · x2 ... sm · xm ⎟ ;
⎜ ⎟ ⎝ ⎠
⎝ ... ... ... ⎠
s1 · y1 s2 · y2 ... sm · ym
1 xm − xc ym − yc

Z and B are column matrices of the size (m+3):

 T  T
Z = p1 , p2 ...pm ; Wc , ψx , ψy , B = 0,0,...0; P; P · xc − My ;P · yc + Mx .

The discretization of the contact domain (the punch bottom) will result in the
location of all the boundary elements in the same plane (the half-space surface
z = 0). Therefore, in order to increase the computation algorithm efficiency and
accuracy we use in Eq. (2.54) the same procedure of analytical calculation of both
singular (i = j) and regular (i = j) surface integrals over an arbitrary boundary ele-
ment with a polygonal contour. This is achieved on the base of algebraic assembling
(according to the choice of sign while moving along a closed circuit) of singu-
lar integrals over triangles with a singularity in the concentrated force application
point. For example, for a triangular boundary element with the apices A, B, and C
and an arbitrary point K(ξ , η), located outside (Fig. 2.9a) and inside (Fig. 2.9b) the
integration domain the same sum

(a) (b)

Fig. 2.9 Geometrical scheme

of the integration domain in
case the point K of
[email protected]
application of a unit
concentrated force being
located (a) outside and (b)
inside the boundary element
2.5 Contact Problems for Rigid Punches Located on the Elastic Base Surface 125
   
= + + .
ABC KAB KBC KCA
is used, corresponding to the positive (anti-clockwise) direction of moving along the
circuit.
The exact calculation of integrals of the form


ω(x, y, ξ , η)dξ dη,

over a triangular domain F = M1 M2 M3 , when the integrand has a singularity in

the first apex, is reduced to the above obtained integral I1 (Sect. 2.3). According to
Eq. (2.23), in a general case, when the triangle is arbitrarily oriented in the elastic
half-space, the latter integral can be found by a combination of logarithmic and
trigonometric functions. In our case, when the punch is located on the half-space
surface (Z1 = Z2 = Z3 = 0), Eq. (2.23) remains unchanged, and the expressions for
its parameters are considerably simplified, having the form

 
D= (X∗ − X1 )2 + (Y∗ − Y1 )2 , R̂ = (X3 − X2 )2 + (Y3 − Y2 )2 ,
X∗ = X2 − (X3 − X2 ) · p, Y∗ = Y2 − (Y3 − Y2 ) · p,
p = ((X2 − X1 )(X3 − X2 ) + (Y2 − Y1 )(Y3 − Y2 )) /R̂2 .
The simplest form of the system (2.53) is achieved for a circular punch at central
(xc = 0, yc = 0) loading (axisymmetric contact problem). Due to the axial symmetry
(Mx = My = 0), (ψ x = ψ y = 0), hence instead of Eq. (2.53) one obtains

*
p1 δi1 + p2 δi2 + ... + pm δim − Wc = 0, i = 1, m,
(2.56)
p1 s1 + p2 s2 + ... + pm sm = P.

The reduction of the system (2.56) is performed similarly to the case of the
axisymmetric problem of a deepened punch indentation (Sect. 2.4.1), assuming
cyclic discretization of the circular (or the ring-shaped) contact domain. A typ-
ical example of such discretization by means of a regular (ϕ = const) grid of
boundary elements whose nodes are obtained by intersection of rays and concentric
circles in a polar coordinate system, is shown in Fig. 2.10. Afterwards the system
(2.56) is transformed in the following way. In each line of this system there are
several terms which, due to the cyclic symmetry, contain the same contact force val-
ues, corresponding to the same boundary element number along the radius (or the
number of the ring-shaped layer). Combine all such terms in the line, correspond-
ing to each ring-shaped layer. If L is the number of the boundary elements along
[email protected]
the punch radius, then the algebraic analogue of the integral equation system on the
contact problem for central indentation of a circular (or ring-shaped) punch with a
flat smooth bottom takes its most simple form
126 2 Static Analysis of Contact Problems for an Elastic Half-Space

(a) (b)

Fig. 2.10 Cyclically symmetrical discretization of circular (a) and ring-shaped (b) contact
domains


L 
L
P·L
Aij · pj − Wc = 0, i = 1,L, pj · sj = (2.57)
m
j=1 j=1

Here the values Aij are found using the influence coefficients δij (i = 1,. . ., L;
j = 1,. . ., m) from the formula


m/L
Aij = δi,j+L(k−1) .
k=1
It is evident that the system (2.57) can be also easily obtained from the system
(2.37) as a degenerate case of a deepened punch with a smooth bottom W (3) (K, N) =
ω(x, y, ξ , η),ρ τ = 0.

2.5.2 Torsion of an Elastic Half-Space by a Rigid Punch

Let the system of the external load on the punch include only a torque M3 , resulting
in the formation of tangential stresses in the contact domain. Neither other torques
M1 , M2 , nor both vertical P3 and horizontal P1 , P2 external forces are assumed. In
this case the punch will rotate with the elastic base, and its position is characterized
only by the rotation angle ψ 3 around the Oz axis. In the system (2.47) let 1 =
2 = 3 = 0, (P1 = P2 = M1 = M2 = 0). Due to the absence of vertical loads
and tilting moments it can be reasonably assumed that the torsion does not affect
the pressure distribution below the punch bottom, hence p3 = 0 [19]. Contrary to
[email protected]
the above case of the punch indentation (Sect. 2.5.1), now the contact interaction of
the punch with the base will be characterized by two functions of contact tangential
stress p1 = px (x, y) and p2 = py (x, y). The integral equations (2.47) of the spatial
2.5 Contact Problems for Rigid Punches Located on the Elastic Base Surface 127

contact problem of torsion for the case of a flat-bottom punch are reduced to the
following system:
⎧   
⎪
⎪ (1)
+ (1)
ds = η · ψ3 ,
⎪
⎪ p1 (N)U (K, N) p2 (N)V (K, N)
⎪
⎨ F
   (2.58)
⎪
⎪
⎪
⎪ p (N)U (2)
(K, N) + p (N)V (2)
(K, N) ds = −ξ · ψ3 .
⎪
⎩
1 2
F

As follows from Eq. (2.48), the system (2.58) is made closed by a single (the
others are fulfilled identically) equilibrium equation, given by


p2 (N)x1 − p1 (N)x2 ds = M3 . (2.59)


In Eqs. (2.58) and (2.59) p1 (N) = px (x, y) and p2 (N) = py (x, y) are the sought
functions of tangential stresses acting in the contact domain F;U(k) and V(k) (k =
1, 2) are the components of the Cerruti solution for the half-space surface displace-
ments under unit concentrated horizontal forces in the direction of the Ox and Oy
axes, respectively.
Having introduced commonly used notations, we present the system (2.58),
(2.59) in a more convenient form for the subsequent analysis:
⎧ 
⎪ 
⎪
⎪ p1 (ξ ,η) 1 (x,y,ξ ,η) + p2 (ξ ,η) 2 (x,y,ξ ,η) dξ dη = y · ψ,
⎪
⎪
⎨ F
 (2.60)
⎪
⎪ 
⎪
⎪ p1 (ξ ,η) 2 (x,y,ξ ,η) + p2 (ξ ,η) 3 (x,y,ξ ,η) dξ dη = −x · ψ,
⎪
⎩
F


p2 (x,y)x − p1 (x,y)y dxdy = M (2.61)


where p1 (x, y) and p2 (x, y) are the sought contact tangential stress functions, i (x,
y, ξ , η), (i = 1, 2, 3) are the Cerruti displacement functions:
 
1+ν x12 1
1 (x,y,ξ ,η) = U (1) (K, N) = ν 3 + (1 − ν) ,
πE R R
ν(1 + ν) x1 y1
= U (2) (K, N) = V (1) (K, N) =
2 (x,y,ξ ,η) ,
[email protected] πE R3
 
1+ν y2 1
3 (x,y,ξ ,η) = V (K, N) = ν 13 + (1 − ν)
(2) ;
πE R R
128 2 Static Analysis of Contact Problems for an Elastic Half-Space

x1 = x – ξ , y1 = y – η, R = x12 + x22 ; M is the external torque rotating the punch
in the horizontal plane by an angle ψ.
Thus, the spatial contact problem on surface torsion of an elastic base by a rigid
flat-bottomed punch at full coupling is reduced to the following: from the two-
dimensional integral equation system (2.60) one should find the angle ψ and two
functions p1 (x, y) and p2 (x, y), satisfying the integral equilibrium condition (2.61).
The exact solutions of the mixed problem of theory of elasticity (2.60) and (2.61)
are known only for circular and elliptical punches [19]. Approximate solutions of
contact problems on torsion of elastic bases of various types have been obtained
by a number of authors [2, 8, 11, 13, 23, 25, 27, 28, 33], but in most cases for cir-
cular and ring-shaped punches. For a rectangular domain, the punch rotation angle
due to a given torque was determined by Mozhevitinov [22]; however, the con-
tact problem was not solved, but a linear distribution of tangential stresses, depend-
ing on the contact point distance from the rotation axis, was suggested. As will be
shown below, the numerical solution of the integral equation system (2.60), (2.61)
for punches with complex-shaped bottom can be efficiently performed using the
boundary-element method in its direct formulation.
If the discretization of the contact domain F is performed by any of the available
methods, then a finite-measure algebraic analogue of the system (2.60), (2.61) can
be readily obtained in the approximation of piecewise constant functions of contact
tangential stresses on the boundary elements similarly to how it has been made in
Sect. 2.4.2. We write the resolving solutions of the torsion contact problem under
consideration in a discrete form:

⎧ M  
⎪ '
⎪
⎪ [p (Q ) (P , Q)dF + p (Q ) 2 (Pi , Q)dF] = yi · ψ,
⎪
⎪
1 q 1 i 2 q
⎪
⎪ q=1
⎪
⎪ Fn Fn
⎪
⎪    
⎨' M
[p1 (Qq ) 2 (Pi , Q)dF + p2 (Qq ) 3 (Pi , Q)dF] = −xi · ψ, (2.62)
⎪
⎪
⎪
⎪
q=1
Fn Fn
⎪
⎪
⎪
⎪
⎪
⎪ '
M 
⎪
⎩ p2 (Pq )xq − p1 (Pq )yq sq = M, q = 1,2,...,m.
q=1

Here the following notations are used: p1 (Qq ), p2 (Qq ) are the averaged values
of the corresponding radial and vertical contact stress within the q-th boundary ele-
ment, Pi are the collocation points (the boundary element gravity centres), Sq is
the q-th boundary element area, m is the total number of the boundary elements.
In the matrix notation the algebraic analogue Eq. (2.62) of the integral equation
system is given by
[email protected]

A·Z = B (2.63)
2.5 Contact Problems for Rigid Punches Located on the Elastic Base Surface 129
 
F2m×2m G2m×1
where A = is a square block matrix of the order (2m + 1),
H1×2m 0

G2m×1 = − (y1 , − x1 ,y2 , − x2 ,...,ym , − xm )T ,

ψ = π q0 /(4G), τ (r) = q0 r(a2 − r2 )−1/2 , (q0 = 3 M/4π a3 ).
Z and B are column vectors of the size (2M+1):

Z = (p1 (P1 ) ,p2 (P1 ) ;p1 (P2 ) ,p2 (P2 ) ;...p1 (Pm ) ,p2 (Pm ) ;ψ)T ,
B = (0,0;0,0;...;M)T ;
 
fiq = 1 (Pi , Q)dF = 1 (xi , yi , ξ , η)dξ dη,
Fq Fq
 
giq = 2 (Pi , Q)dF = 2 (xi , yi , ξ , η)dξ dη,
Fq Fq

hiq = 3 (Pi , Q)dF =f 3 (xi , yi , ξ , η)dξ dη
Fq

are the coefficients of the influence matrix F2m×2m , i, q = 1, 2,. . ., m.

It is evident that the structure of the given matrix representation of the algebraic
analogue of the boundary contact problem with respect to the values of ψ 3 and
p1 (Pq ), p2 (Pq ) (q = 1, 2,. . .m) is similar to the case of the half-space torsion by a
deepened punch in the shape of a rotation body considered above in Sect. 2.4.2
with the only difference that the blocks in the influence matrix F2m×2m for a flat-
bottomed punch are symmetrical. The coefficients fiq , giq , hiq of the influence matrix
for the Cerruti solution used here are determined analytically using the primitives
for the integrals I1 , I4 , I5 , I7 , obtained in Sect. 2.3 in a closed form. Here, similarly
to the case of the punch indentation (Sect. 2.5.1), for the calculation of both singular
[email protected]
and regular integrals the same method is used, based on the algebraic assembling of
integrals over triangles, resting on the sides of flat polygon-type boundary elements.
The formulae of the primitives (2.26) for I1 , I4 , I5 , I7 remain unchanged, and their
130 2 Static Analysis of Contact Problems for an Elastic Half-Space

parameters are determined from simplified 2-D formulae due to the zero values of
the node z-coordinates for all the boundary elements.
We finish the consideration of a half-space torsion by a rigid flat-bottomed punch
by a boundary-element formulation of the problem for the circular (or ring-shaped)
contact domain.
First we perform cyclic discretization into boundary elements (Fig. 2.10). Then
we take into account the axial symmetry of the problem, for which only tangential
stresses are possible. Similarly to the above case, we locate the points of applica-
tion of unit concentrated forces in the boundary-element gravity centres and assume
tangential forces pt = τ, acting in the direction, normal to the radius, to be con-
stant within each boundary element. The algebraic analogue of the integral equation
system of the contact problem of a rigid rotation under the torque of a circular
(or ring-shaped) punch, bound to the half-space, will be obtained from the system
(2.45) as a limiting case of the deepened punch in the shape of a rotation body
(Ut (P, Q) = ˜ (x, y, ξ , η) pr = pz = 0) in the form
(1)

⎧
⎪
⎪ 'L
⎪
⎪ B · τ − yi · ψ = 0, i = 1, L
⎨ j=1 ij j
(2.64)
⎪
⎪ '
L
⎪
⎪ τj · rj · sj = M·L
⎩ m
j=1

where
 
'
m/L
Bij = λi, j+L(k−1) ; λqn = ˜ (Pq , Q)dF = ˜ (xq , yq , ξ , η)dξ dη;
k=1
Fn Fn

˜ (x, y, ξ , η) = − 1 (x, y, ξ , η) · cos ϕ + 2 (x, y, ξ , η) · sin ϕ;

n = 1,. . .,m; m is the total number of boundary elements in the contact domain,
q = 1,. . ., L; L is the number of boundary elements along the radial direction.
Note that the problem of torsion of an elastic half-space by a round punch of the
radius a has an exact solution [19], enabling the rigid rotation ψ of the punch and
the distribution of tangential forces τ(r) below the punch to be obtained:

ψ = π q0 /(4G), τ (r) = q0 r(a2 − r2 )−1/2 (q0 = 3 M/4π a3 ).

An approximate analytical solution of the problem of an elastic half-space torsion

due to a ring-shaped punch rotation was obtained in [8].
In the algebraic analogues (2.53), (2.62), and (2.64) the influence matrix coeffi-
cients are presented for the classic contact model of an elastic homogeneous half-
space. However, all the formulations of spatial contact problems for flat-bottomed
punches considered remain valid also for non-classic bases, for which the influ-
[email protected]
ence functions exist or can be obtained. The algorithm of solution of such contact
problems remains the same, except for the calculation of the influence matrix coeffi-
cients. The literature analysis shows that the great majority of the influence functions
for non-classic elastic bases can be given by
References 131

i (x,y,ξ ,η) = 0 (x,y,ξ ,η) + ∗ (x,y,ξ ,η)

where 0 (x, y, ξ , η) is the influence function for a homogeneous elastic half-space

(a combined Boussinesq-Cerruti solution), and ∗ (x, y, ξ , η) is an additional (non-
classic) component, taking into account nonhomogeneity, anisotropy, lamination
and other mechanical characteristics of the bases. The additional components of
the influence functions ∗ (x, y, ξ , η) are regular and do not introduce any princi-
pal difficulties in numerical integration. Here, similarly to the case of the Mindlin’s
solution integration, a numerical-and-analytical procedure, described in Sect. 2.3
and consisting in analytical integration of the components of the influence functions
0 for the elastic half-space and numerical integration of the non-classic component
∗ , appears to be efficient.
Concerning the above boundary-element formulations of contact problems for a
round (or ring-shaped) punch (2.57) and (2.64), it should be noted that in spite of
the existing corresponding exact solutions, these algebraic analogues are of great
importance for the general development of the proposed version of the numerical
boundary-element method. First, they serve as a convenient tool for numerical algo-
rithm testing. Second, these formulations remain unchanged (except for the tech-
nical procedure of the influence matrix determination) for various spatial contact
models being used and, hence, are a universal tool for the studies of spatial con-
tact interaction of the simplest type (a centrally loaded round punch). The latter
case is important for the identification of the parameters of the existing and con-
stantly elaborated influence functions. The formulations of Eqs. (2.57) are (2.64) are
undoubtedly helpful as well due to the fact they are valid for practically any impor-
tant contact problem with axial symmetry, namely for a flat ring-shaped punch. The
numerical solution of this problem using the proposed boundary-element algorithms
requires, in comparison with the round punch problem, only an obvious slight mod-
ification at the contact domain discretization under the condition of the cyclic sym-
metry being preserved (Fig. 2.10b). Not so many solutions have been obtained for
the ring-shaped punch problem (much less than for the round punch, see Sect. 4.1);
however, the practical interest to it is rather high [18, 32].
In spite of the simplicity and convenience of the given boundary-element for-
mulation of the contact problems with axial symmetry (2.57) and (2.64), one should
take into account that they remain valid only for the base models, for which the influ-
ence functions are symmetrically-differential, i.e. when (x,y,ξ ,η) = (x−ξ ,y−η).
Otherwise, for the numerical solution of contact problems for round and ring-shaped
punches the general boundary-element algorithms of Eqs. (2.53) and (2.62) should
be used.

References
[email protected]
1. Abramowitz M, Stegun I A (1972) Handbook of mathematical functions with formulas,
graphs, and mathematical tables, 10th edn. Dover, New York
2. Aizikovich S M (1978) Torsion of a nonhomogeneous half-space by a round punch. In: Cal-
culation of shells and plates. Rostov University, Rostov, pp. 156–169 (in Russian)
132 2 Static Analysis of Contact Problems for an Elastic Half-Space

3. Aleksandrov A Ya, Solovyov Yu I (1978) Spatial problems of theory of elasticity (application

of methods of theory of complex-variable functions). Nauka, Moscow (in Russian)
4. Alexandrov V M, Kovalenko E V (1986) Problems with mixed boundary conditions in con-
tinuum mechanics. Nauka, Moscow (in Russian)
5. Aleynikov S M, Ikonin S V (1989) Effect of depth factor on the foundation displacements.
Bases and Foundations. Budivelnyk, Kyiv 22:6–8 (in Russian)
6. Aleynikov S M, Ikonin S V, Rashutina G N (1993) Numerical solutions of nonlinear spatial
contact problems of soil mechanics for pile structures. In: Nonlinear soil mechanics. Proc 6th
Russ Conf St Petersburg, vol. 1, pp. 155–160 (in Russian)
7. Banerjee P K, Butterfield R (1981) Boundary element methods in engineering science.
McGraw-Hill, New York
8. Borodachev N M, Borodacheva F N (1966) Torsion of an elastic half-space due to a ring-shape
punch rotation. Inzh Zh Mekh Tverd Tela (issue 1): 94–99 (in Russian)
9. Brebbia C, Telles J, Wrobel L (1984) Boundary element techniques. Springer-Verlag, Berlin
10. Brebbia C A, Walker S (1980) Boundary element techniques in engineering. Newnes-
Butterworths, London
11. Chistyak V I, Ryabokon T N (1988) On the solution of the problem of torsion of a nonhomoge-
neous elastic layer by a punch. In: Stability and strength of structural elements. Dnipropetro-
vsk State University, Dnipropetrovsk, pp. 56–64 (in Russian)
12. Cruse T A (1969) Numerical solution in three dimensional elastostatics. Intern J Solids Struct
5: 1259–1274
13. Dudinskiy V I (1984) Torsion of an elastic nonhomogeneous half-space by a circular die.
Intern Appl Mech 20: 913–917
14. Galin L A (1980) Contact problems of theory of elasticity and viscoelasticity. Nauka, Moscow
(in Russian)
15. Galin L A (ed) (1976) Development of theory of contact problems in the USSR. Nauka,
Moscow (in Russian)
16. Goldshtein R V (1978) Supplement. On the issue of application of boundary integral equa-
tion method for solving problems of mechanics of solid media. In: Mechanics. New in for-
eign science. Boundary integral equation method. Computational aspects and applications in
mechanics. Mir, Moscow, pp. 182–209 (in Russian)
17. Hartmann F (1989) Introduction to boundary elements: Theory and applications. Springer,
Berlin/Heidelberg/New York
18. Instructions on design of plate foundations for tower-type skeleton buildings and structures
(1984) Gersevanov NIIOSP, Stroyizdat, Moscow (in Russian)
19. Johnson K L (1985) Contact mechanics. Cambridge University Press, Cambridge
20. Kovneristov G B (1962) Integral equations of the contact problem of theory of elasticity for
deepened punches. Works of Kyiv Engineering Institute 20: 200–213 (in Russian)
21. Loytsyanskii L G, Lurye A I (1982) Course of theoretical mechanics. Vol. 1: Statics and
kinematics. Nauka, Moscow (in Russian)
22. Mozhevitinov A L (1989) Rotation of a rigid foundation due to a spin torque. In: Bases of
power structures. Izv VNIIG Vedeneyeva 215: 76–80 (in Russian)
23. Naumov Yu A, Chistyak V I (1973) Torsion of a nonhomogeneous rigid layer by a punch. In:
Dnipropetrovsk State University, Dnipropetrovsk, pp. 12–21 (in Russian)
24. Nowacki W (1970) Teoria sprezystosci. PWN, Warszawa
25. Petrishin V I (1988) Torsion of a multilayer base by a ring-shaped punch. In: Stability and
strength of structural elements. Dnipropetrovsk State University, Dnipropetrovsk, pp. 96–99
(in Russian)
[email protected]
26. Polozhiy G I (1963) On the boundary problems of axisymmetrical theory of elasticity. Method
of p-analytical functions of complex variable. Ukr Math Zh 15: 25–45 (in Russian)
27. Protsenko V S (1968) Twisting of a generalized elastic half-space. Intern Appl Mech 4
(issue 6): 56–58
References 133

28. Rostovtsev N A (1955) To a problem of torsion of an elastic half-space. Prikl Mat Mekh 19
(issue 1): 55–60 (in Russian)
29. Shishov O V (1971) Contact problem for axisymmetrical deepened punches. In: Resistance
of materials and theory of structures. Budivelnyk, Kyiv 13: 60–66 (in Russian)
30. Shtayerman I Ya (1949) Contact problem of theory of elasticity. Gostekhizdat, Moscow (in
Russian)
31. Solyanik-Krassa K V (1987) Axisymmetrical problem of theory of elasticity. Stroyizdat,
Moscow (in Russian)
32. Sorochan E A, Trofimenkov Yu G (ed.) (1985) Bases, foundations and underground structures:
Designer’s handbook. Stroyizdat, Moscow (in Russian)
33. Ufliand Ia S (1965) Survey of articles on the application of integral transforms in the theory
of elasticity. North Carolina State University
34. Ugodchikov A G, Khutoryanskiy N M (1986) Boundary-element method in mechanics of
deformable solids. Kazan State University, Kazan (in Russian)
35. Veryuzhskii Yu V, Ikonin S V, Savitskii V V (1982) Application of potential method in calcula-
tion of rigid foundations. In: Bases and foundations. Budivelnyk Kyiv 15: 21–25 (in Russian)
36. Veryuzhskiy Yu V (1978) Numerical methods of potential in some problems of applied
mechanics. Vyshcha Shkola, Kyiv (in Russian)
37. Vinnik A I, Vorona Yu V, Roytfarb I Z, et al. (1981) Calculation of triaxial stressed state
of structural elements using the potential method. In: Resistance of materials and theory of
structures. Budivelnyk, Kyiv 38: 77–82 (in Russian)
38. Yakimchuk D K, Kvitka A L (1979) Application of the boundary integral equation method
for solving spatial problems of static theory of elasticity. Preprint Inst. Problems of Strength,
Kyiv (in Russian)

[email protected]
[email protected]
Chapter 3
Computer Implementation
of Boundary-Element Algorithms

Abstract This chapter deals with practical implementation of the developed

numerical algorithms and substantiation of the reliability of the numerical solutions.
It presents the general characteristics and structure of the Rostwerk software pack-
age for investigating three-dimensional stress-strained states of elastic bases corre-
sponding to the interaction of foundation structures with soil under force factors of
general kind. Procedures for creating input databases are described in detail. Algo-
rithms and modules for automatic formation of boundary element grids in plane and
in space are presented. An original algorithm for triangulation of flat single – and
multiply connected domains, bounded by straight line segments or circle arcs, is
described. An algorithm of generation (according to the given triangulation) of dual
polygonal boundary element grids of Dirichlet cell type is considered. The created
object library of boundary element modules, partitioned into boundary elements,
enabling spatial discretization of complex-shaped surfaces of foundation structures,
is described. Specific features of solving the systems of linear algebraic equations
with asymmetric and close-packed matrices, arising in boundary element analysis,
are considered. For solving such systems by direct (Gauss type) methods a spe-
cial scaling procedure is applied, improving the conditioning of matrices for the
finite-dimensional algebraic analogue of a contact problem. The data about the
reliability of the numerical solutions are presented. The boundary-element method
accuracy and efficiency are demonstrated by the examples of the solved test prob-
lems for flat punches of circular, annular and polygonal shapes. Boundary-element
solutions for spatial contact problems concerning a rigid spherical inclusion and
a cylindrical deepened punch in an elastic half-space are obtained. The final part
of the chapter gives the results for numerical-and-analytical solution of the spa-
tial contact problem on impressing a deepened conical punch into an elastic half-
space. The method of determination of the deformation modulus from tests for
deepened conical indenters with different angles by static loading is substantiated
theoretically.

[email protected]

S.M. Aleynikov, Spatial Contact Problems in Geotechnics, Foundations of 135

Engineering Mechanics, DOI 10.1007/b11479_3,  C Springer-Verlag Berlin Heidelberg 2011
136 3 Computer Implementation of Boundary-Element Algorithms

3.1 Software for Solving Spatial Problems of Contact

of Foundations with Soil Bases
Determination of parameters of contact interaction of foundation structures with
soil still remains a complicated problem. This is related mostly to the two reasons:
(i) appearance of new foundation structure types accepting force and torque loads
of the most general spatial type; (ii) increased demands to the calculation method
accuracy.
In the first case, the existing calculation methods are often unable to take into
account a variety of features of operation of the new-type foundation structures. In
the second case, as a rule, one intends to obtain an economical solution at minimal
safety factor values.
The methods of foundation structure calculations from deformation, having been
used until the recent time, use simplified models what is explained, on one hand,
by the structural complexity of a number of foundation structures and, on the other
hand, by the foundation operation being considered without a detailed account of the
spatial effect from the above-foundation structures. However, the appearance and
development of such a powerful tool for the studies of interaction processes in the
“foundation + soil base” system as the boundary-element method, enables a qual-
itatively new level of calculations of foundation structures of complicated shapes
under the spatial type of external factors affecting the foundation from the load car-
rying structures to be achieved. The calculation model becomes able to take into
account a variety of features of real operation of the foundation structures: a com-
plicated shape of the bottom, the presence of extensions on the lateral surface and its
non-prismatic shape, nonuniformity of the mechanical properties of the base, action
of off-centre loads, impossibility of tensile functioning of soil, etc. Combination
of numerical and analytical approaches in the boundary-element method enables
rather efficient calculation algorithms to be obtained and the possibilities of mod-
ern computer facilities to be fully used. The boundary-element method enables the
calculation of foundations of various structure shapes from the base deformation
to be performed from the general methodological position. The calculator has no
need to make pre-calculation assumptions on the contact stress distribution charac-
ter, on the presence and localization of areas of the foundation uplifting from the
soil, neither to invoke any additional speculations taking into account the specific
features of the stress-strained state on the contact surface of the considered foun-
dation structure type and the soil. All the features of the contact stress distribution
for a foundation structure are revealed in the course of solving the problem using
the boundary-element method. Note that in order to obtain high accuracy of calcu-
lations, the time required for the computation, input data preparation, requirements
for the computer memory size for the boundary-element method are by two or more
orders of magnitude lower than in case finite-element or finite-difference methods
[email protected]
being used.
Basic characteristics of the Rostwerk software. In order to implement the
boundary-element algorithms for solution of spatial contact problems, discussed
in Sect. 2, we have worked out and developed a Rostwerk code, written using
3.1 Software for Solving Spatial Problems of Contact of Foundations with Soil Bases 137

FORTRAN-77 algorithmic language and oriented for personal computers of vari-

ous modifications.
We have chosen FORTRAN language since it is available in a great number of
operating systems and, being rather simple, still enables efficient program debug-
ging and creating object modules. It is also important that different versions of the
FORTRAN language are rather standardized and the transfer of programs from one
environment into another will not make great difficulties. Besides, many graphic
programs, being used in the input/output programs, are graphic extensions of FOR-
TRAN.
The Rostwerk software is intended for the studies of contact interaction of foun-
dation structures with linearly stressed bases under spatial static loading of a gen-
eral type. The software package enables both deep and shallow rigid foundations
of various structural types to be analyzed regarding the second limit state (from the
base deformations). The software takes into account the possibility of partial con-
tact of the foundation with soil as well as a variety of non-classical properties of soil
bases (lamination, non-uniform compressibility, dependence of mechanical proper-
ties on depth etc). The software enables one to calculate deformations of foundation
bases of practically any known shape whose contact surface can be (automatically
or manually) discretized into flat boundary elements. The external load acting on
the foundation should be reduced within the foundation section plane to six com-
ponents of main vectors of forces and moments along the OX, OY, and OZ axes of
the global Cartesian coordinate system. The results of the calculation enable one
to determine stress on the contact surface of the foundation and the soil, foun-
dation displacements and slopes as well as areas of the foundation uplifting from
the soil.
Processor time required for one version of calculation is not more than 10 min
for any PC type in case the contact surface being discretized into 400 boundary
elements with the account of tangential stress. In the case of calculation for shal-
low foundation structures, when one can reduce the problem to the determination
of contact pressures, the calculation time is shortened more than three times. The
calculation time is reduced even more (by an order of magnitude) in case the foun-
dation has the shape of a rotation body under axial loading (axisymmetric contact
problems).
Elaboration of software for boundary-element method was aimed at the following
five main objectives:

(1) the software package should be universal in the sense of its applicability to rigid
foundation structures of various type (both shallow and deep);
(2) the calculations should enclose interaction of the foundation structure with soil
bases characterized by spatial nonhomogeneity;
(3) preparation of the input data should be simple and unified;
(4) computation time required for the input data preparation and requirements for
[email protected]
the computer parameters should be minimized;
(5) data interchange between the software package and other software applications
should be available.
138 3 Computer Implementation of Boundary-Element Algorithms

The first requirement is achieved by the fact that the configuration of the
known relatively complex foundation structures can be without principal difficul-
ties described using typical polygonal, conical, spherical, and cylindrical shapes.
By locating the foundation structure fragments of the above shapes in the required
geometry, one can obtain different types of deep and shallow foundations, used in
engineering practice. The software performs the choice (assignment) of the foun-
dation structure shape with simultaneous mesh of its surface into boundary ele-
ments semi-automatically, what is described in detail in Sects. 3.3 and 3.4. Extensive
practical application of the software has shown the high accuracy of the boundary-
element description of the foundation structure geometry (the errors in the determi-
nation of contact surface area and volume of bodies even for the moderate number
of boundary elements lie within 0.01 ÷ 01%).
The second requirement for the software is satisfied by module assignment of
influence functions for elastic bases, being singular fundamental solutions of three-
dimensional theory of elasticity. The usage of an influence function is related to
the initial data input and can be performed in an explicit way without principal
difficulties for all known up to now contact models of elastic non-classical bases.
The formulations for some influence functions, most often occurring in practical
calculations, have a uniform character and have been considered above in Chap. 1.
In order to satisfy the third demand the input data preparation is organized as
follows. It is assumed that the input data are standardized independently of the
foundation structure type, shape, type of loading and mechanical (contact) model
of the soil base. Similarly the force data (external forces and moments) and geo-
metrical data (boundary-element node coordinates and the direction along the con-
tour) are given. Note that, due to the dicretization of only the contact surface
between the foundation and the soil, the input data size is almost by two orders
of magnitude smaller than for the finite-element or finite-difference methods which
are known to require detailed discretization of the foundation active area in the
soil base.
The fourth requirement is satisfied due to the creation of optimal algorithms
with analytical solutions being provided to maximally possible depth. This first of
all concerns the operation of integration of fundamental solutions for a homoge-
neous half-space (Kelvin’s kernels) at building up the influence coefficient matri-
ces. Numerical integration of singular solutions is known to require long compu-
tation times and is avoided in the elaborated algorithm (Sect. 2.3) due to the fact
that the software contains the analytical expressions for the results of integration
of singular functions over a flat area of boundary elements, arbitrarily oriented in
space. Besides, the economy of time and manual work for input data preparation is
achieved, on one hand, due to the relatively small size of the input data itself, and
on the other hand, due to the elaboration of software modules oriented at automatic
discretization of the contact surface of the foundation structure of a given type. A
considerable decrease of manual labour for input data preparation as well as the
[email protected]
computation time can be achieved for axisymmetric foundation structures as well
as for those possessing symmetry planes in the case the foundation being loaded
axially or under the action of a flat system of forces.
3.1 Software for Solving Spatial Problems of Contact of Foundations with Soil Bases 139

Finally, the fifth requirement is in the simplest way satisfied by a set of intermedi-
ate files being organized in the course of the program performance. These files will
become an input data flow for various specialized or general application software.
Structure and main stages of operation of the Rostwerk software. The software
package has a module structure with partial or complete overlay of common areas
to save the computer RAM at each stage of the problem solution. The package
consists of the main routine and more than forty subroutines which do not require
any additional mathematical software.
The block structural scheme of the software is shown in Fig. 3.1
The WERK module controls the sequence of performance of all steps of the
boundary-element algorithm.
The DAN module is responsible for the input data processing, performs discretiza-
tion of the contact surface into triangular and quadrangular flat boundary elements,
global enumeration of the boundary elements and local enumeration of their nodes
in counterclockwise direction (observed from the side of the external normal to the
contact surface of the foundation structure), as well as formation of nonyero com-
ponents of the load vector. The discretization unit is protected from user errors. In
case of incorrect input data the program does not continue running and shows an
error or a list of possible errors. Besides, in the course of the DAN module being
run, the coordinates of gravity centres of boundary elements, their areas, direction
cosines of external normal vectors are stored on a hard disk. As a result, a file of
input geometrical information is formed, whose graphical processing enables the
user to “view” the discretized surface of the foundation structure on the screen from
any point in space.
After processing the input data, the SBOR module is started, forming the influ-
ence coefficient matrix for the resolving system of linear algebraic equations with
respect to unknown contact stresses. The SBOR module itself has a multilevel struc-
ture, including subdivision of the boundary elements into triangular subelements
with a common vertex in the gravity centres, as well as units responsible for analyt-
ical and numerical integration. In the last case the process is performed adaptively
with the reduction of the number of the integration nodes at the increase of the
distance from the observation point to the centre of the integration domain.
The constructed influence coefficient matrix, complemented by matrix blocks
with the data on the static moments of the boundary elements with respect to the
coordinate axes, is written to the hard disk for on-run storage. This enables the
computation time to be saved at multivariant calculations.
In the course of operation of the GAUSS program module the constructed system
of linear algebraic equations is solved by Gauss elimination method with selection
of the main element in a row. Simultaneously the conditionality number is calcu-
lated, characterizing the closeness of the system matrix to degeneration, the numeri-
cal solution stability, accumulation of rounding-based errors due to arithmetic oper-
ations performed by the computer.
[email protected]
The components of contact stress vectors PX (I) , PY (I) , PZ (I) , I =
1,N, obtained by solving the linear algebraic equation system, are transferred to
the STRESS program module where for each boundary element tangential PT (I)
140 3 Computer Implementation of Boundary-Element Algorithms

[email protected]
Fig. 3.1 Structural scheme of Rostwerk software
3.1 Software for Solving Spatial Problems of Contact of Foundations with Soil Bases 141

and normal PN (I) contact stresses are computed. The numbers of the boundary
elements, for which PN (K)<0, are stored for subsequent elimination while consid-
ering the contact interaction process. For this purpose the influence matrix elements
A (I,J), corresponding to the boundary elements with the number K, are made zero
(except the diagonal ones). As a result, for the boundary elements of such type at
subsequent calculations the contact stress values will always be zero. The above pro-
cess is performed by iteration until none of the boundary elements is characterized
by tensile stress (PN (I)<0).
From the solution, obtained from the iteration process, the OUTPRT software
module prints out (or to a file) the components SU, SV, SW of the displacement
vector for the foundation structure centre, its slopes FX, FY, FZ with respect to
the coordinate axes, the components of contact stress vectors PX (I) , PY (I) ,
PZ (I) , PT (I) , PN (I), acting on the boundary elements as well as auxiliary
geometrical information, required for the description of the results of the numerical
solution of the problem. From the programming point of view, the OUTPRT module
is standard and does not contain new elements. After its operation being finished,
the processing of the computation results can be performed in the user-defined form
using various graphical software providing visualization of 2-D and 3-D images.
Input data preparation. All the input data are divided into the data on the type and
topology of the foundation structure surface, load-and-geometrical data, containing
the data on the structure dimensions and external forces, as well as the data on the
soil deformation properties.
First of all the characteristic of the foundation structure type is specified:

IK = 1 – shallow foundation,
IK = 2 – deep foundation.

Then, for each foundation type the contact surface shape should be specified.
For shallow foundations in the Rostwerk software the following system of char-
acteristics is accepted:

JF = 1 – circle,
JF = 2 – ellipse,
JF = 3 – square,
JF = 4 – rectangle,
JF = 5 – trapezoid,
JF = 6 – quadrangle of general type,
JF = 7 – rectangle with angular cut-outs (cross),
JF = 8 – rectangle with lateral cut-outs (I-beam),
JF = 9 – T-beam,
JF = 10 – L-shaped contact domain,
JF = 11 – closed rectangular-shaped strip bottom,
[email protected]
JF = 12 – symmetrical ring,
JF = 13 – eccentric ring,
JF = 14 – ring with cut-off internal circle,
142 3 Computer Implementation of Boundary-Element Algorithms

JF = 15 – rectangle with a circular cut-out,

JF = 16 – circle with a rectangular cut-out.

For deep foundations in the Rostwerk software the following system of charac-
teristics is accepted:

JZ = 1 – rectangular prismatic,
JZ = 2 – pyramidal,
JZ = 3 – pyramidal with a prismatic head,
JZ = 4 – wedge,
JZ = 5 – stepwise,
JZ = 6 – rectangular prismatic with a set-off in the bottom,
JZ = 7 – prismatic with oblique bottom,
JZ = 8 – slotted with longitudinal cross-section of rectangular shape,
JZ = 9 – slotted rectangular with curved bottom shape,
JZ = 10 – slotted with curved longitudinal cross-section shape,
JZ = 11 – slotted with support widenings,
JZ = 12 – wedge-slotted with a broadened toe,
JZ = 13 – vertical cylindrical,
JZ = 14 – vertical cylindrical with a pile raft of rectangular (JZ = 141) or
circular (JZ = 142) shape,
JZ = 15 – inclined cylindrical,
JZ = 16 – inclined cylindrical with a pile raft of rectangular (JZ = 161) or
circular (JZ = 162) shape,
JZ = 17 – cylindrical with a spheroconical bearing,
JZ = 18 – cylindrical with an anchor plate,
JZ = 19 – common conical (JZ = 191) and biconical (JZ = 192),
JZ = 20 – deepened sphere.

Then, in accordance with the characteristics JF and JZ, the geometrical charac-
teristics of a specific foundation structure are interactively specified using the fol-
lowing parameters:

DL – length,
DS – width,
DH – thickness,
DZ – depth,
DR – radius,
DA1, DB1, DC1,. . ., DR1, . . . – other characteristic dimensions of the
foundation structure (size of set-offs, cut-outs, cross-sections, tilt angles etc.).

The programs for automatic mesh of the contact surface PJF1–PJF15, PJZ1–
[email protected]
PJZ20 perform the mesh into boundary elements in accordance with the user-
specified degree of discretization density, as a result arrays of spatial coordinates
of the boundary-element vertices AX(I,J), AY (I,J), AZ(I,J) being formed,
3.1 Software for Solving Spatial Problems of Contact of Foundations with Soil Bases 143

where I is the global number of the boundary element, J = L(I), L(I) is the
attribute vector whose components are assigned values 3 or 4 for triangular or quad-
rangular boundary elements, respectively.
The library of routines for automatic boundary-element grid formation on the
contact surface of the foundation and soil can be supplemented by user in the course
of using the Rostwerk software. In case the program of automatic mesh being absent,
user can form the corresponding arrays manually and transfer them to the main
routine via shared space

COMMON/ELPS/AX(510,4),AY(510,4),AZ(510.4),L(510).

For the accuracy of the obtained results it is very important to provide rational
mesh of the foundation structure contact surface into boundary elements. These
issues are considered in detail in Sects. 3.3 and 3.4. Here we only note the following
basic recommendations concerning discretization:

(a) it is preferable that boundary elements should have regular geometrical shape:
triangular boundary elements should be close to equilateral triangles, quadran-
gular boundary elements – close to squares; the presence of triangular boundary
elements with obtuse angles or prolate quadrangular boundary elements results
in a decrease of the accuracy of calculation of contact stress values as well of
foundation displacements and slopes,
(b) the grid of boundary elements should be condensed near ribs, angular points,
and edges where a sharp local increase of contact stress occurs (so-called edge
effects in theory of elasticity); if the stress distribution is a priori unknown, in
the contact domain a uniform mesh should be performed with maximal density,
(c) discretization of separate fragments, units, or faces of the contact surface should
be performed in agreement with the discretization of adjacent areas; if this
requirement is fulfilled, the numerical solutions (the calculated contact stress
values) will be obtained with higher smoothness degree.

The input data preparation is finished by specifying the values for the soil defor-
mation modulus E and Poisson ratio PU, nonhomogeneity parameters, lamination
etc as well as external load vector

{PX,PY,PZ,PMX,PMY,PMZ} .

These parameters can be specified both in the input data file and interactively by
typing on keyboard.
The input data can be printed out, output to a file or the screen on the user’s
request.
Description of the numerical modeling results. After the problem solving being
[email protected]
finished, the calculation results are output to a file RESULT.
This file sequentially contains the data on the dimensionality of the discrete
problem:
144 3 Computer Implementation of Boundary-Element Algorithms

KR – number of the boundary elements,

3∗KR + 6 – total number of unknowns at the calculation of a deepened founda-
tion structure,
KR + 3 – total number of unknowns at the calculation of a shallow foundation
structure
as well as
COND – condition number of the matrix of the resolving linear equation system,
S (I), XA (I), YA (I), ZA (I), CNR (I,3) – areas, gravity centre coordi-
nates and direction cosines (components of external normal unit vectors) for
the I-th boundary element.

Then the file contains contact stress arrays in the global coordinate system
OXYZ, normal and tangential stresses, acting in the I-th boundary element plane
(I = 1, 2,. . ., N)

PKX(I),PKY(I),PKZ(I),PN(I),PI(I)

and vectors of linear {SU, SV, SW } and angular {FX, FY, FZ} displacement
of the foundation as a rigid solid.
In case domains of the foundation uplifting being present, the contact stress and
displacement arrays can be, on the user’s request, output to the RESULT file at each
iteration step what enables the development of the foundation uplifting areas to be
traced. Besides, as a result of solving the problem, the value S of the total possible
area of the foundation contact with soil is output as well as the sum of areas of the
boundary elements SK, at which compressive normal stresses act. This enables one
to estimate the fraction of the foundation structure surface which interacts with the
soil mass.
Stages and specific features of practical calculation. Calculation using the Rost-
werk software should be performed in the following sequence:

(1) make the calculation scheme, reducing all forces from the above-foundation
structure to the centre at the foundation edge to the system of forces and
moments P = {PX, PY, PZ}, M = {PMX, PMY, PMZ},
(2) analyze the calculation scheme from the point of view of existence of symmetry
axes and planes and take these data into account in order to reduce the dimen-
sionality and shorten the time of formation of the influence coefficient matrix,
(3) indicate the topological scheme of the foundation structure surface among the
preset schemes (JF = 1, 2,. . ., 15, JZ = 1, 2,. . ., 20) or work out
a program for automatic mesh of the contact area into boundary elements,
(4) input the input data for geometrical and physical characteristics of the “founda-
tion + soil base” system by their formation within the input data file or interac-
tive typing,
[email protected]
(5) using the visualization program, check the correct enumeration of vertices
(counterclockwise) for each boundary element in accordance with the external
normal direction (towards the soil),
3.1 Software for Solving Spatial Problems of Contact of Foundations with Soil Bases 145

(6) run the program according to the requirements of the computer operating
system.

The following conditions should be obeyed at calculations using the Rostwerk

software:

– the total number of boundary elements N, into which the foundation structure con-
tact surface is discretized, at general spatial formulation of the problem should not
exceed 510 (with the account of tangential stress) and 1530 (without the account
of the latter),
– global enumeration of the boundary elements in the software is not strictly deter-
mined, but nodes within each boundary element should be enumerated counter-
clockwise at the observation from the side of the external normal (directed towards
the soil),
– influence functions being used (singular fundamental solutions) are set in the form
of subroutine functions, whose formal parameters are coordinates of points of
observation and application of unit concentrated forces. Mechanical parameters
of the elastically stressed base should be transferred to the subroutines using the
shared space,
– to provide visual interpretation of the numerical solution of the problem all the
data, processed and created by the routines of the Rostwerk software, are stored
in direct-access files in accordance with the requirements of the graphic software
applications used. Therefore, the format of the data storage should be modified in
accordance with the rules of the specific graphic data processing software. There
is a variety of such software tools which are constantly updated.

The above restrictions are determined by the RAM value and can be revised while
using modern hardware. The restriction for the number of the boundary elements
was determined by the RAM size and CPU speed of computers, for which the first
working version of the software was implemented. While more powerful computers
are used, no above noted restrictions are required.
Experience of implementation and work with Rostwerk software. The analysis of
the results of both test and practical calculations for the second limit state, being
considered in detail in the subsequent sections of the book, has shown that the
boundary-element Rostwerk software has a number of clear advantages over the
known software of similar purpose (worked out on the base of finite-element and
finite-difference methods).
The software combines the depth of theoretical studies with the simplicity of
use and enables one to carry out with the required accuracy calculations of contact
interaction of foundations with soil bases which earlier had been considered hardly
possible or totally impossible.
[email protected]
The Rostwerk code can be easily adapted for any computer with various speed
and RAM size, since the routines it contains are written using FORTRAN, a classic
language for technical applications.
146 3 Computer Implementation of Boundary-Element Algorithms

The organization of the Rostwerk software operation with the formation of

intermediate files of input data and calculation results enables the elaborated
boundary-element software to be linked with other calculation applications as well
as computer-aided design tools. In turn, the software can also serve as a base for
working out specialized software for computer-aided design of foundations for both
typical structures and specialized important objects.
High efficiency of the Rostwerk software is also determined by the possibility
of calculations to be performed directly in practice since it requires rather small
computation time and does not impose any special requirements to the computer
hardware. This results in a decrease of manpower effort and reduction of the design
work execution period, ability to optimize the calculation schemes, shape and load-
ing parameters of the known and new foundation structures.

3.2 Specific Features of Numerical Solutions of Linear Algebraic

Equation Systems with Non-symmetrical Matrices, Arising
in Boundary-Element Analysis
Spatial contact problems for rigid punches, interacting with a linearly distorted
base, are reduced, as shown in Chap. 2, to the solution of linear algebraic equa-
tion systems. The dimensionality of such systems, due to the discretization of only
the contact boundary, lower than in the case of finite-element being used; how-
ever, the system matrices are completely filled, nonsymmetrical and not always well
conditioned [32].
Note that in the finite-element method the matrix of a linear equation system
is symmetrical, positively determined and band at corresponding enumeration of
node unknowns. For such systems effective methods for numerical solution have
been elaborated, requiring relatively small computation time, with high accuracy
and numerical stability (square root, block diagonalization, tridiagonalization and
other methods [60]).
In [154] the first schemes for numerical solution of boundary integral equa-
tions were proposed in the form of boundary-superelement method, resulting
in equation systems with symmetrical band matrices. The advantages of the
above approach are easy programming of the numerical algorithm and possibil-
ity to connecting to finite-element method routines. Simultaneously the boundary-
superelement approach has an essential shortcoming: when the equation system
for each superelement is built, a matrix, whose dimensionality is proportional
to the number of the superelement nodes, needs to be inverted. This essentially
reduces the method efficiency and prevents its wide practical application. Tsybenko
and Lavrikov [137] proposed a version of direct formulation of boundary integral
[email protected]
equation method, directly following from the method of weighted residuals [33]
and resulting in a symmetrical and completely filled matrix of resolving equation
system.
3.2 Specific Features of Numerical Solutions of Linear Algebraic Equation Systems 147

In the case under our consideration a matrix equation

A·Z=B (3.1)

corresponding to a resolving system of linear algebraic equations for the spatial

contact problem for a rigid punch, will have a nonsymmetrical matrix A, containing
four blocks. It can be presented in the following form:
 
D C
A= (3.2)
T 0

In Eq. (3.2) D3m×3m is a nonsymmetrical square matrix of influence coefficients,

m is the total number of boundary elements on the contact surface. Due to the prop-
erties of integral operator kernels for problems of theory of elasticity (they are pos-
itively determined functions [81, 144]) the matrix D is positively determined and
possesses a dominant main diagonal. Rectangular sparse matrices C6×3m and T3m×6
are determined by the presence of six linear equations being an algebraic analog of
integral equilibrium equations. The elements of these matrices are the coordinates of
gravity centres, areas and static moments of boundary elements in the global coordi-
nate system (for details see Sect. Section 2.2). First 3m components of the solution
vector X are unknown contact stress values pX , pY , pZ , and the rest six components
are the parameters of the punch displacement as a rigid solid U, V, W, ψ X , ψ Y , ψ Z .
It should be noted that the elements of the matrix D have quite small values in
comparison with the nonzero elements of the blocks C and T. The difference is on
the average by factor of E where E is the soil deformation modulus with typical
values of 10–20 MPa and reaching the values of the order of 40 MPa and higher
for some soil masses of rock type. Hence, the matrix A of the resolving system
of linear equations (3.1) is nondegenerate, almost completely filled, nonsymmetri-
cal, and with a considerable nonuniformity of coefficients. The nonuniformity will
increase with the punch size, i.e. with the increase of the contact surface area of
the foundation and soil. According to [45], the matrix A belongs to the matrices of
storable type since at least (3m)2 of its elements should be stored in the computer’s
RAM.
From the above stated, as well as with the account of recommendations for appli-
cation of direct methods of linear equation systems [93] we conclude that for the
numerical solution of the equation system (3.1) one should apply a known, widely
spread, based on the LU-expansion Gauss method of successive elimination of
unknowns with a choice of the main (with the largest absolute value) element in
a row, column or in the whole matrix. Gauss elimination, expressed in matrix nota-
tion, can be treated as triangular factorization of the matrix A (LU-expansion)

[email protected]
A = LU

and the solution of the system (3.1) is reduced to the solution of two systems
148 3 Computer Implementation of Boundary-Element Algorithms

LY = B, UZ = Y

with the lower L and the upper U triangular matrices.

The most suitable for our calculations are two routines – DECOMP and SOLVE
[45]. The DECOMP routine performs the part of the Gauss elimination, depending
solely on the matrix. Besides, it stores factors and information on the main elements.
The SOLVE routine uses these results to obtain a solution for an arbitrary right-hand
part. It is important that DECOMP also evaluates the matrix condition number, being
a reliable and useful measure of its closeness to non-degeneracy, stability of the
numerical solution, rounding errors, sensitivity to specifying the input data in the
right-hand sides of system of Eq. (3.1). Note that the estimated condition number is
obtained at a rather easy rate after the triangular expansion of the matrix A. having
been found by the Gauss method.
In spite of the Gauss method belonging to direct methods of solution (i.e. it
should give an exact solution after a finite number of actions), no exact solution is
achieved by computer calculation due to computation errors. According to [116],
accumulation of rounding errors at solving linear algebraic equation system by
Gauss method using floating-point computations results in the sought solution being
determined with a relative error

( (
(Z − Z ( 
= o cond(A) · n · 2−t (3.3)
Z

where cond(A) = ||A|| · ||A−1 || is the condition number of the matrix A, n is the
order of the matrix A, t is the number of digits in the mantissa of a floating-point
number in binary representation. For a typical personal computer 2−t is approxi-
mately equal to 10−6 –10−8 .
In the case under consideration direct application of the Gauss elimination
method (routines DECOMP, SOLVE etc.) results in a rather high overall computa-
tion error, evidenced by the increase of condition numbers cond(A), reaching the
order of 106 and higher. Besides, note that if the matrix A of the system (3.1) is
not degenerate, then by increasing the computer word length one can always reduce
the rounding error and obtain a numerical solution close to the exact solution of the
problem. However, calculation with double or triple number of digits accordingly
increases the computation time and computer RAM size which, even at current level
of computer equipment, is not available for the most of the practical problems.
On the other hand, it follows from Eq. (3.3) that in order to reduce the numerical
solution error one should try to increase the input data accuracy or to reformulate the
algebraic analog (3.1) of the integral equation system for the spatial contact problem
with respect to other parameters.
Since the right-hand sides of the system (3.1) are given exactly, and the coeffi-
[email protected]
cients of the matrix A are determined by numerical-and-analytical method with con-
trolled accuracy, then in order to obtain a well conditioned system some artificial-
looking steps should be taken.
3.2 Specific Features of Numerical Solutions of Linear Algebraic Equation Systems 149

First, in order to reduce the condition number of the matrix A we effectively use
scaling of a linear equation system or, in other words, multiplication of the matrix
equation (3.1) at the left-hand side and at the right-hand side by diagonal matrices
Q1 and Q2 . In a general case, scaling of the system (3.1) can be given by

Q1 AQ2 Y = Q1 Y, Z = Q2 Y. (3.4)

Studies of a number of authors show [28, 43] that a successful scaling improves
the computation process, increasing the stability of the numerical solution to the
effect of rounding errors. For many practical problems unilateral scaling, i.e. transi-
tion from the matrix A to the matrix AQ or QA appears to be sufficient.
We have performed numerical experiments on unilateral scaling of the system
(3.1) for a number of models and practical problems of contact deformation of foun-
dations with soil bases. The last six columns of the matrix A (containing coefficients
at unknown punch displacements) were divided by the same constant number μ. As
a result of an extensive series of calculations, we have found an optimal value μ ≈
102 resulting to the decrease of the condition number on the average by 3–4 orders of
magnitude. Note that, in accordance with the rules of linear algebra in the solution,
obtained after scaling, the last six components determine the sought displacement
values with the accuracy of a constant factor, i.e. in the form

μU, μV, μW, μψx , μψy , μψz .

Therefore, in order to obtain a physical solution one should in a {3m+ 6}-

dimensional solution vector divide only last six components by the scaling factor
μ, in this scaling method first 3m components corresponding to invariable contact
stress values, i.e. those which do not need to be corrected. Note also a possible
method of multiparametric (μi , i = 1, 2,. . .6) scaling of the matrix of system (3.1)
over each of the components of displacement vectors of the punch as a rigid solid.
However, such approach requires extensive methodological studies which we have
not performed. This approach can be, undoubtedly, quite useful for the numerical
solution improvement in case specific types of foundation structures and spatial
loading types being considered.
The estimated value cond(A) of 102 , obtained in our practical calculations, is the
evidence for the good condition of the system (3.1) transformed by scaling, i.e. for
the closeness of the approximate numerical solution to the exact one. This can be
confirmed by iterative refinement of numerical solutions, intentionally performed
for many problems. The value of correction to the obtained solution always was
small, and in most cases even zero. From this one can conclude that numerical solu-
tions, obtained using scaling, coincide with the exact solution of the system (3.1)
within the machine word length.
Second, the experience of the numerical calculations performed has shown that
[email protected]
along with the scaling of the matrix A of the system (3.1) a reduction of the condition
number (hence, an increase of accuracy of the numerical solution being obtained)
is possible due to rearrangement of first 3m its variables (unknown contact stress
150 3 Computer Implementation of Boundary-Element Algorithms

values). In the initial formulation of the system (3.1) the vector of unknowns Z was
determined as

Z = (P(1)
x ,Py , Pz , Px , Py , Pz , . . . , Px , Py , Pz , x , y , z , ψx , ψy , ψz ) .
(1) (1) (2) (2) (2) (m) (m) (m)

At such order of unknown contact stress values there is alteration of columns

of different types: projections of contact stress vectors on different coordinate axes
can be essentially different in the order of magnitude. If the vector of unknowns is
transformed to the form

(1) (2) (3) (m) (1) (2) (m) (1) (2)
Z = Px , Px , Px , . . . , Px , Py , Py , . . . , Py , Pz , Pz , . . . ,
(m)
Pz ,x ,y ,z , ψx ,ψy ,ψz ,
then the columns of the matrix A appear arranged in groups with close values of
components and thereby improve the structure of the matrix A.
The corresponding numerical experiments have revealed a quite noticeable
decrease of the condition numbers of the matrix A after the above rearrange-
ment of unknowns. Besides, a general acceleration of the program performance for
solving linear equation systems for matrices with improved structure was noticed
(rearrangement and selection of the main element consume much computation
time). Afterwards all calculations of processes of contact interaction of foundation
structures with soil were carried out with mandatory scaling of the matrix A of the
resolving system of linear equations, and in the cases of deepening – also with the
rearrangement of the system (3.1) over uniform variables.
Finally note that for a more accurate check of the effect of rounding errors being
introduced at each arithmetic operations, Rostwerk software package contains a rou-
tine, using the algorithm of Godunov et al. [51], which takes into account the effect
of the rounding errors at the specific computer used for the calculations and outputs
the result of solving linear algebraic equation systems by Gauss method with a guar-
anteed accuracy. Besides, this program can in the course of computation find out that
a given system possesses a condition level insufficient to guarantee any accuracy at
a computer of the given capacity. In practice, due to high computation time, we used
the routine for solving systems with guaranteed accuracy only for test calculations
before a series of computations for foundations of a given type or while using dif-
ferent contact models of elastic bases for the sake of additional substantiation of
reliability of the numerical solutions being obtained.

3.3 Effective Discretization of 2-D Domains of Complex Shape

at Numerical Solving of Spatial Contact Problems of Theory
of Elasticity
[email protected]
It is well known that numerical implementation of the boundary integral equation
method for solving contact (mixed) problems of spatial theory of elasticity assumes
a stage of mesh of the contact surface into boundary elements, resulting in the
3.3 2-D Domains of Complex Shape at Numerical Solving 151

discretization of the initial problem. The arising linear algebraic equation systems
have asymmetrical, completely filled matrices whose orders correspond to the num-
ber of the mesh elements. With the increase of the number of boundary elements
a trend to worse condition of such matrices is observed, and this affects the accu-
racy of the results being obtained. This feature of the boundary-element method
as well as restricted capacities of the computing facilities urge the elaboration of
such an algorithm of contact surface mesh which could possibly satisfy the fol-
lowing, in general, contradicting requirements. On one hand, it should be aimed
at obtaining minimal number of the mesh elements, on the other hand it should be
universal enough and generate a high-quality mesh, adaptive to the problem require-
ments. These circumstances result in the necessity to apply mostly piecewise con-
stant approximation of the contact pressure field in the boundary-element method
as well as to reduce the number of the mesh elements. Hence, the problems arise,
concerning compensation of the above disadvantages and smoothing of numerical
solution results, obtained for not dense grids.
Section. 3.3.1 describes a universal algorithm of triangulation of a 2-D domain,
aimed at minimizing the number of the mesh elements with the mesh quality
being preserved. The examples of the algorithm application for triangulation of flat
domains of non-canonical shape are also given.
In Sect. 3.3.2 a concept of a dual grid pair is introduced, akin to the concept of
duality in planar graph theory. Application of dual grids in the boundary-element
method is considered. A preprocessor algorithm is described for building up a dual
grid of the type of mesh into Voronoy polygons according to the given triangula-
tion of the domain. The preprocessor application enables the dimensionality of the
resolving linear equation system to be reduced (on the average by factor of 1.8)
with respect to the initial one, and, accordingly, its conditionality to be improved.
A postprocessor is elaborated, enabling a new, more smooth and exact solution to
be obtained based on the simultaneous use of two approximate boundary-element
solutions, found in the nodes of a dual grid pair.

3.3.1 Algorithm of Triangulation in the Boundary-Element

Method
The main idea, pursued at discretization in numerical modeling of contact inter-
action, is to give such a mesh of the calculation domain into boundary elements
that will provide obtaining the most exact solution of boundary integral equations.
For this purpose an algorithm and a software application were elaborated [16], in
particular, intended to perform discretization of flat simply and multiply connected
domains, bounded by straight segments or circular arcs. The following types of
mesh are provided: (i) close to uniform, (ii) with condensation near the boundary or
[email protected]
in separate areas of the domain.
The proposed algorithm, contrary to the procedures quoted in [68, 83, 101, 114,
115], is neither frontal, nor iterative. In the first approximation it can be assigned
to the class of blockwise discretization methods [114]. However, in the case under
152 3 Computer Implementation of Boundary-Element Algorithms

consideration blocks are formed by subdomains of a rather general nature; blocks

can be either selected from the domain by user (in a dialog), or automatically formed
by the software (in the independent mode). The main stages of the proposed algo-
rithm of triangular discretization are the following:
(a) mesh of the domain near the boundary in order to provide the grid condensation,
(b) sequential detachment of large blocks from the rest of the domain,
(c) stratification of the blocks into strips,
(d) mesh of the strips into elements.
Main conventions on the terminology. Part of a plane bounded by a closed con-
nected polygonal line without transversal self-crossing of its segments, will be
called a domain, and the polygonal line itself – its boundary (Fig. 3.2). The parts
of the polygonal line will be called segments, their ends – vertices.
Vertices will be considered enumerated counterclockwise along the boundary
path.
Let us define section (I, J) as a straight-line segment totally inside the domain,
connecting the vertices with numbers I and J of its boundary. A segment is also a
section.
Let H be a number. We will consider a number S comparable with H if H/d ≤ S
≤ dH where d > 1 is a constant (in our consideration d, as a rule, equals 2 or 3).
Let us call a mesh of the polygonal line segments into the minimal number of
equal parts, not exceeding H, anadditional mesh of the polygonal line with a step H.
Let us call a domain a strip in case if in its any point P the minimal diameter does
not exceed the value, comparable with H (Fig. 3.3).
Algorithm of strip discretization according to a given step. Let Q be a strip. Apply
an additional mesh to its boundary with a step H, and let the obtained mesh points
be enumerated from 1 to N. In the course of discretization we will operate with a
movable section (I, J) and a subdomain of the strip at one side of the section. This
section and this subdomain will be called current and possess an ordinal number.
At the initial moment of time I =1, J =1, and the whole strip is the current domain
with the ordinal number 1. Let I1= I +1 and J1= J −1. In the course of the mesh I,
I1, J, J1 will vary, but always I < I1< J1< J.

(a) (b) (c)

[email protected]
Fig. 3.2 To the definition of a domain: (a) a simply connected domain with non-transversal cross-
ing of ribs (6, 7) and (11, 1); (b) a doubly connected domain, bounded by two boundaries; (c) an
example of a polygonal line, which does not bound any domain
3.3 2-D Domains of Complex Shape at Numerical Solving 153

(a) (b)

Fig. 3.3 To the definition of a strip: (a) “typical” strips; (b) example of complex-shaped strips

Then, among the chords, outgoing from I, and the chords, outgoing from J, find
the minimal one, entirely belonging to the current domain. This will be a new sec-
tion. As per construction, it length does not exceed a value, comparable with H. Two
cases are possible.
A. The new section cuts off a triangle from the current domain. Then its sides do
not exceed a value, comparable with H, and it is considered a next element of
discretization. Then evident corrections are made for I, J, I1, J1, and the
process is repeated (Fig. 3.4a).

(a)

(b)

[email protected]
Fig. 3.4 Mesh of a strip: (a) the case of splitting off a triangular element; (b) the case of the
process branching; 1 – initial section, 2 – current section, 3 – new current section, 4 – current
domain, 5 – old current domain, 6 – new current domain
154 3 Computer Implementation of Boundary-Element Algorithms

B. The new chord divides the current domain into two subdomains of a general
nature which, evidently, are also strips (Fig. 3.4b). In this case the subdomain,
containing (I, J), is given the old ordinal number, and for some time is out
of consideration. The other subdomain is given a number, greater by 1, and
becomes the current subdomain along with the new section. Accordingly, I,
J, I1, J1 are changed (Fig. 3.4b), and then the discretization procedure is
applied to a new current domain.
As a result of such a process, first the range of the highest order will be meshed,
then we return to the next-highest order range and repeat the procedure once again.
Finally, such recursive algorithm will result in a complete mesh of the whole strip.
Note 1. If the values of the segments of the strip boundary and their sections are
limited from below (e.g., to H/2), this increases the uniformity and quality of the
mesh.
Note 2. A procedure is provided, preventing appearance of triangles with great
obtuse angle. Here we do not pay special attention to it.
Note 3. Evidently, the algorithm provides a procedure to check whether a new
section is entirely within the strip.
Condensation of mesh near the domain boundary. Cutting off singularities. Let
Q be an arbitrary domain. The mesh can be condensed along separate parts of the
boundary or along the whole boundary. In each of the cases the corresponding part
of the boundary is called the part of condensation. Its mesh is supposed to be already
performed. A number of vertices of the part of condensation, called reference ver-
tices, is displaced into the interior of the domain (Fig. 3.5). The displaced vertices,

[email protected]
Fig. 3.5 Mesh condensation near the boundary and cutting off singularities: 1 – reference vertices
of the part of condensation, 2 – vertices of the new boundary, obtained by shifting of the reference
vertices, 3 – multiple vertices, 4 – the first condensation strip, 5 – the second condensation strip,
6 – domain Q, remaining after the singularities being cut off
3.3 2-D Domains of Complex Shape at Numerical Solving 155

being connected by straight lines, form a part of the boundary (or the whole bound-
ary) of the part Q of the domain Q having remained after the condensation. This
result can be treated as cutting off from Q the parts with the boundary nodes, con-
taining small singularities and segments whose size is much less than the step value
H of the mesh of the remaining part Q of the domain Q.
Note that between the new (displaced) and the old (to be displaced) vertices a
one-to-one correspondence is set (with the account of the repetition multiplicity). It
is indicated by arrows in Fig. 3.5. Between the old and the new parts of the boundary
a subdomain of the domain Q is formed, called the condensation subdomain.
Then the number of condensation layers L and steps H1 , HL , H1 < HL are spec-
ified. Let the sequence H1 < H2 < ... < HK < ... < HL (e.g., geometrical) “con-
nect” H1 and HL . Partition the segments, connecting the corresponding vertices of
the old and the new parts of the boundary, in such a way that their lengths be pro-
portional to H1 , H2 ,. . ., HL (Fig. 3.5). By straight connecting the corresponding par-
titioning points of different segments one stratifies the condensation domain into L
strips whose transverse size varies in accordance with the sequence H1 , H2 ,. . ., HL .
At the next stage at first an additional mesh of the first strip boundary and a mesh
of the first strip with a step H1 is performed, then an additional mesh of the second
strip boundary and a mesh of the second strip with a step H2 etc. until the L-th strip
with a step HL . Thus, the required condensation near the boundary is obtained. The
remaining part Q of the domain is afterwards independently meshed accordingly
with a step H, comparable with HL .
The position of the displaced vertices can be specified in different ways. The
algorithm provides the following options:

(a) let a vertex I be angular for two adjacent segments f1 and f2 . Displace f1 and
f2 parallel to themselves to the interior of the domain by distances d1 and d2 ,
respectively (usually d1 = d2 = d and is comparable with H). The new vertex
position will be placed in the point of intersection of the displaced segments,
(b) the initial position of the displaced vertices is determined similarly to the case
of (a), but afterwards the user makes corrections of position of some vertices
(e.g. combines them in multiple vertices),
(c) the user himself specifies the position of new vertices. For this purpose the soft-
ware provides a graphic cursor.

Note 4. Figure 3.5 shows the procedure to be applicable for an economical (in the
sense of the number of elements) cutting off small singularities and irregularities at
the boundary that simplifies the mesh inside the domain.
Note 5. Based on the above procedure, a final mesh of a doubly connected domain
can be obtained. The idea is clear from Fig. 3.6.
Note 6. The displacement of vertices to the interior can be used for an essential
simplification of the boundary of the remaining domain. This simplifies the mesh
[email protected]
and increases its quality.
In the proposed method the mesh of the remaining domain (close to a uniform
one) is, in general, performed in the following way: the domain is cut (if necessary)
156 3 Computer Implementation of Boundary-Element Algorithms

Fig. 3.6 Example of a mesh

of a doubly connected
domain using a condensation
procedure: 1 – reference
vertices, 2 – shifted vertices,
3, 4, 5 – condensation strips

into “regular” in their coordinate systems blocks. The blocks are stratified into strips
and the strips are partitioned into elements. Now consider stratification into strips
and decomposing into blocks.
Stratification of a regular domain into strips. A domain Q is called regular if
in a Cartesian rectangular coordinate system XOY any straight line, parallel to the
OY axis, intersects Q over a connected set. For such a domain, naturally, upper
and lower boundaries are specified (Fig. 3.7). Besides, it makes sense to distinguish
left and right segments, connecting the initial and end-point of the upper and lower
boundaries, respectively (Fig. 3.7).
Let us introduce order among the points of the plane: A is less than B if the
abscissa of A is less than the abscissa of B.
Let a step H be given, and let the segments of the upper and lower boundaries
be additionally meshed with a step H, and concerning the left and right segments
we assume (without a restriction of generality) that their projection on OX do not
exceed H/2. One should cut Q into strips, the width of whose projections on OX is
not less than H/2 and not more than 2.5–3H (the projection of the whole Q on OX
is assumed more than H/2). We also demand that the projections of all the sections
on OX should not exceed H/2.

Fig. 3.7 Stratification of a

regular domain into strips:
[email protected]
a – left segment (1, 1),
b – right segment (N, M),
c – current section (I, J), d –
new section (I1, J1)
3.3 2-D Domains of Complex Shape at Numerical Solving 157

Let us build sections from left to right. The first section is the left segment (I, J)
(in Fig. 3.7 I= 1, J= 1, a possible case of degeneration of a segment into a point
is shown). Consider a general situation. Let the last built section (I, J) connect the
I-th vertex of the lower and the J-th vertex of the upper boundary, and let at first the
distance along OX between the midpoint of the section (I, J) and the midpoint of
the right segment (N, M) is larger than 3H. Find on the lower and upper boundaries
the closest to I and J, respectively, vertices II and JJ in such a way that that were at
a distance from the centre (I, J) along the OX axis direction not closer than 0.75H.
Choose the largest of them (in the sense of a certain order) and find (in the sense of
the same order) the one closest to it on the opposite boundary. Denote these vertices
I1 and J1. Then the projection of (I1, J1) on OX does not exceed H/2 (since the
boundaries are additionally meshed) and, as can be readily seen, the strip located
between (I, J) and (I1, J1) (denote it as {I, J, I1, J1}) has projections on OX not
smaller than 0.75H and not larger than 2.5H. If the chord (I1, J1) does not entirely
belong to the domain Q, it can be easily corrected by substitution of I1 or J1 by
the vertex appearing to be inside the {I, J, I1, J1} strip. Thus, a new strip with
required properties is cut off from the domain Q. Then the process is repeated with
substitution of (I, J) by (I1, J1).
If the distance between the centres (I, J) and (N, M) is shorter than 2.5H, than
tale the whole strip {I, J, N, M}. If this distance is longer than 2.5H and shorter than
3H, than the strip {I, J, N, M} is divided in two.
A disadvantage of this method of mesh is the possibility of appearance of long
narrow strips that generates, after them having been meshed, a great number of nar-
row elements (Fig. 3.8). The reason for this is the fact that on the boundaries of
the domain Q segments with very large angular coefficients in the XOY coordi-
nate system are possible. Such segments will be called critical. If there are no such
segments, a great number of narrow elements will not arise.
A coordinate system X O Y will be called inverse to the XOY system if
O X ||OY, O Y ||OX. Note that if a segment, critical in XOY, is considered in the


inverse system X O Y , it is no longer critical. This suggests an idea to decompose

Q into blocks, part of which is to be partitioned in strips in the XOY system, and

[email protected]
Fig. 3.8 Example of a strip
containing a critical segment
and narrow elements
158 3 Computer Implementation of Boundary-Element Algorithms

the other part – in the inverse X O Y system. Such operation can be performed
manually, and the software provides such option along with the choice of the cor-
responding coordinate system. We have worked out a procedure to automate this
process. Its idea is in brief described below.
Division of a normal domain into blocks to be stratified into strips in mutually
inverse coordinate systems. A domain Q is called normal if it is regular both in XOY
and X O Y coordinate systems.
Let K > 1 be a number (usually K= 1.7, 2, 3). Boundary segments will be called
critical if their angular coefficients k in XOY are larger than K (the absolute value).
Let M be a set of points on OX which is the projection of all the critical segments
of the boundary, and let [a, b] be a connected component of this set. A part of the
domain Q, located above [a, b], will be called critical (Fig. 3.9). If long narrow set-
offs arise on the left side or on the right side (or on both sides) of this domain, then
emborder the domain at this side by a strip of a width comparable with H/2 at the
expense of the area with precritical slopes of the boundary segments. As a result,
the critical domain is confined between two sections in a domain without narrow
set-offs.
The same procedure can be performed with other components of the set M.
Embordering can result in merging of some critical domains into one. Finally the
whole domain Q will be cut into parts, some of which containing critical segments,
the others – not containing them (Fig. 3.9).
The parts, which do not contain critical segments, are, according to the above
described algorithm, cut into strips and meshed into elements. The others are in turn
considered in the inverse coordinate system X O Y , and the same procedure, as in
the case of the initial domain, is applied to them. Inside them critical and noncriti-
cal domains are again formed. The noncritical domains are immediately meshed in

[email protected]
Fig. 3.9 Decomposing of a normal domain into blocks: 1 – critical domains, 2 – embordering of
critical domains, 3 – non-critical domains (these blocks are meshed in XOY system)
3.3 2-D Domains of Complex Shape at Numerical Solving 159

X O Y , and the critical ones are subject to the above procedure, but in XOY sys-
tem, etc. Hence, Q will be gradually partitioned into smaller and smaller blocks. If
a block is entirely critical both in X O Y and XOY systems, then either the coordi-
nate system should be turned, or K increased, or the block should be meshed as it is.
If a block achieves size comparable with H, it is immediately meshed into elements
as a strip. Thus, the process is finished in any case, and though slim elements can
arise, their number is sharply decreased and their location is dispersed.
Practical application examples. While analyzing the results of the algorithm
operation, note that in a number of simple, but typical cases the described algo-
rithm enables the user’s activity in the discretization process to be reduced to a
minimum. It is sufficient to input the coordinates of the boundary vertices in the
counterclockwise order in a basic coordinate system (further address to the vertices
will be reduced to specifying of their numbers), specify condensation areas, number
of layers L, steps of condensation H1 , HL and the boundary segment displacement
value d. Then the step H of uniform mesh of the remained domain is specified and
automatic operation mode is chosen if the domain, having remained after the con-
densation, is normal. As a result, the program will perform discretization indepen-
dently and output the geometry of the boundary elements in the basic coordinate
system.
Simultaneously, the directed (not iterative) structure of the program enables the
mesh results to be foreseen and the mesh to be actively planned:

– user can in a dialogue mode specify the initial domain sections, the order of mesh
of the arising partitions, steps and coordinate systems of the mesh as well as con-
tinuously check the discretization process visually,
– by means of sections as well as the choice of position of the boundary vertices
being displaced at the condensation, the domain, remaining for the uniform mesh,
can be simplified and made close to regular,
– the routine can be assisted to bypass difficulties arising due to the critical boundary
segments,
– finally, lack of iteration reduces the program runtime duration.

The program code is written in the FORTRAN-77 language.

Figure 3.10 shows the examples of mesh (close to uniform) of domains of circular
and typical polygonal shapes.
For the mesh of the circular domain (Fig. 3.10a) the input data were the circle
radius R, number of vertices of uniform mesh of the boundary (36), the domain
mesh step H = R/5. The domain was meshed independently in the initial coordinate
system.
An example of a uniform mesh of an I-beam is presented in Fig. 3.10b. The
input data were 12 I-beam vertices, boundary additional mesh step; domain mesh
step. By three sections the domain was divided into four equal parts, each of them
[email protected]
being meshed independently in its own coordinate system. In a similar way a uni-
form mesh of a cross-shaped domain (Fig. 3.10c) and a domain with angular cut-
offs (Fig. 3.10d) is carried out. The mesh of the cross-shaped domain is performed
160 3 Computer Implementation of Boundary-Element Algorithms

(a) (b)

(c) (d)

Fig. 3.10 Examples of uniform discretization: (a) circular domain; (b) I-shaped domain; (c) cross-
shaped domain; (d) square with angular cutoffs

independently in the initial coordinate system, while the domain with angular cut-
offs is divided by a horizontal section into two parts, each of them being meshed
independently in its own coordinate system.
An example of mesh of a circular domain with a near-boundary condensation is
shown in Fig. 3.11. The number of the condensation layers is four, the number of
the reference vertices – 12, the displaced vertex has a multiplicity of 12.
Figure 3.12 illustrates the automatic mesh of a circular domain (Fig. 3.12a) and
circular domains with cut-offs (Fig. 3.12b–d) with the optimization of the number
[email protected]
of boundary elements.
The mesh, shown in Fig. 3.12a is performed similarly to the example of
Fig. 3.10a. The number of boundary elements is 110.
3.3 2-D Domains of Complex Shape at Numerical Solving 161

Fig. 3.11 Discretization of a

circular domain with
condensation at the boundary

(a) (b)

(c) (d)

[email protected]
Fig. 3.12 Examples of condensation for a circular domain with cutoffs
162 3 Computer Implementation of Boundary-Element Algorithms

At discretization domain, shown in Fig. 3.12b initially a quarter of the domain

within the sector (π/4, π/4) is considered. It is divided by two vertical sections
into a rectangular subdomain and two circular strips. The boundaries of these parts
are additionally meshed according to the specified step; afterwards, each of them
is independently triangulated. The total mesh of the domain is obtained by triple
rotation. The total number of boundary elements is 208.
Figure 3.12c gives an example of a domain obtained from a circular one by means
of eight cut-offs. For discretization initially an eighth part of the domain is consid-
ered. By a section it is divided into a rectangle and a circular sector. The rectangle
is meshed at once, and the sector is meshed with condensation near the circular arc.
Three condensation layers are seen in the figure. The final mesh of the entire domain
is obtained by means of seven rotations. The total number of boundary elements is
176. The discretization of the domain shown in Fig. 3.12d is performed in a similar
way. The number of boundary elements is 192.
The program, in which the above triangulation algorithm is implemented, pos-
sesses a user interface (in fact, an internal graphic software tool) enabling not only
the input data (domain boundaries, mesh steps, condensation areas etc.) to be spec-
ified by observing them on the screen, but also all the mesh stages to be actively
controlled, choosing the mode which is the most effective for the problem solution.
As a result, the user obtains the triangular element coordinates as well as the neces-
sary geometrical data, required in the boundary-element method. The computation
time for formation of a hundred of mesh elements for a low-speed computer is on
the average 3–5 s.
The elaborated algorithm of triangulation was applied to solve a number of
geotechnical problems on the calculation of parameters of contact interaction of
flexible and rigid punches as well as foundation plates of complex geometrical
shape resting on elastic nonclassical bases. Results on geometrical optimization of
the objects under study at different conditions of spatial loading were obtained (See
Chapter 4). The elaborated software can be without essential changes applied to
obtain numerical solutions of spatial boundary problems of mechanics and mathe-
matical physics with mixed boundary conditions (fracture mechanics, calculations
of structural element strength, heat and mass transfer, electrostatics etc.) based on
boundary integral equations.

3.3.2 Dual Grids and Their Application in Boundary-Element

Method

Here we state the practical approach developed for the solution of the above prob-
lems of reduction of the number of boundary elements and obtaining a smoothed
numerical solution on grids with a moderate number of elements [2, 17]. The main
idea consists in proceeding from a solution on one grid to a pair of approximate
[email protected]
solutions obtained on two grids in duality. Construction of grids in duality is a pre-
processor problem. Joint processing of two solutions, obtained for the dual grids, is
a postprocessor problem. Development and joint inclusion of a preprocessor and a
3.3 2-D Domains of Complex Shape at Numerical Solving 163

postprocessor into the software package is of great practical interest since it enables
solutions of spatial contact problems for complex-shaped punches to be built effec-
tively and with high accuracy without preliminary prediction of the contact pressure
field structure, as well as numerical solutions to be interpolated.
Dual grids on a plane. Main definitions. Let R be a flat polygonal domain and
A = {Aij } is a mesh of R into polygons Ai , i = 1, 2,. . ., M. The polygons Ai will be
called mesh elements or cells, the polygon vertices – nodes the polygon sides – ribs,
and the mesh A itself – a grid. If a node belongs to the domain boundary, it is called
a boundary node, otherwise – an internal node.
The concept of duality, used hereinafter, characterizes mutual location of the
elements and nodes of a pair of grids related to the meshes of one or two domains.
This concept naturally arises at numerical modeling in finite-element and boundary-
element methods for the problems of mathematical physics. Due to the specific fea-
tures of the problems, it is convenient to use several definitions of duality given
below which are close essentially, but differ in details.
If R does not coincide with the whole plane, then by adding to {Aj } a polygon A0 ,
coinciding with the closure of the complement of the domain R to the whole plane,
we come to a mesh A = {Aj }M j=0 of the whole plane. The collection of nodes and
ribs of the grid A (or, what is the same, A) forms a flat graph ÃA without isolated
and pendant vertices [59], for which in graph theory a concept of a dual graph A 

is known. The latter is built in the following way. In each polygon A0 , A1 ,. . . an

internal point b0 , b1 ,. . . is chosen (among them one can be infinitely remote). The
collection of these points forms a set of nodes of the dual graph A  . A rib in   is
A

an arc l , connecting those, and only those bk and bj , for which the cells of the initial
mesh Ak and Aj which contain them, have a common rib l. Such ribs l and l will be
called corresponding.
The graph ÃA is known to be dual to the graph A  up to isomorphism. Con-

sequently, ÃA and A will be mutually dual. The graph A  also generates a mesh

of the plane, denoted as A . It follows directly from the definition that if N, L, M

and N , L , M are the numbers of nodes, ribs and cells for the meshes A and A ,
respectively, then N = M , M = N , L = L .
It follows from the definition of dual graphs that location of their nodes is deter-
mined by far not uniquely. Therefore, while defining plane meshes dual to each
other, with the account of their further application in numerical analysis, we have to
impose additional requirements.
Definition 1. Two grids A and B on a plane are called dual if:
the corresponding graphs ÃA and ÃB are dual,
(1)
(2)
each node of one grid is contained in a cell of the other grid,
(3)
each cell of one grid contains a single node of the other grid,
(4)
all ribs are straight lines and each rib of one grid intercepts only the correspond-
[email protected]
ing rib of the other grid in a point, internal for these ribs (Fig. 3.13a).
It is clear that Definition 1 (with an evident modification of item 4) is also
extended for grids, given on a 2-D sphere or on a smooth surface of a 3-D body,
homeomorphic to a sphere.
164 3 Computer Implementation of Boundary-Element Algorithms

(a) (b)

(c)

Fig. 3.13 Dual grids generated by triangulations: (a) delaunay triangulation and Voronoy poly-
gons; (b) grids, satisfying definition 1; (c) grids, satisfying definition 2

Now let R be a subdomain of a plane and A be a grid on R, A is the corresponding

mesh of the whole plane, and let B be a mesh, dual for A in the sense of Definition 1.
Definition 2. A mesh B of the domain R, consisting of polygons Bj = R ∩ B̄j
where Bj are the elements of the mesh B (Fig. 3.13b), is called dual to the mesh A.
Note that for grids, dual in the sense of Definition 2, the properties (2), (3), and
(4), are, in general, not fulfilled in the pure sense and require a certain modification
of statements.
For such meshes each internal node of one grid is contained in a cell of the other
grid, and different internal nodes are contained in different cells.
[email protected]
Each cell of one grid contains one internal and (or) one boundary node of the
other grid.
Corresponding ribs of the dual grids intercept.
3.3 2-D Domains of Complex Shape at Numerical Solving 165

The meshes A and B are disparate and not mutually dual.

Consider one more definition of duality for grids on domains, not coinciding with
the whole plane.
Definition 3.Two grids A and B, specified on two different domains R and Q,
are called dual if each element of one grid contains inside it a single node of the
dual grid, and the grids themselves can be extended to two dual (in the sense of
Definition 1) grids on the whole plane (Figs. 3.13c and 3.13 a,b).
For such grid the following property is also valid: each internal node of one grid
is contained in a single element of the other grid.
We should emphasize that though for a grid A, specified on R, dual (in the sense
of Definition 3) grids are defined not uniquely, nevertheless they can differ from
each other only in the vicinity of the domain R boundary and should be in fact
identical inside R. Among them one should pay particular attention to those, for
which the domains Q and R are inside each other (the case of maximal and minimal
dual grids).
Thus, the main feature of all the above defined dual grids is a matched location
of their nodes, cells and ribs, for which each internal node of one grid is in the envi-
ronment of nodes of the other grid. This with necessity determines high correlation
of approximate solutions, found independently in the nodes of grids, dual between
each other.
Examples of dual grids
1. Let a domain R be the whole plane and points Pi , i = 1, 2,. . .,N be located on
it in a rather regular way. Using a construction from the book [106], consider a
grid B on R to be a set of Voronoy polygons (Dirichlet cells) with centres in the
points Pi . Consider A to be a grid, in which a rib connects those pairs of points
Pi and Pj , for which the Voronoy polygons have a common rib. Such grids A and
B are dual in the sense of flat graph theory [59, 106]. The grid A is a mesh of
the domain R , coinciding with a convex shell of the set of points Pi , i = 1, 2,. . .,
N, into triangles. Such triangulation is called Delaunay triangulation [66]. In the
general case A and B are not dual in the sense of Definition 1 since the properties
(2), (3), and (4) can be not fulfilled. One can easily check that the grids A and B
are dual in the sense of Definition 1 then, and only then when all elements of the
Delaunay triangulation A are acute-angled (Fig. 3.13a).
2. Let A be a triangulation of the domain R. Let the gravity centres of triangles of
A be the nodes of a mesh B; join by segments those of them which belong to
the triangles Ai and Aj possessing a common side. The grid B is a mesh of a
subdomain of the domain R. It can be easily shown that if for all the triangles Ai
of the grid A with sides ai > bi > ci the condition

aj < bj + 3cj (3.5)

is valid, then the grids A and B
[email protected] are in duality in the sense of Definition 3
(Fig. 3.13c).
3. Figure 3.13b shows two grids A and B on the same domain, satisfying Defini-
tion 2.
166 3 Computer Implementation of Boundary-Element Algorithms

(a) (b)

(c)

Fig. 3.14 Dual rectangular grids (a, b) and dual grids, consisting of arbitrary polygons (c)

4. Let A be a grid of rectangular elements. There exist many grids B, consisting of

rectangular elements, dual to A in the sense of Definition 3 (Fig. 3.14a, b). In
particular, among them there is a dual grid whose nodes are the gravity centres
of elements of A.
5. Figure 3.14c shows a pair of grids, dual in the sense of Definition 1, whose
elements are arbitrary polygons.
It is worth to make some remarks concerning the number of elements of the
grids A and B from the first three examples. Let N be the total number of nodes
of a triangulation A, NÃ being the number of nodes, belonging to the boundary, M
being the number of elements from A, and L being the number of ribs connecting
the nodes. For any triangulation the following formulae are valid:
M = 2(N − 1) − N , L = 3(N − 1) − N . (3.6)
If a grid on the whole surface (boundary) of a three-dimensional finite body is
dealt with, then the corresponding formulae are given by
M = 2(N − 2), L = 3(N − 2) . (3.7)
[email protected]
Hence, if NÃ is small in comparison with N, then the number of elements N of
the dual grid B is approximately twice smaller than the initial number of elements
of the triangulation A.
3.3 2-D Domains of Complex Shape at Numerical Solving 167

The latter remark enables the following approach to the problem of improvement
of conditionality of the system matrix in the boundary-element method to be pro-
posed. In the case of a condensed triangular grid on a 2-D domain or on the boundary
of a 3-D body, which generates a matrix of high dimensionality and bad condition-
ality, resulting due to errors to an inaccurate solution, it is recommended to pass to
a dual grid B with N (N ≈ M/2) elements (of Dirichlet cell type) that will reduce the
dimensionality of the discrete problem and improve its conditionality. One should
keep in mind that the mesh density, accuracy of the boundary approximation, and,
hence, the accuracy of the solution itself in the boundary-element method theoreti-
cally can be reduced.
Preprocessor algorithm. Here we describe the preprocessor algorithm which on
the base of a given grid A creates a grid B, dual to A. The description is given for the
most important case when A is a triangulation of the domain R (examples 2 and 3).
Let a triangulation A of the domain R with N nodes and M elements be given. In
this case at least we know two-dimensional arrays KNOTS (2,N) and ELEM (3,M).
The first one contains coordinates Xi , Yi of all the nodes Pi , i = 1, 2,. . ., N of the
triangulation, and the second one for each k = 1, 2,. . ., M contains the numbers i1 ,
i2 , i3 of those nodes which form the vertices of the element with the number k. Let
the order of i1 , i2 , i3 correspond to counterclockwise direction.
By a simple procedure these simplest data concerning the mesh A is comple-
mented with one-dimensional arrays NNBE (N) and NBE (NK) which explain the
triangulation structure. Namely, NNBE (i) is the number of the triangulation ele-
ments, for which the node Pi is a vertex. Such elements will be called possessions
of the node Pi . The array NBK (NK) contains possession numbers for all the nodes,
recorded consecutively (first for P1 , then for P2 , etc.). Thus,

NK = NNBE(i) = 3 M .
i
Note that the record of numbers of adjacent to Pi elements, contained in the array
NBE, does not yet in any way correspond to their order at counterclockwise encir-
cling Pi . This problem is solved by a separate procedure, accompanied by formation
of further important one-dimensional arrays NNBK (N) and NBK (NK) which for each
node Pi contain the number of neighbouring nodes (i.e. nodes connected by a rib to
Pi ) and the numbers of these nodes (similarly to the NBE array), respectively. Note
that in NBK array the enumeration of the nodes, neighbouring to Pi , already cor-
responds to the counterclockwise direction around Pi . Moreover, if Pi1 , Pi2 ,..., Piq ,
the neighbours of Pi and Pj , belong to the domain boundary, then Pi1 and Piq also
belong to the boundary.
Based on the data obtained, below we describe building up elements Vi of a new
mesh of a subdomain of the domain R, each of which contains one node Pi , that
will finish construction of the dual grid B. Building up Vi differs for the nodes Pi ,
belonging to the interior of the domain R and to its boundary.
[email protected]
Let at first Pi be an internal node and Pi1 , Pi2 ,..., Piq be the neighbours of Pi .
Denote C1 ,...,Cq to be the gravity centres of triangles Pi Pis Pis+1 , s = 1,2,...,q, where
Piq+1 = Pi . Then the polygon line Ãi = C1 , . . . ,Cq which does not contain self-
168 3 Computer Implementation of Boundary-Element Algorithms

(a) (b)

(c)

Fig. 3.15 Schemes of building up dual grids based on a given triangulation: (a) internal node; (b)
boundary node; (c) the case when the grid building method does not work

crossings, bounds a domain Vi (Fig. 3.15a) and, if the triangulation is of sufficient

quality (See, e.g., the condition of Eq. (3.5)), then Vi contains the node Pi . A case
of bad triangulation, i.e. when the algorithm under consideration does not work,
is shown in Fig. 3.15c. It is evident from the construction that the vertices of the
domains Vi , are, in turn, contained in the mesh elements A. Thus, according to
Definition 3, the mesh B, consisting of Vi , where Pi are internal nodes of the domain
R, form a mesh, dual to A, which covers all the nodes of the A grid, except the
boundary ones (Fig. 3.13b).
If the triangulation of A is an approximation of a complete boundary of a
3-D body, then all the nodes of A are internal and, consequently, B is also a mesh,
approximating the boundary of the same 3-D body. By applying the boundary-
element method to A and B we can obtain two approximate solutions of a 3-D
problem, the resolving system matrix for the mesh B being almost twice smaller in
size than the matrix for the mesh A.
Let now Pi belong to the boundary of the domain R. Consider a polygon line
Ãi = Pi D1 C1 . . . Cq−1 Dq where Cs , s = 1,2,...,q − 1 are the same as above and D1 ,
q are the midpoints of the ribs Pi Pi1 and Ui , respectively. Then Fi is closed, does
[email protected]
not contain self-crossings and encloses the domain Vi , including Pi (Fig. 3.15b).
A collection of all Vi , built for Pi both inside the domain R and on its boundary,
forms a new mesh B of the same domain R. B will not be dual to A in the sense
3.3 2-D Domains of Complex Shape at Numerical Solving 169

(a) (b)

(c)

Fig. 3.16 Examples of the preprocessor algorithm application

of Definition 3, since e.g. nodes of B of D1 and C1 type simultaneously belong to

the same element of the mesh A, but will be dual to A in the sense of Definition 2
(Fig. 3.13b). The examples of application of the described method are shown below
(Fig. 3.16) where the initial triangulation is created by a routine of triangular grid
generation (See Sect. Section 3.3.1, [16]).
It should be also noted that the proposed preprocessor algorithm includes the
following important functions: (1) by means of it the dual grids, having been built
before, are redefined in such a way that each internal node of one grid be either
[email protected]
the centre of gravity of the dual grid cell to which it belongs (Fig. 3.17a–c), or
the arithmetic mean of its vertices (Fig. 3.17b–d), (2) for each pair of dual grids a
general triangulation of the whole domain is built, whose nodes are all the nodes
170 3 Computer Implementation of Boundary-Element Algorithms

(a) (b)

(c) (d)

(e) (f)

[email protected]
Fig. 3.17 Pairs of dual grids on a circle: internal nodes of the grids are (a, c): the gravity centres
of the cells to which they belong and (b, d): mean arithmetic points of the vertices of the cells to
which they belong; (e, f): triangulation of a circular domain over the nodes of a dual grid pair
3.3 2-D Domains of Complex Shape at Numerical Solving 171

of both grids (Fig. 3.17e, f). The latter triangulation (being of independent value)
is important for the postprocessor application and at construction of isolines of the
numerical solution, obtained for the pair of the dual grids.
A large series of numerical experiments on dual grids has enabled us to eval-
uate the efficiency of the proposed approach as well as to compare (for a number
of problems) the results of calculations of contact interaction parameters obtained
by boundary-element method and by other known methods. In typical cases sub-
stitution of a triangular grid by a dual grid of generalized Dirichlet-Voronoy cells
results in the reduction of dimensionality of the resolving finite-measure equation
system by factor of at least 1.5. In this case, even if the accuracy of the new solu-
tion id decreased, this decrease is slight (by 5–10%), though often the accuracy is
increased due to a successful choice of the initial grid. Practical examples of solving
the spatial problems of theory of elasticity for non-canonical (simply and multiply
connected) contact domains using dual boundary-element grids, are given in Sect.
3.5.1.
Postprocessor algorithm for dual grids. Let one, as a result of a numerical exper-
iment, in the nodes Pi , i = 1, 2,. . ., M and Qj , j = 1, 2,. . ., N of the grids A and
B, which are in duality, have obtained approximate values Fi , i = 1, 2,. . ., M and
Gj , j = 1, 2,. . ., N of a function z = f(x, y) of two variables. The sets {Fi }, {Gj }
are assumed to be obtained independently and to contain errors (deviations from the
exact values of the function z = f(x, y) both of random character and related to the
errors of the numerical method itself). Since the nodes of one grid are located inside
(and even in the geometrical centre of) the other grid cells, evidently there should
be an internal relationship between the values of {Fi }, {Gj }. The presence of two
approximations enables one to suppose that on their base, with the account of the
interdependent location of the nodes Pi and Qj of the dual grids, a new approxima-
tion for z = f(x, y) can be obtained, being of a better quality than each of {Fi }, {Gj }
separately. Namely, using the approximate value sets {Fi }, {Gj }, one should:

(1) specify new approximate values {Ui }, {Vj }, i = 1, 2,. . ., M; j = 1, 2,. . ., N

simultaneously on the whole system of points Pi , Qj ;
(2) while specifying Ui , Vj , take into account their natural relation to {Fi }, {Gj }
and with each other;
(3) try to specify the new values in such a way that sharp oscillations of the neigh-
bouring approximate values be possibly smoothed and the effect of random
experimental errors be reduced.

Concerning the function f(x, y) itself we assume that within one element of mesh
A or B the sought function f(x, y) is rather well approximated by its first-order Taylor
polynomial. The latter fact means that either the gradf(x, y) variation rate is small,
or the grids A and B are sufficiently dense.
Construct a new approximate solution using the least-square method. For the
[email protected]
sake of simplicity consider the problem for the grids of the example 3. Let Abe a
square grid, containing kk = kk nodes, and B be a square grid, containing kk1 =
(k −1) · (k − 1) nodes. Each node Qj of the second grid is enclosed by some nodes
172 3 Computer Implementation of Boundary-Element Algorithms

Fig. 3.18 Neighbouring

nodes for rectangular dual
grids

Pi1 , Pi2 , Pi3 , Pi4 of the first grid and is their centre of gravity, and each internal
node Pi of the first grid is enclosed by some nodes Qj1 , Qj2 , Qj3 , Qj4 of the second
grid and also is their centre of gravity (Fig. 3.18).
The simplest way is to imply Ui = Fi and Vj = Gj . Then the requirement (1) will
be satisfied, but the requirements (2) and (3) will be not. Nevertheless, it is evident
that Ui and Vj should not essentially be far from Fi and Gj . This means that the value

kk 
kk1
ϕ1 (U, V) = (Ui − Fi )2 + (Vj − Gj )2
i=1 j=1

should not be large. Here U = {Ui },V = {Vj }.

Then, due to the assumption of f(x, y) within one mesh element being close to a
linear function, one concludes its value in the point Qj not to be strongly different
from the arithmetic mean of its values in the neighbouring points Pi1 , Pi2 , Pi3 , Pi4 ,
and the value in the point Pi not to be strongly different from the arithmetic mean
of its values in the points Qj1 , Qj2 , Qj3 , Qj4 . This leads to a conclusion that at the
correct choice of U = {Ui },V = {Vj } the following two values should not be large:

  Ui + Ui + Ui + Ui 2
ϕ2 (U, V) = 1 2 3 4
− Gj
4
j

 Vj + Vj + Vj + Vj 2
+ 1 2 3 4
− Fj ,
4
i

  Vj + Vj + Vj + Vj 2
ϕ3 (U, V) =
1 2 3 4
− Ui
4
[email protected]
j
  Ui + Ui + Ui + Ui 2
+ 1 2 3 4
− Vj .
4
j
3.3 2-D Domains of Complex Shape at Numerical Solving 173

Here i runs the numbers of the internal nodes of the grid A, a j runs the numbers
of all the nodes of the grid B.
From these considerations, the following criterion of choice of Ui and Vj can be
stated. Ui and Vj should be sought in such a way that the minimum value of

ϕ(U, V) = γ1 ϕ1 (U, V) + γ2 ϕ2 (U, V) + γ3 ϕ3 (U, V) (3.8)

be obtained, γ1 , γ2 , and γ3 being weight coefficients.

It is evident that if γ2 and γ3 are very small in comparison with γ1 , then Ui , Vj
coincide with Fi , Gj , respectively. If γ2 is large while γ1 and γ3 are small, then Ui
will be expressed in terms of Gj , and Vj – in terms of Fi . Finally, if γ1 and γ2 are
small in comparison with γ3 , then the solution will be any Ui , Vj , expressed in terms
of arithmetic mean values in the neighbouring nodes of the dual grid, but having no
relation to the initial Fi , Gj . The correct choice of the weight ratio γ1 , γ2 , γ3 can
give the values of Ui and Vj , satisfying to a certain extent all the requirements of the
problem under consideration. By variation of weights an accent can be made on this
or that (possibly, known a priori or a posteriori) property of the sought solution.
The minimum point of the sum of squares of Eq. (3.8) always exists. It can be
found from a system of kk + kk1 equations with kk + kk1 unknowns{Ui }, {Vj },
obtained by making partial derivatives equal zero:

∂ϕ ∂ϕ
= 0 and = 0.
∂Ui ∂Vj
By solving this system, one obtains new approximate values Ui , Vj for the func-
tion f(x, y) in the nodes Pi , Qj of both grids simultaneously. It is worth noticing
that while the system matrix being composed, the analogues for the NNBK and NBK
arrays, but for rectangular grids, are essentially used. Note that in such formulation
the system matrix is obtained symmetrical and sparse.
For carrying out numerical experiments on checking the above method a program
was composed using the FORTRAN 77 language. The results, illustrating the oper-
ation of the described algorithm for a spatial contact problem of off-centre loading
of a square punch, located on a nonhomogeneous elastic half-space, are given in
Sect. 3.5.1.
Additionally, we pay some attention to the properties of a similar postprocessor
which we have developed on the base of an arbitrary triangulation of a general-type
domain. In this case the dual grid are Dirichlet–Voronoy type polygons, formed by
the preprocessor. The solution is sought in the nodes of the general triangulation
of the domain (see the preprocessor algorithm) in such a way that it has minimum
deviation from the two solutions, specified on dual grids in a metric l2 with weights.
Similarly to the postprocessor algorithm on a square, described here, the choice of
[email protected]
weights affects the properties of the sought solution. One of the functions, provided
by the postprocessor, is a possibility of extrapolation of any of the dual solutions to
the whole general triangulation, and then to the whole domain.
174 3 Computer Implementation of Boundary-Element Algorithms

Thus, application of grids in duality in the boundary-element method enables:

– an essentially new numerical solution of the problem, still genetically related to
the one, built on the initial grid, to be obtained without large extra efforts;
– approximate solutions to be easily interpolated to the whole domain since they are
located in the nodes of the two known grids;
– an efficient postprocessor algorithm to be built due to the interdependent location
of nodes of the two grids when each node of one grid is enclosed by the known
nodes of the other one;
– new smoothed solutions to be obtained by means of the postprocessor, which,
inheriting the most important features of the two initial ones, have much less
numerical error.

3.4 Automated Construction of Spatial Grids of Boundary

Elements on the Surfaces of Contact of Deepened
Foundation Structures with Soil
Application of boundary-element method for numerical modeling of processes of
contact interaction of deepened foundation structures with soil media required a
boundary-element grid to be built on a contact surface, the quality of the grid being
to a great extent responsible for the numerical solution to be obtained. Since a cor-
rect calculation should take into account the contact interaction of the foundation
and the soil not only on the bottom, but also on the lateral surfaces, the contact
domains to be discretized are in general for foundation structures of a non-canonical,
piecewise smooth shape. Therefore, the description of geometry of such contact sur-
faces by an explicit functionality of z = f(x, y) type can be used, as a rule, only for
separate areas.
At present, for some spatial problems of computational mechanics considerable
progress has been achieved in plotting calculation grids on complex-shaped bodies.
In spatial problems of gas dynamics, liquid mechanics, heat and mass transfer
formation of high-quality grids on a given boundary surface provides further con-
structions of 3-D finite-element or finite-difference grids [48, 49, 52, 64, 65, 86, 87,
133, 145]. Surfaces whose geometry can be easily described (or whose structural
topology is known) by an initial rough discretization, are considered. Using the the-
ory of Gaussian surfaces for obtaining spatial surface grids with specified properties,
a multistep numerical procedure is built, including:

– parametrization of the surface area within the specified boundaries;

– numerical solution of the boundary problem of elliptical type to determine the
field of parameters in the nodes of a natural or a rough grid on a specified surface;
–[email protected]
determination of new grid nodes on the surface in accordance with the parametric
coordinates found;
– finding Gaussian surface equations provided that the calculation grid nodes belong
to a simplest curved surface.
3.4 Automated Construction of Spatial Grids of Boundary Elements 175

Note that one of the main requirements of most of the algorithms to the properties
of the approximating spatial grids of the given type, is the grid line orthogonality.
The necessity of discretization of a complex-shaped surface arises in the prob-
lems of calculation of thin-walled structures, in particular, complex-shaped shells
whose median surface is not described by a simple analytical expression [58, 78,
148]. After the choice of the reduction surface (in general, of a non-canonical
shape), the main attention is paid to efficient procedures of parametrization of
these surfaces. Shells of various geometry are effectively calculated using the finite-
element method. Therefore, application of flat finite elements as well as those based
on the relations of three-dimensional theory of elasticity, simplifies the surface
parametrization. Surface approximation in theory of shells assumes for a discrete
set of points on the median surface a coordinate system to be introduced and coef-
ficients of the first and the second quadratic forms to be found, i.e. values of the
vector function
r = X(u, v)i + Y(u, v)j + Z(u, v)k

for the introduced coordinate system to be found as well as the derivatives of r over
u and v up to the third order inclusive. As noted in [148], in order to obtain high-
accuracy numerical results on the static and vibrations of shell structures one should
apply either a dense grid of mesh into flat finite elements, or apply approximations
of higher accuracy.
All the experience on the discretization of complex-shaped surfaces, known from
the studies in the above fields of computational mechanics, can be without princi-
pal modifications applied for boundary-element modeling of contact deformation
processes. However, in the spatial contact problems for rigid punches under our
consideration, when flat boundary elements with piecewise constant approximation
of contact pressures are used, direct application of methods of approximation of
higher-order surfaces, developed in liquid mechanics and theory of shells will be
hardly justified. These methods, can be, above any doubt, more efficient for solving
spatial contact problems for flexible foundation structures, i.e. for the problems with
the account of local strength.
In accordance with the above stated, for the approximation of contact surfaces
of rigid foundation structures constant boundary elements (for which the unknown
contact stress has the same value over the whole element [33]) of triangular and
quadrangular shape will be used. The vertices of the elements will be located on
the contact surface itself. Note that at such approach for flat contact domains the
boundary elements will always entirely belong to the foundation structure surface.
For the parts, characterized by a curvature, we assume that the required accuracy
of the surface approximation can be achieved with the increase of the discretization
degree.
It should be noted that our analysis of the geometrical shape of most of the real
foundation structures has shown that they can be divided in two large categories:
[email protected]
block type configurations with flat faces and rotation bodies. Therefore, the descrip-
tion of the discretization methods will be performed in detail for the foundation
structures of these two types. The examples of methods of surface discretization will
176 3 Computer Implementation of Boundary-Element Algorithms

be shown for foundation structures of various shapes, their parts being described by
known equations in a Cartesian coordinate system.
Hereinafter it will be demanded that at any method of the contact surface mesh
into boundary elements there should be no crossings or overlaps, and points on a
non-flat surface should be the nodes of the adjacent boundary elements. In general,
the mesh is supposed to be mixed, i.e. with simultaneous presence of triangular and
quadrangular boundary elements. Global enumeration of elements is used with a
possibility of preferential selection of a certain group of elements for the sake of
convenience of subsequent processing of the numerical results. This is caused by
the piecewise constant approximation of the unknowns and the completely filled
structure of the matrix of the resolving linear equation system. Local enumeration
of nodes for each element is also taken counterclockwise, the observation being
performed from the direction of the external normal to the contact surface (i.e. from
the side of soil).
High efficiency of the automatic generation of boundary-element vertices is
favoured by the concept of boundary macroelements being used, which has been for
a long time applied for computer implementation of the finite-element method [120,
123]. This concept is used mostly for automatic generation of coordinates and num-
bers of node points for flat problems [120]. Some details of the software implemen-
tation of the macroelement concept in the finite-element method for spatial prob-
lems are described in [123]. Spatial surface macroelements of various types will
be also used here in the proposed algorithm of construction of boundary-element
grids. Then some specific features of application of macroelements in the boundary-
element method will be discussed, since there is not much literature available con-
cerning this approach.
For all surface types fragmental discretization is performed that corresponds to a
preliminary mesh of the foundation structure surface into boundary macroelements.
At the next stage boundary macroelements are meshed into separate boundary ele-
ments with automatic generation of coordinates and nodes. Surface fragments with
the simplest topology are taken as boundary macroelements. As a rule, these are flat
quadrangles, parts of cylindrical, conical and spherical surfaces. In specific cases
boundary macroelements, for which surface equations can be given in a determined
way, are used.
The mesh of boundary macroelements into separate boundary elements is per-
formed regularly and not necessarily uniformly. The degree of nonuniformity in a
separate boundary macroelement over different directions is given parametrically
and, if possible, takes into account the presupposed character of contact stress vari-
ation.
An important aspect of the fragmental discretization is adjustment of number of
boundary elements at the lines of conjugation of complex boundary macroelements.
It is required to improve the numerical solution and for the convenience of process-
ing and treatment of the obtained results.
[email protected]
Now consider the procedures, used for automatic construction of the boundary-
element grid for the main types of boundary macroelements.
3.4 Automated Construction of Spatial Grids of Boundary Elements 177

Flat boundary macroelements. The geometry of each flat boundary macroele-

ment is determined by its vertices. Let the global coordinates of the node points of
a triangular (quadrangular) boundary macroelement be known:
 
Xi , Yj , Zi , i = 1,2, . . . , l; l = 3 (4) .

The mesh of boundary macroelements into separate elements is performed on the

base of isoparametrical element technique [22].
The description of geometry in the plane of a quadrangular boundary macroele-
ment is obtained by using interpolation formulae

X = X1 ϕ1 + X2 ϕ2 + X3 ϕ3 + X4 ϕ4 ,
Y = Y1 ϕ1 + Y2 ϕ2 + Y3 ϕ3 + Y4 ϕ4 , (3.9)
Z = Z1 ϕ1 + Z2 ϕ2 + Z3 ϕ3 + Z4 ϕ4

where ϕi , i = 1,4 are simplest linear shape functions given by

1 1
φ1 = · (1 − ξ1 ) · (1 − ξ2 ) , φ2 = · (1 + ξ1 ) · (1 − ξ2 ) ,
4 4

1 1
φ3 = · (1 − ξ1 ) · (1 − ξ2 ) , φ4 = · (1 + ξ1 ) · (1 − ξ2 ) .
4 4
Dimensionless variables ξ1 and ξ2 are identified with local coordinates in the
plane of a standard square |ξ1 | ≤ 1, |ξ2 | ≤ 1. If now a uniform (Fig. 3.19a) or
condensing towards the boundary (Fig.3.19b) grid of elements is applied to a stan-
dard square, then a linear isoparametric transformation (3.9) will transform them
into global boundary elements without violation of interelement continuity and with
required regular enumeration of nodes and vertices.
Based on the above interpolation algorithm a subroutine TRPSPS was developed
for discretization of an arbitrary quadrangular boundary macroelement with the fol-
lowing call:

CALL TRPSPS(M,N,X,Y,Z,ALPHA,K1,K2),

where M, N is the number of the boundary element grid lines along two adjacent
directions, X, Y, Z are 4-component arrays of global coordinates of the boundary
macroelement vertices, ALPHA is a parameter of the boundary-element grid conden-
sation at the boundary element edges (ALPHA = ∅ at uniform discretization), K1 is
the number of boundary elements having been created before calling the subroutine
TRPSPS; K2 is the total number of boundary elements after calling the subroutine
[email protected]
TRPSPS, K2 = K1 + M ∗ N.
Discretization of the contact surface of a foundation structure of the block type
with flat quadrangular faces into boundary elements is reduced in the case under
178 3 Computer Implementation of Boundary-Element Algorithms

(a)

(b)

Fig. 3.19 Local coordinates for a quadrangular boundary macroelement: (a) uniform grid; (b)
non-uniform grid

consideration to specifying the global coordinates of vertices and a subsequent call

to the subroutine TRPSPS. For example, discretization of the contact surface of a
pyramidal foundation (with five flat faces) with a uniform boundary element grid is
obtained as a result of the following set of FORTRAN statements:

K1 = ∅
CALL TRPSPS(N1,N2,X1,Y1,Z1,Æ,K1,K2)
CALL TRPSPS(M,N1,X2,Y2,Z2,∅,K2,K3)
CALL TRPSPS(M,N2,X3,Y3,Z3,∅,K3,K4)
CALL TRPSPS(M,N1,X4,Y4,Z4,∅,K4,K5)
CALL TRPSPS(M,N2,X5,Y5,Z5,∅,K5,KK).

As a result, the common area of global coordinates of vertices and characteristics

of the boundary elements will be filled:

COMMON /ELPS/ AX(KK,4),AY(KK,4),AZ(KK,4),L(KK).

[email protected]
The practical experience of discretization of contact surfaces of various founda-
tion structures of block type with flat faces has shown that the most often repeated
3.4 Automated Construction of Spatial Grids of Boundary Elements 179

Fig. 3.20 Scheme of

discretization of a tetrahedral
boundary macroelement

module (substructure) is a four-faced belt with the height H, symmetrical with

respect to the planes OXZ and OYZ and possessing rectangular shape with the
dimensions 2∗ A1 ×2∗ B1, 2∗ A2 ×2∗ B2 at the depths Z = C1 and Z = C2, (H
= C2 −C1) (Fig. 3.20). For the convenience of discretization of this substructure a
special subroutine NABS was developed, called by the following statement:

CALL NABS(M,N1,N2,A1,A2,B1,B2,C1,C2,K1,K2).

The formal parameters of the NABS subroutine are described as follows:

M is the number of divisions of the boundary-element grid vertically (in depth);

N1, N2 are the numbers of divisions of the boundary-element grid horizon-
tally for two faces of the belt, respectively;
A1, B1 and A2, B2 are half-sizes of the belt rectangles in the Z = C1 and Z
= C2 planes, respectively;
K1, K2 are the parameters of the amount of the boundary elements with the
same sense as in TRPSPS subroutine.

The result of NABS subroutine is, similarly to the case of TRPSPS subroutine, the
filled common area COMMON/ELPS/ for the vertice coordinates and characteristics
of the boundary elements. The total number of boundary elements, formed by NABS
subroutine, is KK = 2 M∗ (N1 + N2).
For the discretization of triangular flat faces (triangular boundary macroele-
ments), the necessity of which happens not so often, the most convenient is applica-
tion of a mesh of a standard triangle (a 2-D simplex) {0 ≤ ξ1 ≤ 1,0 ≤ ξ2 ≤ 1 − ξ1 }
with subsequent interpolation (Fig. 3.21). For a given integer m≥1 in the standard
triangle a system of points is introduced [139]:

ξ1 = i/m, ξ2 = j/m, 0 ≤ i, j ≤ m, 0 ≤ i + j ≤ m .
[email protected] (3.10)

System (3.10) includes the triangle vertices as well as the points obtained by
mesh of each side of the triangle into m equal parts as the points of intercept of the
180 3 Computer Implementation of Boundary-Element Algorithms

Fig. 3.21 Local coordinates for a triangular boundary macroelement

straight lines parallel to the triangle sides and passing through the mesh points. A
uniform mesh of an arbitrary flat triangular face with the vertices

P1 (X1 ,Y1 ,Z1 ) ,P2 (X2 ,Y2 ,Z2 ) , P3 (X3 ,Y3 ,Z3 )

into m2 triangular boundary elements is easily obtained after the application of the
following parametric interpolation [33]:

X = X1 ξ1 + X2 ξ2 + X3 ξ3 ,
Y = Y1 ξ1 + Y2 ξ2 + Y3 ξ3 ,
(3.11)
Z = Z1 ξ1 + Z2 ξ2 + Z3 ξ3 ,
ξ1 + ξ2 + ξ3 = 1 .

This results in an algorithm for discretization of triangular boundary macroele-

ments, which enables the required triangular boundary-element grid condensation
degree to be achieved rather easily. Besides, triangles, similar to the initial macroele-
ment, are formed. If necessary, the obtained boundary-element grid can be easily
condensed near the selected vertex of the boundary macroelement using, e.g. quasi-
uniform grids [69] instead of Eq. (3.10).
Figure 3.22 shows the surface of rectangular prismatic foundation blocks dis-
cretized by means of uniform grids of quadrangular boundary elements. Rigid plates
or monolithic blocks with dimensions as shown in Fig. 3.22a, are used, e.g. to design
foundations for forging hammers as well as molding foundry machines, crushers,
mills, presses etc. [62, 118, 152]. Rectangular prismatic monolithic concrete foun-
dations with relative deepening h/d ∼ 2 ÷ 4 (Fig. 3.22b) are widely applied for
contact-line (console, anchor, fixing etc.) masts [71, 80], bridge piers of caisson
type [140] etc. The experience of foundation engineering of the recent years shows
[email protected]
the efficiency of application of trench (or slotted) foundations (Fig. 3.22c), built by
diaphragm wall method under columns and walls of buildings and structures [34,
98, 99, 132], as well as for high-voltage power line masts [82].
3.4 Automated Construction of Spatial Grids of Boundary Elements 181

(b) (c)
(a)

Fig. 3.22 Rectangular prismatic foundations of various dimensions

(a) (b) (c)

Fig. 3.23 Rectangular prismatic foundations with structural modifications: (a) slotted foundation
with lateral widenings; (b) with a set-off in the bottom; (c) with a pyramidal tip

Rectangular prismatic foundations with simplest structural modifications are

shown in Fig. 3.23. Slotted foundation structures with lateral extensions in the form
of longitudinal ribs of various cross-section configuration [8] (Fig. 3.23a). Calcula-
tions and the experimental data [18, 19, 130] show that foundations with a set-off in
the bottom (Fig. 3.23b) (with intermediate preparation, variable cross-section) work
more efficiently than those with a flat bottom. Rectangular prismatic foundations
with a pyramidal tip (displacement piles) are widely applied in industrial and civil
engineering (Fig. 3.23b).
In industrial engineering the most spread foundation structures are massive
pedestal footings [130]. Figure 3.24a, b show pedestal footing constructions for one
and two treads, respectively. The cross-section shape of such foundations is square
or rectangular for central and off-centre loading, respectively.
[email protected]
In pile foundation construction common prismatic piles are with high efficiency
replaced with foundations of short pyramid-shaped piles in view of their load capac-
ity for the same volume [26, 63, 119]. Figure 3.25 shows discretized contact surfaces
182 3 Computer Implementation of Boundary-Element Algorithms

(a) (b)

Fig. 3.24 Benched pier foundations: (a) with one bench; (b) with two benches

(a) (b)

(c) (d)

[email protected]
Fig. 3.25 Pyramidal foundations: (a) with square cross-section; (b) with rectangular cross-section;
(c) with a prismatic head; (d) a bipyramidal pile
3.4 Automated Construction of Spatial Grids of Boundary Elements 183

(a)
(b)

Fig. 3.26 Block foundations with an inclined bottom: (a) vertical lateral sides; (b) inclined lateral
sides

(a) (b) (c)

Fig. 3.27 Wedge-type foundations: (a) wedge-slotted; (b) asymmetrical wedge pile; (c) wedge
pile foundation with a console block

with soil for pyramidal foundations of most of the known shapes, operating well at
horizontal and momental loads [4, 5, 55, 72, 84].
For structures with strutted elements, e.g. three-hinged frames, application of
foundations with oblique bottom is advisable [130]. The size of such foundation and
the angle of the bottom inclination with respect to horizontal is chosen in such a way
that the eccentricity of load resultant on the foundation founding level be close to
zero [109]. Massive foundations with oblique bottom and vertical or slanted lateral
faces, with plotted boundary-element grids are shown in Fig. 3.26a, b respectively.
To reduce horizontal displacements and to increase the load-carrying capacity
under horizontal forces various wedge foundations are used [71, 73, 98, 117]. The
examples of wedge foundations are presented by a wedge-slotted foundation and an
asymmetric wedge pile shown in Fig. 3.27a, b respectively, along with the boundary-
element grids, as well as wedge pile foundation with a corbel block (Fig. 3.27c).
Figure 3.28 presents variations of post foundations of complex shape, widely
[email protected]
used in practice, for which the contact surface with soil is formed by planes [98, 130,
134]. They can be easily discretized. The extended lateral surface of such foundation
structures is the main factor responsible for the increase of their characteristics.
184 3 Computer Implementation of Boundary-Element Algorithms

(a) (b)

(c) (d)

Fig. 3.28 Complex-shaped foundations with flat sides: (a) double-slotted; (b) double-wedge; (c)
slotted cross-like; (d) box-like

Boundary macroelements on the surfaces of rotation bodies. An axisymmetric

surface is obtained as a result of rotation of a curve around the axis of symmetry.
Assume OZ to be the symmetry axis. Then, due to the axial symmetry, the equation
of the curve in the cylindrical coordinate system is given by

r = F (z) ,F  (z) = 0,r = x2 + y2 , C1 ≤ z ≤ C2 (3.12)

In order to plot the boundary-element grid on the rotation surface (Fig. 3.29) the
segment [C1 , C2 ] is divided into m parts, not necessarily equal, by points zk = C1 +
(C2 − C1 )·q(α, tk ), k= 1, 2, . . ., m+ 1 where tk = (k−1)/m, · q(α, t) is a function,
mapping the [0, 1] segment into itself and chosen, depending on the form of F(z)
for condensation of the points of division near the [C1 , C2 ] segment ends, α is the
condensation parameter.
The interval [0, 2π] of the angular coordinate ϕ variation is meshed into n
equal parts by ϕj =  · j, j= 0, 1, . . ., n−1 planes with a step ϕ = 2π/n.
[email protected]
Considering each two consecutive meridional sections ϕj , ϕj+1 (j = 0, n − 1),
form flat boundary elements by calculating the coordinates of their vertices as
follows:
3.4 Automated Construction of Spatial Grids of Boundary Elements 185

Fig. 3.29 Discretization of a rotation surface into boundary elements

(j)   (j)   (j)

X1 (k) = F (zk ) · cos φj ; Y1 (k) = F (zk ) · sin φj ; Z1 (k) = Zk ;

(j)   (j)   (j)

X2 (k) = F (zk+1 ) · cos φj ; Y2 (k) = F (zk+1 ) · sin φj ; Z2 (k) = zk+1 ;

(j)   (j)   (j)

X3 (k) = F (zk+1 ) · cos φj+1 ; Y3 (k) = F (zk+1 ) · sin φj+1 ; Z3 (k) = Zk+1 ;

(j)   j   (j)
X4 (k) = F (zk ) · cos φj+1 ; Y4 (k) = F (zk ) · sin φj+1 ; Z4 (k) = Zk ;

k = 1,2, . . . , m + 1;j = 0,1,2, . . . , n − 1.

Note that in the case when F(z) = 0 (i.e. in the foundation structure end-
point) the second and third vertices of the boundary elements merge and quad-
rangular boundary elements are degenerated into triangular. The total number of
the flat boundary elements on the rotation body fragment under consideration is
K= m·n, increasing with the mesh number both in depth (m), and in the angular
coordinate (n).
According to the accepted terminology, the rotation surface fragment, approxi-
mated by using flat quadrangular and triangular boundary elements will present a
boundary macroelement on a rotation body. Subroutines for automatical discretiza-
tion of boundary macroelements have been developed for the most often used rota-
tion surface fragments:

cylindrical: F(z) = R = const;

R2 − R1
conical: F(z) = R1 +
[email protected] (z − C1 );
C2 − C1

spherical: F (z) = ± R2 − (z − z0 )
186 3 Computer Implementation of Boundary-Element Algorithms

where R is the cylinder or sphere radius, R1 , R2 are the radii of the upper and
lower bases of a truncated cone, respectively, z0 is the Z-coordinate of the centre of
the sphere.
Another rather efficient method of discretization of boundary macroelements
should be mentioned. It is more convenient than the above one for the cases when
the rotation body surface is given by an equation of the type

z = f (r), r = x2 + y2 (3.13)

We take the advantage of the fact that any rotation surface fragment located
between the z= C1 and z= C2 planes, is always projected onto XOY plane as a circle
or a ring. Then, possessing the discretization of these planar canonical domains and
by applying Eq. (3.13), one can easily approximate boundary macroelements on a
rotation body by planar boundary elements of triangular and quadrangular type, the
requirements of interelement continuity being directly fulfilled. Subroutines, elabo-
rated for the boundary macroelement mesh on a

plane: z = C = const, 
conical surface: z = z0 ± a √x2 + y2 ,
spherical surface: z = z0 ± R2 − r2

employ regular uniform and non-uniform grids on a circle and a ring, enabling the
spatial discretization of the required quality to be obtained.
Full discretization of the foundation structure in the form of a rotation body
is obtained by combining boundary macroelements of simplest shapes considered
above. When the total number of the boundary elements is sufficiently large and
their size is sufficiently small, then, as follows from the knowledge of physics, it is
natural to expect the contact interaction of such foundation structure with soil (after
the surface approximation by an ensemble of flat boundary elements) under external
load not to be much different from the real foundation behaviour, and the approxi-
mate solution of the contact problem to converge to the exact one at the increase of
the number of flat boundary elements and at their size decrease. Practical calcula-
tions, described in the subsequent sections, show a rather good convergence.
Figures 3.30–3.32 show typical examples of mesh of the contact surface of
axisymmetric foundation structures with soil, using flat boundary elements.
Boundary-element discretization is performed most easily when the foundations
have only cylindrical components (Fig. 3.30). In such cases boundary macroele-
ments of only three types are used: a circle, a concentric ring, and the lateral surface
of a cylinder. Rigid cylindrical foundations (Fig. 3.30a) (short single bored piles,
short-length shell-piles, caissons and shell caissons) are now rather widely applied
in industrial, agricultural and transport engineering [25, 50, 74, 79, 100, 153]. They
[email protected]
appear economically justified for construction of industrial and civil buildings with
rather small number of storeys (workshops, garages, warehouses, shops etc.), live-
stock factories, greenhouses, pipe bridges, city transportation contact-line masts,
3.4 Automated Construction of Spatial Grids of Boundary Elements 187

(a) (b) (c)

Fig. 3.30 Foundations of cylindrical components: (a) simple; (b) telescopic; (c) with a collar

(a) (b) (c) (d)

Fig. 3.31 Foundation with shaft widenings of various type: (a) bored pile foundation with a con-
ical widening; (b) bored pile foundation with a spheroconical widening; (c) bored pile foundation
with a hemispherical widening; (d) mushroom-shaped foundation

power transmission towers, bridges. Telescope-shaped cylindrical foundations (Fig.

3.30b) find application for construction of oil and gas drilling platform legs on the
sea shelf [146]. Sometimes near the cylindrical pile heads a stabilizing collar is made
(Fig. 3.30c) what enables the load-carrying capacity of the foundation structure to
be essentially increased [79].
A considerable effect on the increase of the load-carrying capacity of short cylin-
drical foundations can be achieved due to widening of a part of their shank (bottom,
head or the intermediate part) [143, 153]. The surfaces of cylindrical foundations
with different widenings, discretized by means of flat boundary elements, are shown
[email protected]
in Fig. 3.31. In accordance with soil drilling technology, the shape of the widening
surface is determined by the junction of conical, cylindrical, and spherical frag-
ments. As seen from Figs. 3.31a–d application of circular, ring-shaped, cylindrical,
188 3 Computer Implementation of Boundary-Element Algorithms

(a) (b) (c)

Fig. 3.32 Conical foundations: (a) in the shape of a truncated cone; (b) cylindrical pile with a
conical tip; (c) biconical

conical, and spherical boundary macroelements in a required sequence automat-

ically provides consistent discretization of the foundation structures with various
types of widenings.
A relatively new, progressive type of foundations are now short conical piles (Fig.
3.32), especially hollow ones, whose application enables the concrete consumption
to be reduced by 25–30% [25, 27]. The shape of the lateral surface results in an
explicitly pronounced increase of the load-carrying capacity due to the formation
of a compression area around the pile. A good-quality approximation for conical
surfaces is achieved using 2–3 boundary macroelements at relatively small total
number of flat boundary elements (Fig. 3.32a–c).
Figures 3.33–3.35 present the versions of discretization of contact surfaces for
various complex-shaped foundation structures. Each of the configurations consid-
ered is characterized by simultaneous presence of both flat elements and fragments
of second-order surfaces, mostly cylindrical. When the surface fragments are com-
bined in a whole unit with the required conjunction of the boundary-element ver-
tices at adjacent partitions, an individual approach to each foundation structure is
required, preferably taking into account the specific features of the contact stress
distribution.
A version of almost uniform discretization of the contact surface of the soil and a
caisson with rounded (cylindrical) abutting ends is presented in Fig. 3.33 [140]. Two
flat rectangles and two symmetrical halves of straight circular cylinder surface are
taken as boundary macroelements on the lateral surface (Fig. 3.33a). The bottom is
discretized by using three boundary macroelements – a rectangle in the middle and
two semicircles, each of them possessing only two triangular boundary elements, all
[email protected]
other boundary elements being quadrangular (Fig. 3.33b).
Typical examples of discretized surfaces of contact with soil for slotted founda-
tions with non-flat bottom shape [98] are shown in Fig. 3.34. The slotted foundation
3.4 Automated Construction of Spatial Grids of Boundary Elements 189

(a) (b)

Fig. 3.33 Caisson with rounded walls: (a) lateral surface; (b) bottom

(a) (b)

Fig. 3.34 Slotted foundations with non-flat bottom: (a) curved bottom shape in the longitudinal
cross-section; (b) wedge-slotted with an enlarged footing

(a) (b) (c)

Fig. 3.35 Pile raft foundations: (a) vertical cylindrical pile with a rectangular raft; (b) inclined
cylindrical pile with a rectangular raft; (c) vertical cylindrical pile with a circular raft

surface with a curved bottom shape first was presented as a combination of three
[email protected]
boundary macroelements (Fig. 3.34a): a rectangular prismatic one in the upper part,
two symmetrical segments on the lateral surface as well as a ruled surface on the
bottom. Not that the shape of the bottom has no major importance for the program
190 3 Computer Implementation of Boundary-Element Algorithms

module developed to perform the discretization and can be determined by a sub-

routine function. A rather complex structure of a wedge-slotted foundation with
a pedestal [53] was subject to surface discretization using the following standard
boundary macroelements: a four-faced belt-shape part of a pyramid, an open lateral
surface of a circular cylinder with a horizontal axis, and two truncated circles at the
abutting ends (Fig. 3.34b). The program provides consistence of the discretization
while combining of the above boundary macroelements as well as a possibility to
specify different widening shapes and boundary-element grid condensation degree
for the given type of the foundation structures and enables a good-quality surface
approximation to be obtained using a moderate (≈300 ÷ 350) number of flat bound-
ary elements.
Discretization of foundations with pile rafts needs special attention. A low pile
raft is a concrete block of rectangular or circular cross-section, mounted on the pile’s
head to transfer the load from the structure. Such structures combine typical features
of both shallow foundations and pile foundations.
Figure 3.35 shows boundary-element grids on the surfaces of pile raft foun-
dations with single vertical (Fig. 3.35a–c) and inclined (Fig. 3.35b) piles. Such
foundations are effectively used for structures with relatively small number of
storeys as well as for reconstruction of buildings and structures [77]. Since, in
general, a pile and a pile raft can be located eccentrically, special boundary
macroelements of circular and rectangular shape with internal cutoffs of various
shape had to be developed. The discretization on the internal contour was con-
sistent with the boundary-element nodes in the horizontal sections of the pile
substructure.
A rather typical one among the foundations with pile rafts under consideration is
the discretization of the contact surface for a foundation with an inclined cylindrical
pile and a rectangular pile cap (Fig. 3.35b). A scheme, explaining the sequence of
formation of the boundary-element grid on the surfaces of contact of the pile raft and
the pile with soil, is shown in Fig. 3.36. Three consistent boundary macroelements
were used: the lateral surface of a straight circular cylinder, restricted by an orthog-
onal planar section and a planar section, inclined at an angle ϕ (ϕ being the angle
of the cylinder axis inclination to the horizontal plane), a circle on the cylinder’s
abutting end, and a rectangle with an ellipse-shaped cutoff. For the convenience of
computation of the boundary-element vertex coordinates on the inclined cylinder
surface an auxiliary coordinate system O X Y Z is introduced with the origin in
the centre of the cylinder’s abutting end and the coordinate axis O Z coinciding
with the cylinder axis. This corresponds to a counterclockwise rotation of the global
OXYZ system by angle α = π/2–ϕ around the OX axis and a transfer of the coordi-
nate system origin to the (0, Lcosϕ, Lsinϕ) point. The formulae for the coordinate
transformation to the global system are as follows

⎧
[email protected]

⎨x = x ,
y = z cos ϕ + y sin ϕ + L · cos ϕ, (3.14)
⎩
z = z sin ϕ − y cos ϕ + L · sin ϕ
3.4 Automated Construction of Spatial Grids of Boundary Elements 191

Fig. 3.36 Scheme of discretization of the contact surface of soil and an inclined pile with a low
rectangular raft

and the formulae for the coordinate transformation from the global system are given
by
⎧ 
⎨ x = x,
y = y · sin ϕ − z · cos ϕ, (3.15)
⎩ 
z = y · cos ϕ + z · sin ϕ + L

where L is the length of the segment on the cylinder axis between the restricting
upper and lower sections. The equation of the elliptical cut-off in the pile raft

x2 + y2 sin2 ϕ = R2

is obtained as the line of interception of the straight circular cylinder

(x )2 + (y )2 = R2

with the plane

[email protected]
z = 0 (z = y cot ϕ − L) .
192 3 Computer Implementation of Boundary-Element Algorithms

Further mesh of the contact surface into the boundary elements is performed
by sectioning by planes, passing through the cylinder axis. It is convenient at first
to obtain the coordinates of flat boundary-element vertices from the formulae of
Eq. (3.15)-type in the cylinder-linked O X Y Z system, and then to transform them
to the global OXYZ system according to Eq. (3.14). The considered example shows
how one can semiautomatically obtain discretization of various non-axisymmetric
surfaces by preliminary introduction of local coordinate systems for separate surface
fragments (boundary macroelements).
Thus, the developed algorithms of boundary-element discretization of contact
surfaces of deepened foundations with soil are characterized by a simple logical
structure and short computation time (due to the use of the explicit formulae to
determine the coordinates of the boundary-element vertices). Besides, the possibil-
ity of automatic mesh of grids enables a sequence of naturally nested discretiza-
tions to be created. As a result, at the application of condensing grids, numerical
solutions are effectively extrapolated with the estimation of accuracy. The proposed
simplified construction of grid approximations enables the grids to be automati-
cally reconstructed multiply, adapting them by local condensations to the sought
solution.

3.5 Test Examples of Numerical Modeling of Spatial Problems

of Contact Interaction

3.5.1 Contact Problems for Flat Punches with a Smooth Base

Numerical solution of spatial contact problems on regular grids. Consider as the

main example a spatial contact problem for a rigid punch with a circular cross-
section, interacting with an elastic homogeneous half-space. No friction in the
contact domain is considered. Suppose the punch to be loaded at the centre by
a vertical force P. It is known [67] that in such a case the problem has an exact
solution
 
P 1 − v2 P p∗
W= ,p (r) = √ = √ 2
,
2Ea 2π a a − r
2 2
2π 1 − (r/a)
where α is the punch radius, r is the distance from the punch centre to the obser-
vation point, E is the deformation modulus, ν is the Poisson ratio, p∗ = P/α2 . This
problem is one of the few spatial contact problems having an analytical solution
what enables one to estimate the accuracy of its numerical solutions, obtained on
different grids.
Compare the numerical solutions of this problem using the boundary-element
method for four types of the circular domain discretization:
[email protected]
(1) mesh into triangles and quadrangles by coordinate curves in polar coordinate
system with a constant radial step (a uniform mesh, Fig. 3.37a);
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 193

(a) (b)

Fig. 3.37 Regular grids of boundary elements, applied for solving the contact problem of a
centrally loaded circular punch: (a) uniform; (b) with condensation at the boundary

(2) mesh into triangles and quadrangles using the polar coordinate system with a
variable radial step (a non-uniform mesh with condensation towards the bound-
ary, Fig. 3.37b);
(3) cyclic mesh into triangular elements with condensation at the boundary
(Fig. 3.17a);
(4) dual mesh into Dirichlet-Voronoy type cells (Fig. 3.17c), obtained by the
method described in Sect. 3.3.2.

At the numerical solution of the problem in the second case the circular con-
tact domain was discretized into triangular and quadrangular elements by means of
radii and concentric circles condensing towards the boundary. The condensation was
applied to take into account a sharp increase of contact stress near the punch edge.
The radii of the concentric circles were calculated using the following quasiuniform
dependence:
 
exp β · tj − 1 j
rj = a , tj = , j = 1, 2, . . . , L.
exp (β) − 1 L
The condensation degree was controlled by the choice of the parameter β.
The first and the second grids consist of 400 elements and contain 10 concentric
layers. The third grid contains 216 elements and consists of four concentric layers.
Finally, the last grid contains 145 elements, located in five concentric layers, the
first (central) layer containing only one element. Note that the number of Dirichlet–
[email protected]
Voronoy elements of the last grid is by one and half times smaller than the number
of elements of the triangular grid (3) and by factor of 2.76 smaller than the number
of elements of the first two grids.
194 3 Computer Implementation of Boundary-Element Algorithms

Table 3.1 Characteristics of the boundary-element solution of the contact problem for a round
punch on regular grids

Discretization type N S/α2 W/α σ Mean square deviation

Uniform in polar 400 3.12869 0.47802 274.0 4.3265 × 10−3

coordinates (β = 0)
Non-uniform in polar 400 3.12869 0.47114 147.0 3.3212 × 10−2
coordinates
(β = −3)
Triangulation with a 216 3.13761 0.47760 100.6 1.3489 × 10−2
displacement on the
boundary
Dual Dirichlet–Voronoy 145 3.13761 0.46868 274.2 1.0686 × 10−3
grid
Exact solution – 3.14153 0.46875 – 0

The results of the numerical solution of the contact problem for the four dis-
cretization schemes as well as the exact solution data for ν = 0.25, P= Eα2 are
listed in Table 3.1and shown in Fig. 3.38.
As follows from the presented data, the calculation using the uniform grid of
400 elements produces results which are quite acceptable from the engineering
point of view. At the same number of elements the accuracy of the solution can be,
though slightly, increased by the grid condensation near the punch boundary (what
is seen in Table 3.1 for the calculations while choosing the condensation parameter

[email protected]
Fig. 3.38 Exact and approximate solutions of the contact problem for a centrally loaded circular
punch using regular dual grids
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 195

β = –3). In this case the total (mean square) error increases by an order of magnitude
and the condition number of the matrix of the resolving system of linear algebraic
equation is reduced almost twice. This means that at β = –3 the boundary-element
grid rather correctly takes into account the unlimited growth of the solution (contact
pressures) near the punch boundary. The increase of the mean square error here is
most likely due to the extension of the numerical results for the contact pressure
function on the condensed grid to the area where the solution varies with increasing
rate. Evidently, for each non-uniform discretization of the type under considera-
tion an optimal value of the condensation parameter β exists, whose determination
requires additional computation time consumption. For example, the calculations at
β = −5 give the following characteristics of the numerical solution:

W/a = 0.47032, σ = 384.8, mean square error = 2.94958 × 10−1

It is seen that at a slight correction of the relative settlement, which is an integral

characteristic of the numerical solution, the boundary-element function of contact
pressures is determined with higher error near the punch edge. Simultaneously the
condition number strongly increases, which can be related to the degeneracy of the
boundary elements into rectangles of a rather small thickness. In order to obtain
uniformly suitable numerical solution over the whole circular domain one should
simultaneously with the grid condensation near the punch edge increase the dis-
cretization degree over the angular coordinate.
The triangular discretization according to the scheme (3) in comparison with the
above cases contains almost twice less boundary elements, but leads to the numer-
ical results of practically the same degree of accuracy. The accompanying decrease
of the condition number σ by factor of 3 and 1.5, respectively, is related to a suc-
cessful choice of the size of the triangular elements, close in shape to equilateral, as
well as to the decrease of their area with moving away from the centre of the circle.
Proceeding from the triangular grid to the dual mesh of Dirichlet–Voronoy type
polygons enables not only the discretization degree to be reduced almost by factor
of 1.5, but also the numerical results with the improved accuracy both for displace-
ments and contact pressures to be obtained (Table 3.1, Fig. 3.38).
The most essential effect of application of the dual grid of Dirichlet–Voronoy
polygons is a considerable decrease of the mean square error with respect to all the
cases of discretization considered, as well as the possibility of a practically exact
solution to be obtained in a sufficient vicinity to the punch boundary. Besides, the
numerical solution, obtained using the dual grid, gives the values for the boundary-
element vertices what is convenient for interpolation.
Then we give an example of application of a postprocessor to the results calcu-
lated on the base of the boundary-element method. Consider the contact problem for
a square punch 2α × 2α, loaded by an eccentric vertical force P = E0 α2 , εx = 0.1α,
y = 0.1α. The punch rests on an elastic nonhomogeneous half-space with deforma-
ε[email protected]
tion module increasing with depth E = E0 (1 + Bz/h) (See Sect. Section 1.5.3). The
computations have been performed for two dual grids containing 8 × 8 and 7 ×
7 nodes, provided that h/α = 1.0, B= 1.0, and the Poisson ratio ν = 0.25. Level
196 3 Computer Implementation of Boundary-Element Algorithms

(a) (b)

Fig. 3.39 Lines of equal dimensionless contact pressure for an off-centre loaded square punch (a)
before and (b) after the postprocessor application

lines for the contact pressures p̄(x,y) = p(x,y)/pav (pav = P/4α 2 ), built using all the
113 values in the nodes of both dual grids, are shown in dimensionless coordinates
x̄ = x/α,ȳ = y/α in Fig. 3.39a.
The boundary-element solutions were also processed by the postprocessing
method proposed in Sect. 3.3.1. Since the postprocessing algorithm (or simply post-
processor) depends on the ratio γ1 : γ2 : γ3 , then, without any loss of generality,
suppose γ2 = 16 and denote γ1 and γ3 as ω1 and ω2 , respectively. Here ω1 is the
weight of the “old” values and ω2 is the weight of the “new” values in the grid
nodes. To obtain the best result one should choose ω1 and ω2 according to the spe-
cific features of the problem being solved. For the present example by a series of
test calculation the following parameters were chosen: ω1 = 256, ω2 = 16. This has
led for the whole set of 113 nodes to a new approximate solution whose level lines
are shown in Fig. 3.39b. Comparison of the level lines in Fig. 3.39a, b has shown
the prostprocessor to result in a noticeable data smoothing. Unfortunately, for the
contact problem under consideration the exact solution is unknown what does not
enable us to compare the accuracy of the approximate solution before and after the
postprocessing. A considerable excess of γ1 = ω1 = 256 over γ2 = ω2 = 16 and γ3 =
ω3 = 16 was caused by the requirement of the new smoothed solution not to be very
much different from the two initial ones.
In practice, in case the boundary-element method being applied, the contact
domain often has complex shape. For the punches of noncanonical shape under
spatial loading of a general type it is, as a rule, impossible to predict in detail the
solution character. For the correct simulation of the contact pressure field a non-
[email protected]
uniform grid is required with condensations in the expected areas of rapid variation
of the solution. Simultaneously, the triangulation of complex-shaped flat domains by
various methods results, as a rule, in irregular grids with a large number of bound-
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 197

ary elements. By the example of the numerical solution of the contact problem for
a round punch using regular (uniform and non-uniform) grids it was shown that the
application of dual polygonal grids of Dirichlet–Voronoy type results in an essen-
tial decrease of dimensionality of the algebraic analogue of the problem and, due
to the adaptation to the solution, does not decrease its accuracy. Therefore, for the
punches of various cross-section shape at spatial loading of a general type (with
possible uplifting of the bottom from the base) the following procedure of obtaining
a numerical solution is proposed:

(1) a possibly uniform triangulation of the contact domain is performed using a

relatively large number of triangles, close to equilateral (at the a priori known
character of contact pressure variation the triangular grid is condensed in the
areas of rapid variation of the solution);
(2) switching to a dual polygonal grid of Dirichlet–Voronoy cell type is performed,
and on this grid the numerical solution of the contact problem with the pressure
values in the nodes of the initial triangular grid;
(3) the found discrete solution is interpolated for the whole contact domain and
visualized using isobar lines.

Below we present the results of numerical experiments for irregular grids, having
enabled us to check the efficiency of the proposed approach, to evaluate the accuracy
of the boundary-element algorithm by comparing the solutions found to the known
analytical solutions or those obtained on the base of other known methods, as well as
to show the possibilities of the proposed method for finding the contact interaction
parameters in practical problems.
A round punch at central and off-centre loading (irregular grids of boundary ele-
ments). Figures 3.40–3.43 and Table 3.2 present the numerical solutions of the con-
tact problem for a round punch in case it being pressed into a homogeneous elastic
half-space (ν = 0.25) by a central force P. The obtained boundary-element solu-
tions on a sequence of condensed, essentially irregular grids were compared with
the known exact solution of this problem [67]. Consider a dimensionality reduc-
tion coefficient KTV = NT /NV where NT is the number of equations of the resolving
system of the problem on a triangular grid, NV is the number of equations of the
corresponding system on a dual polygonal grid. As one can see, the increase of the
contact domain discretization degree results in stable numerical solutions which at
linear extrapolation over the value (NV )−1 in the limit at (NV )−1 →0 (or (NV )−1
→∞) lead to a result, practically not differing from the exact one (Table 3.2). The
application of dual grids in the examples under consideration has enabled the dimen-
sionality of the discrete analogue of the contact problem to be reduced by the factor
of KTV = 1.5, 1.8, and 1.84, respectively. With the increase of the triangulation den-
sity the KTV value is seen to have, though a slight, but a stable trend to increase what
[email protected]
extremely important while solving real problems, requiring always as minute dis-
cretization as possible.
It is seen from Figs. 3.40c and 3.41c, and 3.42d where the isolines of dimen-
sionless contact pressures p̄(x,y) = p(x,y)/paV (paV = p/S is the average pressure in
198 3 Computer Implementation of Boundary-Element Algorithms

(a) (b)

(c) (d)

Fig. 3.40 To the solution of the contact problem for a circular punch on an elastic half-space
(C37): (a, b) discretization of the contact surface using dual triangular (NT = 56) and polygonal
(NV = 37) grids; (c) isolines of dimensionless contact pressure; (d) exact solution

the contact domain with the area S) are shown, that with the increase of the number
of the triangular grid nodes (or the number of Dirichlet–Voronoy type polygonal
boundary elements) the isobars are still less sensitive to the character of irregularity
of the boundary-element grid and take the shape of concentric circles. As follows
from Fig. 3.40c, d the calculated pressures for a rough grid (C37), though slightly
differ from the exact values in the central part of the punch (r/a≤0.6), still do not
enable one to obtain isolines, independent of the boundary element shape. Appli-
cation of essentially dense grids (C307and C513) results in a contact pressure dis-
tribution, practically coinciding with the exact one almost in the whole punch area
[email protected]
(r/a≤0.9) with the except of a narrow zone near its edges (0.9≤r/a≤1). Compar-
ison of the solutions for two dense grids C307 and C513 shows that the solution
correction near the punch edge due to a mere increase of the number of the bound-
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 199

(a) (b)

(c) (d)

Fig. 3.41 To the solution of the contact problem for a circular punch on an elastic half-space
(C307): (a, b) discretization of the contact surface using dual triangular (NT = 552) and polygonal
(NV = 307) grids; (c) isolines of dimensionless contact pressure; (d) exact solution

ary elements of the irregular (rather uniform, without condensations) grid is rather
small. The solution just moves towards the boundary due to the general decrease of
the boundary element size (Fig. 3.43). A considerable effect in the increase of the
solution accuracy near the boundary condition separation line can be evidently also
achieved, similarly to the case of application of non-uniform regular grids (see Sect.
Section 3.3), when the contact pressure variation near the punch edge is correctly
taken into account.
Application of a detailed discretization of the circular contact domain by a C513
grid has enabled the contact pressure distribution at the asymmetric loading of the
[email protected]
punch to be studied (Fig. 3.44). It follows from the results of the calculations per-
formed that the application of a vertical force even with a slight eccentricity results
in a rather noticeable violation of the contact pressure field symmetry, characterized
200 3 Computer Implementation of Boundary-Element Algorithms

(a) (b)

(c) (d)

Fig. 3.42 To the solution of the contact problem for a circular punch on an elastic half-space
(C513): (a, b) discretization of the contact surface using dual triangular (NT = 904) and polygonal
(NV = 513) grids; (c): isolines of dimensionless contact pressure; (d) exact solution

by an under pressure zone located opposite to the direction of the eccentric displace-
ment of the applied vertical force. The obtained numerical solution of the problem
of the asymmetric full contact of a round punch with an elastic homogeneous half-
space (Fig. 3.44a) has practically coincided (cannot be distinguished on the plot)
with the known exact solution [1, 37, 90, 94]

1 x
p(x,y) = √ (P · a + 3 · · M),
2π a2 a −r
2 2 a

[email protected] x
W(x,y) = δ + εx ·
a
where the axial force P and the overturning moment M with respect to the OY axis
are related to the punch centre displacement δ under impression and the maximal
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 201

Fig. 3.43 Contact pressures for a circular punch on an elastic half-space, obtained using dual grids
of different condensation degree

displacement εx at the rotation, respectively, according to the formulae

 
2a E δ 2 4 Ea3  εx
[email protected]
P= ,M= · · .
2 1−v a 3 1 − v2 a

The solution of a similar problem, when an elastic half-space with the defor-
mation module increasing with depth E = E0 (1 + Bz/h) is considered as a base, is
202 3 Computer Implementation of Boundary-Element Algorithms

Table 3.2 Characteristics of the boundary-element solution of the contact problem for a round
punch on irregular grids

Discretization Contact Contact Slopes Relative Condition Mean square

type domain area pressures ψx · 104 settlement number σ deviation
S/α2 pmin /pmax ψy · 104 W/α

0.5381 −5.695
NV = 37 3.12869 0.49003 93.61 2.2191×10−
NT = 56, 2.7163 −4.944
KTV = 1.514
0.5084 −0.653
NV = 307 3.12869 0.47334 70.76 1.8368×10−
NT = 552, 3.7560 −1.098
KTV = 1.798
0.5065 −0.0321
NV = 513 3.13761 0.47216 115.3 1.7783×10−
NT = 944, 4.2218 −0.6978
KTV = 1.840
0.5036 −0.0004
Extrapolation 3.14242 0.47041 – 1.6911×10−
4.9165 −0.3551
0.5 0
Exact solution 3.14159 0.46875 – 0
∞ 0

shown in Fig. 3.44b (see Sect. Section 1.5). The considered example shows that the
proposed boundary-element method of numerical integration enables to take into
account simultaneously, within a unique approach, complex spatial loading of rigid
punches and nonclassical properties of elastic bases.
Negative (tensile) contact stress is not taken up by an elastically compressed
medium, and at the area of their formation within the contact area an uplifting zone
(gap) is formed. Therefore, the developed algorithm takes into account unilateral
constraints. The specific feature of the calculation in this case consists in the fact
that in the course of the punch loading process the calculation scheme can vary
(a part of the punch surface will not participate in the contact interaction). Thus
modified formulation of the problem qualitatively changes the contact interaction
pattern since the calculation scheme becomes a function of load. Since the uplifting
zone is not known a priori, the solution is sought by iteration, by means of correc-
tion and switching unilateral constraints. For the elements in the contact area it is
(n)
necessary that the condition pi ≥ 0, Wi = 0 be fulfilled, and the elements on
which tensile stress arises (pi <0), are included into the uplifting zone only in case
the condition Wi ≥ 0 being valid, where Wi = Wi+ − Wi− is the gap opening
(n) (n)

value, equal to the difference of the base and the punch displacements. The pro-
cess of more exact specification of the contact domain is performed by iteration,
the equilibrium conditions being fulfilled, until in the subsequent approximation all
the included forces pi are positive and the calculation results coincide with initial
[email protected]
prerequisites. In the problems, having been solved using this algorithm, the number
of approximations seldom exceeded 7 cycles. Note that our calculations and stud-
ies of contact interaction with the account of unilateral constraints were based on
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 203

the general properties of the systems containing unilateral constraints. For this case
convergence of iterative processes of finding the functioning system is shown e.g. in
the book by Reznikov [110], and its uniqueness – in the studies of Rabinovich [107,
108]. It is noted in the latter of the above references that by a limiting transition
the theorem of uniqueness is extended to the systems with infinitely large number
of unilateral constraints, including structures, resting on an elastic continuum with
unilateral consraints.
Figure 3.44c, d demonstrate the abilities of the developed method for solving the
contact problem of an essentially inclined round punch with areas of uplifting from
the elastic homogeneous half-space on its bottom. We used an irregular boundary-
element grid C513 with a sufficient degree of condensation for calculations and for
interpolation of the contact pressures at off-centre punch loading. It is seen from the
calculation data that the increase of eccentricity of the external vertical force appli-

(a) (b)

(c) (d)

[email protected]
Fig. 3.44 Contact pressure isolines at off-centre loading of a round punch (C513): (a, b) ex /à=
0.15; (c) ex /à= 0.35; (d) ex /a= 0.5; (a, c, d) homogeneous elastic half-space; (b) elastic half-space
with the deformation modulus, linearly increasing with depth (B = 1.0, μ = h/a = 1.0)
204 3 Computer Implementation of Boundary-Element Algorithms

cation point results in an appearance and subsequent broadening of uplifting areas

and an increase of the number of iteration process cycles for their determination.
Evidently, the numerical solution of the problem of an essentially inclined punch
will be the more accurate, the higher is the degree and the better is the quality of
discretization of the contact domain into boundary elements.
Note that the solution of the contact problem at the asymmetric loading of a round
punch when the contact domain is a part of a circle, has been considered earlier in
[21, 111, 112, 125]. Rvachev with collaborators [21, 111, 112] have carried out a
study to determine the contact domain of a round punch with a half-space depending
on the eccentricity of the force application using the structural method. They have
found the problem solution in the form of a linear combination of the complete
system of Chebyshev polynomials, and for the calculations they had to make the
type of the function, determining the line L0 of the punch uplifting from the base,
completely specified. This line was assumed to be a two-parameter parabola. The
results we have obtained without any assumptions on the contact domain shape,
show (Fig. 3.44c, d) that the L0 line is not a parabola, its configuration being to a
great extent more complicated with approaching the angular points of the L0 line
interception with the punch contour. In the vicinity of these points a complicated
character of the stressed state is observed, when the pressure varies from zero to
infinity depending on the direction of approach to these points.
A rigid round punch with partial uplifting of its bottom from an elastic homo-
geneous half-space due to asymmetrical impression was also considered in [125].
To determine the contact domain, to calculate contact stress and parameters of the
punch displacement as a rigid solid, a numerical method is applied, much similar
to the boundary-element method. The circular contact domain is meshed into 316
squares, entirely inside the circle, which are identified with circular boundary ele-
ments with the same centres and equivalent areas. For each such circular element a
constant pressure is assumed. Displacements of the homogeneous linearly stressed
half-space surface in the centres of the circular boundary elements due to a uniform
pressure over the circular domain are found from explicit formulae using a combi-
nation of full elliptic integrals of the first and second order. Besides, the proposed
algorithm [125] takes essential advantage of the symmetrical shape of the punch.
The actual contact domain is determined by iteration using a simple switching of
unilateral constraints. The zero-pressure line L0 , bounding the contact domain, is
obtained by square interpolation of discrete values from two points in the directions,
parallel to the diameter, passing through the circle centre and the point of application
of the off-centre vertical force. For different eccentricities of the external force after
such extrapolation the uplifting lines L0 are completely determined in the course of
the numerical solution of the problem and, contrary to the structural method [112],
without any additional assumptions. Essentially, one should confess that the method
of solving the problem of an essentially inclined round punch, proposed in [125],
has set the foundation of direct numerical solving of spatial contact problems for
[email protected]
rigid flat punches in partial contact with elastic bases. Nevertheless, an essential
limitation for this approach is the application of the exact solution of the prob-
lem (expressed via Bessel functions) of a load, uniformly distributed over elements
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 205

(a) (b)

(c) (d)

Fig. 3.45 Square punch with angular cutoffs at central loading: (a) contact domain; (b) irregular
triangular grid (NT = 1118); (c) dual polygonal grid (NV = 608); (d) dimensionless contact pressure
isolines

of circular shape, acting on an elastic homogeneous half-space. Such approach is

related to the contact domain discretization performed using a square grid, whose
cells are to be put into correspondence to circles of equivalent area. However, the
most convenient and widely used method of discretization of flat complex-shaped
domains is non-uniform triangulation with a required condensation in the supposed
areas of the solution variation with considerable gradients. Therefore, it is rather
evident that the boundary-element approach using adaptive irregular grids, proposed
here, is well applicable for the class of spatial contact problems under consideration.
As noted before, such an approach is also indifferent with respect to the contact
model of the elastic base under consideration. If for the base contact model, the
[email protected]
influence function can be integrated analytically over the circular domain, then the
application of the boundary-element method becomes even more efficient. Indeed,
since the calculation grids of Dirichlet–Voronoy type consist mostly of the boundary
206 3 Computer Implementation of Boundary-Element Algorithms

(a) (b)

(c) (d)

Fig. 3.46 Square punch with side cutoffs at central loading: (a) contact domain; (b) irregular tri-
angular grid (NT = 1156); (c) dual polygonal grid (NV = 630); (d) dimensionless contact pressure
isolines

elements, rather close in shape to regular polygons with the number of angles from
6 to 8 (Figs. 3.40b–3.43b, 3.45c–3.56c), the replacement of the latter by circles of
equivalent area will result in a more accurate approximation of the contact domain
after the corresponding aggregation of the elementary circles with less error than in
the case of replacement of traditional square elements [46, 125].
Thus, the performed comparison of the solutions of the problem of an essentially
[email protected]
inclined circular punch has shown that among the known approaches the developed
boundary-element method, including the application of dual, in the general case
irregular grids, possesses high university and universality, being revealed in (a) con-
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 207

(a) (b)

(c) (d)

Fig. 3.47 Square punch at central loading: (a) contact domain; (b) irregular triangular grid (NT =
1202); (c) dual polygonal grid (NV = 656); (d) dimensionless contact pressure isolines

sideration of contact domains of practically any complexity and their discretization

by any of the known methods, (b) interpolation of the numerical solution in auto-
matic mode with simultaneous determination of the a priori unknown shape of the
contact domain, (c) application of a broad spectrum of spatial contact models spec-
ifying the mechanical properties of elastic bases by means of the known influence
functions.
Below we consider the examples of solving spatial contact problems for the
[email protected]
punches of complex (noncanonical) shape, being first of all directly related to the
design of shallow foundation structures, having no exact solutions even in the case
of the classical model of a homogeneous elastic half-space being used as a base.
208 3 Computer Implementation of Boundary-Element Algorithms

(a) (b)

(c) (d)

Fig. 3.48 Square punch with an inclined asymmetric cutoff at central loading: (a) contact domain;
(b) irregular triangular grid (NT = 724); (c) dual polygonal grid (NV = 415); (d) dimensionless
contact pressure isolines

Punches of polygonal cross-section. Figures 3.45 and 3.46 show the dual
boundary-element grids and the results of solving spatial contact problems for
square punches with symmetrical cut-outs, the depth of the latter being 1/8 of the
square side, loaded by a central force. For the punch with angular cut-outs NT =
1118, NV = 608, KTV = 1.839, for the one with lateral cut-outs NT = 1156, NV = 630,
KTV = 1.835. It is seen from the equal contact pressure lines plotted in Figs. 3.45d
and 3.46d that in both cases the stressed states under the punches have characteris-
[email protected]
tic features, being revealed only near the punch boundaries. For the central part of
the punches, due to the small values of the cut-out depth parameter α = D/a, the
contact pressure isolines have the shape of regular ovals, practically the same as the
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 209

(a) (b)

(c) (d)

Fig. 3.49 Polygonal punch (foundation under a locomobile): (a) contact domain; (b) irregular
triangular grid (NT = 659); (c) dual polygonal grid (NV = 370); (d) dimensionless contact pressure
isolines

(a) (b)

(c) (d)

Fig. 3.50 Strip punch with semicircular face ends: (a) contact domain; (b) irregular triangular grid
(NT = 1046); (c) dual polygonal grid (NV = 567); (d) dimensionless contact pressure isolines

corresponding isobars for the square punch without cut-outs (Fig. 3.47). Note that
the punches, considered in Figs. 3.45 and 3.46, are the models of variable-width
foundations. Application of such foundations is economically sound since the soil
[email protected]
in the cut-out area participates in the foundation functioning and at equal external
load they can be used instead of rectangular (constant-width) foundations with the
same dimensions [130].
210 3 Computer Implementation of Boundary-Element Algorithms

(a) (b)

(c) (d)

Fig. 3.51 Eccentric ring-shaped punch: (a) contact domain; (b) irregular triangular grid (NT =
929); (c) dual polygonal grid (NV = 539); (d) dimensionless contact pressure isolines

Along with the punches with symmetrical cut-outs we have also considered (Fig.
3.48a) a square punch with a “deep”, asymmetrical (inclined at 60◦ to the side)
cut-out (a rigid foundation plate with a cut-out for the equipment). The triangula-
tion of the domain contained NT = 724 elements (Fig. 3.48b). Proceeding to a dual
polygonal grid, we have reduced the number of the boundary elements to NV = 415
what corresponds to the value KTV = 1.745 (Fig. 3.48c). The equal contact pres-
sure lines, built from the results of the numerical solution of the problem, indicate
that due to a strong asymmetry at the punch loading in the centre of the exter-
nal square contour, a complex stressed state with the presence of negative (ten-
[email protected]
sile) stress is formed. This means that for the final solution of the contact prob-
lem one should consider in detail the iteration process of determination of the con-
tact domain where the contact pressures will be negative. In practice, variation of
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 211

(a) (b)

(c) (d)

Fig. 3.52 Ring-shaped punch with a cutoff in the inner circle: (a) contact domain; (b) irregular
triangular grid (NT = 710); (c) dual polygonal grid (NV = 419); (d) dimensionless contact pressure
isolines

the shape and load parameters in order to accept a design solution, excluding the
foundation functioning with the bottom uplifting from the base, is usually applied
[102, 113].
A foundation for a locomobile is considered as a typical example of the solution
of a spatial contact problem for a punch of a rather complex polygonal cross-section
shape [135]. The geometrical dimensions of the punch loaded by a vertical force
with the application point in the coordinate origin are shown in Fig. 3.49a. A trian-
gular (NT = 659) and a polygonal (NV = 370) grids in duality (Figs. 3.49b, c) were
used for the numerical solution of the contact problem. The parameter of reduction
[email protected]
of dimensionality of the discrete problem in this example KTV = 1.781. Interpola-
tion of the numerical solution, found in the triangular grid nodes, enabled a detailed
pattern of the stressed state under the punch to be obtained using the lines of equal
212 3 Computer Implementation of Boundary-Element Algorithms

(a) (b)

(c) (d)

Fig. 3.53 Rectangular punch with an elliptical cutoff: (a) contact domain; (b) irregular triangu-
lar grid (NT = 1688); (c) dual polygonal grid (NV = 919); (d) dimensionless contact pressure
isolines

values of dimensionless contact pressures p̄ (x,y) (Fig. 3.49d). Note that the spatial
contact problems for polygonal punches on a homogeneous half-space, shown in
Figs. 3.45, 3.46 and 3.49, have been studied earlier using electrical analogy method
[30, 31]. The comparison of the experimental isobars and our theoretical solu-
tions has shown a good not only qualitative, but also quantitative agreement of the
results.
A strip-shaped punch with half-round abutting ends. The shape and dimensions
of the punch are presented in Fig. 3.50. The punch under consideration corresponds
e.g. to a bottom in caissons with rounded walls constructed as deep foundations for
bridge piers [140]. The dual grids, discretizing the contact domain, are shown in
Figs. 3.50b, c and contain NT = 1046 and NV = 567 elements, respectively (KTV =
1.845). Figure 3.50d shows the plots of isolines of dimensionless reactive pressures,
corresponding to central impression of the punch (a= 3.0, b= 2.85) into an elas-
tic homogeneous half-space. Here we also note that the results of calculation, per-
[email protected]
formed at the parameter values a= 0.2, b= 0.4 with a high degree of accuracy (the
error does not exceed 1.3%) agree with the solution obtained earlier by structural
R-function method [112].
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 213

(a) (b)

(c) (d)

Fig. 3.54 A square punch with a circular cutoff: (a) contact domain, (b) triangular irregular grid
(NT = 1521), (c) dual polygonal grid (NV =848), (d) dimensionless contact pressure isolines

Punches with multiply connected contact domain. Figures 3.51 and 3.52 illus-
trate the solution of the spatial contact problem for centrally loaded asymmetrical
ring-shaped punches. Such punches are the models for foundations under tower-
type structures to be erected on non-uniformly compressed bases or on uniformly
compressed ones subject to momental loads in a given direction [7, 11, 13–15]. For
both punches r1 /r2 = 0.6 is assumed.
For the eccentric ring-shaped punch the value of eccentricity ε0 , by which
[email protected]
the internal circle is displaced with respect to the centre, is taken as 0.2r2
(Fig. 3.51a). The initial triangular grid contains NT = 929 elements (Fig. 3.51b).
The dual polygonal grid is composed of NV = 539 elements (Fig. 3.51c), KTV =
214 3 Computer Implementation of Boundary-Element Algorithms

(a) (b)

(c) (d)

Fig. 3.55 Square punch with a square cutoff: (a) contact domain; (b) irregular triangular
grid (NT = 1448); (c) dual polygonal grid (NV = 814); (d) dimensionless contact pressure
isolines

1.724. The ring-shaped punch, shown in Fig. 3.52a, has the value of the param-
eter D, characterizing the internal circle cut-off, equal to 0.19r2 . The triangular
grid contains NT = 710 elements and has condensed areas in the vicinity of the
angular points on the internal contour of the punch (Fig. 3.52b). By proceeding
to the dual polygonal grid we managed to reduce the discrete problem dimension-
ality with respect to NV = 419 unknown contact pressures by factor of KTV = 1.695
[email protected]
(Fig. 3.52c).
Plotted from the calculation results equal contact pressure lines, reproducing the
features of the stressed state under the asymmetric ring-shaped punches at ver-
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 215

(a) (b)

(c) (d)

Fig. 3.56 Square punch with a triangular cutoff: (a) contact domain; (b) irregular triangular
grid (NT = 1332); (c) dual polygonal grid (NV = 745); (d) dimensionless contact pressure
isolines

tical loading in the centre of the external circle, are shown in Figs. 3.51d and
3.52d. It should be noted that while for the symmetrical ring-shaped punch differ-
ent forms of analytical and approximate solutions are well known [20, 29, 38–40,
47, 56, 57, 96, 112, 141], no spatial contact problem solution for the asymmetri-
cal ring-shaped punches considered here has been proposed before by any of the
methods.
Consider the solution of a practically important spatial problem for the multiply
[email protected]
connected contact domain when the punch has a rectangular shape with an elliptical
cut-out (Fig. 3.53a). Due to such shape of the bottom, the foundation can be made
with smaller support point. The presence of the unsupported elliptical area in the
216 3 Computer Implementation of Boundary-Element Algorithms

central part of the foundation results in the fact that the compressive stress from the
central load grows faster than the tensile stress due to moments [88]. Besides, the
choice of the elliptical shape for the internal contour leads to the minimal decrease
of the moment of inertia of the foundation bottom what enables it to bear higher
bending moments than at any other cut-out shape. Note that the contact domain
under consideration in this example (and its particular case of a square with a cir-
cular cut-out, Fig. 3.54) also arises at the interaction of rectangular pile rafts (for
inclined or vertical cylindrical piles) with the soil surface [12]. Figures 3.53b, c
show a triangular (NT = 1688) and a polygonal (NV = 919) grids (KTV = 1.857) in
duality which were used for the numerical solution of the contact problem for a cen-
trally loaded rectangular punch (a2 = 1.5, b2 = 1.0) with an elliptical cut-out (a1 =
0.5, b1 = 0.25). The obtained solution along the long axis of the punch appeared
graphically indistinguishable from the solution, obtained by structural method (R-
function method) in [112]. The full pattern of the stressed state under the punch is
shown in Fig. 3.53d where the equal contact pressure lines are built based on inter-
polation (using the initial triangular grid) of the boundary-element solution found
on a dual polygonal grid.
Solutions for contact problems at multiply connected contact domain of polyg-
onal shape with rectangular and triangular internal cut-outs in a square punch are
considered in Figs. 3.55 and 3.56, respectively, as examples to demonstrate the pro-
posed boundary-element algorithm, including the dual grid application. The contact
pressure isolines built using the results of the numerical solution on irregular dual
grids (Figs. 3.55d and 3.56d) as well as the solutions of the above considered con-
tact problems for the punches with cut-outs, show wide abilities of the proposed
approach in the case of different doubly connected contact domains bounded by
piecewise smooth curves.
Quantitative characteristics of the developed method based on the results of
its application in 28 contact problems for complex-shaped punches with different
degree of non-uniform discretization of the contact domain are given in Fig. 3.57.
Figure 3.57a shows the dimensionality reduction coefficient KTV = NT /NV and Fig-
ure 3.57b illustrates computation time for dual grid calculations. It is seen that in
problems important for practical applications (with the number of the boundary ele-
ments above 250) the proposed method enables one to reduce essentially the dimen-
sionality of the algebraic analogue of the contact problem (on the average by factor
of 1.8) as well as to decrease the computation time (which is varied as a square of
the system dimensionality) in comparison with the case only triangular grids being
applied.
Thus, the performed comparison of the numerical results with both analytical
solutions and those obtained by other approximate methods enables the boundary-
element method to be successfully applied for solving engineering problems of
calculation of rigid foundation structures from the base deformation. The boundary-
element method is made much more efficient due to the application of not one grid,
[email protected]
but two interlinked (dual) grids for the solution finding domain, which are used
both at the preliminary stage of discretization and directly in the course of compu-
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 217

(a)

(b)

Fig. 3.57 Efficiency characteristics in case irregular dual boundary-element grids being used: (a)
dimensionality reduction coefficient; (b) computation time

tations as well as for subsequent processing of approximate solutions. In particu-

lar, it is shown that the problems of numerical modeling of contact interaction of
rigid punches with nonclassical elastic bases can be without particular difficulties
solved at any shape of bottom. The proposed boundary-element method is appli-
cable with equal convenience to simply and multiply connected contact domains.
The accuracy of the numerical results, by far sufficient for practical purposes, is
obtained at relatively small dimensionalities of the algebraic analogue of bound-
ary integral equations and piecewise constant approximation of the contact pressure
[email protected]
field.
218 3 Computer Implementation of Boundary-Element Algorithms

3.5.2 Contact Problems with the Account of the Deepening Factor

for Axisymmetric Punches, Interacting with an Elastic
Half-Space
Spatial contact problems for rigid inclusions and deepened punches in an elas-
tic medium have been the subject of numerous studies causes by the necessity of
investigation of stress concentration issues in a spatial formulation and their sig-
nificance for the development of mechanics of composite materials [6, 105] as
well as for geotechnical problems regarding the studies of contact interaction of
the foundation structures with soil [24]. By now the problems of such type have
been studied most extensively for the case the elastic medium being infinite space.
Approaches have been developed for studying axisymmetric stress-strained state
of the elastic medium near the cavities or rigid inclusions, as a rule, of spherical
or ellipsoidal shape, when at the infinity the stress components are arbitrary lin-
ear functions of the Cartesian coordinates [103, 104]. In the case of semi-infinite
domains (of elastic half-space or elastic layer type), solution of this class of prob-
lems becomes much more complicated due to the necessity to satisfy the boundary
conditions of the absence of stress on a free surface or the conditions of continuity
of stress and displacements on the surfaces of the layer contact. Approaches with the
application of finite-difference and finite-element methods appear hardly efficient in
this case.
Below we consider the examples of application of numerical-and-analytical
boundary-element method for solving spatial contact problems for an absolutely
rigid inclusion and deepened punches in an elastic isotropic linearly stressed half-
space (with stress-free boundary surface) at their force loading along the symme-
try axis. As an auxiliary state consider an elastic half-space with a cavity whose
boundary is totally identical to the contact surface of the rigid inclusion or the deep-
ened punch. The fundamental singular Mindlin’s solution for an elastic half-space is
used. The structure nonlinearity due to the introduction of unilateral constraints on
the contact between the rigid inclusion and the elastic half-space, not being under
tension and resulting in the violation of the full contact conditions, is taken into
account. Six equilibrium conditions are included into the integral equation system
of the spatial contact problem, which are required to determine the parameters of
the displacement of the inclusion of an arbitrary shape as a rigid solid (See Sect.
Section 2.1).
For numerical solution of these problems direct boundary-element method is
applied (in the assumption of a piecewise constant approximation of the contact
stress function), resulting in linear algebraic equation systems with constraints in
the form of inequalities p(n) ≥0. Discretization of the rigid inclusion contact surface
is performed automatically, using flat boundary elements of triangular and quadran-
gular shape according to the method employing boundary module-elements (Sect.
[email protected]
3.4). A possibility of non-uniform condensation of the boundary element grid in
the areas of contact stress concentration. While calculating the matrix coefficients
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 219

Fig. 3.58 Calculation

scheme for the contact
problem of a rigid sphere in
an elastic half-space

for the resolving system of the linear algebraic equations, the integrals of the sin-
gular part of the Mindlin’s solution are determined analytically, and those of the
regular part are obtained numerically, using quadratures whose order depends on
the size and remoteness of the boundary element under consideration from the col-
location point (See Sect. Section 2.3, See Appendix B). Having solved the finite-
measure analogue of the integral equation system, one can determine the displace-
ments and rotation angles of the rigid inclusion with respect to the chosen spa-
tial coordinate system as well as contact stress vector components for each bound-
ary element, averaged over its area. The stress-strained state in any point of the
elastic half-space, containing the rigid inclusion or the deepened punch, can be
then easily calculated by direct integration of the found solution of the contact
problem.
Test calculations have been performed for few exact solutions of contact
problems for an elastic half-space, known from the literature, as well as for the solu-
tions, obtained numerically in the axisymmetric formulation using other approxi-
mate approaches, available for the case of the half-space. The effect of the elastic
properties of the medium as well as of relative deepening on the characteristic of
contact stress distribution and load-displacement plots for the inclusions and the
punches, is shown.
A rigid spherical inclusion in an elastic half-space. In an elastic homogeneous
half-space z > 0 consider a nonhomogeneity in the form of an absolutely rigid inclu-
sion of spherical type. A vertical force P is acting on a rigid sphere of radius a, deep-
ened inside by distance h; the half-space is considered free of stress at the boundary
surface (Fig. 3.58). In such formulation there is no exact solution of this spatial con-
tact problem. For its numerical solution using the boundary-element method, the
spherical contact surface is subject to discretization into flat triangular and quad-
rangular boundary elements by means of vertical and horizontal sections in a way
shown in Fig. 3.59. Some results of the performed extensive systematic calcula-
tions, enabling the numerical solution convergence to be estimated, are listed in
Tables 3.3 and 3.4. It should be noted that in all cases the solutions of the problem
in the full (Sect. 2.2) and axiially symmetrical (Sect. 2.4.1) formulations have been
always coinciding within the machine accuracy. Since in the first case the conden-
sation degree of the discretization is limited by the computer resources, for numer-
[email protected]
ous calculations using dense boundary element grids the axisymmetric version was
applied.
220 3 Computer Implementation of Boundary-Element Algorithms

Fig. 3.59 Discretization of a

spherical surface using flat
boundary elements

Table 3.3 Normal σ and tangential τ contact stress in meridional section of a spherical inclusion
at relative deepening h/a = 1, 100, 1000

m z/a Without the account of With the account of unilateral

unilateral constraints constraints
σ τ σ τ
1 2 3 4 5 6

h/a =1
1 0.01831 −0.00860 −0.01293 0.0 0.0
2 0.06152 0.17345 0.22415 0.17031 0.22180
3 0.13095 0.46263 0.50427 0.46271 0.50425
4 0.21212 0.53004 0.59668 0.52998 0.59676
5 0.29733 0.42830 0.63391 0.42830 0.63398
6 0.38246 0.28918 0.69239 0.28920 0.69245
7 0.46555 0.17608 0.77105 0.17612 0.77110
8 0.54600 0.10246 0.85703 0.10250 0.85708
9 0.62433 0.06466 0.94225 0.06470 0.94229
10 0.70236 0.05619 1.02382 0.05624 1.02384
11 0.78493 0.08194 1.11162 0.08200 1.11164
12 0.91268 0.14357 1.17952 0.14363 1.17952
13 1.08732 0.29764 1.26111 0.29770 1.26120
14 1.21507 0.43373 1.32531 0.43375 1.32519
15 1.29764 0.53920 1.29886 0.53926 1.29889
16 1.37567 0.64018 1.27954 0.64027 1.27965
17 1.45400 0.74622 1.24718 0.74624 1.24713
18 1.53445 0.85917 1.19827 0.85923 1.19828
19 1.61754 0.97979 1.12833 0.97985 1.12834
20 1.70267 1.10679 1.03076 1.10681 1.03073
21 1.78788 1.23694 0.89749 1.23704 0.89753
22 1.86905 1.36458 0.71814 1.36460 0.71813
23 1.93848 1.48182 0.48317 1.48189 0.48318
[email protected]
24 1.98169 1.53174 0.19160 1.53178 0.19160
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 221

Table 3.3 (continued)

m z/a Without the account of With the account of unilateral

unilateral constraints constraints
σ τ σ τ
1 2 3 4 5 6

h/a =100
1 99.0183 −0.97820 0.12789 0.0 0.0
2 99.0615 −0.94687 0.32283 0.0 0.0
3 99.1309 −0.87206 0.48338 0.0 0.0
4 99.2121 −0.78950 0.60984 0.0 0.0
5 99.2973 −0.70392 0.70811 0.0 0.0
6 99.3825 −0.61843 0.78433 0.0 0.0
7 99.4656 −0.53536 0.84426 0.0 0.0
8 99.5460 −0.45503 0.89124 0.0 0.0
9 99.6243 −0.37681 0.92860 0.0 0.0
10 99.7024 −0.29907 0.95918 0.0 0.0
11 99.7849 −0.21416 0.99599 0.0 0.0
12 99.9127 −0.08870 0.98830 0.0 0.0
13 100.0873 0.08874 0.98831 1.19945 3.39521
14 100.2151 0.21423 0.99611 0.80355 1.73530
15 100.2976 0.29914 0.95922 0.94858 1.66283
16 100.3757 0.37694 0.92886 1.00226 1.50023
17 100.4540 0.45517 0.89144 1.06916 1.37119
18 100.5344 0.53550 0.84433 1.14257 1.24983
19 100.6175 0.61867 0.78465 1.22256 1.12638
20 100.7027 0.70417 0.70830 1.30763 0.99164
21 100.7879 0.78983 0.61007 1.39522 0.83720
22 100.8690 0.87243 0.48356 1.48159 0.65311
23 100.9385 0.94728 0.32298 1.56360 0.43159
24 100.9817 0.97867 0.12792 1.59142 0.16676
h/a =1000
1 999.018 −0.97847 0.12795 0.0 0.0
2 999.063 −0.94702 0.32299 0.0 0.0
3 999.131 −0.87212 0.48316 0.0 0.0
4 999.212 −0.78981 0.61009 0.0 0.0
5 999.297 −0.70408 0.70824 0.0 0.0
6 999.382 −0.61837 0.78478 0.0 0.0
7 999.466 −0.53544 0.84399 0.0 0.0
8 999.546 −0.45508 0.89122 0.0 0.0
9 999.624 −0.37693 0.92861 0.0 0.0
10 999.702 −0.29923 0.95932 0.0 0.0
11 999.785 −0.21420 0.99616 0.0 0.0
12 999.913 −0.08869 0.98836 0.0 0.0
13 1000.087 0.08869 0.98831 1.19940 3.39520
14 1000.215 0.21421 0.99604 0.80363 1.73524
15 1000.298 0.29921 0.95935 0.94884 1.66314
16 1000.376 0.37698 0.92875 1.00247 1.50019
17 1000.454 0.45506 0.89122 1.06917 1.37103
[email protected]
18 1000.534 0.53550 0.84433 1.14274 1.24996
19 1000.617 0.61819 0.78414 1.22209 1.12588
20 1000.703 0.70422 0.70867 1.30777 0.99215
222 3 Computer Implementation of Boundary-Element Algorithms

Table 3.3 (continued)

m z/a Without the account of With the account of unilateral

unilateral constraints constraints
σ τ σ τ
1 2 3 4 5 6

21 1000.788 0.78961 0.60950 1.39514 0.83655

22 1000.869 0.87252 0.48384 1.48178 0.65347
23 1000.938 0.94672 0.32268 1.56316 0.43132
24 1000.982 0.97862 0.12801 1.59143 0.16688

Contact stress values on the surface of the spherical inclusion in a merid-

ional cross-section, obtained from the results of calculation for a sufficiently dense
boundary-element grid (N = 24×72 = 1728), are listed in Table 3.3. Both the tan-
gential τ = pτ /σ0 and normal σ = pn /σ0 stress values are made dimensionless using
a nominal pressure σ0 = P/4πα2 being the average pressure on the surface of the
sphere. At the discretization of the contact surface sectioning by horizontal planes
was performed with condensation towards the centre of the sphere what ha enabled
non-prolate boundary elements (close to squares and equilateral triangles) to be
used. As seen from the data presented, at different values of the sphere deepening
on its rear side (in the pole, closer to the half-space surface) negative both tangen-
tial and normal stress values arise. The calculations show that with the increase of
the deepening parameter the absolute values of the negative contact stress increase
and the contact area, where they are revealed, expands. As could be expected, the
account of unilateral constraints at any deepening parameter h/a ≥ 2 results after
an iterative process in a complete uplifting of the upper half of the sphere from the
elastic medium. In turn, the contact domain decrease leads to the increase of stress
as well as additional displacements acquired by the rigid sphere. Variation of dis-
placements of the rigid spherical inclusion versus depth at different values of the
Poisson ratio ν of the elastic half-space can be traced from Fig. 3.60. The calcu-

Table 3.4 Comparative characteristics of calculation of displacements of a rigid sphere at different

relative deepening values in an elastic half-space on condensed boundary-element grids (ν = 0.3,
P= E · a2 )

W/a
Number of boundary
elements N = m×n S/4πa2 h/a =1 h/a =100 h/a =1000

144 = 6 × 24 0.92279 0.29198 0.163789 0.162152

240 = 10 × 24 0.95780 0.28478 0.161602 0.159967
384 = 16 × 24 0.97416 0.28136 0.160836 0.159200
480 = 20 × 24 0.97886 0.28034 0.160649 0.159014
720 = 20 × 48 0.98305 0.27944 0.160035 0.158401
1440 = 20 × 72 0.98384 0.27928 0.159918 0.158284
1728 = 24 × 72
[email protected] 0.99399 0.27684 0.159815 0.158180

Linear extrapolation 0.99804 0.27624 0.159331 0.157697

According to Eq. (3.16) 1.00000 0.157639

3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 223

Fig. 3.60 Relative displacements of the rigid sphere depending on its depth in the elastic half-
space

lation results here were obtained in the axisymmetric formulation using the most
dense among the boundary-element grids we have used, providing high calcula-
tion accuracy at different depths (See Table 3.4). The calculated curves show that
with the increase of the relative deepening h/a the vertical displacements W/a of
the rigid sphere monotonously decrease, unlimitedly approaching the values deter-
mined based on the known solution of the Roben elastostatic problem for a rigid
sphere, sealed into an unlimited elastic space [92]

(5 − 6ν)(1 + ν)
W= P (3.16)
12π E(1 − ν)a

The displacements of the rigid sphere, calculated from Eq. (3.16), are also shown
in Fig. 3.60 by dotted lines. At high values of the relative deepening h/a the calcu-
lated values of W/a practically coincide with the exact ones, what shows that the dis-
placement values, obtained according to the proposed boundary-element approach,
can be considered reliable. In particular, it is seen from the plots that the displace-
ments of the rigid sphere monotonously grow with the increase of the Poisson ratio
ν of the elastic half-space.
Similarly to the considered contact problem for a rigid sphere, the cases of
inclusions of different shapes can be also studied. The most frequent in the liter-
ature [41, 42, 91, 97, 124, 151] are the studies of axisymmetric problems for rigid
inclusions in the shape of rotation bodies: rotational ellipsoids (spheroids), finite
cylinders, smoothly conjugated with half-spheres at abutting ends, circular cones,
[email protected]
smoothly conjugated with spherical segments, and some more. In the boundary-
element approach being developed, the inclusion can have any shape and undergo
a spatial load of a general type. The accuracy of the contact problem solution will
224 3 Computer Implementation of Boundary-Element Algorithms

mostly depend on the degree and quality of the contact surface discretization and
finally will be determined by the computer resources.
In geotechnics, rigid flat inclusions of various shape in an elastic half-space are
considered as a mathematical idealization of anchor foundation plates [121]. A solu-
tion of the axial loading problem for an ellipsoid, for whose semiaxes the following
condition is valid
a1 = a2 = α, a3 → 0,
is often used in order to estimate the displacement versus load dependences of such
anchors. Under such consideration, the inclusion becomes an infinitely thin circu-
lar disc (penny-shaped). Due to the above mentioned advantages of the boundary-
element approach it becomes possible to obtain rigidity characteristics of the
“deepened anchor + elastic geomaterial” system independently of the simplify-
ing assumptions concerning the anchor shape, thickness, direction of the pulling
load, etc.
Note that the problem of a rigid sphere deepened into an elastic half-space is
especially important for geotechnical calculations. This problem models the func-
tioning of a spherical enlarged footing of bored piles. Experimentally the absence
of limit equilibrium areas under the sphere and uplifting from the soil is observed
only in the upper part of the enlarged footing [23]. Besides, there is a good coinci-
dence of normal contact stress diagrams. This confirms the validity of application
of theory of elasticity to calculate the footings of bored piles with enlarged bases. It
should be also noted that in a prelimit state the soil under the tip of a pile without
enlarged footing also moves together with the pile. In the course of impression and
at pile driving a condensed core is formed which should be treated together with the
pile as a whole unit body. The core, independently of the tip shape, acquires a shape
close to hemispherical [85]. Note that for both flat and conical pile footing the core
shape has no significant differences. Therefore, while performing pile calculations
for the impression load, one can specify boundary conditions not for the tip surface,
but for the hemispherical core surface. Consequently, while developing engineer-
ing calculation methods, solution of contact problem for a rigid sphere in an elastic
half-space enables a more substantiated, with the account of the deepening factor,
evaluation of pile displacements and soil resistance at their tip.
Cylindrical punch, deepened in an elastic half-space. Figure 3.61a shows a cal-
culation scheme for the axisymmetric contact problem for an absolutely rigid cylin-
der, deepened into an elastic half-space and undergoing a vertical impressing load
P. Consider the punch and the half-space to be soldered at the contact surface. Due
to the axial symmetry of the problem, in the elastic half-space only radial pr and
vertical pZ stresses arise, and the punch will be displace only in the direction of the
force P.
In such formulation the axisymmetric contact problem for a deepened cylinder
was first solved numerically by Shishov in 1971 [126]. He averaged the contact
[email protected]
stresses over m cylindrical belts of a finite height on the punch lateral surface and
ring-shaped belts of a finite width on its bottom. The integral equation system of
theory of elasticity and the punch equilibrium equations are, due to such averaging,
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 225

(b) (c)

(a)

Fig. 3.61 Axisymmetrical contact problem for a cylindrical punch, deepened in an elastic half-
space: (a) calculation scheme; (b) discretization of the contact surface by flat boundary elements;
(c) location of collocation points on the generatrix

reduced to an algebraic equation system of the (2m+ 1)-th order with respect to 2m
unknown contact stress values and vertical displacement Z . The coefficients at the
unknowns are found by numerical integration of the fundamental solution over the
circular coordinate. The results of the problem solution with respect to normal σ and
tangential τ stresses in fractions of the mean contact stress σav = P/2πRh (R being
the cylinder radius, h – its height) are given in the end of Table 3.5. The calculations
were performed for the relative punch deepening h/R = 3 and for the contact surface
discretization into 20 belts (5 on the bottom and 15 on the lateral surface, Fig. 3.61c).
A characteristic feature of the solution is the presence of negative (tensile) normal
stress at the punch lateral surface in the area close to the bottom.
Our calculations, performed using the boundary-element method, are presented
in Table 3.5 at different density of the contact surface mesh into flat boundary ele-
ments. For all cases the cylindrical punch surface approximation was agreed with
the discretization into cylindrical and ring-shaped belts of the numerical study of
Shishov (15 + 5 = 20 boundary elements) in each meridional section θk = 2πk/n,
= 1, 2,. . .n (Fig.3.61b). The problem was solved both based on a full spatial for-
[email protected]
mulation and on the axisymmetric approach. For the cases of n= 12 (240 boundary
elements) and n= 20 (400 boundary elements) the calculation results have coincided
within the machine accuracy.
226 3 Computer Implementation of Boundary-Element Algorithms

Table 3.5 Normal σ and tangential τ contact stress in the axial section of a cylindrical punch on a
succession of condensing boundary-element grids (h/R = 3, ν = 0.3)

n = 12 n = 20
m z/a r/a σ τ r/a σ τ

1 0.1 0.966 1.97843 0.72064 0.988 1.99296 0.71949

2 0.3 0.966 0.87163 0.58824 0.988 0.87118 0.59347
3 0.5 0.966 0.61466 0.62024 0.988 0.61681 0.61874
4 0.7 0.966 0.45167 0.64667 0.988 0.45410 0.64528
5 0.9 0.966 0.34663 0.67429 0.988 0.34899 0.67323
6 1.1 0.966 0.27256 0.70191 0.988 0.27469 0.70115
7 1.3 0.966 0.21733 0.72988 0.988 0.21915 0.72939
8 1.5 0.966 0.17409 0.75911 0.988 0.17555 0.75890
9 1.7 0.966 0.13854 0.79091 0.988 0.13960 0.79104
10 1.9 0.966 0.10758 0.82723 0.988 0.10818 0.82783
11 2.1 0.966 0.07842 0.87125 0.988 0.07848 0.87251
12 2.3 0.966 0.04770 0.92846 0.988 0.04717 0.93077
13 2.5 0.966 0.00850 1.01241 0.988 0.00739 1.01725
14 2.7 0.966 −0.03929 1.11874 0.988 −0.04046 1.13133
15 2.9 0.966 −0.38419 1.98107 0.988 −0.40651 2.00987
16 3.0 0.874 2.60622 0.48425 0.893 2.56068 0.47496
17 3.0 0.681 1.53689 0.08493 0.696 1.49349 0.07438
18 3.0 0.489 1.35948 0.05645 0.500 1.32977 0.05175
19 3.0 0.301 1.25883 0.02817 0.307 1.23436 0.02596
20 3.0 0.129 1.21725 0.01037 0.132 1.19175 0.00957

n = 24 n = 40
m z/a r/a σ τ r/a σ τ

1 0.1 0.994 1.99712 0.72097 0.997 1.99927 0.72252

2 0.3 0.994 0.87088 0.59342 0.997 0.87071 0.59255
3 0.5 0.994 0.61744 0.61783 0.997 0.61780 0.61734
4 0.7 0.994 0.45481 0.64478 0.997 0.45519 0.64449
5 0.9 0.994 0.34969 0.67288 0.997 0.35006 0.67270
6 1.1 0.994 0.27532 0.70092 0.997 0.27566 0.70077
7 1.3 0.994 0.21969 0.72925 0.997 0.21998 0.72917
8 1.5 0.994 0.17599 0.75886 0.997 0.17622 0.75881
9 1.7 0.994 0.13991 0.79109 0.997 0.14008 0.79114
10 1.9 0.994 0.10835 0.82802 0.997 0.10844 0.82813
11 2.1 0.994 0.07849 0.87288 0.997 0.07849 0.87309
12 2.3 0.994 0.04700 0.93146 0.997 0.04691 0.93183
13 2.5 0.994 0.00696 1.01844 0.997 0.00668 1.01913
14 2.7 0.994 −0.03990 1.13430 0.997 −0.03937 1.13541
15 2.9 0.994 −0.41540 2.02530 0.997 −0.42099 2.03601
16 3.0 0.898 2.54350 0.47215 0.901 2.53226 0.47007
17 3.0 0.700 1.47932 0.07000 0.703 1.47161 0.06755
[email protected]
18 3.0 0.504 1.32144 0.05058 0.505 1.31710 0.04998
19 3.0 0.309 1.22781 0.02533 0.310 1.22472 0.02501
20 3.0 0.133 1.18297 0.00933 0.133 1.17720 0.00918
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 227

n = 60 n = 72
m z/a r/a σ τ r/a σ τ

1 0.1 0.988 2.00045 0.72366 0.999 2.00083 0.72395

2 0.3 0.988 0.87062 0.59176 0.999 0.87059 0.59152
3 0.5 0.988 0.61800 0.61709 0.999 0.61806 0.61704
4 0.7 0.988 0.45540 0.64433 0.999 0.45547 0.64428
5 0.9 0.988 0.35027 0.67258 0.999 0.35034 0.67256
6 1.1 0.988 0.27585 0.70071 0.999 0.27591 0.70068
7 1.3 0.988 0.22015 0.72913 0.999 0.22021 0.72910
8 1.5 0.988 0.17636 0.75880 0.999 0.17641 0.75878
9 1.7 0.988 0.14019 0.79114 0.999 0.14023 0.79113
10 1.9 0.988 0.10851 0.82817 0.999 0.10855 0.82816
11 2.1 0.988 0.07852 0.87320 0.999 0.07855 0.87318
12 2.3 0.988 0.04688 0.93200 0.999 0.04691 0.93203
13 2.5 0.988 0.00655 1.01946 0.999 0.00656 1.01947
14 2.7 0.988 −0.03898 1.13589 0.999 −0.03881 1.13591
15 2.9 0.988 −0.42433 2.04240 0.999 −0.42514 2.04359
16 3.0 0.903 2.52557 0.46857 0.903 2.52438 0.46812
17 3.0 0.704 1.46757 0.06629 0.704 1.46666 0.06604
18 3.0 0.406 1.31548 0.04966 0.506 1.31553 0.04960
19 3.0 0.311 1.22326 0.02487 0.311 1.22297 0.02484
20 3.0 0.133 1.17372 0.00908 0.133 1.17286 0.00906

Calculation by Shishov [126]

m z/a r/a σ τ

1 0.1 1.0 1.952 0.715

2 0.3 1.0 0.877 0.600
3 0.5 1.0 0.614 0.620
4 0.7 1.0 0.451 0.646
5 0.9 1.0 0.346 0.674
6 1.1 1.0 0.272 0.702
7 1.3 1.0 0.216 0.730
8 1.5 1.0 0.173 0.760
9 1.7 1.0 0.137 0.792
10 1.9 1.0 0.106 0.829
11 2.1 1.0 0.076 0.875
12 2.3 1.0 0.045 0.934
13 2.5 1.0 0.005 1.024
14 2.7 1.0 −0.044 1.149
15 2.9 1.0 −0.424 2.013
16 3.0 0.9 2.488 0.443
17 3.0 0.7 1.490 0.078
18 3.0 0.5 1.322 0.006
19 3.0 0.3 1.223 0.000
[email protected]
20 3.0 0.1 1.221 0.000
228 3 Computer Implementation of Boundary-Element Algorithms

Fig. 3.62 Calculation

scheme to the problem of
impression of a deepened
cone into an elastic half-space

The calculations for n = 28, 40, 60, and 72, due to the limitations of the computer
RAM size, were performed only in the axisymmetric approach and have shown a
stable increase of the approximate solution accuracy with the increase of the number
of the boundary elements (the correction regards the 4–5-th digit after the decimal
point, Table 3.5). The calculations performed by boundary-element method at n =
20 gave the result which appeared to be the closest to the numerical calculations of
Shishov. Additional calculations for this case of discretization when the cylindrical
punch radius was increased in such a way that the boundary-element polyhedron
surface corresponded to the contact surface for the initial cylinder and the colloca-
tion points on the bottom were chosen not in the gravity centres of the boundary
elements, but in the geometrical centres of the circular sectors, has led to a practi-
cally full coincidence of the numerical boundary-element values with the calcula-
tion data of Shishov (the difference was observed only for the 5th, only sometimes
for the 4th digit after the decimal point). Consequently, the results of the contact
problem solution for a deepened cylindrical punch using the proposed boundary-
element method, are reliable and the calculation accuracy can be increased due to
the increase of the number of boundary elements. To obtain the results with an accu-
racy, sufficient for practical purposes, the contact surface discretization degree can
be limited to 400–500 boundary elements.
Note that the successful approbation of the contact problem for a deepened cylin-
drical punch subsequently enabled the boundary-element approach to be applied
with better substantiation for the calculation of base deformations for rigid foun-
dation structures of vertical and inclined cylindrical bored piles as well as bored
piles with footings under spatial loading of general type (see Sects. Section 5.4 and
Section 5.5).
Impression of a cone into an elastic half-space. Consider a spatial contact prob-
lem for a deepened conical punch being impressed by an axial force P into an elastic
homogeneous half-space with the deformation modulus E and the Poisson ratio ν.
[email protected]
Consider the punch to be soldered with the half-space which has a conical cavity
of the depth h and radius a (the angle at the cone vertex α = 2arccot(h/a)). The
calculation scheme of the problem and the notations used are shown in Fig. 3.62.
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 229

(a) (b)

Fig. 3.63 Discretization of the contact surface of a deepened conical punch by flat triangular and
quadrangular boundary elements: (a) lateral view; (b) view from below

Since no analytical solution of the problem under consideration is available, we

study it numerically using the boundary-element method. The contact surface of
the cone and the half-space is meshed into flat boundary elements whose nodes are
formed at the intercepts of horizontal (equidistant) and vertical (passing through
the cone axis and forming equal angles between the neighbouring ones) planes
(Fig. 3.63). Such discretization of the cone lateral surface enables the contact prob-
lem to be solved in both axisymmetric and general spatial formulation for the same
boundary-element grid. Note that, similarly to the above contact problems for the
rigid sphere and cylinder, comparison of the results in both formulations at the num-
ber of boundary elements not higher than 500 gives the results, coinciding within
the machine accuracy. This is the evidence for the applied boundary-element algo-
rithm to be correct and reliable. On the other hand, since central loading of the
conical punch is of the main practical interest, this enables us hereinafter to solve
the contact problems in the axisymmetric formulation what increases practically
unlimitedly the number of boundary elements to obtain numerical solutions with a
required accuracy.
The results of the numerical solution of the contact problem for a conical punch
with a depth h = 2a(α/2 = 26.57◦ ), being impressed by a force P = Ea2 into
an elastic half-space with the Poisson ratio ν = 0.3, are presented in Tables 3.6
and 3.7and in Fig. 3.64. The cone generatrix was meshed into ten parts of equal
length (m= 10), in the centres of which the values of normal σ and tangential τ
contact stress were determined, normalized by the average contact stress σav = P/S
where S is the cone lateral surface area. The calculation data give the evidence for a
rather fast convergence of the numerical solution with the increase of the number of
[email protected]
the boundary elements n in the circular direction, i.e. with the increase of the degree
of the contact surface approximation by an ensemble of flat boundary elements.
Table 3.6 contains also contact stress extrapolation, obtained from the calculation
230 3 Computer Implementation of Boundary-Element Algorithms

Table 3.6 Normal σ and tangential τ contact stress in the axial section of a deepened conical
punch (h/a = 2.0, ν = 0.3, P = E · a2 )

n = 40 n = 60
m z/a r/a σ τ σ τ

1 0.0983 0.94795 1.89056 0.54453 1.89209 0.54555

2 0.2980 0.84836 0.71133 0.56688 0.71052 0.56578
3 0.4978 0.74880 0.60811 0.59814 0.60822 0.59758
4 0.6974 0.64927 0.54249 0.63204 0.54267 0.63131
5 0.8969 0.54982 0.52223 0.67833 0.52255 0.67730
6 1.0963 0.45046 0.53094 0.74526 0.53163 0.74389
7 1.2952 0.35129 0.56938 0.84678 0.57088 0.84517
8 1.4933 0.25255 0.65315 1.01607 0.65623 1.01557
9 1.6889 0.15508 0.83164 1.37025 0.83421 1.37094
10 1.8667 0.06646 1.23047 2.56431 1.22676 2.55185

n = 100 n = 160
m z/a r/a σ τ σ τ

1 0.0983 0.94795 1.89379 0.54561 1.89744 0.54428

2 0.2980 0.84836 0.79814 0.56466 0.70856 0.56345
3 0.4978 0.74880 0.60861 0.59665 0.60938 0.59558
4 0.6974 0.64927 0.54331 0.63017 0.54435 0.62917
5 0.8969 0.54982 0.52370 0.67604 0.52524 0.67517
6 1.0963 0.45046 0.53353 0.74259 0.53566 0.74212
7 1.2952 0.35129 0.57395 0.84438 0.57677 0.84465
8 1.4933 0.25255 0.66101 1.01712 0.66453 1.01923
9 1.6889 0.15508 0.83772 1.37386 0.83942 1.37689
10 1.8667 0.06646 1.22158 2.53778 1.21563 2.52637

data in the assumption of a linear dependence of the numerical solutions on n−1 . It

is seen that at n= 400 the relative error δ of determination of dimensionless contact
stress seldom exceeded 0.5% and relative punch displacement W/a has three cor-
rect digits after the decimal point (Table 3.7), what is by far sufficient for practical
calculations. In order to visualize the contact interaction character, normal and tan-
gential stress diagrams, built along the cone generatrix from the numerical solution
extrapolation data, are shown in Fig. 3.64.
Note that the solution of the above contact problem for the cone with the same
vertex angle had been performed earlier by Aleksandrov and Solov’ev [3] using a
contour integral method. Unfortunately, the quoted reference does not contain the
Poisson ratio value ν, for which the numerical results were obtained. Therefore,
we have performed additional calculations for various most widely used ν values.
The calculated displacement data are presented in Table3.8. The numerical results,
[email protected]
obtained by different methods, are seen to be close. It should be noted that applica-
tion of the contour integral method in [3] implied neglection of horizontal displace-
ments in the points of the punch contact with the base what, in turn, resulted in a
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 231

n = 400 Extrapolation δ (%)

m z/a r/a σ τ σ τ σ τ

1 0.0983 0.94795 1.91321 0.53968 1.92372 0.53661 0.55 0.57

2 0.2980 0.84836 0.70244 0.56152 0.69836 0.56023 0.58 0.23
3 0.4978 0.74880 0.61113 0.59408 0.61230 0.59308 0.19 0.17
4 0.6974 0.64927 0.54612 0.62806 0.54730 0.62732 0.22 0.12
5 0.8969 0.54982 0.52753 0.67447 0.52906 0.67400 0.29 0.07
6 1.0963 0.45046 0.53825 0.74195 0.53998 0.74184 0.32 0.02
7 1.2952 0.35129 0.57951 0.84524 0.58134 0.84563 0.31 0.05
8 1.4933 0.25255 0.66711 1.02167 0.66883 1.02330 0.26 0.16
9 1.6889 0.15508 0.83813 1.37988 0.83727 1.38187 0.10 0.14
10 1.8667 0.06646 1.20232 2.50906 1.19345 2.49752 0.74 0.46

hardly probable law of contact stress distribution over the lateral surface of the cone
(the normal and tangential stress diagrams appeared to be practically identical). A
removal of this limitation would undoubtedly enable even better compatibility of the
results of the solution of the contact problem under consideration for the deepened
cone using two different approaches.
Consider in more detail the results of the boundary-element solution of the con-
tact problem for obtuse-angle conical punches, deepened into an elastic half-space
whose Poisson ratio is taken as 0.3. The punch contact surface approximation in
this case was performed using 4000 boundary elements (50 along the cone gener-

[email protected]
Fig. 3.64 Profiles of dimensionless tangential τ and normal σ contact stress for a deepened conical
punch (h/a = 2.0; ν= 0, 3; P = Ea2 )
232 3 Computer Implementation of Boundary-Element Algorithms

Table 3.7 Comparative characteristics of settlements of a deepened cone, calculated for a

sequence of condensed boundary-element grids (ν = 0.3, P = E · a2 ,h/a = 2.0)
√
Number of boundary elements N = m × n S/π a a2 + h2 W/a

400 = 10 × 40 0.99836 0.34571

600 = 10 × 60 0.99927 0.34538
1000 = 10 × 100 0.99974 0.34516
1600 = 10 × 160 0.99990 0.34498
4000 = 10 × 400 0.99998 0.34446
Linear extrapolation 1.00002 0.34411

Table 3.8 Comparison of the calculated values of relative displacementsW/aof a deepened conical
punch depending on the Poisson ratio ν of the elastic half-space (h/a = 2.0, P = E  a2 )

ν Contour integral method Boundary-element method Relative error (%)

0.3 0.31955 0.34525 7.4

0.35 0.33184 0.34493 3.8
0.4 0.34413 0.34302 0.32

atrix and 80 along the circular coordinate), what for the case of the axisymmetric
formulation of the problem corresponds to a system of 101 linear algebraic equa-
tions with respect to the averaged 50 vertical pz and 50 radial τrz stress values along
the generatrix as well as the punch vertical displacement W in the direction of the
external force. The cone radius a and the external impressing force value P = Ea2
were fixed at the calculations. The h/a ratio, i.e. the relative deepening of the punch
was used as a variable parameter. The calculations were performed for the values
h/a= 0.5. 0.3, 0.2, 0.1, 0.03, and 0.01, what corresponds to a gradual increase of
the angle at the cone vertex. In the limit (at small h/a) it tends to the solution of
the problem of a flat circular punch, linked to a half-space, which has been solved
analytically at least several times [70, 138, 142]. In the most convenient visual form
(on the base of the numerical data) the solution of the problem in question is pre-
sented in [129] and is treated here as the case of a degenerate conical punch (h/a=
0). The above limiting transition is illustrated by Fig. 3.65where an example with
vertical contact stress diagrams along the cone generatrix is shown. Complementing
Fig. 3.65, one should note that the contact stress diagrams, built for h/a= 0.01, have
practically coincided with the corresponding diagrams for the circular punch with
flat bottom (dotted curve in Fig. 3.65).
The calculated settlement values for the series of the above considered obtuse-
angle conical punches are shown in Fig. 3.66. In the same figure the value W0 /a=
0.044234 for the flat-base punch settlement is indicated, determined at ν = 0.3 and
P = Ea2 using the formula [129]

[email protected](1 − 2ν)(1 + ν)
W= P.
2Ea · ln (2 − 4ν)
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 233

Fig. 3.65 Profiles of dimensionless vertical contact stress σ z = pz Ea2 /P along the cone generatrix
at different depths ; ν =0.3

Fig. 3.66 Dimensionless settlements of obtuse-angled cones at different relative depths; ν= 0.3;
P = Ea2

The relative settlement value for the conical punch with the deepening parameter
h/a= 0.03, calculated by the boundary-element method, is 0.44222 what is rather
close to the exact result for the flat-base punch. At 0 ≤ h/a≤ 0.01 the numerical
values of relative displacements of the cone within the machine accuracy were prac-
tically indistinguishable between each other and from the W0 /a value. As one can
see, the calculated data of the settlements of conical punches with increasing vertex
angle show the applicability of the proposed boundary-element algorithm and the
reliability of the results obtained using this approach. Besides, from the performed
[email protected]
numerical study an important feature of contact interaction of obtuse-angle conical
punches with an elastic half-space should be noted, namely the practically linear
law of their settlement decrease with the relative depth increase (Fig. 3.66).
234 3 Computer Implementation of Boundary-Element Algorithms

The performed formulation and boundary-element solution of the spatial con-

tact problem for the deepened conical punch is highly important in geotechnics for
the development of punch (field-based and laboratory-based) methods of determi-
nation of mechanical properties of soils. In practical studies of soils for engineering
purposes, in the recent years intense improvement of methods of determination of
physical and mechanical properties of soils by various punches and tips, namely
flat circular, conical (penetrating and immersing), spherical, torvane, cylindrical,
pile, etc. [35, 44, 75, 76, 89, 95, 127, 128, 136, 149, 150]. Application of vari-
ous tips increases the reliability of the obtained soil characteristics at their natural
conditions. Since a considerable interval of the soil functioning belongs to the com-
pression phase with a linear (or close to linear) relation between the load and the
settlement, one should note that while calculating foundation bases, solutions of
linear theory of elasticity can be used for the absolute majority of the soils. In turn,
the experimentally obtained plots of the indenter (tip) settlement versus the load (or
pressure) being transferred to it, form a basis for the determination of deformational
properties of the soil and for checking its strength properties.
The most important deformational characteristic of the soil is its deformation
modulus E. The correct evaluation of this parameter is quite important for the accu-
racy of determination of settlements of buildings and structures. Variation of another
deformational parameter, the Poisson ratio ν, is known to affect much less the stress-
strained state of soils. Usually it is sufficient to take the Poisson ratio from handbook
data, depending on the soil type [131]. Besides, while assigning the Poisson ratio it
is rather convenient to apply the known Wet formula [147].
At present the most elaborated both from theoretical and practical point of view
is soil examination method by a test static load using test circular punches (first
of all, in order to determine the deformation modulus) [35, 127, 131]. However, in
spite of the simplicity of the equipment, the flat punch test methods possess a num-
ber of essential disadvantages [44]. The main of them is low representability of the
obtained results for characterization of the soil properties from the deposit profile.
The test results for the base surface, which usually have a random character, are
extended to the whole depth of the soil mass. Then, the contact stress under the flat
punch is characterized by a considerable nonuniformity due to a sharp increase of
gradients with approaching the punch edges. This results in residual plastic deforma-
tions along the punch perimeter, determined by the soil strength properties and not
characterizing its compactability (elastic compressibility). Discrepancies between
theoretical solutions and experimental data are often revealed, especially in case
punches of small size being used [44, 128]. Besides, soil tests using flat circular
punches are rather labour-consuming and expensive, therefore wide application of
this method for examination of construction sites for residential building develop-
ment or, moreover, for engineering-and-geological survey, is still limited.
Based on a review of physical and mechanical characteristics, which, according
to the construction codes and regulations, should be used for foundation and base
[email protected]
calculation, and which are determined by different tips, Yeltsov has convincingly
shown the advantage of conical methods of soil tests [149]. These tests are suitable
both for the Earth and other planets [36], both for field-based and laboratory studies.
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 235

They are characterized by simplicity and portability, low price and high performance
speed, are easily automated and enable continuous mechanical characteristics across
the soil section to be obtained at minimal labour and power expense. Application of
conical tips in the measuring device is especially important for road construction,
which is characterized by long road extension and requires determination of soil
properties in a very great number of points [35].
In spite of the mentioned advantages, wide application of tips, first of all conical,
is retarded by insufficient development of experimental and theoretical fundamen-
tals of determination of mechanical properties of soils by static load. A number of
authors note that the existing theoretical solutions in many cases considerably differ
from the experimental data and are even contradictory [75, 136]. Then boundary-
element results of a systematic and rather strict solution of the problem on cone
impression into an elastically compressible medium are analyzed in detail. The
investigation performed has enabled data of fundamental character to be obtained,
which are required for working out a substantiated method of soil testing using
deepened conical punches.
Figure 3.67 presents the values of a dimensionless function

π WEa
= q(α,ν) (3.17)
P

[email protected]
Fig. 3.67 Calculated values of a dimensionless function q(α,ν) which determines the settlement
of cones, deepened into an elastic half-space
236 3 Computer Implementation of Boundary-Element Algorithms

obtained from the results of settlement calculations for cones, deepened into an
elastic half-space, with various vertex angles (α/2 = 15◦ , 18.43◦ , 30◦ , 45◦ , 60◦ ,
63.43◦ , 71.57◦ , 75◦ ) and at the Poisson ratio ν values from 0.2 to 0.5 with a step
0.05 what in a rather detail covers the real possible interval of variation of volume
compressibility of solid media (ν = 0.27 for macrofragmental soils, ν = 0.3 for
sands and clayey sands, ν = 0.35 for clayeys, ν = 0.42 for clays [127]). A detailed
analysis of the calculation data has shown that at fixed cone opening angle α the
values of q can be rather successfully approximated by a parabolic dependence
 
q(α,ν) = K(α) · 1 + A(α) · ν − B(α) · ν 2

its parameters as well as the high approximation accuracy being indicated in

Table 3.9. For the intermediate values of a with approximation, sufficient for practi-
cal purposes, the coefficients K, A, and B can be obtained from the data of Table 3.9
by linear interpolation. For the sake of convenient application of Eq. (3.18) Figure
3.68 contains auxiliary plots of K(α), A(α), B(α), obtained from square approxima-
tion. The corresponding formulae are given by

  2
K = K0 + K1 ha + K2 ha ,
  2
A = A0 + A1 ha + A2 ha , (3.19)
  2
B = B0 + B1 ha + B2 ha

where h/a = cot(α/2) is the relative deepening of the cone, Ki , Ai , Bi (i= 0, 1, 2)

are regression parameters, listed in Table 3.10. Note that for obtuse-angle conical
punches (h/a ≤ 1) the K(h/a), A(h/a), and B(h/a) plots have appeared practically
linear (Fig. 3.68). This enables one to use for the cones with the vertex angles
α ≥90◦ less complicated, but not less exact linear approximations instead of the
dependences given by Eq. (3.19):

Table 3.9 Parameters of the approximation dependence Eq. (3.18) for deepened cones with dif-
ferent vertex angles

h/a α/2, deg K A B Mean square deviation

0.26795 75 1.342 0.46319 1.40716 5.38 × 10−6

0.33333 71.57 1.32848 0.46983 1.38527 5.01 × 10−6
0.5 63.43 1.29322 0.48806 1.33147 4.24 × 10−6
0.57735 60 1.27657 0.49679 1.30787 3.94 × 10−6
1.0 45 1.18423 0.54349 1.19669 2.78 × 10−6
1.73205 30 1.03221 0.60650 1.06071 1.67 × 10−6
[email protected]
2.0 26.57 0.98241 0.62364 1.02334 1.42 × 10−6
3.0 18.43 0.82771 0.66912 0.91932 8.33 × 10−7
3.73205 15 0.74099 0.69074 0.86571 5.99 × 10−7
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 237

Table 3.10 The parameters of square, Eq. (3.19), and linear, Eq. (3.20), approximations for the
coefficients K, A, Bin Eq. (3.18)

i Ki Ai Bi K̃i Ãi B̃i

0 1.41259 0.42289 1.47791 1.40053 0.43332 1.47888

1 −0.25241 0.12901 −0.3072 −0.21584 0.11003 −0.28596
2 0.019314 −0.01589 0.03893
Mean square 7.28 2.71 55.6 0.1508 0.02713 9.57
deviation (s 105 )

Fig. 3.68 Coefficients of Eq. (3.18) depending on the cone relative depth, α =2arccot(h/a)


K = K̄0 + K̄1 ha ,

A = Ā0 + Ā1 ha , (3.20)

B = B̄0 + B̄1 ha

whose coefficients are also listed in Table 3.10.

Consider an example of calculation the cone settlement using the obtained depen-
dences. Let the settlement value for a deepened conical punch with the vertex angle
α = 100◦ and radius a = 5 cm at the half-space surface, be required to estimate.
The punch is deepened into a fine grained sand by a vertical force P= 15 kN. The
deformation parameters of the elastic base are taken as follows: E= 28 MPa, ν =
0.333.
First consider the relative punch depth h/a = cot(α/2) = cot(50◦ ) = 0.8391. Using
Eqs. (3.19) with the account of Table 3.10, one can find the coefficients
[email protected]
K = 1.2143, A = 0.52591, B = 1.24754
238 3 Computer Implementation of Boundary-Element Algorithms

and then the value q(100◦ , 0.333) = 1.25896. The sought value of the cone settle-
ment will be W1 = qP/πEa = 0.42937 cm. Since the angle at the cone vertex is
obtuse, the calculation can be also performed using a more simple equation (3.20),
what results in a value W2 = 0.43154 cm, what is slightly different from the one
obtained earlier. A direct calculation using the developed program employing the
boundary-element method, leads to a result W3 = 0.43935 cm which, as could be
expected, is close to the settlements obtained earlier, and the differences between
the three values are within the measurement accuracy.
The developed numerical-and-analytical approach to determine the relationship
between W and P for a deepened cone, similarly to the flat punch [54, 127], can be
used to determine the soil deformation modulus according to the formula

q(α,ν) · P
E= (3.21)
π a · W

where P and W are the variations of load and settlement, respectively, at static
tests within the soil linear displacement phase. This dependence for the determina-
tion of the deformation modulus is the consequence of Eq. (3.17), the values of P
and W being replaced with their increments, since the settlement at the initial load-
ing stages will always be related to deviations due to a loose contact of the punch
and the soil in the cavity, breaking down surface irregularities, etc.
The procedure of testing and processing of the results for static loading of a cone
using Eq. (3.21) is similar to that for flat-bottom punches and consists in application
of a central force to the conical punch with subsequent stepwise loading with a
step P whose value should depend on the soil grain composition, dampness and
density, consistency, and porosity coefficient. Contrary to the flat punch tests, when
the settlement should be determined in several points of the punch, for impression of
a cone it is sufficient to determine the axial displacement. Note that, similarly to the
case of the flat punch, for a conical punch the critical pressure (corresponding to the
proportionality limit) can be determined from a sharp kink of the experimental curve
W = W(P). Thus, application of static loading of a conical punch can be regarded as
improvement of the existing methods of determination of deformational parameters
and checking the strength properties of soils.
An important feature of static examination based on the theoretical solution of
Eq. (3.21) consists in the fact that impression of a cone, having been deepened into
the soil before, is performed. When disturbed soils are examined, the samples are
formed using liners, forming conical cavities after their removal [76]. In the natural
soil samples the cavity is made using a special device of helical drill type. Near
walls and pits, as well as in order to determine anisotropic properties of the soil, the
cone can be impressed both in horizontal and in vertical direction.
Equation (3.21) enables the impression of a conical punch, deepened from the
elastic base surface. Application of this formula for the cases of impression of the
[email protected]
cone on the bottom of a pit, a trench, or a well will introduce an error into the
deformation modulus value, depending on the conical punch location depth. The
evidence for this is the above considered solution of the contact problem for a
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 239

Fig. 3.69 Calculation

scheme of the contact
problem for a conical punch
inside an elastic half-space

deepened spherical inclusion as well as the solution of the problem for a flat cir-
cular punch, deepened into an elastic half-space [95]. Evidently, the settlement of
the cone inside the half-space will be smaller than the settlement of the cone on the
surface, since part of the transferred load will be accepted by the soil mass, located
above the punch level. This was confirmed by the calculations which we performed
for a conical punch with the vertex angle 60◦ at the relative depth H/a from 1 to
1000. The calculation scheme of the contact problem for the conical punch inside
an elastic half-space is shown in Fig. 3.69. As follows from the calculation data
(Table 3.11), the similarity function q0 = q(H/a)/q(0), taking into account the cone
depth, is characterized at first by a rather sharp decrease (below H/a≈ 50), fol-
lowed by a rather slow stabilization towards the limiting value at H/a→∞. Note
that the volume compressibility has practically no effect on these tendencies. The
dependence q0 (ν), obtained by approximation of the calculation data at high relative
depth of the cone (H/a= 1000), is shown in Fig. 3.70. The formula

b0 + b 1 ν + b 2 ν 2
W= P, (3.22)
π Ea
found by numerical-and-analytical method, where b0 = 0.43808, b1 = 0.51284,
b2 = −0.67609, mean square error 2.6877 × 10−5 , can be treated as a kind
of analogue of the Roben equation for the cone with the vertex angle α =
60◦ . In case of necessity similar dependences can be without major difficulties
also obtained for cones with different vertex angles and appear helpful for the-
oretical substantiation of issues concerning static probing by cone-shaped tips
[44, 136].
The data of Table 3.11 and Eq. (3.22) are, in principle, quite sufficient to deter-
mine the deformation modulus with the account of depth correction for the standard
cone with the vertex angle α = 60◦ . Note that such angle is recommended for con-
ical tips by European standard on probing. Meanwhile, the analytical expression of
[email protected]
the dependence of q0 on H/a can be helpful for the calculations. However, approx-
imation of slowly decreasing functions of q0 (H/a) type on a half-infinite interval is
known to be a rather difficult problem to solve. For example, we have managed to
240 3 Computer Implementation of Boundary-Element Algorithms

Table 3.11 Similarity function q0 (ν) values at different Poisson ratios and depths for a standard
cone (α = 60◦ )

z/H ν = 0.2 ν = 0.25 ν = 0.3 ν = 0.35 ν = 0.4 ν = 0.45 ν = 0.5

0 1.00000 1.00000 1.00000 1.00000 1.00000 1.00000 1.00000

1 0.81687 0.82181 0.82688 0.83197 0.83686 0.84116 0.84396
2 0.71404 0.72023 0.72629 0.73195 0.73666 0.73938 0.73729
3 0.66061 0.66715 0.67335 0.67885 0.68291 0.68412 0.67859
4 0.62916 0.63585 0.64209 0.64744 0.65109 0.65142 0.64400
5 0.60869 0.61547 0.62172 0.62699 0.63039 0.63018 0.62166
6 0.59437 0.60121 0.60748 0.61268 0.61592 0.61536 0.60614
7 0.58381 0.59069 0.59698 0.60214 0.60526 0.60448 0.59478
8 0.57571 0.58263 0.58893 0.59407 0.59711 0.59615 0.58612
9 0.56930 0.57626 0.58256 0.58769 0.59066 0.58958 0.57929
10 0.56411 0.57109 0.57741 0.58252 0.58545 0.58427 0.57379
15 0.54819 0.55526 0.56161 0.56669 0.56949 0.56804 0.55703
20 0.54005 0.54715 0.55352 0.55859 0.56134 0.55976 0.54852
30 0.53177 0.53892 0.54532 0.55038 0.55309 0.55139 0.53993
40 0.52758 0.53476 0.54117 0.54624 0.54892 0.54718 0.53561
50 0.52506 0.53225 0.53867 0.54374 0.54641 0.54463 0.53301
100 0.51997 0.52719 0.53364 0.53871 0.54135 0.53952 0.52779
200 0.51742 0.52466 0.53111 0.53617 0.53881 0.53696 0.52518
1000 0.51536 0.52262 0.52908 0.53415 0.53678 0.53489 0.52308

Fig. 3.70 Quadratic approximation of the function of similarity q0 (ν) for a standard cone
(α= 60◦ ) at H/a=1000

approximate the dependence q0 = q0 (H/a) at ν = 0.35, 1≤H/a≤200 rather accu-

rately (Fig. 3.71) only in logarithmic coordinates X = ln(1−H/a), Y = ln(1−q0 )
[email protected]
using a 9-th order polynomial approximation

Y = a0 + a1 X + a2 X 2 + a3 X 3 + a4 X 4 + a5 X 5 + a6 X 6 + a7 X 7 + a8 X 8 + a9 X 9 (3.23)
3.5 Test Examples of Numerical Modeling of Spatial Problems of Contact Interaction 241

Fig. 3.71 Approximation of the function of similarity q0 (H/a) for a standard cone (α= 60◦ ) in
logarithmic coordinates; ν=0.35

where a0 = −4.6179, a1 = 9.50151, a2 = −13.9894, a3 = 13.3122, a2 =

−8.08656, a5 = 3.12577, a6 = −0.761261, a7 = 0.11267, a8 = −0.00922487,
a9 = 0.000319464, mean square error is 0.017685.
Though the approximation of Eq. (3.23) provides a good accuracy of approxima-
tion (Fig. 3.71), its practical application is not always effective due to computational
difficulties. For the same reason hardly applicable is Shapery method [122], often
used for approximation of slowly decreasing functions, where the approximation is
performed by a set of a great number of exponents. Our calculation experience has
shown a rather good effectivity (at fixed ν values) of interpolation of the considered
dependences q0 = q0 (H/a) using cubic splines on the base of a numerical algorithm,
proposed in [45] and implemented in a Fortran routine SPLINE. This approach can
be especially convenient for numerical calculations with direct application of the
data of Table 3.11. Finally, we show an example of our successful, in our opinion,
approximation of the dependence q0 = q0 (H/a) (Table 3.11, ν = 0.35), obtained
based on a combination of two exponential splines [61]:
   
q(t) = q(0) 1 − ctα e−χ t h(t) − b 1 − e−β(t−ξ ) h(t − ξ ) ,

where h(t) is the Heaviside function, t = H/a, b = 1−q(1000) = 0.46585, c =

0.1901, α = 0.87511, β = 0.8689, χ = 0.15545, ξ = 5.82654. From Table 3.12 one
can see a rather small discrepancy between the calculation data and the values of
analytical expression of Eq. (3.24) (mean square error is 2.1489 × 10−3 ).
Finalizing the analysis of solution of the contact problem for a deepened cone,
[email protected]
note that application of the proposed boundary-element procedure of solving soatial
contact problems for an elastic half-space has enabled an alternative (along with
the application of flat punches) method of determination of the most important
242 3 Computer Implementation of Boundary-Element Algorithms

Table 3.12 Approximation of the similarity function q0 (ν = 0.35, α = 60◦ ) by two exponential
splines at various depths of a standard cone

z/H BEM calculations Spline approximation δ (%)

0 1.00000 1.00000 0.0

1 0.83197 0.83727 0.64
2 0.73195 0.74449 1.71
3 0.67885 0.68812 1.37
4 0.64744 0.65659 1.41
5 0.62699 0.64264 2.49
6 0.61268 0.63456 3.57
7 0.60214 0.60549 0.56
8 0.59407 0.58532 1.47
9 0.58769 0.57174 2.71
10 0.58252 0.56302 3.35
15 0.56669 0.55524 2.02
20 0.55859 0.56208 0.63
30 0.55038 0.56238 2.18
40 0.54624 0.55238 1.12
50 0.54374 0.54387 0.02
100 0.53871 0.53434 0.81
200 0.53617 0.53415 0.38
1000 0.53415 0.53415 0.0

deformational characteristic of soil, the deformation modulus E, from the results of

impression of deepened cones with different vertex angles [9, 10]. The obtained the-
oretical dependences are not much more complicated that the equations used in the
procedure of standard examination by a test static load of experimental punches with
a flat bottom [54, 95, 127, 136]; however, they have higher generality. In particular,
a possibility of theoretically substantiated experimental studies for identification of
deformational properties of materials based on the results of impression of different-
angled (both acute-, and obtuse-angled) cones is being open. In our opinion, it will
enable a fast and rather reliable determination of the deformation modulus of soils
with different consistency, porosity and genetic type. Finally, the corrected values
of the soil deformation modulus will result in higher reliability of determination
of the calculated values of foundation settlements at construction of building and
structures at various engineering-and-geological conditions.
The analysis of numerous examples, presented in this subsection, is a rather
convincing evidence for the broad opportunities being opened by applying the
boundary-element method to solve spatial contact problems. The test and model cal-
culations, which we have performed using the Rostwerk software, have been com-
pared with other results for the problems having analytical solutions or obtained
by alternative approximate methods. Besides, high efficiency and reliability of the
proposed boundary-element algorithms has been shown. These algorithms are char-
[email protected]
acterized by module structure, low computation time and computer RAM require-
ments. In comparison with other popular numerical methods (e.g., finite-difference
or finite-element methods which require discretization not only of the boundaries,
References 243

but also inside the calculation domains) the input data size has been considerably
reduced and their preparation procedure has been simplified. Vast methodological
calculations enable us to conclude on a good convergence of the numerical results
and a sufficiently high for practical purposes accuracy of the boundary-element
method even at piecewise constant approximation of the contact pressure field on
the contact surfaces of rigid punches (at different depth). The possibilities of the
boundary-element approach to be applied for spatial contact deformation studies
are definitely shown, in particular, to perform structural analysis of rigid founda-
tions from base deformations, to determine soil base deformation parameters, to
calculate rigid anchor bearing plates of various plate depth.

References
1. Abramov V M (1939) Study of a case of asymmetric pressure of a punch of round
cross-section on an elastic half-space. Dokl Akad Nauk SSSR 23(issue 8):759–763
(in Russian)
2. Aleinikov S M, Sedaev A A (1999) Dual grids and their application in the boundary-element
method. Comput Math Math Phys 39:228–241
3. Aleksandrov A Ia, Solov’ev Iu I (1964) The solution of the three-dimensional axisymmetric
problem in the theory of elasticity by means of line integrals. J Appl Math Mech 28:1106–
1112
4. Alekseyev V M, Lipson G A, Mitrenko Yu A (1980) Study of the bearing capacity of pyra-
midal piles on a vertical load. In: Soil mechanics, bases and foundations. Voronezh State
University, Voronezh, pp. 72–81 (in Russian)
5. Alekseyev V M, Lipson G A, Mitrenko Yu A (1984) Studies of bearing capacity of pyramidal
piles under an inclined force. In: Studies of rational structures of foundations. Voronezh State
University, Voronezh, pp. 17–24 (in Russian)
6. Alexandrov V M, Smetanin B I, Sobol B V (1993) Thin stress concentrators in elastic solids.
Nauka, Moscow (in Russian)
7. Aleynikov S M (1996) Ring-shaped foundation with an internal cut-off circle. In: Calcu-
lation and design of bases and foundations at complex geotechnical conditions. VGASA,
Voronezh, pp. 9–16 (in Russian)
8. Aleynikov S M (1997) Calculation of slotted foundations in spatial stress-strain state of soil
base. In: Proc 14 Intern Conf Soil Mechanics and Found Eng (6–12 Sept. 1997, Hamburg),
Balkema, Rotterdam, vol 1, pp. 629–632
9. Aleynikov S M (1998) Method for the determination of modulus of deformation. Patent
2145655, MPK6 E02D 1/00; G01N 3/42. No. 98117270/28 (Russia)
10. Aleynikov S M (1999) Deformation of deformation modulus from soil test data by inden-
tation of a conical punch. In: Geotechnics-99. Proc Int. Conf. PGASA, Penza, pp. 9–12 (in
Russian)
11. Aleynikov S M, Ikonin S V (1994) Design and calculation of ring-shaped foundation on
variable-thickness bases. Energ Stroit (issue 11):66–70 (in Russian)
12. Aleynikov S M, Ikonin S V (1995) Calculation of inclined foundations for high-voltage
power line metal supports from the base deformations. Energ Stroit (issue 3):63–66 (in
Russian)
[email protected]
13. Aleynikov S M, Ikonin S V (1995) Foundation for tower building on tapered base. Patent
2043462 RU, MKI6 E 02 D 27/42. No. 92010566/33 (Russia)
14. Aleynikov S M, Ikonin S V (1996) Control of foundation shape and loading parameters to
preserve uniform settlement. Commun Numer Methods Eng 12:745–754.
244 3 Computer Implementation of Boundary-Element Algorithms

15. Aleynikov S M, Ikonin S V (1996) Prevention of nonuniform settlement of foundations.

Build Res J 44 (issue 2):69–89
16. Aleynikov S M, Sedaev A A (1995) Algorithm of grid generation in the boundary-element
method for flat domains. Mat Modelirovanie 7(issue 7):81–93 (in Russian)
17. Aleynikov S M, Sedaev A A (1998) Implementation of dual grid technique in BEM analy-
sis of spatial contact problems. In: Numerical methods in continuum mechanics: Proc. 7th
Intern. Conf. (High Tatras, Slovak Republic, 6–9 October 1998), pp. 113–118
18. Anan’ev I V, Andronov Ya L, Shlafman Sh M (1988) Three-dimensional problem of interac-
tion between a rigid die and an inhomogeneous foundation. Mech Solids 23 (issue 3):76–79
19. Andronov Ya L (1985) Numerical solution of the spatial problem of the interaction of a rigid
punch with a protrusion in the footing, with a nonhomogeneous base. Rostov Institute of
Engineering, Rostov-na-Donu, VINITI No. 2631-85Dep (in Russian)
20. Antipov Yu A (1987) Exact solution of a problem of impression of a ring-shaped punch into
a half-space. Doklady AN UkrSSR Ser A (issue 7):29–33 (in Russian)
21. Arkhipova E S, Protsenko V S, Rvachev V L (1974) Indentation in an elastic half-space of a
plane, circular, inclined stamp. Int Appl Mech 10:567–573
22. Banerjee P K, Butterfield R (1981) Boundary element methods in engineering science.
McGraw-Hill, New York
23. Baranov V S, Romanov D A, Romanov K D, Sidorchuk V F (1973) Investigation of the
pressure distribution in the soil around piles with spherical enlarged bases. Soil Mech Found
Eng 10:96–99
24. Barbakadze V Sh, Murakami S (1989) Calculation and design of engineering structures in
deformable media. Stroyizdat, Moscow (in Russian)
25. Bartolomei A A (1994). Modern state and problems of pile foundation engineering. In: Proc.
4th Int. Conf. on Pile Foundation Engineering Problems. Pt 1. Improvement of methods of
calculation and technology of pile foundations. PGTU, Perm, pp. 38–43 (in Russian)
26. Bartolomei A A, Pilyagin A V (1988) Stress-strain state of the beds of pyramidal pile foun-
dations. Soil Mech Found Eng 25:136–140
27. Bartolomey A A, Ponomaryov A B, Yushkov B S (1994) Foundations of hollow conical
piles. In: Bases and foundations in geological conditions of the Urals. PPI, Perm, pp. 13–17
(in Russian)
28. Bayer F L (1963) Optimally scaled matrices. Numer Math 5:73–87
29. Borodachev N M, Borodacheva F N (1966) Impression of a ring-shaped punch into an elastic
half-space. Inzh Zh Mekh Tverd Tela (issue 4):158–161 (in Russian)
30. Borodachev N M, Tarikov G P (1972) Determination of the reaction pressure under a foun-
dation by the electrical modeling method. Soil Mech Found Eng 9:428–429
31. Borodachev N M, Tarikov G P (1974) Solution of three-dimensional contact problems in
elasticity theory by the method of electrical simulation. Mech Solids 9 (issue 3):84–87
32. Brebbia C A, Walker S (1980) Boundary element techniques in engineering. Newnes-
Butterworths, London
33. Brebbia C, Telles J, Wrobel L (1984) Boundary element techniques. Springer-Verlag, Berlin
34. Bykov V I (1995) Foundations for individual residential buildings. Soil Mech Found Eng
32:66–68
35. Cherkasov I I (1976) Mechanical properties of soils in road engineering. Transport, Moscow
(in Russian)
36. Cherkasov I I, Shvarev V V (1970) Basics of Moon soil studies. Nauka, Moscow (in Russian)
37. Egorov K E (1948) Deformation of the base of a round rigid foundation under an eccen-
tric load. In: Bases and foundations. Issues of soil mechanics. Tr NIIOSP 11: 119–138 (in
Russian)
[email protected]
38. Egorov K E (1958) On the issue of the base calculation under a foundation with a ring-
shaped footing. In: Bases and foundations. Soil Mech Tr NIIOSP 34:34–57 (in Russian)
39. Egorov K E (1963) Indentation of a punch with a flat ring-shaped bottom into a half-space.
Izv AN SSSR Mekh i Mashinostr (issue 5):187–196 (in Russian)
References 245

40. Egorov K E (1965) Calculation of bed for foundation with ring footing. In: Proc. 6th Intern.
Conf. Soil Mech. Found. Eng., vol. 2, pp. 41–45
41. Eshelby J D (1957) The determination of the elastic field of an ellipsoidal inclusion and
related problems. Proc R Soc Lond A 241:376–396
42. Eshelby J D (1959) The elastic field outside an ellipsoidal inclusion. Proc R Soc Lond A
252:561–569.
43. Faddeev D K, Faddeeva V N (1968) Problems of scaling for linear systems. In: Mod-
ern numerical methods: Proc Intern School on Numerical Methods (Kyiv). VC AN SSSR,
Moscow, issue 1, pp. 76–84 (in Russian)
44. Ferronskii V I (1969) Penetration and carotage methods of engineering and geological stud-
ies. Nedra, Moscow (in Russian)
45. Forsythe G E, Malcolm M A, Moler C B (1977) Computer methods for mathematical com-
putations. Prentice-Hall, Englewood Cliffs
46. Galanov B A (1981) Approximate solution of some problems of elastic contact between two
bodies. Mech Solids 16(issue 5):61–67
47. Galin L A (ed.) (1976) Development of theory of contact problems in the USSR, Nauka,
Moscow (in Russian)
48. Giannakoglou K C, Chaviaropoulos P, Papailiou K D (1994) A numerical structured grids
of desired properties, on complex 3-D surfaces. Proc. 2nd ECCOMAS Conf. Stuttgart (Sept.
5–8, 1994), pp. 701–710
49. Giannakoglou K C, Chaviaropoulos P, Papailiou K D (1996) Boundary-fitted parameteriza-
tion of unstructured grids on arbitrary surfaces. Adv Eng Software 27:41–49
50. Glotov N M, Luga A A, Silin K S et al. (1975) Pile foundation. Transport, Moscow (in
Russian)
51. Godunov S K, Antonov A G, Kirilyuk O P et al. (1988) Guaranteed accuracy of solving
linear equation systems in Euclid spaces. Nauka, Novosibirsk (in Russian)
52. Godunov S K, Zabrodin A V, Ivanov M Ya et al. (1976) Numerical solution of multidimen-
sional problems of gas dynamics. Nauka, Moscow (in Russian)
53. Goncharov B V, Yenikeyev A Kh, Kogan V L (1991) Slotted foundations in tamped pits. In:
Pile foundations. Stroyizdat, Moscow, pp. 87–90 (in Russian)
54. GOST 20276-85 (1985) Soils. Methods of field determination of deformability characteris-
tics. Standard Publishers, Moscow (in Russian)
55. Gotman A L (1987) Analysis of tapered piles under combined action of vertical, horizontal,
and flexural loads. Soil Mech Found Eng 24:7–12
56. Gubenko V S, Mossakovskii V I (1960) Pressure of an axially symmetric circular die on an
elastic half-space. J Appl Math Mech 24:477–486
57. Gubenko V S, Nakashidze G M, Pyatovolenko V G (1986) Exact solution of the problem of
a ring-shaped punch. Dokl AN UkrSSR Ser A (issue 3):40–44 (in Russian)
58. Gulyayev V I, Bazhenov V A, Gotsulyak Ye A et al. (1990) Calculation of complex-shaped
shell. Budivelnyk, Kyiv (in Russian)
59. Harary F (1969) Graph theory. Addison-Wesley, Reading
60. Ikramov Kh D (1988) Numerical methods for symmetrical linear systems. Nauka, Moscow
(in Russian)
61. Ilyin V P, Maltsev L Ye, Sokolov V G (1991) Calculation of engineering structures of vis-
coelastic materials. Stroyizdat, Leningrad (in Russian)
62. Instructions on design of foundations for machines subject to dynamic loads (1982) Gerse-
vanov NIIOSP, Stroyizdat, Moscow (in Russian)
63. Instructions on design of pyramidal pile foundations (1983) Glavn Upr Project Organiz,
Moscow (in Russian)
[email protected]
64. Ivanenko S A (1993) Adaptive grids and grids on surfaces. Zh Vychisl Matem Matem Fiz
33:1333–1355 (in Russian)
65. Ivanenko S A, Prokopov G P (1997) Methods of adaptive harmonic grid generation. Comput
Math Math Phys 37:627–645 (in Russian)
246 3 Computer Implementation of Boundary-Element Algorithms

66. Joe B (1986) Delaunay triangular meshes in convex polygons. SIAM J Sci Stat Comput 7:
514–539
67. Johnson K L (1985) Contact mechanics. Cambridge University Press, Cambridge
68. Kalinnikov A Ye, Yefremov S M, Vakhrushev F V (1985) Algorithm for automatic mesh of a
two-dimensional domain for solving contact problems by boundary-element method. Probl
Prochnosti (issue 2):106–108 (in Russian)
69. Kalitkin N N (1978) Numerical methods. Nauka, Moscow (in Russian)
70. Kapshivyi A N, Maslyuk G F (1967) The solution of the mixed axisymmetric problem from
the theory of elasticity for a half-space by the method of p-analytical functions. Int Appl
Mech 3 (issue 7):13–16
71. Kazarnovskii I I (1997) Experience of design of monolithic foundations for contact-line
supports. Transp Stroit (issue 7):21–23 (in Russian)
72. Khazin V I (1984) Foundation structures for rural buildings on short pyramidal piles. Budi-
velnyk, Kyiv (in Russian)
73. Khilobok V G, Yudin A V (1984) On some specific features of functioning of asymmetric
wedge-shaped piles. In: Study of rational design of foundations. Voronezh State University,
Voronezh, pp. 78–81 (in Russian)
74. Klein G K, Cherkasov I I (1985) Foundations of urban transport structures. Stroyizdat,
Moscow (in Russian)
75. Klemyatsionok P L (1987) Indirect methods of determination of parameters of soil proper-
ties. Stroyizdat, Leningrad (in Russian)
76. Klemyatsionok P L (1994) Obtaining and studies of theoretical balance dependences of the
penetration method of determination of soil deformation modulus. In: Reliability of bases
for transport structures. SPBGUPS, St. Petersburg, pp. 41–48 (in Russian)
77. Konovalov P A (1988) Soil bases and foundations of buildings under reconstruction. Stroy-
izdat, Moscow (in Russian)
78. Kornishin M S, Paymushin V N, Snigiryov V F (1989) Computational geometry in the prob-
lems of shell mechanics. Nauka, Moscow (in Russian)
79. Kosorukov I I (ed.) (1974) Pile works. Vysshaya Shkola, Moscow (in Russian)
80. Kryukov K P, Kurnosov A I, Novgorodtsev B P (1975) Construction and design of metal
and ferroconcrete power line supports. Energiya, Leningrad (in Russian)
81. Kupradze V D, Gegeliya T G, Basheleyshvili M O, Burchuladze T V (1979) Three-
dimensional problems of the mathematical theory of elasticity and thermoelasticity. North-
Holland, Amsterdam
82. Kuzma I, Nikitenko M I (1984) Experimental studies of slotted foundations for the supports
of high-voltage lines. Soil Mech Found Eng 25:176–180
83. Kuznetsov A Yu (1990) Algorithm of construction of two-dimensional triangulation.
preprint VC SO AN SSSR No. 909, Novosibirsk (in Russian)
84. Lapshin F K (1986) Baes and foundations in diploma project design. Saratov State Univer-
sity, Saratov (in Russian)
85. Levachev S N, Fedorovskiy V G, Kurillo S V, Kolesnikov Yu M (2002) Piles in hydrotech-
nical engineering. Balkema, Rotterdam
86. Liseykin V D (1991) Technology of construction of three-dimensional grids for aerome-
chanical problems. In: Vopr Atom Nauki Tekhn Ser Mat Modelirovaniye Fiz Protsessov
31:1670–1683 (in Russian)
87. Liseykin V D (1996) Review of methods for construction of structural adaptive grids. Zh
Vychisl Mat Mat Fiz 36: 3–41 (in Russian)
88. Luchinskiy I A, Maliy E Yu, Povolotskaya I R, Ostrovskiy B S (1992) Foundation. USSR
Authors’ Certificate 1758169, MKI5 E 02 D 27/04, 27/44. No. 4737430/33 (in Russian)
[email protected]
89. Lupan Yu T, Klemyatsionok P L (1988) Tests of soils by penetration and combined probing.
UMK VO UkrSSR, Kyiv (in Russian)
90. Lur’e A I (1964) Three-dimensional problems in the theory of elasticity. Interscience, New
York
References 247

91. Lur’e A I (1967) Elastostatic Roben problem for a triaxial ellipsoid. Inzh Zh Mekh Tverd
Tela (issue 1):80–83 (in Russian)
92. Lur’e A I (1970) Theory of elasticity. Nauka, Moscow (in Russian)
93. Molchanov I N (1987) Machine methods for solution of applied problems: algebra and
approximation of functions. Naukova dumka, Kyiv (in Russian)
94. Muki R (1960) Asymmetric problems of the theory of elasticity for a semi-infinite solid and
a thick plate. In: Sneddon I N and Hill R (eds.). Progress in solid mechanics. North-Holland,
Amsterdam, vol. 1, pp. 399–439
95. Ogranovich A B (1993) On the issue of determination of soil deformation modulus from
punch test data. Izv Vuzov Stroit (issue 10):129–131 (in Russian)REF>
96. Olesiak Z (1965) Annular punch on elastic semi-space. Arch Mech Stosow 17: 633–648
97. Ovsyannikov A S, Starikov V A (1989) Method of superposition of singular solutions in
axially symmetrical problems of theory of elasticity. Naukova dumka, Kyiv (in Russian)
98. Pavlov V V (1992) Slotted foundation of buildings. Stroyizdat, Krasnoyarsk (in Russian)
99. Pavlov V V, Katsov K P, Smorodinov M I (1992) Slit foundations of industrial and civic
buildings in the middle Urals. Soil Mech Found Eng 29:140–143
100. Perley Ye M (1969) Tubular ferroconcrete piles for industrial and civil engineering. Stroyiz-
dat, Leningrad (in Russian)
101. Piskunov V G, Prisyazhnyuk V K (1985) Calculation of nonhomogeneous plates on a non-
homogeneous half-space. Stroit Mekh Raschet Sooruzh (issue 1):25–28 (in Russian)
102. Plevkov V S, Polishchuk A I (1990) Size determination of off-centre loaded foundations of
various geometrical shape. TPI, Tomsk (in Russian)
103. Podilchuk Yu N (1979) Three-dimensional problems of theory of elasticity. Naukova
Dumka, Kyiv (in Russian)
104. Podilchuk Yu N (1984) Boundary problems of statics of elastic bodies. Naukova Dumka,
Kyiv (in Russian)
105. Popov G Ya (1982) Concentration of elastic stress near punches, cuts, fine inclusions and
supports. Nauka, Moscow (in Russian)
106. Preparata F P, Shamos M I (1985) Computational geometry: an introduction. Springer-
Verlag, New York
107. Rabinovich I M (1961) On the problem of calculation of statically undetfinable sys-
tems with unilateral constraints (proof of the uniqueness of the solution). In: Gvozdev
A A (ed.) Studies in theory of structures, Gosstroyizdat, Moscow, issue 10, pp. 163–169
(in Russian)
108. Rabinovich I M (1975) Problems of theory of static calculation of structures with unilateral
constraints. Stroyizdat, Moscow (in Russian)
109. Recommendations on the calculation of bases under foundations with an inclined bottom
(1983) Gersevanov NIIOSP, Moscow (in Russian)
110. Reznikov R A (1971) Solving problems of engineering mechanics using a computer. Stroi-
telstvo, Moscow (in Russian)
111. Rvachev V L, Protsenko V S (1970) On the structure of solution of a contact problem with
an inclined punch. Dokl AN UkRSSR Ser A 11:1023–1026 (in Russian)
112. Rvachev V L, Protsenko V S (1977) Contact problems of theory of elasticity for nonclassical
domains. Naukova Dumka, Kyiv (in Russian)
113. Rybin V S (1990) Design of foundations of buildings under reconstruction. Stroyizdat,
Moscow (in Russian)
114. Sabonnadière J-C, Coulomb J-L (1986) La méthode des éléments finis. Du modèle. . . a la
CAO. Hermes, Paris – Londres – Lausanne
[email protected]
115. Sakalo V I, Shkurin A A (1985) Universal program for triangulation of a two-dimensional
domains of an arbitrary shape with grid condensations. Probl Prochn (issue 1):106–108 (in
Russian)
116. Samarskiy A A, Gulin A V (1989) Numerical methods. Nauka, Moscow (in Russian)
248 3 Computer Implementation of Boundary-Element Algorithms

117. Savchenko F M, Oding B S (1988) Economically efficient foundations under strutted

structures of agricultural buildings. In: Pile foundation studies. Voronezh State University,
Voronezh, pp. 162–167 (in Russian)
118. Savinov O A (1979) Modern design of foundations under machines and their calculation.
Stroyizdat, Leningrad (in Russian)
119. Sazhin V S, Shishkin V A, Zhanalinov B N (1991) Design and technology of erection of
foundations on heaving soils. Gylym, Almaty (in Russian)
120. Segerlind L J (1976) Applied finite element analysis. J. Wiley & Sons, New York
121. Selvadurai A P S (1994) On the problem of a detached anchor plate embedded in a crack.
Int J Solids Struct 31:1279–1290
122. Seo T, Mura T (1979) The elastic fields in half space due to ellipsoidal inclusions with
uniform dilatational eigen-strains. Trans ASME J Appl Mech 46:568–572
123. Shabrov N N (1983) Finite-element method in calculation of details of heat engines.
Mashinostroyenie, Leningrad (in Russian)
124. Shapery R A (1962) Approximate methods of transform inversion for viscoelastic
stress analysis. In: Proc. 4th US Nat. Congr. Appl. Mech. ASME, New York, vol. 2,
pp. 1075–1085
125. Shibuya T, Koizumi T, Takakuda K, Takagi T (1982) Partial contact problem of an elastic
half-space pressed aslant by a circular rigid punch. Bull JSME 25:1366–1372
126. Shishov O V (1971) Contact problem for axisymmetrical deepened punches. In: Resistance
of materials and theory of structures. Budivelnyk, Kyiv 13:60–66 (in Russian)
127. Shvets V B, Lushnikov V V, Shvets N S (1981) Determination of engineering properties of
soils: handbook. Budivelnyk, Kyiv (in Russian)
128. Sidorov N N, Sipidin V P (1972) Modern methods of determination of characteristics of
mechanical properties of soils. Stroyizdat, Leningrad (in Russian)
129. Solovyov L Yu, Solovyov Yu I (1995) Distribution of stress at the action of a circular punch
on an elastic half-space. Izv Vuzov Stroit (issue 9):22–26 (in Russian)
130. Sorochan E A (1986) Foundations of industrial buildings. Stroyizdat, Moscow (in Russian)
131. Sorochan E A, Trofimenkov Yu G (ed.) (1985) Bases, foundations and underground struc-
tures: designer’s handbook. Stroyizdat, Moscow (in Russian)
132. Sorochan E A, Piven V G, Rybnikov A M (1991) Monolithic foundations with an effective
lateral surface. Soil Mech Found Eng 28:99–102
133. Spekreijse S P, Boerstoel J W, Kuyvenhoven J L, Van der Marel M J (1992) Surface grid
generation for multi-block structured grids. Comput Fluid Dynam 2:937–944.
134. Studies of interaction of wedge-slotted foundations (double wedge) with a nonhomoge-
neous loess base (1989) Research report. Rostov Institute of Engineering, Rostov-na-Donu
(in Russian)
135. Telezhnikov V Ye (1956) Foundations under the equipment of small and medium power
plants. RSFSR Ministry of Public Service Publishers, Moscow (in Russian)
136. Trofimenkov Yu G, Vorobkov L N (1981) Field methods for the studies of constructive
properties of soils. Stroyizdat, Moscow (in Russian)
137. Tsybenko A S, Lavrikov S A (1984) Direct symmetrical formulation of the method of bound-
ary integral equations. Probl Prochn (issue 11):48–50(in Russian)
138. Ufliand Ya S (1977) The dual equation method in the problems of mathematical physics.
Nauka, Leningrad, 219pp. (in Russian)
139. Ugodchikov A G, Khutoryanskiy N M (1986) Boundary-element method in mechanics of
deformable solids. Kazan State University, Kazan (in Russian)
140. Ukhov S B, Semenov V V, Znamenskii V V et al. (1994) Soil mechanics, bases and founda-
tions. Association of Engineering Institutions, Moscow (in Russian)
[email protected]
141. Valov G M (1968) Infinite elastic layer and half-space under the action of a ring-shaped die.
J Appl Math Mech 32:917–930
142. Valov G M (1989) Method of solving the main mixed problem of theory of elasticity for a
half-space with circular line of separation of boundary conditions. In: Applied problems of
References 249

strength and elasticity. Methods of solution. Gorkiy Technical University, Gorkiy, pp. 77–82
(in Russian)
143. Vershinin V P, Kovalev I P, Chelnokov Ye L (1978) Bored pile foundations with support
widenings. Stroyizdat, Leningrad (in Russian)
144. Vladimirov V S (1976) Generalized functions in mathematical physics. Nauka, Moscow (in
Russian)
145. Warsi Z U A (1986) Numerical grid generation in arbitrary surfaces through a second-order
differential-geometric model. J Comput Phys 64:82–96
146. Werno M (1994) Statecznosc platformy Petrobaltic. In: Geotechnika w inzynierii wodnej i
ladowej. Gdansk, pp. 35–44
147. Wet J (1966) Application of energy hypothesis in soil mechanics. In: Soil mechanics and
foundation engineering. Proc 5th Intern Conf. Stroyizdat, Moscow, pp. 105–114 (in Russian)
148. Yakupov N M, Serazutdinov M N (1993) Calculation of elastic thin-walled structures of
complex geometry. KNC, Kazan (in Russian)
149. Yeltsov Yu A (1991) Studies of soil with indentors. Izhevsk mechanical Institute, Izhevsk
(in Russian)
150. Yeltsov Yu A (1993) Mechanics of effect of indentors on soil. Udmurt University, Izhevsk
(in Russian)
151. Yu H Y, Sanday S C (1990) Axisymmetric inclusion in a half space. Trans ASME J Appl
Mech 57:74–77
152. Zabylin M I, Linovskii S V, Nuzhdin L V (1991) Design of pile foundations under machines.
NISI, Novosibirsk (in Russian)
153. Zhukov N V, Dondysh A M, Pogosyan Z G et al. (1972) Short piles in rural engineering.
ONTI CNIIEPSelstroy, Moscow (in Russian)
154. Zienkiewicz O C, Kelly D W, Bettess P (1977) The coupling of the finite element method
and boundary solution procedures. Int J Num Method Eng 11:355–375

[email protected]
[email protected]
Chapter 4
Contact Interaction of Shallow Foundations
with Nonhomogeneous Bases

Abstract In the fourth chapter the results of the boundary-element solutions of spa-
tial contact problems for complex-shaped punches, located on the surfaces of elas-
tic nonclassical bases, are analyzed. The problems under consideration correspond
to the modeling of contact interaction of shallow foundations with elastic nonho-
mogeneous bases. Contact pressure fields under punches of various shape under
an eccentric load (a contact problem on a strongly inclined punch) are obtained.
The influence of non-uniform (over the area) compressibility as well as depth-
dependent nonhomogeneity of the base deformational properties on the formation
and development of detachment zones, settlements and slopes of punches with the
increase of the absolute values of overturning moments is shown. An algorithm to
calculate the boundaries of the section core for rigid complex-shaped foundation
plates from the stress values is described. Some optimization problems are solved
for load and shape parameter control in order to provide uniform settlement of rigid
foundation. As an example for the application of the developed boundary element
method, a contact problem is solved and the elastic base stress-strained state is deter-
mined for a rigid strip foundation of variable width. In the same chapter a spatial
contact model of the base is built taking into account nonlinear elastic soil proper-
ties. A procedure for the model parameter characterization based on the direct punch
test data is considered. Finally, the chapter contains the studies of contact problems
of bending of orthotropic plates situated on elastic nonclassical bases, performed by
BEM combined with finite difference method.

Foundations with small ratio of the depth to the width (the smaller side) of the
bottom are called shallow foundations. A characteristic feature of shallow founda-
tions is load transfer to the soil mostly through their bottom [258, 269]. Therefore,
at the calculation of shallow foundations, their depth is neglected and the calcu-
lation scheme for a foundation, located on the base surface, is used. Besides, in
order to simplify the calculations for the most of shallow foundations, the latter are
considered not as a rigid body (punch) on a natural base whose properties are deter-
mined by the chosen mechanical model of the soil. If the knowledge of the average
[email protected]
settlement of the structure is sufficient and the calculation of the foundation strength
is not required, then the distribution of reactive pressures under its bottom has no

S.M. Aleynikov, Spatial Contact Problems in Geotechnics, Foundations of 251

Engineering Mechanics, DOI 10.1007/b11479_4,  C Springer-Verlag Berlin Heidelberg 2011
252 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

special importance and load transfer to the soil can be considered as a distributed
load, i.e. using the calculation scheme for a flexible punch. If an economical design
solution for foundation structures of steel and reinforce concrete is required, the
issue of reactive pressure distribution becomes very important since the determina-
tion of bending moments and shearing forces in the foundation structures depends
on its solution. Various models of soil bases being used as well as a representation of
the foundation as a rigid punch enable the slopes of the structures to be determined
and, the forces in the foundation structures to be assigned more reasonably, though
with a certain allowance, after the contact problem of the soil mechanics having
been solved. Note that the foundation of a structure can be treated as a rigid body
in case the main condition being fulfilled: the foundation deformations should not
make a noticeable effect on the base deformations.
Shallow foundations on a natural base are the most widely spread in our coun-
try and constitute about 75% of the gross volume of the foundation engineering.
For their construction annually about 40 million m3 of concrete; 11.9 million tons
cement, 1.69 million tons metal are spent. The costs of their construction in 1992
was estimated as about 4.4 billion roubles [197, 257]. Hence, the rational solution
for a foundation essentially determines the cost of its construction and, therefore,
the reduction of materials consumption for shallow foundations is an important eco-
nomic problem.
Design of shallow foundations (free-standing, plate foundations, etc.) is a rather
complicated and labour-consuming problem, especially under several combinations
of loads. The main stages in this case are the choice of the foundation depth, deter-
mination of its bottom dimensions, determination of pressures over the bottom, cal-
culation of the foundation settlements and their nonuniformities, calculation of the
foundation strength etc. The calculation of off-centre loaded shallow foundations
(taking into account their rigidity and using modified base models) is rather com-
plicated, especially in the case of a complicated foundation shape. At present such
calculations are performed in a rather approximate way, the solutions being often
ambiguous what not always provides the choice of economical solutions. A particu-
lar uncertainty is induced by an assumption of a linear distribution of contact forces
over the foundation bottom [124, 162, 197, 226, 249]. Many important features of
engineering-and-geological conditions (in particular, non-uniform compressibility
of the base) that are rather essential for large-size foundations, are not taken (or
practically not taken) into account. In this chapter we apply the algorithm of calcu-
lation of settlements, slopes and contact pressures over the bottom of centrally and
off-centre loaded separate shallow foundations of various geometrical shape, con-
sidered in Sect. 2.5. The developed method is shown to enable not only the contact
stress distribution features, displacements and slopes for foundations, loaded by a
spatial force system of a general type to be studied in detail (Sects. 4.2, 4.4 and
4.6, Appendix B), but also important practical problems related to the determina-
tion of section kernels of complex-shaped foundations to be solved (Sect. 4.5). The
[email protected]
section kernel boundaries are a convenient geometrical characteristic of the foun-
dation stability under various technical and geological conditions. The proposed
boundary-element approach combined with intentionally developed procedures of
4.1 Spatial Contact Problems for Rigid Flat-Bottom Punches 253

Fig. 4.1 Calculation scheme

for a rectangular punch on a
variable-thickness elastic
layer

shape and load parameter control provides design solutions on uniform settlement
of shallow foundations at the conditions of the spatial stress-strained state of the
base (Sect. 4.3). Using the developed semi-empirical models of contact deforma-
tion, the elaborated boundary-element algorithm enables the deformations of natu-
ral and artificial bases of structures and buildings to be predicted with the account
of nonlinear properties of soils (Sect. 4.6). Combination of boundary-element and
finite-difference methods has enabled contact problems of bending of orthotropic
foundation plates of finite rigidity, resting on soil bases, to be effectively solved in
the framework of elastic non-classical models (Sect. 4.7). At various type of load-
ing and boundary conditions at the plate contours the effect of the plate material
orthotropy degree on the stress-strained state of the plates (on the deflections, shear-
ing forces, bending moments and torques) is determined. Elastic layers of constant
or variable thickness (Fig. 4.1) as well as an elastic half-space with deformation
modulus, increasing with depth, are used as models for the soil bases. The latter
case corresponds to the interaction of foundation structures with real soil bases,
nonhomogeneous both over the area and in depth [86]. Due to a convenient module
structure of the software developed for the elaborated numerical algorithm, the use
of influence functions for any of the known models of linearly deformable base of
non-classical type does not result in any substantial difficulties and can lead only to
the total computation time increase.
Before considering special issues, the chapter begins with a review of main stud-
ies devoted to the solution of spatial contact problems for rigid flat-bottom punches
and directly related to the foundation construction problems (Sect. 4.1).

4.1 Spatial Contact Problems for Rigid Flat-Bottom Punches

The problem of contact pressure distribution over the foundation bottom of various
shape is of high practical importance, especially when their dimensions are being
[email protected]
assigned and the foundation plate reinforcement is determined.
At present, two-dimensional (planar and axisymmetric) problems are solved
rather successfully, contrary to the three-dimensional solutions which are very rare
254 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

[8, 92, 98, 99, 117, 128, 159, 160, 203, 205, 225, 278]. Meanwhile, it is well
known [194] that for a number of geotechnical problems the approximations of flat
deformation and axially deformed stress-strained state will not correspond to the
real interaction pattern in the “foundation – soil base” system, or they will lead to
unreasonably high allowances. First of all, among such problems there are calcula-
tions of foundation bases of rectangular or more complex (e.g. polygonal) shape of
solid plate type under buildings and structures, roads as well as foundations under
machines and equipment etc. [121, 123, 125, 197, 226, 257]. Therefore, the develop-
ment of the corresponding spatial solutions is of considerable interest to substantiate
economical technical solutions.
Experimental studies. An all-round experimental study of the mechanism of
interaction of foundation models and soil bases is a rather complicated problem
requiring the improvement of the experimental technique, its automation, process-
ing of the measured data [178]. As a result, at the present stage of development
of soil mechanics and foundation engineering experimental studies strongly lag
behind theoretical investigations. Simultaneously, along with the difficulties in using
new experimental techniques, carrying out expensive and labour-consuming experi-
ments, the development of the experimental studies is hindered by a practically total
absence of strict theoretical solutions of essentially spatial problems of mechan-
ics of continuous media. The presence of such solutions enables the measurement
accuracy to be estimated and the experiment to be planned more substantially, the
measurement gauges to be located, the correspondence of the boundary conditions
for the experiment and theory to be determined, etc.
The available experimental data on the spatial distribution of contact pressures,
obtained from field and laboratory studies, often results to disputable and even con-
tradictory conclusions [72, 83, 96, 144–147, 177, 178, 264]. In most cases the exper-
imental studies enable the data to be obtained only for the specific conditions which
can be extended to other cases only with great precaution. It is mostly explained by
the use of different measurement techniques. In order to compare the data of differ-
ent experiments, the size and shape of the foundation models should be determined
from the similarity conditions and the experimental data processing should be per-
formed in the same way. At the present stage of development of the experimental
techniques one should consider the most reliable the data obtained at symmetrical
loading under a rigid circular or rectangular punch, for which rather detailed studies
have been performed concerning the contact stress profile transformation with load
increase, the distribution of stresses and their invariants in the soil mass, the base
deformation along the punch axis in horizontal and vertical directions, compressible
soil mass depth, corresponding to the depth of the sand base density variation range.
Besides the field and laboratory experimental studies, a certain progress in the
description of the processes of spatial contact interaction of foundations with soil
bases can be achieved by the method of mathematical analogies. The contact prob-
lems possess a certain similarity with the problems of electrostatics. From this point
[email protected]
of view one should mention the studies of Borodachev and Tarikov [45, 46] where
the results of the experimental solution of spatial contact problems of theory of
4.1 Spatial Contact Problems for Rigid Flat-Bottom Punches 255

elasticity for complex-shaped punches are given using electrical modelling. In such
approach the problem of determination of the reactive pressure under the punches in
accordance with the similarity of partial differential equations is reduced to a prob-
lem of determination of charge density on an electrically conductive plate with a
shape of the punch base. After an additional mathematical processing of the exper-
imental data by least-square method, the distribution of the reactive pressure under
the punches at their translational motion due to the vertical force, applied at the
punch gravity centre, is found. In all cases under consideration the contact area was
supposed to coincide with the flat area of the punch base. The effect of cutout in
the bases of punches of polygonal shape is studied. The electric modelling method
has an advantage over other (direct) experimental methods due to its high stability;
meanwhile the measurement accuracy is sufficient for the practical engineering pur-
poses. Nevertheless, the electric modelling of contact interaction can be realized yet
only in the framework o the elastic half-space model as well as at the conditions of
direct contact of the punch with the base. Although practical problems of determi-
nation of reactive pressure under punches (foundations) of rather complex shapes
can be solved, the experiments carried out by this method are still expensive and
labour-consuming.
In spite of the great scope of the experimental studies having been performed, a
number of issues still remain unsolved, regarding experimental procedures for the
studies of essentially spatial processes of deformation of soil masses under founda-
tions. In particular, this concerns experimental studies devoted to the contact pres-
sure distribution over the punch foundation bottom under an off-centre load. The
scope of these studies having been performed so far, is still insufficient [145, 178,
197]. This is one of the reasons for the fact that in practical design (according to the
valid regulations [124, 162, 197, 226, 258]) under an off-centre load the shape of
the contact pressure profiles is simplified and for separate foundations is assumed
to be linear. Depending on the load application eccentricity, rectangular, triangular,
or trapezoidal type profiles are used. Note that in a number of situations the reac-
tive pressure distribution scheme can be used only as a tentative one, mostly for the
rectangular-shape foundations.
Exact analytical solutions. The analysis of numerous literature sources has
shown that the contact pressure distribution over a rigid foundation bottom depends
on the bottom shape, load application eccentricity value, nonuniformity degree and
many other factors. In practical design preference is given only to such calcula-
tion models of soil bases, the use of which enable the theoretical law of contact
pressure distribution to be obtained. The main calculation models of soil bases, for
which the greatest number of theoretical solutions have been obtained and which to
a sufficient extent provide an adequate description of spatial functioning of shallow
foundations, are an elastic half-space and a finite-thickness layer.
The known exact analytical solutions of spatial contact problems for a linearly
deformed half-space have been built mostly for circular, elliptical and ring-shaped
[email protected]
punches. Note that the necessity of the contact problem solution for the circular and
ring-shaped punches is caused by the wide application of circular and ring-shaped
256 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

foundations under tower-type structures (funnels, cooling towers, water towers, TV

towers, etc.) in engineering.
The most effective solution has been obtained for the problem of a circular punch,
being impressed without friction into an elastic half-space. Boussinesq was the first
to solve the problem of a rigid circular punch impression by a central force [55]. He
obtained a formula to determine the pressure under the punch:


P
p (r) = √ , r= x2 + y2 < a
2π a a2 − r2
where a is the punch radius, r is the distance from the punch centre to the observation
point, P is a force, acting on the punch. This formula was the first to show the gen-
eral law of the contact pressure being unlimited near a fixed smooth boundary of a
contact domain. Boussinesq has not determined the stress-strained state in the inter-
nal points of the elastic base. Such study for an axisymmetric case was performed by
Dinnik [71]. In particular, he has shown a fact, very important for the strength cal-
culations, that the maximal tangential stress is reached under the punch at a distance
from the half-space surface, approximately equal to half radius of the contact area.
Much later Egorov and Shelest [81] have given the full pattern of the stress-strained
state in an elastic half-space under a centrally loaded round foundation in the form
of isolines of vertical and horizontal displacements as well as tangential and normal
stresses. Comparison with the corresponding results for a flexible foundation has
shown an essential effect of the foundation rigidity and the Poisson ratio of the soil
upon the stress-strained state of the base, mostly near the half-space surface. The
obtained formulae for the displacement and stress components were concluded to
be very useful for the design of bases under tower-type structures (cooling towers,
funnels, reservoirs, etc.).
Leonov [153] and Mossakovskii [170, 171] have solved the problem of pressure
of a circular punch on an elastic half-space in the most general formulation already
in 1953, the expressions for the contact pressure having been obtained in a closed
form – as integral-differential operators from given functions.
This problem has been further considered mainly with the aim to obtain the
results in a more convenient form, more suitable for computations, as well as to
evaluate the efficiency of methods being used to solve contact problems. Detailed
reviews of studies regarding the spatial contact problem for round punches and for
different equations of the punch bottom were published in a book [99] and a topical
review [174]. The analysis of the studies of the problem of a round punch being
impressed without friction into an elastic half-space, has shown that at present this
problem is fully studied. For a punch with a polynomial base the problem solution is
expressed in elementary functions. Such solution is a convenient object to estimate
the efficiency of approximate methods, aimed at the investigation of more compli-
cated contact problems with non-canonical contact domains and for non-classical
[email protected]
bases. Most of the contact problem solutions quoted in [99, 174] are applicable for
flat-bottom punches, which to the greatest extent correspond to shallow foundation
structures design.
4.1 Spatial Contact Problems for Rigid Flat-Bottom Punches 257

In the overwhelming majority of the investigated contact problems the lack

of friction forces between the punch and the half-space is assumed. In a rather
recent paper [256], based on the spatial problems of theory of elasticity solved
using analytical functions of a complex variable [8], the solution of the contact
problem for a rigid punch, fully coherent with an elastic half-space, is analyzed.
The studies of the stressed state inside the half-space and under the bottom of
the punch subject to a pressing force and a displacive force as well as an over-
turning moment, have been carried out. The effect of contact stress oscillation
in the vicinity of the punch edge (similar to the Abramov effect [1] in the pla-
nar problem), typical for the contact problems of linear theory of elasticity with
the account of the friction forces has been studied. Due to the complicated char-
acter of the formulae in the obtained solution, they were treated using numerical
integration.
Lur’e [159, 160] has performed analytical studies of asymmetrical pressure of
rigid planar punches of circular and elliptical cross-section, completely adjacent to
the elastic half-space.
Relatively recently exact solutions of the contact problem of a centrally loaded
ring-shaped punch (or action of a hollow circular cylinder on a half-space) were
obtained [25, 113]. For this problem the exact solution in [113] is expressed in
terms of a sequence of powers of infinite-measure matrix, while in [25] the solution
is found as a dual series, for its coefficients explicit formulae having been obtained.
The results of both works [25, 113] have mostly a pure theoretical significance, since
their application requires rather complicated computations.
The book by Rvachev and Protsenko [225] describes exact solutions of spatial
contact problems of an infinite stripe punch and a system of infinite stripe punches,
located on an elastic half-space, obtained as improper integrals of series over Math-
ieu functions. It is shown that the problem of a punch, whose cross-section is
restricted by two osculant circles, can be reduced by Kelvin transformation to the
problem of a stripe-shaped punch. The same book also quotes the exact solution for
a spatial problem of a simultaneous contact of a system of stripe-shaped punches
with a combined elastic Shtayerman base [245], being a combination of the elastic
half-space model with the Winkler model. In spite of the obtained formulae for the
contact pressure function being rather complicated, the application of the combined
base model enables the character of the contact pressure distribution to be preserved
similarly to the case of the elastic half-space; however, in this case their values at
the punch edges are not restricted.
Thus, the known exact solutions of the contact problems are obtained for par-
ticular, relatively simple domains and classical base models under full contact. The
methods being used for this purpose settle upon a classical mathematical appara-
tus – theory of potential, Fourier method, integral transformations, integral equa-
tions, solution of infinite systems of algebraic equation and some others. Solution
of almost each type of the problems requires a high qualification of the researcher,
[email protected]
enabling considerable mathematical difficulties to be overcome. As a result, for
practical purposes engineers use only few analytical solutions, the most simple from
the computational point of view [92, 117, 205, 258].
258 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

Approximate analytical solutions. Due to the difficulties of solving spatial con-

tact (mixed) problems of theory of elasticity, the possibilities of construction of their
exact analytical solutions without introducing simplifying assumptions regarding
the geometrical (circular and elliptical line of separation of boundary conditions)
and mechanical (isotropic and homogeneous medium) parameters are rather lim-
ited. Therefore, most of the studies of spatial contact problems now are related to
the construction of various approximate solutions.
A mixed boundary (contact) problem of theory of elasticity can be, using integral
transformations, reduced to paired or triple functional (integral or ordinary) equa-
tions [99]. As a result, finally such equations are transformed into a second-order
Fredholm integral equation that can be effectively solved by one of the approximate
methods.
The method of paired integral equations for spatial contact problems of theory
of elasticity was first applied by Abramov [2], where a solution of the problem on
impression of a round punch with a flat base into an elastic half-space under an
eccentrically applied vertical force was obtained. The contact pressures were deter-
mined provided the whole punch bottom being in contact with the elastic medium.
Later the same problem for a finite-thickness layer was considered by Egorov [74].
The solution obtained there is still being used in Russian handbooks and regula-
tions [125, 258]. The handbook [258] quotes a table to determine slopes of round
foundation plates depending on the eccentricity of the vertical component of the
resultant load and the relative depth of the base mass being compressed. Graphs
for the determination of settlements and slopes of a round absolutely rigid founda-
tion on a linearly deformed half-space and a finite-thickness layer under off-centre
vertical load are presented in [125]. An approximate formula for the calculation of
horizontal displacements in an elastic layer under a centrally loaded round-shaped
foundation is given in [82]. For different ratios of the compressed layer thickness to
the foundation radius tables are composed and isolines of horizontal displacements
of soil are built which should be taken into account in case of design of tower-type
structures which produce high load on the foundation.
The method of paired integral equations was rapidly developing in the 1950–
1970s. As noted by Popov [92], at present this method is the most flexible and
universal among the analytical methods for solving mixed problems of mathemat-
ical physics. Reviews of contact problem studies, performed using the method of
paired integral equations, are given in books by Ufliand [266, 267] and a book edited
by Galin [99]. Mostly the problems of determination of contact stresses under a
round punch, coherent with an elastic layer, including an anisotropic one, have been
solved.
Among the recent studies of applications of the theory of paired integral equa-
tions, one should mention two papers by Borodachev [36, 37] regarding the
generalized Hankel operator. An axisymmetric mixed boundary problem of the pres-
sure of a circular punch on a nonhomogeneous elastic half-space (in the absence of
[email protected]
friction and coherence forces), investigated in [36], is generalized for the case of
off-centre impression in [37]. A non-classical model of the elastic base is consid-
ered, for which the elastic modulus and the Poisson ratio vary with the base depth
4.1 Spatial Contact Problems for Rigid Flat-Bottom Punches 259

in a sufficiently arbitrary way. The problems under consideration are shown to be

reduced to a succession of one-dimensional second-order Fredholm integral equa-
tions, in case the punch equilibrium equations being fulfilled. A periodical law of
variation of elastic properties of the half-space material with depth was found, for
which the obtained integral equations admit exact solutions to be constructed. The
regularities of the influence of the base nonhomogeneity on the characteristics of its
interaction with the rigid foundation are shown. As follows from the author’s esti-
mations, the account of the elastic base nonhomogeneity can result in the settlement
variation for a round foundation of up to 33% in comparison with the homogeneous
base model. The contact problem for a centrally loaded elliptical punch on an elas-
tic base with a variable Poisson ratio is considered using the Hankel transformation
in [35, 39].
One should mention rather scarce [53, 63, 151, 175, 204, 212, 231, 248, 286]
exact and approximate analytical solutions of spatial contact problems for circu-
lar and elliptical punches, having been obtained rather long ago and included into
handbooks [205].
Distribution of vertical stresses at various depths under an asymmetrically loaded
round punch, coherent with an elastic half-space, was obtained by Muki [175].
Earlier Bycroft [63] had obtained formulae for horizontal displacements for a round
punch at the same coherence conditions at horizontal loading. A solution for slopes
of a round punch, coherent with a half-space, under momental load, was first
obtained by Borowicka [55], while Sneddon obtained the distribution of contact
pressures and vertical stresses in the half-space for the problem in question [248].
Reisner and Sagoci obtained a solution for the rotation angle of a round punch,
coherent with a half-space, under a torque load [212]. Approximate solutions for
slopes and contact pressures for a round punch under a momental load on a finite-
thickness layer were obtained by Egorov and Nitchiporovich [286]. Poulos deter-
mined the contact pressures, vertical compressive stresses and displacements along
the vertical axis of a centrally loaded round punch with a smooth base, located on
an elastic layer of finite thickness [204]. A centrally loaded elliptical punch on an
elastic half-space was studied in [231] where a distribution of contact pressures and
vertical compressive stresses under the centres of ellipse of different eccentricity
was obtained. An exact solution for horizontal displacements of the half-space sur-
face under an elliptical punch was obtained by Lee [151].
An efficient method of solution of a spatial contact problem for punches with
a cross-section close to circular, on an elastic half-space, is given in the book
[172]. Mossakovskii and co-authors have reduced the contact problem to a problem
of potential theory for a half-plane and is solved approximately. Pressures under
punches with a flat bottom in the shape of a rectangle, a parallelogram, a trian-
gle, a hexagon are determined as well as for the case of the punch bottom being
an area restricted by elliptical arcs, when the impression is performed without an
inclination (the contact area is a priori known: it coincides with the punch bottom).
[email protected]
The unknown functions of the problem are found using an expansion over a small
parameter characterizing the contact area shape deviation from a circle. Reliability
of the obtained solutions was estimated using a similarity between pressure isolines
260 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

and Prandtl function level lines at a rod torsion with a cross-section corresponding
to that of the punch [255].
Galin in his book [97] was the first to formulate the contact problem of a ring-
shaped punch, having set an analogy with a problem of electric charge distribution
on a ring-shaped conductor. He had not performed specific calculations concern-
ing the problem because the specific functions, contained in the solution, were not
tabulated.
Egorov has made a considerable progress in investigating the problem of a ring-
shaped punch [75, 77], using an approximate approach based on fulfillment of only
the stress boundary conditions. He obtained approximate formulae for settlements
and contact pressures under a centrally loaded ring-shaped punch at different ratios
of the inner and outer ring radii as complete second-order elliptical integrals. The
numerical results have enabled one to reveal that the ring-shaped punch settlement
remains almost unchanged with the variation of the inner-to-outer radii ratio within
0 < R2 /R1 < 0.6 in case the overall load on the punch remaining constant. It was
shown that at m = R1 /R2 > 0.9 the contact pressure distribution is similar to the case
of a rigid strip foundation. Besides, in [77] the dimensionless values of the ring-
shaped punch vertical displacements were tabulated depending on the parameter m.
In a number of subsequent publications the axisymmetric contact problem of a
ring-shaped punch was solved by various techniques based on the method of triple
integral equations [43, 76, 186, 272], power expansion of a singular integral equa-
tion kernel [112], application of the work reciprocality theorem [176], using asymp-
totic representations separately for large and small values of the parameter char-
acterizing the relative thickness of the ring-shaped contact area [9], using toroidal
coordinates [28], by expansion of the contact pressure function in a dual series with
subsequent determination of the coefficients from recurrent relations [214], using
a representation of the contact pressure function on the base of the expansion into
trigonometrical series [240], by collocational fulfillment of boundary equations in
case contact pressures being presented as a sum of boundary values of two analyti-
cal functions of a complex variable [173]. Note that the results of the first Egorov’s
study [75] appeared to be in a good agreement with subsequent more accurate data
[112, 176, 240]. Besides, since now the exact solutions for an axisymmetric con-
tact problem of a ring-shaped punch on an elastic half-space are known, the quoted
papers [9, 28, 43, 75–77, 112, 173, 176, 185, 214, 240, 272] are mostly of method-
ological character. The approaches, developed in these studies, have enabled the
contact problems of the pressure of a ring-shaped punch on an elastic layer of a
finite thickness to be solved approximately (Valov [272], Gubenko [111]), the solu-
tion of non-axisymmetric problems of an elliptical punch with a circular opening
of a small radius to be obtained (Aleksandrov [9]) as well as the one for a ring-
shaped punch with a flat oblique base (Hara, Shibuya et al. [116], Popov [200],
Borodacheva [47, 48]), the action of a ring-shaped punch on a nonhomogeneous
half-space to be studied (Popov [200], Protsenko [208]) as well as its action on mul-
[email protected]
tilayer bases (Lamzyuk, Privarnikov [149], Nikishin and Shapiro [180]). One should
also mention important from the practical point of view studies of Borodacheva
[49–52] where the field of displacements and stresses in an elastic half-space due to
4.1 Spatial Contact Problems for Rigid Flat-Bottom Punches 261

a ring-shaped punch was investigated as well as a work by Borodachev [42] where

the character of distribution of pressures under a ring-shaped punch is studied in
detail.
An approximate analytical solution of a rather general form for a flat ring-shaped
punch on elastic homogeneous and nonhomogeneous (with a power dependence
of the deformation modulus E = E0 zn on depth) bases under an eccentrically
applied vertical force is given in [225]. Ordinary assumptions are made regarding
the absence of friction in the contact area and the absence of the bottom uplifting
from the base. The problem is reduced to a system of triple integral equations which
is solved by expansion of the sought functions into power series whose coefficients
are found in the third approximation using a small parameter method. The value
ε = a/b where a and b are the inner and outer radii of the ring, respectively, is taken
as the small parameter. Unfortunately, the found approximate analytical solution
has a limited applicability in construction of foundations since it is valid only for
very broad punches and non-realistic bases (with zero deformation modulus on the
surface).
The asymptotical method and its modifications for solving contact problems were
developed in numerous studies by Vorovich, Aleksandrov, and Babeshko. A gen-
eralizing book by these authors [281] contains mixed (contact) problems of two-
dimensional and three-dimensional theory of elasticity on the interaction of strip,
circular, and elliptical punches with non-classical bases of layer and wedge type.
Mostly so-called asymptotic methods of ‘large and small λ’ are used. In partic-
ular, in the problem of interaction of a round punch with an elastic layer the λ
parameter is the ratio of the layer thickness to the characteristic size of a contact
area, equal to the punch diameter. The proposed method is based on an idea of a
transition from the known solution of the classical problem of the punch action on
an elastic half-space to the approximate solution of impression of the same punch
into an elastic body with finite dimensions in one direction. The main advantage of
such an approach is obtaining the contact pressure function in a closed and rather
simple analytical form, convenient for engineering applications. However, the gen-
erality and efficiency of the method are low – the asymptotic approach has been
developed only in the case of building approximate solutions of spatial problems
for punches of circular and elliptical cross-section impressed into an elastic layer.
The problem of pressure of a punch with a wedge-shape cross-section on an elas-
tic half-space in the presence of a load outside the punch was first solved by Galin
[97]. Later Rvachev solved this problem without an additional load and simultane-
ously revealed a singularity of contact pressures in the vicinity of the wedge ver-
tex [222]. He found an approximate solution using a semi-inverse method together
with Galerkin method. This solution is especially important for the determination
of singularity indices in the angular points of polygonal domains in case the contact
problems being solved by numerical methods. It should be also noted that later the
problem of an infinite wedge was studied by the asymptotic method by Aleksandrov
[email protected]
and Babeshko [10].
The problem of an off-centre loaded punch of an arbitrary cross-section rest-
ing on an elastic homogeneous linearly deformed half-space was considered in
262 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

[84]. An approximate semi-inverse analytical method of determination of the punch

inclinations with respect to the coordinate axes is proposed. The method does
not require preliminary calculation of contact pressures. The general formula for
the angular displacements of the punch is rather complex, being an infinite series
of integrals of the hypergeometrical Gaussian function. In the case of an ellip-
tical punch this method leads to an exact result. By truncation of the solution
to five harmonics, the developed approach is applied to punches with the cross-
section in the shape of a rectangle, a rhombus, a circular sector, or a segment.
The obtained results, having been compared with the known numerical solutions
[225], show the applicability of the developed approximate formulae in practical
design.
Rvachev in his earlier studies on theory of contact problems paid main attention
to the influence of geometrical factors on the character of the contact stress distribu-
tion. The search for approaches to the solution of contact problems for punches of
polygonal cross-section has led him in 1963 to the development of a new mathemat-
ical approach – the R-function method [223], combining logic algebra methods with
classical methods of mathematical physics and computational mathematics. In the
book [225] the structural method developed by Rvachev (the R-function method)
is applied to find approximate solutions of the contact problems for punches of a
rather arbitrary cross-section shape (the contact area can be bounded by a piecewise
smooth curve). The contact problems for punches on a half-space with contact areas
in the shape of polygons, an elliptical ring, a rectangle with an elliptical cutout, a
part of a circle, etc. have been solved. A characteristic of this approach is the con-
struction of series using coordinate successions for non-traditional (non-classical)
contact domains in the framework of elementary functions, exactly satisfying the
boundary conditions of the contact problem being solved by a variation method of
Bubnov-Galerkin type.
When the contact area is known (in axisymmetric problems as well as at central
loading of punches of canonical shape), an approximate method consisting in setting
the contact pressure function as an infinite series over known (coordinate) functions
but with unknown coefficients, is rather widely spread. The boundary conditions are
satisfied approximately when a finite number of terms in the series are preserved. As
noted in [99], similar approaches as well as the R-function method are, in general,
badly conditioned and can result in big errors if the coordinate functions (polyno-
mials) are chosen for the given geometry of the contact area boundary in a wrong
way. The examples of solution of contact problems using expansions into power
series as well as into series over orthogonal polynomials (Gegenbauer, Chebyshev,
Jacobi, Laguerre, Hermit, etc. [3, 259]) are described in detail in known books [105,
201–203, 235, 241, 242].
In a rather simple and most widely spread in practice case of a rectangular foun-
dation, the solutions of the contact problem at central and off-centre loading (under
full contact) were obtained by various methods only approximately, with different
[email protected]
degree of accuracy in [31, 41, 57, 105, 154].
Whitman and Richart [285] have obtained approximate solutions for vertical
displacements of a rectangular punch with a smooth base at central vertical load.
4.1 Spatial Contact Problems for Rigid Flat-Bottom Punches 263

Barkan in his book [30] gives an approximate solution for horizontal displacements
of a rectangular punch undergoing a horizontal load in the direction of one of the
central axes. His results are presented in a tabulated form depending on the ratio
of the punch sides and Poisson ratio values for the soil base. Way has obtained
an approximate solution for the slopes im of rectangular foundations, loaded by a
moment in one of the directions and presented tabulated values of im at different
ratios of the punch sides [284]. The results of the studies [30, 284, 285] were later
quoted in a handbook [205].
The contact problem solution for a narrow punch of a rectangular cross-section
(a narrow girder) on an elastic half-space in an approximate formulation was con-
sidered by Borodachev and Galin [40, 44]. It was implied that the pressure value
by unit length of the girder is subject to be determined and contact pressures in
each cross-section agree with the solution of the corresponding flat problem. At
such assumptions the problem is reduced to a one-dimensional integral first-order
Fredholm equation with a kernel containing a complete first-order elliptical inte-
gral. After polynomial approximation of the regular part of the kernel the problem
is finally reduced to the solution of an infinite system of linear algebraic equations.
This system was approximately substituted by a finite system of the 11th order and
solved for different values of a small parameter ε = δ/a = 0.02÷0.2 where 2a and
2δ are the length and the width of the rectangular contact domain of the punch
(the girder). The distribution of pressures per unit length of the punch was obtained
as well as its settlement under a central vertical force. An important result of the
investigation performed was the consideration of an issue regarding the range of
applicability of the Winkler hypothesis in the engineering theory of beam bending
on an elastic base.
Burmistrov [62] suggested an asymptotic method of solution of the contact prob-
lem for the punches of prolate shape, not necessarily with a straight axis. In order to
simplify the problem, a small parameter is used, related to the narrowness of the con-
tact domain. The initial problem with a two-dimensional integral equation is shown
to split into two one-dimensional integral equations, being solved sequentially. A
number of asymptotically exact and one polynomially exact solution have been
obtained (for an elliptical domain). In the author’s opinion, the developed method
is very important for the problem of contact of a roller with a bearing ring. How-
ever, in our opinion, the asymptotic method of Burmistrov can also be successfully
used for the calculations of structures of prolate shape (beam type structures or strip
foundations of finite length) on elastic bases.
Numerical solutions. The performed analysis of exact and approximate solutions
has shown that a strict mathematical study of the processes of spatial contact inter-
action can be performed for a known contact domain only for the punches of the
simplest geometry and, as a rule, at central loading. Mathematical difficulties while
finding the exact and approximate solutions of the spatial contact problem still
remain insurmountable in case the punch geometry being more complex as well
[email protected]
as with the account of various sorts of nonhomogeneity, laminarity, anisotropy, etc.
for the elastic base [59, 73, 179, 239]. Therefore, of high practical importance are
contact interaction studies, carried out using computer modelling.
264 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

In case spatial contact problems being solved, including those with an a pri-
ori known contact domain, the most efficient are numerical methods based on dis-
cretization with subsequent piecewise constant, linear, or quadratic approximation
of the contact stress function. Among the numerical method of this trend, used for
the design of foundation bases, the most widely spread are finite-difference method
(FDM), finite-element method (FEM), and boundary-element method (BEM). It
should be noted that the finite-difference and finite-element methods assume the
discretization of the base area, which is active for the foundation (punch) being cal-
culated. In the case of spatial problems both methods, though formally allow the
physical and geometrical nonlinearity of the base material as well as bulk forces
to be formally taken into account, still result in the solution of algebraic equation
system of a very high order. On the other hand, in the boundary-element method
the account of the above factors makes the problem solution much more labour-
consuming; thus such main advantages of the method as discretization of only the
domain boundary and small amount of input and output data, can vanish.
Finite-difference method. Using the finite-difference method, Vinokurov in the
end of the 1960s elaborated an iterative method of calculation of bases and foun-
dations in planar and spatial formulations [279]. The numerical algorithms and
solution method constructed have enabled the elastic-plastic-viscous soils to be
considered with and without the account of structural strength for a number of
cases when solutions in the closed form are missing. As a field of application of
the elaborated method stress-strained states of soil bases are considered, including
the case of being subject to the action of a rigid punch. Main finite-difference equa-
tions to determine settlements and slopes of a rigid square punch located on a trans-
versely isotropic base and loaded by a vertical force and moments in two orthogonal
planes. Both the settlements and slopes can be found only after the calculation of
the stress-strained state of the base. The book [279] contains only theoretical issues
of calculation of bases and foundations. The practical application of the elaborated
finite-difference approach to the solution of the essentially spatial contact problem
for a rectangular punch on a linearly deformed base was given much later (in almost
two decades) in the studies by Shevchuk [133, 238] which now, in fact, determine
the level having been achieved for the solution of spatial contact problems using the
finite-difference method.
Shevchuk has worked out a finite-difference algorithm to solve the spatial con-
tact problem for punches of rectangular cross-section located on an anisotropic and
simultaneously nonhomogeneous with depth layer [238]. The formulation of the
problem includes complete determination of the stress-strained state of the layer
under the flat punch undergoing all six components of load. The finite-difference
approximation has been performed for differential equations of equilibrium, Cauchy
equations, and physical equations of an anisotropic nonhomogeneous body. On a
bounded (prismatic) domain of the layer under the punch a regular spatial grid is
built. The condition of full coherence of the punch and the elastic layer is accepted.
[email protected]
Equations, obtained from the condition of normal and tangential stress being equal
to zero, are used to calculate the displacements of the finite-difference nodes of the
free surface. Due to the great number of the finite-difference equations and to the
4.1 Spatial Contact Problems for Rigid Flat-Bottom Punches 265

coefficient matrix being very sparse, an iterative Seidel method is used to obtain
a solution. The studies of the iterative scheme have resulted in a condition of the
numerical solution stability relating the Poisson ratio of the layer and the ratio of
moduli of elasticity in the horizontal and vertical directions. Comparison of the
results obtained for the grids of various step has shown the approximation errors to
be localized, as it should be expected, near the punch edges at a distance from them
of half of the punch bottom width. The effect of remoteness of the lateral boundaries
of the base approximation area is traced only for the overall settlement of the punch.
The effect of the lateral sides is shown to be insignificant in case they being remote
from the punch centre by more than the triple width of its bottom. Comparison of
the finite-difference calculations with the results of experiments and field studies of
real structures has shown rather high discrepancies (up to 12%), especially near the
contact surface.
The necessity of solving contact problems for various types of structures inter-
acting with soil bases, requires influence functions to be used. Shevchuk managed
to build a discrete analogue of the latter, using the finite-difference approach in
the following way. From the calculation for an elastic anisotropic layer loaded by a
unit vertical force, uniformly distributed within one cell of the finite-difference grid,
influence coefficients are determined, numerically equal to vertical displacements of
the layer surface nodes. From the influence coefficients an influence matrix is built
in a standard way which was used for the combined calculation of the frame system
and the base.
Only a rather small number of publications available touch the problem of
account of physical nonlinearity of the base material when the problems are solved
using the finite-difference method [161, 207, 243, 244].
Malyshev and Proskuryakov in [207] present the results of numerical solution
of the problem of the stress-strained state of a half-space subject to a flexible load
uniformly determined over a square area. The half-space is characterized by non-
linear deformational properties, described by nonlinear relations of deformational
theory of plasticity. The bulk deformation modulus is taken constant and the gen-
eralized secant shear modulus is determined by the Botkin dependence [54]. The
dependences between the stresses and deformations are taken in the form of Hencky
equations. The problem is solved by the finite-difference method in displacements.
An integral-interpolational method is used to construct the difference scheme. The
obtained difference equation system is solved iteratively using the Seidel method.
In view of the symmetry of the calculation scheme only a quarter of the base and
the load area is considered. The account of the nonlinearity of the soil physical
properties was carried out incrementally: the nonlinear problem of the action of a
load of a given intensity on a half-space is reduced to a succession of problems
of the action of an incremental load with constant deformation moduli. As noted
in [207], convergence of the incremental iteration method was not always easily
achieved, what made the computations much longer (the number of iterations in
[email protected]
each approximation was not less than 10). Therefore, the need of the discretization
degree to be increased in the areas of high displacement gradients for the case of
the three-dimensional soil mass requires a sharp increase of the computation time.
266 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

Probably, this did not enable the authors of [207] to solve the contact problem of a
rigid rectangular punch by finite-difference method even for the case of a symmet-
rical calculation scheme.
The contact problem for a rigid punch on a physically nonlinear base is more effi-
ciently solved by the finite-difference method in the case of axial symmetry what
practically corresponds to the planar formulation. The contact problem of the defor-
mation of a nonlinearly elastic base under a round rigid punch, studied by Solomin
et al. using the finite-difference method [161, 243, 244], is discussed below when
the numerical solutions of the contact problem for round punches is specially con-
sidered.
In opinion of a number of authors, among the main disadvantages of the finite-
difference method, revealed at the contact problems being solved, one should men-
tion considerable difficulties of finite-difference formulations for different boundary
conditions as well as violation in a number of cases of symmetry of the resolving
algebraic equations what finally does not enable this method to be recognized as a
sufficiently universal one.
Finite-element method. Recently the finite-element method has become more
widely applied for numerical solving geotechnical problems than the finite-
difference method. The finite-element method is especially efficient to evaluate the
stress-strained state of bodies of a finite size and a complex shape both in planar
and spatial problems of mechanics of solid deformed bodies. In the framework of a
unique approach this method is used to solve many complicated geotechnical prob-
lems of joint deformation of a foundation and soil with the account of geometrical
and physical nonlinearities, factors of nonhomogeneity, anisotropy, consolidation,
etc. Different approaches to the investigation of the stress-strained state of soil bases
using the finite-element method can be found in the studies of prominent geotechni-
cians [24, 70, 85, 107, 114, 142, 194, 196, 216, 218, 236, 268, 287–289, 292–294],
devoting the main attention to the solution of planar and axisymmetric problems.
In comparison with the planar and axisymmetric problems, the number of
degrees of freedom in the finite-element method for three-dimensional (essentially
spatial) problems increases by several orders of magnitude. This results in such
insurmountable yet difficulties in practical application of the finite-element method
to solve geotechnical spatial problems as the required computer RAM size and speed
increase as well as the increase of input and output data size. Therefore, even at
contemporary level of computer systems application of the finite-element method
in spatial geotechnical problems still remains rather limited. The situation is even
more complicated for the solution of problems (nonlinear, with a priori unknown
contact domains, etc.), requiring various iterative processes to be organized when
the elastic calculation procedure is to be repeated many times.
Pilyagin and Kazantsev in their papers [193, 195] and book [194] were the
first to give the solution of a spatial elastoplastic problem of evaluation of the
stress-strained state of bases. They also describe a software package with numerical
[email protected]
algorithms of the problem produced using the finite-element method. The FEM is
considered there for the case of symmetrical foundation structures at central loading.
This has enabled the solution of the discrete problem to be essentially simplified by
4.1 Spatial Contact Problems for Rigid Flat-Bottom Punches 267

storing only the symmetrical part of the rigidity matrix in the computer RAM and
using effective methods of solving systems of linear equations with symmetrical
matrices.
Based on the results of the solution of a spatially symmetrical elastoplastic prob-
lem, the analysis of the influence of rectangular foundation size on the development
of settlements at different physical and mechanical parameters of the soil was car-
ried out. The influence of the foundation bottom shape on the soil base deformation
was studied only by variation of the ratio η = l/b of its length to its width. In partic-
ular, at η≥6 the rectangular foundation settlement is shown to be practically equal
to that of a strip foundation of the same width. For an absolutely rigid square punch
being impressed into a homogeneous soil base by a central force, the calculated
results are compared with the experimental studies by Rabotnikov and Kovanev
[209]. The experimental data and the FEM theoretical solution for the settlement
versus load dependences appeared to be in a good agreement. As one could expect,
the difference between the calculated and experimental curves increases with the
load increase. The different soil strength conditions being used have made the set-
tlement values for the linearly elastic solution more exact, but still differing from
the actual values by up to 7% at the Mohr-Coulomb strength condition, and up to
14% at the Mises-Schleicher-Botkin strength condition.
The method of solving a mixed spatial problem of theories of elasticity and plas-
ticity, quoted in [193–195], as well as the analysis of influence of various factors
on the variation of the stress-strained state of the soil base have shown a conceptual
possibility of the finite-element method to be applied in practical design of foun-
dations for buildings and structures. However, the variety of factors affecting the
stress-strained state of the bases, the long computation time for spatial elastic and
elastoplastic problems strongly encumber the application of the three-dimensional
FEM in practical engineering. This has made Pilyagin and Kazantsev [194] to sug-
gest a way to apply the solutions of mixed problems of soil mechanics for elu-
cidating the factors affecting the settlement, for choice of versions most widely
used in real design of bases and foundations as well as for obtaining correspond-
ing approximate dependences or making alignment charts. It should be noted that
due to the above difficulties in solving essentially three-dimensional problems, the
obtained finite-element solutions in [194] are made ready for practical application
in the form of alignment charts and multifactor power models only for the strip
type foundations, i.e. for the conditions of flat deformation of the soil bases when
the two-dimensional formulation of the contact problems essentially simplifies the
contact interaction analysis.
The known general approaches to the solution of static planar and axisymmetric
contact problems by the finite-element method are discussed in detail in [198]. The
influence of various contact conditions, rheological properties of the material (creep
and plasticity) and contact domain variation in the course of loading on the charac-
ter of functioning of the contacting structure elements, mostly of machine-building
[email protected]
type, has been studied. The analysis of the contact interaction of elastic and elasto-
plastic bases with rigid punches, performed in [198], has a direct similarity with the
problems of calculation of bases and foundations. A test solution of the classical
268 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

problem in an axisymmetric formulation of impression of a rigid circular punch

into an elastic half-space and a constant-thickness layer has clearly shown some
shortcomings of using the finite-element method for solving contact problems for
infinite domains. For numerical modelling the semi-infinite elastic half-space was
approximated by a finite-size domain (0 ≤ r/a ≤ 30.0 ≤ z/a ≤ 150), at the boundaries
of which the resting conditions were formulated. In order to obtain a satisfactory
numerical solution, the meridional cross-section of the deformed base domain was
meshed into finite elements with condensation towards the contact domain bound-
ary. The total number of the grid nodes was 1230, among them only 15 (i.e. slightly
above 1%) being on the contact area. Comparison of the numerical values for pres-
sures under the punch with the data of analytical solutions, an exact one for the
half-space [128] and an asymptotic one for the layer [281] appeared interesting.
The general conclusion of the comparison of the above problems was made that in
the areas adjacent to the punch edge, a considerable error of the numerical solution
is observed. For these areas the stress-strained state of the base is of complicated
character and cannot be approximated in a satisfactory way by bilinear coordinate
functions of the finite element. It is noted that obtaining more exact numerical results
in the presence of areas with rapidly growing solutions makes the contact problems
of the class under consideration (for rigid punches on infinite bases) rather labour-
consuming in case the finite-element method being used.
While performing calculations for the stress-strained states of bases under
punches the most convenient is to apply linear theory of elasticity, but the load value
should be restricted in accordance with the average pressure over the bottom (cal-
culated soil resistance) [258, 269]. This restriction is abandoned in case nonlinear
(elastoplastic) models of soil bases being used. Axisymmetric problems regarding
stresses and displacements of a nonlinearly deformed half-space, on whose surface
a circular or a ring-shaped centrally loaded punch being located, are considered in
[12, 64, 101, 161, 198, 243, 244, 258, 269]. In all cases no slippage of the soil under
the bottom was assumed to occur, i.e. there were no horizontal displacement over
the contact of the soil and the punch.
The problem of a circular rigid punch on a soil base is the simplest problem
of spatial type and the most widely spread in soil mechanics and foundation engi-
neering [12]. On one hand, a wide experimental material has been obtained for the
problem of a circular punch. On the other hand, a nonlinear solution of this problem
can be used for the adjustment of equations of the soil state and practical calcula-
tions of strip and columnar foundations. Therefore, it is this contact problem, from
which many authors begin the application of a new calculation method; besides, the
solutions of this problem are complex enough and enable one to judge upon the
efficiency of the calculation method applied.
A method of calculation of soil bases for rigid circular foundations at axisym-
metric conditions using the finite-element method in combination with the initial
stress method is proposed in [12]. A model of soil as a strengthening elastoplastic
[email protected]
medium of modified “Cam-Clay” model is considered [232]. The node coordinates
of the calculated finite-element grid were given analytically as the intercepts of con-
focal ellipses and hyperbolae [292]. This discretization has enabled a rather large
4.1 Spatial Contact Problems for Rigid Flat-Bottom Punches 269

base area (by 3.8R in radius and by 3.6R in depth where R is the punch radius) to be
embraced using a small amount of elements (24 finite elements bound by 20 nodes).
Based on the calculations, settlement versus load plots are built, which enabled the
effect of initial porosity on the deformational properties of the soil to be revealed.
A typical more intense increase of stress under the punch edge (in comparison with
the centre) with load increase was found.
Glushkov has obtained solutions for mixed elastoplastic problems in an axisym-
metric formulation for ring-shaped and circular foundations [101]. Using the finite-
element method, the effect of strength characteristics (ϕ and c) and deformational
characteristics (E and ν) of soil as well as the foundation size on the variation of the
stress-strained states of bases to be studied. The stressed state of the soil medium
was determined only under an external load. The soil in the prelimit state was con-
sidered as a solid linearly deformable medium, being transformed under further
loading to a limit (plastic) state in accordance with the Mohr-Coulomb criterion of
fluidity (strength). The calculations were performed using a stepwise procedure of
load application. Regarding the contact of the foundation bottom with the base full
adhesion is assumed. With the account of the natural stressed state the calculations
were carried out to determine the effect of strength characteristics (ϕ and c) and
deformational characteristics (E and ν) of the soil on the stress-strained state in the
foundation active zone. The analysis of the results has shown that the settlement
versus load dependences are generally nonlinear (bilinear). With the exception of a
small initial part these dependences are almost linear, and their slope and the load
bearing capacity of the base are essentially dependent of the soil strength parameters
ϕ and c as well as of the ratio of the inner and outer radii of the ring. Detailed iso-
line patterns of horizontal and vertical displacements of the soil, main and tangential
stresses in the foundation base were also obtained. An important result of the investi-
gation carried out in [101] consists in finding the laws of formation and development
of a plastic deformation zone in the base active area under an axisymmetric load-
ing of ring-shaped and circular foundations. These zones arise in an area, directly
adjacent to the bottom, and develop laterally and downward with the increase of the
load on the foundation. The character of formation and development of the plastic
zones mostly depends on the strength characteristics of the soil: the smaller are ϕ
and c values, the earlier are the plastic zones in the base formed and the more intense
is their growth rate. Similarly to the case of strip foundations [194], a rather high
computation time for the axisymmetric problems with a variety of factors affect-
ing the stress-strained state of the ring-shaped and circular foundation bases did not
enable Glushkov to apply the elaborated method in the engineering practice. Sim-
ilarly to [194], for wide application of the solutions of axisymmetric problems of
soil mechanics, the author proposes to determine the parameters (strength-related,
deformational, geometrical), affecting the settlement, search through the variants,
the most widely involved at real conditions, and obtain the corresponding approxi-
mate dependences and homographs. In particular, a multifactor power dependence,
[email protected]
assumed in [101] for the settlement of a round foundation under a central force, has
shown a good agreement with the experimental data and can be applied in practical
design. However, for practically important cases of off-centre loading no effect of
270 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

the strength and deformational properties of soil on the settlements and slopes of
ring-shaped and circular foundations have been found using the FEM in the quoted
papers [12, 101, 198].
The authors of [64] propose to use local variation method for solving the axisym-
metric problem of impression of an absolutely rigid flat-bottomed punch into an
elastoplastic base. The unlimited base area in polar coordinates r ≥ 0, z ≥ 0 is sub-
stituted by a finite-size domain 0 ≤ r ≤ R, 0 ≤ z ≤ H where R and H are sufficiently
large. At the boundary of the domain under investigation r = R, z = H the condi-
tions for displacements are given from the known solution of the problem of action
of a concentrated force on an elastic half-space (from the Boussinesq problem). The
main equations of the elastoplastic problem under consideration, besides the equilib-
rium equation, included the Mises plasticity condition and the Prandtl-Reiss equa-
tion of state. The variational principle of theory of plastic flow is applied, according
to which the real increments of vertical and radial displacements are the realizations
of the minimum of the increment energy functional. The proposed algorithm for
solving the variational problem has the following stages: (i) meshing of the stressed
base domain (0 ≤ r ≤ R, 0 ≤ z ≤ H) into triangular cells by a non-uniform grid, (ii)
combining the neighbouring triangular cells into one cell and calculation of inte-
grand functions in the energy functional using the finite-element expressions, (iii)
solving the obtained variation-difference problem using the local variation method.
The calculation data enabled the development of plastic zones, arising at the
punch edges, to be traced. Comparison of the exact solution with a numerical solu-
tion in a purely elastic formulation has shown the difference not to exceed 2.5% for
the calculation domain discretization by a 7 × 7 grid and 1% for the 11 × 11 grid.
Due to the presence of several nested iterative processes as well as the formulae used
being rather cumbersome, the scope of calculations required to solve the problem,
is very large. In spite of the possibility of complex rheology to be taken into account
(both in the framework of the plastic flow theory and for the deformational theories
of plasticity) and a rather high calculation accuracy, the algorithm of solving the
contact problem of a punch on an elastoplastic base, proposed in [64], is elaborated
only for solving two-dimensional and axisymmetric problems.
In order to solve the contact problem of a round punch, in [161, 243, 244] the
relation between the stress and deformation components was taken in the form of
the generalized Hooke’s law where the shear modulus and the modulus of dilatation
are scalar functions of the stress tensor invariants. The form of the functional depen-
dence was determined experimentally. The contact problem is reduced to a system
of nonlinear differential equations with respect to the vertical and radial components
of the displacement vector. The solution is obtained using the successive approxi-
mation method (Ilyushin elastic solutions) in combination with the finite-difference
method. While the problem being solved, the half-space was substituted by a cylin-
der with a radius 16R and height 8R where R is the punch radius. The calculations
have revealed the effect of the variation character and value of the secant shear
[email protected]
modulus on the stress-strained state of the base at the same value of the modulus of
dilatation. The issue of expediency of account of the soil structural strength is stud-
ied at the example of a sand base. Two solutions are compared. The first one was
4.1 Spatial Contact Problems for Rigid Flat-Bottom Punches 271

obtained using the Botkin model whose parameters were taken from the test data
for the sand of disturbed structure. The second solution takes into account the struc-
tural strength at shear. A quantitative discrepancy in the data consisted mostly in
the fact that the account of the structural strength enabled the calculated settlement
to be essentially reduced (based on the calculation data – almost by three times).
Besides, the account of the structural strength enabled a number of new qualita-
tive effects to be revealed, corresponding to the experimental data: formation and
development of a compressed zone; formation of an elastic kernel; concentration of
vertical deformations on the foundation axis directly under the elastic kernel; more
intense damping of displacements with depth and corresponding to the experimen-
tal data more complicated (in comparison with the Botkin model-based solution)
character of transformation of the reactive pressure profile.
Some important features of the numerical solution of problems of interaction
of punches with soil bases were noted by Likhovtsev and Estrin [155]. They have
performed comparative calculations of interaction of a punch with a base (of a finite-
thickness layer type) both for a cinematic (traditional) and a force loading schemes
in the planar problem formulation.
In the case of the cinematic loading scheme the punch is considered absolutely
rigid. At each loading step an increment of the punch settlement is given, a contact
pressure profile is determined and from the latter the loading step value is deter-
mined. This process is performed iteratively until the given load is reached or the
base loses its carrying capacity. At this loading scheme there is no necessity to
discretize the foundation itself (this reduced the computation time and makes the
computer RAM less loaded), the matrix of the resolving system of linear algebraic
equations is obtained well conditioned and, consequently, when the global rigid-
ity matrix is formed there are no truncation errors (due to the arithmetic operation
between numbers of the same order of magnitude). Nevertheless, such approach
does not take into account deformational characteristics of the punch; besides, the
problems with off-centre and horizontal application of load cannot be considered.
Note that numerical data for the case of the spatial contact problem when the finite-
element method at the cinematic loading scheme was used for a rectangular punch
on an elastic two-layer base, were obtained in [95].
At the force loading scheme, the punch (with a finite rigidity) and the base are
considered as a compound body with different mechanical characteristics. Similarly
to the base, the punch is subject to discretization into finite elements. The prob-
lems of interaction of the foundation with the base at off-centre load application can
be solved, the effect of the deformational characteristics of the punch on the con-
tact pressures and the stress-strained state of the base can be estimated, stress and
bending moments in the foundation can be determined with the account of its defor-
mational characteristics. Consequently, the application of the force scheme enables
a broader class of problems to be solved in comparison with the cinematic loading
scheme. However, the calculations have shown [155] that numerical application of
[email protected]
the force loading scheme (especially, using a computer), should be performed with
double-precision data format. This requirement follows from the appearance of trun-
cation errors at the formation of the global matrix of rigidity for the different-type
272 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

materials whose deformational characteristics differ by several orders of magnitude.

According to the data of [155], the single-precision calculations, performed accord-
ing to the force scheme (at central loading), have shown a symmetry violation for the
displacements and stresses in the nodes located symmetrically with respect to the
punch axis, i.e. the results become erroneous. Thus, when the force loading scheme
for solving the spatial problems of interaction of rigid punches with soil bases by
finite-element method is used, the requirement of the calculations to be performed
with double precision (to avoid numerical errors) will require a rather essential extra
increase of the computer RAM size and the computation time.
Boundary-element method. As seen form the analysis of the available literature,
for solving planar and axisymmetric contact problems as well as spatial contact
problems in case the contact domain size being comparable with the characteristic
size of the contacting bodies, a boundary-element method is efficiently used [22].
For solving essentially spatial problems in soil mechanics and foundation engineer-
ing when one of the deformed bodies (base) is infinite (half-space, unlimited layer,
wedge, etc.), the boundary-element method is being used more and more. In this
method continuous contact forces are substituted by discrete values in accordance
with the mesh of the contact domain only into boundary (contact) elements.
While searching for the discrete contact forces, maximally satisfying the bound-
ary conditions, as a rule, two approaches are used [128]: a direct approach when the
boundary equations are satisfied exactly in the given collocation points (These are
usually taken in the boundary-element centres), and a variational approach when
the discrete force values are chosen from the conditions of the energy function
minimum. Aleksandrov was one of the first to propose a method of solving three-
dimensional problems of theory of elasticity by numerical implementation of the
integral equation method [6]. He constructed solutions for circular and ring-shaped
punches with a flat bottom based on the elaborated method of determination of
stresses and displacements due to a load, uniformly distributed over a ring [7]. The
proposed method was tested by a problem of pressure of a circular flat-bottomed
punch on a half-space. The possibilities of this variant of the boundary-element
method appeared to be essentially dependent of the accuracy of the contact pressure
calculation from the algebraic analogue of the boundary integral equations of the
problem. Simultaneously, determination of tangential stress field in the half-space
by this method gives the results that are in a good agreement with the experimental
data [174].
Most of the essentially spatial problems were solved using different variations
of the BEM for the foundation models with rectangular cross-section [34, 68, 140,
165, 166, 182, 186, 187, 192, 213, 270, 276]. Only several studies consider circular,
ring-shaped, and elliptical punch foundations [6, 88, 140, 276].
Relatively recently Bokiy and Petrishin elaborated an algorithm for numerical
solution of the pressure of a smooth punch of square cross-section on an elastic
half-space [34]. A case of central loading is considered, resulting in the symme-
[email protected]
try of the unknown pressures on elementary squares, discretizing the square con-
tact domain. The coefficients of the resolving system of integral equations, being
the dual integrals of the Boussinesq solutions for a vertical concentrated force, are
4.1 Spatial Contact Problems for Rigid Flat-Bottom Punches 273

calculated in the finite form via elementary functions. The approximate solutions of
this problem at various degrees of discretization, obtained in [34], correct the errors
in the solution of the same problem, quoted in a manual [213] and using the tabu-
lated values of the influence matrix coefficients (according to Zhemochkin), correct
only for certain directions.
Novotný and Hanuška also studied the contact problem for a rigid rectangular
punch resting on an elastic homogeneous half-space [182]. As in similar works by
most of other authors, while determining the contact pressures, horizontal stress
was not accounted. A unilateral constraint on the contact of the punch and the base,
preventing from tensile contact stress, was taken into account. This requires the
integration domain (contact domain) to be defined while solving the main integral
equation of the problem and the equations of equilibria. Numerical solution of the
problem is performed using the boundary-element grid of right-angle triangles, on
each of which the contact stress is taken according to a linear law. The algorithm
of the problem solution includes an iterative process of removal of negative contact
stresses. The calculations have shown the non-uniform discretization with conden-
sation of triangular elements near the punch angles to produce essentially better
results than a uniform one. Detailed studies of the ratio of the angle uplifting and
the centre settlement for a square punch at eccentric application of the resultant
vertical force have been performed.
Veryuzhskii et al. [276] proposed a general approach to the solution of spa-
tial contact problems regarding the interaction of soil and rigid foundations, using
potential methods. The main concepts of the potential method, stated in [277], are
based on the consideration of unit point forces acting in a soil medium, and enable
the method in question to be considered as a numerical-and-analytical form of the
boundary-element method. Based on the application of Betti’s theorem to the basic
state under consideration and an auxiliary state constructed, an algebraic analogue of
the Somigliana integral equation is derived. In general, Mindlin’s solutions are used
as the integral equation kernels [169]. The main attention in the work is paid to the
increase of efficiency of the contact problem solution method due to analytical deter-
mination of antiderivatives for the integrals contained in the algebraic analogues of
the Somigliana formulae. In order to estimate the abilities of the numerical-and-
analytical potential method, test problems for rigid round and square punches rest-
ing on the surface of an elastic half-space and loaded by a central vertical force, were
solved in [276]. The punch settlement values and the contact stress profiles over the
punch bottom were obtained for a different density of the contact surface mesh.
For a square punch the settlement value obtained by the numerical-and-analytical
potential method on a 6 × 6 boundary-element grid appeared practically the same
as the one found from the Schleicher formula [260]. A rather good coincidence with
the Boussinesq solution is also noted for the contact stress distribution in the soil
under a rigid round punch: the discrepancies obtained are within the accuracy of
graphic interpolation. The circular contact domain was approximated by a regular
[email protected]
hexadecagon, meshed using 68 contact (boundary) triangular and quadrangular ele-
ments. In spite of the approach proposed being rather general, the contact problems
for punches of complex-shaped cross-sections, resting on non-classical bases and
274 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

loaded by off-centre forces (what results in the bottom uplifting from the base),
have not been solved.
Numerical implementation of the integral equation method in contact prob-
lems for an orthotropic half-space was performed by Martynenko, Knyazeva, and
Romanchik [140, 165, 166]. An approach to the numerical solution of integral equa-
tions developed in [167], was used to solve the problem of pressure of a coherent
(without slippage) [165] and a smooth [166] punches of rectangular cross-section on
a half-space. The rigid punch displacement parameters are determined by invoking
six equilibrium equations. Three contact stress functions are proposed to be sought
in the form of a product of a specific part of the solution and a linear combina-
tion of Sobolev δ-shaped averaging functions [103] whose coefficients are found by
collocation method. In order to describe the contact domain boundaries R-function
representations are used. A number of problems for complex-shaped punches with
the cross-section of triangular, rectangular, elliptical shape and their combinations
have been also solved [140].
A number of experiments performed have shown that many important features
of the engineering and geological conditions, affecting the contact interaction of the
foundation with the soil base can be successfully taken into account in the frame-
work of the boundary-element method: the compressible soil mass variability [186,
187], non-uniform salinity [192] and laminarity [68, 88, 270] of the soil bases.
Onopa and Fedorovskii have solved the spatial problem of the settlements and
slopes of a rectangular punch on a wedge-shaped base, i.e. on a compressed layer
whose thickness varies linearly [186, 187]. A kernel of the wedge-shaped base
contact model, found earlier [89], is used, relating the pressure to the vertical
displacements on the base surface. A punch can be subject to an off-centre vertical
load, reduced to a central force and two moments, acting in orthogonal directions.
As a result of the discretization by the punch bottom into rectangular boundary
elements, the unknown averaged contact pressures on each of them as well as the
punch settlement and slopes with respect to the coordinate axes are found from the
solution of a linear algebraic equation system being an analogue of the equality of
the punch and the base vertical displacements as well as three integral equations of
equilibrium. The authors of [186, 187] have succeeded to avoid the difficulties of
the numerical integration of singular functions at the calculations of the influence
matrix coefficients due to the analytical calculations of the terms in the contact
model corresponding to the Boussinesq equation for a unit vertical concentrated
force. Dependences of the punch settlements and slopes on the wedge angle at var-
ious Poisson ratio values ν, relative depth of the compressed mass under the punch
centre H/b, its side length ratio a/b were obtained for three main combinations of
the external loads: at separate action of the vertical force P and each of the moments
Mx and My . Subsequently, based on the results of [14], recommendations were
made for the calculations of the settlements and slopes of rectangular foundations
on a wedge-shaped base [211]. The studies of contact pressure fields while solving
[email protected]
the above problems were not detailed enough [186]. Practically significant cases of
complex-shaped foundations were not considered as well as off-centre loaded with
the bottom uplifting from the base. The latter case is especially important to be taken
4.1 Spatial Contact Problems for Rigid Flat-Bottom Punches 275

into account while designing foundation structures at complicated geotechnical

conditions (on bases with compressible soil mass depth, variable over their area).
Petrukhin [192] proposed a method to calculate non-uniform settlements of
free-standing rigid smooth-bottomed foundations on a nonhomogeneous base of
salinated soils, including the case of the presence of oblique layers. Using the
boundary-element method, for a rectangular foundation contact pressures and set-
tlements are determined as well as slopes in two planes due to the action of a com-
plex external load including a vertical force and an overturning moment of arbitrary
direction. In the proposed approach the total settlement of a base composed of the
salinated soils, is found as the sum of the base settlement prior to the beginning of
the suffusion process (using the scheme of a linearly deformed half-space and the
total deformation modulus of the soil in a natural state), and the base settlement
due to the suffusion, determined by integration over the depth of relative suffu-
sion deformation, depending on the desalination degree and the value of the vertical
stress in the soil. The resolving system of equations of the discrete contact problem
was solved iteratively: at each approximation step the suffusion-related part of the
base deformation at each boundary element is determined according to the experi-
mental curves between the relative suffusion-related compression of the soil and the
vertical pressure applied. The performed boundary-element calculations for a single
shallow square foundation on a nonhomogeneous base (with three layers of soil of
various of salinity level) at the discretization of its bottom even by a rather rough
grid (five elements along each side of the square) produced results which agree well
with the experimental data. This has enabled the author of [192] to propose, after
a required refinement, the boundary-element method of solving the contact prob-
lem for a shallow rigid foundation on a salinated soil for other types of structurally
unstable soils as well.
Dempsey and Li applied the boundary-element method to solve the contact prob-
lem for a rigid centrally loaded rectangular punch on an elastic layer resting on
a non-compressible underlayer base [68]. They used the Burmister solution [60]
as the influence function. A similar problem had been solved earlier by Urisman
[270] without studying the distribution of vertical compressive stress in the base
at the off-centre loading of a rectangular punch located on a finite-thickness elas-
tic layer with the Egorov influence function [75]. Urisman considered the contact
surface of the rectangular punch as a single boundary element, the unknown pres-
sure function on which was presented in the form of a two-dimensional polynomial
of the 12th order, similarly to [105]. The unknown polynomial coefficients were
proposed to be determined by the least-square method. With this purpose, the coor-
dinates of points, belonging to the punch domain, re substituted into the main inte-
gral equation of the problem; the number of these points is taken greater than the
number of the parameters to be determined, thus enabling a redetermined system of
linear algebraic equations to be built. Thus formed redetermined system is solved
by the least-square method, after which the punch displacement parameters and a
[email protected]
polynomial function of reactive pressures are found. Subsequently, using the soft-
ware elaborated by Urisman, extensive calculations of settlements and slopes were
performed for rectangular rigid foundations on a linearly deformed half-space and a
276 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

finite-thickness layer due to a vertical off-centre load of a general type (with eccen-
tricities along the longitudinal and transverse axes). The corresponding graphs were
included into an appendix to a manual for design of plate foundations for frame
constructions and tower-type structures [125].
Using the boundary-element method, Fedorovskiy and Dokhnyanskiy have
solved an axisymmetric contact problem for rigid ring-shaped and circular smooth
punches (foundations) on the base formed by an elastic layer resting on a more
rigid half-space [88]. The domain of the contact of the punch and the base was
meshed into ring-shaped boundary elements of equal area, the pressure on each
of them being assumed constant. In order to calculate the coefficients of the
influence matrix of the resolving system of linear algebraic equations, a repre-
sentation of displacements and stresses in a multilayer elastic medium by inte-
gral Hankel transformation is used in the form, proposed earlier by Fedorovskii
et al. [78]. According to the algorithm elaborated, automated calculations were per-
formed using a high-speed Koltso software. Based on the great scope of multivari-
ant calculations performed using this software, simple formulae (of the Schleicher
formula type) were constructed for the settlements of ring-shaped and circular
foundation punches depending on the punch radii ratio and the deformational
properties of the base. The boundary-element approach to the calculation of the
settlements of ring-shaped foundations and the stress-strained state of their bases,
proposed by Fedorovskii and Dokhnyansky in [88], at the presence of the influ-
ence functions for the layered bases is more general and accurate than the approxi-
mate method of evaluation of settlements of ring-shaped foundations on a combined
base [80]. A comparison of the calculation data with the measurements of the
average settlements of the ring-shaped foundations performed in [80, 88], con-
firmed the application of the solutions of the contact problem of a rigid ring-shaped
punch to be correct for the design of foundations under high funnels. According
to [80, 88], the ring-shaped foundations can be effectively used instead of expen-
sive pile foundations and solid round foundation plates at complicated geotechnical
conditions.
Thus, the review performed enables one to conclude that the abilities of most of
the numerical methods available for solving essentially three-dimensional contact
problems of foundation engineering, problems still remain rather limited.
Finite-difference and finite-element methods enable a considerable success to be
achieved in planar and axisymmetric problems of foundation base calculation. How-
ever, they appear rather ineffective for solving essentially spatial problems even
for elastic linearly stressed bases. Size assignment of the foundation active area,
required in case these methods being used, is an independent and labour-consuming
problem, and in order to solve it one should take into account the mechanical prop-
erties of the base and the load type as well as the problem solution method. The
requirements of a detailed discretization of the base stressed area result in too high
computer RAM size and computation time required. The quantity and size of the
[email protected]
finite elements should be chosen in such a way that the effect of the planes, restrict-
ing the foundation active zone, be reduced to minimum.
4.1 Spatial Contact Problems for Rigid Flat-Bottom Punches 277

The broad application of the finite-element methods for planar and axisymmet-
ric contact problems is caused by the possibility of taking into account the material
nonhomogeneity and anisotropy, complex rheological properties, slippage condi-
tions, etc. at rather high calculation speed. Therefore, the amount of publications
devoted to the FEM applications in soil mechanics and foundation engineering is
extraordinarily high: almost every third paper in geotechnical journals is somehow
related to the FEM [85]. The account of nonlinear regularities, inherent to soils,
is undoubtedly essential, but results in a considerable complication of the contact
problem. The requirements to the computer speed and RAM size increase by sev-
eral orders of magnitude in the case of spatial problems of nonlinear soil mechanics.
This is the reason for the essentially spatial problems having not been considered
using the finite-element method.
Comparison of the solutions, obtained for rigid punches in an axisymmetric for-
mulation by boundary-element and finite-element methods, have shown [198] that
using the boundary-element method provides higher accuracy and working effi-
ciency in the areas of high stress gradients (below the punches where the solution
for the theory of elasticity tends to infinity) even in case piecewise constant approx-
imation of the contact pressure field being used. Besides, the data preparation pro-
cedures (discretization into the boundary elements) are reduced by several orders of
magnitude in comparison with the expenses for discretization in the finite-element
method what finally results in a considerable decrease of the computer resources for
solving the problems of the class under consideration.
Thus, among the numerical methods one of the most promising and efficient
for solving mixed problems of theory of elasticity is the boundary-element method
which successfully competes in a number of aspects with the finite-element method.
The BEM is characterized by an essential reduction of the data preparation proce-
dure and a decrease of the dimensionality of the resolving system of equations. At
present, the BEM has an explicit advantage over the FEM for the problems of the-
ory of elasticity, especially for infinite domains. The boundary-element method to
a considerable degree uses the superposition principle and can be applied to the
problems where the initial differential equations are linear or assume a stepwise lin-
ear approximation with respect to the increments. This has enabled the authors of
the known book regarding the BEM applications [29] to conclude that, in principle,
only a very few problems exist which can be solved by the finite-element method
and which could not be solved with at least the same efficiency solved using the
BEM. The BEM has not been yet studied in such a detail as the FEM, and not all its
possibilities are revealed, both concerning the increase of the calculation accuracy
and the reduction of the computation labour consumption. The known few solutions
of the spatial contact problems for rigid punches, obtained by the boundary-element
method, enable a high efficiency of this method to be predicted in case the elastic
base models with different influence functions being used.
The analysis of the reference data also shows that the difficulties in creation of
[email protected]
universal algorithms of mathematical modelling of processes of interaction of foun-
dations with soil masses are related to the necessity of the simultaneous account
of the geometrical and physical nonlinearities of the soil base, contact friction, and
278 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

slippage effects on the contact surface of the foundation and the soil, search for
the a priori unknown contact domains of complex shape, etc. The FEM computa-
tion schemes, due to the superimposement of various iterative procedures, appear
so cumbersome that they result in a deterioration of the general convergence of
the numerical solution processes and extraordinarily high consumption of computer
resources. Therefore, elaboration of new efficient methods of solving the contact
problems of interaction of foundations with soil, with the account of the complex
shape of their section, essentially spatial loading conditions, close to the real phys-
ical and mechanical properties of soils and deformational characteristics of foun-
dation structures, is still an up-to-date task. In our opinion, combined application
of the boundary-element and finite-element methods should enable more effective
solutions of important and practically significant problems of interaction of founda-
tions with soil in spatial formulation.
Optimal application of advantages of the boundary-element and finite-element
methods appeared rather fruitful for solving the modified problems of contact inter-
action of foundations with soil bases. In this view one should mention a series of
publications of Rozin et al. [217–221], devoted to working out a combined numeri-
cal method of solving the problems for infinite domains of complex structure. The
proposed combined method is based on iterative combination of the boundary-
element and finite-element methods. The iterative process of solving the spatial
problem of interaction of the foundation with the soil is proposed to be built in
the following way [217–221]. At each step the finite-element method is applied in
a finite domain of complex structure (being the foundation together with the area of
plastic deformation of the base) using the Somigliana formula on a boundary sur-
face of the rest of the infinite homogeneous area of the base. All the calculations in
the specified finite domain can be performed using traditional grid methods (FDM,
FEM). The iterations continue until the displacements on a certain finite convo-
lute (auxiliary) surface will differ within the given computation error. Solutions of
a number of test (non-contact) problems for infinite domains, performed by Rozin
et al., have shown higher accuracy and considerable computational benefits in case
the combined method being used, in comparison with the application of the FEM
with truncation of the infinite base domain.
Evidently, the effective application of the combined method for contact inter-
action calculations in the “foundation + soil base” system will require the use of
the boundary-element solutions of contact problems for infinite elastic bases in an
essentially spatial formulation.

4.2 Contact Problems for Rigid Rectangular Punches, Resting

on Elastic Nonhomogeneous Bases
[email protected]
Consider spatial contact problems for rigid rectangular punches, resting on elastic,
in general case spatially nonhomogeneous bases. Elastic solutions. corresponding
to such formulation, are important for the design of extensively used foundations
4.2 Rigid Rectangular Punches, Resting on Elastic Nonhomogeneous Bases 279

with a rectangular bottom shape: under elevators, under columns of buildings and
structures, for equipment, etc.
Let us consider an absolutely rigid punch of a rectangular shape with a flat bottom
to undergo a spatial static load, reduced to a vertical resultant force P and moments
Mx , My (Fig. 4.1). We also assume that vertical displacements of the punch and
the base surface are equal, and there is no tangential stress in the contact plane.
The latter assumption is based on the results of solving similar problems for a half-
plane and a half-space with the account of tangential stress in the contact domain
which have shown that the effect of tangential reactions under the punch can be
neglected while determining the main characteristics of the contact interaction (First
of all, normal forces) [1, 91, 127, 128, 135]. Note that the account of friction forces
in the contact domain is required in a number of machine-building problems, e.g.
the evaluation of wear strength of contacting elements in various mechanisms and
machines, estimation of the probability of fatigue crack nucleation, etc. when the
friction coefficient depends on the acting stress and relative velocity of the bodies
in contact [26, 108].
As noted above, there are no exact solutions for the punches of rectangular cross-
section. Therefore, for a rectangular punch, as well as for the punches of more com-
plex shape, one should apply numerical methods for solving the contact problem.
For numerical solution of the contact problems by the boundary-element method
the contact domain was preliminarily discretized into rectangular boundary ele-
ments. Figure 4.2 shows a scheme of a uniform mesh of a rectangular domain into
96 quadrangular boundary elements. At the calculations of the contact interaction
of punches with the variable-thickness elastic layer the number of elements in the
direction of the OX axis was always equal or exceeding the number of elements
along the OY axes. This is required for the correct account of the contact inter-
action characteristics in the direction of the maximal variation of the elastic layer
thickness. The optimal number of the elements at the calculations was taken as
12 × 8 = 96 what enables one to obtain relatively exact values of integral character-
istics of the contact interaction (the settlements and slopes of the punches as a rigid
body).
Since the contact stresses near the edge (under a full contact with the base) are
of singular character, in order to increase the accuracy of the solution of the contact
problem for a half-space, on the edge boundary elements the known asymptotic was
taken into account [27, 31, 183, 225]:

in the vicinity of the right angle p(r,ϕ) = C1 (ϕ) · r−γ1 , γ1 ≈ 0.7;

at straight sides p(r,ϕ) = C2 (ϕ) · r−γ2 , γ2 ≈ 0.5

where r = 0 is a point on the punch contour.

No similar estimations have been performed for punches on elastic non-classical
bases. Therefore, at numerical calculations for bases of different type, in order to
[email protected]
“soften” the contact pressure field near the punch edge. The rectangular boundary-
element grid was condensed towards the edges of the rectangular punch contact area.
The nodes of the boundary-element grid were chosen at the intercepts of straight
280 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

Fig. 4.2 Uniform

discretization of a rectangular
domain into boundary
elements

lines, located at the distances from the punch symmetry axes, given by a quasiuni-
form dependence

eγ tm − 1 m−1
ρm = ρ0 , tm = , m = 1,2,...,L
eγ − 1 L
where ρ0 is a half of the rectangular punch side length, 2L is the number of nodes in
one of the directions. The condensation degree was controlled by the choice of an
empirical parameter γ . The series of test calculations for test problems have enabled
the value γ = –2 to be chosen, at which the numerical results of the reactive pres-
sure distribution at the full contact of the rectangular punches with the base are
obtained with acceptable accuracy. To calculate the integral characteristics of the
contact interaction (settlements and slopes), a boundary-element grid of the optimal
density 10 × 10 was effectively used (Fig. 4.3a). A quasiuniform grid of rectan-
gular boundary elements, enabling reliable contact pressure fields to be obtained at
complex loadings on elastic bases of different type, is shown in Fig. 4.3b.
At the contact domain discretization for rectangular punches, located in the XOY
plane of arbitrary orientation, as well as for punches of quadrangular cross-section
of general type (not necessarily rectangular), the following transformation was used
[56]:

U = ϕ1 U1 + ϕ2 U2 + ϕ3 U3 + ϕ4 U4

with the interpolating functions

[email protected]
1
ϕ1 = (1 − ξ1 ) (1 − ξ2 ) , ϕ2 =
1
(1 + ξ1 ) (1 − ξ2 ) ,
4 4
1 1
ϕ3 = (1 + ξ1 ) (1 − ξ2 ) , ϕ4 = (1 − ξ1 ) (1 + ξ2 )
4 4
4.2 Rigid Rectangular Punches, Resting on Elastic Nonhomogeneous Bases 281

(a) (b)

Fig. 4.3 Quasiuniform discretization of a square domain into (a) 10 × 10 and (b) 20 × 20 bound-
ary elements (γ = –2)

where Ui (i = 1,4) are the X or Y coordinates of the four angular points of the
contact domain, ξ1 and ξ2 are the coordinates of points of a standard square |ξ1 | ≤ 1,
|ξ2 | ≤ 1.
The subsequent subsections of this section contain the results of numerical stud-
ies of the contact interaction of rectangular punches resting on elastic bases with
non-uniform compressibility.
For an elastic layer of variable thickness, the calculation results for punches,
whose centre is located at a distance xc from the elastic wedge rib x = 0, –∞<y<∞,
z = 0, and loaded by a vertical central or off-centre force P, are analyzed. For the
calculations the following values of the punch loading parameters and the elastic
base characteristics were used:

Mx = My = 0, P = (10 · a)2 E kN, E = 10 MPa.

In all cases, except those directly specified otherwise, the calculations were per-
formed with the Poisson ratio ν = 0.25. The constant vertical load was applied at the
point with the coordinates x = xc + εx , y = yc + εy where (εx , εy ) is the eccentricity
of the force P with respect to the punch centre C(xc , yc ), 2a is the characteristic size
of the punch.
In a similar formulation the contact problem solution for punches of circu-
lar cross-section, resting on an elastic layer of variable thickness, is described in
[email protected]
Appendix B. Settlements and slopes of a rectangular punch on a wedge-shaped base
without the analysis of contact pressure fields and without the account of unilateral
constraints in the contact domain were studied in [187].
282 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

4.2.1 Contact Interaction at Central Loading

Elastically compressible bases of constant and variable thickness. The results of
calculation of parameters of interaction of rectangular punches with an elastic half-
space and layers of variable and constant thickness are given in Figs. 4.4–4.12 and
4.14 and in Tables 4.1–4.3.
Figure 4.4 shows the calculated dependences of contact pressures for a centrally
loaded square punch at different values of the tilt angle α of the underlayer and
fixed values ν = 0.25 and xc = 2a. For the sake of comparison the same figure
contains the contact pressure curve, corresponding to the calculations for an elastic
half-space. The contact pressure values pi , calculated at the solution of the bound-
ary problem in the centres of gravity of the elements located along the OY axes,
were afterwards interpolated using cubic splines [93]. It is seen from the compar-
ison of the calculated dependences, presented in Fig. 4.4, that the pressure field
under the square punch on a variable-thickness layer is essentially asymmetrical.
With the increase of the tilt angle α of the rigid underlayer the difference between
the solution for the half-space and the calculation results for the elastically com-
pressible wedge becomes smaller, this being the consequence of the increase of the
thickness of the compressible layer under the punch. A more detailed pattern of the
non-uniform distribution of the contact pressure under the square punch is seen in
Figs. 4.5 and 4.6 where the equal contact pressure lines are presented, plotted using
the interpolation of functions of two variables. For the square punch on an elastic
half-space the obtained solution is in a qualitative agreement with the one given in
[154]. For this solution the isoline pattern for the contact pressure is symmetrical
with respect to the symmetry axes of the square and its centre (Fig. 4.5), in the
greater part of the contact domain the equal pressure lines being practically indis-
tinguishable from circles. With the increase of the distance form the centre the equal
pressure lines are taking the shape of squares with rounded angles.

[email protected]
Fig. 4.4 Contact pressures in the y = 0 section for a centrally loaded square punch, resting on a
variable-thickness elastic layer (xc = 2a) at α = 30◦ , 45◦ , 60◦ (1–3); (4) elastic half-space
4.2 Rigid Rectangular Punches, Resting on Elastic Nonhomogeneous Bases 283

(a) (b)

(c) (d)

Fig. 4.5 Equal contact pressure linesp/p∗ for a centrally loaded square punch, resting on an elastic
half-space (a) and on a variable-thickness elastic layer (xc = 2a), α = 30◦ (b), 45◦ (c), 60◦ (d)

For a square punch on an elastically compressible layer of variable thickness,

contact pressure isolines are always asymmetrical with the contours displaced from
the centre of the square towards the increase of the layer thickness. An evident con-
densation of the isolines occurs in the same direction. The asymmetry in the pattern
of the equal pressure lines is noticeably manifested in the central part of the punch
with the decrease of the α values. However, with the increase of α, as well as with
approaching the punch contour, the effect of the elastic layer thickness variation
is smaller. As follows from the calculations (Figs. 4.5 and 4.6), in the considered
cases of the contact interaction of a square punch with a non-uniformly compress-
[email protected]
ible base, a rather considerable contact domain exists, in which the pressure is prac-
tically constant. With the decrease of α these domains are distorted, acquiring a
prolate shape along the OY axis, and shift towards the increasing thickness of the
284 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

(a) (b)

(c) (d)

Fig. 4.6 Equal contact pressure linesp/p∗ for a centrally loaded square punch, resting on an elastic
half-space (a) and on a variable-thickness elastic layer (xc = 3a), α = 30◦ (b), 45◦ (c), 60◦ (d)

elastically compressible wedge (normally to its rib). As follows from the calcula-
tions performed, for the contact domains, more remote from the elastic wedge rib, at
a fixed angle α value the described characteristics of the contact interaction are also
revealed, but to a much smaller extent: the domains of constant contact pressure are
of more rounded shape and the displacement from the punch centre is essentially
smaller. Already for α = 60◦ and xc /a = 3 the contact pressure field both quali-
tatively and quantitatively practically does not differ from the corresponding field
for an elastic half-space. Contact pressure field variations with the increase of the
[email protected]
punch distance from the elastic wedge rib at a fixed value of α = 45◦ are illustrated
by Fig. 4.7. As seen from the figure, with the increase of the xc /a values the asym-
metry of the contact pressure profile rather quickly vanishes, and already at xc /a
4.2 Rigid Rectangular Punches, Resting on Elastic Nonhomogeneous Bases 285

Fig. 4.7 Equal contact

pressure linesp/p∗ for a
centrally loaded square punch
in they = 0 section, resting on
a variable-thickness elastic
layer (α= 45◦ ) at xc =a(1),
3a(2),30a (3); (4) elastic
half-space

Fig. 4.8 Dependence of

pressures in the centre of a
square punch on the distance
from the elastic wedge rib (ε x
= εy =0) at α = 30◦ (1), 45◦
(2), 60◦ (3); (4) elastic
half-space

= 30 the contact pressures for a square punch on the half-space and on the elastic
layers (with various values of α) practically coincide.
Figures 4.8–4.11 illustrate the variation of the integral characteristics of the con-
tact interaction of a square punch at different values of xc /a and α: the pressure
pc and the centre settlement Wc , as well as the punch slopes ψ x . As follows from
the calculations, presented in Fig. 4.8, the contact pressures pc in the square punch
centre for xc /a ≤10 are essentially dependent of the values of α, growing with the
increase of the latter. However, for xc /a ≥10 no dependence of the pressure values
c on α is observed, and the corresponding solutions for the half-space can be used
[email protected]
to estimate pc without any essential error.
The punch centre settlements increase with the increase of its distance from the
elastic compressible wedge rib, as well as with the increase of the tilt angle α of
286 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

(b)

(a)

Fig. 4.9 Settlements of the centre (a) and slopes (b) of a centrally loaded square punch versus the
distance from the elastic wedge rib at α = 30◦ (1), 45◦ (2), 60◦ (3); (4) half-space

the lower surface of the distorted base (Fig. 4.9a). With the increase of xc /a the
settlement values unlimitedly approach the punch settlement values on the half-
space what is related to the increase of the compressible layer thickness under the
punch and the corresponding small range of the compressed soil mass thickness
variation.
The influence of the relative distance xc /a and angle α on the slopes at central
loading of the square punch is shown in Fig. 4.9b. As follows from the calculations,
with the increase of xc /a values the punch slope ψx in the interval a≤xc ≤30a rapidly
decreases to zero, the higher values of a corresponding to smaller slopes and their
more rapid decrease. Note that the results of calculations for the centrally loaded
square punch, shown in Figs. 4.8 and 4.9, are in full qualitative agreement with
those we have obtained from the calculations of pressures pc , settlements Wc and
slopes ψx of a centrally loaded round punch (See Appendix B).
Figure 4.10a–c illustrate the quantitative effect of the punch geometrical shape
on the pressures pc , settlements Wc and slopes ψx of the round and square punches,
loaded by equal vertical forces and having√the same contact area. Centrally loaded
punches of circular (with a radius r = a π/2) and square (with the lateral size
2a) shape of equal area were compared. It follows from the obtained calculation
data (Fig. 4.10a) that the round punch centre settlements are always smaller than
[email protected]
those of the square shape, independently of the elastic base type. The larger is the
angle α at the elastically compressible wedge rib, the greater is the difference in the
settlements. It is seen from the plots of Fig. 4.10b that the circular punch slopes are
4.2 Rigid Rectangular Punches, Resting on Elastic Nonhomogeneous Bases 287

(a) (b)

(c)

Fig. 4.10 Settlements of the centre (a), slopes (b) and pressures in the centre (c) of rigid punches
(ε x =εy =0) of equal area, resting on a variable-thickness elastic layer for α = 30◦ (1,2), 60◦ (3,4);
solid line – square punch, dashed line – circular punch

larger, though slightly, than those for the square one of the same area. The absolute
increment of the round punch slopes is the most considerable in the area, adjacent to
the elastic wedge rib. The influence of the punch bottom shape is the most essential
[email protected]
for the contact pressure distribution. The pressures pc in the centres of the square
and the round punches of equal area are plotted in Fig. 4.10c versus the relative
distance of their centres to the elastic wedge rib for the values of α = 30◦ and 60◦ .
288 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

(b)
(a)

Fig. 4.11 Settlements of the centre (a) and slopes (b) of a centrally loaded square punch at ν = 0
(1), 0.25 (2), 0.5 (3). Solid line – a variable-thickness elastic layer (α = 30◦ ), dashed line – elastic
half-space

Fig. 4.12 Distribution of contact pressures in the y = 0 section of a centrally loaded square punch
for constant-thickness Hc =2a (1) and variable-thickness layers at α = 15◦ (2), 30◦ (3), 45◦ (4),
60◦ (5)
[email protected]
4.2 Rigid Rectangular Punches, Resting on Elastic Nonhomogeneous Bases 289

Table 4.1 Characteristics of the contact interaction of a centrally loaded square punch on elastic
bases of different type (ν = 0.25)

Settlement of the Pressure in the punch

punch centre (Wc /a) × centre
Elastic base type 102 Slope ψ x × 10 pc /p∗ , p∗ = P/a2

Half-space 0.4209 0 0.1305

Constant-thickness 0.2672 0 0.1414
layer Hc = 2a
Elastic wedge:
α = 15◦ 0.2495 0.1004 0.1477
α = 30◦ 0.2461 0.2170 0.1464
α = 45◦ 0.2339 0.4003 0.1457
α = 60◦ 0.1989 0.8579 0.1371

Table 4.2 Settlements and slopes of a square punch with a side 2a on a variable-thickness elastic
layer (ν = 0.25; α = 45◦ , xs = 2a) at different orientation angles β in the contact plane

β 0◦ 22.5◦ 45◦

(Wc /a) × 102 0.2339 0.2338 0.2337

ψ x × 10 0.4003 0.4101 0.4197
ψ y × 103 0 0.8958 0

Table 4.3 Characteristics of the contact interaction of rectangular punches with different side ratio
a/b, resting on a variable-thickness elastic layer (xc = 3a, ν = 0.25, α =45◦ )

a/b ψ x × 10 pc /p∗ (Wc /a) × 102

1.0 0.2097 0.1367 0.2840

0.25 0.1586 0.1409 0.2575
4.0 0.2452 0.1477 0.2454

As seen from the calculated plots, due to the concentration of the pressure field
near the angular points, the contact pressures in the central part of the square punch
are essentially lower than in the central part of the circular punch. At α ≥30◦ for
both the circular and the square punches, variation of the contact pressures pc in
the centre occurs only in the area, adjacent to the rib of the elastically compressible
wedge, with the size x÷10a. For xc ≥10a the contact pressures pc in the centre of
both the circular and the square punches are the same as the corresponding values for
the elastic half-space (Fig. 4.10c). Thus, the performed comparative analysis of the
contact interaction of the circular and square punches of equal area with the elastic
layer of variable thickness shows the necessity of a more detailed discretization of
[email protected]
the contact domain towards the decrease of the base depth to be performed at the
design of rigid and flexible structures on elastic bases with variable compressible
thickness.
290 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

The calculated dependences, plotted in Fig. 4.11 show the effect of the Poisson
ratio on the punch centre displacements and slopes with the increase of the dis-
tance from the elastic compressible wedge rib. As seen from the plots, the increase
of ν most essentially affects the settlement of the square punch centre. With the
increase of the punch distance from the rib (x = 0) the corresponding settlement
values asymptotically tend to the values, corresponding to the punch settlements on
the half-space (Fig. 4.11a). Numerous calculations have shown that the increase of
α leads to the size of the domain where the settlement values for the square punch
centre on a variable-thickness elastic layer will be practically indistinguishable for
the Poisson ratio values ν in the whole range of its possible variation. The numerical
calculations have also shown the increase of ν to cause the growth of pc values. Note
that the contact pressure values in the centre of the square punch become practically
equal for the different values of the Poisson ratio ν at essentially lower values of
the relative distance xc /a in comparison with those, for which the levelling of the
relative settlements Wc /a occurs. The dependence of the punch rotation angles ver-
sus the Poisson ratio values is illustrated by the calculated plots, given in Fig. 4.11b
for the angle α = 30◦ . Similarly to the circular punch (Appendix B), at xc ≥ 5a the
square punch slopes are practically independent of the Poisson ratio value.
The obtained numerical values for the parameters of the contact interaction for
the square punch on the elastic layer with a linearly varied thickness were compared
with the calculation data based on the known Egorov solution [75] for a concentrated
normal load on the surface of an elastic layer of a constant thickness H. The depth
of the layer under the square punch centre was fixed (Hc /a= 2), the tilt angles of the
lower surface of the elastic compressible layer of variable thickness were varied.
The analysis of the data presented in Fig. 4.12 and Table 4.1 gives the evidence for
the necessity of the account of the elastic base thickness variation, especially for the
estimation of such parameters of contact interaction of rigid punches as settlements
and slopes.
The above considered calculation data for the square punch correspond to such
its location when one of the sides is parallel to the elastic compressible wedge rib.
In order to investigate the effect of the punch rotation angle in the XOY plane on
the characteristics of its contact interaction, additional calculations were performed,
their results being presented in Table 4.2. An angle β ≤ π/4 between the symmetry
axis of the square and the OX axis is used as a parameter, determining the punch
location on the free surface of the elastic wedge (Fig. 4.13). As it should be expected,

[email protected]
Fig. 4.13 Scheme of a
square punch orientation on
the surface of an elastic
wedge-type base
4.2 Rigid Rectangular Punches, Resting on Elastic Nonhomogeneous Bases 291

Fig. 4.14 Contact pressures in the y = 0 section of rectangular punches with the side ratio a/b: 1.0
(1), 0.25 (2), 4.0 (3), resting on a variable-thickness elastic layer (xc = 3a, α = 45◦ )

the slope ψx is always nonzero, and at β=0◦ , 45◦ one also has ψy = 0. Note that the
ψx slope values are more than by an order of magnitude higher than the ψy slopes.
The relative settlement values Wc /a remain practically unchanged. As follows from
the numerical calculations, for a centrally loaded square punch its rotation at a fixed
value xc = 2a results in the corresponding variation of ψx not more than by 5%.
Therefore, for square punches of arbitrary orientation on a variable-thickness elastic
layer, the values of Wc and ψ x , obtained at β= 0, can be used with a sufficient
accuracy. At different β one should correctly estimate the corresponding slope ψ y
variations.
Consider some results of calculations for punches of rectangular cross-section
shape with different side ratio. The calculation results for centrally loaded rectan-
gular punches of equal area with the side ratio a/b = 0.25, 1, 4 at α = 45◦ , xc /a= 3
are presented in Fig. 4.14 and Table 4.3. For the discretization of the contact domain
grids of 100 square boundary elements arranged as 4 × 25, 10 × 10, 25 × 4, respec-
tively, were applied. Such system of the contact domain discretization has enabled
the contact interaction to be reproduced at the numerical calculations with minimal
errors. The relative settlement Wc /a of a rectangular punch is always smaller than for
the square punch of the same area (Table 4.3). On the other hand, relative pressures
pc /p∗ in the rectangular punch centre will be higher than for the square punch. The
calculations have shown the relative pressure pc /p∗ in the centre of the rectangular
punch with longer dimension along the rib (a/b = 0.25) of the elastic compressible
wedge, to be by 4.6% lower than for the rectangular punch with longer dimension
(a/b = 4) along the OX axis. The influence of the a/b ratio is the most essential for
the angular displacements ψx of rectangular punches. For a rectangular punch with
the side ratio a/b = 0.25 the slope ψ x decreases by 24.3% with respect to the slope
of the square punch. Simultaneously, for a rectangular punch with longer dimension
in the direction of the elastic layer depth increase (a/b = 4), its slope ψ x increases
[email protected]
by 17% in comparison with that of the square punch. The calculations for rectangu-
lar punches have also shown that, contrary to the square punch, the slope values ψ x
and ψ y are comparable and strongly depend on the a/b ratio at different values of
the angle β.
292 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

Elastic half-space with an increasing deformation modulus. Consider a spatial

contact problem for a rigid rectangular punch resting on an elastic nonhomoge-
neous base with an increasing deformation modulus, loaded by a central force
P. Fundamental solutions ω(n) (x–ξ , y–η) for a nonhomogeneous base with linear
(n=1) and parabolic (n=2) dependences of the deformation modulus on the depth
(Sect. 1.5.3) are used as influence functions. The Poisson ratio, as a rule, varies only
slightly, its value does not produce any essential effect on the stress-strained state
characteristics and hereinafter is assumed constant. Since there is a known relation-
ship between the elastic constants G = E/2(1+ν), at such consideration the shear
modulus G = G(z) will be variable with depth.
The main regularities of the contact interaction can be traced from the calcu-
lation results, obtained for a square punch (a/b = 1). In Table 4.4 the calculation
data for the settlements W0 /a and contact pressures in the centre p0 /pav (pav =
P/S being the average pressure over the bottom, S = 4ab) illustrate convergence
of the numerical solutions with the increase of the discretization degree for a punch
located on an elastic half-space (B = 0) and elastic non-classical bases with differ-
ent nonhomogeneity degree (B = 5, n = 1, 2). The upper values in the table cells
correspond to dimensionless pressures, the lower ones – to dimensionless settle-
ments. The calculations were carried out for the values n = 0.25, P = E0 × a2 ,
and μ = h/a= 0.5. The same table contains the corresponding extrapolated values
obtained using the Richardson method in the assumption of the second order of
accuracy of the numerical method being used. As follows from Table 4.4, with the
increase of the discretization degree a stable increase of the accuracy of the obtained
results is observed. On a 20 × 20 grid the relative error of the numerical solutions
does not exceed 1.5% for the settlements and 2% for the contact pressures what is
quite sufficient for practical calculations.
Contact pressure p = p(x, y)/pav isolines for the punch on an elastic homogeneous
half-space (B = 0) and on elastic punches with different degrees of nonhomogene-
ity of the deformational properties (B = 1, 5, n = 1, 2) are plotted in dimensionless
coordinates x = x/a, y = y/a in Fig. 4.15. In all cases the contact pressure distribu-
tion patterns are qualitatively similar: in the punch bottom centre a constant-pressure

Table 4.4 Studies of convergence of numerical solutions of the contact problem of a square punch
on nonhomogeneous elastic bases

N x = Ny
Base Extrapolation
parameters 8 12 16 20 24 values

0.5296 0.5144 0.5069 0.5025 0.4996 0.4931

B=0
0.4236 0.4181 0.4153 0.4137 0.4125 0.4099
B = 5. 0.6026 0.5844 0.5758 0.5707 0.5674 0.5598
n=1
[email protected]
0.2599 0.2557 0.2536 0.2524 0.2516 0.2498
B = 5. 0.8118 0.7957 0.7873 0.7820 0.7785 0.7704
n=2 0.114 0.1093 0.1082 0.1075 0.1071 0.1060
4.2 Rigid Rectangular Punches, Resting on Elastic Nonhomogeneous Bases 293

Fig. 4.15 Contact pressure p/p∗ isolines for a square punch on the bases of different nonhomo-
geneity degree atμ=1: (a)B = 0; (b)n = 1, B = 1; (c) n = 2, B = 1; (d) n = 2, B = 5

area with a nearly circular shape exists, with the increase of the distance from the
centre the isobars from the circles are transformed into symmetrical ovals, and in the
vicinity of the foundation edges they take the shape of similar squares with rounded
corners. For both contact models of the elastic nonhomogeneous base an unlimited
growth of the contact pressures occurs at the foundation boundary contour. Simul-
taneously, as seen from the calculation data presented, the contact pressure field is
rather sensitive with respect to the distributive properties of the base, determined
by the nonhomogeneity parameters. The circular central constant-pressure area has
the minimal size in the homogeneous half-space case (Fig. 4.15a). For nonhomoge-
neous bases the area in question broadens with the increase of both the n parameter
(Fig. 4.15b, c), and the B parameter (Fig. 4.15c, d). Simultaneously the values of
the contact pressures themselves increase for these areas as well (Figs. 4.15b–d and
Fig. 4.16a). Thus, with the increase of the nonhomogeneity parameters, due to the
variation of the distributive properties, for different contact models of nonhomoge-
neous bases redistribution (levelling) of the reactive pressures in the contact domain
occurs with the pressure decrease near the boundaries and increase in the central
area.
The relative settlements W0 /a of the punch are plotted across the values of the
deformation modulus B variation range on nonhomogeneous bases of different type
(n = 1, 2) at μ = 0.5, 1, and 2 in Fig. 4.16b. It is seen from the figure that an increase
[email protected]
of the nonhomogeneity parameter B within one order of magnitude can result in the
punch settlement decrease more than by factor of 5 what is related to the relax-
ation of the base distributive properties with the increase of their nonhomogeneity
294 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

Fig. 4.16 Contact pressures (a) and settlements (b) in the centre of a square punch on nonhomo-
geneous bases with increasing deformation modulus

degree. A rather considerable (≈70%) settlement decrease occurs already in the

case when at the depth of half the punch side width (μ = 1) the deformation
modulus increases not more than twice (B ≤ 1). Besides, numerous calculations have
shown that at a fixed μ the above noted character of the settlement decrease is essen-
tially affected by the type of the base nonhomogeneity: at the linear increase of the
deformation modulus with depth (n = 1) the settlements are always larger than at
the parabolic variation (n = 2). Depending on the deformation modulus variation
law employed, the difference in the settlement determination can reach up to 45%.
It should be noted that the account of nonhomogeneity, inherent to real soil bases,
is rather important for the evaluation of internal forces in complex-shaped foun-
dations since it enables the above considered trend to the contact pressure field
[email protected]
levelling to be estimated at the design stage. The more uniform pressure over the
foundation bottom than in the case of the homogeneous base will be helpful for the
decrease of the slopes and levelling of the settlements.
4.2 Rigid Rectangular Punches, Resting on Elastic Nonhomogeneous Bases 295

4.2.2 Contact Interaction at Off-Centre Loading with the Account

of Unilateral Constraints
Multivariant numerical studies have shown that at certain conditions related to the
action of external moments or an eccentric load on a punch of a regular geometrical
shape as well as at contact interaction of complex-shaped punches with elastic bases
of half-space type, constant- or variable-thickness type, negative pressure zones can
arise [13, 16, 17, 76, 225]. Since the soil medium does not work for tension, then,
for a correct description of the real pattern of interaction of foundations with the
base a unilateral character of constraints in the contact domain should be introduced
into consideration.
Consider the results of calculations for off-centre loaded rectangular punches.
Equal pressure lines p = p/p∗ (hereinafter p∗ = P/a2 ) in the contact domain of a
square punch under an eccentric load on an elastic wedge for α = 45◦ , xc =3a are
plotted in Fig. 4.17. As seen from the calculations performed, the off-centre punch
loading results in the formation of areas with rather low contact pressure (Fig. 4.17a)
or even negative pressure (Fig. 4.17b) in the contact domain. In accordance with the
unilateral constraint principle, in the presence of negative (tensile) stress the calcu-
lation was performed iteratively, excluding the boundary elements, for which pi <0.
A large series of the calculations performed has shown the iterative process to be,
as a rule, convergent even at considerable eccentricities of the vertical load applied
and the maximum iteration number not to exceed eight. By interpolation of the con-
tact pressure function the contact zone configurations in the presence of the punch
uplifting are plotted in Fig. 4.18 and the contact domain variation for the square
punch loaded by an eccentric force for εx = 0, εy = 0.7, xc = 2a, ν = 0.25, α = 45◦
in the course of the iterative process is shown. After the iteration process being

(a) (b)

[email protected]
Fig. 4.17 Equal contact pressure lines p/p∗ at off-centre loading of a square punch, resting on a
variable-thickness elastic layer (α = 45◦ , xc = 3a): (a) (εx = – 0.1a, εy = 0.3a), (b) (εx = –0.2a;
εy = 0.4a)
296 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

Fig. 4.18 Contact domain

variation at off-centre loading
(εx = 0; εy = 0.7a) of a square
punch, resting on a
variable-thickness elastic
layer (α= 45◦ , xc = 2a) in the
course of iterations,
N = 1,2,3,6

finished, when all the values pi ≥0, the zone of the punch uplifting from the elastic
base surface and, accordingly, the contact zone are determined. With the increase
of the distance of the load application point from the rectangular punch centre the
uplifting zone covers broader area (Fig. 4.19). The analysis of the great amount of
numerical calculations performed shows that at εx , εy ≥0.8a a square punch under-
goes a sharp increase of slopes what finally results in its overturn. In the case when
εy = 0, the contact area of a rectangular punch with its side being parallel to the
elastic wedge rib, is symmetrical with respect to the OX axis at any values εx = 0.
At εx = 0, εy = 0 simultaneously as well as in the case when εx = 0, due to the
variable thickness of the elastically compressed layer the area of contact of the rect-
angular with the base will be asymmetrical at any values εy = 0 (Fig. 4.19). The
analysis of the contact area shape has shown the asymmetry degree to increase with
eccentricity εy .
Figures 4.20 and 4.21 show the results of calculations of the settlements of the
centre of a square punch and its slopes for various cases of off-centre loading with
and without the account of unilateral constraints in the contact area. The distance
of the punch centre from the elastic wedge rib xc = 2a and the Poisson ratio of
the stressed base ν = 0.25 were considered constant. As follows from the cal-
culation performed (Fig. 4.20), the dependences of the rectangular punch centre
settlements on εx (at εy = 0) and on εy (at εx = 0) are strictly linear what fully
corresponds to the superposition principle in linear theory of elasticity. Similarly
to the case of a round punch (Appendix B), at εy = 0, εx = 0 (Fig. 4.20a), the
angular coefficients for the corresponding straight-line dependences of the variable-
thickness layer are obviously nonzero and weakly depend on the angle α. It should
be noted that at εx = 0 for an elastically compressed variable-thickness layer as
well as for a half-space (Fig. 4.20a, b) the settlement of the punch centre does not
[email protected]
depend on εy .
The results of the calculations performed show that an approach without the
account of unilateral constraints does not enable one to describe the rigid punch
4.2 Rigid Rectangular Punches, Resting on Elastic Nonhomogeneous Bases 297

Fig. 4.19 Contact domains at off-centre loading (ε x =0, ε y /a= 0.5, 0.6, 0.7, 0.8) of a square punch,
resting on a variable-thickness elastic layer (α = 45◦ , xc = 2a)

uplifting from an elastic base and the unlimited increase of its slope with the exter-
nal force application point approaching the punch boundary. These shortcomings
of the contact problem solution in the classical (linear) formulation are corrected
by the account of the unilateral character of constraints in the contact area. From
the calculations performed it follows (Fig. 4.20) that when the zones of the punch
uplifting form the elastic base surface arise and the contact area decreases, a non-
linear dependence of the slopes ψx , ψy and settlements Wc of the rectangular punch
centre on the external force eccentricity and, hence, on the external overturning
moments, is observed. The structure-type nonlinearity, being revealed in such a way
[email protected]
at the contact interaction, results in the inapplicability of the superposition princi-
ple. The calculations, performed for a square punch on a variable-thickness elastic
layer (xc = 2a), have shown (Fig. 4.20) that on the ranges 30◦ ≤ α ≤ 60◦ and
298 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

(b)
(a)

Fig. 4.20 Settlements of a centre of a square punch, resting on a variable-thickness elastic layer
(xc =2a) depending on the force eccentricity: (a) εx (εy = 0), (b) ε y (ε x = 0) at α = 30◦ (1), 45◦
(2), 60◦ (3); (4) elastic half-space. Solid line – with the account of uplifting of the punch from the
base, dashed line – without the account of the punch uplifting

–0.5×a≤ εx ≤ 0.4×a, |εx |≤ 0.4×a the contact interaction of the punch and the elas-
tic base with the variable-thickness compressible soil mass goes without uplifting
and, hence, the application of the superposition principle is correct.
Thus, it has been clearly shown with particular examples, that for rigid punches,
using numerical calculations, one can determine the section kernel, i.e. the area of
the vertical load application, for which the punch uplifting from the base does not
arise. The problem of determination of the section kernel boundaries for rectan-
gular punches at different values of a/b, α, xc , ν, and β requires extensive special
numerical calculations. The issues related to the calculation of the section kernel
boundaries for foundation plates of non-canonical shape, often used for the design
of new and renovated buildings, are considered systematically in Sect. 4.5. Here we
only note that, contrary to the approximate estimation of the section kernel dimen-
sions of rigid rectangular plates on an elastic half-space from the pressures or set-
tlements [105], the numerical approach with the account of unilateral constraints
proposed here enables reliable automatic determination of the section kernel bound-
aries for punches of a given shape resting on elastic bases of various type (constant-
or variable-thickness layers, layered elastic bases, half-space with variable physical
properties, etc.).
The analysis of the numerical calculations performed for the contact problems
of off-centre loading of rectangular punches has shown that, similarly to the case
[email protected]
of central loading, independently of the eccentricity of the impressing vertical force
application, the settlements of the centre of the punch resting on an elastic half-
space, also exceed the corresponding settlement values for the variable-thickness
4.2 Rigid Rectangular Punches, Resting on Elastic Nonhomogeneous Bases 299

(a)

(b)

Fig. 4.21 Slopes of a square punch, resting on a variable-thickness elastic layer (xc = 2a), versus
the force eccentricity: (a) ε x (ε y = 0), (b) ε y (ε x = 0) at α = 30◦ (1), 45◦ (2), 60◦ (3); (4) half-space

bases with various α angle values. It follows from the calculated curves, plotted
in Fig. 4.20a, that the dependences of the settlements of the centre of the square
punch on the angle α at fixed values of eccentricity of the external force applica-
tion point are monotonous and qualitatively similar. The square punch slope also
monotonously depend on the eccentricity values (Fig. 4.21), but do not possess such
a pronounced dependence on the angle α, as the settlements do. At full contact of
the punch with the base (when the solutions with and without account of unilateral
constraint coincide), the superposition principle holds and, as one should expect,
the dependences of ψx and ψy on the eccentricity of the vertical resultant are practi-
cally linear. It is important to note that the dependences of slopes, corresponding to
[email protected]
the off-centre loading of rectangular punches for εy = 0 (Fig. 4.21a), enable one to
determine rather exactly (even graphically) the values of eccentricities εx , at which
the punches do not undergo any slope (ψx = 0), i.e. have a uniform settlement.
300 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

The issues concerning providing uniform settlement of punches of various cross-

section, resting on non-uniformly compressed bases, are considered in detail in
Sect. 4.3.
With the beginning of formation of the zones of the square punch uplifting from
the elastic base, the absolute values of the punch slopes vary nonlinearly, sharply
increasing at |ε x |, |ε y |≥ 0.6, what finally results in the punch overturning. As follows
from the calculations, performed for εy = 0 (Fig. 4.21a), the effect of the angle α at
the vertex of the elastically compressible variable-thickness layer at εx < 0 is more
essential than for εx > 0. In the case when εx = 0, the effect of the angle α on the
dependence of the slopes ψ y on εy is insignificant (Fig. 4.21b).
Thus, the numerical approach developed here enables the distribution of contact
pressure over the foundation bottom to be studied in detail without any specula-
tions or assumptions about the contact pressure character under external loading of
spatial type. The numerical boundary-element algorithm being used is rather effec-
tive for solving spatial contact problems of essentially oblique punches when the
area of the punch uplifting from the base is comparable or even larger than the
contact area. Multivariant calculations have enabled us to find out that the account
of spatial nonhomogeneity in the framework of the existing elastic base models or
those proposed in Chap. 1, results in a considerably lower distributivity of the soil,
essential reduction of the foundation settlements, redistribution of reactive pressures
under their bottom in comparison with the most widely used homogeneous elastic
half-space model. Since all real soils possess a rather pronounced nonhomogeneity
of their elastic properties, it becomes possible to describe the contact interaction
processes more accurately and substantiate the decrease of the calculated forces in
shallow foundations and, hence, provide more economical use of materials at their
construction.

4.3 Control of the Parameters of Loading and Shape to Provide

a Uniform Settlement of Rigid Foundation Plates
The problem of providing a uniform settlement of punches (models of rigid founda-
tions) under a vertical load is of great practical interest for foundation engineering.
Formation and development of foundation slopes results in the appearance of
cracks and irreversible deformations of breakdown character in structures above the
foundation. Special measures for the foundation reconstruction and reinforcement,
required in such cases, are related to additional materials and labour expenses.
For homogeneous bases, the main reason for a slope in foundations with two
or more symmetry axes is an off-centre load application. The presence of a single
symmetry axis or a totally asymmetrical shape is another reason for the appearance
of a foundation slope.
[email protected]
Non-uniform base compressibility on a relatively small area of a building or a
structure, typical for a number of construction sites [86], results in slopes appearing
for the foundations of symmetrical shape even at central loading. There are a lot of
4.3 Control of the Parameters of Loading and Shape 301

references in the literature about breakdowns in the course of construction or func-

tioning of objects of industrial and civil engineering, resulting from non-uniform
deformations of soil bases (See, e.g. [90]).
The studies [122, 187] as well as the calculations we have performed for rectan-
gular and round punches (Sect. 4.2, Appendix C) show that the variable thickness
of the compressible layer in the base is the reason for an essentially non-uniform
settlement and slopes of foundations, untolerable in most cases for engineering con-
structions.
In order to prevent dangerous foundation slopes, exceeding the limiting values
(set by the regulations for the design of bases for building and structures), and to
get rid of them, as a rule, the size of the foundation bottom is increased. In case of
buildings being erected on non-uniformly compressible soils, an approach to elimi-
nate slopes and level settlements by introduction of compensating systems into the
base is also applied [211]. Note that both methods, depending on the soil conditions,
result in an increase of the construction expenses.
Below a theoretical substantiation and numerical confirmation is given for two
possible approaches to prevent punch slopes by controlling their shape and loading
parameters.
The essence of the first approach is that, for a given punch shape and size, a
location of the resultant vertical load is determined, for which the punch settlement
occurs without a slope. The advantage of such approach consists in the fact that the
problem solution always exists and its search does not require iterative processes
for a complex shape of the punch. However, not always the solution found can have
practical value since the resultant external load can appear too close to the punch
edge or even beyond its boundary.
In the second approach, the punch bottom area and the load application point are
fixed, and then a search for the punch bottom shape, providing its uniform settle-
ment, is carried out. For the shape search, using a control parameter, a multipara-
metric family of contours of a certain type, bounding the punch, is given. For each
punch shape a direct problem is solved, and then, by approximation of the solu-
tions obtained, the control parameter value is determined, for which the punch slope
is zero. This approach does not guarantee the uniqueness of the solution, requires
time-consuming iterative procedures, but in most cases the results obtained are use-
ful from the practical point of view.
For both approaches the area of the punch contact with the base remains
unchanged and, hence, the corresponding practical implementation will not require
any additional consumption of materials for the foundation.

4.3.1 Formulation of the Problem and Its Numerical

Implementation
[email protected]
In both approaches the contact problem is solved by direct boundary-element
method, using a piecewise constant approximation of the contact pressures
302 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

(Sect. 2.5.1). In the first approach the resolving equation system is given by
⎧ n  
⎪ '
⎪
⎪ pj · δij − Wc = 0, i = 1, n ,
⎪
⎪
⎪
⎪ j=1
⎪
⎪ 'n
⎪
⎪ pj · sj = P,
⎪
⎪
⎨ j=1
'n
(4.1)
⎪
⎪ pj · sj · xj + P · εx = P · xc ,
⎪
⎪
⎪
⎪
j=1
⎪
⎪ 'n
⎪
⎪ pj · sj · yj + P · εy = P · yc ,
⎪
⎪
⎪
⎩ j=1
pj ≥ 0

where the unknowns are contact pressures pj within the boundary elements, the
punch settlement Wc and eccentricities εx , εy of application of the resultant external
load P with respect to a fixed punch point with the coordinates (xc , yc ). The amount
of the boundary elements in general is n, the j-th element area is sj , and the coordi-
nates of its centre of gravity are (xj , yj ). The coefficients of the canonical equations
are calculated from
 
δij = ω (xi ,yi ,ξ ,η) dξ dη
Fj

where ω(xi , yi , ξ, η) is the settlement of the point (ξ, η) within the domain of the
j-th boundary element Fj due to a unit vertical force, applied to the base surface
at the point (xi , yi ), coinciding with the i-th element gravity centre. The function
ω(x, y, ξ , η) is given in accordance with the elastic base model being used. Note that
the δ ij coefficients characterize the foundation shape.
While the system (4.1) is being solved, the eccentricities εx and ε y play the role
load control parameters. The fact that they are contained explicitly in the resolving
linear equation system, essentially simplifies their calculation. For this purpose it is
sufficient to solve the system (4.1) by any known method of linear algebra.
The second approach is based on the study of a somewhat different resolving
equation system which can be written in the following way:
⎧ n  
⎪ '
⎪
⎪ pj · δij − ψx · (xi − xc ) − ψy · (yi − yc ) − Wc = 0 i = 1, n ,
⎪
⎪
⎪
⎪ j=1
⎪'
⎪ n
⎪
⎪ pj · sj = P,
⎪
⎪
⎨ j=1
'n
(4.2)
⎪
⎪ pj · sj · xj = −P · xc − My ,
⎪
⎪
⎪ j=1
[email protected]
⎪
⎪
⎪ 'n
⎪
⎪ pj · sj · yj = P · yc + Mx ,
⎪
⎪
⎪
⎩ j=1
pj ≥ 0.
4.3 Control of the Parameters of Loading and Shape 303

Here the same notations are used as those in Eq. (4.1). The unknowns are the
contact pressures pj , the punch slopes ψ x and ψ y and its settlement Wc due to the
vertical force P and overturning moments Mx and My applied at the point (xc , yc ).
The parameters, controlling the punch shape, are not contained explicitly in the
equation system (4.2) what essentially encumbers the search for their optimal values
corresponding to the condition ψ x = ψ y =0. Therefore, for solving the problem in
question the following algorithm is proposed. At the first stage the control parame-
ters (u, v) are chosen, enabling the punch shape to be intentionally distorted along
the main directions of variability of the base deformational properties. Then speci-
fied numerical values are assigned to the parameters with a step. These values cor-
respond to the punches with different cross-section shape, but with the same area.
For each punch shape the system (4.2) is solved with new coefficients. As a result,
sets of tabulated functions ψ x (u, v), ψ y (u, v) are obtained.
From a combined consideration of equations
6
ψx (u,v) = 0,
(4.3)
ψy (u,v) = 0,

the control parameter values can be found, resulting in such shape of the contact
domain, at which the punch will have a zero slope.
The both approaches discussed were tested for solving problems for punches
of various shape on a non-uniformly compressible base in the form of an elastic
isotropic wedge with a pinched lower surface [89].

4.3.2 External Load Control

The solutions, proposed for consideration, were obtained for the punches of square,
L-shaped and round shape. For each of the problems we determined the coordinates
of the point, at which the resultant vertical external load should be applied in order
to obtain a zero punch slope.
Square punch on an elastic wedge. The calculation scheme, used for this prob-
lem, is shown in Fig. 4.22. The numerical solutions were obtained at the wedge
angles α = 30◦ and 60◦ and a relative distance of the punch centre from the wedge
rib 2 ≤ xc /a ≤ 20 where a is a half of the punch side. The wedge-shaped base
was characterized by the deformation modulus E = 10 MPa and the Poisson ratio
ν = 0.25. The calculations were preformed at a vertical load P = 100 kN, the punch
side 2a = 2 m. In order to describe the domain of the punch contact with the base,
a uniform 12 × 8 boundary-element grid was used. The grid density along the OX
axis is taken by factor of 1.5 higher than along the OY axis, following the direction
of the greatest variation of the compressible soil mass thickness. The total amount of
the unknowns in the problem was 99. The solution results in a dimensionless form
[email protected]
are given in Table 4.5.
It follows from the table that in the whole range of variation of the geometrical
parameters α and xc /a under investigation, the calculated eccentricity εx of the
304 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

Fig. 4.22 Calculation

scheme to determine the
external load eccentricity,
corresponding to the uniform
settlement of a square punch

Table 4.5 Eccentricity ε x of the resultant external vertical load, providing a uniform settlement
Wc of a square punch on a wedge-shaped base

εx /a (Wc /a) × 102

xc /a α=30◦ α=60◦ α=30◦ α=60◦

2 0.11579 0.72108 × 10–1 0.16996 0.27540

3 0.60385 × 10−1 0.33134 × 10−1 0.22605 0.32092
5 0.25334 × 10−1 0.12219 × 10−1 0.28937 0.35979
8 0.10534 × 10−1 0.4694 × 10−2 0.33444 0.38245
12 0.45538 × 10−2 0.1791 × 10−2 0.36219 0.39521
20 0.17548 × 10−2 0.76294 × 10−3 0.38536 0.40547

external load P, for which the punch slope will be zero, does not extend beyond
the punch section kernel, i.e. a condition |ε|< ρ holds, where ρ is the radius of the
section kernel of a circle, inscribed into the square, ρ = a/3. Besides, for xc /a ≥ 3
an inequality εx /a < 7ε holds, where 7ε = a/15 is the value of a random relative load
eccentricity, usually taken at the foundation structure calculations.
The increase of xc /a and α parameters results in a regular decrease of the required
displacement value ε x of the application point of the force P with respect to the
punch centre. With a double increase of α (from 30◦ to 60◦ ) the ε x /a value decreases
almost by the same factor. The increase of the distance xc /a by factor of two results
in a more pronounced decrease of the εx /a parameter – up to 5–6 times. Thus,
the distance of the punch from the wedge rib affects the punch slope value much
stronger than the wedge base angle α.
Regarding the punch settlement Wc /a, one should mention that at xc /a ≥ 20 and
α = 30◦ ÷60◦ it is practically comparable with the settlement value for punch on an
elastic half-space Wc /a = 0.0042. As follows from the calculations performed, the
discrepancy does not exceed 10%.
L-shaped punch on an elastic compressible wedge and on an elastic half-space.
The specific feature of the problem under consideration is the fact that even in the
case of a homogeneous base and if the resultant of external vertical load is applied
[email protected]
exactly in the gravity centre of the L-shaped punch, this will anyway result in its
slope. This circumstance is due to the incomplete symmetry of the punch or its
absence at all.
4.3 Control of the Parameters of Loading and Shape 305

The contour of the punch, considered in the numerical experiments, is shown at

the calculation scheme, given in Fig. 4.23. For building up the boundary-element
grid the domain of the punch contact with the base was uniformly meshed into 72
rectangular boundary elements with sides of 0.167a along the OX axis and 0.25a
along the OY axis, a being the characteristic size of the punch (Fig. 4.24). Two base
models were used for the calculations: a linearly deformable homogeneous isotropic
half-space and a linearly deformable homogeneous isotropic wedge with the vertex
angle α= 45◦ . The distance from the coordinate system origin OXY to the wedge
rib, as shown in Fig. 4.23, was chosen as L = 2a. In both cases the Poisson ratio ν=
0.25, the deformation modulus E = 10 MPa, the resultant vertical load P = 100 kN.
The results obtained are shown in a dimensionless form in Table 4.6.
The analysis of the table data enables one to conclude on a noticeable difference
of the calculated eccentricity values ε x , εy from the corresponding distances to the

Fig. 4.23 Calculation

scheme to determine the
external load eccentricity,
corresponding to the uniform
settlement of an L-shaped
punch

Fig. 4.24 Contact domain

discretization for an L-shaped
punch

Table 4.6 Contact interaction characteristics and eccentricities ε x , ε y of the resultant external
vertical load, providing a uniform settlement Wc of an L-shaped punch on elastic bases

Base type |ε x |/a |ε y |/a ψ x × 10 ψ y × 10 (Wc /a) × 102

[email protected]
0.1472 0.1472 0 0 0.4608
Elastic half-space
0 0 0.1281 0.1281 0.4986
◦ 0.2316 0.1265 0 0 0.2554
Elastic wedge, α=45 , xc =2a
0 0 0.1617 0.1165 0.3075
306 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

Fig. 4.25 Dimensionless contact pressures along the perimeter of an L-shaped punch with a uni-
form settlement at xc =2a, α= 45◦ : (a) on an elastic half-space and (b) on an elastic wedge-shaped
base

punch gravity centre xc = yc = a/6 = 0.1667a. Note that for the half-space the value

(εx )2 + (εy )2 does not exceed the random eccentricity value 7
ε = a/15.
Note that for the input data assumed the punch settlement on the half-space
obtained from the calculation was almost twice the punch settlement on the wedge-
shaped base.
Figure 4.25a,b illustrates the contact pressure profiles, plotted from the values
in the boundary-element gravity centres for the elastic-half-space and the elastic
wedge, respectively. The contact pressure values are given in the dimensionless form
p = pa2 /P. In both cases the punch slope is zero. It should be noted that since both
models of the base are elastic, then in the angular points and near the punch edges
the stress condensation occurs. The base model noticeably affects qualitatively on
the character of the contact pressure distribution, and quantitative differences of
stress values within similar boundary elements reach 30%.
Round punch on an elastic wedge. Consider some numerical results, obtained for
a round punch. The calculation scheme, used for the problem, is shown in Fig. 4.26.
The following parameter values, characterizing the elastic wedge, were assumed: the
wedge angle α = 15◦ , deformation modulus E = 10 MPa, Poisson ratio ν = 0.35.
The vertical load on the punch was P = 103 kN. The punch surface was meshed
into 96 boundary elements by equidistant concentric circles and rays, centred in the
punch centre. The distance of the punch centre from the wedge rib was assumed
L = 2r where r is the punch radius.
With the parameter values assumed, the contact problem solution was obtained,
[email protected]
according to which the relative eccentricity value for the vertical load P, at which
the punch slope is zero, should be εx /r =0.111. In this case the contact pressure
profile in the XOZ plane will have the shape shown in Fig. 4.27. In spite of the
profile being essentially asymmetrical, the punch settlement is uniform.
4.3 Control of the Parameters of Loading and Shape 307

(a) (b)

Fig. 4.26 Calculation scheme to determine the external load eccentricity, corresponding to a uni-
form settlement of a circular punch: (a) vertical section, (b) plan

Fig. 4.27 Contact pressure

under a round punch with a
uniform settlement on an
elastic wedge-shaped base
(α= 15◦ , xc = 2r, ν= 0.35)

4.3.3 Shape Parameter Control

Below we consider solutions, illustrating how a uniform settlement of a punch on
a nonhomogeneous with depth base at a given external load can be achieved by the
punch shape control.
Trapezoidal punch on an elastic wedge. The calculation scheme for this problem
is shown in Fig. 4.28. A square punch with a side 2awas taken as the initial configu-
ration. In the course of the numerical studies its shape was intentionally distorted by
a control parameter d/a. The control parameter was varied within 0≤d/a≤0.5. Along
with the shape, the boundary-element grid was also distorted, the total number of
the boundary elements was taken 96 (Fig. 4.29).
The wedge angle was α = 45◦ . The base parameter was the following: Poisson
[email protected]
ratio ν = 0.25, deformation modulus E = 10 MPa. The distance from the application
point of the resultant vertical load P= 100 kN to the wedge rib was taken as xc =2a.
The calculation results are presented in Table 4.7.
308 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

Fig. 4.28 Calculation

scheme to determine the
optimal size of a trapezoidal
punch with a uniform
settlement on a wedge-shaped
elastic base

Fig. 4.29 Contact domain

dicretization for a trapezoidal
punch

Table 4.7 Slopes ψx and relative settlements Wc /a of a trapezoidal punch on a wedge-shaped

base depending on the control parameter d/a values

d/a 0 0.1 0.237425 0.3 0.5

ψ x ×10 0.40033 0.23431 –0.04968 –0.10667 –0.47326
(Wc /a) × 102 0.23395 0.23365 0.22737 0.23560 0.24190

Processing of the data of Table 4.7 by least-square method with subsequent

numerical testing of the square approximation results enables us to conclude that
the absence of the punch slope (ψ x = 0) corresponds to the control parame-
ter value d/a = 0.2133. In this case the relative settlement of the foundation is
Wc /a = 0.22692·× 10−2 .
It is known that at oblique position of the underlayer of a more dense soil the
expected non-uniform settlements of a foundation can be reduced to the maximal
tolerable value by broadening its bottom in the direction of the weak soil layer
thickening [265, p. 93, Fig. III-26b]. The consideration of the contact problem for a
trapezoidal punch performed here served as a calculation model enabling the foun-
dation geometrical parameters, its settlement and contact stress over its bottom to be
determined depending on the deformational and geometrical characteristics of the
elastic layer [19].
[email protected]
Ring-shaped punch on a wedge base. The calculation scheme for this problem is
shown in Fig. 4.30. The absence of the punch slope is achieved by shifting the inner
circle of the ring by a value ε= ε0 which serves as a control parameter.
4.3 Control of the Parameters of Loading and Shape 309

(a) (b)

Fig. 4.30 Calculation scheme for an eccentric ring-shaped punch: (a) vertical section, (b) plan

Fig. 4.31 Contact domain

discretization for an eccentric
ring-shaped punch

The resultant external load is applied at the symmetry centre of the outer circle
of the punch, its value is P = 103 kN. The wedge-shaped base is characterized
by the deformation modulus E = 10 MPa, Poisson ratio ν = 0.35, wedge angle
α = 30◦ . The radii ratio of the inner and outer circles of the ring-shaped punch is
taken r1 /r= 0.6, and the distance from the wedge rib to the ring-shaped punch centre
is L = 8r. The boundary-element grid was built using a fractionally linear function
of a complex variable ζ =(Az–1)/(A–z), conformally mapping a uniform grid, built
in a circle, onto the ring interior, A is the transformation parameter, determined in
terms of ε, r, and r1 . In order to obtain stable numerical results at various values
of ε, the number of the boundary elements reaches 156 (Fig. 4.31). The calculation
results are presented in Table 4.8.
It follows from the table that, contrary to the problems considered before where
all unilateral constraints were involved into functioning, for the ring part of unilat-
eral constraints appears to be open, i.e. a partial uplifting of the punch from the base
[email protected]
observed. The account of the uplifting essentially affects the calculation results.
Based on the approximation of the numerical values of the slope ψ x , quoted in
Table 4.8, the optimal value of the control parameter ε0 = 0.2088r is found, for
which there will be no slope of the ring-shaped punch.
310 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

Table 4.8 Slopes ψ x of a ring-shaped punch depending on the control parameter ε 0 /r

ε0 /r ψ x × 103
With the account of Without the account of
unilateral constraints unilateral constraints

0 5.3907 −
0.2 0.34688 −
0.208 0.072936 0.14653
0.20777 0.35083 × 103 0.85922 × 101
0.209 −0.049834 0.48411 × 101
0.21375 −0.24178 −0.10291
0.25 −2.1049 −7.4181

Fig. 4.32 Contact pressures

for an eccentric ring-shaped
punch with a translational
motion on a
variable-thickness elastic
layer

Figure 4.32 shows the contact pressure profile under the ring-shaped punch in the
XOZ plane at ε 0 /r= 0.2088. Due to the contact pressure redistribution, the variation
of the compressible base thickness will be compensated and the punch motion in
this case will be translational.
Table 4.9 contains the optimal values of the control parameter ε0 /r for typical
cases which can be practically important for construction of foundations under
tower-type structures (funnels, water towers, TV towers, etc.) on a wedge-shaped
base. The radii ratio here, similarly to the above case, is taken as r1 /r= 0.6.
By using the data of Table 4.9 for the foundation design, one can choose the
required displacement ε 0 of the inner circle with respect to the outer one, for which
the settlement of the ring-shaped foundation under consideration under a verti-
cal load will be uniform. Application of the results of the solution of the contact
[email protected]
problem for an eccentric ring-shaped punch for the design of foundations under
tower-type structures in complicated engineering and geological conditions (on
variable-thickness bases) is shown in Appendix C.
4.4 Spatial Stress-Strained State of the Base of a Rigid Strip Variable-Width Foundation 311

Table 4.9 Optimal values of the control parameter ε 0 /r (for which the ring-shape punch slope is
zero) depending on the wedge angle α and on the relative distance xc /r of the punch centre from
the wedge rib

xs /r α= 30◦ α= 45◦ α= 60◦

4 – 0.3398 0.3071
8 0.2088 0.1667 0.1413
12 0.1274 0.09828 0.08247

Thus, the numerical studies performed using the boundary-element method

enable the following conclusions to be made:

– from the condition of providing the punch settlement uniformity, the use of control
parameters while solving spatial contact problems enables the punch shape to be
rationally assigned and the point of application of the resultant external vertical
load to be shown;
– a considerable sensitivity of the punch slope to the control parameter variation
as well as the comparability of the calculated and random values of eccentricity
of the vertical load on the punch, revealed in the numerical experiments, prove
the necessity of thorough monitoring of geotechnical survey results, as well as
strict observance of the project prescriptions in the course of construction being
performed on non-uniformly compressible bases;
– the computer software, worked out on the base of the proposed algorithms of the
numerical solution of spatial contact problems, enables punches of a rather com-
plex shape resting on elastic bases, for which fundamental solutions are known or
can be obtained, to be studied under various combination of loads.

4.4 Spatial Stress-Strained State of the Base of a Rigid Strip

Variable-Width Foundation
Variable-width foundations (VWFs) are rifid foundations of a progressive type
[191]. Such foundations have a complex shape of the domain contacting with base,
formed by a periodical system of cutouts of different configuration, depending on
the foundation construction technology as well as on the specific feature of the
above-foundation structures. Application of VWFs enables, in comparison with
the solid plates, i.e. constant-width foundations (CWFs), concrete and steel con-
sumption to be essentially reduced and rational design solutions in construction and
reconstruction of foundations to be achieved in order to provide uniform settlement
of structures (without slope).
At present no reliable methods of contact stress calculation of for VWFs have
[email protected]
been elaborated, and the stress-strained state characteristics in the active zone of the
base of such foundations are found rather approximately. In order to simplify the
calculations, VWFs, as a rule, are treated as a regular repetition of similar fragments
312 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

along the foundation length, each of the fragments, in turn, consisting of a narrow
block and a wide block. The pressure of each block on the soil is assumed constant.
In spite of the limited foundation strip length, the edge effect revealed as the contact
stress increase near the bottom boundaries, is not taken into account. At the design
stage for the strip VWFs with different shape of cutouts the main task is the decrease
of bending moments what reduces the consumption of materials (concrete and steel)
in comparison with solid plates. The relevant instructions in the regulations result in
cumbersome, very approximate and often not sufficiently substantiated calculations
with a vast graphic and tabulated material being used. Reliable determination of
the bending moments due to the reactive pressures in the VWFs sections requires a
detailed account of the cutout shape.
The simplest example of a VWF is a foundation with a cross-shaped bottom
which is used instead of square or rectangular foundations. Square and cross-shaped
foundation tests with the load increase up to the limiting value have enabled the
advantages of the foundations with angular cutouts to be estimated [257]: substitu-
tion of square foundations with cross-shaped ones with the same outer dimensions
enables the metal consumption to be reduced by 26% and concrete consumption –
by 15% for the same external load being transferred to the foundation.
The results of field tests of complex-shaped strip foundations with a flat surface
of contact with a sand base are presented in [83, 257]. The studies have shown that
the contact pressure distribution for a VWF essentially depends on its shape and the
external load. It was proven experimentally that the ability of plates with cutouts
to accept increased loads is caused by the specific features of the contact stress
distribution. In the load range below the crack formation limit and at relatively small
length VWFs on dense soils work as rigid punches.
Below we present some calculation data, obtained for strip CWFs and VWFs
under central loading [18, 20]. Application of the elastic base model in the numerical
algorithm elaborated is performed as a separate module and does not require the
whole program resetting while different models of the elastic base with a known
influence function being used.

4.4.1 Contact Problem for a Variable-Width Strip Foundation

As an example of a VWF consider a shallow foundation obtained by making ten
symmetrical rectangular cutouts with the dimensions 1.2 × 0.4 m in a solid rect-
angular plate of 13.2 × 1.6 m. Since in design practice the base depth is to be set,
from which the stress in the soil can be determined as for a CWF with the equal
bottom area, the contact interaction parameters for the VWF were compared with
the corresponding calculated characteristics for a rectangular foundation of the same
length and a constant width of 1.237 m with equal area. The shapes of the VWF and
[email protected]
CWF of equal areas with the corresponding contact surface discretization schemes
are presented in Fig. 4.33. The calculation scheme for a rigid beam on an elastic
base is accepted. Numerical modelling was performed for the case of the base being
4.4 Spatial Stress-Strained State of the Base of a Rigid Strip Variable-Width Foundation 313

Fig. 4.33 Discretization of contact domains for rigid strip foundations of (a) variable and (b)
constant width with equal areas and lengths, using 396 boundary elements

treated as an elastic half-space whose deformational properties are characterized by

the elastic modulus E = 10 MPa and the Poisson ratio ν = 0.25. The foundation
was assumed to undergo a vertical load, reduced to a central force P = 1000 kN.
Multivariant calculations have shown that with the increase of the discretization
degree the accuracy of the numerical results steadily grows. In order to estimate
the convergence of the approximate solutions, the approximate values, obtained by
Richardson method in the assumption of the second order of accuracy of the numer-
ical method being applied, were used. On a uniform grid of 396 boundary elements
(Fig. 4.33) the relative error of the boundary-element solutions did not exceed 1.5%
for the settlements and 2% for the contact pressures, what is quite sufficient for
practical calculations with a detailed graphic interpretation of the results. Thus, the
approach used to calculate the VWFs with a given cutout shape, using the algo-
rithms of numerical solution of spatial contact problems, is rather efficient: there
are no difficulties with convergence, and the required accuracy does not demand
high computation time and computer RAM resources.
Contact pressure isolines p= p(x, y) on the foundation bottom are plotted in
Fig. 4.34. The contact pressure distribution patterns for the CWF and the VWF are
qualitatively different. In the first case for the rectangular strip foundation (of the
constant width) an oval area of practically contact pressures exists, extended along
the longitudinal axis of the bottom (Fig. 4.34b). With the increase of the distance
from the centre the symmetrical ovals are transformed, and directly near the foun-
dation edges the isobars have the shape of similar rectangles with rounded angles.
On the foundation boundary contour the contact pressures increase unlimitedly.
As follows from the calculations, cutouts in the strip foundation bottom results
in an essential transformation of the contact pressure field, which becomes non-
uniform first of all due to the presence of additional angular points on the bound-
ary contour (Fig. 4.34a). The shape of the isobars become complex and, at a given
average pressure over the bottom, a distinct nonuniformity of the reactive pressures
[email protected]
exists. In the areas of the bottom, corresponding to the larger foundation width,
isolated discharge areas appear, being repeated along the whole foundation length.
The numerical studies have shown that the perimeter increase due to the shape
314 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

Fig. 4.34 Equal contact pressure lines (MPa) for foundations of (a) variable and (b) constant width

complication of the boundary contour plays the decisive role in the formation of the
soil reactive pressures over the foundation bottom. At equal bottom area, its perime-
ter for the VWF (in the case of the cutout system under consideration, Fig. 4.33a) is
by 30% longer than for the CWF.
Contact pressures in a vertical plane, including the longitudinal axis of the VWF
and the CWF, are plotted in Fig. 4.35. As seen from the figure, almost along the
whole VWF length, due to the repeated foundation widenings, the axial contact
pressures oscillate with respect to the monotonously varying contact pressures for
the CWF. In a narrow zone with the width of about the broadening step, adjacent to
the edge part of thee foundations of both types, an unlimited growth of the contact
pressures occurs. Due to this zone being broader for the VWF, the contact pressures
there are essentially lower than the corresponding values for the CWF. The regu-
larities of the contact pressure distribution in the area of the strip foundation width
variation will be analyzed in detail in Sect. 4.4.3.
Note that, according to the calculations performed, the mentioned features in the
contact pressure distribution are in a qualitative agreement for different models of
elastic bases with distributive capability being used [205].
It is seen from the presented calculation data that the numerical method employed
enables the contact pressures over the VWF bottom to be calculated from the given
VWF shape. The detailed account of the contact pressures is quite important for the
[email protected]
evaluation of internal forces in the foundation structures. When the considered reg-
ularity of the contact pressure field formation can be estimated at the design stage, a
more strict substantiation of the choice of calculated forces in the VWF becomes
4.4 Spatial Stress-Strained State of the Base of a Rigid Strip Variable-Width Foundation 315

Fig. 4.35 Contact pressures σ (MPa) along the longitudinal axes of (1) variable-width and (2)
constant-width foundations

possible; hence, the materials for its construction can be spent more rationally.
Besides, the revealed features of the reactive stress distribution over the variable-
width foundation bottom can serve as a basis to create an engineering method to
calculate bending moments in the calculated foundation sections.

4.4.2 Stress-Strained State of a Strip Foundation Base

The increase of nonuniformity in the contact pressure distribution due to the pres-
ence of a cutout system in a VWF results in a decrease of its settlement. For the
example under consideration the VWF settlement decrease was 2.8% in compari-
son with the CWF. In practice the settlement value will be even smaller than the
calculated one since the soil in the cutout area participates in the VWF operation.
The obtained result means that at a fixed load the bottom area decrease to a cer-
tain limit, related to the shape and number of cutouts in the bottom of a rectangular
strip foundation, not only will not result in the settlement increase, but will also pre-
serve it within the tolerance. The aforementioned fully agrees with the result of [83]
where for a VWF on a medium-density sand the maximum size of curved cutouts
was determined, for which the settlements do not increase. The mentioned trend
should be taken into account at various stages of design when the VWF shape is
chosen. Note that cross-shaped foundation tests also confirm the positive effect of
angular cutouts on the decrease of the foundation settlements [257].
When foundations are designed with the account of their mutual influence, the
compressive stress distribution in the thickness of the soil base should be taken
[email protected]
into account. There is no special literature with instructions regarding the deter-
mination of additional pressure in the soil from the variable cross-section foun-
dations. While calculating compressive stress in any base point both below the
316 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

foundation and outside it, the action of a uniformly distributed load on each rect-
angular boundary element is summated according to the known Love formula
[92]. For the curved shape of the cutouts, we performed the most successful dis-
cretization of the contact surface, using triangular boundary elements (See Section
3.3.1 and [21]). Therefore, it is convenient to apply a more general approach
[227], enabling vertical stress to be calculated at any depth in an elastic half-
space, resulting from a load, uniformly distributed over triangular or quadrangu-
lar areas by algebraic summation of stresses under the vertices of acute angles of
right-angle triangles formed by perpendiculars to the sides of the uniformly loaded
domain.
The distribution of vertical compressive stresses σCW and σVW in the active zones
of the CWF and VWF bases, obtained from the results of some of the calculations
performed, is illustrated by Figs. 4.36–4.38. Figure 4.36 shows at various depths
vertical compressive stresses in a vertical plane containing the CWF and VWF
longitudinal axis. The comparison is performed for four different depths, starting
from the near-contact area (z = 0.1 m) and moving away from the bottom sur-
face (z = 1.0, 1.5, 2.0 m). As one should expect, the maximal difference in the
calculated values was observed for z ≤ 0.1 m. The compressive stresses along the
VWF longitudinal axis are oscillating due to the periodical variation of the foun-
dation width. At depths smaller than the average foundation halfwidth, due to the
higher concentration of contact pressures under the narrow parts and the presence
of the discharge zones under the broadenings, the stressed state in the soil qualita-
tively reproduces the stress distribution over the bottom. Compressive stresses for
the CWF practically become averaged with respect to the corresponding oscillat-
ing values for the VWF. The maximal relative deviation of the discussed values
at z = 0.1 m reaches 13% (Fig. 4.36). For the depths of the order of the foun-
dation characteristic width, the vertical stresses for the VWF practically along the
whole longitudinal axis are smaller than for the CWF and their oscillation ampli-
tudes sharply decrease. Starting from the depth z = 2 m, the presence of the cutouts
in the foundation bottom shows practically no effect on the characteristics of the
stressed state of the soil. The compared compressive stress values become graphi-
cally indistinguishable and can be calculated as in the case of a rectangular foun-
dation, i.e. using well known methods. The difference of the compressive stresses
σCW and σVW at all depths of the soil base along the longitudinal axis outside the
foundations is also quite insignificant (Fig. 4.36, x >6.6 m). Isolines of the verti-
cal stress in horizontal planes at the depths 0.1, 0.5, 1.0, and 1.5 m for the CWF
and VWF are plotted in Figs. 4.37 and 4.38. The zones of the foundation shape
effect on the stressed state in the mass soil are clearly seen, enabling the “active”
and “passive” zones of the base operation to be distinguished. Using similar plots
for the investigated stages of the foundation loading, one can estimate at various
depths the degree of the stressed state being close to the limiting one (in view of
its strength) both under the centre and under the boundary contour of the foun-
[email protected]
dation bottom. Now the combined effect of the VWF shape and the mechanical
properties of the soil on the limiting load value for rigid strip foundations can be
revealed.
4.4 Spatial Stress-Strained State of the Base of a Rigid Strip Variable-Width Foundation 317

Fig. 4.36 Vertical compressive stress σ (MPa) in the soil base under the longitudinal axes of
variable-width (solid lines) and constant-width (broken lines) foundations at the depth z = 0.1,
1.0, 1.5, 2.0 m (1–4)

Fig. 4.37 Lines of equal vertical compressive stress σ (MPa) in the soil base for a variable-width
foundation at the depths z = 0.1, 0.5, 1.0, 1.5 m (a–d)

4.4.3 Contact Pressure Distribution in the Area of the Strip

Foundation Width Variation
[email protected]
From the results of [100], based on the data of [215], in the majority (69.5%) of the
representative amount of the examined buildings in Moscow the ratio of minimal
and maximal pressures over the foundation bottom was within 1.2–2.0.
318 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

Fig. 4.38 Lines of equal vertical compressive stress σ (MPa) in the soil base for a constant-width
foundation at the depths z = 0.1, 0.5, 1.0, 1.5 m (a – d)

Evidently, one of the reasons for this is the fact that when the load on the foun-
dation strip is changed, its width is not always changed even in the case of absence
of the expansion joints. Meanwhile, the pressures under a foundation could be to
a considerable extent leveled by varying its width even by one nominal size grade
according to the foundation plate production list. The load on the foundation can be
treated as uniformly distributed. Such foundation is made as a strip consisting of the
footing plates and several rows of wall blocks. Besides, such foundation supports
an above-foundation structure – a carrying wall not less than 2.5 m high (the height
of a storey). It is natural to consider such foundation structure as a rigid beam on an
elastic base.
Here we study the contact pressure distribution in the area of the stepwise vari-
ation of the strip foundation width and quantitatively estimate additional pressures
resulting in the decrease of the load due to the foundation on its narrow part and its
increase on the broad part.
For the numerical solution of the problem we present the VWF as a strip consist-
ing of the blocks A and B, located symmetrically (Fig. 4.39). The area, occupied by
the strip foundation, is meshed in the longitudinal direction into regularly repeated
fragments of variable cross-section, consisting of rectangular parts A and B. The
latter are meshed into elementary rectangles (hereinafter elements), extended verti-
cally. Within each element we consider the distribution of pressure on the base to
be constant and equal to the pressure value in its gravity centre. This enables the
variation of the non-uniform contact pressure field in the longitudinal direction to
[email protected]
be studied, averaging it across the foundation strip.
The main difficulty for the compiling of the canonical equation system consists in
the determination of influence coefficients which, in general, are found by numerical
4.4 Spatial Stress-Strained State of the Base of a Rigid Strip Variable-Width Foundation 319

Fig. 4.39 Calculation scheme for a fragment of a strip foundation of variable cross-section for the
case Ay < By

Fig. 4.40 To the calculation

of the influence coefficients
for rectangular boundary
elements with centres, located
on the longitudinal axis of a
strip foundation

integration over the quadrature formuae of different orders. In the case under con-
sideration all the boundary elements are rectangular with sides, parallel to the coor-
dinate axes and centres lying on the same straight line. These circumstances enable
the practical calculations to be facilitated by using an easily programmable formula.
In the case of an elastic half-space the canonical equation system coefficients δ ij are
calculated in the finite form in terms of elementary functions
     
R1 + b R2 − b R1 + x 1
δij = x1 · ln + x2 · ln + 2b · ln
R1 − b R2 + b R2 − x 2

where x1 = x + a, x2 = x − a, x = xi − xj , R1,2 = b2 + x1.2 2 , a and b are the

half-element dimensions in the direction of the X and Y axes, respectively, xi is the

element centre, xj is the observation point (Fig. 4.40). This formula is a particular
case of a more complicated Love formula [156] for the settlements of the elastic
half-space points due to a uniform load, distributed over a rectangular domain.
The pressures, determined from the given calculation scheme, were used to con-
struct the corresponding profiles according to the following method.
First the pressures were calculated due to a VWF and two CWFs, one of which
[email protected]
had the width Ay and the other – By . The pressure due to the VWF was determined
for two variants of mutual arrangement of the blocks A and B: variant 1: Ay < By
(Fig. 4.39) and variant 2: Ay > By .
320 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

Fig. 4.41 Plots of contact pressures under the foundation atP= 100 kN: (1) foundation with a
constant width Bconst = Ay = 0.8 m; (2) the same, Bconst = By = 1.0 m; (3) foundation with a
variable width Ay = 0.8 m, By = 1.0 m

Then in each point a relative pressure deviation was determined as follows


padd = (p1 − p2 ) /p1 · 100%,

where padd is the relative deviation (in %) of the pressure under the VWF in com-
parison with the pressure under the CWF (hereinafter – the additional pressure), p1
and p2 are the higher and the lower (respectively) of the pressures under the VWF
(hereinafter pV ) and under the CWF (hereinafter pC ) for each measurement point
along the foundation length.
Then the profiles of relative additional pressures were built.
Figures 4.41 and 4.42 show the plots of the contact pressures in absolute
(Fig. 4.41) and relative (Fig. 4.42) values for a foundation with widths Ay = 0.8 m
and By = 1.0 m. The shape of these plots is typical for the VWFs with the size Ay
and By considered. The plots, shown in Fig. 4.41, were obtained at a central force
P = 100 kN and the soil deformation parameters E = 10 MPa, ν = 0.3.
At further consideration the following restrictions are assumed.
The plates of width, specified by the GOST standard, were used [109]. With the
account of this factor, the ledge values were assumed YL = 0.1 m and YL = 0.2 m.
For each ledge only the minimal and maximal values of the block widths were taken.
For example, for YL = 0.2 m only the blocks with Amin y = 1.6 m and Amax
y = 3.2 m
are accepted.
[email protected]
The block lengths Ax = Bx = 10 m were considered. A decrease of the length
Ax increases the influence on the block A from the block B, opposite to the
boundary considered. For example, while considering the right-side part of the
4.4 Spatial Stress-Strained State of the Base of a Rigid Strip Variable-Width Foundation 321

Fig. 4.42 Diagrams of additional pressures (in %) under the foundation with the size Ay = 0.8 m
and By = 1.0 m: (1) foundation with a constant width Bconst = Ay = 0.8 m, (2) the same, Bconst =
By = 1.0 m

foundation (Fig. 4.39), the decrease of Ax will result in the increase of influence
of the left block B. Taking this circumstance into account, the pressures at Ax =
Bx = 10 m and Ax =Bx = 20 m were compared. This study has shown that the
lengths assumed (10 m) result in an excessive pressure. For the blocks A the exces-
sive pressure is higher than for the blocks B. With the increase of the widths Ay and
By the value of such error decreases within each ledge value. At YL = 0.2 m the error
is higher than at YL = 0.1 m. However, in all cases this error does nor exceed 0.81%.
Such value can be neglected and, therefore, the numerical studies were performed at
Ax = Bx = 10 m.
Besides, the arrangement of the blocks A and B according to the variant 2
(Ay > By ) enabled, at the data processing, padd for the broad block to be taken from
the calculation results according to the variant 1, and for the narrow block – accord-
ing to the variant 2. This has also enabled the influence of the block B, opposite
to the boundary under consideration, to be reduced. The profile in Fig. 4.42 was
constructed using this approach.
The results of the studies are presented in Table 4.10. The signs “+” and “–”
indicate the additional loading, downward with respect to the foundation bottom,
and unloading, upward (respectively) action of the additional pressure. The addi-
tional pressure profile, characteristic for the cases under consideration, is shown in
Fig. 4.42. In Table 4.10 the values of pmin for the block A were calculated according
[email protected]
to the variant 2 at the distance Ax from the boundary between the blocks A and B.
The analysis of the results of the calculations performed has enabled the follow-
ing conclusions:
322 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

Table 4.10 Additional pressures (padd ) under a variable-width foundation with respect to the pres-
sure under a constant-width foundation

Block padd value at the segments, %

width, m
YL , m A B
Ay By pmin ps pmax pmax ps pmin

0.1 0.8 1.0 +1.64 +1.95 +9.54 −10.68 −1.86 −1.54

1.4 1.6 +1.07 +1.33 +5.76 −6.10 −1.28 −1.04
0.2 1.6 2.0 +1.90 +2.36 +9.40 −10.28 −2.21 −1.76
2.8 3.2 +1.28 +1.64 +5.60 −5.84 −1.56 −1.20

• padd , arising in the area of a sharp variation of the foundation width, have dif-
ferent directions (under the block A they are unloading, and under the block B –
additional loading);
• the maximal values of padd are achieved at the boundary between the blocks A
and B; its value depends on the YL value and on the block widths and lies within
5.6–9.54% under the block A and within 5.84–10.68% under the block B;
• the maximal values of padd are achieved at YL = 0.1 m for the narrowest blocks
(A = 0.8 m, B = 1.0 m);
• for each ledge size the value of padd is inversely proportional to the block width;
• with the decrease of the relative ledge value ( Y L = YL /Ay ) the maximal padd
values decrease;
• the most considerable variations of padd occur near the boundary; the length, at
which the maximal pressure is reduced twice, is not more than 1.5 m for the block
size combinations under investigation.

Using the relative values (%) has enabled the results, which can be extended to
various loads and soils, to be obtained in a compact form.
The analysis performed has shown that the pressure redistribution in the area of
the strip foundation width variation exists and should be taken into account. Addi-
tional pressures, arising in the area of the foundation width variation, are unloading
for the narrow foundation parts and additional loading for the broad parts. In a short-
length area (up to 1.5 m), adjacent to the boundary between the blocks, such pres-
sures sharply increase, for the block widths under investigation the increase reach-
ing 9.54% (for the narrow blocks) and 10.68% (for the broad blocks) of the pressure
value arising under a constant-width foundation made of broad blocks loaded by
above-foundation structures.
Thus, the account of regularities of stress distribution on the contact surface and
in the soil at different geometrical shape of the foundation enables the methods for
design of foundation structures to be improved. The calculations of the VWF bases of
different type (periodical profile, symmetrical and asymmetrical with respect to the
[email protected]
longitudinal axis, of sleeper type, etc.) can be carried out with higher reliability, the
optimal size of such foundations can be chosen automatically, providing their minimal
settlement without over-expenditure of expensive construction materials.
4.5 Calculation of the Section Kernel Boundary for Rigid Foundation Plates 323

4.5 Calculation of the Section Kernel Boundary for Rigid

Foundation Plates
As follows from the investigations performed (Sects. 4.2 and 4.3), in Sect. 2.5, based
on the boundary-element method, an efficient and a rather general approach was
developed to the numerical solution of spatial problems of contact interaction with
the account of structural nonlinearity (unilateral constraints in the contact domain)
for absolutely rigid punches of complex (practically any) shape on non-classical lin-
early deformable bases. Under a given spatial static load linear and angular punch
displacements are determined, the pressure distribution in the contact domain is
found. The account of unilateral constraints in the proposed numerical algorithm
enables the rigid punch uplifting from the elastic base to be described as well as the
pressure redistribution over the contact domain and slope increase with the external
load application point approaching the punch boundary (or with the increase of an
external overturning moment). Numerous calculations show that with the appear-
ance of the zones of the punch uplifting from the elastic base surface and with
the decrease of the contact area a nonlinear variation of the slopes and displace-
ments of punches of various shapes with the external vertical force eccentricities (or
external overturning moment values) is observed. Structural nonlinearity, revealed
in such a way under the contact interaction, makes impossible the application of
the superposition principle which is applied for almost all engineering calculation
methods. Therefore, for a given punch geometry, the determination of the verti-
cal load application area without the punch uplifting from the elastic base, is quite
important.
A convenient geometrical characteristic of a shallow foundation bottom is section
kernel [105], which denotes the part of the punch (a rigid foundation plate), within
which a vertical load can be applied without causing a negative reactive pressure or,
more exactly (since the soil does not work for tension), the punch uplifting from the
base in any other point of the punch bottom.
Note once again that under an external vertical load within the section kernel the
contact interaction of the punch with the elastic base occurs without uplifting and,
hence, the superposition principle of linear theory of elasticity, being the basis for
most engineering calculations, can be applied.
For a circular punch, as directly follows from [2, 74, 159], contact stress will be
compressive in all points of the punch bottom only under a condition that the vertical
force application point is located at a distance from the punch centre, not exceeding
a/3 where a is the punch radius. Thus, the section kernel of a circular punch is also
a circle whose radius is three times smaller than the punch radius. In [160] a contact
problem is solved for an elliptical punch resting on an elastic half-space, loaded by a
vertical eccentric force. From the solution obtained it follows that the section kernel
of the elliptical punch is an ellipse with semiaxes three times smaller than the punch
[email protected]
semiaxes.
When the vertical force goes beyond the section kernel, the punch uplifting
from the base occurs, accompanied by the change of the contact stress distribution
324 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

character (the problem of an essentially oblique punch). Investigation of the prob-

lem of an essentially oblique circular punch, using the structural method, shows the
possibility of the contact problem solution to be obtained for asymmetrical load-
ing with the account of the punch uplifting from the base [225]. The contact area
of the punch with the half-space, being a part of a circle, is determined depend-
ing on the force application point. The contact domain, bounded by the boundary
condition separation line, will be unknown and is determined in the course of the
problem being solved by iteration from the contact pressure sign change. However,
the application of the structural method in the problem of an essentially oblique
punch is encumbered by the necessity of the function, determining the family of the
boundary condition separation lines for the contact stresses to be fully specified for
the calculations. For a circular punch this function is assumed to be a biparametric
parabola [225]. It is clear that for solving contact problems for essentially oblique
complex-shaped punches the application of the structural method (R-function) is
rather difficult from the practical point of view due to the impossibility of the con-
tact domain on the punch bottom to be predicted.
At present, the complete data on the section kernel configurations neither for
rectangular punches, nor (moreover) for the punches of more complex geometrical
shapes, interacting with an elastic half-space, are available in the literature. A great
practical interest is also attracted by the problems of design and estimation of sta-
bility of shallow foundations when the finite and variable depth of the compressible
soil mass, layered superposition of rock strata, variation of the elastic properties
with depth, etc. should be taken into account.
Here we describe the procedure of an approximate (with a given accuracy) deter-
mination of the section kernel boundary for punches of a complex geometrical
shape, contacting with elastic bases, by invoking a numerical algorithm of solution
of the spatial contact problem (Sect. 2.5.1).
In the punch bottom plane we introduce a polar coordinate system with a pole
in an internal point (centre) of the punch (xc , yc ) (Fig. 4.43). This point is the most
convenient to be taken in the gravity centre or on the symmetry axis of the punch
bottom. Then we fix a direction ϕ, 0 ≤ ϕ < 2π, and carry out numerical calculations
of the contact interaction of the punch with the elastic base for a vertical load, whose
application point is varied along the segment ρ 1 (ϕ) < ρ(ϕ) < ρ 2 (ϕ) (the choice
of its ends will be specified below). The action of the vertical force in this case
is equivalent to the action of a force P, applied in the punch centre (xc , yc ), and
moments Mx = –P×ρsinϕ, My = P×ρcosϕ. Then we apply a procedure similar to
the bisection method of solving nonlinear equations.
At each value of ϕ, the polar radii of the zeroth approximation, bounding the sec-
(0)
tion kernel boundary point, are chosen in such a way that for ρ = ρ1 the contact
interaction is without uplifting, and for ρ = ρ2(0) at least for one boundary element
(0) (0)
ρ i < 0. The interval ρ1 < ρ < ρ2 is divided into the simples case in two and
[email protected]
perform the calculations for ρ = (ρ1(0) + ρ2(0) )/2. If at ρ = ρ there will be a con-
(1) (1) (0)
tact interaction without uplifting, then ρ1 = ρ, ρ2 = ρ2 is chosen as the first
approximation. At the presence of areas of the punch uplifting from the elastic base
4.5 Calculation of the Section Kernel Boundary for Rigid Foundation Plates 325

Fig. 4.43 Scheme for the

algorithm to determine the
section kernel of a rigid
complex-shaped foundation
plate, resting on an elastic
base

(1) (0) (1)

for ρ = ρ, one assumes ρ1 = ρ1 , ρ2 = ρ. Then the calculations are performed
according to the above scheme by iteration, until the condition
+ + + +
+ (k) (k) + + (k−1) (k−1) +
+ρ1 − ρ2 + ≤ ε · +ρ1 − ρ2 +

 ε is the
is fulfilled, where  given relative accuracy of the calculations. Monotonous
sequences ρ1(k) , ρ2(k) , obtained from the calculations, are limited and, within
the given accuracy, after a finite number of iteration cycles each of them achieves
its limiting value ρ ∗ (ϕ) which determines the point (ρ ∗ cosϕ,×ρ ∗ sinϕ) of the punch
section kernel for the chosen direction ϕ.
In the practical calculations of the section kernel boundary, depending on the
complexity of the punch bottom shape, the angular interval {ϕ: 0≤ϕ<2π} was pre-
sented by discrete values

2π
ϕ = ϕ · (m − 1) , ϕ = , m = 1,2,...,M.
M
The calculation procedure is essentially shortened by the account of the external
punch contour symmetry. It is also worth noticing that, in order to accelerate the
iteration process convergence, we used the following technique: at each iteration
step the values ρ(ϕ) were determined as

(k) (k)
ρ1 + λk ρ2
ρ (k) =
1 + λk
where λk is a dimensionless parameter, characterizing at the k-th step the ratio of
the total area F1 of boundary elements, for which the contact stress is compressive
(ρ i > 0), to the total area F2 of boundary elements, for which the contact stress is
tensile (ρ i < 0), F1 ∪ F2 = Φ, Φ being the whole punch bottom area.
[email protected]
As a test example to estimate the working efficiency of the algorithm proposed,
we have chosen determination of the section kernel boundary for a round punch
resting on a homogeneous elastic half-space. As noted above, this problem has an
326 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

(a) (b)

Fig. 4.44 Section kernel for a square punch on an elastic half-space: (a) location with respect to
the punch contour, (b) calculated values of the polar radius of the boundary points

exact solution, ρ ∗ = a/3, a being the punch radius. Numerical calculations on the
base of the method proposed with a uniform discretization of the circular contact
domain into triangular and quadrangular boundary elements using 20 radii and 5
concentric circles, gives for ε = 10−4 an approximate solution ρ∗ = 0.33486a. The
relative error is seen to be quite small, much less than 1%.
The results of calculations of the section kernel boundary for a square punch
on an elastic half-space are presented in Fig. 4.44. Due to the symmetry of the
problem, the calculations were performed for the angular interval {ϕ: 0≤ϕ≤π/4}
with ϕ = 3◦ . As seen from the calculation data, the shape of the section kernel
is a square-type figure with rounded angles, whose symmetry axes make an angle
π /4 with the symmetry axes of the punch itself (Fig. 4.44a). For the example under
consideration, the linear size of the section kernel is almost 2.8 times smaller than
the punch itself.
The problem of pressure of a flat square punch, loaded by an eccentric force,
was first solved approximately by Leonov, Posatskiy and Ivashchenko [154]. They
applied the method of solving the problem of pressure under the punch bottom,
resting on an elastic half-space, when the contact domain is close to circular. A
circle is inscribed into the square contact domain. The pressure under the punch
within the circle consists of the pressure, arising at the impression of a circular
punch, and an additional pressure due to the application of a normal load, applied
outside the circular punch, on the areas, complementing the circle to the square
domain, required to provide the settlement under the punch on the whole square
contact domain to coincide with the given settlement value. Solution of the problem
[email protected]
at an eccentrically applied vertical force has shown that in order to obtain non-
negative contact pressures under the punch, a condition for the eccentricity along
the axis, normal to the square side with the length 2a,
4.5 Calculation of the Section Kernel Boundary for Rigid Foundation Plates 327

ρ ≤ 0.45a

should hold. Note that the accuracy of the obtained value for the estimation of the
section kernel size ρ ∗ = 0.45a is low due to a rather approximate integration of
the main equation of the problem with a large step α = 5◦ along the angular
coordinate as well as not quite efficient of subsequent approximation of the solution
found. Besides, the method proposed in [154] does not possess high generality and,
as noted earlier, is applicable only for contact domains, close to circular.
The results we have obtained are in a rather good agreement with the estimations
for the section kernel boundaries for rigid rectangular plates on an elastic half-space,
found in [105] by an approximate method, using the expansion of the contact pres-
sure function into power series with a finite number of terms. It was also shown there
that a point in the centre of the punch side is more dangerous in view of the punch
uplifting than an angular point. The limit eccentricity value, above which the square
punch uplifing from the soil occurs, obtained for this point in case the reactive pres-
sure equation being taken as a sixth-order polynomial for a centred vertical load and
a fifth-order polynomial for a momental load, equals ρ∗ /a = 0.425 (a being half the
side of the square). More exact calculations with eighth- and seventh-order polyno-
mials being applied for the vertical and momental loads, respectively, give the value
ρ ∗ /a = 0.429. According to our calculations for a square punch, discretized into
10 × 10 square boundary elements, at ε = 10−4 the corresponding value will be
ρ ∗ /a = 0.4282 (Fig. 4.44b).
The proposed numerical algorithm enables the problems of essentially oblique
punches of a rather arbitrary shape to be solved and their section kernel bound-
aries to be found with a given accuracy, depending on the punch bottom domain
discretization degree (number of the boundary elements) and the mesh grid qual-
ity. The calculation of the section kernel boundary for rigid punches with complex
contours on elastic bases of various type is fully automated. From the geometrical
point of view the problem is reduced to a rather detailed discretization of the flat
domain, using triangular and quadrangular boundary elements, taking into account
the contact stress concentration at the punch outer boundary. Application of any
known software for building up high-quality grids for flat domains with a piecewise
smooth boundary is possible, including those extensively used for finite-element
calculations (See, e.g., [131, 228, 229]). We employed an intentionally elaborated
software for solving spatial contact problems of theory of elasticity (FORTRAN-77
language) on the base of the algorithm described in Sect. 3.3.1. It enables discretiza-
tion of flat domains, restricted by straight segments and arcs of smooth curves to be
performed, what corresponds to the geometrical characteristics of foundation
bottoms of various shape, used for construction and reconstruction of buildings and
structures of known types.
In practical engineering, especially for construction of new foundations at con-
strained conditions of subterranean parts of buildings under reconstruction [226],
[email protected]
complex-shaped foundations must often be used, as a rule, with a polygonal contour.
Under considerable momental variable-sigh loads, or when there is a need to reduce
the bottom length, foundations with I-shaped and T-shaped bottoms are applied. In
328 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

the cases of high momental loads, acting in perpendicular planes, L-shape bottom
foundations are used. Cross-shaped foundations are used instead of rectangular ones
in case lugs being provided for mounting closely located pillars or other structures
of various design. Round foundations are often substituted by polygonal (hexa- or
octagonal), since they are simpler for construction and their formwork production
is less labour-consuming. These foundation types as well as other foundations with
polygonal bottom are often used under various equipment and other technological
structures [124, 257].
The abilities of the proposed method for estimation of stability of rigid founda-
tion plates (interacting with the soil without their bottom uplifting) under an off-
centre vertical load are illustrated (Figs. 4.45–4.48) by the calculations of the sec-
tion kernel boundaries for polygonal punches (trapezoidal, cross-shaped, I-shaped,
L-shaped), resting on an elastic half-space.
A common fact for all the examples under consideration is the presence of kinks
at the section kernel boundaries, corresponding to the internal angular points on the
contour of the punch itself. The presence of an external angle on the punch contour
results in a smooth curving (rounding) between the practically linear segments at
the section kernel boundaries. Besides, from the comparison of the section kernel
configurations for a square and a trapezoidal punches with one common base
(Figs. 4.44 and 4.45) it is seen that the punch contact domain broadening results
in the stretching and displacement of the section kernel in the corresponding
direction.
As follows from the calculations performed, the proposed numerical approach
enables, from a given punch contour shape, the section kernel boundary (the domain
of contact interaction without uplifting) to be effectively determined. This, in turn,
provides important information at the stage of the complex-shaped foundation

(a) (b)

[email protected]
Fig. 4.45 Section kernel for a trapezoidal punch on an elastic half-space: (a) location with respect
to the punch contour, (b) calculated values of the polar radius of the boundary points for a square
(1) and a trapezoidal (2) punch
4.5 Calculation of the Section Kernel Boundary for Rigid Foundation Plates 329

Fig. 4.46 Section kernel for a cross-shaped punch on an elastic half-space: (a) location with
respect to the punch contour, (b) calculated values of the polar radius of the boundary points

Fig. 4.47 Section kernel for an I-shaped punch on an elastic half-space: (a) location with respect
to the punch contour, (b) calculated values of the polar radius of the boundary points

design when a strict account of the conditions of the punch bottom uplifting from
the base is required. The relevant instructions available in the regulations [162, 249],
result in cumbersome, very approximate and often not enough substantiated calcu-
lations with the use of vast graphic and table data. Therefore, the determination of
the section kernel boundaries for foundations with bottom of a complex geometri-
cal shape enables the methods of design and calculation of shallow foundations to
be improved. In a number of practical cases, arising at the reconstruction of build-
[email protected]
ings, the maximal possible values of vertical load application eccentricities can be
reliably determined. Thus, design of foundations with uplifting is avoided and high
quality of the project solutions is provided.
330 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

(a) (b)

Fig. 4.48 Section kernel for an L-shaped punch on an elastic half-space: (a) location with respect
to the punch contour, (b) calculated values of the polar radius of the boundary points for a square
(1) and the L-shaped (2) punches

Application of the elastic base model in the elaborated numerical algorithm is

performed as a separate module and does not require resetting of the whole pro-
gram in case any other model of the elastic base being used, for which the influence
functions have been obtained (See, e.g., [39, 73, 75, 89, 210, 225]). As a rather
helpful illustration of the abilities of the proposed numerical method we consider a
spatial contact model of an elastically compressible variable-thickness layer in the
shape of a spatial wedge 0 ≤ x < ∞, –∞ < y < ∞, 0 ≤ z ≤ x × tanα with a central
angle α at the rib, 0 < α < 90◦ [89, 158]. The oblique plane, restricting the wedge
from below, is completely restrained. The presence of an oblique underlayer makes
the elastic base nonhomogeneous in depth and enables the account of variation of
the compressed soil layer thickness under the bottom of large-scale foundations.
Figures 4.49–4.54 present the calculated section kernel configurations for founda-
tion plates of various, mostly polygonal shape, resting on an elastic compressible
variable-thickness layer. It follows from the numerical calculations performed that
for the foundations of the geometrical shapes under consideration their section ker-
nels on an elastic compressible wedge always have smaller size than in the classical
case (on an elastic homogeneous half-space). The corresponding reduction of the
section kernel area, related to the non-uniform compressibility of the base, are rather
essential and depend on the angle α of inclination of the rigid underlayer and on the
punch distance from the elastic wedge rib, corresponding to the coordinate origin.
With respect to the punch centre, one can observe a sort of the section kernel shift
towards the wedge rib and its size decrease at the opposite side.
For the punches with a smooth contour (a circle, an ellipse) the section kernel
boundaries are distorted and shift, remaining practically similar to the section kernel
[email protected]
in the case of the half-space (Figures 4.49 and 4.50). However, for the polygonal
punches, characterized by the presence of angular points on the contour, the section
kernel boundaries acquire distinct straight segments and kinks, their location being
4.5 Calculation of the Section Kernel Boundary for Rigid Foundation Plates 331

Fig. 4.49 Section kernel for

a round punch with a radius a
on a variable-thickness elastic
layer. 1: xc = 1.5a,α = 15◦ ; 2:
xc = 1.5a, α= 30◦ ; 3: xc = a,
α= 45◦ ; 4: elastic half-space

Fig. 4.50 Section kernel for an elliptical punch (x–xc )2 /a 2 + y 2 /b 2 = 1 on a variable-thickness

elastic layer (α=30◦ ). 1: xc = 2.5a, b = a/2; 2: xc = 1.5a, b= 2a; 3: elastic half-space

dependent of the nonhomogeneity degree of the compressed soil mass under the
punch bottom (Figs. 4.51–4.54).
Note that the section kernel boundary configurations for the rigid foundation
plates resting on elastic half-space-type bases with depth-dependent characteristics,
[email protected]
calculated by the method proposed here, are similar to those obtained for the homo-
geneous half-space (Figs. 4.44–4.48), but their size depends on the nonhomogeneity
parameters. Numerous calculations have shown that the method elaborated enables
332 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

Fig. 4.51 Section kernel for a square punch on a variable-thickness elastic layer. 1: xc = 1.5a, α=
15◦ ; 2: xc = 1.5a,α= 30◦ ; 3: xc =a,α= 45◦ ; 4: elastic half-space

Fig. 4.52 Section kernel for a trapezoidal punch: (1) on a variable-thickness elastic layer, xc = 3a,
α= 15◦ and (2) on an elastic half-space
[email protected]
4.5 Calculation of the Section Kernel Boundary for Rigid Foundation Plates 333

Fig. 4.53 Section kernel for an L-shaped punch on a variable-thickness elastic layer (xc = 2.5a).
1:α= 15◦ ; 2: α= 30◦ ; 3: elastic half-space

Fig. 4.54 Section kernel for a cross-shaped punch: (1) on a variable-thickness elastic layer, xc =
2a, α= 30◦ ; and (2) on an elastic half-space
[email protected]
334 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

the effect of the elastic base nonhomogeneity degree (both in depth and over the
area) on the size and shape of the rigid foundation plate section kernels to be esti-
mated within the guaranteed accuracy.
Thus, the numerical algorithm elaborated enables the section kernel boundaries
to be calculated from the given shape of the rigid foundation plate bottom and the
mechanical characteristics of the elastic base and thereby the domains of applica-
tion of a vertical resultant external load to be determined, for which the contact
interaction of the rigid foundation plate occurs without the bottom uplifting from
the soil base. The section kernel configurations have been obtained for the circu-
lar, elliptical, rectangular, trapezoidal, cross-shaped, I-shaped, L-shaped rigid foun-
dation plates on spatially nonhomogeneous bases, for which obtaining analytical
solutions has come across unsurmountable difficulties.
The obtained calculation data on the section kernel boundaries are important
from the point of view of estimation of the complex-shaped rigid foundation plate
stability for the spatial stress-strained state of the soil and can provide recommenda-
tions regarding the determination of optimal values of foundations operating without
uplifting of the bottom from the base. The method requires short time for data prepa-
ration and numerical computations, provides stability and accuracy of the results
obtained, sufficient for engineering applications.

4.6 Numerical Algorithms of Solving Boundary Integral

Equations in Spatial Contact Problems for a Nonlinearly
Deformable Base
The effective account of physical nonlinearity of soils is important for the ratio-
nal design of bases and foundations [67, 94, 283]. At present, however, there is no
unique and reliable approach to the solution of spatial problems of soil mechanics
with the account of a wide variety of strength-related and deformational properties.
As a rule, this is related to the lack of reliable data from stabilometric measurements
of mechanical characteristic of soils for complex stress-strained states as well as
with too high computer resources required for the calculations. Besides, one should
note a rather narrow orientation of the existing software requiring large amount
of information to prepare the input data for three-dimensional finite-element mod-
elling. Therefore, it seems quite important to work out rather simple contact models
with an ability to the account of nonlinear properties of various bases. Application of
such models in combination with the efficient boundary-element method, enabling
the problem dimensionality to be reduced, for a number of practical situations can
lead to a considerable simplification of calculations, reduction of their scope, and
a sufficiently accurate estimation of the qualitative and quantitative aspects of the
contact interaction of the structures with the soil mass. Besides, as shown below,
[email protected]
such an approach widely employs traditional numerical methods, popular for solv-
ing problems in the linear formulation with a wide range of contact models being
applied, taking into account the base distributive properties.
4.6 Numerical Algorithms of Solving Boundary Integral Equations 335

4.6.1 Spatial Contact Model for a Nonlinearly Deformable Base

In practical calculations of soil bases for the linear deformation stage spatial contact
models for a half-space [92, 128], a constant- or variable-thickness layer [75, 89],
layered and nonhomogeneous with depth half-spaces, etc. have gained good repu-
tation. The above noted contact models are the solutions of problems of the base
surface settlement under a concentrated vertical load. The contact models, based on
the strict solutions of a mixed spatial problem of theory of elasticity, are a relatively
simple and convenient means to set a relationship between the pressure on the base
and its settlement as given below:
 
1 − ν2
W (x,y) = p (ξ ,η) ω (x,y,ξ ,η) dξ dη (4.4)
πE
S

where the function ω(x, y, ξ , η) is chosen, depending on the base model assumed.
The contact models applied are linear and differ in the prediction of the base dis-
tributive properties. For the real conditions of soil deformation under the load
increase, a nonlinear character of settlement-versus-load relation is revealed, and
Eq. (4.4) requires a generalization.
Suppose the settlement in a given base surface point to be dependent of the load-
ing level p/p∗ . It can be given by

dW (x,y) = ϕ (p,p∗ ) ω (x,y,ξ ,η) dξ dη

where ϕ(p, p∗ ) is a function of the loading level, defining the law of nonlinear defor-
mation of the surface under the load, p∗ is the limiting contact pressure on the base
surface, ξ and η are the coordinates of the centre of an elementary area, x and y are
the coordinates of the point under consideration.
From mechanical speculations, qualitative conclusions on the form and proper-
ties of ϕ(p, p∗ ) can be made. Thus,

ϕ (p,p∗ ) ∼ p at p << p∗ ,
ϕ (p,p∗ ) → ∞ at p → p∗ ,
ϕp (p,p∗ ) > 0, ϕpp (p,p∗ ) > 0, 0 < p < p∗ .
The p∗ value is considered as the contact model parameter, and we call it a lim-
iting contact stress, since it has the dimensionality of stress and plays the role of a
characteristic scale.
Specify the function ϕ(p, p∗ ) using the following analytical approximation:

p
ϕ (p,p∗ ) =  2 n m
[email protected] 1 − p p∗
where m, n > 0 are dimensionless parameters which, along with p∗ , are subject
to identification based on the analysis of the experimental data. Hereinafter we
336 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

Fig. 4.55 Load function

plots at the nonlinearity
parameters n= 1; m = 0 (1),
0.2 (2), 0.5 (3), 2.0 (4), 5.0
(5)

consider n = 1. Parametric dependences, reflecting the character of the nonlinear

behaviour of ϕ(p, p∗ ) with the increase of the m parameter, are plotted in a dimen-
sionless form in Fig. 4.55. Note that at small values of m one has the case of linear
settlement-versus-load relationship practically in the whole range of the active stress
variation.
Thus, the settlement of the nonlinearly deformable base surface under an exter-
nal vertical load p(x, y), distributed over the finite area S, will be presented in the
following generalized form:
 
1 − ν2
W (x,y) = p (ξ ,η) ϕ (ξ ,η,p∗ ) ω (x,y,ξ ,η) dξ dη (4.6)
πE
S

where ϕ (ξ ,η,p∗ ) = ϕ (p,p∗ ) /p∗ . The form of Eq. (4.6) enables boundary contact
problems to be built for given ϕ (p,p∗ ) or ϕ (p,p∗ ) and for various fundamental solu-
tions ω(x, y, ξ , η), known from the literature.
Note that in [282] a nonlinear dependence of the base settlement on the load is
taken into account for the calculations of rigid beams on a nonhomogeneous Winkler
base, obtained from an approximate formula

1 − ν2 p
W= ω·b (4.7)
E 1 − p/p∗
where ω is a coefficient, depending on the rectangular foundation side ratio, p∗ is
the pressure corresponding to the loss of the carrying capacity by the base. Equation
[email protected]
(4.7) was also used to set the coefficients of increase of the standard pressures on the
soil in the mining underworking areas as well as to estimate the effect of nonlinear
relationship between the foundation settlements and loads at the determination of
4.6 Numerical Algorithms of Solving Boundary Integral Equations 337

forces in buildings erected on the underworking territories. In [290], the shooting

method was used for solving the problem of calculation of a rectangular plate on a
nonlinearly deformable base, modelled by a solid half-space with the deformation
modulus being a function of the load
 γ
1/γ
p
E = E0 1 −
p∗
where E0 is the initial deformation modulus, corresponding to the proportional
dependence between the pressure on the soil and the displacements on its surface,
p∗ is the limiting pressure on the soil, determined from the punch tests, γ is the
nonlinearity parameter.

4.6.2 System of Nonlinear Contact Equations of the Contact

Problem for Absolutely Rigid Punches of a Complex Shape
with a Flat Base
The spatial contact model of Eq. (4.6), using the dependences ϕ(p, p∗ ) and
ω(x, y, ξ , η), is applied to determine contact pressures p(x, y), vertical displacements
Ws and slopes ψ x , ψ y of rigid punches of arbitrary shape resting on the surface of a
nonlinearly deformable base. Consider the punch to undergo a static load, reduced
to a vertical resultant force P and moments Mx , My . Suppose also the vertical dis-
placements of the punch and the base surface to be equal, and tangential stress in
the contact plane to be zero.
For the spatial contact problem the equality of the vertical displacements of the
punch with the area F and of the base surface results in the following nonlinear
boundary integral equation to determine the unknown contact pressure and parame-
ters of the punch displacement as a rigid solid:
 
p (ξ ,η) ϕ (p, p∗ ) ω (x,y,ξ ,η) dξ dη = Wc + ψx · (x − xc ) + ψy · (y − yc ) (4.8)
F

where F is the area of the contact between the punch and the base, p(x, y) is the
sought contact pressure function, Ws is the vertical displacement of the punch cen-
tre, xc , yc are the coordinates of the reduced external force application point. On the
free surface of the base (z = 0) outside the contact domain p(x, y)= 0. Besides, the
equilibrium equation system for the punch
 
p (ξ ,η) dξ dη = P,
  F   (4.9)
[email protected]
p (ξ ,η) ξ dξ dη = P · xC − My , p (ξ ,η) ηdξ dη = P · yC − Mx
F F

should be satisfied.
338 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

Exact solutions of the formulated spatial contact problem exist only for a uniform
isotropic linearly deformable half-space when the contact domain is of an elliptical
or circular shape [74, 92, 98, 128, 203]. For punches of a more complex shape,
as well as under interaction with the nonlinearly deformable base, the numerical
boundary-element method will be used, which is successfully applied in the case of
linearly deformable bases (See Sects. Section 4.2 – Section 4.5).
The nonlinear integral equation (4.8) is solved numerically in combination with
the conditions of Eq. (4.9). For this purpose, the contact domain is meshed into
N triangular or quadrangular elements. In the simplest case a piecewise constant
approximation of the contact pressure function is assumed, hence within a separate
element p(x, y) = const. In case of impossibility of the contact domain to be dis-
cretized into a rather large amount of elements and in order to increase the accuracy
of the numerical solution, the calculations should be performed using the piecewise
constant contact pressure function.
We successively substitute the coordinates of centres of gravity of all elements
into Eq. (4.8) and replace the dual integrals over the area F by a sum of integrals
over each element. The unknown contact pressures pi on the elements (i=1,..,N)
as well as the parameters Ws , ψ x , and ψ y of displacement of the punch as a
rigid solid are determined from a system of (N+ 3) nonlinear (over pi ) algebraic
equations

⎧
⎪
⎪ p1 ϕ (p1 ,p∗ ) δi1 + p2 ϕ (p2 ,p∗ ) δi2 + ... + pN ϕ (pN ,p∗ ) δiN − WC −
⎪
⎪
⎪
⎪
⎪
⎪ −ψx · (xi − xC ) − ψy · (yi − yC ) = 0, i = 1,N,
⎨
p1 S1 + p2 S2 + ... + pN SN = P, (4.10)
⎪
⎪
⎪
⎪
⎪
⎪ p1 S1 x1 + p2 S2 x2 + ... + pN SN xN = P · xC − My ,
⎪
⎪
⎩
p1 S1 y1 + p2 S2 y2 + ... + pN SN yN = P · yC + Mx .


Here, similarly to the above case, δij = ω (xi ,yi ,ξ ,η) dξ dη is a vertical dis-
Fj
placement of the surface of an elastic linearly deformable base in the point (xi ,yi ),
coinciding with the centre of gravity of the i-th element, due to a unit load, uni-
formly distributed over the area Fj of the j-th element, Si is the i-th element
area.
The numerical studies of the problem under consideration were carried out using
a software, programmed by means of the FORTRAN language for punches of an
arbitrary shape. According to the algorithm elaborated, the contact domain is suc-
cessively discretized, the δij coefficients are calculated from the given function
ω(x, y, ξ, η), determining the distributive ability of the base, the equation sys-
[email protected]
tem (4.10) is formed and solved for different values of the external forces
and moments. The δij coefficients are calculated semianalytically, as mentioned
in Sect. 2.5.1.
4.6 Numerical Algorithms of Solving Boundary Integral Equations 339

4.6.3 Iterative Processes of Solving a Finite-Measure Analogue

of the Spatial Contact Problem for a Nonlinearly Deformable
Base
As a result of the discretization of the contact domain and the piecewise constant
approximation of the contact pressure field, a finite-measure analogue of the spa-
tial contact problem is obtained in the form of a system of N nonlinear equations
(4.10), resulting from the nonlinear integral equation (4.8), as well as three linear
equilibrium equations (4.9).
At m = 0, i.e. when the contact model is linear and the settlement of each point of
the base surface is directly proportional to the contact pressures, Eqs. (4.10) become
a system of linear algebraic equations and, due to the diagonal predominance, have
good conditionality. This, in turn, enables standard methods of Gauss type to be
used for the solution, without any special methods of regularization. For this case
a rather general approach is developed for solving the spatial problems of contact
interaction with the account of unilateral constraints for complex-shaped punches
on no-classical linearly deformable bases (Sect. 2.5.1, Sects. 4.2–4.5).
At m= 0 the system (4.10) becomes nonlinear. Its solutions can be found by dif-
ferent methods, among which we used iterative ones [188, 230, 262]. Based on the
experience of organization of different iterative algorithms, one can conclude about
the efficiency of their application for the class of problems under consideration.
The simple iteration method can be the most easily realized for the system (4.10).
First the system (4.10) is solved for the case of a linearly deformable base (m = 0).
(0) (0) (0) (0) (0) (0)
In this case all ϕ (p, p∗ )=1. The solution obtained (p1 ,p2 ,...,pN ;Wc ,ψx ,ψy )
is then used as an initial approximation in the following iterative process:

⎧ α+1
⎪ p1 ϕ (pα ,p∗ ) δi1 + p2α+1 ϕ (pα ,p∗ ) δi2 + ... + pN
α+1
ϕ (pα ,p∗ ) δiN − WCα+1 −
⎪
⎪
⎪
⎪
⎪
⎪ −ψxα+1 · (xi − xC ) − ψy · (yi − yC ) = 0, i = 1,N,
⎪
⎨
p1α+1 S1 + p2α+1 S2 + ... + pN α+1
SN = P,
⎪
⎪
⎪ α+1
⎪ α+1 α+1
⎪
⎪ p1 S1 x1 + p2 S2 x2 + ... + pN SN xN = P · xC − My ,
⎪
⎪
⎩ α+1
p1 S1 y1 + p2α+1 S2 y2 + ... + pN α+1
SN yN = P · yC + Mx
(4.11)
where α is the iteration number. Numerous calculations have shown that the iterative
process of Eq. (4.11), as a rule, converges for different values of the parameter m and
punches of various configurations at a wide choice of the known influence functions
ω(x, y, ξ , η) and external loading parameters. However, as follows from the prac-
tice of numerous calculations, direct application of the simple iteration method is
hardly effective due to its slow convergence. With the increase of the m parameter
values the convergence rate sharply decreases. Therefore, in order to speed up the
[email protected]
convergence of the succession of approximation, the δ 2 -Eitken transformation [115,
262] was used, convenient for the calculations and enabling the convergence of the
iterative process of Eq. (4.11) for the class of problems under study to be speeded
340 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

up by dozens of times. The studies performed have shown that the use of such iter-
ative processes as nonlinear relaxation methods, Jacobi or Seidel methods [230]
have not revealed any advantages in comparison with the simple iteration methods
and possess linear convergence. Meanwhile, the differential properties of nonlin-
ear functions, contained in the equations of the system (4.10), enable the Newton
method and its modifications to be used for its solution with any preset accuracy.
The Newton method for the system of nonlinear equations (4.10) is built in the
following way. If pαi , i = 1,N,; Wc ,ψx ,ψy are already known, then
(α) (α) (α)

piα+1 = pαi + pαi , i = 1,...,N;

WCα+1 = WCα + WCα , ψxα+1 = ψxα + ψxα , ψyα+1 = ψyα + ψyα ,

where the values pαi , i = 1, N,; WCα , ψxα , ψyα are found from the following
linear equation system:
⎧      
⎪
⎪ δi1 ϕ pα1 pα1 + δi2 ϕ pα2 pα2 + ... + δiN ϕ pαN pαN −
⎪
⎪  
⎪
⎪
⎪ −WCα − ψxα · (xi − xC ) − ψyα · (yi − yC ) = −δi1 ϕ pα1 −
⎪
⎪
⎪    
⎪
⎪
⎪
⎪ −δi2 ϕ pα2 − ... − δiN ϕ pαN + WCα + ψxα (xi − xC ) +
⎪
⎪
⎪
⎪
⎪
⎪ +ψyα (yi − yC ) , i = 1,2,...,N;
⎨
pα1 S1 + pα2 S2 + ... + pαN SN = P − pα1 S1 − pα2 S2 − ... − pαN SN ,
⎪
⎪
⎪
⎪
⎪ pα1 S1 x1 + pα2 S2 x2 + ... + pαN SN xN = P · xC − My − S1 x1 pα1 −
⎪
⎪
⎪
⎪
⎪
⎪
⎪ −S2 x2 pα2 − ... − SN xN pαN ,
⎪
⎪
⎪
⎪
⎪
⎪ pα1 S1 y1 + pα2 S2 y2 + ... + pαN SN yN = P · yC + Mx − S1 y1 pα1 −
⎪
⎪
⎩
−S2 y2 pα2 − ... − SN yN pαN .
(4.12)
As an initial approximation, here we also take the solution, corresponding to the
case of a linearly deformable base. Due to the choice of such initial approxima-
tion and to the square convergence of the method, the succession of approximations
converges rather quickly (3–5 iterations) in a broad range of variation of the nonlin-
earity parameter m. Note that in the case of large values of m > 5 the approximation
processes considered give stable solutions together with the loading method with
the parameter m [184].
Evidently, the boundary-element solutions pi , i = 1, N, of the integral equation
system (4.10) have the physical sense of contact stresses. Indeed, since the function
ϕ (ξ ,η,p∗ ) is dimensionless, the dimensionality of p(ξ , η) does not change in case
a nonlinear model being used. Besides, the nonlinear contact problem solutions,
besides the integral representation of Eq. (4.8), with necessity satisfy the integral
[email protected]
equilibrium conditions (4.9). Finally, in the limiting cases p∗ →∞ or m→∞ the
solutions obtained coincide with the contact stresses in the classical problem of
theory of elasticity.
4.6 Numerical Algorithms of Solving Boundary Integral Equations 341

As follows from the theoretical study presented, the physical sense of the func-
tion ω(x, y, ξ , η) in the nonlinear contact model is not changed, i.e. it is supposed to
be used from the consideration of linear deformation laws. It should be emphasized
that the semiempirical model with the approximation of the loading function in the
form of Eq. (4.6) should be treated as a complex representation of the dependence
between the load and the settlement in the integral sense with invoking a dimension-
less loading level function ϕ (p,p∗ ), and the functions p(ξ , η) and ω(x, y, ξ , η) have
their initial sense each (from the linear theory) – contact pressure and the influence
function for a linearly deformable base without relation to any modifications.
According to the proposed numerical algorithm, the calculations are performed
using a unique procedure, applicable for the cases of linear and nonlinear deforma-
tion. The influence functions are specified in a separate software module what does
not require resetting of the whole algorithm.

4.6.4 Contact Problem for a Round Punch on a Nonlinearly

Deformable Base

Consider, as an example, a spatial contact problem for a circular punch, interacting

with a nonlinearly deformable base. We restrict our consideration by a punch loaded
only by a vertical force P, as it is conventional at punch tests [110, 263]. To solve
the problem numerically, the circular contact domain is discretized (Fig. 4.56) into
triangular and quadrangular elements, using radii and concentric circles, condensing
towards the boundary, what is related to a sharp change of the stressed state near the
punch edge. The radii of the concentric circles are calculated using the following
quasiuniform dependence:

[email protected]
Fig. 4.56 Discretization of a
circular contact domain
342 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

eβtj − 1 j−1
rj = α , tj = , j = 1,...,L.
eβ − 1 L

The condensation degree is controlled by the choice of the parameter β.

In the case when the influence function is symmetrically differential, ω(x, y, ξ ,
η) = ω(x–ξ , y–η), the system (4.10) for the round punch is essentially simplified
⎧
⎪ '
L  
⎪
⎨ Aij pj ϕ pj , p∗ − Wc = 0, i = 1,L;
⎪
j=1
(4.13)
⎪
⎪ '
L
⎪
⎩ pj Sj =
P·L
N
j=1

where the Aij values are found using the influence coefficients δ ij (i, j = 1,...,N) from
the formula


N /L
Aij = δi,j+L(k−1) .
k=1

The dimensionality of the system (4.13) is L+ 1 where L is the number of the

boundary elements in the radial direction.
Then consider the results of numerical modelling of the process of contact inter-
action of a round punch for the most popular models of elastic bases of a half-space
type and of a finite-thickness layer type. The significance of such studies follows
from the necessity of recovering nonlinear deformation diagrams from the punch
test data.
Nonlinearly deformable half-space. Take the advantage of the known Boussinesq
solution on the action of a unit concentrated force normally to the surface of the
elastic linearly deformable half-space [128]. The influence function here is given by

1 − ν2 1
ω (x,y,ξ ,η) =  . (4.14)
πE
(x − ξ )2 + (y − η)2

In the case of interaction of a centrally loaded punch with a linearly deformable

half-space, the problem has an analytical solution [98]
 
P 1 − ν2 P
WC = ,ψx = ψy = 0, p (r) = √
2Ea 2π a a2 − r2
[email protected]
where a is the punch radius, r is the distance from the centre to any point under
the punch. This solution was a test for the algorithm constructed. Contact stress
values of sufficient accuracy were obtained using β = –1.0 and the circle domain
4.6 Numerical Algorithms of Solving Boundary Integral Equations 343

Fig. 4.57 Contact pressures

under a round punch on a
nonlinearly deformable
half-space at P = a 2 p∗ , m =
0 (1), 0.8 (2), 3.0 (3)

Fig. 4.58 Relative

settlements of a round punch
on a nonlinearly deformable
half-space versus the
nonlinearity parameter m at
different loading values

discretization preformed using 96 elements, formed by six concentric circles and

sixteen rays (Fig. 4.56).
The results of the calculations of contact pressures and displacements for a rigid
punch on a nonlinearly deformable half-space are given in the dimensionless form
in Figs. 4.57–4.59. It is seen from Fig. 4.57 that with the increase of the nonlinearity
parameter m the contact pressure profiles are more flat, and the main change of the
contact pressures occurs near the punch edge. The punch relative settlements at a
given load increase nonlinearly with the m parameter increase, what is seen from
Fig. 4.58. Nonlinear settlement-versus-load dependences are plotted in Fig. 4.59 at
[email protected]
various values of the m parameter.
Nonlinearly deformable finite-thickness layer. The model of a linearly
deformable half-space, in spite of wide application, still possesses certain
344 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

Fig. 4.59 Settlement versus

load plots for a round punch
on a nonlinearly deformable
half-space at different values
of the nonlinearity parameter
m =0.2 (1), 0.4 (2), 0.8 (3)

shortcomings. In particular, the main of them is an assumption of the soil base

being a spatially homogeneous semi-infinite medium. This essentially idealizes the
situation and is not always confirmed at practical calculations. Application of a lin-
early deformable finite-thickness layer provides much better correlation with the
foundation settlements on a soil base and enables the calculated values of forces
and deformations in structures, resting on an elastic half-space, to be reduced.
In accordance with [203], the influence function ω(x, y, ξ , η), determining the
vertical displacements of the surface points of a linearly deformable finite-thickness
layer, caused by a unit normal concentrated force on its surface, is determined from
the following integral dependence:

∞
1 − ν2
ω (x,y,ξ ,η) = Q (H,t) · J0 (R · t) dt (4.15)
πE
0


where R = (x − ξ )2 + (y − η)2 , J0 is the first-order Bessel function of the 0th
order.
The kernel Q(t) of the integral representation (4.15) is given by

2 · sinh2 (Ht)
Q (H,t) = (4.16)
2Ht + sinh (2Ht)

where 0 < H < ∞ is the thickness of an unlimited elastic layer, resting on an abso-
[email protected]
lutely rigid base. Equation (4.16) was obtained using the Hankel transformation in
the assumption of the absence of friction forces at the contact of the layer with the
base. For a homogeneous elastic half-space Q = 1.
4.6 Numerical Algorithms of Solving Boundary Integral Equations 345

Table 4.11 Approximation parameters of the contact model for a constant-thickness elastic layer

k 1 2 3 4

Ak 0.8 1.4 2.0 2.6

Bk 0.426 −6.051 7.395 −2.770

The finite-thickness elastic layer model is in the mathematical sense more gen-
eral than the elastic half-space model. At relatively high layer thickness H >> a,
Eq. (4.15) is transformed into Eq. (4.14), i.e. calculations, based on Eq. (4.16), result
in solutions of contact problems for the half-space as well.
It is known [150], that in the case of interaction of a centrally loaded punch with
a linearly deformable finite-thickness layer, the contact problem solution is based
on its reduction to even integral equations. This enables the punch displacements
and contact stresses to be expressed in terms of an auxiliary function, satisfying
the integral Fredholm equation with a continuous symmetrical kernel. Though this
equation can be solved with a required degree of accuracy, this can be done only
numerically.
While solving the contact interaction problem for punches of arbitrary shape
with a finite-thickness elastic layer under a complex spatial loading, according to
the numerical algorithm proposed here, the use of the integral representation of
Eq. (4.15) is related to additional computational difficulties regarding the calcula-
tion of improper integrals containing oscillating Bessel functions. Therefore, when
contact problems for a finite-thickness layer are to be solved, it is convenient to
present the integrand of Eq. (4.16) in terms of a finite series of exponential func-
tions. This enables the improper integrals to be calculated analytically, and thereby
the computation accuracy to be increased and the computation time to be consider-
ably shortened. In particular, a highly accurate approximation

2 · sinh2 (α)  4
=1+ Bk exp (−Ak · α),
2α + sinh (2α)
k=1
obtained by the least-square method in [33], enables the approximate solution for a
concentrated normal force on the surface of a layer with a thickness H to be used in
the form [75]
 
1 
4
1 − ν2 Bk
ω (x,y,ξ ,η) = +  (4.17)
πE R (Ak H)2 + R2
k=1

where the coefficients Ak , Bk , k=1,4 are listed in Table 4.11.

The first term in Eq. (4.17) corresponds to the Boussinesq solution and the rest
[email protected]
of the terms, dependent of the layer depth H, can be treated as modifying terms. It is
important to note that the function Q(H, t) used is determined by one parameter of
length dimensionality (the thickness of virtually compressible soil mass H), which
346 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

Fig. 4.60 Relative settlements of a round punch on a nonlinearly deformable finite-thickness layer
versus the nonlinearity parameter m at different thicknesses of the elastic layer P = a 2 p∗

is the single geometrical parameter of the base. Due to this fact, the coefficients in
Eq. (4.17) are fixed and, what is quite important, independent of the characteristics
of mechanical properties of the base and its depth.
Numerical studies of the contact interaction of a round punch for the nonlin-
early deformable finite-thickness layer, performed using the algorithm proposed,
are in qualitative agreement with the calculation data for the half-space and are
partly shown in Figs. 4.60–4.62. As one should expect, the contact pressure profiles
become more flat both with the increase of the nonlinearity parameter m and with
the decrease of the compressible layer thickness H (or a parameter s =H/a). The
punch relative settlements under a given load nonlinearly increase with the increase
of the m and s parameters, not exceeding the corresponding values for the half-space
(Fig. 4.60).With the load increase this effect is revealed more noticeably (Fig. 4.61).
The nonlinear character of the settlement-versus-load plots at different values of the
m and H parameters is illustrated by Fig. 4.62.
As mentioned above, the calculations we have performed, are carried out in the
assumption that there is no friction between the elastic layer and the rigid base. The
presence of a contact surface between the elastic layer and the incompressible base
[email protected]
in the soil at the depth H results in a nonuniformity of the stress-strained state. The
studies, performed by a number of authors (See, e.g., [104]), have shown that the
influence of the incompressible base on the stress concentration within the layer
4.6 Numerical Algorithms of Solving Boundary Integral Equations 347

Fig. 4.61 Relative

settlements of a round punch
on a nonlinearly deformable
finite-thickness layer (H/a=
1.0) versus the nonlinearity
parameterm at different load
values

Fig. 4.62 Dimensionless

settlement-versus-load plots
for a round punch on a
nonlinearly deformable
finite-thickness layer W
=(1–ν 2 )ap∗ /2E

becomes negligible only at s = H/a >5. The concentration degree at s <5 essentially
depends on the conditions of the elastic layer sliding on the incompresible base.
In particular, a detailed analysis shows that the condition of absence of tangential
stress in comparison with the condition of the absence of displacements (restraint)
on the lower boundary of the layer results in an increase of vertical displacements
[email protected]
and compressive stresses. For real soil conditions, evidently, one should employ a
boundary condition of the type of elastic constraints between an elastic body and an
absolutely rigid body at their contact
348 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

1−ν
uj (x, y, H) = − μ · τjz (x, y, H) , j = x,y, uz (x, y, H) = 0,
G
proposed in [102], where μ is the coherence coefficient, resulting in an increase of
the stress and displacement absolute values. The limiting cases of such conditions
are coherence and smooth contact: at μ = 0 the case of coherence of the layer and
the base is obtained, at μ→∞ the layer and the base can slide without friction along
their whole common boundary.
A further approximation to the real deformation condition should be, evidently,
considered modelling of a natural base as a scheme of an elastic finite-thickness
layer on an elastic half-space. For example, when the elastic layer freely (μ = 0)
rests on the elastic half-space, the influence function is given by [11]

cosh 2α − 1 + χ (2α + sinh 2α)

Q (α,χ ) =  . (4.18)
2α + sinh 2α + χ cosh 2α − 1 − α 2

This base model contains a ratio χ =θ 1 /θ 2 as a parameter, where θi = Ei /2

(1 − νi2 ), i = 1,2 are the mechanical characteristics of the layer and the half-space,
respectively.
At χ →0 one arrives at the scheme of an elastic layer on a rigid base, at H/a→∞
– at the elastic half-space scheme.
According to the module structure of the numerical algorithm developed, appli-
cation of the influence function for the generalized model of a linearly deformable
base of Eq. (4.18) does not result in any principal computational difficulties, leading
only to the total increase of the computation time.

4.6.5 Estimation of Nonlinear Deformation Effects from Punch

Test Results
From the calculation data presented in Sect. 4.6.4 it follows that, at the identifica-
tion of the m, p∗ , and H parameters, the contact model under consideration can be
used to describe the deformational characteristics of bases as well as for the calcu-
lations of spatial contact interaction of foundation structures with soil at nonlinear
deformation stage.
The qualitative character of the calculated dependences (Figs. 4.59–4.62) indi-
cates the model applicability for the description of nonlinear deformation under
foundations for both clay soils of soft and tough plastic consistency and sand soils
of medium density with deformation diagrams without strengthening phenomena.
One of the possible algorithms of punch test data processing regarding the non-
linearly deformable base model under consideration can be the following.
First, the p∗ parameter value can be estimated from the results of traditional
[email protected]
punch tests [263], or the known formulae by Berezantsev [32] or Egorov [79] for
initial critical load on the soil in the round punch case can be used for the first
approximation.
4.6 Numerical Algorithms of Solving Boundary Integral Equations 349

Note that the calculation results, presented in Figs. 4.58, 4.60, and 4.61, are a sort
of loading diagrams for a load series Pt = Kt ×(a 2 p∗ ), i=1, 2, ...M (M is the number
of the experimental points on a settlement versus load plot). If the experiments for
the given loads Pi are carried out, then from the experimental data available (W)i one
can, using the calculation diagrams, easily determine the corresponding {mi } values.
Then a standard statistical processing of the obtained {mi } data array is performed
[199] and the most probable (average) value for it m is evaluated. If the experimental
data spread is small and the calculated data correspond to the experimental ones in
accordance with the Fisher criterion at the given significance level, then the model
based on the corresponding agreement criteria will be adequate to the experimental
data and the m value is taken as a nonlinear model parameter.
As follows from the presented theoretical analysis of the spatial contact interac-
tion at the nonlinear deformation stage, p∗ and m are not independent parameters of
the contact model, since the choice of the formula for the limiting contact pressure
p∗ (or its experimental determination) affects the m estimation.
The calculations performed (Fig. 4.62) are in agreement with the known results
of practical observations indicating that one of the factors, essentially affecting the
settlement value, is the depth H of the compressed soil base. It is determined rather
accurately when the specific geotechnical conditions are known. The most typi-
cal are situations when, for example, under a compressible soil at a certain depth
practically incompressible rock occurs; often under a structure season thawing out
occurs to a finite depth, below which ever-frozen soils are located. Then it is nat-
ural to consider the whole deformable layer to be the compressible mass. Besides,
there is a great number of calculation methods [104] to determine the compressible
mass thickness: from the foundation width, from punch tests in combination with
solutions for theory of elasticity, from the comparison of the natural and excessive
pressures, etc.
The most popular approach to obtain the thickness H is its conditional calcu-
lation as the depth, at which the excessive pressure from the punch (the founda-
tion model) is 10–20% of the natural value at the same depth. However, none of
the known approaches to the compressible base thickness assignment is free from
shortcomings. The values of the compressible soil mass depth, found for the same
conditions by different methods, sometimes differ by factor of 2–5. Thus, the com-
pressible base thickness is a conditional value being introduced into the calculation
due to the difference of the real conditions from the half-space calculation model.
This value should be assigned from the requirement of coincidence of the calculated
and the actual settlements. Therefore, it is convenient to include the H value to the
contact model nonlinear parameters and to determine it from the conditions of the
calculated settlements approaching the real values. At such approach, the nonlinear
character of the observed base thickness increase with load will be naturally taken
into account as well as the influence of the punch area, soil condensation and a
number of other factors.
[email protected]
The nonlinear parameters m, n, H of the model can be all together found by
means of minimizing the mean square error at the mathematical processing of the
punch test results by one of the nonlinear programming methods [60], e.g. by the
350 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

simplex method when a direct search is performed, using only the function values
(without derivatives).
One should pay attention to the relation of p∗ and m parameters to the form of the
ω(x, y, ξ , η) function. According to the linearly deformable base theory [105], the
influence function ω(x, y, ξ , η) determines the distributive properties of soil bases for
the calculation model which in practice substitutes a real natural base. Many of these
functions are at present widely used for the design and calculation of structures on
an elastic base and are included into the regulations in action. A linearly deformable
half-space, a linearly deformable finite-thickness layer, and the Fuss-Winkler model
are most widely used in the calculation practice. Influence functions for many non-
classical elastic bases have been obtained, with e.g. layered superposition of rock
strata [210], variation of the elastic properties with depth [73], etc. being taken into
account. Obviously, the choice of ω(x, y, ξ , η) at a given punch geometry determines,
at different p∗ and m parameter values, the characteristic settlement versus load dia-
gram and thereby the evaluation of the nonlinearity parameters following from the
experimental data processing. Calculations for the most widely used structures with
the influence functions of Eqs. (4.14) and (4.17) are performed in Sect. 4.6.4. The
choice of another ω(x, y, ξ , η) functions for the round punches will result in new
calculated deformation diagrams, based on which, after the punch test experimental
data processing, the model parameters p∗ , m, and n will be identified. The presence
of additional free parameters for the ω(x, y, ξ , η) function (e.g. the H parameter – the
compressible soil base depth, or γ – the degree of nonuniformity of the mechanical
properties with depth, etc.), enables the accuracy of approximating the experimen-
tal data to be increased. The example of a nonlinear contact problem for a finite-
thickness elastic layer, considered above in Sect. 4.6.4, clearly and convincingly (in
particular, the calculations presented in Figs. 4.60–4.62) shows a strong relationship
of the p∗ and m parameters with the influence function ω(x, y, ξ , η).
For the sake of completeness of the consideration one should make a note regard-
ing the choice of the influence functions ω(x, y, ξ , η) themselves. It is well known
[98, 128, 283] that until now the influence function ω(x, y, ξ , η), which would enable
the behaviour of various soils to be described in a broad range of deformation condi-
tions, has not been found. The results of comparison of different influence functions,
referred to in the ample literature, do not enable one to determine, which of the influ-
ence functions available is the best for a satisfactory description of displacements
and stresses for structures resting on elastic bases. Though each influence function
ω(x, y, ξ , η) has its own physical background, at present it is impossible to make a
preference for any of these functions. Therefore, the choice of the influence function
ω(x, y, ξ , η) does not extend beyond the researcher’s individual intuition based on
a rational compromise between the complexity of the mathematical representations
and the results of predictions regarding the functioning of bases for buildings and
structures. Note once again that the use of an influence function ω(x, y, ξ , η) essen-
tially affects the contact problem solution (Figs. 4.60–4.62), or, in other words, the
[email protected]
nonuniformity degree of the reactive pressure distribution, which, in turn, affects
the variation of the loading diagrams W(m, H, p∗ ). The latter serve to estimate the
nonlinearity parameters.
4.7 Contact Problem for Orthotropic Foundation Plates 351

It is important to know that the nonlinearity parameters m, H, and p∗ for a

semi-empirical model are found from the condition of the best description of the
experimental data at the non-linear stage of deformation. Such approach is gener-
ally accepted and widely used at the analysis of nonlinear mechanical models. The
approach is most convincingly demonstrated in a number of publications (See, e.g.,
reviews [87, 181]), devoted to the issues of construction of nonlinear rheological
calculation models of soil media. A lot of other references can be quoted, where the
efficiency of the semi-empirical approach to describe various nonlinear mechanical
phenomena is clearly demonstrated. Concerning the calculation of structures on an
elastic base, two most important studies should be pointed out [282, 290]. In these
publications, the ideology being used in this chapter, had been applied for the cal-
culation of beam-type and plate-type structures, resting on an elastic nonlinearly
deformable base.
Thus, a semiempirical contact model of a soil base is proposed, relating the non-
linearity parameters p∗ , m, n with the distributive properties of the base. In the limit-
ing cases (p∗ →∞, or m→0) the contact model is linear, predicts a linear settlement-
versus-load diagram and is fully characterized by the chosen influence function
ω(x, y, ξ , η) for a linearly deformable base.
It is quite important that while formulating the calculation model proposed, con-
trary to the classical integral representations of theory of elasticity, the issue of the
relationship between stress and deformation is not discussed. This is the specific
feature and a certain advantage of the nonlinear integral representation of Eq. (4.6),
relating the displacements and contact pressures only on the elastic base surface. The
dimensionality of the foundation base calculation problems is reduced, and in a spa-
tial (three-dimensional) formulation they become two-dimensional. One can treat
the situation as a quite similar one to the application of the calculation models of
one or two coefficients of subgrade reaction (Fuss-Winkler, Pasternak, or Filonenko-
Borodich) which also do not imply the corresponding dependence between the stress
and deformation tensor components. For practical engineering calculations, nonlin-
ear contact models with coefficients of subgrade reaction depending on contact pres-
sures or displacements, are successfully used [137, 139], and the variety of the soil
properties and the complexity of their deformation processes quite justify the fur-
ther development of the calculation methods invoking various modified calculation
models of contact interaction.

4.7 Contact Problem for Orthotropic Foundation Plates

with the Account of the Specific Features of Spatially
Nonhomogeneous Base Deformation
Reinforced concrete plates are structural parts, widely used in practical engineering.
[email protected]
When the plates serve as a foundation for a building or a structure, their functioning
essentially affects the stress-strained state of the whole subterranean structure. The
calculated values of separate structural parts in this case to a great extent depend on
the soil base calculation model being employed.
352 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

4.7.1 Static Calculations of Foundation Plates on Elastic Bases

Reinforced concrete plate foundations of a rectangular, round, and ring-shaped
cross-section are extensively applied for construction on natural bases (especially,
on weak and nonhomogeneous soils) of multistorey buildings, elevators, funnels as
well as dock bottoms, sluices, floors of industrial structures, airfield coverings, etc.
[125].
Theoretical studies of plate foundation calculations are characterized by a quite
great number of publications, based on some of which the computer software for the
plate foundation calculations was worked out [105, 125]. Contrary to the traditional
methods of calculations of proper plates of different shape and thickness, the calcu-
lations of the plate foundation structures are more complicated due to the necessity
of account of the influence of a rigid above-foundation structure of the frame type, a
system of constrained diaphragms, carrying walls of a very rigid above-foundation
structure of funnel type, etc. The main specific features of the calculations result
from the non-uniform base compressibility due to the natural soil nonhomogeneity
within the plate foundation structures with large area.
The most popular scheme for the calculation of foundation plates on elastic bases
is the one for the base with a variable rigidity coefficient which approximately takes
into account the base nonhomogeneity with depth and over the area. Approaches,
using this calculation scheme, as well as related ones based on the application of
contact models with one or two coefficients of subgrade reaction (Winkler-type
models), have been applied since the 1930s and are still used in the design prac-
tice. They are reflected in standards and regulations (See, e.g., [120, 121, 125]), and
their theoretical substantiation with calculation examples is presented in detail in a
vast number of books [5, 130, 134, 136, 143, 148, 189, 190, 257, 273, 274, 275,
280]. Note a publication [66] on an exact solution, directly related to the theory of
design of rectangular foundation plates on Winkler-type elastic bases, which has not
been included into books or handbooks, where a solution of the problem of bending
of a plate with free edges, resting on a Winkler base, under a symmetrical load. This
solution has not attracted the attention of design engineers due to the complexity
of the expressions obtained for deflections, moments and transverse forces in the
form of dual Fourier series whose coefficients should be found from the solution of
infinite systems of linear algebraic equations. For practical purposes the foundation
plate calculation by one of numerous methods (finite-difference, finite- or boundary-
element) always appears more convenient due to its universality for the formulation
of boundary conditions of various type, for the account of the complicate plate con-
figuration and for a combined action of force and momental loads.
In spite of the disadvantages of the hypotheses concerning one and two coef-
ficients of subgrade reaction, revealed in ambiguous dependences of the model
parameters on the method of their determination, in the approximate account of the
base distributive properties, in an additional concentrated lineal reaction along the
[email protected]
plate contour, the base models with the use of the rigidity coefficients still remain
applicable for practical calculations in view of their mathematical simplicity and
convenience for software development.
4.7 Contact Problem for Orthotropic Foundation Plates 353

Most of the above disadvantages of the Winkler-type elastic base contact models
are overcome by using the base models, representing a spatial elastic continuum – an
elastic half-space [105] or an elastic layer of constant and finite thickness [75]. Note
that the model of an elastic layer, underlayered by an incompressible base, closer
corresponds to the soil base properties and possesses the highest generality, since
at considerable layer thicknesses it is reduced to the elastic half-space model, while
at low thicknesses it is similar to the models of Winkler type (with an appropriate
choice of the coefficients of subgrade reaction). Foundation plate calculations, using
the contact models of spatial continuum type, enable the base distributive properties
to be taken into account (formation of a sedimentary boot). While describing the
base properties, these models take into account the deformation modulus and the
Poisson ratio as parameters being, contrary to the coefficients of subgrade reaction,
physical characteristics of the soil. It is especially important that the application of
the elastic half-space and finite-thickness layer models enables the stress in the soil
mass environment to be determined more reliably, the effect of foundations on each
other to be taken into account, etc.
Theoretical studies devoted to the application of base models in the form of a
half-space and a finite-thickness layer for the calculations of foundation plates, have
been performed mostly by Russian researchers and are considered in detail in [105,
106, 291].
Gorbunov-Posadov [105] elaborated a method for approximate calculation of
round and rectangular plates of various flexibility on an elastic half-space, based
on the expansion of the plate deflections as a two-dimensional polynomial with
unknown coefficients, determined by satisfying the plate bending differential equa-
tion, the boundary conditions on its contour, the equilibrium equations and the iden-
tity of the settlements of the plate and the base (half-space). A similar approach
with some improvements in the technique of formation of the resolving system of
linear equations was later implemented by Gorlov and Serebryanyi [106] for the
calculations of rectangular foundation plates, resting on an elastic layer of a finite
thickness.
In a method proposed by Zhemochkin [292], for the determination of the con-
tact stress between the foundation plate and the base, vertical rod constraints are
introduced, the forces in which are the resultants of the stresses in the soil near the
surface of the foundation and the base contact. In such formulation the problem
of calculation of the structures and the foundation is reduced to the determination
of the forces in the constraints, i.e. to the construction and solution of a canonical
system of equations of the mixed method of engineering mechanics.
The Zhemochkin approach is more labour-consuming than that of Gorbunov–
Posadov, since it is related to much larger computation scope. However, it also pos-
sesses higher generality, enabling the variable rigidity of the plate to be taken into
account as well as the base calculation model to be somewhat modified. The com-
parison of both methods for beams working at the conditions of the spatial problem,
[email protected]
i.e. rectangular plates with the side ratio 7:1 and higher, has shown the difference
in the application of these calculation methods to be small [106]. From the contem-
porary point of view, the method, proposed by Zhemochkin for a half-space, is in
354 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

fact the simplest form of the boundary-element method with a piecewise constant
approximation of the contact pressure function.
Already in 1960 Solomin was one of the first to use the finite-difference method
for the calculation of rectangular plates on an elastic half-space [250]. Later, finite-
difference calculations for rectangular plates, resting on a finite-thickness elastic
layer, were performed [251]. Further, the application of the variation-difference
approach [271] has enabled Solomin et al. to extend the possibilities of the grid
method to calculate complex-shaped foundation plates [253, 254]. The elaborated
method of calculation appeared very efficient, and subsequently served as a basis
for creation of a great number of professional software for foundation plate compu-
tations [105, 125].
The finite-difference method was of great importance at earlier stages of com-
puter development, since the difference equation systems, having a simple and reg-
ular structure, can be efficiently solved using less powerful computer systems.
The book by Kaczkowski
˛ [130], which is one of the most complete studies
concerning static plate calculations, ranks the finite-difference method as the first
among the approximate calculation methods. A special chapter there, devoted to the
fundamentals of the finite-difference method application, proves that computations
using the finite-difference method, do not encounter any serious programming dif-
ficulties, and the use of a sufficiently fine grid can provide results with an accuracy,
quite sufficient for the practical purposes, to be obtained. The same chapter quotes
the main studies concerning the finite-difference method application for plates of
specific shapes (triangular, hexagonal, parallelogram-type, skew-angle).
The most significant among the recent papers, using the finite-difference method
to calculate foundation plates on bases in the form of a rigid body, are [58, 126].
Reference [58] considers a plate (of beam type) on an elastic half-space, for its cal-
culation a 40 × 10 finite-difference grid being employed. Variational principles are
invoked to determine the contact domain of the plate and the base. An intentionally
developed finite-difference software is used in [126] to perform extensive calcula-
tions, using a succession of condensing grids, of processes of interaction of a rect-
angular plate with an elastic half-space. A quite good convergence of the numerical
results for the basic types of the external load (angular, edge, internal) is shown. The
estimations of the finite-difference method efficiency have shown the computation
time decrease by factor of 2–3 in comparison with the finite-element algorithms of
a comparable accuracy. It is concluded that even now (in spite of the known popu-
larity of the finite-element method) the finite-difference method is an effective tool
for geotechnical calculations, especially for multiparametric studies.
Meleshenkov and Ozherelyev [168] present the results of a numerical experi-
ment, performed for an example of plate bending problem solution by the finite-
difference and finite-element methods. Depending on the density of a rectangular
uniform grid, imposed on the plate, the accuracy of the solution and the number of
arithmetic operation performed at the stage of solving the algebraic equation sys-
[email protected]
tems was compared. The finite-difference method is shown to be quite competitive
with the finite-element method, and in some cases (a hinged plate under a uniformly
distributed load) is has advantage over the finite-element method both in accuracy
4.7 Contact Problem for Orthotropic Foundation Plates 355

and in the operation number. The finite-difference method is concluded to be rather

efficient for solving some classes of problems of engineering mechanics and still
remains a powerful tool for solving partial differential equations.
Application of the finite-element method for the calculation of foundation plates,
resting on an elastic half-space, was first described in [65]. A quite detailed dis-
cussion of the calculation results obtained until 1979, including those obtained by
the finite-element method, for foundation plates, resting on an elastic half-space,
was performed by Selvadurai [234]. Due to the convenience of application of the
finite-element method for the calculation of complex-shaped plates and simulta-
neous account of various nonlinear laws of their deformation, a lot of the devel-
oped software tools for foundation plate design use the finite-element method as
the computation basis. However, the search performed has not revealed any profes-
sional software simultaneously including the use of the finite-element method and
the elastic continuum models. Probably, this is explained by a sharp increase of the
computer resources required and, consequently, by low efficiency of this type of
algorithms in computer-aided design systems.
Among the publications, devoted to the finite-element calculation of of founda-
tion plates, interacting with soil bases modelled by elastic spatial continua, some
theoretical studies of the recent years should be mentioned [184, 206, 233, 237].
Contact interaction of two square plates, one of them being homogeneous, the
other – three-layered, with an elastic half-space under a concentrated force, applied
to the centre, is studied in [206]. Rectangular finite elements with a linear reactive
pressure distribution were used.
In order to determine the deflection of a rectangular plate on an elastic half-space,
Novotný and Hanuška [183] employed triangular finite elements, for which the plate
deflection function is given by a full polynomial of the 5th order. Appearance of
tensile contact stresses is excluded (the account of unilateral constraints), and the
solution is found by iteration.
The Reisner square plate on an elastic half-space is calculated in [233] by using
square isoparametric elements. The plate was supposed to bear on two opposite
sides, the other two being gripped, and in the centre the plate bearing on a pole.
The author claims this method, consisting in a choice of interpolation of the same
order for the finite element geometry, displacement and reactive load fields, to result
in a more exact and general solution than the known ones, since this solution is
applicable not only for thin, but also for thick plates.
Contact pressure distribution under plates, resting on an elastic half-space, is
investigated in [237]. The finite-element formulation of the problem is given with
the application of eight-node isoparametric quadrangular elements. The numerical
algorithm developed is capable of taking into account various shape of plates, whose
material properties and thickness can vary from one element to another.
In the case of large areas the assumption of the soil working as a homogeneous
isotropic elastic half-space is known to result in the overestimated values of deflec-
[email protected]
tions and bending moments in the foundation plates [105]. One of the reasons is
a rather pronounced nonuniformity of the elastic properties of most soils over the
depth. The most popular way of taking the base homogeneity into account is the use
356 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

of the deformation modulus, varying according to a power law. The layered super-
position of soils is often taken into account as well as the Poisson ratio variation with
depth. However, the rejection of the classical theory results in considerable compli-
cations of the calculation formulae, and only a small number of papers is devoted
to the plate calculation with the account of the specific features of deformation of
bases with variable physical and mechanical characteristics. In [73], the calculation
for a rectangular plate, the soil deformation modulus under which varies with depth
according to a power law, was performed by the finite-difference method. Mostly
the studies are devoted to the calculations for round and ring-shaped plates, resting
on bases with variable physical and mechanical characteristics under axisymmetric
loading [4, 11, 38, 119, 164]. From the results of these studies it follows that the
account of the increase with depth of the physical and mechanical parameters of the
base can result in a reduction of the calculated values of the bending moments by
up to 20%.
It follows from the analysis of the available literature that the issues of account
of the base variable compressibility over the area, using the base continuum mod-
els for plate foundation design, have not been sufficiently developed. This results
in a low reliability of the calculations for foundation plates, located on slope edges,
near high-angle soil layers, on the bases with the compressible soil mass thickness
increasing (decreasing) in a certain direction, as well as for a number of other com-
plicated geotechnical conditions when the compressible soil mass thickness varies
within the foundation plate area [125].
In the great majority of the calculation schemes at the studies of interaction of
foundation plates with a soil base the plate material is assumed isotropic. How-
ever, even conventional ferroconcrete plates, reinforced in different directions, are
characterized by different rigidity at bending, depending on its direction. This was
noticed already by Huber [118] and Marcus [163], considering a ferroconcrete plate
with orthogonal reinforcement as an orthotropic one. The anisotropic properties of
plates should be taken into account in order to carry out more reliable calculations
of mutual deformation of the above-foundation structure and the plate on a com-
pressible base, to choose the reinforcement, to check the crack widths, to determine
the calculated forces in the foundation.
Theory of calculations for anisotropic plates has been developed rather long ago
and is in quite a detailed way described in a number of books, among which some
are worth to be mentioned first of all [23, 152]. A particular case of anisotropy
is orthotropy, which is characterized by three perpendicular directions of elastic
symmetry in a plate. In practical engineering this type of anisotropy is the most
common.
The orthotropic properties of plate foundations are in most cases induced by rein-
forcement, by smooth variation of thickness in a rather narrow range, by a one-side
finning in one or two directions, by formation of cavities, by construction of box-like
structures with ribs in two directions, etc. [105, 130]. The systems under consider-
[email protected]
ation are, in general, plates of isotropic materials, but their bending rigidity varies
over the area due to the variation of the plate cross-section. Many researchers come
to a conclusion that the use of the mechanical model of a homogeneous orthotropic
4.7 Contact Problem for Orthotropic Foundation Plates 357

plate with constant rigidities for the consideration of an elastic system, nonhomo-
geneous over the area, is evidently convenient for calculations (in case the averaged
rigidities being determined correctly). The formulae to calculate the averaged elas-
tic parameters of orthotropy for the above noted orthotropic systems, depending on
the geometrical characteristics of the cross-section and the mechanical parameters
of the plate material, are given, for instance, in [130].
In spite of the lack of principal difficulties for the calculation of anisotropic plates
and the importance of the account of the foundation plate rigidity variation under
bending in different directions, the programs, intended to be used for wide appli-
cation in this field [105, 125], still do not take into account this orthotropy factor
which is important for the foundation plate design.
Theoretical studies of calculation of orthotropic foundation plates are rather rare
[129, 138, 247] and do not use, with a very rare exception [69], the soil base calcu-
lation models of spatial continuum type.
Klepikov and Malikova [138] formulated a problem of an orthotropic plate bend-
ing on an elastic Winkler base with a coefficient of subgrade reaction, variable in
the plane of the contact between the plate and the base. The base displacements
(subsidence) are taken into account, which essentially influence the stress-strained
state of the whole structure, erected on underworked territories and collapsing soils.
Unfortunately, the publication does not contain the calculation results.
The major part of a book by Kiselev [134] is devoted to the calculation of
orthotropic rectangular plates, resting on an elastic base with two coefficients of
subgrade reaction. It is considered that the plates can be supported differently at all
four sides (hinge or fixation). The solutions are found, depending on the type of
the boundary conditions in dual Fourier series or by an originally elaborated initial
parameter method. In order to obtain the solution, a preliminary expansion of the
external load in a Fourier series is required.
Smirnov [247] suggested to use an original numerical method employing differ-
entiation matrices [246] to perform calculations for orthotropic rectangular plates,
resting on an elastic base with a coefficient of subgrade reaction, variable over the
area. The problems are solved for the case of the plate loading by a uniform load in
case of a hinge-type support on a contour.
Kocatürk [141] performed calculations for a rectangular orthotropic plate with
free edges, located on an elastic unilateral Winkler base and subject to a uniformly
distributed load, a concentrated force and a moment at arbitrary points. The solution
is obtained using the generalized Galerkin method, resulting in infinite systems of
linear algebraic equations. The calculations have shown an essential influence of
the orthotropic properties of the plate on the characteristics of the processes of its
mutual contact deformation with the base.
The finite-element algorithm of calculation for a rectangular orthotropic piece-
wise homogeneous plate, resting on a Winkler base and undergoing an arbitrarily
distributed normal load, is worked out in [129]. Software, developed on the basis of
[email protected]
this algorithm, enables one to determine deflections, bending moments and stresses
in the plates of road covering or floors of engineering structures, accepting the
applied load and the base reaction.
358 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

The subsequent Sects. 4.7.2–4.7.4 are devoted to the account of the soil base
compressible thickness variation and influence of conditions of different type on
the stress-strained state of the orthotropic rectangular foundation plate structures.
Such formulation of the contact problem is performed for further approaching of
the calculation data to the real conditions of foundation plate interaction with non-
homogeneous and structurally unstable soils, when the account of the real non-
uniform compressibility of the base under the large-size foundation plates bottom
is required. The calculation method employed is applicable for any known contact
models for elastic bases. As examples, numerical results for spatially nonhomoge-
neous bases of constant- and variable-thickness elastic layers are considered [75,
211]. No restrictions are assumed to be imposed on the characteristics of the plate
flexibility, related to the relation of the deformation properties of the plate material
and the base. An external transverse load on the plate can be given according to an
arbitrary law.

4.7.2 System of Integro-Differential Equations of Bending

of a Plate, Resting on an Elastic Base
Assuming the plate deflections small in comparison with its thickness, we consider
the Kirchhof–Love hypotheses to be true. Then the differential equation for the
curved surface of an orthotropic plate under a transverse load, resting on an elas-
tic base (Fig. 4.63), is given by [152]:

∂4 W ∂4 W ∂4 W
D1 + 2D3 + D 2 = q(x,y) − p(x,y) (4.19)
∂x4 ∂x2 ∂y2 ∂y4

where W(x, y) is a vertical displacement of the median plane of the plate, q(x,y) is
the external load intensity, p(x,y) is the contact pressure, D1 and D2 are cylindrical
bending rigidities for the basic directions of elasticity, Di = Ei h3 /12(1 − ν1 ν2 ),
i = 1, 2, D3 = D1 ν2 + 2Dk = D2 ν1 + 2Dk , Dk = Gh3 /12 is the torsion rigidity,

[email protected]
Fig. 4.63 Calculation
scheme for an orthotropic
plate, resting on an elastic
base
4.7 Contact Problem for Orthotropic Foundation Plates 359

E1 , E2 , ν 1 , ν 2 are the elastic moduli and the Poisson ratios of the plate material,
respectively, G is the shear modulus, h is the plate thickness.
For an orthotropic plate [152] the following relation holds:

E1 ν2 = E2 ν1 .

In the isotropic plate case

E1 = E2 = E, ν1 = ν2 = ν, G = E/2(1 + ν),

and all the rigidities are reduced to a single one

D1 = D2 = D3 = D = Eh3 /12(1 − ν 2 ).

Similarly to the traditional approaches [105, 125, 252], friction at the contact
domain is not taken into account; it is also assumed that the plate is completely
adherent to the base, i.e. the vertical displacements of the plate and the base surface
are equal. For the spatial contact problem the equality of the vertical displacements
of the plate and the elastic base surface results [252] in an integral equation to deter-
mine the contact pressure
 
1 − ν02
W(x,y) = W(x,y) + A + B · x + C · y = ω(x,y,ξ ,η)p(ξ ,η)dξ dη (4.20)
π E0
S

where A, B, C are the parameters of the plate displacement as a rigid solid, E0 and
ν 0 are the elastic modulus and the Poisson ratio for the base, S is the domain of the
plate contact with the base. The function ω(x, y, ξ, η) is given in accordance with
the elastic base model being used.
The boundary conditions at the plate contour age given by
+
∂W ++
W|1 = 0, = 0 (pinching), (4.21)
∂n +1
W|2 = 0, Mn |2 = 0 (hinge), (4.22)
 +
∂Mtn ++
Mn |3 = 0, Qn + = 0 (free edge) (4.23)
∂s +3

where Mn and Mtn are the bending moments and torques, respectively, Qn is the
shearing force, ∂/∂n is a derivative over the normal, ∂/∂s is the derivative over the
plate contour arc.
Note that the second equation in Eqs. (4.23) combines two necessary conditions
[email protected]
Qn = 0, Mtn = 0 on a free smooth contour of the plate when the distributed torque
pairs are statically equivalent to the shearing force. If the free contour contains angu-
lar points, such kind of substitution results in concentrated forces, appearing in the
360 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

angular points [134, 157, 224, 261]. For example, in a detailed notation the formula-
tion of the boundary conditions (4.23) for the most popular case of foundation plates
of rectangular shape (a and b being the plate sides) with the account of the absence
of concentrated forces in the angles, proportional to the torques

∂2 W
Mxy = Myx = −2Dk ,
∂x∂y
takes the following form with respect to the deflections: at x= 0, x= a

∂2 W ∂2 W ∂3 W ∂3 W
+ ν 2 = 0, + ε 2 = 0; (4.24)
∂x2 ∂y2 ∂x3 ∂x∂y2

at y= 0, y= b

∂2 W ∂2 W ∂3 W ∂3 W
+ ν1 2 = 0, + ε1 2 = 0; (4.25)
∂y 2 ∂x ∂y 3 ∂x ∂y

at x= 0, y= 0; x= 0, y= b; x= a, y= 0; x= a, y= b

∂2 W
= 0. (4.26)
∂x∂y

Here the notations ε1 = (D3 + 2Dk )/D2 , ε2 = (D3 + 2Dk )/D1 are introduced.
The problem is closed by a system of equilibrium equations for the plate, loaded
by an external distributed load q(x, y):
   
p(ξ ,η)dξ dη= q(ξ ,η)dξ dη,
 S F 
p(ξ ,η)ξ dξ dη = q(ξ ,η)ξ dξ dη, (4.27)
 S  F
p(ξ ,η)ηdξ dη = q(ξ ,η)ηdξ dη
S F

where S is the domain of the plate contact with the base, F is the domain of action
of the external distributed load.
Thus, the mathematical formulation of the problem under consideration is
reduced to the combined solution of a differential equation (4.19) and integral equa-
tions (4.20), (4.27) with the boundary conditions (4.21–4.23) on the plate contour.
The values of moments and transverse forces after the contact problem solution
[email protected]
are determined by differentiating the deflection function and can be found from
[152]
4.7 Contact Problem for Orthotropic Foundation Plates 361
   
∂2 W ∂2 W ∂2 W ∂2 W
Mx = −D1 + ν2 , My =−D2 + ν 1 , Mxy
∂x2 ∂y2 ∂y2 ∂x2
∂ W
2
=−2Dk ,
∂x∂y
   
∂3 W D3 ∂ 3 W ∂3 W D3 ∂ 3 W
Qx = −D1 + , Qy = −D2 + . (4.28)
∂x 3 D1 ∂x∂y2 ∂y3 D2 ∂x2 ∂y

4.7.3 Calculation of Rectangular Orthotropic Plates Based

on Combining Finite-Difference and Boundary-Element
Methods
Solving the spatial contact problem, formulated in Sect. 4.7.2 for any elastic base
model requires considerable difficulties; for rectangular plates no exact solutions are
known.
Our numerical calculations will be based on a combination of the grid method
and the boundary-element method. A finite-difference grid of regular structure with
the cell size of x and yin the direction of the coordinate axes, is imposed on the
plate (Fig. 4.64). The finite-difference equation system will be given by

[H] {W} = {q} − {p} .

Here [H] is the differential operator matrix with respect to the deflection vec-
tor {W}; {q} and {p} are the vectors of the external normal load and the reac-
tive pressure, considered in the finite-difference grid nodes. The details of the
finite-difference representation of the differential equation (4.19) and the boundary

[email protected]
Fig. 4.64 Contact domain
discretization
362 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

conditions (4.23) for orthotropic rectangular plates with free edges are given in
Appendix E.
In order to obtain a matrix analogue of the integral equation (4.20), along with
the regular finite-difference grid, consider an auxiliary grid (denoted by a dotted line
in Fig. 4.64) whose cells are formed by straight lines, passing through the median
points of the main grid cell sides. In assumption of the contact pressures being con-
stant within each cell of the auxiliary grid, the integral equation (4.20) can be pre-
sented in the following discrete form of canonical equations:

{W} = [B] {p}

( (
where [B] = (δij ( is the influence matrix, the elements of which are calculated as
 
δij = ω(xi ,yi ,ξ ,η)dξ dη
Fj

and represent vertical displacements of the elastic half-space surface in the point
(xi , yi ), coinciding with the i-th element gravity centre, due to a unit load, equally
distributed over the area Fj of the j-th element. The elements of the matrix [B] are
calculated by a procedure, described in detail in [15]. Solving the matrix equation
(4.30) with respect to the contact pressures {p}, one obtains

{p} = [V] {W}

where [V]=[B]−1 is the elastic base rigidity matrix. By substituting {p} from
Eq. (4.31) to Eq. (4.29), one obtains the solution of the problem under consider-
ation

{W} = [#]−1 {q} , [#] = [H] + [B]−1 .

The contact pressures {p} for the obtained solution {W} are obtained using
Eq. (4.31), and the distribution of the force factors in the plate is found based on
a finite-difference approximation of the dependences (4.28).
After discretization the linear equation system for the deflection function in the
node points of the plate is given by

[#] {W} = {q}

where {W} is a vector of unknown deflections in the plate nodes, {q} is a vector of
generalized load parameters, [#] is a matrix of the “plate + elastic base” contact sys-
tem, being formed of the coefficients of the canonical equations of the force method
[email protected]
for Eqs. (4.20), (4.27) and the coefficients of difference equations for Eq. (4.19).
Thus, the elements of the matrix [S] depend on the stressed states of the base and
4.7 Contact Problem for Orthotropic Foundation Plates 363

the plate. The system (4.32) is solved by the Gauss method with elimination of the
main element in a row.
When the Gauss method is implemented for computations, arithmetic operations
are performed by the processing unit with a fixed accuracy with, as a rule, 7 digits
at single-precision and 16 digits with double-precision calculations. In the first case
it results in rounding errors due to the truncation or rounding of the data and due to
the accumulation of errors in the course of solving the problem. With the increase of
the amount of unknowns, single-precision calculations may be not sufficient for the
Gauss method. Then one should use the double-precision solution of the problem or
apply a subsequent correction based on an iterative process.
At the single-precision calculations, due to the finite data length and an approx-
imate calculation of derivatives, practically the computer is used to solve a
system

(# + #) (W + W) = (q + q) .

The error of its solution is given by

# q
≤ σ (#)
# q
( (
where σ (#) = # · (# −1 ( is the conditionality number [93]. If σ(#) is great then
the algebraic equation system is badly conditioned, and one should apply special
methods of regularization in order to solve it.
In practical computations for the matrix of the linear equation system (4.32) we
calculated its conditionality number s which shows how far are the results from the
case of degeneracy. The estimations of conditionality, performed on the basis of
numerous calculations, has shown that, though the matrix of the system (4.32) is
nor sufficiently conditioned, still solving the equation system (due to the diagonal
predominance of the influence matrix B) by the Gauss elimination method leads
to quite quite acceptable results as the first approximation. In order to improve the
solution obtained, it was further processed by iterative method. An extensive series
of the calculations performed has shown that after 3–5 iterations the solution, as a
rule, converges within the machine accuracy ε =10−5 .
The numerical algorithm described above is implemented in a software module
ORTOPLIT, written using the FORTRAN language, and includes formation of the
system matrix and iterative refinement of the calculation results.
In the next subsection, using the calculation data obtained, the effect of spatial
nonhomogeneity of the compressible soil mass, external transverse load distribu-
tion, different combinations of the plate fixation at its contour, and the degree of
orthotropy of its material on the stress-strained states of the plate and the base is
[email protected]
analyzed. An elastic linearly deformable half-space, a constant-thickness layer, and
an elastic compressible wedge are considered as the soil base models.
364 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

4.7.4 Examples of Numerical Modelling of the Contact Interaction

of Plates with Elastic Bases
Comparative calculations for an isotropic plate on an elastic half-space. In order to
estimate the convergence of the numerical solutions and to illustrate the ability of
the method proposed, consider a known example of calculation of a plate with free
edges, resting on an elastic half-space [105, 183].
The input data for the calculation are the following: a reinforced concrete plate
with the elastic module E = 2.65·× 104 MPa, Poisson ratio ν= 0.1667 and size 2
× 2 × 0.2 m, is located on a soil base which is treated as an elastic half-space with
the characteristics E0 = 49 MPa, ν = 0.4.

3  
The plate flexibility index r = · π E0 a3 1 − ν 2 /Eh3 1 − ν02 ≈ 10.
2
The plate is subject to a uniformly distributed load of the intensity q = 1 MPa.
The results of calculation of deflections, contact pressures, and bending moments
in the plate centre along with the information on the conditionality number of the
matrix of coefficients of the linear algebraic equation system (4.32) at various den-
sity of the finite-difference grid are shown in Table 4.12. For the sake of comparison,
the table also contains the results, obtained by the finite-element method [183] and
based on the Gorbunov-Posadov solution [105].
As follows from Table 4.12, with the increase of the discretization degree the
accuracy of the obtained results increases; for the grids with the density of not less
than 8 × 8 it is sufficient for carrying out engineering calculations. One can also
easily estimate the limiting values from the data of Table 4.12. For the case of a
linear dependence of the solution on n −1 where n× n is the grid size, at n = 8 and
n = 10 one obtains

Table 4.12 Characteristics of convergence of numerical solutions for a square plate on an elastic
half-space

n×n σ(#) Wc , m pc /q 4Mx /a 2 q

Finite-difference method
4×4 0.6487 × 103 0.02401 0.5204 0.07655
6×6 0.2206 × 104 0.02625 0.5286 0.08559
8×8 0.5826 × 104 0.02744 0.5448 0.08865
10 × 10 0.7442 × 104 0.02817 0.5539 0.08975
12 × 12 0.1435 × 105 0.02867 0.5598 0.09010
14 × 14 0.2449 × 105 0.02904 0.5647 0.09026
16 × 16 0.3847 × 105 0.02928 0.5671 0.09008
18 × 18 0.6694 × 105 0.02951 0.5697 0.08998
20 × 20 0.9135 × 105 0.02968 0.5719 0.09017
[email protected]
22 × 22 0.1217 × 106 0.03003 0.5780 0.09092
FEM [183], 10 × 10 0.03063 0.5747 0.1038
Gorbunov-Posadov [105] 0.0327 0.630 0.061
4.7 Contact Problem for Orthotropic Foundation Plates 365

WC = 0.03109 m, p = pc /q = 0.5903, M x = 4Mx /a2 q = 0.09415.

The corresponding limiting values, obtained using the finite-element method

[183], are the following:

WC = 0.03115, m, p = 0.5815, M x = 0.0899.

The difference of the limiting calculated values, obtained by the two numerical
methods, is seen not to exceed 1.5–4.5%. The result obtained enables a sort of the
optimal density of the finite-difference grid to be chosen as 10 × 10, at which the
accuracy of the deflection calculations is achieved, exceeding 92%.
Note that, contrary to the finite-difference method, application of the finite-
element method to solving the equations of theory of thin plates is encumbered
by a high degree of derivatives, contained in the initial equations and, consequently,
in the expression for the potential energy functional. This results in the application
of complex elements with a great number of degrees of freedom and is equivalent
to the increase of the number of unknowns. The comparison of the finite-element
and finite-difference methods that we have performed shows that both methods give
practically the same accuracy of the solution. However, the finite-difference method
applied here is expressed by a more compact linear equation system and requires
essentially less computation time and computer RAM size. We should also, as noted
in [183], point out an insufficient accuracy of the Gorbunov–Posadov solution, what
is related to the use of partial sums of power series with insufficient number of terms
for the sought functions.
Orthotropic plate with free edges on a variable-thickness elastic layer. Applica-
tion of the proposed algorithm to estimate the effect of the thickness and nonho-
mogeneity of the compressible base on the plate bending will be illustrated by the
results of calculations for a square plate of reinforced concrete (E = 2×104 MPa,
ν = 0.167) with a side a = 16 m and thickness h = 1 m. For the account of the
orthotropic properties of the plate we imply E1 = 2E, ν 1 = 2ν, E2 = E, ν 2 = ν, G =
0.429·E. The following base characteristics are chosen: E0 =29.1 MPa, ν 0 = 0.25.
The plate flexibility index in this case is estimated by a value

3  
r= · π E0 a3 1 − ν 2 /Eh3 1 − ν02 ≈ 29.
2
First consider the case of the plate being subject to a uniformly distributed load
q = 105 N/m. The results of the calculated deflections of a plate with a non-fixed
contour on an elastic half-space and on a variable-thickness layer are shown in Fig.
4.65. Figure 4.65a shows isolines of equal deflections for an isotropic plate on an
elastic half-space, which are symmetrical with respect to the plate centre and prac-
tically do not differ from circles. With the increase of the distance from the centre,
[email protected]
in the angles, the equal deflection lines become almost straight, parallel to the diag-
onals of the square. Figure 4.65b shows the deflection isolines for the same plate,
but resting on an elastic compressible wedge (α = 30◦ , the plate centre is located
366 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

Fig. 4.65 Equal deflection W (cm) lines for a plate with free edges under a uniformly distributed
load: (a): half-space, (b), (c), (d): elastic compressible wedge, α = 30◦ (a), (b): isotropic plate, (c),
(d): orthotropic plate (Ex =2E, Ey =2E, respecively)

at a distance x = 16 m from the wedge rib). As one can see, the equal deflection
line pattern is of an asymmetrical character with the contours being shifted with
respect to the plate centre towards the increase of the non-uniformly compressed
layer thickness. At the chosen values of the calculation parameters, the deflection
values in the case of the elastic wedge appeared almost twice smaller than for the
[email protected]
half-space. This is explained by the fact that soil under the plate on the half-space
is deformed over an infinite depth. The calculation data for the deflections, obtained
for an orthotropic plate on an elastic wedge, when Ex = E1 , are shown in Fig. 4.65c,
4.7 Contact Problem for Orthotropic Foundation Plates 367

and Fig. 4.65d corresponds to a situation when the same plate is rotated with respect
to its centre by a right angle (i.e. Ey = E1 ). It is seen from the presented data that
variable thickness of the elastic layer and variation of orientation of the orthotropy
axes result in an essential asymmetry in the distribution of deflections. The maxi-
mal deflection area always shifts towards the base thickness increase and acquires
the shape of an ellipse with the higher axis along the direction with the smallest
deformation properties. As follows from the calculations, with the increase of the
orthotropy degree and the elastic compressible layer thickness interval, at cylindri-
cal bending this nonuniformity in the plate deformation character increases.
Thus, the calculations show that in order to reduce the deflection nonuniformity
and, hence, to reduce the bending moments in the calculated sections of a plate, one
should take into account the orientation of the orthotropy axes with respect to the
rib of the elastic compressible wedge. As a recommendation for design, one should
also keep in mind the requirement of the plate thickness to be increased or the load
to be reduced in the direction of the increase of the compressed base thickness.
The proposed calculation method enables one to obtain reasonable values of the
corresponding parameters of the contact interaction in the “foundation plate + base”
system.
Orthotropic plate with partly free, hinged, and pinched edges on elastic layers of
variable and constant thickness. Consider the influence of the type of fixing at the
plate contour in two cases: (i) a plate with two adjacent pinched sides (Fig. 4.66)
and (ii) a plate with one hinge-born and one pinched side (Figs. 4.67 and 4.68). In
both cases two sides are free.
Deflection isolines for a plate with the first type of the boundary conditions,
loaded by a uniformly distributed load, are shown in Fig. 4.66. The equal deflection
lines are plotted for an isotropic plate on a half-space (Fig. 4.66a) as well as on a
variable-thickness layer at the rigid underlayer base tilt angle α = 45◦ (Fig. 4.66b).
In order to take into account the orientation of the elasticity directions, similarly to
Fig. 4.65, deflection isolines for an orthotropic plate are shown (Fig. 4.66c, d). It
is seen from the calculations performed that the influence of the rigid underlayer
tilt angle on the plate bending characteristics is also quite essential, similarly to
the case of the plate with a totally free contour. This is the consequence of the non-
uniform compressibility of the base within the foundation structure under considera-
tion. Simultaneously, the comparison of the calculation data has shown that with the
increase of the angle α the difference between the solutions for the half-space and the
non-uniformly compressible base becomes smaller. The numerical results, obtained
from the calculation of the plate, interacting with a variable-thickness elastic layer
(Figs. 4.66 and 4.67), and their analysis show that substitution of pinching by hinge
on a part of the plate contour makes a slight effect on the stress-strained state of the
plate independently of the spatial nonhomogeneity of the compressed soil mass and
the elastic base depth.
Consider the calculation results for a plate with the second type of the mixed
[email protected]
boundary conditions. Deflection isolines for the plate with the account of the base
thickness variation range, loading type, as well as the change of the orthotropy axes
direction, are plotted in Figs. 4.67 and 4.68. Figure 4.67 shows the isolines for a
368 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

Fig. 4.66 Equal deflection W (cm) lines for a plate with the fixation conditions of type I under a
uniformly distributed load: (a): half-space, (b),(c),(d): elastic compressible wedge, α= 45◦ ; (a),
(b): isotropic plate, (c), (d): orthotropic plate (Ex =2E, Ey =2E, respecively)

plate under a uniformly distributed load, and Fig. 4.68 – for a plate, loaded in the
centre by a force whose value is Q = 0.256 × 104 N – the total force from the uni-
formly distributed load q. The force Q was distributed over the area 1.6 × 1.6 m2 ,
corresponding to one cell of the finite-difference grid. The calculated dependences,
[email protected]
plotted in Figs. 4.67a and 4.68a, correspond to the isotropic plate interaction with
a half-space, while Figs. 4.67b and 4.68b illustrate the effect of the non-uniform
base compressibility (elastic wedge, α = 45◦ ) on the isotropic plate deflections
4.7 Contact Problem for Orthotropic Foundation Plates 369

Fig. 4.67 Equal deflection W (cm) lines for a plate with the fixation conditions of type II under a
uniformly distributed load: (a), (b), (c), (d): same as Fig. 4.66

under a uniformly distributed and a concentrated load, respectively. The effect of

the orthotropic properties of the plate material on its bending is clearly seen from
Figs. 4.67c and 4.68c (Ex = 2E), as well as from Figs. 4.67d and 4.68d (Ey = 2E).
As one can see, the effect of the type of the plate loading (action of the concen-
[email protected]
trated and the uniformly distributed load) on its bending consists in a considerable
difference in the distribution of deflections, bending moments, and reactive pressure
intensity. As it should be expected, the stress-strained state of a plate of reinforced
370 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

Fig. 4.68 Equal deflection W (cm) lines for a centrally loaded plate with the fixation conditions of
type II. (a), (b), (c), (d): same as Fig. 4.66

concrete, loaded uniformly over its area, depends on the spatial nonhomogeneity of
the base more essentially than under a concentrated force.
The numerical results, obtained for the variable-thickness layer, were compared
with the results of calculations for a constant-thickness elastic layer. The depth Hc of
the constant-thickness layer under the plate centre was determined from the condi-
[email protected]
tion Hc = xc ·tanα. In Tables 4.13 and 4.14 two values for the maximal deflection are
presented: the upper one corresponds to the variable-thickness layer with different
angles α, and the lower one – to the constant-thickness layer. As follows from the
4.7 Contact Problem for Orthotropic Foundation Plates 371

Table 4.13 Contact interaction characteristics for an orthotropic plate (boundary conditions of the
first type)

Elastic base type Wmax, cm 4Mx /a2 q pc /q Wmax, cm 4Mx /a2 q pc /q

E1 = 2 × E, E2 = E E1 = 2 × E, E2 = E

Variable-thickness layer:
0.658 0.791 0.438 0.720 0.504 0.456
α=15◦ , Hc =0.27a
0.949 0.704 0.445 0.949 0.446 0.445
1.087 0.851 0.315 1.182 0.576 0.320
α=30◦ , Hc =0.58a
1.468 0.785 0.304 1.468 0.547 0.304
1.400 0.852 0.270 1.457 0.590 0.272
α=45◦ , Hc =a
1.716 0.807 0.252 1.716 0.572 0.252
1.587 0.844 0.248 1.627 0.593 0.249
α=60◦ , Hc =1.73a
1.853 0.813 0.231 1.853 0.581 0.231
Half-space 1.984 0.816 0.212 1.984 0.587 0.212

Table 4.14 Contact interaction characteristics for an orthotropic plate (boundary conditions of the
second type)

Elastic base type Wmax, cm 4Mx /a2 q pc /q Wmax, cm 4Mx /a2 q pc /q

E1 = 2 × E, E2 = E E1 = 2×E, E2 = E

Variable-thickness layer:
0.812 0.952 0.587 0.883 0.505 0.545
α=15◦ , Hc =0.27a
1.036 0.885 0.595 1.075 0.470 0.535
1.311 1.156 0.441 1.437 0.622 0.395
α=30◦ , Hc =0.58a
1.633 1.113 0.424 1.703 0.598 0.379
1.640 1.219 0.339 1.754 0.656 0.340
α=45◦ , Hc =a
1.990 1.197 0.375 2.022 0.642 0.320
1.866 1.234 0.366 1.953 0.667 0.316
α=60◦ , Hc =1.73a
2.200 1.232 0.350 2.190 0.659 0.296
Half-space 2.442 1.261 0.330 2.356 0.672 0.275

data quoted in Tables 4.13 and 4.14, the contact interaction characteristics (maximal
settlements Wmax , bending moments Mx and reactive pressures pc in the plate cen-
tre) essentially depend on the base model being used. The analysis of the obtained
results shows that variation of the base compressed soil mass thickness under the
foundation structure affects first of all the character of distribution and the values
of deflections. The deflection maximum Wmax always shifts towards the increase
of compressed base thickness. Variable compressibility of the base increases the
nonuniformity of the stress-strained state of the plate. This trend should be taken
into account at the design of foundation structures, since in the case of a momen-
tal stressed state an intense development of inelastic deformations in concrete is
[email protected]
possible, leading to the formation of cracks [132].
The account of orthotropic properties of the plate material at a given type of
it being fixed at its contour is also important for the estimation of the contact
372 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

interaction characteristics. In comparison with a freely resting plate, fixing of its

sides has led to a decrease of the deflections and an increase of the bending
moments. For the othotropic plate (Ex = 2E) at the boundary conditions of the first
type (Table 4.13) the bending moments Mx in the plate centre are by factor of about
1.4 higher than the corresponding values for the plate, rotated with respect to its
centre by a right angle. Note that for the boundary conditions of the second type
(Table 4.14), at the same orthotropy parameters for the plate material, the bending
moment values Mx increase almost twice. As follows from numerous calculations
of bending of plates made of reinforced concrete, their orthotropic properties can
essentially affect both the stress-strained state of the plate and the characteristics of
the base reactive pressure.
Thus, the proposed numerical approach, based on a combination of the finite-
difference and the boundary-element methods, enables effective solving of contact
problems for rectangular plates, resting on non-classical elastic bases. The finite-
difference method being used in the numerical algorithm is quite suitable to satisfy
the boundary conditions which can be found in practice, and to perform the calcula-
tions on a computer far from the high-end level. The calculation examples presented
confirm the applicability of the elaborated algorithm for the analysis of bending of
rectangular plates on elastic spatially nonhomogeneous bases and illustrate its abil-
ities regarding the design of foundation plates for complicated geotechnical condi-
tions. The developed software module ORTOPLIT can be without any difficulties
applied for the studies of complicated cases of loading and different combinations
of boundary conditions at the plate contour. Besides, the application of a contact
model of the elastic base is performed as a separate module and does not require the
whole algorithm resetting.

References
1. Abramov V M (1937) Problems of contact of an elastic half-plane with an absolutely rigid
foundation with the account of friction forces. Dokl AN SSSR 17 (issue 4):173–178 (in
Russian)
2. Abramov V M (1939) Study of a case of asymmetric pressure of a punch of round cross-
section on an elastic half-space. Dokl AN SSSR 23 (issue 8):759–763 (in Russian)
3. Abramowitz M, Stegun I A (1972) Handbook of mathematical functions with formulas,
graphs, and mathematical tables, 10th edn. Dover, New York
4. Aizikovich S M, Trubchik I S, Shklyarova Ye V (1992) Calculation of a circular plate resting
on a vertically inhomogeneous half-space. Mech Solids 27 (issue 4):163–171
5. Aleksandrov A V, Lashchenikov B Ya, Shaposhnikov N N et al. (1976) Computer-aided
methods of calculation of rod systems, plates and shells. Stroyizdat, Moscow (in Russian)
6. Aleksandrov A Ya (1955) Some solutions of axially symmetrical contact problems of the-
ory of elasticity. In: NIIZHT Works. Transzheldorizdat, Moscow, issue 11, pp. 29–61 (in
Russian)
7. Aleksandrov A Ya (1955) Stresses and displacements in an elastic space and half-space
[email protected]
under a load, uniformly distributed over a ring. In: NIIZHT Works. Transzheldorizdat,
Moscow, issue 11, pp. 62–88 (in Russian)
8. Aleksandrov A Ya, Solovyov Yu I (1978) Spatial problems of theory of elasticity (application
of methods of theory of complex-variable functions). Nauka, Moscow (in Russian)
References 373

9. Aleksandrov V M (1967) Axially symmetrical contact problem of the action of a ring-shaped

punch on an elastic half-space. Inzh Zh Mekh Tverd Tela (issue 4):108–116 (in Russian)
10. Aleksandrov V M, Babeshko V A (1972) On the pressure on an elastic half-space by a
wedge-shaped stamp. J Appl Math Mech 36:78–83
11. Aleksandrova G P (1973) Contact problems of bending of plates resting on an elastic base.
Izv AN SSSR Mekh Tverd Tela (issue 1):97–106 (in Russian)
12. Aleksandrovich V F, Fedorovskii V G (1979) Round punch on an elastoplastic strengthening
soil base. In: Experimental and theoretical studies of nonlinear problems in the field of bases
and foundations. NPI, Novocherkassk, pp. 35–43 (in Russian)
13. Aleynikov S M (1991) Spatial contact problems for polygonal punches on elastic tapered
basement. VISI, Voronezh, VINITI No. 1393-B91Dep (in Russian)
14. Aleynikov S M (1998) Method for the determination of modulus of deformation. Patent
2145655, MPK6 E02D 1/00; G01N 3/42. No. 98117270/28 (Russia)
15. Aleynikov S M, Ikonin S V (1990) Three-dimensional deformation of the surface of a
tapered elastic layer. Soil Mech Found Eng 27:218–222
16. Aleynikov S M, Ikonin S V (1993) Numerical calculation of contact stresses in the base of a
rectangular punch at off-centre loading. In: Issues of strength and plasticity. Dnipropetrovsk
State University, Dnipropetrovsk, pp. 95–101 (in Russian)
17. Aleynikov S M, Kozlovtsev A M (1992) Numerical solution of spatial contact problems
for rectangular punches on a variable-thickness elastic layer with the account of nuilateral
constraints. Stroit Mekh Raschet Sooruzh (issue 3):18–23 (in Russian)
18. Aleynikov S M, Neiburg E V (1995) Contact pressure distribution in the area of strip foun-
dation width variation. Izv Vuzov Stroit (issue 4):19–23 (in Russian)
19. Aleynikov S M, Neiburg E V (1997) Foundation. Patent 2080030 RU, MKI6 E 02 D 27/00.
No. 94019239/33 (Russia)
20. Aleynikov S M, Nikolayev S N (1999) Spatial stress-strained state of the base of a variable-
width rigid strip foundation. Izv Vuzov Stroit (issue 4):15–21 (in Russian)
21. Aleynikov S M, Sedaev A A (1995) Algorithm of grid generation in the boundary-element
method for flat domains. Mat Modelirovanie 7 (issue 7):81–93 (in Russian)
22. Altenbach J, Sacharov A S (eds.) (1982) Die Methode der Finiten Elemente in der Festkör-
permechanik. Fachbuchverlag, Leipzig
23. Ambartsumyan S A (1967) Theory of anisotropic plates. Nauka, Moscow (in Russian)
24. Amusin B Z, Fadeyev A B (1975) Finite-element method at the solution of problems of
mining geomechanics. Nauka, Moscow (in Russian)
25. Antipov Yu A (1987) Exact solution of a problem of impression of a ring-shaped punch into
a half-space. Doklady AN UkrSSR Ser A (issue 7):29–33 (in Russian)
26. Atopov V I, Serdobintsev Yu P, Slavin O K (1988) Contact stress modelling. Mashinostroye-
nie, Moscow (in Russian)
27. Babeshko V A, Glushkov Ye I, Glushkova N V (1981) On the singularities in the angular
points of spatial punches in contact problems. Dokl AN SSSR 257 (issue 2): 289–294 (in
Russian)
28. Babloyan A A, Tonoyan V S (1967) On the impression of a ring-shaped punch into an elastic
half-space. Izv AN ArmSSR Mekh (issue 1):3–16 (in Russian)
29. Banerjee P K, Butterfield R (1981) Boundary element methods in engineering science.
McGraw-Hill, New York
30. Barkan D D (1948) Dynamics of bases and foundations. Stroyvoyenmorizdat, Moscow (in
Russian)
31. Basant Z P (1974) Three-dimensional harmonic functions near termination or intersection
[email protected]
of gradient singularity lines: a general numerical method. Intern J Eng Sci 12:221–243.
32. Berezantsev V G (1970) Calculation of bases of buildings.Stroyizdat, Leningrad (in Russian)
33. Birman S Ye (1953) On the settlement of a rigid punch on an elastic layer resting on an
incompressible base. Dokl AN SSSR 93 (issue 5):791–794 (in Russian)
374 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

34. Bokiy I B, Petrishin V I (1986) Algorithm of approximate solution of the problem of the
pressure of a smooth square punch with a flat bottom on an elastic half-space. In: Stability
and strength of structural elements. Dnipropetrovsk State University, Dnipropetrovsk, pp.
38–41 (in Russian)
35. Borodachev A N (1984) Pressure of an elliptical punch on a nonhomogeneous elastic half-
space. Dokl AN UkrSSR Ser A (issue 7):30–33 (in Russian)
36. Borodachev A N (1985) Interaction of a rigid foundation with a nonhomogeneous elastic
base. Izv Vuzov Stroit Arhit (issue 10):38–41 (in Russian)
37. Borodachev A N (1990) The influence tensor for an elastic medium with Poisson’s ratio
varying in one direction. J Appl Math Mech 54:242–249
38. Borodachev A N, Dudinskii V I (1986) Bending of a round plate on a nonhomogeneous
elastic base. Izv Vuzov Stroit Arhit (issue 11):26–30 (in Russian)
39. Borodachev A N, Dudinskii V I (1986) Contact problem for an elastic half-space with a
variable Poisson ratio. Mech Solids 21 (issue 1):86–91
40. Borodachev N M (1970) Impression of a punch with a footing in the shape of a narrow
rectangle into an elastic half-space. Izv AN SSSR Mekh Tverd Tela (issue 4):135–141 (in
Russian)
41. Borodachev N M (1976) Contact problem for a stamp with a rectangular base. J Appl Math
Mech 40:505–512
42. Borodachev N M (1976) On the nature of the contact stress singularities under an annular
stamp. J Appl Math Mech 40:347–352
43. Borodachev N M, Borodacheva F N (1966) Impression of a ring-shaped punch into an elastic
half-space. Inzh Zh Mekh Tverd Tela (issue 4):158–161 (in Russian)
44. Borodachev N M, Galin L A (1974) Contact problem for a stamp with narrow rectangular
base. J Appl Math Mech 38:108–113
45. Borodachev N M, Tarikov G P (1972) Determination of the reaction pressure under a foun-
dation by the electrical modeling method. Soil Mech Found Eng 9:428–429
46. Borodachev N M, Tarikov G P (1974) Solution of three-dimensional contact problems in
elasticity theory by the method of electrical simulation. Mech Solids 9 (issue 3):84–87
47. Borodacheva F N (1968) Action of a vertical eccentric force on an annular foundation situ-
ated on a compressible base. Soil Mech Found Eng 5:34–38
48. Borodacheva F N (1969) On indentation of a ring-shaped punch into an elastic half-space
under an off-centre force. Izv Vuzov Stroit Arkhit (issue 8):15–20 (in Russian)
49. Borodacheva F N (1971) Determination of displacements of the boundary of an elastic half-
space due to the action of a ring-shaped foundation. Izv Vuzov Stroit Arkhit (issue 5):80–86
(in Russian)
50. Borodacheva F N (1972) Displacements and stresses in the base of a rigid, symmetrically
loaded ring foundation. Soil Mech Found Eng 9:217–221
51. Borodacheva F N (1973) Stresses in internal points of a half-space due to the action of a
ring-shaped punch. Izv Vuzov Stroit i Arkhit (issue 3):58–65 (in Russian)
52. Borodacheva F N (1975) On the determination of main stresses in an elastic half-space under
a ring-shaped punch. Izv Vuzov Stroit i Arkhit (issue 10):45–48 (in Russian)
53. Borowicka H (1943) Über ausmittig belaste starre Platten auf elastisch-isotropen Under-
grund. Ing Arch 1:1–8
54. Botkin A I (1940) On the strength of friable and fragile materials. Izv VNIIG Vedeneeva
26:205–236 (in Russian)
55. Boussinesque J (1885) Applications des potintiels: l’etude de l’equilibre et du mouvement
des solides e’lastiques. Gauthier-Villars, Paris
56. Brebbia C A, Walker S (1980) Boundary element techniques in engineering. Newnes-
[email protected]
Butterworths, London
57. Brothers P W, Sinclair G B, Segedin C M (1977) Uniform indentation of the elastic
half-space by a rigid rectangular punch. Intern J Solids Struct 13:1059–1072.
58. Bufler H, Lieb H, Meier G (1982) Frictionless contact between an elastic stamp and elastic
foundation. Ing Arch 52:63–76.
References 375

59. Bugrov A K, Golubev A I (1993) Anisotropic soils and bases of buildings. Nedra, St. Peters-
burg (in Russian)
60. Bundy B D (1984) Basic optimization methods. Edward Arnold, Baltimore
61. Burmister D M (1956) Stress and displacement characteristics of a two-layer rigid
base soil system: influence diagrams and practical applications. Proc Highw Res 35:
773–81
62. Burmistrov A N (1987) Contact problem of theory of elasticity for punches of oblong shape.
In: Study of non-stationary motion of a solid medium. MFTI, Moscow, pp. 23–30 (in Rus-
sian)
63. Bycroft G N (1956) Forced vibrations of a rigid circular plate on a semi-infinite elastic space
and on elastic stratum. Phil Trans Roy Soc London Ser A 248:327–368.
64. Chernousko F L, Banichuk N V (1973) Variational problems of mechanics and control.
Numerical methods. Nauka, Moscow (in Russian)
65. Cheung Y K, Zienkiewicz O C (1965) Plates and tanks on elastic foundation: an application
of finite element method. Intern J Solids Struct 1:451–461
66. Darevskii V M (1977) Bending of a rectangular plate with free edges. Izv AN SSSR Mekh
Tverd Tela (issue 1):79–90 (in Russian)
67. Decisive laws of soil mechanics (1975) Mir, Moscow (in Russian)
68. Dempsey J P, Li H (1989) A rigid rectangular footing on an elastic layer. Geotechnique
9:147–152
69. Derkach V F (1971) Approximate calculation of thin constructive orthotropic ribbed plates
on an elastic half-space. Izv Vuzov Stroit Arkhit (issue 4):50–54 (in Russian)
70. Desai C S, Christian J T (eds.) (1977) Numerical methods in geotechnical engineering.
McGraw-Hill, New York
71. Dinnik A N (1952) Selected works. Vol 1. Impact and compression of elastic bodies. Ukr
Acad Sci Publ, Kyiv (in Russian)
72. Dovnarovich S V, Pol’shin D E, Baranov D S, Sidorchuk V F (1981) Dependence of the
stress state of the base on the form of the foundation in plan. Soil Mech Found Eng 150:
340–344
73. Durayev A Ye (1991) Calculation of structures on a soil base with a deformation modulus
increasing with depth. Mordovian University, Saransk (in Russian)
74. Egorov K E (1948) Deformation of the base of a round rigid foundation under an eccen-
tric load. In: Bases and foundations. Issues of soil mechanics. Tr NIIOSP 11:119–138 (in
Russian)
75. Egorov K E (1958) On the deformation of a finite-thickness base. In: Bases and foundations.
Soil mechanics. Tr NIIOSP 34:5–33 (in Russian)
76. Egorov K E (1963) Indentation of a punch with a flat ring-shaped bottom into a half-space.
Izv AN SSSR Mekh i Mashinostr (issue 5):187–196 (in Russian)
77. Egorov K E (1965) Calculation of bed for foundation with ring footing. In: Proc 6th Intern
Conf Soil Mech Found Eng, Vol. 2, pp. 41–45
78. Egorov K E, Barvashov V A, Fedorovskii V G (1973) On the application of theory of elas-
ticity to the calculation of bases under structures. In: Proc 8th Intern Congr Soil Mech Found
Eng. Stroyizdat, Moscow, pp. 72–83 (in Russian)
79. Egorov K E, Finaeva T I (1984) Initial critical load on the ground in the case of a circular
foundation. Soil Mech Found Eng 21:269–272
80. Egorov K E, Kitaykina O V, Dokhnyanskii M P (1986) Calculation of bases for rong-shaped
foundations. In: Efficient foundation design and methods of base preparation. Tr NIIOSP
85:104–116 (in Russian)
81. Egorov K E, Shelest L A (1985) Distribution of stresses and displacements in the base of a
round foundation of tower structures. In: Numerical methods of solving the problems of soil
[email protected]
mechanics and calculation of foundations at complex geotechnical conditions. Tr NIIOSP
84:46–64 (in Russian)
82. Egorov K E, Shelest L A, Sokolov N S (1988) Determination of the horizontal displacement
of the base of a rigid round foundation on a finite-thickness layer. In: Theoretical and exper-
376 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

imental substantiation of new design and technology in foundation engineering. Tr NIIOSP

89:3–13 (in Russian)
83. Ermashov V P (1985) Effect of foundation shape on normal contact pressure distribution.
Soil Mech Found Eng 22:61–64
84. Fabricant V I (1986) Inclined flat punch of arbitrary shape on an elastic half-space. Trans
ASME J Appl Mech 53:798–806
85. Fadeev A B (1987) Finite-element method in geomechanics. Nedra, Moscow (in Russian)
86. Fedorov V I (1993) Design and construction of foundations for structures on a complex
relief. Dalpress, Vladivostok (in Russian)
87. Fedorovskii V G (1985) Modern methods of description of mechanical properties of soils.
VNIIS, Moscow (in Russian)
88. Fedorovskii V G, Dokhnyansky M P (1988) Settlements of round and ring-shaped founda-
tions: prediction and comparison with field observation data. Baltic Conf on Soil Mechanics.
Tallinn, Vol 2, pp. 99–106 (in Russian)
89. Fedorovskii V G, Onopa I A (1985) Surface settlements of a wedge-shape bed under a
concentrated load. Soil Mech Found Eng 22:79–84
90. Filatov A V, Adigamov R Sh (1990) Emergencies and deformations of industrial buildings
and structures. Soil Mech Found Eng 27:43–47
91. Florin V A (1936) On the calculation of structures on weak soils. In: Gidrostroyproekt
Works: Gidrotekhnika, issue 1, pp. 21–34 (in Russian)
92. Florin V A (1959) Fundamentals of soil mechanics. Stroyizdat, Moscow/Leningrad (in Rus-
sian)
93. Forsythe G E, Malcolm M A, Moler C B (1977) Computer methods for mathematical com-
putations. Prentice-Hall, Englewood Cliffs
94. Fourie A B, Beer G (1989) An illustration of the importance of soil non-linearity in soil-
structure interaction problems. Comput Geotechnics 8:219–241
95. Fraser R A, Wardle L J (1976) Numerical analysis of rectangular rafts on layered founda-
tions. Geotechnique 26:613–630
96. Galashev Yu V, Dyba V P, Murzenko A Yu (1979) Experimental studies of the depth od
a compressible base loaded by a round punch. In: Experimental and theoretical studies of
nonlinear problems in the field of bases and foundations. NPI, Novocherkassk, pp. 128–134
(in Russian)
97. Galin L A (1961) Contact problems in the theory of elasticity. North Carolina State College,
Raleigh
98. Galin L A (1980) Contact problems of theory of elasticity and viscoelasticity. Nauka,
Moscow (in Russian)
99. Galin L A (ed.) (1976) Development of theory of contact problems in the USSR, Nauka,
Moscow (in Russian)
100. Gilman Ya D, Gilman Ye D (1989) Reinforcement and reconstruction of buildings on loess
soils. Stroyizdat, Moscow (in Russian)
101. Glushkov V Ye (1990) Stress-strained state of elastoplastic bases of ring-shaped and cir-
cular foundations. In: Nonlinear methods of calculation of bases and foundations. Mariy
Polytechnic Institute, Yoshkar-Ola, p 44–46 (in Russian)
102. Godes Yu Ya (1987) Functions of compliance of a multilayer base with elastic constraints
between the layers. In: Nonlinear problems of hydroaeromechanics and theory of elasticity.
Dnipropetrovsk State University, Dnipropetrovsk, pp. 92–97 (in Russian)
103. Godes Yu Ya, Privarnikov A K (1983) Quadrature formulae for Hankel inte-
grals. Dnipropetrovsk State University, Dnipropetrovsk, VINITI No. 327-84Dep (in
Russian)
[email protected]
104. Goldshtein M N, Kushner S G, Shevchenko M I (1977) Calculation of settlements and
strength of bases of buildings and structures. Budivelnyk, Kyiv (in Russian)
105. Gorbunov-Posadov M I, Malikova T A, Solomin V I (1984) Calculation of structures on an
elastic base. Stroyizdat, Moscow (in Russian)
References 377

106. Gorlov A M, Serebryaniy R V (1968) Automated calculation of rectangular plates on an

elastic base. Stroyizdat, Moscow (in Russian)
107. Gorodetskiy A S, Zavoritskiy V I, Lantukh-Lyashchenko A I, Rasskazov A O (1989)
Automation of calculations of transport buildings. Transport, Moscow (in Russian)
108. Goryacheva I G, Dobychin M N (1988) Contact problems in tribology. Mashinostroyeniye,
Mosow (in Russian)
109. GOST 13580-85. Ferroconcrete plates of strip foundations. (1986) Standard Publishers,
Moscow (in Russian)
110. GOST 20276-85. (1985) Soils. Methods of field determination of deformability characteris-
tics Standard Publishers, Moscow (in Russian)
111. Gubenko V S (1960) Pressure of an axially symmetrical ring-shaped punch on an elastic
layer and an elastic half-space. Izv AN SSSR Mekh Tverd Tela (issue 3):60–64 (in Russian)
112. Gubenko V S, Mossakovskii V I (1960) Pressure of an axially symmetric circular die on an
elastic half-space. J Appl Math Mech 24:477–486
113. Gubenko V S, Nakashidze G M, Pyatovolenko V G (1986) Exact solution of the problem of
a ring-shaped punch. Dokl AN UkrSSR Ser A (issue 3):40–44 (in Russian)
114. Gudehus G (ed.) (1977) Finite elements in geomechanics. Wiley, New York
115. Hamming R W (1962) Numerical methods for scientists and engineers. McGraw-Hill, New
York
116. Hara T, Shibuya T, Koizumi T, Nakahara I (1978) An asymmetric mixed boundary value
problem of the elastic half-space subjected to moment by an annular ring punch. Bull JSME
21:566–571.
117. Harr M E (1966) Foundations of theoretical soil mechanics. McGraw-Hill, New York
118. Huber M T (1914) Die Grundlagen einer rationellen Berechnung der Eisenbetonplatten. Z
Österr Ing und Architektur Vereins:557
119. Il’inkova N I (1992) Bending of a plate of linearly variable thickness on an elastic half-space
with modulus of elasticity varying along the depth. Intern Appl Mech 28:484–489
120. Instruction on design of blast-furnace foundations (1972) VSN 001-71, USSR Ministry of
Ferrous Metallurgy, Stroyizdat, Moscow (in Russian)
121. Instruction on design of road coatings of non-rigid type (1985) VSN 46-83, USSR Ministry
of Transport Engineering, Transport, Moscow (in Russian)
122. Instructions on design and erection of foundations for buildings and structures at hill slopes
using the base settlement alignment method (1970) DVPromstroyniiproekt, Vladivostok (in
Russian)
123. Instructions on design of foundations for machines subject to dynamic loads (1982) Gerse-
vanov NIIOSP, Stroyizdat, Moscow (in Russian)
124. Instructions on design of foundations on a natural base of subcolumn for buildings and indus-
trial structures (1978) Stroyizdat, Moscow (in Russian)
125. Instructions on design of plate foundations for tower-type skeleton buildings and structures
(1984) Gersevanov NIIOSP, Stroyizdat, Moscow (in Russian)
126. Ioannides A M (1988) Finite difference solution for plate on elastic solid. J Transp Eng
114:57–75
127. Ivanov P L (1991) Soils and bases of hydrotechnical structures. Vysshaya shkola, Moscow
(in Russian)
128. Johnson K L (1985) Contact mechanics. Cambridge University Press, Cambridge
129. Kachamkin A A, Tsukrov I I, Shvayko N Yu (1987) Application of the finite-element
[email protected]
method to the calculation of variable-rigidity anisotropic plates on an elastic base. In: Issues
of strength and plasticity. Dnipropetrovsk State University, Dnipropetrovsk, pp. 39–40 (in
Russian)
130. Kaczkowski
˛ Z (1968) Plyty – obliczenia statyczne. Arkady, Warszawa
378 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

131. Kalinnikov A Ye, Yefremov S M, Vakhrushev F V (1985) Algorithm for automatic mesh of a
two-dimensional domain for solving contact problems by boundary-element method. Probl
Prochnosti (issue 2):106–108 (in Russian)
132. Karpenko N I (1976) Theory of deformation of ferroconcrete with cracks. Stroyizdat,
Moscow (in Russian)
133. Kim Yu A, Shebchuk L I (1978) Deformation of soil under a rigid punch. In: Theoretical
and applied mechanics. Vysheyshaya Shkola, Minsk 5:109–112 (in Russian)
134. Kiselev V A (1973) Calculation of plates. Stroyizdat, Moscow (in Russian)
135. Kizyma Ya M (1973) Pressure of a circular punch on an elastic layer in the presences of
tangential forces in the contact zone. Intern Appl Mech 9:901–904
136. Klepikov S N (1967) Calculation of structures on an elastic base. Budivelnyk, Kyiv (in
Russian)
137. Klepikov S N (1972) Calculation of beams on a nonlinearly deformed Winkler base. Osno-
vaniya Fundamenty Mekh Gruntov (issue 1):8–10 (in Russian)
138. Klepikov S N, Malikova T A (1982) Bending of a variable-rigidity orthotropic plate on a
nonlinearly deformable displaced base. In: Bases, foundations, and underground structures.
Tr NIIOSP. 73:15–19 (in Russian)
139. Klepikov S N, Tregub A S, Matveyev I V (1987) Calculation of buildings and structures on
subsiding soils. Budivelnyk, Kyiv (in Russian)
140. Knyazeva L P (1984) Solving mixed spatial problems of theory of elasticity of anisotropic
media using the boundary integral equation method. Ph D thesis, Minsk (in Russian)
141. Kocatürk T (1995) Rectangular anisotropic (orthotropic) plates on a tensionless elastic foun-
dation 31:378–386
142. Konik A Ya, Demin I I (1982) Calculation of temperature fields and stress-strained state of
soil bases using the finite-element method. MISI, Moscow (in Russian)
143. Korenev B G, Chernigovskaya Ye I (1962) Calculation of plates on an elastic base. Gosstroy-
izdat, Moscow (in Russian)
144. Krivorotov A P (1963) Experimental studies of distribution of normal stress over the contact
area for a punch with a sand base. Osnovaniya Fundamenty Mekh Gruntov (issue 2):8–12
(in Russian)
145. Krivorotov A P (1965) On the distribution of normal stress over the bottom of a rigid punch
at the variation of the vertical load eccentricity. Izv Vuzov Stroit (issue 7):33–38 (in Russian)
146. Krivorotov A P (1969) State of stress of a sand mass under a plate lying at a shallow depth.
Soil Mech Found Eng 6:22–24
147. Krivorotov A P, Babello V A (1981) Results of investigation of the stress state of a sandy
bed for rigid plates. Soil Mech Found Eng 18:117–122
148. Kuznetsov V I (1952) Elastic base. Gosstroyizdat, Moscow (in Russian)
149. Lamzyuk V D, Privarnikov A K (1977) Mixed problems of theory of elasticity for multilayer
bases with circular boundary condition separation lines. In: Mixed problems of mechanics
of deformable bodies. Abstr 1st Sov Conf, Rostov-na-Donu, Pt 1, pp. 37–38 (in Russian)
150. Lebedev N N, Ufliand Ia S (1958) Axisymmetric contact problem for an elastic layer. J Appl
Math Mech 22:442–450
151. Lee I K (1962) Bearing capacity of foundations with particular reference to the Melbourn
area. J Instn Engrs Austr 34:283
152. Lekhnitskii S G (1947) Anisotropic plates. Gostekhizdat, Moscow/Leningrad (in Russian)
153. Leonov M Ya (1953) General problem of a pressure of a circular punch on an elastic half-
space. Prikl Mat Mekh 17:87–98 (in Russian)
154. Leonov M Ya, Posatskii S L, Ivashchenko A N (1956) Calculation of a square-shaped foun-
dation. Sci. Rep. Inst. Machinery and Automation, Ukr.Acad, Sci. 5 (issue 4):141–151 (in
Russian)
[email protected]
155. Likhovtsev V M, Estrin I Yu (1985) Some aspects of numerical implementation of solving
problems of interaction of a punch and a base by the finite-element method. In: Numerical
methods of solving problems of soil mechanics and calculation of foundations in compli-
cated geotechnical conditions. Tr NIIOSP 84:174–181 (in Russian)
References 379

156. Love A E H (1929) Stress produced in a semi-infinite solid by pressure on part of the bound-
ary. Philos Trans Roy Soc London A 228:377–420.
157. Love A E H (1959) A treatise on the mathematical theory of elasticity, 4th edn. Cambridge
University Press, Cambridge
158. Lubyagin I A, Pozharskii D A, Chebakov M I (1991) Generalization of Boussinesq and
Cerruti problems for the case of an elastic spatial wedge. Dokl AN SSSR 321 (issue 1):
58–62 (in Russian)
159. Lur’e A I (1964) Three-dimensional problems in the theory of elasticity. Interscience,
New York
160. Lur’e A I (1970) Theory of elasticity. Nauka, Moscow (in Russian)
161. Malyshev M V, Zaretskii Yu K, Shirokov V N, Cheremnykh V A (1973) On the combined
functioning of rigid foundations and a nonlinearly deformable base. In: Proc 8th Intern
Congr Soil Mech Found Eng. Stroyizdat, Moscow, pp. 97–104 (in Russian)
162. Manual for the design of bases for buildings and structures (for the Engineering Rules and
Regulations 2.02.01.-83) (1986) Gersevanov NIIOSP, Stroyizdat, Moscow (in Russian)
163. Marcus H (1932) Die Theorie elastischer Gewebe und ihre Anwendungen auf die Berech-
nung biegsamer Platten, 2. Auflage, Springer-Verlag, Berlin
164. Martynenko M D, Dashkevich A A (1987) Bending of round plates on a nonhomogeneous
base with a variable in depth Poisson ratio with the account of their thickness deformability.
In: Non-axisymmetrical problems of hydromechanics and theory of elasticity. Dnipropetro-
vsk State University, Dnipropetrovsk, pp. 151–153 (in Russian)
165. Martynenko M D, Knyazeva L P, Romanchik V S (1979) Pressure of an absolutely rigid
punch on an orthotropic half-space under cohesion. In: Current problems of mechanics
of deformable media. Dnipropetrovsk State University, Dnipropetrovsk, pp. 156–161 (in
Russian)
166. Martynenko M D, Knyazeva L P, Romanchik V S (1979) Pressure of an absolutely rigid
punch on an orthotropic half-space in the absence of friction. Izv AN BSSR Ser Fiz Mat
Nauk (issue 6):40–44 (in Russian)
167. Martynenko M D, Romanchik V S (1977) On one method of solving the main integral equa-
tion of the contact problem of theory of elasticity. Izv AN BSSR Ser Fiz Mat Nauk (issue
3):42–47 (in Russian)
168. Meleshenkov Ye I, Ozherelyev V A (1986) Some results of comparison of finite-element and
finite-difference methods for the example of the problem of bending of a plate. In: Numerical
methods in the studies of engineering structures. Kucherenko CNIISK Works, pp. 70–76 (in
Russian)
169. Mindlin R D, Cheng D H (1950) Nuclei of strain in the semi-infinite solid. J Appl Phys
21:926–931
170. Mossakovskii V I (1953) General solution of a problem of determination of pressure under
the bottom of a round punch without the account of friction. Scientific Records Inst Machin-
ery and Automation Acad Sci UkrSSR 2 (issue 1):41–53 (in Russian)
171. Mossakovskii V I (1953) Pressure of a round punch on an elastic half-space. Scientific
Records Inst Machinery and Automation Acad Sci UkrSSR 2 (issue 1):9–40 (in Russian)
172. Mossakovskii V I, Kachalovskaya N Ye, Golikova S S (1985) Contact problems of mathe-
matical theory of elasticity. Naukova dumka, Kyiv (in Russian)
173. Mossakovskii V I, Kovura A B (1976) On one method of solving the problems of theory of
potential and its application in theory of elasticity. Dokl AN UkrSSR Ser A (issue 1):44–47
(in Russian)
174. Mossakovskii V I, Kovura A B (1980) Contact problems for an elastic half-space with round
and near-round boundary condition separation lines. In: Dnipropetrovsk State University,
[email protected]
Dnipropetrovsk (issue 5):74–89 (in Russian)
175. Muki R (1960) Asymmetric problems of the theory of elasticity for a semi-infinite solid and
a thick plate. In: Sneddon I N and Hill R (eds.) Progress in solid mechanics. North-Holland,
Amsterdam, Vol. 1, pp. 399–439
380 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

176. Muravskii G B (1967) On the application of the Betti theorem for solving some problems
for an elastic base. In: Works Mosc. Inst. Rail Transport Eng. 236:51–57 (in Russian)
177. Murzenko Yu N, Revenko V V (1976) Experimental studies of distribution of normal and
tangential stresses in the base of a round punch using load-cell stress gauges. In: Bases and
foundations. NPI, Novocherkassk, pp. 3–12 (in Russian)
178. Murzenko Yu N, Tarikuliyev Z A, Arinina E V, Revenko V V (1979) Results and prospects
of experimental studies of the stress-strained state of bases of buildings. In: Experimental
and theoretical studies of nonlinear problems in the field of bases and foundations. NPI,
Novocherkassk, pp. 121–128 (in Russian)
179. Nikishin V S, Shapiro G S (1973) Problems of theory of elasticity for multilayer media.
Naula, Moscow (in Russian)
180. Nikishin V S, Shapiro G S (1976) Problem of incomplete contact of annular or circular
stamp with elastic laminated medium. Mech Solids 11 (issue 5):27–38
181. Nikolaevsky V N (1972) Mechanical properties of soil and theory of plasticity. VINITI AN
SSSR, Moscow (in Russian)
182. Novotný B, Hanuška A (1985) Tuhy rasník obdĺžnikoveho podorysu na pruznom podlozí.
Staveb. Cas. 33:521–540
183. Novotný B, Hanuška A (1987) Rectangular plate on an elastic half-space. Staveb Cas
35:359–376
184. Oden J T (1972) Finite elements of nonlinear continua. McGraw-Hill, New York
185. Olesiak Z (1965) Annular punch on elastic semi-space. Arch Mech Stosow 17:633–648
186. Onopa I A, Fedorovskii V G (1984) Calculation of contact stress under a rigid rectangular
punch on a wedge-shaped base. In: Mechanics of soils and calculation of bases and founda-
tions according to the limiting states. Tr NIIOSP 82:45–50 (in Russian)
187. Onopa I A, Fedorovskii V G (1984) Settlements and slopes of a rectangular punch on a
wedge-shaped base. Izv Vuzov Stroit Arkhit (issue 5):47–50 (in Russian)
188. Ortega J M, Rheinboldt W C (1970) Iterative solution of nonlinear equations in several
variables, Academic Press, New York
189. Palatnikov Ye A (1964) Rectangular plate on an elastic base. Stroyizdat, Moscow (in
Russian)
190. Pasternak P L (1954) Fundamentals of a new method of calculation of foundations
on an elastic base using two coefficients of subgrade reaction. Stroyizdat, Moscow (in
Russian)
191. Perich A I (1994) Effective strip foundations. Zhilish Stroit (issue 4):11–13 (in Russian)
192. Petrukhin V P (1995) Calculation of suffosion deformations in saline soils. Soil Mech Found
Eng 32:159–162
193. Pilyagin A V, Kazantsev S V (1986) Mixed elastoplastic problem of calculation of a soil base
in a spatial formulation. In: Geotechnics of the Volga area, Kazan, pp. 44–47 (in Russian)
194. Pilyagin A V, Kazantsev S V (1990) Design of bases and foundations with the account of
elastoplastic properties. Krasnoyarsk State University, Krasnoyarsk (in Russian)
195. Pilyagin A V, Kazantsev S V, Kazantseva N N (1985) Elastoplastic calculation of a soil base
at the conditions of a spatial stress-strained state. Mariy Polytechnical Institute, Yoshkar-Ola,
VNIIIS No. 5713 (in Russian)
196. PLAXIS – Finite Element Code for Soil and Rock Analyses. Ver. 7. General Information
and Tutorial Manual (1998) Balkema, Rotterdam
197. Plevkov V S, Polishchuk A I (1990) Size determination of off-centre loaded foundations of
various geometrical shape. TPI, Tomsk (in Russian)
198. Podgorniy A N, Gontarovskii P P, Kirkach B N et al. (1989) Problems of contact interaction
of structural elements. Naukova Dumka, Kyiv (in Russian)
[email protected]
199. Pollard J H (1979) A handbook of numerical and statistical techniques. Cambridge Univer-
sity Press, Cambridge
200. Popov G Ya (1967) On an approximate method of solving a contact problem of a ring-shaped
punch. Izv AN ArmSSR Mekh 20 (issue 2): 19–36 (in Russian)
References 381

201. Popov G Ya (1976) Mathematical problems of contact problems. Odesa State University,
Odesa (in Russian)
202. Popov G Ya (1982) Concentration of elastic stress near punches, cuts, fine inclusions and
supports. Nauka, Moscow (in Russian)
203. Popov G Ya (1982) Contact problems for a linearly deformable base. Vyshcha Shkola,
Kyiv/Odesa (in Russian)
204. Poulos H G (1968) The behaviour of a rigid circular plate resting on a finite elastic layer.
Civil Eng Inst Engrs Aust CE10:213–219
205. Poulos H G, Davis E H (1974) Elastic solutions for soil and rock mechanics. Wiley,
New York
206. Prisyazhnyuk V K, Marchuk A V (1987) Boundary-element method in the problems of
contact interaction of multilayer rectangular plates with an elastic half-space. Stroit Mekh
Raschet Sooruzh (issue 4):9–12 (in Russian)
207. Proskuryakov S M, Malyshev M V (1979) Numerical solution of a spatial problem of a
flexible load on a half-space. In: Experimental and theoretical studies of nonlinear problems
in the field of bases and foundations. NPI, Novocherkassk, pp. 35–43 (in Russian)
208. Protsenko V S (1968) On the pressure under an annular punch. Intern Appl Mech 4 (issue
9):S51–53
209. Rabotnikov A I, Kovanev B M (1970) Experimental studies of deformations and stresses
along the base depth under a rigid punch. In: Bases and foundations. Budivelnyk, Kyiv 3:59–
64 (in Russian)
210. Rappoport L M (1948) Boussinesq problem for a layered elastic half-space. Works of
Leningrad Polytechnical Institute (issue 5):3–18 (in Russian)
211. Recommendations on the calculation of settlements and slopes for rectangular foundations
on a wedge-shaped base (1985) Gersevanov NIIOSP, Moscow (in Russian)
212. Reisner E, Sagoci H (1944) Forced torsional oscillation of an elastic half-space. J Appl Phys
15:652–662
213. Rekach V G (1984) Instructions for solving problems of applied theory of elasticity.
Vysshaya Shkola, Moscow (in Russian)
214. Roitman A B, Shishkanova S F (1973) The solution of the annular punch problem with the
aid of recursion relations. Intern Appl Mech 9:725–729
215. Roitman A G, Smolenskaya N G (1978) Renovation and reconstruction of residential and
public buildings. Stroyizdat, Moscow (in Russian)
216. Rozin L A (1971) Computer-aided calculation of hydrotechnical structures. Finite-element
method. Energia, Leningrad (in Russian)
217. Rozin L A (1990) Combination of the finite-element method and the Green-Somigliana for-
mula. In: LPI Works 434:3–10 (in Russian)
218. Rozin L A (1998) Problems of theory of elasticity and numerical methods of their solution.
SPBGTU, St. Petersburg (in Russian)
219. Rozin L A, Rukavishnikov V A (1995) Development of a combined method of calculation
of structures and their bases of infinite length. Izv Vuzov Stroit Arkhit (issue 11):37–42 (in
Russian)
220. Rozin L A, Rukavishnikov V A (1997) Problem of numerical calculation of structures, inter-
acting with infinite bases. Izv Vuzov Stroit Arkhit (issue 4):47–52 (in Russian)
221. Rozin L A, Yevdokimov B M (1993) Method, based on a combination of the potential and
finite-element methods. In: Issues od dynamics and strength, Zinatne, Riga, 43:49–55 (in
Russian)
222. Rvachev V L (1959) The pressure on an elastic half-space of a stamp with a wedge-shaped
planform. J Appl Math Mech 23:229–233
[email protected]
223. Rvachev V L (1982) Theory of R-functions and some of its applications. Naukova Dumka,
Kyiv (in Russian)
224. Rvachev V L, Kurpa L V (1987) R-functions in the problems of theory of plates. Naukova
Dumka, Kyiv (in Russian)
382 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

225. Rvachev V L, Protsenko V S (1977) Contact problems of theory of elasticity for nonclassical
domains. Naukova Dumka, Kyiv (in Russian)
226. Rybin V S (1990) Design of foundations of buildings under reconstruction. Stroyizdat,
Moscow (in Russian)
227. Rybin V S, Zakharchenko V E (1981) Vertical stresses in a bed stratum due to a load uni-
formly distributed over the area of a polygon. Soil Mech Found Eng 18:202–205
228. Sabonnadière J-C, Coulomb J-L (1986) La méthode des éléments finis. Du modèle. . . a la
CAO. Hermes, Paris – Londres – Lausanne
229. Sakalo V I, Shkurin A A (1985) Universal program for triangulation of a two-dimensional
domains of an arbitrary shape with grid condensations. Probl Prochn (issue 1):106–108 (in
Russian)
230. Samarskiy A A, Gulin A V (1989) Numerical methods. Nauka, Moscow (in Russian)
231. Schiffman R L, Aggarwala D B (1961) Stresses and displacements produced in a semi-
infinite elastic solid by a rigid elliptical footing, In: Proc 5th Intern Conf Soil Mech Found
Eng, Vol. 1, pp. 795–801
232. Schofield A, Wroth P (1968) Critical state soil mechanics. McGraw-Hill, London
233. Sekulović M (1988) Metod konačnih elemenata. Gradevinska knjiga, Beoqrad
234. Selvadurai A P S (1979) Elastic analysis of soil-foundations interaction. Elsevier, Amster-
dam
235. Seymov V M (1976) Dynamic contact problems. Naukova Dumka, Kyiv (in Russian)
236. Shapiro D M (1996) Finite-element calculation of structures and bases. VGASA, Voronezh
(in Russian)
237. Sharma K G, Nagpal A K, Garg M K (1984) Finite element analysis of rafts resting on elastic
half space. Indian Geotechn J 14:28–39
238. Shevchuk L I (1988) Method of calculation of frame systems on an anisotropic elastic layer.
PhD thesis, Minsk (in Russian)
239. Shevlyakov Yu A (1977) Matrix algorithms in theory of elasticity of nonhomogeneous
media. Vyshcha Shkola, Kyiv/Odesa (in Russian)
240. Shibuya T, Koizumi T, Nakahara I (1974) An elastic contact problem for a half-space
indented by a flat annular rigid punch. Intern J Eng Sci 12:759–771
241. Shirinkulov T Sh (1969) Calculation of structures on a solid base. FAN, Tashkent (in
Russian)
242. Shirinkulov T Sh (1972) Calculation of engineering structures on a nonhomogeneous elastic
base. FAN, Tashkent (in Russian)
243. Shirokov V N, Solomin V I, Cheremnik V A, Malyshev M V, Zaretsky Yu K (1971) A
circular rigid plate on a nonlinearly deforming base. In: Proc 4th Budapest Conf Soil Mech
Found Eng. Akademiai Kiadó, Budapest, pp. 757–76
244. Shirokov V N, Solomin V I, Malyshev M V, Zaretski Yu K (1970) Stress state and displace-
ments of a ponderable, nonlinearly deformable soil half-space under a circular rigid footing.
Soil Mech Found Eng 7:2–8
245. Shtayerman I Ya (1949) Contact problem of theory of elasticity. Gostekhizdat, Moscow (in
Russian)
246. Smirnov V A (1978) Calculation of plates of a complex contour. Stroyizdat, Moscow (in
Russian)
247. Smirnov V A (1983) Calculation of orthotropic plates resting on an elastic base with a coef-
ficient of subgrade reaction, variable over the area. In: Applied theory of elasticity. SPI,
Saratov, pp. 12–21 (in Russian)
248. Sneddon I N (1946) Boussinesq’s problem for a flat-ended cylinder. Proc Camb Phil Soc
42:29–39
[email protected]
249. SNIP (Construction Rules and Regulations) 2.02.01-83. Bases for buildings and structures
(1985) Stroyizdat, Moscow (in Russian)
250. Solomin V I (1960) Calculation of rectangular plates on an elastic half-space by the grid
method. Stroit Mekh Raschet Sooruzh (issue 6):12–17 (in Russian)
References 383

251. Solomin V I, Shirokov V N, Komarov E A (1968) Computation of rectangular plates on

elastic layer of limited thickness. Soil Mech Found Eng 5:294–298
252. Solomin V I, Shmatkov S B (1986) Methods of calculation and optimal design of ferrocon-
crete foundation structures. Stroyizdat, Moscow (in Russian)
253. Solomin V I, Sytnik A S (1974) Design of foundation slabs of complex configuration and
variable stiffness. Soil Mech Found Eng 11:300–305
254. Solomin V I, Sytnik A S, Vysokovskiy V L (1977) Calculation of foundation plates of
complex configuration. Stroit Mekh Raschet Sooruzh (issue 2):39–41 (in Russian)
255. Solomon L, Zamirescu I (1963) Some approximate formulae and solutions in the contact
problem for punches with a plane bounded non-elliptical basis. In: Applications of the theory
of functions in continuum mechanics: Proc Intern Symp Tbilisi, pp. 400–409
256. Solovyov L Yu, Solovyov Yu I (1995) Distribution of stress at the action of a circular punch
on an elastic half-space. Izv Vuzov Stroit (issue 9):22–26 (in Russian)
257. Sorochan E A (1986) Foundations of industrial buildings. Stroyizdat, Moscow (in Russian)
258. Sorochan E A, Trofimenkov Yu G (ed.) (1985) Bases, foundations and underground struc-
tures: designer’s handbook. Stroyizdat, Moscow (in Russian)
259. Szegö G (1975) Orthogonal polynomials, 4th edn. Amer Math Soc, Colloquium Publ., Vol.
23__33 Providence, Rhode Island
260. Timoshenko S P, Goodier J N (1970) Theory of elasticity, 3rd edn, McGraw-Hill, New York
261. Timoshenko S, Wojnowski-Krieger S (1959) Theory of plates and shells. McGraw- Hill,
New York
262. Traub J F (1982) Iterative methods for the solution of equations. Chelsea, New York
263. Trofimenkov Yu G, Vorobkov L N (1981) Field methods for the studies of constructive
properties of soils. Stroyizdat, Moscow (in Russian)
264. Tsytovich N A, Abelev M Yu, Sidorchuk V F, Polishchuk A I (1979) Experimental inves-
tigation of the stress-strain state of loess soils in the base of rigid plates. Soil Mech Found
Eng 16:140–145
265. Tsytovich N, Berezantsev V, Dalmatov B, Abelev M (1974) Foundation soils and substuc-
tures. Mir, Mocow
266. Ufliand Ia S (1965) Survey of articles on the application of integral transforms in the theory
of elasticity. North Carolina State University, Raleigh
267. Ufliand Ia S (1977) Method of coupled equations in mathematical physics problems. Nauka,
Leningrad (in Russian)
268. Ukhov S B (1973) Calculation of structures and bases using the finite-element method. MISI,
Moscow (in Russian)
269. Ukhov S B, Semenov V V, Znamenskii V V et al. (1994) Soil mechanics, bases and founda-
tions. Association of Engineering Institutions, Moscow (in Russian)
270. Urisman V S (1976) Settlement and tilting of a rigid rectangular foundation on a compress-
ible foundation bed of finite thickness. Soil Mech Found Eng 13:270–275
271. Vainberg D V, Gerashchenko V M, Roytfarb I Z et al. (1967) Grid equations of plate bend-
ing derived by variational method, In: Resistance of materials and theory of constructions.
Budivelnyk, Kyiv, pp. 23–33 (in Russian)
272. Valov G M (1968) Infinite elastic layer and half-space under the action of a ring-shaped die.
J Appl Math Mech 32:917–930
273. Varvak P M, Varvak L P (1977) Method of grids in problems of calculation of engineering
[email protected]
structures. Stroyizdat, Moscow (in Russian)
274. Varvak P V, Ryabov A F (ed) (1971) Handbook on theory of elasticity (for engineers and
builders). Budivelnyk, Kyiv (in Russian)
275. Vaynberg D V, Vaynberg Ye D (1970) Calculation of plates. Budivelnyk, Kyiv (in Russian)
384 4 Contact Interaction of Shallow Foundations with Nonhomogeneous Bases

276. Veryuzhskii Yu V, Ikonin S V, Savitskii V V (1982) Application of potential method in

calculation of rigid foundations. In: Bases and foundations. Budivelnyk Kyiv 15:21–25 (in
Russian)
277. Veryuzhskii Yu V (1978) Numerical methods of potential in some problems of applied
mechanics. Vyshcha Shkola, Kyiv (in Russian)
278. Vinokurov L P (1956) Direct method of solving of spatial and contact problems for arrays
and foundations. Kharkiv State University, Kharkiv (in Russian)
279. Vinokurov Ye F (1972) Iterative method of computation of bases and foundations. Nauka i
tekhnika, Minsk (in Russian)
280. Vlasov V Z, Leontiev N N (1960) Beams, plates and shells on an elastic base. Phys. Math.
Liter. Publ., Moscow (in Russian)
281. Vorovich I I, Aleksandrov V M, Babeshko V A (1972) Nonclassical mixed problems of
theory of elasticity. Nauka, Moscow (in Russian)
282. Vronskii A V (1969) Considering the non-linear relationship between the settling of the
foundation bed and load in the computation of rigid beams on non-uniform foundations.
Soil Mech Found Eng 6:160–164
283. Vyalov S S (1978) Rheological principles of soil mechanics. Vysshaya shkola, Moscow (in
Russian)
284. Way S (1940) Some observations on the theory of contact pressures. Trans ASME J Appl
Mech 7:147–157
285. Whitman R V, Richart F E (1967) Design procedures for dynamically loaded foundations. J
Soil Mech Founds Divn ASCE 93:169–193
286. Yegorov K E, Nitchiporovich A A (1961) Research on the deflection on foundations. In:
Proc 5th Intern Conf Soil Mech Found Eng, Vol. 1, pp. 861–866
287. Yerzhanov Zh S, Karimbayev T D (1975) Finite-element method on problems of rock
mechanics. Nauka, Almaty (in Russian)
288. Zaretskii Yu K (1989) Lectures on modern soil mechanics. Rostov State University, Rostov
(in Russian)
289. Zaretskii Yu K, Lombardo V N (1983) Statics and dynamics of earth-fill dams. Energoat-
omizdat, Moscow (in Russian)
290. Zavarykin L G, Mazo B M (1989) Calculation of plates on a linearly deformable base by
shooting method. In: Resistance of materials and theory of structures. Budivelnyk, Kyiv
54:43–46 (in Russian)
291. Zhemochkin B N, Sinitsyn A P (1962) Practical methods of calculation for beams and plates
on an elastic base without Winkler hypothesis. Gosstroyizdat, Moscow (in Russian)
292. Zienkiewicz O C (1971) The finite element method in engineering science, McGraw-Hill,
London
293. Zienkiewicz O C, Cheung Y K (1968) The finite element method in structural and continuum
mechanics. McGraw-Hill, London
294. Zienkiewicz O C, Lewis R W, Stagg K G (eds.) (1979) Numerical methods in offshore
engineering. Wiley, London

[email protected]
Chapter 6
Spatial Contact Problems for Porous
Elastic Bases

Abstract The sixth chapter presents solutions of spatial problems of applied

geomechanics related to variation of pore pressure in the soil. The influence of the
pore pressure decline on the soil settlement and cracking as well as the induced seis-
micity and other environmental hazards due to pumping out gas and oil deposits or
intense removal of underground water at industrial or civil engineering is discussed.
The methods for numerical modelling of soil mass deformations due to the reduc-
tion of the pore pressure are described. The approach is based on the application of
integral representations for displacements in a half-space saturated with liquid (or
gas) according to the theory of linear pore-elasticity (filtration consolidation). Spa-
tial deformation of the earth surface due to operating horizontal gas-and-oil wells or
water drains is studied with the account of the run-off mode. Finally, the results for
boundary-element solutions of the spatial contract interaction of structures with the
soil at reduced pore pressure are presented.

Underground mining, intense employment of ground waters, oil and gas extraction
result in violation of equilibrium in the geological environment and, first of all,
in the changes in geostatic and geodynamic stress fields in the surrounding rock
mass [22].
A pore pressure decline results in a volume decrease of the environment con-
stituents due to the pore volume decrease. In the limiting case, if the ground
skeleton is not rigid enough, pores can be completely occluded that causes vol-
ume reduction by the magnitude of the initial porosity, i.e. from 20 to 40% of the
initial value. Pores and cracks can disappear due to their collapse, whose mecha-
nism is related to the deformation process instability, as well as due to slow flow-
ing processes in essentially viscous media. These changes can cause rock mass
movement and land surface subsidence, resulting in complications of the condi-
tions of functioning of surface and underground constructions as well as min-
ing activities on vast areas. Land surface subsidence and trough formation are
the most hazardous. They can lead to shear and tensile stresses in above-ground
[email protected]
structures and finally to overstresses, inadmissible deformations, slopes and even
breakdowns.

S.M. Aleynikov, Spatial Contact Problems in Geotechnics, Foundations of 505

Engineering Mechanics, DOI 10.1007/b11479_6,  C Springer-Verlag Berlin Heidelberg 2011
506 6 Spatial Contact Problems for Porous Elastic Bases

Cases of considerable decline of depressive sinkhole with underground water

withdrawal by 100 m or more, and even up to 1000 m in coal fields have been
reported (see [22]). Underground water level decline causes essential subsidence of
the daylight surface over vast areas measuring sometimes hundreds of square kilo-
meters. For example, according to the observation of scientists from Mexico Engi-
neering Research Institute, during the period of 1900–1969 the territory of Mexico
City gave subsidence to 7 m; moreover, in some places subsidence was non-uniform,
which caused strong deformation of fundamental buildings such as the Theater Cen-
ter, Shrine Museum and others. In Tokyo (Japan) where Artesian water is used, the
subsidence has reached 3.3 m for 50 years, however at some places the subsidence
rate was up to 18 cm per year. Only during 1971 in Tokyo and its suburbs the ground
surface level had lowered by 10–20 cm (as a result of underground water withdrawal
for the purposes of industrial and civil engineering), some sections of the city sub-
way lines gave subsidence to 10 cm during a year and a half. Permanent land surface
subsidence by tens of centimeters has been observed in Bangkok (Thailand) due
to underground water withdrawal and rapid growth of skyscrapers, requiring deep
foundations and deep water lowering. In Long Beach (the USA) land surface subsi-
dence due to oil extraction has reached 7.6 m. In Niigata (Honshu Island, Japan) land
surface subsidence due to gas extraction by up to 50 cm per year was observed and
even after water having been pumped into the exhausted gas wells the settlement was
just reduced to 25 cm per year. Land surface subsidence in Osaka (Japan) between
1935 and 1958 reached 175–190 cm, and in order to eliminate it, the land surface
level was raised by putting in 26.5 million m3 of soil and making a 124-km long
levee.
At oil and gas field development sites the layer pressure decline is a serious envi-
ronmental hazard. An increase of seismicity in oil and gas extraction areas has been
observed in Russia (Northern Caucasus, the Urals and the Volga area, Sakhalin), in
the Western part of Ukraine, and in Uzbekistan. It is related not only to the land
surface subsidence and its cracking, but also to the induced seismicity due to the
interaction of the changing effective stresses and tectonic anomalies. As a result,
seismic activity bursts are possible, resulting in devastating earthquakes.
The most widely known are strong earthquakes in the Gazli gas field
(Uzbekistan) on April 8, 1976 with a magnitude of M = 7.0 and on March 20,
1984 with M = 7.2 [13], as well as the earthquake with M = 6.7 at the depth of
9–10 km under Coalinga oil field in a non-seismic area of Southern California on
May 2, 1983 [19]. In the latter case, all-round investigations have shown the seis-
micity to be related to a trigger action of the stressed state changes, generated by
consolidation of the productive layer. At the earthquake, due to the man-triggered
stress the limiting equilibrium was broken and relative motion of the break edges
started.
The gas and oil fields in question were developed using vertical wells. Employ-
ment of horizontal wells for gas extraction [23], thereby intensifying the extraction
[email protected]
process, resulting in the pore pressure changes in much greater layer volumes, can
lead to a more significant effect on the rock mass stressed state. The length of the
horizontal wells can reach 2 km. The pore pressure decline, arising in the vicinity
6 Spatial Contact Problems for Porous Elastic Bases 507

of wells, is spread unevenly over the layer, depending on the anisotropy of the layer
permeability, over the whole thickness of the productive layer. It is important to
perform mathematical modelling for horizontal gas wells functioning in relatively
thick layers (3–20 m) with a width of 400 m or more.
Development of oil and gas fields leads to specific man-triggered geodynamic
processes. References [18, 22] report on apparent consequences of processes related
to land surface subsidence, pipe casing failures and crushing, pore liquid squeezing
out of non reservoirs into the pore space of reservoirs, deformation of engineering
and industrial constructions and, what is quite essential, the well sealing breakdown,
changes of the reservoir characteristics of rocks due to the deformation of layers,
containing mass, etc.
The experience of development of hydrocarbon fields using the horizontal wells
has shown that man-triggered consequences sometimes result in a situation when
the economic gain in development is less than the man-triggered damage [6, 26].
These consequences can be manifested as follows:

– man-triggered land surface subsidence (stable lowering) with a total amplitude up

to several metres, as a rule non-uniformly distributed over the area;
– man-triggered “localized” earthquakes in seismic activity areas with a seismic
focus, located much deeper than the object depth (Neftekamsk, Russia, 1995).

Such consequences lead to a whole series of complications in construction and

functioning of civil and industrial structures:

– breaking the well airtightness, resulting in the gas eruptions into the upper layers
and to the surface;
– deformation of industrial and engineering structures.

Displacements of the Earth surface points are caused by a complicated three-

dimensional field of stress and deformation of both the reservoir bed itself and the
surrounding rock mass. However, in the framework of the present-day approaches
the reservoir bed deformation is studied separately from the deformation of its top
and bottom as well as of the overlying and underlying rocks. Besides, the produc-
tive layer is usually considered to be of a very simple geometrical shape, and the
loads are considered axisymmetric. In fact, the reservoir has a very complex shape,
and the loads vary over the productive layer volume and surface. Calculations of the
arising fields of rock displacements, deformations and stresses cannot be performed
by the known methods. Elaboration of new mathematical models and new methods
for solving the problem is required. In order to study and predict geodynamic pro-
cesses occurring at oil, gas and water extraction, new mathematical models should
be developed that would combine the reservoir bed and the surrounding rock mass
into a single system.
[email protected]
In Russia, land surface subsidence, resulting from human activity, has been sys-
tematically observed at a number of fields since 1976, occasional observations hav-
ing been carried out even before. The registered land surface subsidence over gas
508 6 Spatial Contact Problems for Porous Elastic Bases

fields has been ranging from 19 cm (North Stavropol oil field) to 27–37 cm per year
(Gazli oil field, Uzbekistan). High-precision measurement data for spatial defor-
mation of the land surface in the Shebelynka gas field area (the right bank of
the Siverskiy Donets river, Kharkiv province, Ukraine) are analyzed in [5]. The
comparison of the reference point coordinates, measured in different years, has
shown the biggest displacements to have occurred before 1975, during the period
of the most intense gas extraction. The observation point displacements are noted
to occur at a variable rate even now, in the central area of the gas field they are
larger – near 13 mm per year, at other points – 13 mm per year. Some coor-
dination between vertical and horizontal displacements has been noticed, but not
explained.
The predicted land surface subsidence values above the Astrakhan (Russia,
seam thickness 100 m) and Qaragandy (Kazakhstan, seam thickness 1500 m)
fields are equal to 0.3–0.4 and 1.0–2.0 m, respectively. One should note that the
Earth’s surface subsidence can be considerably (up to 5 times) larger than the
productive layer thickness change due to the compression of overlying water-
saturated sandy and clay soils. The subsidence of these soils is caused by
squeezing out water into the pore space of the reservoir bed at the pressure
decrease as well as by change of the level of near-surface and underground
water, being pumped out for industrial and economic needs. Thus, the actual
land surface subsidence can only be obtained by taking into consideration the
deformation of the whole earth stratum thickness rather than solely the reservoir
bed.
In the course of developing deformable reservoirs in the surrounding soil mass a
field of normal and tangential stresses is formed. For large-thickness layers (of the
order of 1000 m) tangential stress values can amount up to 50% of the seam pressure
decline value, and if the seam thickness is about 200 m – not more than 20–30%.
However, even in this case the stress can cause one-way flow of plastic soil along
the stratification planes. In this case one has to do with transverse-and-longitudinal
bending, the most hazardous kind of loading for the elements of industrial under-
ground structures.
A rather detailed analysis of the most noticeable land surface subsidence cases
due to decline of water table as well as gas and oil field development, having
taken place in the world practice, is performed in reviews [7, 14, 18]. Note that
the problem of land surface subsidence, caused by the pore pressure change due
to development of oil and gas seams at shallow depth, has a rather complicated
character. Traditional methods of estimation of the stress-strained state around the
excavations at great depth are inapplicable in this case. The situation is aggra-
vated by layered rock stratification and considerable nonhomogeneity. Hence, effi-
cient investigations require invoking new numerical methods of structural and soil
mechanics in spatial formulation. This will enable one to improve the existing pre-
diction methods used to estimate the effect of large-scale hydrotechnical high-rise
[email protected]
and underground construction on the environment and the existing buildings, in par-
ticular when erecting deepened structures and large-scale structures for oil and gas
industry.
6.1 Soil Mass Deformation Due to the Pore Pressure Decline 509

6.1 Soil Mass Deformation Due to the Pore Pressure Decline

6.1.1 Integral Representation of Displacements in a Porous

Elastic Medium
Consider a porous elastic medium saturated with a fluid or a gas. The pore fluid rate
is expressed, according to the Darcy’s law, as follows [15]:

∂H
Vi = −k
∂xi
where k is isotropic permeability, H = − ρg P
+ H0 is the total head, H0 = const is the
excessive head, ρ is the liquid density, g is the acceleration of gravity, P is pressure.
Since the fluid flow rate from a unit volume of a porous body is given by eV =
∂Vi
∂xi , then for an incompressible fluid the value eV should be equal to the volume
decrease rate which, in turn, is equal to the volume decrease rate with the opposite
sign εV = εii .
Thus,
 2 
∂εV ∂ H ∂ 2H ∂ 2H
=k + 2 + 2
∂t ∂x2 ∂y ∂z
or

∂εV k
= − P , (6.1)
∂t γf

where γ f is the fluid specific gravity,  is the Laplace operator.

By introducing effective σij and total σij stresses in a typical way for saturated
porous media [4, 10], given by

σij = σij + Pδij , (6.2)

one obtains
 
∂εV 1 ∂σ0 1 ∂σ0 ∂P
= = − , (6.3)
∂t K ∂t K ∂t ∂t

where K is the bulk modulus of the soil skeleton deformation (K = E

3(1−2ν) ); σ0 =
σii
3 and σ0 = σ3ii are the mean effective and total stresses, respectively; E is the
modulus of elasticity, ν is the Poisson ratio.
Thus, from Eqs. (6.1) and (6.3) one arrives at the equation, which determines the
pressure change in a porous medium:
[email protected]
∂P ∂σ0 kK
CP = − , C= . (6.4)
∂t ∂t γf
510 6 Spatial Contact Problems for Porous Elastic Bases

The equations of equilibrium

∂σij
= 0, i,j = 1, 2, 3,
∂xj
with the account of Eq. (6.2), enable the pore pressure and the effective stresses to
be related by

∂σij ∂P
+ =0.
∂xj ∂xi
Assuming the soil skeleton material to obey the Hooke’s law
 
ν
σij = 2G εij + σij εkk ,
1 − 2ν
where G = 2(1+ν)E
is shear modulus, εij = 12 (ui,j + uj,i ) is the deformation tensor,
one arrives at three differential equations of elliptic type in displacements

2G(1 − ν) ∂εV ∂P
· + G ui + =0. (6.5)
(1 − 2ν) ∂xi ∂xi
Equations (6.4) and (6.5) are interrelated and should be solved together. Note that
the principal complications arise from the term ∂σ ∂t = q(xi ,t) in Eq. (6.4), whose
0

value should be set from physical considerations [24]. When ∂σ ∂t = 0 or σ0 = const

0

(theory of simple consolidation), Eqs. (6.4) and (6.5) are separated from each other,
and the situation is essentially simplified. In this case the system under consideration
is similar to the thermoelasticity equation system [11], the equations of both systems
Eαt
coinciding at the substitution of P by – 1−2ν T, C by κ 2 , where at is linear thermal
expansion coefficient, κ is thermal conductivity coefficient, T is temperature.
Equations (6.4) and (6.5) should be complemented by boundary conditions for
total surface forces

σij nj = σij nj + Pni = ti + Pni = σi (N) , N ∈ Sσ , (6.6)

displacements

ui (N) = fi (N) , N ∈ Su , S = Su + S σ , (6.7)

as well as the initial and boundary conditions for the pore pressure

P(N,O) = Pf (N) when t = 0 , N ∈ V ,

P(N,t) = Pg (N,t) , N ∈ S1 ,
(6.8)
∂P
[email protected]
− ni = Ph (N,t) , N ∈ S2 , S = S1 + S2 , t>0.
∂xi
In order to determine the field of displacements in a medium due to the pore
pressure change only (uniform boundary conditions over Su and Sσ ), one can use
6.1 Soil Mass Deformation Due to the Pore Pressure Decline 511

Maizel’s method for thermoelasticity [9, 11], based on the application of the theorem
of reciprocity in an auxiliary state chosen in a special way (with zero temperature).
Then an integral representation for the field displacement component uk is obtained:


(k)
uk (ξ ,η,ς ) = − P(N,t)Uj,j (N,K)dV(N) , t>0, (6.9)
V

(k)
where N = (x1 ,x2 ,x3 ), and Uj,j is the dilatation of the stressed state due to a con-
centrated unit force acting at the point in the direction parallel to the xk axis. In Eq.
(6.9) the pressure function P(N, t) is assumed to be known (e.g., from the experi-
mental data) or to be found according to Darcy’s equation (6.4) with the initial and
boundary conditions of Eq. (6.8).
The pore pressure can be obtained in an explicit form using a general integral
representation [2]

   
∂P
αP(N,t) = F∗P−G∗ dS + (G ∗ ψ + f · G)dV , (6.10)
∂n
S V

where α = 1 for N∈ V; α = 0 for N∈ / V; ψ is the volume source;

∂G
F(N,t; K,τ ) = − · ni is the directed flow due to G;
∂xi
G is the Green’s function (for an unlimited space)

 
r2
G(N,t; K,τ ) = exp − · [4π (t − τ )]3/2 ;
4(t − τ )
ni are the direction cosines of an external normal to the boundary S of the domain
V; r2 = (xi –ξ i ) · (xi –ξ i ). The asterisk in Eq. (6.10) means the Riemann convolution,
determined according to the rule

t
(ϕ ∗ χ )(x,t) ≡ ϕ(x,t − τ )χ (x,τ )dτ .
0
In practice, the determination of the pressure field P(N, t) for complex-shaped
domains using the general integral representation (6.10), is encumbered, hence
finite-difference-based [17] and finite-element-based [1] numerical methods as well
as approaches, including Laplace transformations and step-by-step processes of time
variation with spatial discretization according to the boundary-element method [2],
[email protected]
are applied.
Thus, in the case the dilatation function being known, from the given or calcu-
lated pore pressure decline variation, displacements of any points of an elastic mass
can be obtained using direct spatial integration with the representation of Eq. (6.9).
512 6 Spatial Contact Problems for Porous Elastic Bases

6.1.2 Dilatation Relations

The efficiency of application of the general integral representation (6.9) for the
calculation of displacements of a porous elastic massif due to the pore pressure
decline to a great extent depends on the availability of the explicit expression of the
(k)
dilatation function εV = Uj,j for the spatial stressed state. This, in turn, requires
the availability of fundamental singular solutions of spatial problems of theory of
elasticity Ui(k) (influence functions). In our case, accurate analytical solutions for
concentrated forces for the case of infinite domains with boundary condition inter-
faces are of great interest. First of all, this category of domains includes an elastic
homogenous half-space, two coupled half-spaces with different elastic properties
(k)
and layered half-spaces. For these cases the Green’s tensor Ui (N, K) is given in
Chap. 1 and in Appendix A. Application of such contact models is well suited for
simulating elastic deformations of a layer and the surrounding rock mass in case of
functioning of horizontal wells to be considered below.
Elastic half-space. The Green’s tensor components in an expanded form
( (  
( (k) (
(Ui ( = U (k) , V (k) , W (k) .

for a rectangular (Cartesian) coordinate system are given in Appendix A. Using

(k)
these formulae and Hooke’s law, the expressions σij for stresses arising in an elas-
tic half-space due to a concentrated unit force, applied at a point K(ξ ,η,ζ ) in the
direction of axis xk , can be determined in a standard way. Then the dilatation rela-
tions are obtained in a standard way according to the known dependences

1 − 2ν (k) 1 − 2ν (k)
εV(k) = $ = (σxx + σyy
(k)
+ σzz(k) )
E E
or directly by the definition

(k) ∂U (k) ∂V (k) ∂W (k)

εV = + + .
∂x ∂y ∂z
Detailed computations in the general case lead to extremely bulky equations.
The most simple form of the dependences under consideration is obtained in the
case of the concentrated force being applied on the half-space surface, when ς = 0,
R1 = R2 = R (i.e. when Boussinesq and Cerruti solutions are used) [8]

1 − 2ν (1)
(1) (1 − 2ν)(1 + ν) x1
εV = θ =− · 3 , (6.11)
E πE R
1 − 2ν (2) (1 − 2ν)(1 + ν) y1
εV(2) =
[email protected] θ =− · 3 , (6.12)
E πE R
(3) 1 − 2ν (1 − 2ν)(1 + ν) z
εV = θ (3) = − · 3 . (6.13)
E πE R
6.1 Soil Mass Deformation Due to the Pore Pressure Decline 513

Coupled elastic half-spaces. In order to calculate the deformation of surfaces,

bounding the layer, one should know the dilatation functions of the stressed state
caused by the concentrated unit forces on the interface of the elastic half-spaces.
For this purpose, first of all, the Plevako fundamental solution is suitable [12].
(k)
Based on the equations of Sect. 1.3 for the Green’s tensor components Ui =
{U (k) , V (k) , W (k) } for the lower half-space, after conventional but cumbersome
transformations, one arrives at the sought dilatation equations
6
1 x1 x 1 z2
εV(1) = (β1 − 2λ + τ − φ) 3 + 3(β1 + τ ) 5 +
16π (1 − ν1 )G2 R R
x1 z(2R + z) x1 x1 z(R − z)
+ 4χ 3 − 4δ + 2χ 4 + (6.14)
R (R + z)2 R(R + z)2 R (R + z)

x1 (R − z)(3R + z) x1 z(5R + 3z)(R − z)(2R + z)
+δ −χ ,
R3 (R + z)2 R5 (R + z)2
6
(2) 1 y1 y1 z2
εV = (β1 − 2λ + τ − φ) 3 + 3(β1 + τ ) 5 +
16π (1 − ν1 )G2 R R
H1 Z (2R + z) y1 y1 z (R − z)
+ 4χ − 4δ + 2χ 4 + (6.15)
R3 (R + z)2 R (R + z)2 R (R + z)

y1 (R − z)(3R + z) y1 z(5R + 3z)(R − z)(2R + z)
+δ − χ ,
R3 (R + z)2 R5 (R + z)2
(3) z 2φ (2R + z)(R − z)
εV = (τ − t) 3 − +φ . (6.16)
R R(R + z) R3 (R + z)

In Eqs. (6.14), (6.15), and (6.16) the following notations are assumed:

χ = 2(α − β1 ) ,
δ = (3 − 4ν2 )α + [1 − 2(ν1 − ν2 )]β1 − 2λ ,
φ = (3 − 4ν2 )α − [3 − 2(ν1 + ν2 )]β1 ,
τ = (2α − β1 ) ,
t = 2(ν2 − ν1 )β1 + (3 − 4ν2 )α ,
β1 = β + 1 ,
μ−1 G2 E2 (1 + ν1 )
β= ,μ= = ,
μ(3 − 4ν1 ) + 1 G1 E2 (1 + ν2 )
8μ(1 − ν1 )[μ(1 − ν1 ) + (1 − ν2 )]
α= ,
(μ + 3 − 4ν2 )[μ(3 − 4ν1 ) + 1]
μ
λ = 4(1 − ν1 ) ; x1 = x − ξ , y1 = y − η.
μ+1
In case concentrated forces acting at a given depth in the lower (upper) half-
space, dilatation relations are obtained in a quite similar way. They are not quoted
[email protected]
here because of being too cumbersome. Note that for the case of a three-layer system
with an internal productive layer, the dilatation relations are given in Chap. 1 in
Fourier transforms (Sect. 1.4.3).
514 6 Spatial Contact Problems for Porous Elastic Bases

6.2 Distribution of Pressure in a Layer in Case of Functioning

Horizontal Wells

6.2.1 Distributed Sources of Predetermined Intensity

Consider a well, whose axis presents a segment of a spatial straight line L between
points M1 and M2 , whose length is denoted as l (Fig. 6.1).

Fig. 6.1 To the

determination of pressure
near a well of finite length

Let sinks q(s) of a specified intensity, calculated per unit length of the straight
line L, be distributed along this segment. Then, the velocity potential ϕ(x, y, z) due
to the action of the sinks distributed along straight line L within an infinite space, is
given by [15]

1 q(s)ds
ϕ(x,y,z) =  , (6.17)
4π (x − ξ )2 + (y − η)2 + (z − ς )2
L

where s is the path length, measured from the starting point M1 of the straight line
L; ξ , η, ζ are Cartesian (rectangular) coordinates of points of this line, M(x, y, z) is
an arbitrary observation point.
Since the velocity potential can be written in the form [15]
 
P
ϕ(x,y,z) = −k +z +C ,
ρg
then for the pore pressure one obtains

γf q(s)ds
P = P0 − γf · z −  (6.18)
4π (x − ξ )2 + (y − η)2 + (z − ς )2
L

where k is the coefficient of filtration, ρg = γf is the fluid or gas specific gravity,

P0 = const is the reference pressure value, with respect to which the extra pressure
is evaluated.
Let us reduce Eq. (6.18) to a form, suitable for practical calculations. We deter-
[email protected]
mine direction cosines of the segment M1 M2 , based on the relations

x2 − x1 y2 − y1 z2 − z1
cos α = , cos β = , cos γ = ,
& & &
6.2 Distribution of Pressure in a Layer in Case of Functioning Horizontal Wells 515

then

ξ = x1 + s · cos α , η = y1 + s · cos β , ς = z1 + s · cos γ

and Eq. (6.18) is finally written as

&
γf q(s)ds
P(x,y,z) = P0 − γf · z −  . (6.19)
4π r12 − 2r1S s + s2
0

Here the following notations are used:

r12 = (x − x1 )2 + (y − y1 )2 + (z − z1 )2 ,

r1S = (x − x1 ) cos α + (y − y1 ) cos β + (z − z1 ) cos γ .

At a given function of the sink intensity q(s), the regular integral, contained in
Eq. (6.9), can be evaluated numerically with any accuracy, based on a suitable rule
of integration. In practice, the q(s) function is most often linear

q2 − q1
q(s) = q1 + s,
&
where q1 and q2 are the sink intensity values at the beginning and at the end of the
well. In this case, the integration is exact and Eq. (6.19) takes the following form:
 
γf q2 − q1
P(x,y,z) =P0 − γf · z − & − 2r1S & + r1 − r1 +
2 2
4π &
 ⎤
  &2 − 2r1S & + r12 + & − r1S
(6.20)
q2 − q1 ⎦ .
+ q1 + r1S ln
& r1 − r1S

Finally, in the simplest case, when q1 = q2 = q = const,


γf & − r1S (& − r1S )2 + r̄2
P(x,y,z) = P0 − γf · z − ln =
4π r1 − r1S (6.21)
γf (r1 + r1S )(r2 + r2S )
= P 0 − γf · z − ln
4π r̄ 2
[email protected]
where r̄ is the distance from the point M (x, y, z) to the segment M1 M2 , r̄2 = r12 −r1S
2 ,

r2 is the distance from the point M (x, y, z) to the point M2 (x2 , y2 , z2 ), r22 = (x −
x2 )2 +(y−y2 )2 +(z−z2 )2 , r2S is the projection of the segment MM2 on the direction
M1 M2 , r2S = (x2 − x) cos α + (y2 − y) cos β + (z2 − z) cos γ = & − r1S .
516 6 Spatial Contact Problems for Porous Elastic Bases

6.2.2 Account of the Finite Radius of the Well

Potential representation of the pore pressure, Eq. (6.19), and its special cases, Eqs.
(6.20) and (6.21), are convenient for engineering calculations, but do not contain
some important characteristics of the inflow process to the wells, including the well
radius and the size of the feed area. Hence, for the model calculations, along with
Eqs. (6.19), (6.20), and (6.21) the following representations will be used. Let a cylin-
drical surface of radius R2 (Fig. 6.2a) be treated as the feed area boundary for a well
with radius R1

(b) (c)

(d)
(a)

Fig. 6.2 Horizontal well feeding area (a, b, c) and pore pressure distribution in the near-face
area (d)

In the simplest case, the well axis is considered horizontal, located at a depth z0 .
The axes of the spatial rectangular co-ordinate system are chosen in a way, shown in
Fig. 6.2a–c. The well length is considered to be equal to l. The pressure distribution
is considered to be axisymmetric. In each section, perpendicular to the well axis
(Fig. 6.2d), it varies according to a logarithmic law

P − Pk ln (r/R2 )
= ,r= y2 + (z − z0 )2 . (6.22)
Pc − Pk ln (R1 /R2 )
Furthermore, it is assumed that the pressure at the feed contour Pk and at the well
Pc varies only along the well axis, this variation being linear:

(1) (2) (1) y − y1
Pk = Pk + Pk − Pk , (6.23)
y2 − y1

(1) y − y1
Pc = P(1)
c + Pc − Pc
(2)
. (6.24)
y2 − y1

Here Pk(i) , Pc(i) , (i = 1, 2) are given constant pressure values at the left and the
right well ends.
The model representations under consideration are easily generalized for the case
of an inclined well with a given orientation within the layer as well as for more
[email protected]
complicated laws of the pressure distribution in the feed area of the well. They can
be used for quantitative and qualitative description of the processes of spatial fluid
flow in a saturated porous medium.
6.3 Contact Problems for Foundation Structures at a Reduced Pore Pressure in the Soil 517

6.3 Contact Problems for Foundation Structures at a Reduced

Pore Pressure in the Soil
Sections 6.1 and 6.2 are devoted to the method of numerical modelling of soil mass
deformation caused by the pore pressure decline due to oil and gas extraction or
functioning of water supply drains. The approach is based on application of inte-
gral representations for spatial displacements, according to theory of linear pore
elasticity. Mindlin solutions (for elastic half-spaces), Plevako solutions (for coupled
half-spaces) as well as any other influence functions with known dilatation functions
under spatial deformation can be used as fundamental solutions for the soil skeleton
( (
Ui = (U (k) , V (k) , W (k) (. In this section, the formulations obtained are used for
(j)

the description of the processes of spatial contact interaction of shallow and deep
foundation structures with the soil mass. Formulations of the contact problems of
such type are important due to the following reasons. If the area of the pore pres-
sure reduction is comparable with the size of the foundation structure itself and is
located close to it, then such a foundation undergoes inadmissible slopes and settle-
ments which are sometimes very significant and lead to hazardous deformations of
above-foundation structures. Besides, it is known that the pore fluid tends to flow
to pile foundations what thereby results in the formation of a reduced pore pressure
area in the foundation active area [3, 21]. In the course of a pile foundation loading,
its additional indentation occurs with the formation of a contact layer on the surface
of the stem. At a shear settlement, the grain structure of the contact layer tends to the
state with a critical density. Thus, along the contact surface of the pile the conditions
for water draining out of the soil are formed, i.e. a drain layer is formed around the
pile. As a result, the existing load is increased by growing forces due to the pore
pressure decline. All this leads to an additional settlement and sometimes even to
the pile breakdown. The latter is especially dangerous in the case of compound pile
structures. The analyzed features of behaviour of foundations are not manifested
directly after the foundation having been constructed. The described contact filtra-
tion effect is revealed after some time, usually, a long period after the end of the
foundation construction. As a result, the arising settlements are unexpected and it is
difficult to find their reasons at once and to explain them correctly.
Therefore, it is quite clear that the studies of spatial contact interaction of
deepened structures in case of a reduced pore pressure in the soil require further
development.

6.3.1 Integral Equations of a Spatial Contact Problem

Consider a spatial contact problem for foundation structures located on the ground
surface or deepened into the soil either in the vicinity of oil and gas deposits or close
[email protected]
to watershed areas. A conventional consideration of the contact interaction here is
complicated due to additional settlements caused by the pore pressure decline in the
soil.
518 6 Spatial Contact Problems for Porous Elastic Bases

For further consideration the following assumptions will be made: (1) the founda-
tion structure under study is an absolutely rigid body of a rather arbitrary shape; (2)
the structure is subjected to the action of a static spatial system of forces, reduced
to a basic force vector and a basic moment which are supposed to be known; (3)
the soil base is treated as an isotropic, porous elastic, linearly deformable, homoge-
neous, weightless half-space; (4) the structure and the soil base are coupled over the
contact surface.
The main integral equation of the contact problem is obtained, based on the sim-
ilarity of consolidation and thermoelasticity problems, by introduction of a basic
state and an auxiliary state, using Betti’s theorem of reciprocity [11].
 
(i) (i)
ui (K) = σj (N) Uj (K,N) d + P (N) k,k (K,N) dV (N) . (6.25)
 V

Here K(ξ , η, ζ ) is the point of application of the concentrated unit force, σ j

are the components of the stress vector, distributed along the surface Γ of the soil
base contact with the structure, ui (K) is the displacement of an arbitrary point K,
situated on the contact surface. Equation (6.25) expresses the fact that the value of
the displacement ui (K) is equal to the sum of the works of the arising contact stresses
(i)
σ j (N) in the basic state on the displacements Uj of the auxiliary state, plus the pore
(i)
pressure work at the relative changes of the auxiliary state volume UK,K , the values
of the index i =1, 2, 3 denoting the direction of the concentrated unit force parallel
to the X, Y, Z axes, respectively. In Eq. (6.25), as usual, summation over repeated
indices is assumed.
Then, similarly to the Sect. 2.1, we take into account the fact that the displace-
ments of any point of the contact surface can be expressed in terms of displacements
x , y , z of the whole rigid construction as well as angles of its rotation ψ x , ψ y ,
ψ z with respect to the coordinate axes as follows:

U = x − z · ψy + y · ψz ,
V = y − x · ψz + z · ψx , (6.26)
W = z − y · ψx + x · ψy .

The system of integral equations (6.25) in an expanded form is given by

 
σx (N)U (1) (K,N) + σy (N)V (1) (K,N) + σz (N)W (1) (K,N) d−
  
(1)
− P(N)Uk,k (K,N)dV = x − ζ · ψy + η · ψz ,
  V 
σx (N)U (2) (K,N) + σy (N)V (2) (K,N) + σz (N)W (2) (K,N) d−
   (6.27)
(2)
− P(N)Uk,k (K,N)dV = y − ξ · ψz + ζ · ψx ,
[email protected]
  V 
σx (N)U (3) (K,N) + σy (N)V (3) (K,N) + σz (N)W (3) (K,N) d−
  
(3)
− P(N)Uk,k (K,N)dV = z − η · ψx + ξ · ψy .
6.3 Contact Problems for Foundation Structures at a Reduced Pore Pressure in the Soil 519

In order to make the system of Eqs. (6.25) and (6.27) we use, as usual, six static
equations of equilibrium

  
σx (N)d = Px , σy (N)d = Py , σz (N)d = Pz ,
    
σz (N)y − σy (N)z d = Mx , σx (N)z − σz (N)x d = My , (6.28)
  
σy (N)x − σx (N)y d = Mz .


Note that the presence of the terms in Eqs. (6.27), obtained from the volume
integration over the domain V of the pore pressure variation which do not contain
unknown values, makes the pore pressure a loading parameter in the contact prob-
lem under consideration together with Px , Py , Pz , Mx , My , and Mz .

6.3.2 Finite-Dimensional Algebraic Analogue of the Integral

Equation System
The spatial contact problem under consideration is represented in a rather general
formulation by the system of integral equations (6.27) and (6.28), which can be
solved numerically by the boundary-element method.
In order to obtain a finite-dimensional algebraic analogue of the system of Eqs.
(6.27), we proceed similarly to Sect. 2.2. For this purpose, at first, the contact sur-
face is discretized into triangular and quadrangular boundary elements is performed.
Let m be a total number of the boundary contact elements. Then each surface inte-
gral in Eqs. (6.27) and (6.28) is substituted by the sum of integrals over individual
boundary elements. The approximation of the contact stress functions is assumed to
be piecewise constant, and the unknown contact forces are considered in the cen-
tres of the gravity of the boundary elements. The integral equations (6.27) with the
account of the surface discretization only, are given by

 
'
m   
σx (Ni ) U (j) (K,N)d + σy (Ni ) V (j) (K,N)d + σz (Ni ) W (j) (K,N)d =
i=1 i⎧ i i
 ⎨ x − ζ · ψy + η · ψz , j = 1,
(j)
= P(N)Uk,k (K,N)dV + x − ξ · ψz + ζ · ψx , j = 2,
⎩
v x − η · ψx + ξ · ψy , j = 3.
(6.29)
The system of Eqs. (6.29) contains 3m unknown forces σ x (Ni ),
σ y (Ni ), σ z (Ni ) (i = 1,...,m) and six unknown parameters x , y , z , ψ x , ψ y ,
ψ z characterizing the displacement of the structure under consideration as an
[email protected]
absolutely rigid body. The series of the first 3m equations is obtained by the
collocation method by sequential substitution of coordinates of the gravity centre
of the boundary elements into each equation of Eqs. (6.29) which is true for any
point of the contact domain
520 6 Spatial Contact Problems for Porous Elastic Bases
 
'
m   
σx (Ni ) U (j) (Kf ,N)d + σy (Ni ) V (j) (Kf ,N)d + σz (Ni ) W (j) (Kf ,N)d =
i=1 i ⎧ i i
 ⎨  x − ζf · ψ y + ηf · ψ z , j = 1,
(j)
= P(N)Uk,k (Kf ,N)dV + x − ξf · ψz + ζf · ψx , j = 2,
⎩
V x − ηf · ψx + ξf · ψy , j = 3, f = 1,...,m.
(6.30)
After the surface integral discretization, the equations of equilibrium (6.28) take
the form


m
σx (Ni )si = Px , (6.31)
i=1

m
σy (Ni )si = Py , (6.32)
i=1

m
σz (Ni )si = Pz , (6.33)
i=1

m

σz (Ni )yi − σy (Ni )zi si = Mx , (6.34)
i=1

m

σx (Ni )zi − σz (Ni )xi si = My , (6.35)
i=1

m

σy (Ni )xi − σx (Ni )yi si = Mz , (6.36)
i=1

Thus, one obtains a finite-dimensional algebraic analogue of the system of inte-

gral equations (6.27) and (6.28) in the form of (3m+6) linear algebraic equations
(6.30)–(6.36) with respect to 3m contact stresses σ x (Nj ), σ y (Nj ), σ z (Nj ) (j = 1,...,m),
three displacements x , y , z , and three rotation anglesψ x , ψ y , ψ z .
The obtained system of Eqs. (6.30)–(6.36) is similar to the system of Eqs. (2.13)
and (2.14) for the contact problem without the account of the effect of the pore
pressure reduction, the only difference being the fact that Eqs. (6.30)–(6.36) include
the terms, caused by the volume integration. Finally they only result in a change of
the load vector (the right-hand side vector).

6.3.3 Numerical Algorithm of Solution of the Contact Problem

[email protected]
The matrix notation of Eqs. (6.30)–(6.36) is given by

A·Z=B (6.37)
6.3 Contact Problems for Foundation Structures at a Reduced Pore Pressure in the Soil 521
 
D3m×3m C3m×6
where A = is a square matrix of the (3m+6)-th order, Z =
T6×3m 0
 
Q
is the column vector of (3m+6) unknowns, these factors exactly coincide with
q
 
H
the corresponding ones in Eq. (2.15). The vector of the right-hand side B =
h
of the same dimensionality as Z includes six components of the external load as
well as a 3m- dimensional vector H, responsible for the pore pressure reduction in
the domain V. In a detailed notation these vectors are given by
⎛ ⎞ ⎛ ⎞
σx1 ⎛ ⎞ Hx1 ⎛ ⎞
⎜ σy1 ⎟ x ⎜ Hy1 ⎟ Px
⎜ ⎟ ⎜ y ⎟ ⎜ ⎟ ⎜ ⎟
⎜ σz1 ⎟ ⎜ ⎟ ⎜ Hz1 ⎟ ⎜ Py ⎟
⎜ ⎟ ⎜ z ⎟ ⎜ ⎟ ⎜ ⎟
⎜ ⎟ ⎟, H = ⎜ ⎟ Pz
⎟, q = ⎜ ⎟, h = ⎜ ⎟,
Q=⎜ .. ..
. ⎜ ψx ⎟ ⎜ . ⎜ ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ Mx ⎟
⎜ σxm ⎟ ⎝ ψy ⎠ ⎜ Hxm ⎟ ⎝ ⎠
⎜ ⎟ ⎜ ⎟ My
⎝ σym ⎠ ψz ⎝ Hym ⎠ Mz
σzm Hzm
where
 (1)  (2)
Hxi = P(N)εv (Ki ,N)dV, Hyi = P(N)εv (Ki ,N)dV,
 V  (3)
V
Hzi = P(N)εv (Ki ,N)dV, i = 1,2,...,m.
V
Thus, the proposed numerical algorithm includes the following stages: (i) dis-
cretization of the contact surface of the foundation structure and the soil into bound-
ary elements; (ii) discretization of the 3-dimensional domain of the pore pressure
decline into finite 3-dimensional elements; (iii) calculation of the matrix coefficients
at the unknowns and the load vector; (iv) formation and solving of the system of
Eqs. (6.37) for various external force values and the given field of the pore pressure
variation.
As mentioned above (see Sect. 3.2), the obtained algebraic systems have good
conditionality, related to the diagonal predominance that enables one to use the
standard solution methods of the Gauss type, without application of special regu-
larization methods.
Here some explanations regarding the determination of the values Hxi , Hyi , Hzi ,
i = 1,...,m should be made. The details of the numerical integration procedure over
standard elements like pyramids, triangular and quadrangular prisms, are given in
Appendix H. One should only discretize the given domain V of the pore pressure
decline into 3-dimensional elements of the types mentioned. Here it seems possible
to apply any algorithms for the discretization of complex-shaped spatial objects
into 3-dimensional elements. First of all, this applies to the algorithms that are
[email protected]
widely used at the finite-element simulation [1, 25]. However, taking into account
the specific character of the problems under consideration, we propose the follow-
ing approach. As a rule, the source of the pore pressure decline in the soil is a direct
522 6 Spatial Contact Problems for Porous Elastic Bases

sink (drainage, absorbing well with a central axis, etc.), and the form of the whole
domain of the pore pressure decline is such that the sections, orthogonal to the sink
axis, are of a similar shape. Hence, if a flat domain of a section, which is orthog-
onal to the preferred axis, is subjected to disretization, the whole domain V can
be rather easily discretized into elementary pyramids, triangular and quadrangular
prisms. The discretization of the flat domain itself into triangles and quadrangles can
be performed using any of the known algorithms (see e.g. [16, 20]). The algorithm
of automatic discretization of flat domains of a rather general type is described in
detail in Chap. 3. It is sufficiently flexible and optimal and enables the procedure
of generation of spatial grids for representation of the 3-dimensional domain of
the pore pressure variation by a set of elementary pyramids as well as triangular
and quadrangular prisms to be built up based on the geometrical characteristics of
the cross-sections. This algorithm enables the mesh grid to be parametrically con-
densed in the vicinity of the source of the pore pressure decline (the corresponding
examples of the discretization will be shown in Sect. 6.4). All this enables the vol-
ume integrals, contained in Eqs. (6.30)–(6.36), to be effectively computed using the
numerical integration procedures for standard domains (see Appendix H).

6.3.4 Contact Problem for Shallow Foundations

For a visual illustration of the abilities of the approach considered in Sect. 6.3.3,
we make some simplifying assumptions. As an object interacting with the porous
elastic mass, we consider a shallow foundation structure (an absolutely rigid punch
with a flat base of a rather arbitrary shape), resting on the surface of a half-space
(z = 0). Similarly to the case, considered in Sect. 2.5.1, we do not take into account
any friction in the contact domain and assume the punch to be loaded only under
a static load reduced to a vertical force Pz and tilting moments Mx and My . Then,
as usual, we assume that vertical displacements of the punch and the base surface
are equal and there is no load outside the punch. Taking into consideration the pore
pressure variation in the known domain of the soil mass (of the porous elastic half-
space), one should determine the distribution of reaction pressures under the punch
and the displacement parameters of the punch as a rigid body.
In this case, instead of three boundary integral equations (6.27), one has only
one equation expressing the geometrical condition of the contact of the punch and
half-space:
    
1 − ν2 p(ξ ,η)dξ dη (3)
 + P(N)εv (K,N)dV(N) =
πE (x − ξ )2 + (y − η)2
F F (6.38)

=Wc + ψx · (x − xc ) + ψy · (y − yc )
[email protected]
where F is the area of the punch contact with the porous elastic base, p(x, y) = σ z
(x, y, 0) is the sought contact pressures function, Wc is the vertical displacement of
the punch centre (xc , yc ), ψ x , ψ y are the punch slopes with regard to the OX and
6.3 Contact Problems for Foundation Structures at a Reduced Pore Pressure in the Soil 523

(3)
OY axes, respectively, εv (K,N) = (1−2ν)(1+ν)
πE · Rz3 is the dilatation function for the
elastic half-space due to a unit vertical concentrated force acting on the free surface.
Taking into account that among the six equilibrium conditions of Eqs. (6.28) in the
formulation of the contact problem under consideration. it is sufficient to analyze
three integral equations which should be satisfied by the contact pressure field:


p(ξ ,η)dξ dη = Pz ,
F  (6.39)
p(ξ ,η)ξ dξ dη = Pz · xc − My , p(ξ ,η)ηdξ dη = Pz · yc − Mx .
F F

Thus, the spatial contact problem for a punch with a flat bottom, resting on the
surface of a porous elastic half-space, is reduced to finding the parameters Wc ,
ψ x , ψ y , determining the punch location and the distribution of contact pressures
p(ξ , ζ ) over its bottom. After the punch being indented under the conditions of the
pore pressure variation, they will be determined by solving the system of Eqs. (6.23)
and (6.24).
For numerical solution of the system of integral equations (6.38) and (6.39) in
the approximation of a piecewise constant function of contact pressures p (ξ , ζ ) =
const, one obtains the following finite-dimensional algebraic analogue in a more
compact form than Eqs. (6.30)–(6.36):

⎧
⎪
⎪ p δ + p2 δi2 + ... + pm δim − Wc − ψx · (xi − xc ) − ψy · (yi − yc ) + Hiz = 0, i = 1,m;
⎨ 1 i1
p1 s1 + p2 s2 + ... + pm sm = Pz ,
⎪
⎪ p s x + p2 s2 x2 + ... + pm sm xm = Pz · xc − My ,
⎩ 1 1 1
p1 s1 y1 + p2 s2 x2 + ... + pm sm xm = Pz · yc − Mx ,
(6.40)
being a linear equation system with (m+3) unknowns.
 Here pi = p(ξ , ζ ) are contact pressures on the boundary elements, δij =
ω(xi ,yi ,ξ ,η)dξ dη are vertical displacements of the base surface at a point
Fj
(xi , yi ), coinciding with the centre of gravity of the i-th element, due to a unit load,
uniformly distributed over the domain 8 Fj of the j-th element, si is the area of the
  
i-th element, ω (xi ,yi ,ξ ,η) = 1 − v2 π E (x − ξ )2 + (y − η)2 is the influence
function (Boussinesq solution).
While solving the system of Eqs. (6.40), we use the matrix notation:

a · z = b, (6.41)

[email protected]
where  
dm×m c3×m ( (
a= is a square matrix of the order (m+3), d = (δij ( , i,j = 1,m;
t3×m 0
524 6 Spatial Contact Problems for Porous Elastic Bases
⎛ ⎞
1 x1 − xc y1 − yc ⎛ ⎞
⎜1 s1 s2 · · · sm
x2 − xc y1 − yc ⎟
c = −⎜
⎝ ···
⎟ , t = ⎝ s1 x1 s2 x2 · · · sm xm ⎠ ,
········· ········· ⎠
s1 y1 s2 y2 · · · sm ym
1 xm − xc y1 − yc
Z and B are column matrices of the size (m +3),
   
Q H
Z= , B= ,
q ⎛ ⎞ h ⎛ ⎞
p1 Hz1 ⎛ ⎞ ⎛ ⎞
⎜ p2 ⎟ ⎜ Hz2 ⎟ Wc Pz
⎜ ⎟ ⎜ ⎟
[6pt]Q = ⎜ . ⎟ , H = ⎜ . ⎟ , q = − ⎝ ψx ⎠ , h = − ⎝ Pz · xc − My ⎠ .
⎝ .. ⎠ ⎝ .. ⎠ ψy Pz · yc + Mx
pm Hzm
Then the numerical-and-analytical method, described in Sect. 2.3, is used to
calculate the surface integrals (both singular and regular) over an arbitrarily ori-
ented boundary element with a polygonal contour. The components of the vector
H are computed numerically, using the volume integration formulae in accordance
with the discretization of the domain V into pyramids, triangular and quadrangu-
lar prisms (Appendix H). An example of the numerical calculation to estimate the
pore pressure decline influence on the contact interaction process is given below in
Sect. 6.4.3.

6.4 Examples of Numerical Calculations

The technique proposed in Sects. 6.1 and 6.2 for calculating the land surface defor-
mation at oil and gas extraction and underground water withdrawal, is implemented
in a PC software Subsidence-01, which does not require high-end PCs. According to
the algorithm developed, all the above described stages are sequentially performed,
the dilatation representations being applied are programmed as a separate routine
that enables one to use a wide set of the known fundamental solutions for elastically
deformable media, including nonclassical ones with a spatial non-uniformity of
physical and mechanical parameters or anisotropy. According to the gas field extrac-
tion techniques, variation of the pore pressure in a layer is provided in the developed
software product both based on the experimental database and using direct mathe-
matical modelling of the filtration processes. The software outputs the values of the
vertical and horizontal displacements of the land surface points and the soil mass
for the specified set of observation points. The sought values are recommended to
be obtained in the nodes of triangular or quadrangular grids for visual presentation
of the results obtained on the basis of the numerical solution interpolation.
Numerical implementation of the method for the calculation of the land sur-
[email protected]
face deformation and the contact deformation of the foundation structures with
soil for the case of horizontal wells, developed in Sects. 6.1–6.4 on the basis of
the concepts of spatial theory of elasticity, implies the stage of discretization of
6.4 Examples of Numerical Calculations 525

the contact surfaces and spatial domains of the well feed into boundary and finite
(3-dimensional) elements, respectively. For this purpose we use a program for dis-
cretization of single and multiply connected flat domains, bounded by linear seg-
ments or circular arcs, and a specially developed program for spatial discretiza-
tion of complex-shaped 3-dimensional objects (mostly with geometrically similar
configurations in sections, orthogonal to the specified directions, e.g. to a well axis,
arbitrarily oriented within the layer under consideration). Figures 6.3, 6.4, and 6.5
show typical examples of discretization of flat doubly connected sections of spatial
domains, bounding the well feed area for different engineering-and-geological con-
ditions. As seen from the figures, the plotted grids enable one to characterize with
sufficient accuracy the pore pressure field nonuniformity in the well feed area with
the account of the filtration flow nonuniformity.
For visual representation of the possibilities of the approach developed for qual-
itative and quantitative estimation of the effect under investigation, model calcula-
tions have been carried out, part of which is presented below in Sects. 6.4.1–6.4.3.

(a) (b)

Fig. 6.3 Discretization of a ring-shaped centrosymmetric (a) and eccentric (b) feeding area with
condensation near the horizontal well borehole

(a) (b)

[email protected]
Fig. 6.4 Discretization of a centrosymmetric (a) and eccentric (b) feeding area with a rectangular
contour near the horizontal well
526 6 Spatial Contact Problems for Porous Elastic Bases

(a) (b)

Fig. 6.5 Discretization of an elliptic centrosymmetric (a) and eccentric (b) feeding area near the
horizontal well

6.4.1 Spatial Deformation of the Land Surface

Numerical calculations of spatial displacements of the land surface points are shown
in Figs. 6.6–6.12. They were obtained for a horizontal gas well with a length l,
whose axis is positioned at a depth z0 in an elastic homogeneous half-space with the
Poisson ratio v = 0.3. The feed contour of the well was bounded by the productive
(1)
layer thickness H2 = 2 h, at the boundaries of which (R2 =  h) one assumes Pk =
(2) (1) (2)
Pk = 0. At the well contour (R1 = 0.01 h) a constant Pc /Pc = Q = 1 or

(1) (2)
linearly variable along the well axis Pc /Pc = Q = 1 sink was given. In the
calculations presented we use dimensionless spatial coordinates
2 2 2
x̄ = x h, ȳ = y h, z̄ = z h

and displacements

U·E V ·E W ·E
Ū = , V̄ = , W̄ = ,
P · h P · h P · h
[email protected]
(1)
where P = −Pc ; E is the soil deformation modulus.
From the calculation data presented one can clearly see the effect of the well
depth on the deformation of the elastic half-space free surface (Figs. 6.6, 6.7).
6.4 Examples of Numerical Calculations 527

Fig. 6.6 Profiles of dimensionless vertical (a) and horizontal (b) displacements of the land sur-
face in the cross-section orthogonal to the horizontal well axis (l = 5 h) at different depths,
1: z0 = 2.5 h, 2: z0 = 5.5 h

Vertical displacements of the land surface are always of monotonous descending

character, their absolute values decrease very fast as they move away from the well.
Horizontal displacements, on the contrary, vary non-monotonously, with a specific
extremum, located at a distance of the order of the longitudinal size of the well. It
is important to note that the maximum absolute values of horizontal displacements
appeared almost three times lower than the corresponding values for the vertical
displacements, however, their decay rate was sufficiently lower. This feature should

[email protected]
Fig. 6.7 Profiles of dimensionless vertical (1,3,5) and horizontal (2,4,6) displacements of the land
surface in the vertical cross-section containing the axis of horizontal wells of different length, (1,2):
l = 10 h, (3–6): l = 5 h, at different depths (1–4): z0 = 2.5 h, (5,6): z0 = 5.5 h; (1–4): sinks with
linear variation of intensity (Q = 3), (5,6): sinks with constant intensity (Q = 1)
528 6 Spatial Contact Problems for Porous Elastic Bases

Fig. 6.8 Bowl of vertical deflections of the land surface formed due to a horizontal well
functioning

be taken into account while estimating the consequences of functioning of horizon-

tal wells. As one should expect, the irregularity of sink along the well axis makes
the pattern of the land surface deformation essentially asymmetrical with respect to
the transverse axis of the well (Figs. 6.7, 6.10, 6.11, and 6.12).

[email protected]
Fig. 6.9 Typical function of
horizontal displacements of
the land surface at a
horizontal well functioning
6.4 Examples of Numerical Calculations 529

(a)

(b)

[email protected]

Fig. 6.10 Isolines of vertical displacements W/(ΔP·h/E) of the land surface for horizontal wells
of different length with a linear sink intensity variation (Q =3), (a): l = 5 h, (b): l = 10 h
530 6 Spatial Contact Problems for Porous Elastic Bases

(a)

(b)

[email protected]

Fig. 6.11 Isolines of horizontal displacements U/(ΔP·h/E) of the land surface for horizontal wells
of different length with a linear sink intensity variation (Q =3), (a): l = 5 h, (b): l = 10 h
6.4 Examples of Numerical Calculations 531

(a)

(b)

[email protected]

Fig. 6.12 Isolines of horizontal displacements V/(ΔP·h/E) of the land surface for horizontal wells
of different length with a linear sink intensity variation (Q = 3), (a): l = 5 h, (b): l = 10 h
532 6 Spatial Contact Problems for Porous Elastic Bases

The analysis of the calculations performed has shown the horizontal well depth
and length, in comparison with its feed area radius, to be the factors providing the
strongest effect on the land surface deformation.

6.4.2 Surface Deformations of the Layer

From the known dilatation relations (6.14–6.16) one can estimate the influence of
the elastic properties of the layer and the surrounding rock mass on the deformation
of their interface with the variation of the pore pressure in the productive layer. This
is important in order to provide stability of the “layer + elastic mass” system and
reliability of the horizontal well functioning.
We restrict our consideration to the vertical displacements of the contact surface
of two coupled elastic half-spaces (Fig. 6.13). Figure 6.14 shows the calculation
data (in the dimensionless form) for relative vertical displacements of two contact-
ing half-infinite elastic masses for ν 1 = 0.3 with the variation of the pore pressure
in the oil- or gas-bearing (bottom layer No. 2 in Fig. 6.13). As follows from the
calculations performed, the pore pressure decline results in a decrease of deforma-

Fig. 6.13 Concentrated

forces, acting on the
boundary of two coupled
elastic half-spaces

[email protected]
Fig. 6.14 Effect of elastic properties on the displacements of the boundary of the seam and the
containing mass
6.4 Examples of Numerical Calculations 533

tion the interface between the upper elastic layer and the oil- or gas-bearing layer in
comparison with the system, possessing the same deformation parameters E1 = E2 .
The influence of the layer compressibilities (ν 1 , v2 ) is negligible and can be revealed
only if the upper layer deformation modulus is by an order of magnitude less than
the deformation modulus of the oil- or gas-bearing layer, E1  E2 .

6.4.3 Settlements and Slopes of Rigid Foundation Plates

Consider an example of numerical solution of a contact problem. Let a square punch
with a side 2a rest on the surface of a porous elastic half-space and undergo a central
loading by a vertical force P. A coordinate system is chosen with the origin in the
punch centre, axes X and Y being directed perpendicularly to the sides of the square
and axis Z is directed into the half-space.
A source of the pore pressure decrease (a horizontal well, a drain, etc.) with a
length l is assumed to be located at a depth z = z0 (Fig. 6.2a).. While performing
simulations, we consider a cylindrical surface with a radius R2 (Fig. 6.2b, c) as the
feed area boundary for the well with a radius R1 . Pressure distribution is considered
to be axisymmetric, in each section perpendicular to the well axis (Fig. 6.2d) and
varying according to the logarithmic law Eq. (6.22). The pressure at the feed contour
Pk and at the well Pc linearly varies along the well axis according to Eqs. (6.23) and
(6.24).
Numerical calculations of the contact interaction parameters for the spatial prob-
lem under consideration were performed for a porous elastic half-space with the
(1) (2)
Poisson ratio ν = 0.3. At the feed contour of the well (R2 = a/2), Pk = Pk = 0
was assumed. At the well contour (R1 = 0.01a), a sink, linearly varying along the
well axis, was specified (Pc(1) /Pc(2) = Q = const). z0 = 2.5a; l = 5a, Q = 3, P = Ea2
was taken, E being the soil deformation modulus.
The calculations performed have shown the contact deformation characteristics
to be essentially affected by the well location depth and its length ratio to the
feed contour radius as well as by the sink intensity parameters. For the example
considered, a well located symmetrically with respect to the punch, leads to the
punch slope ψ y = 4.1671·10−3 and settlement Wc /a = 0.44675·10−2 due to the
non-uniform run-off along its axis. One should note for the sake of comparison that
the corresponding classical solution of the contact problem for an elastic half-space
gives the value of Wc /a = 0.42032·10−2 with zero slope. As one can see, besides the
slope arising, the relative error in the determination of settlement with the account
of the pore pressure decrease for the situation considered is 6.3%. The analysis of
numerical computations has also shown the influence of the pore pressure variation
on the redistribution of contact stresses p(x, y) over the punch footing is negligible.
This is explained by a specific character of the spatial contact problem, revealed in
[email protected]
an unlimited growth of contact stresses while approaching the punch boundary. If
the flexibility of the foundation structure is taken into account, the effect of the pore
pressure decrease will be revealed to a greater extent.
534 6 Spatial Contact Problems for Porous Elastic Bases

Thus, an effective method for calculating the land surface deformation under
functioning of horizontal wells in a complete spatial formulation has been devel-
oped. It takes into account the nonuniformity of the mechanical properties of
the productive layer and the surrounding mass, a non-uniform sink mode along
the well as well as its spatial orientation and location depth. The elaborated
Subsidence-01 software has a convenient module structure and enables the sought
values of displacements to be obtained with high numerical stability. Discretization
of spatial well feed areas of different complexity levels is provided, meeting the
practical needs. It seems quite promising to develop an application software pack-
age based on the elaborated method that would enable, for a given trajectory of
horizontal well systems on a vast territory of industrial extraction at complicated
engineering-and-geological conditions, the processes of the well interference and
the land surface lowering to be simulated, the optimal modes of oil and gas field
operation, applicable from both environmental and technological point of view, to be
interactively chosen. Such approach is helpful for the studies of large-scale effects
in well drainage areas, providing more realistic predictions of gas extraction char-
acteristics and application of various methods of intensification of intralayer cross
flows on vast areas. In other words, the studies of geotechnical processes at oil, gas
and water extraction enable, on the one hand, the material loss caused by these pro-
cesses to be predicted and reduced and, on the other hand, these processes to be
employed for controlling the filtration flows for industrial purposes.
Additionally, the model calculations performed have shown that based on the
method proposed, one can carry out detailed studies of contact interaction of the
soil mass with rigid foundation structures, located close to industrial mining areas
and undergoing strong land surface deformation effects caused by the pore pres-
sure decrease. Meanwhile, slopes and settlements of foundation structures with the
development of oil and gas deposits or withdrawal of underground water are pre-
dicted with a sufficient accuracy. The corresponding algorithm of numerical calcu-
lation is based on the boundary-element method and enables one to estimate the
deformation of bases of shallow and deep foundation structures under a system of
spatial loads of general type. From the obtained solution of the contact problem one
can calculate the stress-strained state of soil in the foundation active area and predict
its slopes and displacements with the account of the processes of the pore pressure
decline in the soil.
Among the important advantages of the approach proposed one should mention
the absence of any iterative algorithmic processes, numerical stability of the results
obtained that follows from the integral representation of the numerical solutions, as
well as acceptable computation times required for practical purposes.

References
[email protected]
1. Altenbach J, Sacharov A S (eds.) (1982) Die Methode der Finiten Elemente in der Festkör-
permechanik. Fachbuchverlag, Leipzig
2. Banerjee P K, Butterfield R (1981) Boundary element methods in engineering science.
McGraw-Hill, New York
References 535

3. Bartolomei A A, Omelchak I M, Yushkov B S (1994) Principles of the prediction of pile

foundation settlements. Stroyizdat, Moscow (in Russian)
4. Biot M A (1941) General theory of three-dimensional consolidation. J Appl Phys 12: 155–164
5. Bondar A L, Subbotin I E (1993) Study of land surface deformation in the area of Shebelynka
gas field. Geod Kartogr (issue 4): 22–24 (in Russian)
6. Chernykh V A (1989) Geodynamics of processes of development of oil and gas fields. In:
Problems of increase of efficiency of technology of natural gas field development. VNIIGAZ,
Moscow, pp. 12–21 (in Russian)
7. Corapcioglu M Y (1984) Land subsidence – a state of the art review. In: J. Bear and M J.
Corapcioglu (eds.) Fundamentals of transport phenomena in porous media. Noordhoff, Dor-
drecht, pp. 371–444
8. Johnson K L (1985) Contact mechanics. Cambridge University Press, Cambridge
9. Maizel V M (1951) Temperature problem of theory of elasticity. Ukr Acad Sci Publ., Kyiv (in
Russian)
10. Nikolaevsky V N, Basniyev K S, Gorbunov A T et al. (1970) Mechanics of saturated porous
media. Nedra, Moscow (in Russian)
11. Nowacki W (1970) Teoria sprezystosci. PWN, Warszawa
12. Plevako V P (1969) A point force inside a pair of cohering half-spaces. Soil Mech Found Eng
6: 165–169
13. Plotnikova L M, Flyonova M G, Machmudova V I (1990) Induced seismicity in the Gazly gas
field region. Gerlands Beitr. Geophysik. Leipzig 99: 389–399
14. Poland J (ed.) (1984) Guidebook to studies of land subsidence due to groundwater withdrawal.
UNESCO, Paris
15. Polubarinova-Kochina P Ya (1977) Theory of motion of ground waters. Nauka, Moscow (in
Russian)
16. Sakalo V I, Shkurin A A (1985) Universal program for triangulation of a two-dimensional
domains of an arbitrary shape with grid condensations. Probl Prochn (issue 1): 106–108 (in
Russian)
17. Samarskiy A A, Gulin A V (1989) Numerical methods. Nauka, Moscow (in Russian)
18. Scott R F (1979) Subsidence: a review. In: Saxena S K (ed.) Evaluation and prediction of
subsidence. ASCE, New York, pp. 1–25
19. Segall P, Yerkes R F (1990) Stress and fluid-pressure associated with oil-field operations: a
critical assessment of effects in the focal region of the earthquake. In: Geological survey.
Professional paper, pp. 259–272
20. Segerlind L J (1976) Applied finite element analysis. J Wiley & Sons, New York
21. Sobolevskiy D Yu (1994) Strength and bearing capacity of dilatating soil. Nauka i Tekhnika,
Minsk (in Russian)
22. Tsytovich N A, Ter-Martirosyan Z G (1981) Fundamentals of applied geomechanics in engi-
neering. Vysshaya Shkola, Moscow (in Russian)
23. Word Oil’s Handbook of Horizontal Drilling and Completion Technology (1991). Gulf Pub-
lishing Company, Houston
24. Zaretskii Yu K (1967) Theory of soil consolidation. Nauka, Moscow (in Russian)
25. Zienkiewicz O C (1971) The finite element method in engineering science. McGraw-Hill,
London
26. Zotov G A, Chernykh V A (1992) Geodynamic processes at development of hydrocarbon
fields. In: Geotechnical problems of development of natural gas fields. VNIIGAZ, Moscow,
pp. 24–30 (in Russian)

[email protected]
Chapter 5
Calculation of Bases for Rigid Complex-Shaped
Deepened Foundations According to the Second
Limiting State in a Three-Dimensional
Formulation

Abstract In the fifth chapter boundary-element method is applied to calculate con-

tact interaction of foundation structures with soil, taking into account the deepen-
ing factor. The need for spatially based calculation of bases of deepened founda-
tions is explained. The principles for foundation structure calculations from the base
deformations are briefly reviewed as well as the existing problem formulations and
solution methods for spatial problems of contact interaction of deepened founda-
tion structures with soil bases. Solutions of spatial contact problems for deepened
monolithic-type foundation structures most widely used in the recent years are also
considered, namely for (1) pyramidal piles; (2) foundations made of short vertical or
inclined bored piles with caps; (3) bored pile foundations with support extensions;
(4) slot foundations with the longitudinal cross-section of various shape. Heteroge-
neous stress-strained states of the base are taken into account as well as the forma-
tion of cavities between the soil and the foundation structures. The effect of the foun-
dation shape on its displacement and slope at various spatial loading is estimated
quantitatively. Numerous examples show the results of the boundary-element mod-
eling to be in good agreement with the experimental measurements performed for
spatial foundation structures, in most cases boundary-element method results being
closer to the experiment than those obtained by other known calculation methods.

Large scale of engineering and related capital expenditures have set the problem
of reduction of the engineering cost due to the decrease of consumption of materi-
als and labour as well as the increase of the level of technology. Solution of these
problems requires a rapid increase of development and implementation of efficient
engineering structures, in particular, in foundation engineering.
There is a great variety of modern foundation structures related to the specific
features of buildings and structures, methods of their construction, properties of soils
in their bases, combinations of loads acting on them, etc. [19, 26, 31, 47, 50, 59, 65,
78, 90, 91, 94, 95, 97, 98, 100, 102, 136, 138, 149, 157, 160, 171, 172, 199, 202,
[email protected]
205, 206, 211, 218, 220, 224, 215, 227, 236, 244, 247]. Most of such foundations are
traditional, the technology of their construction and methods of calculation having
been well elaborated in practical engineering.

S.M. Aleynikov, Spatial Contact Problems in Geotechnics, Foundations of 385

Engineering Mechanics, DOI 10.1007/b11479_5,  C Springer-Verlag Berlin Heidelberg 2011
386 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

However, along with the advantages, many of the known foundation types, from
the point of view of modern science and engineering, do not fully provide the
required efficiency of foundation structures which should combine reliability, low
labour consumption, speed of construction, reduced consumption of materials, etc.
In order to reduce the labour consumption for the construction of foundation, new
types of foundations are developed, more economical in comparison with the foun-
dations of traditional type [224].
New solutions are related mostly to the improvement of the shape of the deep-
ened parts of foundations and, consequently, to the development of special tech-
nologies of their construction. Among the new types of foundations elaborated in
the recent years one should note some efficient modern structures with a complex-
shaped deepened part, which have not yet been described in special literature.
At present, the most industrial foundation type are pile foundations, hence for
them numerous attempts have been made to increase the pile bearing capacity by
variation of their shape. Examples of successful application of some types of piles
(triangular, pyramidal, trestle, with widenings, etc.) are quoted in the literature.
An effective trend to reduce the cost of foundations under buildings with strutted
structures is the application of compaction piles of wegde-shaped and pyramidal
type [79]. An improvement of the design of these piles consists in a cradle for a
half-frame toe along the pile diagonal. The studies performed show the specific
bearing capacity of such piles to be increased by 20–25% due to the increase of the
soil back pressure area and application of a vertical load with high eccentricity.
A pyramidal displacement pile with a T-shaped cross-section in the medium part
and a solid head was developed and found applications in Belarus [99]. The studies
have shown that pyramidal displacement piles with a T-shaped cross-section, hav-
ing the volume smaller than solid pyramidal piles (by a factor of about 1.9), are
not worse than the latter in the soil base bearing capacity and meet all regulation
standards in rigidity, crack-proofness, and strength.
Piles with self-unfolding blades with increased bearing capacity, were developed
in Perm Technical University [11, 22, 23]. A pile consists of a shaft, in the lower
part of which a widening in the shape of the self-unfolding blades is mounted. Due
to afterdriving or under an applied load the pile settles together with the soil above
the area of the blades, i.e. works as a small pile group. The experiments show that
in the base of such piles a complex stress-strained state of the soil of the active area
is formed, affecting the further functioning of the loaded pile.
A hollow conical pile is an example of a relatively new efficient foundation struc-
ture. The results of extensive experimental studies [14, 15, 144] have shown the
hollow conical piles to possess high engineering and economical parameters. The
efficiency of their application increases with their length.
A collaboration of technical universities of Perm and Mariy El [12] has resulted
in the development and implementation of piles with a longitudinal sector-shaped
cutout. Due to a broader involvement of soil into functioning, pile foundations
[email protected]
with a longitudinal cutout provide high specific bearing capacity and low material
consumption.
5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations 387

Among non-traditional efficient solutions of pile foundations for structures being

erected on slopes, one should mention piles of reinforced concrete with a cross-
section in a shape of an equilateral triangle [48]. Piles with triangular cross-
section possess a developed lateral surface (with respect to the required direction)
and, in comparison with traditional prismatic piles, higher bending rigidity. Piles
with triangular cross-section are also efficient for construction of structures on a
flat terrain when they are subject to horizontal loads, constantly acting in a fixed
direction.
For construction of low-rise buildings a design and technology for strong foun-
dations in the shape of reverse arches, deepened into soil, is proposed [191]. The
internal contour of the arch elements is parabolic and the external one is a circu-
lar arc of a larger radius. The functioning of the arch-shaped foundations cannot
be reduced to the conditions of a flat problem due to their length and width being
approximately the same.
For a number of objects of rural engineering, being constructed in rock, sand, and
clay soils in a stabilized state, foundations with an inclined bottom are used [160].
Construction of the foundations with an inclined bottom is most appropriate in the
case of a considerable horizontal load to be transferred to the base, i.e. when the
above-foundation structure is a strutted system of arc or frame type. A shortcom-
ing of the foundations of such type is a low accuracy of their calculation method,
involving numerous tables and nomographs.
Pile foundations and piles with a non-traditional shape of the below-grade part
can be an efficient way to reduce vibrations for foundations under machines [199].
The studies of behaviour of piles with a variable shaft geometry (trapezoidal, pyra-
midal, with the shaft widening at the tip (alpha piles) and in the upper part) in a
broad range of variation of the vibration direction and frequency, as well as dynami-
cal load value at vertical and horizontal loading are reported in [81–83, 200, 237]. In
particular, application of piles with one or several widenings of the shaft is shown to
be appropriate. For the foundations under a horizontal load, application of piles with
widenings in the upper part of the shaft is recommended. The main conclusion of the
investigations performed is that construction of foundations with a complex shape
of the below-grade part enables multiple reduction of the below-grade part volume
with a corresponding saving of materials, the rigidity properties being preserved or
even improved.
An example of short foundation structures of non-traditional shape can be anchor
structures, applied for high metallic funnels, power transmission line supports,
pipeline supports at irrigated territories. Among the anchor structures a special atten-
tion should be paid to the one with a shaft of a variable cross-section, proposed
in [80] to reduce labour consumption and increase its reliability. In the longitudi-
nal direction the structure has a wedge-like shape with anchor lugs on the inclined
sides, their widened side facing the soil surface like a herringbone. Evidently, in
order to provide efficient and reliable functioning of such anchor structures, espe-
[email protected]
cially under a pulling load, a calculation method for the base deformations in an
essentially spatial formulation should be developed. Since in this case the contact
388 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

between the anchor and the soil will be violated, the method should undoubtedly
take into account unilateral constraints.
Spatial foundations of structural type (with inclined panels) for ever-frozen soils
are less investigated, but promising in view of the efficiency of their functioning and
reduction of material consumption [145], as well as cross-shaped piles with diag-
onal reinforcement for bearing-wall residential buildings [32], pyramidal-prismatic
foundations of short piles under strutted structures [113], prismatic, T-shaped, and
pyramidal piles with a cantilever under strutted loads for construction sites on bases
with an upper layer of humified or fill soil [170].
The quoted references to the literature regarding foundation structures with a
complex-shaped deepened part do not exhaust the known sources and are just typical
examples which can be further extended.
A considerable progress has been achieved in theory of foundation structure cal-
culation what follows from a vast literature, including manuals, instructions, and
handbooks. Nevertheless, many issues have still not been developed enough for
practical solutions. For example, calculations of foundation structures, providing
their reliable functioning, should, on one hand, be as reliable as possible, and, on
the other hand, avoid excessive safety factors, implying unjustifiable expenditures
of concrete and reinforcement. An optimal solution for these contradictory prob-
lems can be found only as a result of high-precision calculations for the foundation
+ base system.
For some foundation structure types there are either no recommendations for
calculations, or they are not described sufficiently in special literature. Therefore,
designers often encounter a number of difficulties and often use rather approximate
and insufficiently substantiated calculation methods. As a result of such a situation,
an additional safety factor is introduced into the approximately calculated founda-
tion structures, this leading to the increase of material consumption. On the other
hand, due to the low accuracy of the calculations, the foundation structures appear
not reliable enough what results in a decrease of their lifetime or in emergency
events.
Requirements for compliance with the Construction Rules and Regulations
2.02.01–83 [207], instructions for designers, and handbooks lead to cumbersome
and labour-consuming calculations (especially for the foundations of a complex
geometrical shape). In many cases the calculations are performed by iteration, with
numerous tables and handbook data being invoked. As a consequence, in order to
simplify the calculations, the dimensions of the foundation structures are taken a pri-
ori larger than the optimal required values. For the foundations with given dimen-
sions, load values are reduced by using reliability coefficients that are not always
substantiated.
For many foundation types, no special calculation methods, taking into account
their structural features, friction resistance over the lateral surface and various com-
binations of acting forces and moments, have been created at all. For the calcula-
[email protected]
tion of new types of foundation structures a specially corrected scheme is created,
based on assumptions and pre-conditions resulting, as a rule, in the application of
an already known calculation method, in some cases being quite approximate. Often
5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations 389

the calculations are performed using an analogy with the known simplified methods,
the simplest elastic soil model, based on the Winkler hypothesis, is used. Calcula-
tions according to the Construction Rules and Regulations 2.02.01–83 do not take
into account the interaction of the foundation lateral surface with the soil and, there-
fore, the character of the unilateral constraint functioning in the areas of the structure
uplifting from the base is not reflected.
Thus, due to the specific structural features of foundations, the insufficiency of
the existing approaches to the foundation base calculation is evident. Broader and
more universal approaches to the calculation of foundation structures are required,
including non-traditional foundations, demanding more precise calculation meth-
ods. It should be also noted that the issue of the correct choice of the modified
calculation scheme is of extreme importance as well in view of the spatial type of
functioning of most of the foundation structures due to a combined action of verti-
cal, horizontal, and momental loads.
Among the modern numerical methods that have found applications for calcu-
lation of structures in various fields of engineering, one should first of all mention
the boundary-element method. Due to its sufficient universality, indifference to the
shape of the structures, convenient data input, and possibility of consideration of
infinite domains, this method is undoubtedly effective to calculate bases for founda-
tions with a complex-shaped deepened part.
In the present chapter, the boundary-element method is used to analyze the
results of mathematical modelling of the joint functioning of soil bases with
complex-shaped rigid foundations under spatial loading of general type. The cal-
culations are performed according to the method of solving a spatial contact prob-
lem of a rigid deepened punch, modelling the foundation. Numerical experiments
are performed in two variants – with and without the account of unilateral con-
straints on the contact of the foundation with the soil. The calculation method
is implemented in the Rostwerk software that does not require large computer
resources.
Examples in Sect. 5.3–5.6 demonstrate the most typical calculations of base
deformations for deepened foundation structures of monolithic type, having become
most widely spread in the recent years:

(1) pyramidal piles,

(2) foundations of short vertical or inclined bored piles with pile rafts,
(3) bored pile foundations with support widenings,
(4) slotted foundations with different longitudinal cross-sections.

These sections are preceded by Sect. 5.1 where general information is given
regarding the principles of calculation of foundation structures from the base defor-
[email protected]
mation, and Sect. 5.2 with a brief review of the existing formulations and methods
of solving spatial problems of contact interaction of deepened foundation structures
with soil bases.
390 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

5.1 General Information on the Calculation of Bases for

Foundation Structures from the Deformations
Design of foundation structures required an optimal solution to be found, providing
fail-safe functioning of the structure. This is achieved by performing the calculations
according to the principle of the limiting states [67, 101, 115, 215, 224]: forces,
acting on the base, and stresses, displacements, and deformations, arising in the
latter, should be restricted by corresponding limiting values.
Calculations according to the limiting states are divided into two groups [67, 115,
215, 224].
The first group includes carrying capacity-based calculations when the restric-
tions are introduced, enabling one to avoid excessive irreversible plastic deforma-
tion, deformations of creep, damage, resonance vibrations, etc. In case the limiting
conditions of this group being violated, the base becomes completely unsuitable for
the foundation structure functioning.
The second group is formed by deformation-based calculations, enabling such
values of settlements, upheavals, horizontal displacements, deflections, slopes,
vibration amplitudes, etc. to be set, for which no problems arise with normal func-
tioning of the structures and their long-term reliability. Violation of the second-
group calculation conditions, depending on the excess of the joint deformations of
bases and structures over the limiting values, can result either in a state of the struc-
ture, unsuitable for normal functioning, or in its complete functional unsuitability.
Construction Rules and Regulations 2.02.01–83, taking into account a variety of
features of interaction of foundation structures with bases, ordain the necessity of
calculations for the bearing capacity in four most dangerous cases and for deforma-
tions – in all cases without exception [207]. Thus, for industrial and civil engineer-
ing, calculation according to the second group of limiting states (base deformations)
is decisive.
Base deformations are caused by various reasons and can be of the following
type [224]:

settlements – deformations due to the soil compression (without changes of

their structure) under an external load as well as the influence of neighbour-
ing structures, the intrinsic weight of the soil,
collapsing – deformations due to the soil compression (with a change of their
structure) under an external load and the intrinsic weight of the soil as well as
due to such processes as soaking of collapsing soils, thawing of frost soils,
etc.,
upheavalor shrinkage of the base surface – deformations, related to the change
of volume of some soils due to physical or chemical processes (frost heave,
swelling, chemical ground stabilization, etc.),
subsidence – deformations of the land surface due to mining, soil excavation at
[email protected]
underground construction work, intense water intake, etc.,
horizontal displacements – common deformations of bases and structures at
high values of horizontal components of the total load (at the erection of sheet
5.1 General Information on the Calculation of Bases 391

pile constructions, retaining walls, foundations of spacer structures, etc.) as

well as at large vertical displacements of the ground surface at settlements
and collapsing,
non-uniform deformations of the base – due to the soil nonhomogeneity within
the construction site and misadjustment of loads on separate foundations.

The following basic types of the common deformations of a structure and its base
are considered.
An important quantitative characteristics of settlement or horizontal displace-
ment of a separate foundation is its absolute displacement W, determined as vertical
and horizontal displacements of a selected point of the foundation (on its bottom,
edge, on the base surface level, etc.) or as their average displacement, respectively
[70, 115, 215, 224].
From the W values, known for different foundations, the non-uniformity of the
base deformations is evaluated. For example, the average settlementof the base W
can be estimated from the equation

'
n
Wi A i
i=1
W=
'
n
Ai
i=1
where Wi is the absolute settlement of the i-th foundation, Ai is its base area. Know-
ing the values Wi (i=1,. . .n) along with W, one can choose the design solutions to
reduce the settlement non-uniformity.
In order to evaluate additional forces, arising in structures under a non-uniform
deformation of bases, a relative settlement non-uniformity for two foundations is
introduced into consideration:

|W2 − W1 |
Ŵ =
L
where L is the distance between the foundations.
The foundation slope ψ in the simplest case can be defined as a ratio of the
difference of settlements of the extreme points of its bottom to the distance between
them

Wα − Wβ
ψ= .
L
The foundation slope value enables one to estimate additional forces in the foun-
dations as well as in the above-foundation structures.
In case forces acting on the foundation ends in different directions, a torsionof
[email protected]
the structure arises. In order to characterize the spatial functioning, a relative torsion
angle θ is introduced into consideration, its value enabling the additional forces in
bearing structures and ceilings to be evaluated.
392 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

Non-uniform settlements of foundation structures arise additional stress in the

latter, what can result in the formation of cracks and even to the structure breakdown,
while slopes and deflections can violate the normal functioning of the equipment
installed. Therefore, the design of foundation structures implies a condition of all
kinds of deformation not to exceed the limiting values set by the project (the second
group of limiting states) [115, 215]:

W ≤ WLIM (5.1)

where W is the generalized displacement of the foundation (the foundation edge,

the pile head, etc.), determined by calculation according to one of the recommended
methods, WLIM is the admissible generalized displacement of the foundation, given
as a reference value, depending on the building strength and deformability as well
as on the requirement to the building functioning and installation of equipment. The
WLIM value can be modified according to the design assignment or to the combined
calculation of the base + foundation + building system.
In some cases, according to the tables of the Construction Rules and Regulations,
deformations due to the soil shrinkage and swelling as well as to the suffusion-
related base settlement, are determined.
The values of W and WLIM can mean all the above mentioned deformation types:
(1) absolute vertical settlement of the base of a separate foundation, (2) horizontal
displacement of the foundation, (3) average settlement of the base of the structure,
(4) relative non-uniformity of settlements of two foundations, etc.
Thus, the deformation-based calculation of the foundation bases is performed
with the account of the common functioning of the building and the base. The limit-
ing values of the common deformation of the base and the structure in the course of
the building functioning (soil compression, swelling, collapsing, suffusion-related
settlement) are set from the technological or architectural requirements as well as
from the conditions of strength, stability and crack-proofness of the structure. Only
in the case the structures being not intended to bear the forces arising at the com-
mon functioning with the base, and the WLIM values being not set in the design
assignment, the limiting values of the base deformations (relative settlement differ-
ence, slopes, average or maximal settlements, etc.) for different structures are set in
accordance with the tables of the Construction Rules and Regulations [207].
The condition of Eq. (5.1) is reduced to a condition

P ≤ PLIM (5.2)

where P is the load on the foundation; PLIM is the load corresponding to the admissi-
ble displacement. The conditions of Eqs. (5.1) and (5.2) are written for the simplest
cases of loading in a selected direction. At complex spatial loading, in the general
[email protected]
case vertical, longitudinal, and transverse displacements arise; consequently, the dis-
placement and load become vectors and the W and P values mean the absolute values
of the corresponding displacement and load vectors, respectively
5.1 General Information on the Calculation of Bases 393
 
W= Wx2 + Wy2 + Wz2 , P = P2x + P2y + P2z .

Besides, for some cases of spatial loading a restriction of the type of Eq. (5.2) should
be also introduced for momental loads.
Consider some examples of typical conditions of calculations of foundations
from the base deformations.
Instructions [67] give recommendations regarding calculation and design of
foundations for machines with operation-related dynamical loads due to the action
of unbalanced forces and moments, motion of masses, impacts of moving or
falling parts of the machines, including foundations for machines with rotating
parts, machines with crank mechanisms, forging hammers, precast reinforced con-
crete forming machines, pile drivers, crushers, rolling equipment, metal-cutting
machines, and rotating furnaces.
According to [67], the foundation vibration amplitudes should satisfy a condition

A ≤ Aadm (5.3)

where A is maximal amplitude of the foundation vibrations determined from the

calculations, Aadm is the maximal admissible amplitude of the foundation vibrations.
The Aadm value can be modified according to the design assignment.
If due to the machine operation vertical AV , horizontal longitudinal AHL and
transverse AHT vibrations arise, then the condition of Eq. (5.3) is recommended
to be used in the form

A2V + A2HL + A2HT ≤ Aadm .

The vibration amplitudes AV , AHL , and AHT should be determined from a sys-
tem of differential equations of vertical and horizontal-rotational vibrations of the
foundation [67].
Calculation of pile foundations according to the second-group limiting condi-
tions (based on deformations) under vertical loading is carried out, based on a con-
dition [224]

S ≤ SU

where S is the deformation of the pile foundation base (settlement and relative set-
tlement difference), determined by the calculations, SU is the limiting admissible
value of the pile foundation base deformation, determined from the Construction
Rules and Regulations 2.02.01–83 or set by the design assignment.
[email protected]
The pile foundation settlement is determined as for a nominal foundation, trans-
ferring to the soil a uniformly distributed pressure in the horizontal plane passing
through the lower ends of the piles. Due to the functioning of the lateral surface of
394 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

the piles, the nominal foundation bottom size is taken bigger than the pile group
lateral size.
Foundations of piles, working as end-bearing piles, do not require calculations
based on the deformations due to the vertical load.
Calculation for pile foundation displacements due to horizontal loads and
moments consists in the fulfillment of conditions

U p ≤ U u ; ψ p ≤ ψu

where Up and ψp are the calculated values of the pile head horizontal displacement
and its rotation angle, respectively, Uu and ψu are their limiting values, set in the
design assignment.
Prediction of additional settlements of bases under reconstructed buildings is rec-
ommended [225] to be performed by a special calculation in two stages: (1) calcu-
lation of the initial (before the reconstruction) stress-strained state of the base, (2)
determination of the additional settlement of the foundation after the reconstruction.
The first stage enables one to estimate the base settlement value over the period
before the reconstruction and to build up a scheme of local non-uniformity in the
base of the existing foundation with pointing out the soil compression (decompres-
sion) zones, determining their size and configuration. This enables the variants and
schemes of the foundation reinforcement to be assigned in a more substantiated way.
Further, the deformation-based calculation of the base under reconstruction can be
carried out from the condition

S ≤ SU , J ≤ JU

where S is the additional settlement of the foundation as per the calculation, due to
the load increase under the reconstruction, SU is the limiting additional settlement,
J is the calculated skew of two neighbouring foundations after the reconstruction,
JU is the limiting skew value.
The limiting value of the additional settlement of a building after the reconstruc-
tion SU can be within the first approximation determined from the condition

SU = KC · Smax ,U

where Smax,U is the maximal limiting settlement, recommended by Appendix D of

the Construction Rules and Regulations 2.02.01–83 for a new construction, KC is
the coefficient of reduction of the limiting settlement value due to the structure
ageing (at the wear of 20% KC =0.4, at the wear above 30% KC =0.3). In order
to determine the additional settlements, various numerical methods are used as
well as the methods of layer summation, equivalent layer, limited compressible
thickness, etc.
[email protected]
Base deformations due to the soil compression are determined by calculation
from analytical and approximate solutions, obtained by simplifying assumptions
[70, 101, 115, 215, 224] (the method of a linearly deformable half-space or a
5.1 General Information on the Calculation of Bases 395

limited-thickness soil layer; the layer summation method, the equivalent layer
method), as well as by approximate numerical methods of calculation (the finite-
difference method, the finite-element method, the boundary-element method), using
the apparatus of mechanics of condensed media, applied mathematics and compu-
tational engineering [66, 116].
In order to calculate foundation bases according to the second-group of limiting
states we apply the BEM, being the most efficient among the known numerical
methods for solving engineering problems of essentially spatial type for linearly
deformable soil masses. Appraisal and improvement of the developed method will
further enable this method to be spread to the cases of elastoplastic deformation of
soil bases, processes of consolidation, creep, etc. when numerous specific features
of the soil media are taken into account and the linear solutions remain the basic
ones.
As noted above, most of the practical methods of calculation of settlements of
structures are based on using a linear dependence between the settlement and the
load at any intensity of the latter. In fact, the development of domains of plastic
deformation with the load increase results in a nonlinearity of the settlement-vs-
load plot. Therefore, in case the foundation structures being designed using linear
calculation methods, the following condition is introduced:

σ ≤R

where σ is the average stress over the foundation bottom and R is the calculated soil
resistance, admitting the development of areas of plastic deformations to the depth
of a quarter of the foundation width and being determined from the Construction
Rules and Regulations [207]. In other words, at σ>R the Construction Rules and
Regulations do not guarantee the correct value of the settlement of the structure,
determined from the linear calculations. It should be also noted that the value of the
linearity limit of the quarter of the foundation width is rather arbitrary and not sub-
stantiated for different types of soils. For example, the calculations [70] have shown
that at high internal friction angles of soils ϕ > 20◦ the linearity of the settlement-
vs-load plot is preserved far beyond R. Practically, for such soils the solutions of
theory of elasticity can be used for settlement determination within a small error up
to the loads of 2∼=3 R and even more.
To a considerable extent the introduction of the value R is caused by insufficient
development and practical implementation of solutions of nonlinear problems of
soil mechanics [70, 224]. With the development of techniques for mathematical
modelling of elastoplastic (nonlinear) deformation of soils, more possibilities are
open to determine settlements at any stage of loading, up to the limiting value.
The only possibility to estimate the reliability of the existing methods for the
calculation of foundation bases is the comparison of the base settlement values,
determined from the calculation (Wc ) and from the field observations (Wf ). Sotnikov
[email protected]
[216] generalized and analyzed long-term field observation data regarding the set-
tlements of a great number of buildings and structures in comparison with the cal-
culation data for the same values. In order to perform such a comparison, over
396 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

200 research papers have been analyzed with the observation data regarding the
settlements of several hundreds buildings and structures in different countries of
the world after 1938. Average settlement values were compared as well as charac-
teristics of the settlement non-uniformity (maximal difference, deflections, skews,
slopes). Objects under investigation were residential and industrial buildings of vari-
ous type (brick, bearing-wall, skeleton-type), round-type structures (funnels, towers,
tanks), rectangular-shape structures, extended structures (concrete dams, sluices),
etc. A both-side spread of the calculated values was revealed, the errors reaching
hundreds per cent both sides. Based on the results of observations and calculations
from many publications, it was concluded that coincidence of the calculation results
with the measurement data can be treated as a mere accident. In most cases calcu-
lations resulted in considerable errors. It is also noted that for loose soils (oozes,
varved clays, peaty sand-clays, etc.) the settlement calculation results in lowered
data (in 72% cases). For dense soils (bedrock of various composition – clays, marls,
limestones, till of semihard and hard consistency, sand, etc.) Wc in most cases (78%)
appeared higher than Wf (the calculations result in overstated data). In the first case
the understatement of Wc is dangerous, since the error does not increase the safety
factor. However, the increase of Wc by factor of 1.5–2, proposed by some authors,
found no reflection in the Construction Rules and Regulations. The main result of an
extensive and detailed study of Sotnikov [216] is the conclusion: reliability of deter-
mination of the calculated values of Wc at present stage of development of founda-
tion construction and soil mechanics is absolutely insufficient for efficient design.
Section 5.2 contains a review of the existing approaches to the calculation of
foundation bases (with the account of the deepening factor), their specific features
being noted and a short comparative characteristics being given. It also contains
information on the practical possibilities of the calculation methods proposed in the
literature. Restrictions for their application for the design of foundation structures
of an essentially spatial type are also given.

5.2 Spatial Problems for Calculation of Foundation Bases with

the Account of the Depth Factor

An extensive literature is devoted to the problem of foundation base calculation

based on their deformation. Studies [51, 71, 99, 243, 244] give a review of formation
of modern calculation schemes, taking into account the spatial character of the base
and foundation functioning under a load. The variety of the calculation methods is
related to the existence of a great number of models, being various modifications
of the Winkler model, the elastic half-space model as well as numerous nonlinear
(elastoplastic) models.
Winkler model. Due to its simplicity and convenience, the most widely spread
in practice ate methods based on the hypothesis of a coefficient of subgrade reac-
[email protected]
tion (Winkler model). The simplest solutions in this case are obtained when the
foundation calculation scheme is taken as an absolutely rigid body [51, 87, 88, 98,
217, 244]. Note that most of the shallow foundations correspond to such calculation
5.2 Calculation of Foundation Bases with the Account of the Depth Factor 397

scheme. In this case, depending on the longitudinal-to-transverse dimension ratio,

the foundations are considered as three-dimensional bodies or as rods. In the latter
case the load from the rod to the base is supposed to be transferred via fictitious
areas whose dimensions are chosen in a specific way. The proposed methods differ
in a number of components of the soil reactive forces being taken into account. In
[51, 87, 98] two coefficients of subgrade reaction (in vertical and horizontal direc-
tions, respectively), determining the soil back pressure normally to the foundation
surfaces, are introduced into the calculation. For a more adequate modelling of con-
tact interaction over the foundation bottom, Zavriyev and Shpiro [244] introduce an
additional coefficient characterizing the foundation shift in the horizontal direction
(exactly equal to half of the coefficient of subgrade reaction in the vertical direc-
tion). Kiselev [88] and Ten [217] in their solutions fully take into account the base
reaction, using all components of the soil reactive back pressure. For a better fit
of the proposed calculation models to the real functioning of a foundation in the
soil, the coefficients of subgrade reaction are supposed to be linearly [51, 87, 88,
98] or arbitrarily [244] variable with depth. The sought foundation settlements, dis-
placements, and slopes are found by invoking equilibrium equations, displacement
method, and the principle of superposition.
A scheme of a foundation in the form of an elastic bar has been widely spread
for the calculation of long piles. In this case, when the soil base is described using
the Winkler model, the calculation methods [84, 111, 209, 242] are well elabo-
rated, experimentally substantiated, and form a basis for the instructions on pile
foundation design [68]. A theoretical basis for the calculation methods used in the
quoted papers, is a fourth-order differential equation of simple bending of beams
[203]. By invoking a system of equations of the initial parameter method [226] solu-
tions have been obtained at various boundary conditions at the pile ends [111, 242].
For the coefficients of subgrade reaction triangular [242] or trapezoidal [111, 209]
profiles were used. Reference [84] gives numerical solutions for a triangular and
a trapezoidal coefficient of subgrade reaction profiles, practically coinciding with
the results of [209], however, being obtained on the base of a specially developed
numerical approach using the possible displacement method and integral equilib-
rium equations.
Elastic half-space model. Method of calculation of deepened foundations based
on base deformations using the elastic half-space model have not been widely devel-
oped due to essential mathematical difficulties. This concerns calculation schemes
of foundations not only as three-dimensional bodies, but also as an elastic or a
rigid rod.
The problem of a horizontally loaded absolutely rigid rod, interacting with an
elastic half-space, was first considered by Zhemochkin [245]. The essence of the
proposed approach is reduced to a substitution of the contact between the rod and
the elastic isotropic linearly deformable half-space by a point contact using arbitrary
rod constraints. The resolving system of canonical equations is obtained based on
[email protected]
the mixed method of engineering mechanics after cutting the above constraints and
introduction of a rigid fixation on the rod axis. Friction forces on the contact of
the fixation and the soil are not taken into account. The canonical equation system
398 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

coefficients are sought using the fundamental solution for a unit horizontal force
acting near the half-space boundary. Later, in [117] the fundamental Zhemochkin
solution was generalized for a soil medium with a variable deformation modulus.
A further development of the Zhemochkin method for the calculation scheme
of a foundation as a nondeformable rod was performed by Levenstam [108]. He
considered a granular medium of Kandaurov type [73] as the soil base model and
elaborated a corresponding fundamental solution for a unit horizontal force. One
should note that the results, obtained for the granular medium and for the elastic
half-space, are qualitatively similar.
The simplest implementation of the Zhemochkin method is proposed in [19, 100]
where a simplified approach is developed by means of restriction of the number
of constraints by two pairs only. It also takes into account the anisotropy of the
soil base properties in the horizontal and vertical directions. For this purpose, the
soil deformation modulus, determined using a vertical punch, is multiplied by a
reduction coefficient 0.8. Such assumption enables the higher deformability of the
base in the horizontal direction to be taken into account in the simplest way. Note
that solutions, using the Zhemochkin method for a linearly deformable half-space
[19, 117, 245], later were used as a basis for the recommendations for calculation
of pile foundations for vertical and horizontal loads [93].
A pile deformation in the approximation of simple bending is considered by
Simvulidi [203], using the nonhomogeneous half-space model with a smoothly vari-
able deformation modulus. After the pile length having been partitioned, simplifying
assumptions are made. First, the friction forces at the contact between the pile and
the soil base are neglected. Second, no mutual influence of the pile partitions is sup-
posed. Then the soil deformation modulus within each of the partitions is averaged
and the function, approximating the base reaction, is taken as a third-order poly-
nomial of the reduced coordinate for each partition. After the force method being
applied, the proposed approach enables a closed system of linear equations to be
formed for the unknown coefficients of each polynomial.
For the design and evaluation of functioning of deepened foundations like bored
piles, anchor foundations, etc., a solution of the problem of a deepened punch set-
tlement in a spatial formulation is required. For example, bored and pyramidal piles
transfer a considerable part of load to the soil by means of a footing whose base is a
rigid deepened punch. Besides, the foundations of various structures, as a rule, are
located not on the surface, but inside the soil. Rather approximate solutions for the
settlement of rigid foundations incorporated into a linearly deformable half-space,
have been obtained based on the Mindlin’s solution. They are determined as aver-
age vertical displacements of a uniformly loaded circle [198] or rectangle [106],
treated as nominal rigid foundations. In order to obtain more exact settlement val-
ues, one should apply solutions of spatial contact problems of deepened punches.
Direct solving of such problems is known to be rather difficult and not always justi-
fied for engineering calculation purposes.
[email protected]
The labour-consuming solution of contact problems can be avoided if, due
to some speculations, one can predict the contact pressure distribution over the
foundation bottom. In [109, 110] approximate values for the rigid deepened punch
5.2 Calculation of Foundation Bases with the Account of the Depth Factor 399

settlements are obtained using the semi-inverse method, when for setting the con-
tact pressures a known exact solution of the problem of a circular rigid punch on a
half-space surface is invoked. In order to take into account the displacement pile tip
functioning, in [204] the Boussinesq solution for the settlement of a round punch
on a half-space surface is used, with an additional coefficient 0.5 being introduced.
This assumption is rather rough, it can be used only for the displacement piles, for
which the cross-section radius is small in comparison with their length.
Settlements of rigid foundations with small relative depth values (h/a=0.5, 1.0)
were determined in [197] using the solution for a rigid punch on a half-space. The
contact pressure distribution under the deepened punch is given similarly to the
punch on a half-space, with a 6th order polynomial being added to take into account
the foundation deepening. The contact pressure profile under a deepened rigid punch
appeared to be practically the same as for a punch on the surface of an elastic half-
space. This enabled the author of [163] to find the settlement of a round rigid deep-
ened punch as well as vertical compressive stress under, based on the Mindlin’s solu-
tion with the contact pressure distribution over the bottom according to the Boussi-
nesq law. A good agreement of the obtained solution with the results of [198] (the
difference does not exceed 4%) has shown the assumptions, performed in [163] for
the determination of a circular rigid deepened punch settlement, to be correct. For
practical convenience, [163] also gives tabulated values of the dimensionless punch
settlements for different Poisson ratio values ν and relative depths h/a.
The elastic base settlements are shown to be noticeably reduced with depth [106,
198], what, in turn, affects the displacements of rigid foundation structures and the
internal force distribution in flexible structures. Therefore, for rational design of
foundation beams and plates, as exact as possible solutions are required, based on
the analysis of spatial contact problems for deepened rigid punches.
Ogranovich [133, 134] has solved numerically spatial problems of determination
of the settlement of centrally loaded round and rectangular foundations located at
a given depth form the surface of a linearly deformable elastic isotropic half-space.
In both cases the solution is reduced to the canonical equations of the mixed
method of engineering mechanics. The author uses the improved Zhemochkin
method of piecewise uniform elements [246], being, in fact, the simplest form of
the boundary-element method. The canonical equation system matrix is determined
using the Mindlin vertical displacement function. For the rigid punch the contact
domain is meshed into separate elements with a shape of a circle (the central ele-
ment) and rings with their thickness decreasing near the contour. For the rectangular
punch the contact domain is meshed into a grid or rectangular elements, condensing
towards the angular points. Within each element the soil back pressure was assumed
constant. The optimal number and size of the elements are determined by test
calculations for punches located on the half-space surface, when there is an exact
solution for a circle (according to Boussinesq [219]) or an approximate solution
for a rectangle [54]. The calculations have convincingly shown that the account
[email protected]
of the foundation deepening results in an essential decrease of the settlement in
comparison with the same foundation resting on the half-space surface (for the
circular foundation at the relative depth h/b=5 and ν=0.35 the settlement decrease
400 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

was 66%). Simultaneously an important for practical purposes conclusion is made

on a quite small error, introduced by non-accounting of the pit, on the bottom of
which the foundation structure is located. Note that the same author considered in
the spatial formulation the problem of a beam, deepened into an elastic half-space,
loaded by a concentrated force or a uniformly distributed load [132]. The deepening
effect was also revealed in an essential decrease of bending moments both under
the concentrated force (of the order of 30% at relative depth h/b=10 and flexibility
index t=10) and under a uniformly distributed load (of the order of 100%).
Solutions of contact problems of plate-type objects deepened into an elastic
medium are of considerable interest for the estimation of functioning loads on
anchor-type structures. The calculation scheme in the form of a low anchor plate,
fully bonded with the soil, presents a rather adequate model for the case of a flat
anchor domain, formed in a saturated soil mass. In this case the contact is provided
by the cohesive forces and the intrinsic weight of the soil weakly affects the axial
rigidity at low depth.
The solution of the problem of a rigid anchor disk, fixed into an isotropic elastic
infinite medium, was first given in [24]. This solution was developed by Selvadurai
[176–178] for a transversely isotropic elastic medium with the account of the plate
flexibility as well as under an eccentric load. The problem of the anchor plate can
be considered as a limiting case of a spherical anchor subject to an axial load. Such
solutions were developed in [74, 175] based on the application of the singularity
method and using the expansions over spherical harmonic functions. In [251] these
studies were extended to the case of transversely isotropic media. Non-classical
effects related to partial uplifting of the anchor plate from the soil, were considered
by Keer [77] and Selvadurai [183]. Similar phenomena, related to anchor plates,
deepened into crumbling domains, localized in an elastic space, were studied in
[181, 188]. The generalized rigidity-related properties of anchor plates, deepened
into a transversely isotropic medium or resting on a surface of separation of two
contacting half-spaces, were studied in [179, 184]. Due to essential mathematical
simplifications, many solutions for anchor plates in an infinite elastic space can
be obtained in an analytical form. However, one should keep in mind that the
assumption of the elastic medium infinity is justified only for the cases when the
anchor plate is deepened into the elastic base by a sufficiently large distance from
the surface.
In the case when the anchor plate is located near the elastic-half-space boundary,
the load-free plane makes an essential influence on the rigidity-related character-
istics of the deepened anchor plate. The mathematical formulation of the contact
problem for a rigid disc-shaped anchor, deepened into an elastic half-space, was
first considered in [62]. Besides the main attention in this paper being paid to the
consideration of the effect of the compressed elastic medium, the authors also con-
sider an uplifting effect on the contact surface of the plate and the soil. Besides, for
shallow soil anchors the solution of the contact problem of a disc, deepened into
[email protected]
an elastic half-space, was used for in situ tests on loading conventional and helical
plates [185, 186], and Rower and Booker [165] numerically studied contact prob-
lems for anchors, horizontally deepened into an elastic half-space.
5.2 Calculation of Foundation Bases with the Account of the Depth Factor 401

Contact problems for anchors, deepened into a homogeneous and a nonhomo-

geneous half-space, were studied by means of variation schemes and discretization
procedures in [152, 155]. An axisymmetric problem for a rigid disc, deepened into
a half-space, when the conditions for common displacements on the boundary sur-
face being fulfilled only partially, is considered in [135]. Some specific aspects for
a disc anchor in an elastic half-space were considered in the review papers [121,
122]. Later Selvadurai [189] solved an interesting from the practical point of view
problem of the influence of the surface load on a disc anchor, deepened into an elas-
tic half-space. An elastostatic problem for an absolutely rigid anchor disc (a round
plate of neglectably small thickness), deepened into a half-space, welded with the
elastic medium and subject to an axial loading, is considered in [182]. The prob-
lem was solved using the Hankel transformation and reduced to a system of the
second-order Fredholm equations with subsequent numerical implementation. The
calculation results were presented by the plots of the plate settlements versus its
distance from the half-space surface and, in fact, represent the anchor axial rigidity
depending on its depth.
Less known are solutions for the calculation scheme of a foundation as a spatial
block, when for the soil base an elastic half-space model is applied and, conse-
quently, Mindlin equations are to be used for the displacements under concentrated
forces in the directions of the coordinate axes. The possibility of the Mindlin’s solu-
tions to be used was first substantiated by Gersevanov already in 1948, with the
limits of applicability being indicated [49]. However, due to considerable mathemat-
ical difficulties at the integration of the fundamental Mindlin’s solutions, no general
approaches to the calculation of deepened foundations of spatial type according to
the second group of limiting states have been developed yet. In scarce studies some
Mindlin’s solutions were used for geotechnical calculations (under the action of a
vertical or a horizontal force). Lapshin, using the Mindlin’s solutions, considered an
issue of determination of additional horizontal and vertical stresses on the bound-
ary of a bored pile under loading due to the resistance of a clay soil over the lateral
surface above the considered cross-section [102]. Separate Mindlin’s solutions were
applied by Vasilenko, Petrenko, Orobchenko, Tsymbal for the calculation of piles
in sands [139, 140].
In relatively recent papers by Ogranovich [129, 130] the Mindlin’s solutions
were successfully used for the calculation of rigid piles of variable cross-section
(pyramidal and conical) subject to horizontal, vertical, and momental loads.
Besides, the same author, for the calculation of rigid pyramidal piles under
horizontal and momental loads, took into account formation of a gap between
the foundation and the soil [131]. The engineering method of calculation of pile
foundations with the account of the depth factor, developed in detail by Ogranovich,
is discussed in Sect. 5.3.1, specially devoted to the calculation of rigid pyramidal
piles under spatial loading.
A combined deformation of cylindrical piles and the surrounding homogeneous
[email protected]
mass under a central load was considered in [125, 127]. The soil was considered as a
linearly deformable half-space with the average deformation modulus Eo and lateral
expansion coefficient ν0 . The pile deformability was taken into account according
402 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

to the Hooke’s law as compression of a straight short beam with a deformation mod-
ulus E. The contact is supposed to be a full cohesion of the pile and the soil. The
mutual influence of the pile and the soil is approximated by a system of fictitious
forces, acting along the pile axis and related by boundary conditions at the contact
surface. The pile length is divided into n equal partitions. The authors assume the
fictitious forces to act in the centre of each partition and at the tip level. Tangentail
stress within the lateral surface of each partition is assumed constant. It is calculated
using the Mindlin’s solution for a horizontal unit force applied inside an elastic half-
space. The problem is solved by the finite-difference method, based on the relation-
ship between the variation of the fictitious forces and the tangential stresses at cho-
sen neighbouring partitions of the pile. The proposed approach enables the pile set-
tlements to be determined with the account of its geometry (length and cross-section
diameter) without any preliminary assumptions on the tangential stress distribution
character. The numerical investigations performed have shown that at a considerable
relative depth the pile material compressibility essentially affects the stress distri-
bution character and the pile settlements. In particular, at the diameter d=0.3 m,
length L=6 m, n=10, Eo =30 MPa, E=15·104 MPa, and axial load N=200 kN the
pile material compression was 0.6 mm, or 22% of the pile head settlement (2.7 mm).
Later the method, having been proposed in [125, 127], was applied for the problem
of studying stress and displacements in the soil in case a vertical static load being
transferred by a conical pile [126]. In order to simplify the calculation scheme, the
conical pile was proposed to be considered as a telescopic one with the account of
the load transfer by ledge rings and cylindrical lateral surface of each telescopic part
of the pile.
A method, applicable for rigid off-centre loaded pier foundations [104], appeared
to be rather general and simultaneously experimentally substantiated. The method
is based on the application of partial solutions for a separate partition of the contact
surface, obtained on the base of the Mindlin’s solutions for an isotropic and homo-
geneous elastic half-space. In order to find the general solution, the finite-difference
method is applied. The calculation results appeared to be in a good agreement with
the experimental data.
The elastic half-space model was used for the calculation of short cylindrical
bored piles subject to a spatial load system of a general type [63, 64]. The pile body
was assumed to be non-deformable, no slippage of the soil particle with respect to
the pile surface was assumed. The calculation was based on the potential method
[228]. The contact interaction model took into account unilateral constraints, not
working in tension, and under compression having an unlimited resistance to
friction (structural nonlinearity). The Mindlin’s solution is taken as the fundamental
one. The analysis of numerous calculations has shown the profiles of normal and
tangential contact stress to be in a good qualitative agreement with the known
experimental data. The only exceptions are the boundary parts of the profiles where
[email protected]
concentration of stress is observed, typical for the contact deformation of ideally
elastic media. No such concentration was observed experimentally. A satisfactory
description of kinematics of the pile interaction with the surrounding soil base.
Comparison of the calculations with the experimental plots and the calculation data
5.2 Calculation of Foundation Bases with the Account of the Depth Factor 403

based on the Construction Rules and Regulations 2.02.03–85 (using the Winkler
model) for horizontal displacements, settlements, and slopes of the pile under a
horizontal load has shown that a much better agreement with the experimental
data is observed for the elastic half-space model calculations than for those based
on the Construction Rules and Regulations 2.02.03–85. An important conclusion
is made that the development of deformation-based calculation methods for short
bored pile bases under a spatial load system with the application of the soil model
in the form of an elastic half-space is advisable and promising. This model enables
the soil mass behaviour under a load to be described in a more adequate way
than the hypothesis of the coefficient of subgrade reaction. Unfortunately, the
approach of [63, 64], based on the numerical-and-analytical potential method, has
appeared not to be sufficiently universal. Its application for solving the problems
of deepened punches (foundation models) with an arbitrary configuration of the
contacting surfaces could not be realized. This is related to the method of analytical
calculation of the influence matrix coefficients being developed only for the case of
flat boundary elements, parallel to the coordinate axes, approximating the contact
surface between the foundation and the soil. Therefore, the performed calculations
were restricted by the consideration of cylindrical bored piles (with a special
boundary-element grid) as well as prismatic and benched block foundations.
It is known that deepened piled foundations can be subject to loading by a torque
due to different off-centre lateral forces. In particular, foundations undergo due to
storm winds and earthquakes. In the case of foundations under machines, unbal-
anced masses at the operation of engines also result in torque loads. Another typical
example are bored foundations of power transmission line supporting structures,
subject to considerable torsional loads due to non-uniform stress in the electric
cables and due to wind or snow. Consequently, the evaluation of torsional rigidity of
deepened foundation structures is important for the static and dynamical analysis of
the foundation+structure system. However, this important geotechnical problem has
not found its proper development yet. In this connection one should mention [28,
60, 76, 112, 148, 153, 154, 156, 164, 180, 187], devoted to the analysis of contact
interaction of deepened structures under torsion.
The stress-strained state of a circular cylinder and a half-space welded to it,
subject to a torque, applied to the free end of the cylinder, is investigated in [60].
The cylinder and the half-space are elastic isotropic materials with different defor-
mational characteristics. The torque is considered to rotate the cylinder’s free end as
a rigid body, i.e. radial displacement in the cylinder varies in each section linearly
along the radius. Besides, the lateral surface of the cylinder and the half-space sur-
face beyond the cylinder are assumed to be free from the external load. The Fourier
method is used for finding the solution for the cylinder, and in the case of the half-
space the integral Hankel transformation is applied. Explicit formulae for all the
values, characterizing the stress-strained state, are obtained. If the half-space is an
absolutely rigid body, the obtained formulae give the known solution of the problem
[email protected]
of strength of materials regarding a round rod torsion. In another limiting case, cor-
responding to the geotechnical problem formulation, when the cylinder is an abso-
lutely rigid body, results in the Reisner and Sagoci formulae [164] to determine the
404 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

contact stresses and displacements in a half-space subject to torsion by a rigid cylin-

drical punch. Finally, the problem solution is reduced to the solution of an infinite
algebraic equation system, having a symmetrical matrix with a determinant without
singularities. Numerical calculations, performed in [60] for a 16th order system,
enabled the stress-strained state characteristics in the cylinder and in the half-space
to be obtained, areas of influence of fixation (with respect to the cylinder depth) to
be determined for various ratios of the elastic properties of the cylinder and the half-
space. Subsequently the solutions of the problem under consideration were used
to describe the torsional behaviour of rigid piled foundations, using the function of
torsion rigidity in a homogeneous half-space [148, 180] as well as in a layered base
when a layer and a half-space beneath it are assumed to be homogeneous [76, 112,
153, 187]. Torsion of rigid foundations, deepened into a nonhomogeneous elastic
soil with a residuum, is studied in [154]. The residuum shear modulus is assumed
to decrease linearly with depth and the shear modulus of the underlying nonhomo-
geneous soil – to increase with depth. Problems for rigid foundations of flat circular
and cylindrical shape are considered. The boundary integral equation method, based
on the Green’s functions for displacements and forces in a nonhomogeneous elastic
medium, is applied. Numerical solutions with respect to the torsion rigidity function
are presented for various depths, shear modulus gradients, and residuum thickness.
A detailed analysis of the obtained results has shown the torsional rigidities to vary
linearly at the variation of the shear modulus gradient and the residuum thickness
and parabolically at the variation of the cylindrical foundation length.
Based on the approaches, considered in [60, 76, 112, 148, 153, 154, 164, 180,
187] to the setting of the stress field in the near-pile space, different approximate
solutions were obtained in a closed form [28, 156]. Randolph [156] presented an
approximate analytical solution, based on a simplification of the stress field in the
soil, and obtained torsional rigidities of an elastic pile, deepened into a homoge-
neous soil as well as into a linearly nonhomogeneous soil with zero shear modulus
on the daylight surface. The main result of this study was a strict consideration of
the influence of the underlying soil nonhomogeneity on the rotation of the deep-
ened rigid foundation. Based on the simplest assumptions on the stress field in the
soil, a closed form of the solution for a torsion rigidity of an absolutely rigid or
flexible single cylindrical pile in a homogeneous isotropic soil was obtained [28].
Using the same finite-difference method, a numerical solution was found for the
rotation angle of the pile in the soil whose shear modulus varies arbitrarily with
depth. The calculations were performed in the assumption of soil to be deformed
as a linear elastic medium. The calculation results were compared with the Poulos
experimental data [148]. A good correlation of the calculated and the measured data
was observed. In the authors’ opinion, in order to describe a more realistic picture
of the pile behaviour under torsion (up to its breakdown), a more adequate concept
should be introduced into the numerical algorithm, based on the dependence of the
shear stress versus the rotation angle along the pile length, similar to the known
[email protected]
hypothesis of the coefficient of subgrade reaction for horizontally loaded piles.
Nonlinear soil base models. To calculate the foundation settlements for a lin-
ear settlement-vs-load plot, an elastic solution is applied, taking into account only
5.2 Calculation of Foundation Bases with the Account of the Depth Factor 405

the deformational characteristics of the soil (E – deformation modulus, ν – Poisson

ratio). At the pressure over the foundation bottom, exceeding the calculated pres-
sure, the settlement-vs-load plot becomes nonlinear, and the account of strength-
related and elastoplastic properties of the soil is required to calculate the settlements
(recommendations of the Construction Rules and Regulations 2.02.01–83, Sect. 24).
Fedorovskii [43, 107] has proposed a method for the calculation of nonlinear
settlements (at the soil deformation in the shear phase) of deep supports of bored
pile type. The calculation method according to the second limiting state is based on
the application of theory of elasticity to the interaction of a cylindrical rod (pile)
with a soil base. The soil is considered as an elastic half-space with an incorpo-
rated cylindrical non-deformable pile. The soil daylight surface is free from loads.
The bored pile is loaded by a vertical force, it can be with or without an broad-
ened footing. In the shear phase, i.e. at loads, exceeding the proportionality limit,
the soil slippage occurs at the pile lateral surface, due to its insufficient cohesion
with the pile, or to an extremely stressed state arising in a thin layer, surround-
ing the pile. Since at the pile sliding practically all the soil mass remains in the
elastic state, the calculation of the pile settlement can be performed using the meth-
ods of theory of elasticity. Gorbunov-Posadov was the first to pay attention to this
fact [55].
In the approach, proposed by Fedorovskiy, the stress-strained state of the soil
mass is presented as a superposition of Mindlin’s solutions for a concentrated force
in a half-space. The half-space with a cutout, corresponding to the pile volume,
is substituted by a full half-space, and a fictitious load in the form of vertical and
double horizontal forces is distributed along the pile axis. Such approach is based
on the studies performed by Poulos [150] who has shown the difference between
the solutions obtained based on the fictitious load distribution along the pile axis
and over its lateral and end surfaces to be quite small and noticeable only in small
areas near the pile ends. In order to increase the calculation accuracy, Fedorovskiy
applied concentrated forces at the pile head and tip: vertical forces as well as dou-
ble vertical and horizontal forces. A principal difference from the known boundary
integral equation method (BIEM) is in the fact that fictitious forces are applied not
on the contact boundary of the pile and the soil, but at a certain distance from it
(along the pile axis), what is quite justified only for long piles. The integral equa-
tions are not written in an explicit form, but substituted by conditions at sepa-
rate points (collocation points), lying on frontal and rear generatrices of the pile
in the centre of the constant load areas. Due to such simplifications, all the cal-
culations, preceding the solution of the resolving equation system, are performed
analytically, and no numerical integration is required, as in the boundary-element
method.
The fictitious load distribution along the pile axis in the fedorovskiy method is
assumed piecewise linear. The sought unknowns are the fictitious force values at
the points on the pile’s axis as well as concentrated load values at the pile ends.
[email protected]
The influence matrix (displacements and stresses due to unit fictitious loads) of the
resolving equation system is obtained, using the Mindlin formulae and integrals of
them. At the initial stage of calculation, the conditions of full cohesion between the
406 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

soil and the pile both over the lateral surface and over the tip, i.e. the conditions of
equal displacements of the pile and the soil at their contact, are set as the contact
conditions. Then, with the load increase, in the upper part of the pile the soil
slippage areas arise, gradually widening and involving the whole lateral surface and
the tip. The plot of the pile settlement versus load s=s(p) becomes curved and the
settlements increase in comparison with a purely elastic calculation. In order to take
this increase into account, the method implies a search for the boundary of separa-
tion of the areas of slippage and full cohesion. In the slippage area, the condition of
equality of tangential (with respect to the pile surface) displacements is substituted
by the slippage condition, which is taken in the form of the Coulomb’s friction law

τ = σn · tan ϕ + C

where ϕ and C are the parameters of the contact friction between the soil and the
pile, τ and σn are the friction and normal pressure on the pile at the contact points,
calculated with the account of the initial stressed state and additional stresses due to
the pile loading. In the case of the presence of a broadened footing at the pile end,
which is assumed spherical, the calculation tales into account the experimentally
observed [9] uplifting of the soil from the upper part of the widening. At the
numerical implementation of the proposed approach, the author recommends the
dependence s=s(p) to be found by a semiinverse method, i.e. the load on the pile is
determined from the given settlement.
The method for calculation of deep pile settlements, proposed by Fedorovskii,
was subsequently implemented as a computer code. Additional features are intro-
duced into the calculation algorithm, taking into account the pile shaft compress-
ibility, what is essential for the piles with a high ratio of the length to the diame-
ter. Besides, in order to make a transition from homogeneous to multilayer bases,
the Mindlin formulae for the calculation of the stress-strained state of the medium
under concentrated forces are substituted by the corresponding formulae for a lay-
ered medium according to the method, described in [44]. Based on the elaborated
software and on the results of the performed numerical studies, recommendations
for the design of bored piles with and without a broadened footing, as well as shell
piles in sand and clay soils were elaborated [161], helpful (along with static tests)
for the optimal choice of piles at the assignment of load on the piles of a given
size. It should be noted that, though, in the authors’ opinion, the method elabo-
rated produces quite acceptable results. it is not applicable for short piles and piles
with considerable widenings. Besides, the loading is assumed to be strictly axial,
and small widenings of spherical shape should be located under the pile footing.
Finally, the development of zones of extremely stressed state, resulting in additional
pile settlement increase, is not taken into account.
Using a method, identical to the one, proposed by Fedorovskii for the problem
of calculation of a vertically loaded pile with the account of contact slippage [43,
[email protected]
107], later the problems of functioning of a single pile or a pile group in an elastic
homogeneous half-space under a horizontal load were solved [46]. Comparison of
the calculation results with the known solutions and data of model experiments [146,
5.2 Calculation of Foundation Bases with the Account of the Depth Factor 407

147] have shown the account of the pile uplifting from the soil in the course of
loading considerably approaches the calculated mutual influence of the piles in the
group by means of the soil to the experimental values.
The results of theoretical studies of nonlinear (elastoplastic) displacements of a
pile, subject to an axial load, are given in [86]. The pile absolute rigidity is assumed,
as well as the equality of displacements of the pile shaft and the contacting soil layer.
The mathematical model of the soil in the framework of theory of plastic flow with
strengthening is applied [85]. The author characterizes the stressed state of the base
by the Lode parameter, equal to zero. Different strength hypotheses can be applied
(Mises–Schleicher–Botkin, More–Coulomb, Hill–Tresk). It is noted by the author
that practical application of his formula for a pile settlement has essential difficul-
ties, related to the experimental determination of mechanical characteristics of the
soil: in order to obtain the deformation-related and strength-related parameters, a
special non-standard equipment should be applied or even constructed, what leads
to complicated and expensive studies. This made the author of the proposed calcu-
lation method to apply approximate equations of state as well as to neglect elastic
deformations. The results of the calculations performed and the field studies give
a satisfactory qualitative agreement. Quantitatively, a satisfactory agreement of the
calculation and the experiment is observed up to the load value, not exceeding about
60% of its carrying capacity. Note also that the proposed method of the deformation-
based pile calculation contains an assumption of zero influence of the pile tip, i.e.
is restricted to the consideration of a pile of infinite length. Therefore, recommen-
dations, setting the limits of applicability of the proposed method regarding the
pile relative depth, are required. This limit can be set only based on comparison of
numerous test data and model calculations. Hence, the proposed method of estima-
tion of nonlinear settlements of piles is far from being complete and its practical
application is encumbered.
One of the main tasks of soil mechanics and foundation construction is devel-
opment of such a method of calculation of bases and foundations, which would
enable the deformation processes to be described simultaneously in the whole
range of loads, including the limiting values, with the account of both deformation-
related and strength-related characteristics of soils (C – cohesion, ϕ – internal
friction angle) [123], and in case the behaviour of a structure with time being
considered, viscosity-related (fluidity-related) properties should be also taken into
account [36].
The class of problems under consideration is characterized by a strong nonlin-
earity. Their complex solution is possible only in case efficient numerical methods
being used on modern computers. However, in spite of overall intense application of
powerful computers for solving nonlinear problems in the recent years, this complex
problem of soil mechanics and foundation construction is still far from its admissi-
ble solution. Noticeable success has been achieved, using the finite-element method
for deepened foundations in the form of rotation bodies (piles, anchors, round foun-
[email protected]
dations) under axisymmetric loading [36]. For such foundations their geometry and
loading conditions are independent of the azimuthal coordinate. Therefore, they are
considered in a two-dimensional space of cylindrical coordinates (z-coordinate and
408 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

polar radius), what is to a great extent similar to the case of well studied flat prob-
lems [6, 35, 42, 56, 174, 223].
In a series of papers by Fadeev and collaborators [7, 36, 37, 41], a method of
boundary-element solution of elastoplastic and elastoviscoplastic problems of deter-
mination of the stress-strained state of the base + foundation system is developed
with the account of rheological properties of soils at axisymmetric deformation.
This method is implemented in the Geomechanics Creep-3 computer code, using
the initial stress procedure [37]. The program enables one to model axisymmetric
problems, arising in the practice of industrial, civil, hydrotechnical, and transport
engineering, and provides a series of elastoplastic solutions for a given sequence of
loading by gravitational forces, stepwise application of construction loads, stage-
wise excavation of ditches or below-grade workings by introduction of structural
elements and given displacements in the nodes at any stage. A bilinear elastoplas-
tic model of soil with the Coulomb fluidity criterion serves as a rheological model
in the proposed calculation scheme. It is characterized by parameters, determined
at conventional engineering-and-geological studies: deformation modulus, Poisson
ratio, cohesion, internal friction angle, density. In the numerical finite-element pro-
cedure linear ring-shaped elements of triangular cross-section are used. They are
most convenient for the solution of elastoplastic problems. Their rotation results in
a triangular torus. The domain of calculation can contain up to 20 element types
with different set of properties. Obtaining the elastoplastic solution is based on a
multiple repetition of elastic solutions with modified values of the node forces, the
system rigidity matrix remaining constant (i.e. the rigidity matrix inversion being
performed only once). The natural stressed state is modelled by an all-round hydro-
static stress tensor which is summed up with the actual principal stress value. At
first the problem is solved for the given node force values and boundary condi-
tions. Then, in order to find the elastoplastic solution in each of the elements, a
sequential procedure is performed, stated in detail in [6, 35, 36, 41, 42, 56, 174,
223]. For the stress-strained state analysis with the account of the soil creep the
Shvedov–Bingham model is applied [229]. In this case, for each finite element the
characteristics of long-term strength and viscosity coefficients are added, and then
the elastoviscoplastic solution is found similarly to the elastoplastic one, using a
temporal stepwise procedure.
Using the Geomechanics Creep-3 software, calculations for some typical
schemes of foundation design were performed and analyzed.
A central loading of a low-deepened (H/d=0.125) rigid punch was considered in
[41]. Plastic areas in the base were found and their development with the external
load increase was traced. The obtained settlement-vs-load curves enabled the levels
of the average pressure over the foundation bottom, for which the settlements on an
elastic and on an elastoplastic base practically coincide, to be separated from those,
for which the presence of plastic deformations results in an enhanced increase of
settlements in comparison with the purely elastic solution.
[email protected]
The analysis of settlements and carrying capacity of a single conical pile is given
in [7]. A full cohesion is assumed both over the lateral surface of the foundation
and over the bottom of the foundation and the base. The pile rigidity is given by the
5.2 Calculation of Foundation Bases with the Account of the Depth Factor 409

description of its geometrical shape and the concrete characteristics. An axisymmet-

ric elastoplastic solution has shown a satisfactory coincidence (the discrepancy did
not exceed 15%) with the experimental settlement-vs-load plots, built based on the
data of static tests of equivalent 4.4–70–20 pyramidal piles, equal in volume and in
length to the conical one, assumed for the calculation.
The analysis of calculations of the stress-strained state in the active area of a
foundation, performed in [7, 36, 37, 41], has shown quite convincingly that the pro-
posed finite-element method for the calculation of settlements and carrying capac-
ity of axisymmetric foundations enables a combined calculation to be performed
according to the first and the second limiting conditions and settlements at the lim-
iting loads to be predicted. It is quite important that the calculation enables the non-
homogeneity of the base soil layers to be taken into account, a combined calculation
of the base + pile foundation system to be performed with the account of formation
of the pile compression area at the pile driving and loading, plastic deformation
development areas to be detected with the account of the load on the foundation.
Aleksandrovich and Fedorovskii [2] also used the finite-element method for solv-
ing the problem of interaction of a rigid deepened foundation and a base represented
by a model of an elastoplastic strengthening medium (of Cam-Clay type [45, 173])
for the case of axial symmetry under vertical central loading. The base model, cho-
sen for the investigation is based on the associated plastic flow law. The plasticity
potential includes a strengthening parameter, depending on the stress-strained state.
Finally, the problem becomes nonlinear and is solved by the incremental method
with iterations at each loading step. Over the foundation – base contact special
contact elements were located, simulating the Coulomb friction and the foundation
detachment from the soil. The elaborated approach enabled foundations of various
depth in the range from surface punches to piles to be calculated in the same way.
A series of calculations performed with input data, differing only in the foundation
depth, has enabled the character of the foundation interaction with the base to be
traced with deepening. The nonlinear character of the settlement-vs-load curve is
revealed in a more pronounced way with depth, ending with the foundation fall-in.
The authors explain this deepening effect by an increase of the lateral surface con-
tribution into the total area of the contact between the base and the foundation and
by the increasing role of the lateral surface in this interaction.
Boyko and co-workers [1] applied finite-element modelling for automated cal-
culations of interaction of deepened foundation structures with the soil medium at
elastoplastic deformation of the base. This method was implemented into a Rosinka
software [168], based on the Mises–Schleicher–Botkin criterion of limiting equilib-
rium and theory of plastic flow in the form of the non-associated law. The usage of
the latter is due to the fact that the associated flow law, widely used for calculations,
is not confirmed for non-rock soils, as follows from most experiments. The main
advantage of this approach is the account of formation of limiting state areas in the
soil, what enables the stress-strained state of the soil mass to be traced of the founda-
[email protected]
tion + base system operation: from the beginning of loading to the loss of the bear-
ing capacity. As an example of application of the proposed method for calculation
of foundations, the problems of interaction of deepened punches with various tip
410 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

configurations with the soil medium were solved. Numerical solutions, confirmed
by field experiments, have shown a slight (below 5%) discrepancy in the character-
istics of the stress-strained state for all punch shapes. The authors explain this fact
by the decisive role of formation of an elastic kernel under the punch footing.
A method for calculation of bored piles in the case of nonlinear behaviour of
the soil base is proposed in [238, 239, 241]. The mathematical model, used for
the description of the mechanical properties of the soil [240], is formulated in the
framework of theory of plastic flow. The parameters of three strengthening func-
tions, being contained in the corresponding loading functions, are specified from
the data of shear and compression soil tests. The model is implemented by means of
a finite-element solution of axisymmetric boundary problems. The results of inter-
action of the bored piles with the soil base are considered. The calculation domain is
approximated by second-order triangular elements with a condensation at the con-
tact of the pile with the soil. The pile was modelled by a set of "rigid" elements with
the elastic moduli, by 3–4 orders of magnitude higher than the soil moduli. Full
cohesion was assumed as a boundary condition at the contact between the founda-
tion and the soil. Thus, a possible pile slippage with respect to the soil can occur
only with a "capture" of part of the soil. As reasonably noted in [239], the latter
circumstance is quite realistic for bored piles and is determined by the technology
of their production. At the calculations for displacement piles and other types of
foundations, boundary conditions should be introduced with the account of slip-
page. Based on the calculation, the pile settlement is plotted vs the vertical load.
With the pile length increase a linear part of the calculated settlement-vs-load curve
becomes more pronounced. The areas of the limiting state of the soil are formed
along the pile lateral surface, starting from its lower end, and develop with the load
increase. The proposed method enables one to trace the transformation of the tan-
gential stress profile as well as horizontal pressures along the pile contact with the
soil, which have appeared to be linear practically in the whole load range. The bored
pile settlements are essentially different for different assumptions of the initial load-
ing surface in the mathematical soil model even for the same soil conditions. The
most important advantages of the proposed method are the possibilities of the bored
piles calculation for the base collapsing, as well as for two limiting states (both the
admissible deformations and the bearing capacity) in the framework of one theory.
The axisymmetric calculation scheme was also employed by Shapiro for solving
the soil elastoplastic problem in his studies, devoted to the modelling of a pile load-
ing by an axial impressive force [193, 196]. The calculation procedure is based on
the initial stress method in a combination with the finite-element method. The theo-
retical basis of the method is the soil description as a solid isotropic medium, mod-
elled according to theory of plastic flow. The axisymmetric version of the numerical
calculation method is employed, which had been approved earlier at the design of
structures with a calculation scheme corresponding to the flat deformation condi-
tions [192, 194, 195]. Continual ring-shaped finite elements of a triangular cross-
[email protected]
section are used. The calculation simulates the pile loading by a stepwise increasing
force. The natural stressed state is assumed to be hydrostatically distributed. The
following hypotheses are applied for the formulation of the elastoplastic problem:
5.2 Calculation of Foundation Bases with the Account of the Depth Factor 411

(1) the nonlinearities, being taken into account, include plastic deformation of the
shape change type for the complex stressed state, unobstructed deformation
under tension, shear along the pile lateral surface,
(2) for the complex stressed state the total deformation includes an elastic (linear)
a plastic component, the latter arising after the stressed state having reached the
strength limit in according with the Mises–Schleicher–Botkin condition [30],
(3) the vectors of principal plastic deformations and principal stresses are assumed
coaxial,
(4) at the plastic deformation stage, the non-associated flow law is applied with a
dilatancy rate being introduced.

Based on the calculations performed for cylindrical piles, the boundaries of the
prelimit and limiting stressed state areas of the soil are determined and a settlement-
vs-load plot is built. The results of the mathematical modelling are compared with
the static test data for bored piles and pile foundations in boreholes. The calculation
data on the settlement-vs-load plots are noted to be close to those obtained from
the static tests of the piles in the boreholes. This is, to a considerable extent, due
to the mechanical characteristics of the soil, determined from the all-round studies
of the foundation active areas being taken as the input data. The results of calcula-
tions for bored piles of 1.7-m diameter for bridges show a possibility of increase (in
comparison with the Construction Rules and Regulations 2.02.03–85) of the pre-
dicted bearing capacity of both single foundations and components of pile rafts. For
both types of pile foundations many settlement-vs-load dependences appeared prac-
tically linear and the number of iterations, required for the achievement of a given
accuracy level, is quite significant (30–50). In the author’s opinion, in spite of well
checked and extensively used assumptions, the proposed method of pile foundation
calculations still requires further studies and modifications.
Among rare studies, devoted to essentially spatial interaction of foundations
with nonlinearly deformed bases, one should first of all note a series of studies by
Fadeyev and Matveyenko [38–40] regarding a method of solving the spatial contact
problem by means of a semianalytical finite-element method, first proposed by Wil-
son [232] and developed by Zienkiewicz et al. [233, 248]. The approach, elaborated
in [38–40], is applicable for axisymmetric rigid foundations, loaded by vertical and
horizontal forces, and can be implemented in both elastic and elastoplastic formula-
tions. The base stress-strained state characteristics (node loads and displacements)
are determined as a product of two functions, one of which being found by the
finite-element method, and the other one being given in the form of a trigonomet-
ric series. In case the trigonometric function orthogonality and the axial symmetry
of the domain under consideration being taken into account, the three-dimensional
problem is separated into a series of independent two-dimensional problems. The
axisymmetric domain of the soil base calculation is discretized by a set of ring-
shaped triangular finite element, interconnected by node circles. A procedure of
[email protected]
obtaining an elastoplastic solution by the initial stress method for a medium with
internal friction is proposed in [6, 35]. The soil is considered as an ideal elastoplas-
tic medium with a fluidity surface, described by the Botkin criterion in octahedral
412 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

stresses. The elastoplastic solution is reached by multiple repetition of elastic solu-

tions with variable node force values and a variable rigidity matrix of the system.
The numerical algorithm of spatial problem solution for axisymmetric foundations
is implemented in a Geomechanics ADA code. This software was tested for the elas-
tic stage of the base functioning by a comparison with the known theoretical solution
by Egorov [33] for the problem of an off-centre loading of a round foundation plate.
A rather good agreement was obtained for the solution of an axisymmetric problem
of a rigid shallow foundation in an elastoplastic base [41]. The studies of efficiency
of the semianalytical finite-element method have shown that, in comparison with
the full three-dimensional analysis, the developed method enables the calculation
time for the stress-strained state of bases under axisymmetric foundation structures
subject to a horizontal load to be reduced by an order of magnitude.
Today the elaborated approach using the semianalytical finite-element method
should be considered one of the most efficient for the finite-element prediction of
spatial stress-strained states of axisymmetric foundation bases. However, no other
studies have been performed for deepened foundations, except for solving one prob-
lem of pile behaviour under a horizontal load [39].
Among other studies regarding nonlinear calculation of bases for foundations of
spatial type with the account of the depth factor, one should note the most perfect, in
our opinion, finite-element approaches [143, 231] as well as the calculation of pile
foundation settlements using transformational functions [61].
A professional software PLAXIS, worked out in the Delft University (the
Netherlands) [143], is at present one of the most successful software tools, intended
for numerical studies of interaction of foundation structures with the soil. The
PLAXIS software employs the finite-element method, using two type of elements
fifteen (their amount can be up to 200) and six (up to 800) node triangular elements.
It is designed for solving a class of axisymmetric problems at elastoplastic defor-
mation. On the surface of interaction of the foundation and the soil, special contact
elements are introduced, for with the strength-related parameter values (internal fric-
tion angle and cohesion) are used, different from the corresponding values for the
soil. In such a way a thin near-surface layer of the soil is modelled, in which the
shear deformation processes are the most intense. The PLAXIS software allows sev-
eral soil models to be employed in order to describe the mechanical behaviour of
the soil:

– a linearly deformable medium (a zeroth order model – for strong bedrock bases),
– the Mohr-Coulomb model (a first-order model – for the most of engineering-and-
geological conditions),
– the Cam-Clay model (a second-order model – for soft soils).

Note that the second-order model is practically very seldom used because of
its high sensitivity to the soil nonlinearity parameters: overconsolidation and trans-
[email protected]
verse stress coefficients, which are determined in the geotechnical practice with
great difficulties. PLAXIS has no new conceptual achievements, in comparison with
the Russian software. Only due to quite convenient service and graphical functions
5.2 Calculation of Foundation Bases with the Account of the Depth Factor 413

this software is widely apply the professional version of PLAXIS, whose first release
was produced in 1988. Numerous examples of practical application of PLAXIS for
the analysis of geotechnical problems are quoted in [143]. The PLAXIS-based results
of numerical modelling of interaction of a group of four C4–30 prismatic piles with
the surrounding soil are presented in [128]. The problem is solved in an axisymmet-
ric formulation. The ideal elastoplastic Mohr-Coulomb model is taken as the soil
base model. The numerical modelling results show two important features of the
PLAXIS software. First, an additional study should be performed regarding the cor-
rect assignment of the dimensions of the pile foundation active area. Second, when
the characteristics of the soil, compressed due to the pile submergence, were intro-
duced into the calculation, doubtful results were obtained. In the author’s opinion
[128], after the load being applied, the plastic deformation areas begin to develop
very rapidly and a considerable soil squeeze out towards the half-space boundary
occurs. As a result, the obtained pattern appears very far from being realistic.
Essentially three-dimensional finite-element calculations were performed [231]
for a rigid deepened prismatic foundation under an eccentric inclined force due
to the organization of comparative test calculations in 1990 in order to study the
possibilities of modelling of an intentionally planned experiment at the Institute
of Soil Mechanics and Rock Mechanics at the University of Karlsruhe (Germany).
The results, presented in [231], appeared to be the closest to those obtained from
field measurements among the 17 calculation predictions from different countries
(Great Britain, Germany, France, Slovakia, Italy). A MONOT model of elastoplas-
tic type with double strengthening [120] was taken as the model to describe the
rheological properties of the soil medium (sand). The three-dimensional version of
the model was calibrated using the results of standard punch tests of the base. Three-
dimensional calculations (with the account of a symmetry plane in the calculation
scheme), in spite of being labour-consuming, resulted in exaggerated settlements,
and no satisfactory prediction of the base rigidity was obtained. The finite-element
calculation results are shown to be essentially determined by the friction at the con-
tact of the foundation and the soil.
In the Technical University of Gdańsk, since the 1970s the studies have been
performed to find the analytical dependences of nonlinear settlements of vertical
cylindrical piles on vertical loads in order to describe the whole settlement-vs-load
plot based on the geotechnical parameters, corresponding to the properties of the soil
base before the pile installation. A method of so-called transformational functions
[61], describing the dependences between the lateral resistance and the settlement
(t–z curves) as well as between the pressure under the pile and its settlement (q–z
curves) is developed. Using the approximate theoretical solutions and the experi-
mental data, an extensive analysis of the transformational functions is performed. In
order to identify the empirical parameters being contained in the transformational
functions, a colossal amount of the experimental data was required to be processed
(260 piles with the account of several variants of choice of transformational func-
[email protected]
tions for each of them). The calculations were carried out, using an intentionally
elaborated PALOS software, automatically performing iterative process of approx-
imation from the experimental pile settlements and the given nonlinear elasticity
414 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

law. In each separate case a good agreement of the calculated values with the field
test data was obtained. However, the method is hardly applicable for practical pur-
poses in view of the requirement of a group of empirical parameters, obtained from
a great number of penetration tests, to be given for each transformational function.
The method is mostly oriented at cylindrical piles of a large diameter under vertical
central loading. Another restriction for its application is related to the fact it being
essentially based on Polish engineering regulations.
It follows from the presented analysis that a certain success has been achieved
in the methods of foundation calculations with the account of the depth factor. A
sufficient amount of calculation schemes has been proposed, reflecting a variety of
properties of soils, loading type, relative depth, and other factors. However, each of
the methods has its own specific features, and all of them are different in accuracy
and convenience of application.
The simplest and most convenient for engineering calculations are methods,
using the Winkler base model and its modifications. Meanwhile, they do not enable
the desired accuracy of the project solutions, require the superposition principle to
be fulfilled, what is known not to be confirmed experimentally in most cases. Tan-
gential contact stresses, comprising up to 50% of total reactive resistance of the soil,
depending on the shape of the foundation deepened part and on the loading type, are
either not taken into account by these methods, or taken into account by assignment
of the corresponding coefficients of subgrade reaction without proper substantiation
of their numerical values.
Calculations, using the elastic half-space model, are more preferable. They take
into account the distributive and deformational properties of the soil base in a rather
adequate way. However, in view of essential mathematical difficulties, there are no
closed solutions for deepened foundation structures at all, and obtaining numerical
solutions is quite labour-consuming even for the simplest-shape foundations. For
practical calculations, solutions for centrally loaded circular and rectangular areas
are used, simulating the functioning of separate parts of the foundations (s a rule,
only its footing). None of the desired solutions of contact problems for foundations
with a given shape of the deepened part under spatial force and momental loading
of general type have been found yet.
As follows from the reference data, application of the soil base nonlinear models
enables the calculation conditions to be essentially shifted towards higher loads, up
to the limiting values. In most cases the nonlinear calculation methods are based
on the finite-element modelling of axisymmetric modelling of axisymmetric foun-
dation structures (of cylindrical or conical shape), subject to a central loading. The
authors pay the greatest attention to the studies of influence of the strength-related
soil properties on the character of the settlement-vs-load dependence. All methods
of evaluation of deformability of nonlinear (elastoplastic) bases are characterized
by high labour consumption in input data preparation, a great number of iteration
cycles at each loading step as well as insufficient substantiation of the choice of
[email protected]
an elastoplastic deformation law. A disadvantage of the existing nonlinear meth-
ods of foundation base calculation consists in the spatial functioning of the nonlin-
early deformed base being taken into account only for the simplest case of centrally
5.3 Calculation of Bases for Pyramidal Piles 415

loaded foundations in the form of rotation bodies, what is actually equivalent to a

consideration of the problem of a flat stress-strained state. In such a formulation it
is not possible to perform deformation-based foundation calculations for off-central
loading, the most widely used in design practice. The above speculations are even
more applicable to foundations with a complex geometrical shape of the deepened
part, which for any loading conditions determine the spatially non-uniform stress-
strained state in the foundation active area, i.e. in all cases determine the complex
functioning of a base.
The subsequent sections of the chapter contain the results of boundary-element
modelling for the most popular foundations with a complex (non-traditional) shape
of the deepened part, applied in the modern practice of enabling works. The design
of such foundations requires the account of real depth, specific features of the geo-
metrical shape of the contact surface with the soil above the footing, load applica-
tion eccentricity, etc. Calculation of bases for such foundation structures leads with
necessity to the solution of spatial contact problems, since reliability in the determi-
nation of the foundation settlements, horizontal displacements and slopes depends
to a great extent on the accuracy of determination of tangential and normal stress on
the contact surface of the foundation with the soil.

5.3 Calculation of Bases for Pyramidal Piles Under Vertical,

Horizontal, and Momental Loads

It is well known from the literature on foundation engineering that traditional foun-
dations of long prismatic piles have low economic efficiency [13, 102, 149]. One of
the alternative solutions for loose soils, covered by a layer of a stronger soil (3–6 m
thick), is the application of short pyramidal piles.
Short pyramidal piles began to be introduced into engineering in the mid-1970s
and soon became very popular due to their high efficiency in comparison with the
prismatic piles [78]. The experiments [4, 13, 31, 94, 172] show the pyramidal piles
to have carrying capacity by factor of 1.5–3 higher than prismatic piles of the same
volume. Such increase of the pyramidal pile bearing capacity is explained by a con-
siderable compression of soil around them and a specific character of their inter-
action with the surrounding soil. In particular, pyramidal piles have large lateral
surface and, due to their faces being inclined, the soil resistance to normal stress
increases. Due to these circumstances, rather large loads can be transferred to the
piles, even for loose soils in their bases.
Pyramidal piles are successfully used for objects of industrial, civil, rural engi-
neering [13, 53, 57, 58, 433]. Construction on such piles is economically justified
for the most of engineering-and-geological conditions, including the bases consist-
ing of nonhomogeneous, collapsing, swelling, heaving soils [8, 213, 249, 250]. As
shown by recent all-round studies [172], application of pyramidal piles in freezing
[email protected]
heaving soils can provide reliable functioning of light buildings and structures with
an essential reduction of the foundation cost. Note that introduction of pyramidal
piles instead of strip foundations enables the foundation cost to be reduced by half,
416 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

labour consumption for their construction – by up to 60%, concrete consumption –

by up to 70%, the foundation material consumption – by a factor of 3 [69].

5.3.1 Existing Approaches to the Calculation of Piles

with a Variable Cross-Section

The experimental proof of the efficiency of pyramidal pile application in engineer-

ing became an impetus to the development of a theory of their calculation. The
problem of creation a method for pyramidal pile calculation has been studied exten-
sively since the 1970s. The relevant papers [7, 13, 52, 57, 58, 69, 102, 103, 129–131,
249, 250] should be mentioned as giving the most complete characteristics of the
existing approaches to the pyramidal pile calculations.
A calculation of pyramidal piles under a combined action of vertical, horizontal,
and momental loads is proposed in [57, 249, 250]. A pile is considered as a rigid
volumetric punch and the base – according to the Fuss-Winkler model. Components
of the soil reactive back pressure over all four inclined sides of the pile are taken
into account in [57, 249]. In [250] the soil back pressure only at the front and rear
sides of the pile lateral surface is taken into account. In these papers the method of
determination of coefficients of subgrade reaction are also different, the later being
introduced in the calculation as constant values. The authors of [58] propose the
coefficients of subgrade reaction, nonlinearly dependent of the contact stress values
and the calculated soil resistance, to be applied for the pyramidal pile calculations.
A simplified solution of the elastoplastic problem for a vertically loaded pyramidal
pile is obtained by the finite-element method in a flat formulation in [13]. A similar
approach is applied in [52] to the studies of obliquely and off-centre loaded pyrami-
dal piles. The authors of [8, 69, 151] recommend the principle of independent action
of forces to be applied for the pyramidal pile calculation. At such approach the pile
calculation is performed separately for the vertical load, on one hand, and on the hor-
izontal and momental loads, on the other hand. In the case of calculation for the hor-
izontal and momental loads, a pyramidal pile is modelled as a rigid trapezoidal plate,
restrained in the base. In [69] the soil properties are described by the Fuss-Winkler
model, while in [131] an elastic linearly deformable half-space is used. When the
calculation for vertical and horizontal loads is performed, the pile is treated as a
rigid volumetric punch, fixed in a linearly deformable elastic half-space [129, 130].
A paper [7] and a book [102] are devoted to the pyramidal pile calculation for a ver-
tical load in an axisymmetric formulation when the pile is modelled by a truncated
cone, equal in height and in volume. An advantage of these studies is that the calcu-
lation scheme takes into account the soil nonhomogeneity due to its compression at
the pile driving. In [7] the finite-element method is employed for the studies of the
pyramidal pile interaction with soil, and in [102] an approximate analytical solution
of the mixed problem of theory of elasticity and plasticity is obtained.
[email protected]
Thus, at present several methods for base deformation-based calculation of pyra-
midal piles have become popular. The main shortcomings of these methods should
be mentioned as follows:
5.3 Calculation of Bases for Pyramidal Piles 417

– lack of account of combined action of vertical, horizontal, and momental loads,

– introduction of coefficients of subgrade reaction for determination of horizontal
and vertical reactive pressures of the soil and their rather approximate assignment
for different soil types and states,
– application of a rectangular profile for the coefficients of subgrade reaction which
does not take into account their variation with depth,
– presence of extensive auxiliary tables, applicable only for separate nominal sizes
of pyramidal piles and physical and mechanical parameters of soils,
– substitution of a pyramidal pile by a deepened conical punch with a surface equal
to those of the pile, and application of solutions of the axisymmetric problem of
limit equilibrium theory.

A required increase of the pyramidal foundation efficiency can be achieved by

application of solutions of mixed problems of theory of elasticity and plasticity in a
spatial formulation [142]. However, in this case essential mathematical difficulties
arise. Even the attempts to apply well elaborated numerical methods (FEM) for the
studies of soil bases in a flat or axisymmetric stress-strained state due to static loads
result in a great increase of the required memory size and computation time [141].
Besides, consideration of the soil as an ideal elastoplastic medium in the framework
of the associated and non-associated plastic fluidity laws does not enable the solu-
tions of essentially three-dimensional problems to be applied for the development
of engineering methods for the calculation of foundations of various type since they
are mostly of theoretical character.
To our knowledge, among the papers, devoted to pyramidal pile calculations,
only [129, 130] contain calculations according to the second group of limiting states,
based on the application of solutions of spatial problems of theory of elasticity.
The boundary-element method in its simplest form is applied. A pyramidal pile of
square cross-section is substituted by a telescopic absolutely rigid body. At vertical
or horizontal loading, only the Mindlin’s solution components, corresponding to the
load direction, are used, with the soil Poisson ratio ν=0.35. Integration of displace-
ment functions over each vertical lateral side of the telescopic body is substituted by
summation of displacements from our concentrated forces as well as sixteen con-
centrated forces at the lower horizontal end. This calculation method was applied
for comparison with the test results [53] for a pyramidal pile with a square cross-
section of 70 × 70 cm2 near the surface and 10 × 10 cm2 at the depth of 5 m. The
soil deformation modulus was determined by punch tests and at the construction site
it was within 2.5 to 3 MPa. Under the load on the pile of 500 kN the settlement was
5 cm. Even for a rather small number of telescopic part (n= 5) and the deformation
modulus, taken as the average value over the site 2.75 MPa, the settlement value
of 0.0467 m was obtained, the discrepancy with the experiment being only 6.6%
[130]. This enabled a conclusion to be made that the application of rectangular and
"telescopic" elements in the calculations of pyramidal pile settlements considerably
[email protected]
simplifies the solution in comparison with the finite-element method and even for
a small mesh (5–7 elements) is quite applicable for obtaining results, sufficiently
accurate for practical purposes.
418 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

Fig. 5.1 Calculation scheme

for the spatial loading of a
pyramidal pile

Developing the above approach for the pyramidal pile calculation under simul-
taneous action of vertical, horizontal, and momental loads (Fig. 5.1), we apply an
original modern numerical method of solving the spatial contact problem, based on
the direct boundary-element method.
While formulating the contact problem, we present the pile calculation scheme
as a rigid volumetric deepened punch in the shape of the pyramidal pile to be calcu-
lated. We assume a linearly deformable half-space with a cavity, whose boundaries
coincide with the pile contact surface, as a model for the base. For modelling the
contact interaction of the pile with the soil, we apply unilateral constraints, which do
not work in tension and possess an unlimited resistance to friction under compres-
sion. The system of equations (2.9) and (2.10), corresponding to the spatial contact
problem of theory of elasticity for a deepened punch whose surface is bonded with
the base, is complemented by a restriction

p(n) (N) ≥ 0 (5.4)

where p(n) (N)is the contact surface vector projection in the point N onto a normal,
external to the pile structure surface, contacting with the soil.
The system of Eqs. (2.9) and (2.10) with a restriction of Eq. (5.4), due to the
presence of an inequality, results in a class of contact problems with a construction
type nonlinearity; hence, the principle of independence of forces is not applicable
for solving it. For each separate case of spatial loading of the pile, a separate calcu-
lation should be performed.
The system of Eqs. (2.9), (2.10), and (5.4) is solved, using the direct boundary-
element method (Sect. 2.2) in accordance with the following modification, taking
into account the presence of unilateral constraints. If there is no preliminary infor-
mation on the contact domain, then after the coefficient matrix having been formed,
[email protected]
an iterative program mode is entered, at each step of which the solution of the sys-
tem of Eqs. (2.13) and (2.14) is determined. At the full contact of the pile founda-
tion with the base the system of Eqs. (2.13) and (2.14) is a closed system of (3M+6)
5.3 Calculation of Bases for Pyramidal Piles 419

linear algebraic equations where M is the total number of the boundary elements.
At each step of iteration the values p(n) (Nt )≥0 (t=1, 2,. . .M) are analyzed in order
to determine the boundary elements, at which tensile normal stress arises. The rows
and columns of the matrix of the equation system (2.13) and (2.14), corresponding
to such elements, are made zero at subsequent steps. The iteration process is finished
when there is no tensile stress along the normal to the foundation contact surface.
The experience of the program operation shows that the iteration number is within
4–7, no program cycling being observed.
The routine for the pyramidal pile calculation under a combined action of ver-
tical, horizontal, and momental loads from the elaborated Rostwerk software kit is
simple in operation and requires the following input data: total number of the bound-
ary elements, into which the pile contact surface is meshed (Fig. 5.2), the size of the
upper and lower pile ends and the pile height, projections of force and moment vec-
tors, applied to the pile head, the deformation modulus and the Poisson ratio of the
soil.
Note that in the numerical method applied here for pyramidal piles, the geomet-
rical shape of the foundation is exactly described and the tangential conditions at
the contact surface of the pile and the soil are taken into account. Therefore, the
boundary-element approach in the full spatial formulation with the application of
the pyramidal pile discretization into several hundreds of boundary elements will
undoubtedly provide higher level of accuracy of characteristics of the contact inter-
action of the pyramidal piles with the soil than the "telescopic element" method.
The efficiency and reliability of the proposed method for pyramidal pile calcula-
tion are estimated in the subsequent sections by comparison of the numerical results
with the experimental data.

[email protected]
Fig. 5.2 Discretized surface
of the pyramidal pile contact
with the soil
420 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

5.3.2 Calculation for the Vertical Load

The results of static tests of a pyramidal pile under a vertical load are given in [53].
The pile was 5 m long, the side of the upper square end of the pile was 0.7 m,
the lower end side 0.1 m. The pile base was a high-plasticity clayey, the average
deformation modulus value E=2.75 MPa was found from the punch test data. The
density and strength-related parameters of the clayey were the following: γ=16.3
kN/m3 , ϕ=10◦ , C=15 kPa. The Poisson ratio of the clayey is assumed ν=0.35.
Figure 5.3 shows the settlement-vs-load plots for the pyramidal pile under con-
sideration, obtained experimentally (line 1) and from the calculation (line 2). In the
same figure the pile settlement curve is shown (line 3), obtained in Appendix F
according to the method, quoted in the instruction [69]. Comparison of the experi-
mental and calculated data (lines 1 and 2) leads to a conclusion that the settlement-
vs-load plots, obtained experimentally and from the solution of the spatial contact
problem, are practically parallel, with the exception of the initial part (in particular,
this is the evidence for the correct evaluation of the high-plasticity clayey deforma-
tion modulus from the punch test data). Note that the contact problem solution can
be made even closer to the experimental results, if one takes into account the fact that
when the stress in the soil is lower than its structural strength, the soil deformations
are extremely small. According to the experimental data (limit 1), the soil structural
strength limit is overcome at the vertical load P3 ≈ 50 kN. Then the contact prob-
lem solution with the account of the soil structural strength can be presented by a
plot (line 4) which is essentially closer to the experimental curve. Note also that the
results of the pile settlement calculation (line 3) according to the instruction [69]
are in a good agreement with the experiment only at the very initial stage of the pile
loading, and at other stages they are quite far from the reality.

5.3.3 Calculation for the Action of a Horizontal Load

The results of the numerical studies were also compared with some of the experi-
mental data published in [57]. Horizontally loaded pyramidal piles of two nominal
sizes were considered: one pile with the length of 5.5 m with the head size 0.6 × 0.6
m and the footing size 0.2 × 0.2 m, and a second pile with the length of 4.4 m and
the head and the footing size 0.75 × 0.8 m and 0.3 × 0.3 m, respectively. The piles
were tested at four experiment sites, three of which were on water-saturated clayeys
and one – on water-saturated clays. Table 5.1 presents the results of our calculations
of piles along with the average values of physical and chemical parameters of soils
of the experiment sites which were kindly provided by the author of [57].
The analysis of the data, quoted in Table 5.1, has shown that the boundary-
[email protected]
element results give a satisfactory agreement with the experimentally measured pile
displacements. Note that for the piles Nos. 4 and 6 the boundary-element method
gives the results close to the field measurements than the finite-element method [57].
 5.3

Table 5.1 Parameters of the test site soil physical and mechanical properties and results of pyramidal pile calculations

Parameters of the test site soil physical and mechanical Horizontal displacement of
properties Pile dimensions Load on pile the pile head Δ1 , mm
Pile no. Deformation Calculation
according Density in Unit Internal modulus Calculation according
to the table natural state cohesion S, friction (punch)E, Poisson Head Horizontal Momental according to the

[email protected]
in [57] γ , g/cm3 MPa angle ϕ, deg MPa ratio ν Length, m P1 ,kN M2 , kNm Test to [57] BEM
Tip,cm
Calculation of Bases for Pyramidal Piles

60 × 60 7.7
2 1.95 0.043 12.3 11 0.47 5.5 160 30 10 9.0
20 × 20 8.1
60 × 60 9.0
3 2.02 0.027 5 8.5 0.49 5.5 144 30 10 10.0
20 × 20 9.5
60 × 60 9.8
4 1.83 0.041 19.7 11 0.42 5.5 200 40 10 13.9
20 × 20 10.3
75 × 80 10.2
6 1.90 0.047 16.6 6.0 0.44 4.4 115 20 10 11.3
30 × 30 10.8
Note. The last column of the table contains the displacement values with (denominator) and without (numerator) unilateral constraints on the pile–soil
contact.
421
422 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

Fig. 5.3 Dependences of the pyramidal pile settlements on the vertical load: 1– experiment, 2–4
– calculation

5.3.4 Calculation for the Action of an Inclined Load

Figure 5.4 shows the calculation scheme and the discretized surface of a prismatic-
pyramidal pile loaded by a vertical and a horizontal force. The pile length is 3 m,
the head size 0.6 × 0.6 m, the footing size 0.1 × 0.1 m, in the upper part the pile
is prismatic, this partition length being 0.8 m. The given pile was tested under an
inclined load at two test sites [5]. The first site was located on a stift clay with the fol-
lowing physical and mechanical parameters: γ=19.7 kN/m, ϕ=19◦ , C=0.033 MPa,
E=18 MPa, ν=0.426. The angle of inclination of the external load, applied to the
pile head, with respect to the horizon, was α=55◦ . The second site was located on
a stift clayey: γ=17.7 kN/m, ϕ=20◦ , C=0.015 MPa, E=14 MPa, ν=0.417. At this
site the pile was tested by the load at an inclination angle α=68◦ .
The horizontal displacement of the pile head is plotted across the inclined load in
Fig. 5.5. Lines (1) and (2) in the figure correspond to the experimental dependences
[5], and lines (3) and (4) are plotted according to the calculation, performed by the
boundary-element method. Comparison of the results, shown in Fig. 5.5, leads to
a conclusion that in the major part of the whole load range the experimental and
theoretical dependences are close to each other. This range is restricted by a load
[email protected]
of 65–70% of the pile bearing capacity what, in turn, corresponds to the admissible
calculation load on a pile at the reliability coefficient near 1.4, what is specified by
the Construction Rules and Regulations for the pile foundation design.
5.3 Calculation of Bases for Pyramidal Piles 423

Fig. 5.4 Calculation scheme

(a) and discretized surface of
contact with the soil (b) for a
pyramidal-prismatic pile
loaded by an inclined force

(a) (b)

Fig. 5.5 Dependence of the pile head horizontal displacement on the joint action of vertical and
horizontal loads: 1,2 – experiment [76], 3,4 – calculation; (a) site No. 1 (semihard clay); (b) site
No. 2 (semihard clayey)

5.3.5 Calculation for the Combined Action of an Inclined Force

and a Moment
We have analyzed an example of calculation of a pyramidal pile with the upper end
size 0.68 × 0.68 m, the lower end size 0.1 × 0.1 m, length 2.9 m, loaded by a ver-
tical load of 155 kN, a horizontal load of 83 kN, and a momental load of 22 kN·m,
taken from the instruction book [69]. The soil is represented by a fine damp sand
with the following physical and mechanical characteristics: γ=17.4 kN/m3 , ϕ=28◦ ,
[email protected]
C=3 kPa, E=18 MPa. The soil Poisson ratio is taken as ν=0.25. The calculation
results for this pile from the instruction book [69] and obtained by the method pro-
posed here, are listed in Table 5.2.
424 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

Table 5.2 Parameters of contact interaction of the pyramidal pile with the soil, loaded by an
inclined force and a moment

Calculation method Settlement z , mm Horizontal Slope

displacement x , ψ102 , rad
mm

Instruction book [69] 6.2 5.0 0.32

Boundary-element method 3.7 3.6 0.16

It follows from the table that, similarly to the earlier comparison, for a centrally
loaded pyramidal pile the numerical boundary-element solution of the spatial con-
tact problem for the case of loading by an inclined force and a moment, results
in essentially smaller deformations of the pile base than for the predictions of the
instruction book [69].
Thus, the studies performed show the evidence for the proposed approach being
promising for the pyramidal pile base calculations in the most general spatial for-
mulation. Its further improvement is related to a detailed experimental testing at
different engineering-and-geological conditions that will enable reliable range of
the method applicability to be established and the method itself to be suggested for
application in the development of regulations.

5.4 Interaction of Bases and Rigid Bored Foundations with

Vertical and Inclined Piles
In many countries the fraction of pile foundations is above 25% of the gross founda-
tion volume, and in some regions up to 40–70% buildings and structures are erected
on pile foundations [10]. Development and implementation of new pile structures
with high technological parameters is considered to be one of the major trends to
solve the problem of pile foundation efficiency. This can provide an essential eco-
nomic effect and qualitatively improve the figures.
Extensive application of rational structures of rigid pile foundations in engineer-
ing [20, 75, 114, 210] is hampered by a low reliability level of the practical calcula-
tion methods, based on separate empirical data and extremely simplified theoretical
models of soil. For example, according to the design regulations, accepted in Rus-
sia, pile foundation base deformations are studied for a vertical load using a linearly
deformed half-space model, and separately for a horizontal and a momental load
using the coefficient of subgrade reaction (Winkler) model [208]. Dependence of
the foundation calculation scheme and the base model on the external load direction
is the reason for the low reliability of the results of calculations to determine the pile
foundation displacements in the soil at the stage of its functioning.
[email protected]
A considerable economy due to a reduction of materials consumption can be
obtained due to the account of functioning of a low pile raft in pile foundations
[190]. As follows from the estimations, a pile raft can accept from 20 to 70% of
5.4 Interaction of Bases and Rigid Bored Foundations with Vertical and Inclined Piles 425

the total load on the foundation. The method, used in the Construction Rules and
Regulations 2.02.03–85 [208] for the design of pile foundations, does not take into
account the soil distributive properties in case the load being transferred by the pile
lateral surface and tip, as well as the low pile raft functioning.

5.4.1 Structure, Design, and Specific Features of Calculation

of Rigid Pile Foundations with Short Piles and a Pile Raft

Pile foundations with short rigid piles have found applications in various fields of
engineering [151, 169, 235, 247]. In particular, for rural engineering (cattle farms,
warehouses, hangars, etc.), with bearing structures in the form of three-hinge frames
or arches as well as vaulted folded plates, one of the most rational foundation types
are short vertical and inclined bored piles with a raft in the form of a washer or a
cantilever [105, 222].
When inclined piles are used, their longitudinal axis is put along the direction
of the resultant of the main external load combination. This results in an essential
improvement of conditions of functioning of the soil, surrounding the pile. As fol-
lows from the field experiments data [118, 119], the foundation slope is reduced,
horizontal displacements of the mounting groups of bearing structures are reduced
to a minimum. The stress in the pile cross-sections is purely compressive, what
essentially favours the decrease of the bending moments and transverse forces in
the pile body and enables one to use the only reinforcement to conjugate the pile
shaft with the pile raft. In case the horizontal loads exceed the vertical ones, the pile
rafts are made in the form of an unloading cantilever what is also helpful to reduce
the foundation slope and its horizontal displacements. Such pile foundation struc-
ture is one of the most rational foundation structures, applied under legs of aerial
power transmission line supports [100].
In Russia, no special regulations for the design of pile raft foundations with ver-
tical and inclined piles have been developed so far. Therefore, for substantiating the
engineering calculations, in the first approximation the regulations for the design of
conventional pile foundations are used. However, the calculation schemes, provided
by the Construction Rules and Regulations [208], do not reflect some essential fea-
tures of interaction of bases and pile raft foundations. In particular, the calculations
according to the Construction Rules and Regulations do not take into account the
soil reaction over the pile raft bottom as well as mutual influence of the pile raft
and the pile. This is the reason for the behaviour of inclined foundations under an
operating load being hardly predictable with sufficient reliability.
Another rational field of application for short piles are foundations under pipeline
trestles, power line supports, contact-wire line supports [96, 166, 167]. In this case
[email protected]
the lower part of the support, sunk into a preliminarily drilled well, is often used
as a pile. In order to increase the bearing capacity for a horizontal load at the base
surface level, the piles are reinforced with washers or soleplates [17, 72].
426 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

Technology of construction of short pile foundation with rafts is highly mecha-

nized. After the site planning, wells are drilled. Compression of the residual slime
at the hole bottom is performed by impression of the working part of the drill with-
out rotation into the well bottom. After drilling, reinforcing cages are installed and
fixed in the wells. Then pile construction process is begun. A concrete-delivery truck
arrives directly at the construction site and supplies concrete directly from the agi-
tator to the well through a filling funnel. The concrete mixture compaction is per-
formed using vibration tools. The concrete should have pourable consistency. After
the piles having been produced, reinforcing fabrics and raft formwork are set. Con-
crete is supplied into the formwork directly from the agitator with its subsequent
compaction by a platform vibrator. The productivity is even more increased in case
a gap between the concrete placement into the pile and the raft can be avoided.
In order to achieve higher adequacy between the calculation and experimental
data, we use the spatial contact problem formulation for a complex-shaped rigid
punch, deepened into a soil base, performed in Chap. 2. The calculations are per-
formed according to the second limiting state for rigid pile foundations of a constant
cross-section, both with and without a raft.
We accept initially that the external load (from the power line support or other
above-foundation structures) is transferred to the base not only by the pile contact
surface, as it is set in [208], but also by the pile raft bottom. Interaction of the raft
lateral surface with the soil is not taken into account due to a high friability of
the back fill soil as well as due to the fact that for power line supports and many
other engineering structures, as a rule, higher pile rafts are used, located on the base
surface.
Since the flexibility of a short pile foundation is low (with respect to the soil
base), it can be treated as an absolutely rigid volumetric deepened punch. Small
pile length (2–4 m) provides, as a rule, their location within the same engineering-
and-geological layer. Therefore, in the first approximation for the simulation of the
soil distributive property and the possibility of account of mutual influence of the
pile raft and the pile on the stressed state of the soil, the base model is taken in the
form of a homogeneous isotropic linearly deformable half-space with a vertical or
inclined cavity. The pile raft bottom is resting on the half-space surface, and the
cavity is fully filled with the pile.
Since the soil practically does not work in tension, the numerical modelling
should take into account the unilateral character of the foundation contact with the
base: in the areas of the foundation uplifting from the soil, all contact stress compo-
nents are zero.
Some examples of automatic boundary-element discretization of contact surfaces
of pile foundations with inclined and vertical piles and pile rafts of various shapes,
used for practical calculations, are shown in Figs. 5.6 and 5.7.

[email protected]
5.4 Interaction of Bases and Rigid Bored Foundations with Vertical and Inclined Piles 427

(c)
(a) (b)

Fig. 5.6 Discretized surfaces of contact of soil and a vertical pile with (a) a circular and (b) a
rectangular eccentric pile raft, as well as (c) of soil and an inclined pile with a rectangular pile raft

(a) (b)

(c) (d)

[email protected]
Fig. 5.7 Discretized surfaces of contact of soil with (a) a circular eccentric pile raft for a vertical
cylindrical pile, (b) a rectangular pile raft for an inclined cylindrical pile, (c) and (d) a circular pile
raft for piles with square cross-section
428 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

Fig. 5.8 Cylindrical punch,

deepened into an elastic
half-space: (a) calculation
scheme, (b) deconvolution of
the lateral surface and
discretization of the contact
domain

5.4.2 Vertical Cylindrical Piles Under an Inclined Load

Piles without a pile raft. A series of numerical experiments was devoted to the stud-
ies of the influence of a cylindrical pile length and the direction of an inclined load
on the character of the pile displacement in the soil. The calculation scheme, corre-
sponding to the case in question, is presented in Fig. 5.8. The same figure shows the
discretization of the contact surface of the pile and the soil into boundary elements.
The external load values, contained in the equation system (2.13) and (2.14), are
taken as follows:

P1 = 0, P2 = R cos ϕ, P3 = R sin ϕ, M1 = M2 = M3 = 0.

Calculations of characteristics of contact interaction of separate cylindrical piles

depending on the inclined load direction are presented in Figs. 5.9 and 5.10. Due
to the calculation scheme symmetry with respect to the vertical pile axis, the cal-
culations are performed for the external force inclination angles within the range of
90◦ ≤ϕ≤180◦ . In order to plot the results graphically (most of the plots appeared to
be quite monotonous), the calculations were performed with a step of 6◦ . In some
cases the step was reduced to 3◦ , 1◦ , and even 0.5◦ . The relative pile length H/r1 was
3, 10, and 20, r1 being the pile cross-section radius. The calculations were carried
out in two variants – with and without the account of unilateral constraints at the
pile–soil contact and the values of R/Er1 2 = 1, ν=0.3.
Figures 5.9 and 5.10a show the calculation data for a short pile (H/r1 = 3).
Numerical results, obtained under a vertical load without the account of unilateral
constraints, were the test data and are in part described in Sect. 3.5.2. With the
increase of the angle of inclination of the force R the horizontal component of the
inclined load Rsinϕ increases while its vertical component Rcosϕ simultaneously
[email protected]
decreases. This results in the absolute values of the vertical and horizontal displace-
ments decrease and increase, respectively. Simultaneously the slope ψx increases,
its maximal value being achieved under a purely horizontal load.
5.4 Interaction of Bases and Rigid Bored Foundations with Vertical and Inclined Piles 429

The account of the unilateral constraints in the contact domain results in con-
siderable quantitative changes. This is related to the contact area decrease due to
the formation of contact zones where tensile normal stresses act. As follows from
the numerical experiments, the zones with negative stress arise at the lower end of
the pile shaft (Fig. 5.9) and increase towards the foundation bottom, encompassing
larger areas both with the increase of the load absolute value R and with the relative
deepening parameter H/r1 . Here a "wedging" effect is revealed: on the lower part of
the pile lateral surface a tension zone is formed as well as a gap between the pile
and the soil, while on the upper part a compression zone arises.
The strongest effect of the "wedging" on the pile behaviour is for the ϕ angles,
close to 90◦ , for the solutions, obtained with the account of the gap formation
between the pile and the soil. In our opinion, this effect is to a considerable degree
related to the specific features of the contact pressure formulation. They consist in
the fact that on the boundary elements where compressive normal stress acts, no
possibility of the pile slippage with respect to the base is provided and, as a conse-
quence, no restrictions are imposed on the tangential stress values. In the course of
the numerical solution of the contact problem, at each iteration step, boundary ele-
ments on the pile lateral surface are consecutively (from the bottom up) excluded,
since the stresses on these elements are tensile (See the test example from Sect.
3.5.2 as well as Fig. 5.9). In the course of the iterations part of the lateral surface
of the vertically loaded pile is excluded from the contact domain between the pile
and the soil. As a result, the load is transferred from the pile to the soil only by its
bottom and a ring-shaped area of the lateral surface, represented by the upper rows
of the boundary elements. With the account of the possibility of the pile slippage

[email protected]
Fig. 5.9 Profiles of tangential (τ ) and normal (σ ) contact stresses at central loading of a short pile
(H/r1 =3) without a pile raft
430 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

with respect to the base surface, the influence of the "wedging" effect on the pile
behaviour can be smoothed.
For all external load inclination angles, the "wedging" effect results in an increase
of the absolute values of horizontal displacements y and slopes ψx . The maximum
relative difference for these values, due to the account of the unilateral constraints,
for a short pile (with a relative depth of H/r1 = 3) reaches 25% and 37%, respec-
tively. With the increase of the resultant force inclination angle, the z values, with
the account of the unilateral constraints, almost up to ϕ≈125◦ exceed the corre-
sponding values, obtained without the unilateral constraint account. At ϕ≈125◦ the
two plots intercept (Fig. 5.10a) what is the evidence for the fact that the contact
interaction with the formation of a gap between the pile and the soil at such ϕ does
not lead to any changes of vertical displacement. At further increase of the resul-
tant force angle ϕ, the account of the unilateral constraints results in an increase

(a) (b)

(c)

[email protected]
Fig. 5.10 Displacements and loads of a cylindrical pile depending on the inclined load direction
without (1) and with (2) the account of unilateral constraints; H/r1 = 3 (a); 10 (b); 20 (c)
5.4 Interaction of Bases and Rigid Bored Foundations with Vertical and Inclined Piles 431

of the absolute values of horizontal displacements y and slopes ψx . However, the

increase of the external load horizontal component, increasing with ϕ, results in a
sharp decrease of vertical displacements z . At the load, practically close to hori-
zontal (ϕ≥ 170◦ ), there are practically no vertical displacements near the pile head,
and with the further increase of the load inclination angle the pile can even go up
(z <0). In other words, at horizontal (or close to horizontal) pile loading it can be
squeezed out of the soil towards the base surface. With the increase of the rela-
tive pile length this effect becomes less noticeable (Fig. 5.10b, c). Note that in case
unilateral constraints being not taken into account, no such effect is observed, i.e.
always z ≥0 and the pile does not move upward.
Consider the effect of the relative depth H/r1 on the linear and angular dis-
placements of the pile. It can be analyzed from the comparison of the calculated
dependences, plotted in Fig. 5.10. The numerical analysis performed has shown
all the calculated dependences to be qualitatively similar. As one could expect,
the pile displacement under the same external load essentially decreases with the
increase of the relative pile length. With the relative length decrease, the discrepan-
cies between the solutions, obtained with and without the account of unilateral con-
straints, become more and more noticeable. The pile settlement behaviour depend-
ing on the conditions, accepted for the contact surface, is ambiguous. At the angles
ϕ, close to 90◦ , the account of unilateral constraints results in the pile settlement
increase. In the range of ϕ variation from 90 to 180◦ , starting from a certain value
of ϕ, depending on the pile length, a reverse dependence is observed, namely the
settlement decrease with the account of the unilateral constraints. Variation of the
H/r1 parameter provides the strongest effect on the vertical pile displacements. The
relative depth increase plays an important role only under almost vertical loading
(ϕ≈90◦ ±6◦ ), when the effect of horizontal forces is small yet, and the decisive role
is played by the "wedging" effect. At such loading conditions, the account of uni-
lateral constraints results in considerable additional settlements, whose values can
be compared with the settlement values calculated without the gap formation, using
Table 5.3.
The obtained calculation data for the settlements and slopes of separate short,
medium, and long cylindrical piles enable, with the account of the pile uplifting
from the soil base, the displacements and slopes under an arbitrary inclined load,
the settlements and slopes under an arbitrary inclined load to be evaluated without
the hypothesis of the coefficient of subgrade reaction. Besides, the data quoted here,
will be used in subsequent chapters for comparative estimations of the obtained

Table 5.3 Effect of relative depth on the settlement of piles without a raft under vertical loading
without (numerator) and with (denominator) the account of unilateral constraints

H/r1 3 10 20
[email protected] 0.406 0.106 0.0685
z /r1 0.436 0.165 0.2355
Settlement increment, % 7.4 54 244
432 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

Fig. 5.11 Calculation

scheme for a vertical pile
with a circular pile raft

calculation results for pile raft foundation structures with both vertical and inclined
piles.
Vertical short piles with a low circular pile raft. The presence of a low pile raft is
known [20, 167] to produce an essential effect on the load transfer to the base and,
in particular, to enable the load increase on a pile foundation.
We studied the effect of the pile raft size on the interaction of the pile foundation
with the soil for the example of vertical cylindrical piles with a circular cylindrical
pile raft. The calculation scheme for this case is presented in Fig. 5.11. The general
outlook and the discretization of the contact surface of the pile foundation with the
soil into boundary elements are shown in Fig. 5.6a. The discretization of the pile
raft contact surface is illustrated by Fig. 5.7a.
For the calculations the following parameters were fixed: the pile diameter r1
and its length H = 10r1 . The pile raft radius (r2 /r1 = 0, 2, and 4) and the external
load vector value were varied. The latter value was chosen from the condition of the
average contact pressure being constant for the piles with different pile rafts. The
load direction was varied within 0◦ ≤ ϕ ≤ 180◦ . In a separate study the influence of
the pile raft location eccentricity with respect to the pile axis on the characteristics
of the pile foundation displacement in the soil. The solutions were obtained both
with and without the account of unilateral constraints on the contact of the pile with
the soil.
The results of the performed numerical studies of processes of interaction of
separate piles with a low circular raft with soil are shown in Figs. 5.12, 5.13, and
5.14.
Comparison of the plots, presented in Figs. 5.12 and 5.13, shows that the presence
of a circular pile raft with a radius of r2 = 2 r1 (the pile frat of such size increases
the surface of the foundation contact with the soil by 14.3%) does not result in any
essential qualitative changes of the contact interaction characteristics, in comparison
with the case of a pile without a raft. At a fixed external load value, the decrease of
horizontal displacements both with and without the account of unilateral constraints
in the example under consideration achieves 20.1% with the slope decrease up to
[email protected]
25.8%. The most essential effect of the pile raft presence is for the case of vertical
displacements z at a vertical or near-vertical load. This is revealed, first of all, in
a decrease of settlements for the pile with a raft. With the general pattern of the
5.4 Interaction of Bases and Rigid Bored Foundations with Vertical and Inclined Piles 433

(a) (b)

Fig. 5.12 Horizontal (a) and vertical (b) displacements of pile foundations (H/r1 =10) with a
circular pile raft depending on the inclined load direction without (solid lines) and with (dashed
lines) the account of unilateral constraints: (1) r2 =4r1 , R = r1 2 ·E; (2) r2 =2r1 , R = r1 2 ·E; (3)
r2 =4r1 , R =1,5r1 2 ·E

"wedging" effect being preserved, at the increase of the load inclination angle the
presence of the pile raft results in a smoother decrease of settlements at the account
of unilateral constraints and, on the contrary, to their sharper decrease in case the gap
formation not being taken into account. With the further increase of the pile raft size
(r2 = 4 r1 ) at vertical or near-vertical loading the "wedging" effect is still preserved
(Figs. 5.12 and 5.13); however, it is less pronounced. It follows from the calculations
performed that the pile raft size increase essentially affects the displacements and
slopes at a load close to horizontal, when a considerable part of the pile raft is
excluded from the interaction with the soil. The data for the comparative analysis of
the displacement and slope numerical values under vertical and horizontal loading
are presented in Table 5.4.

Table 5.4 Effect of pile raft size on the characteristics of contact interaction of the soil and the pile
foundation (H/r1 =10) under vertical (z ) and horizontal (y , ψx ) loading without (numerator)
and with (denominator) the account of unilateral constraints

Horizontal loading Vertical loading

r2 /r1 y /r1 ψx (rad) z /r1

0.184 0.0193 0.102

2
0.221 0.0244 0.138
4 0.121 0.0107 0.0842
0.164 0.0166 0.0943
[email protected]
4 0.181 0.0160 0.126
(R=1.5·R) 0.246 0.0249 0.141

R is the increased load (by factor of 1.5).

434 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

Fig. 5.13 Slopes of vertical cylindrical piles (H/r1 =10) with a circular pile raft depending on
the inclined load direction without (solid lines) and with (dashed lines) the account of unilateral
constraints. (1) r2 =4r1 , R = r1 2 ·E; (2) r2 =2r1 , R = r1 2 ·E; (3) r2 =4r1 , R =1,5 r1 2 ·E

As one should expect, the slopes of a pile foundation structure decrease with the
pile raft size (Fig. 5.13). An increase of the load absolute value results in a practi-
cally similar increase of slopes and displacements. The comparison of calculations,
performed under a fixed average contact pressure for pile structures with rafts of
different size and piles of the same length, shows the pile raft size increase to result
in an essential slope decrease.
Independently of the pile raft size and the external load absolute value, there is
a direction of the load action, for which the account of unilateral constraints does
not result in any change of the pile structure settlements. In this case the decisive
parameter is the relative pile depth H/r1 . As shown by the calculations, at H/r1 = 10
such situation is observed at ϕ≈135◦ –140◦ (Fig. 5.12b). Horizontal displacements
(without the account of unilateral constraints) decrease, though slightly, with the
pile raft size increase. This is the evidence for the main effect on the horizontal
displacements in the case of a full contact of the pile foundation with the soil is
produced by the pile length. On the other hand, with the increase of the pile raft
size, as seen from Fig. 5.12a, the account of unilateral constraints results in the y
displacement increase with the increase of the horizontal load component.
At 45◦ ≤ ϕ ≤ 135◦ the settlements of a pile structure with a larger raft are higher
than for the one with a smaller raft, while at a horizontal or near-horizontal loading
the situation is opposite (Fig. 5.12b). This is related to the fact that, when the vertical
component of the external load exceeds the horizontal one, the major part of force is
[email protected]
taken up by the pile tip and its lateral surface. The redistribution of forces over the
pile raft surface makes less effect on the contact interaction characteristics. At a hor-
izontal load, a considerable part of the structure with a pile raft does not participate
5.4 Interaction of Bases and Rigid Bored Foundations with Vertical and Inclined Piles 435

(a) (b)

Fig. 5.14 Displacements and slopes of a vertical cylindrical pile (H/r1 =10) with (a) a concentric
and (b) an eccentric ( x /r1 =0.3) round pile raft (r2 =2r1 ) depending on the inclined load direction
(1) without and (2) with the account of unilateral constraints

in the contact interaction with the soil and, to a considerable extent, the increase of
the specific contact pressure results in the increase of vertical displacements for the
foundation with a larger pile raft.
By comparing the plots, presented in Fig. 5.14, one can estimate the effect of an
eccentrically located pile raft on the slopes and settlements of a pile structure in the
soil. First of all, as follows from the comparative analysis, even a relatively slow
eccentricity in the pile raft location with respect to the pile symmetry axis (x /r1 =
0.3) results in rather noticeable slopes in the opposite direction. This is an additional
theoretical confirmation for the applicability of asymmetrical foundation structures
of console type, accepting simultaneous horizontal and vertical loads from struc-
tures with a spacer-type scheme [105, 113, 170]. For the horizontal and vertical
displacements the effect of an asymmetrical pile raft is less noticeable: the highest
asymmetry is observed in the difference of horizontal displacements at ϕ=0◦ and
ϕ=180◦ . For the example under consideration this difference is negligibly small
without the account of unilateral constraints in the contact domain and reaches 4.5%
in case them being taken into account.

5.4.3 Foundations with Inclined Piles and a Rectangular Pile Raft

A separate series of numerical experiments was carried out for the estimation of the
effect of the pile inclination angle on the pile raft foundation displacements. Piles
[email protected]
with a given angle α of deviation from the vertical direction were considered, conju-
gated with a rectangular pile raft. The calculation scheme is shown in Fig. 5.15. The
shape of the surface of the foundation and the soil contact, subject to discretization,
436 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

is shown in Fig. 5.6c. The total number of the boundary elements on the pile and the
pile raft did not exceed 500.

Fig. 5.15 Calculation

scheme for an inclined pile
with a square pile raft

Figures 5.16 and 5.17 show the calculated dependences of slopes and displace-
ments for a foundation with piles of a fixed length H/10r1 , inclined to the base
surface at angles α=30◦ and α=60◦ , for the different angles of the resultant inclina-
tion. The pile raft was assumed to be a square with a side 4r1 and the inclined axis
of the pile to pass through its centre. The inclined load acts in the symmetry plane
of the pile foundation (Fig. 5.15).
As follows from the data obtained (Figs. 5.16 and 5.17), all the calculated curves
are considerably asymmetrical. It is seen from Fig. 5.17 that there are no horizon-
tal displacements at the load inclination at an angle ϕ = ϕ0 ≈ 75◦ . The value of
ϕ0 is practically not related to the angle α of the inclination of the pile itself and

Fig. 5.16 Slopes of

foundations with an inclined
pile (H/r1 =10) and a square
pile raft depending on the
[email protected]
inclined load direction (1)
without and (2) with the
account of unilateral
constraints
5.4 Interaction of Bases and Rigid Bored Foundations with Vertical and Inclined Piles 437

(a) (b)

Fig. 5.17 Relative displacements of foundations with an inclined pile (H/r1 =10) and a square
pile raft depending on the inclined load direction (1) without and (2) with the account of unilateral
constraints; α=30◦ (a), 60◦ (b)

is determined mostly by the foundation structure size. Up to the value of ϕ0 the

account of unilateral constraints plays an important role for the determination of the
displacement values. The largest difference is observed at a horizontal load (ϕ =
0◦ ), when the foundation structure is most uplifted from the soil. The same holds
for slopes (Fig. 5.16). With the increase of the angle ϕ of the resultant force inclina-
tion, its horizontal component decreases and the conditions are achieved, at which
the structure and the soil are in contact over the entire surface of the pile and the
pile raft. At α=30◦ the mentioned angle of the load inclination is ϕ≈15◦ , while
for α=60◦ ϕ≈45◦ . However, when the load direction slightly deviates from the pile
axis (ϕ≈α±15◦ ), the "wedging" effect is noticeably revealed, leading to additional
displacements both in the horizontal and the vertical direction. In this case the pile
structure slope due to the account of unilateral constraints, is small what is impor-
tant for the foundation design. A subsequent (up to ϕ≈120◦ –135◦ ) increase of the
resultant inclination angle practically does not result in any noticeable uplifting of
the structure from the soil. With the further increase of the ϕ angle the role of the
horizontal component of the external load vector becomes increasingly important.
Due to the increasing uplifting of the structure from the soil, an additional slope
arises and the horizontal displacement absolute values increase. Simultaneously, the
uplifting of the structure from the soil does not make any noticeable influence on
the development of vertical displacements.
Thus, from the calculations using the proposed boundary-element method, one
can obtain the main features of contact interaction of pile structures with low pile
rafts of various geometry. This will enable the pile foundation displacements and
[email protected]
slopes depending on the external force direction and value to be reliably determined
at the stage of design by variation of the pile raft shape and location as well as the
pile size and inclination angle.
438 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

5.5 Spatial Contact Problem for a Bored Pile Foundation

with a Widening
Bored pile foundations are successfully used for construction of buildings and struc-
tures of various type (industrial, social, residential, rural, etc.) on cohesive (clay)
soils, especially under high concentrated vertical and horizontal loads as well as
on construction sites with complicated geotechnical conditions [97, 162, 166, 247].
High bearing capacity and lower settlements of bored pile foundations in compari-
son with those, performed in open trenches, are explained by the preserved soil base
structure, rational bottom shape and good contact of the foundation with soil.
Application of efficient structures of bored pile foundations enables the degree of
work mechanization to be increased, material consumption for the foundation to be
decreased, the work to be performed in any season, the construction duration to be
reduced. The expenses for the foundation construction are reduced on the average
by 60%. Application of bored pile foundations enables one to decrease the con-
sumption of metal, fuel, and power resources, to reduce essentially transportation
expenditures, as well as to exclude the pile cut-off.
Bored pile foundations are also effective from the point of view of technical
solutions:

– in case of necessity of the piles to be bored through embankments with hard inclu-
sions (in the form of fragments of destroyed stone, concrete, ferroconcrete build-
ings and structures) or through natural layers of hard clay soil, soils with frequent
boulders, etc., when pile driving or immersion cannot be performed;
– on constrained sites where it is difficult to transport and install displacement piles;
– near the existing buildings and structures where inadmissible deformation of ele-
ments of bearing structures or equipment can arise due to the pile driving or
vibroimmersion.

The practical experience of foundation engineering has shown bored pile foun-
dations to be suitable for bearing-wall buildings with bearing basement and ground
floor outer walls as well as for bearing-wall or brick buildings with pile rafts or
edge beams. Besides, bored pile foundations are applicable for skeleton-type build-
ings without basements. In this case the excavation bulk-up is minimal and equal to
the volume of excavation work under the foundations. The most advisable (from the
technical and economical point of view) is application of bored piles in case of the
possibility of their placement without strengthening the walls of the well.
The bearing capacity of bored pile foundations can be increased both by the
pile length increase (this is, however, related to the considerable work complexity
[email protected]
and labour consumption) and due to the bored piles with widenings resulting in
the increase of the bearing capacity due to the development of a part of the shaft
(footing, head, or the intermediate part).
5.5 Spatial Contact Problem for a Bored Pile Foundation with a Widening 439

5.5.1 Production and Structures of Bored Pile Foundations

with a Support Widening
Technology of production of bored pile foundations consists of the following oper-
ations: well sinking using a rock drill, construction of the widening, installation of
the reinforcement, and placement of concrete.
Widenings are mostly constructed by soil drilling out and compression by an
impact or a static load.
Bored piles with radial widenings, produced by the soil compression by an
impact load, are used for construction of rural objects on collapse soils of I type
[234]. The experience of construction work testifies for a high speed of the whole
cycle of construction of each well with a broadened footing. Short bored piles with
a radially broadened footing are effectively applied in clay soils (including collapse
soils of type I and swelling soils) of hard, semihard, and low-plasticity consistency.
For construction of foundations under supports of aerial power lines in ever-
frozen soil, for which labour consumption reaches 40%, an efficient technological
solution is anchoring of metallic piles of steel pipes [92]. This enables the bearing
capacity with respect to pulling loads in summer as well as with respect to the frost
heave to be increased. The most simple and natural is application of the pile body
widening as anchor elements. An effective energy source, providing the pile shape
change and impression of the pile body widening into the soil, is a concentrated
explosive charge, exploded in the pile cavity, since the produced pressures are suf-
ficient for the deformation of soil of any strength. In the course of the pile shaft
deformation, the widening shape is a part of a toroidal surface, described by a semi-
circle, rotating around the pile axis. Due to the widening formed, the force provided
by conventional lifting means is insufficient and the bearing capacity considerably
exceeds the operational requirements.
Bored pile foundation structures with spheroconical widenings have appeared to
be quite reliable [211, 212, 227]. Wide possibilities for the variation of the widen-
ing diameter and depth enable one to adjust the bored pile foundation settlements
at drastically different loads on separate foundations, at the variation of the defor-
mational characteristics of the soil base under the building as well as in the case
of oblique soil base layers. The advantages of the foundations of this type are rela-
tive simplicity of the fully mechanized construction technology, absence of the soil
collapse from the walls of the conical part of the widening during the whole work
cycle. Corresponding typical solutions of bored pile foundations have been worked
out, used for residential houses of 5–14 storeys. The high efficiency of the bored
pile foundations has been shown by the construction of 14-storeyed brick buildings;
the settlements of a building on a bored pile foundation appeared close to those of a
building, constructed on displacement piles. Technical and economic parameters of
conventional pier foundations and bored pile foundations with spheroconical widen-
[email protected]
ings were determined on the base of estimate calculations (for bored pile founda-
tion being used instead of conventional pier ones, the excavation work is shortened
by 94%, concrete consumption – by 19%, metal consumption – by 63%, general
expenses – by 41%, and labour consumption – by 55%) [211].
440 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

Fig. 5.18 Bored pile (a) (b)

foundations with widenings:
(a) with a single widening,
(b) with a double widening

Bored pile foundations with widenings were extensively applied in India for con-
struction on swelling soils as anchor type piles as well as load-carrying piles [149].
For anchor piles, as a rule, one widening is used (Fig. 5.18a); for load-carrying piles
several widenings can be used (Fig. 5.18b). Bored pile foundations with a single
widening are used in the same way as piles with a broadened footing, the widening
can be located above the footing.
For bored pile foundations with two or more widenings, separated by a certain
distance, the soil between the widenings functions as a part of the pile in such a way
that the total soil resistance can be taken on the A–A’ surface of a cylinder with
a diameter, equal to the widening, and height, equal to the Lo step (Fig. 5.18b).
These features of the contact interaction were confirmed by model experiments
[149]. Besides, it was shown by the Indian researchers that in order to achieve the
maximal efficiency, the optimal step between the widenings in a multiple-widening
foundation should be within 1.25–1.5 of the widening diameter. As an example
for economy of materials, the application of bored pile foundations with multiple
widenings, working in London clays, is mentioned: for the same bearing capacity
the concrete volume in a homogeneous pile should be increased by more than four
times.
Finally note that a cylindrical footing, present for all bored pile foundation struc-
tures under consideration, is required not only from the point of view of their pro-
duction technology, but also from the point of view of the structure stability; in this
case the possibility of a unilateral hear of the widening base is excluded and the
vertical character of displacements is provided.
In case developed and perfect methods of calculation and design as well as appro-
[email protected]
priate technical and economic substantiation being used, bored pile foundations
with widenings can be applied for various types of industrial and civil buildings. As
shown by the experience of engineering [227], the method for construction of bored
5.5 Spatial Contact Problem for a Bored Pile Foundation with a Widening 441

pile foundations with support widenings is a promising one, and, hence, requires
further development of the corresponding methods for calculation and design.

5.5.2 Engineering Methods for Calculation of Bored Pile

Foundation Bases from the Base Deformation
For the further development of the existing engineering methods for calculation of
bored pile foundations according to the limiting states and for working out new
methods one should have a clear and full concept of the character of functioning of
each separate part of the foundation. First of all, it is important to elucidate the role
of the widening and the cylindrical footing of the foundation in the load transfer
from the building to the soil, i.e. to determine the quantitative distribution of load
between the main structural parts of the foundation.
Field tests have enabled the qualitative pattern of interaction of the main parts of
foundation with the base to be studied, the soil deformation character and the con-
ditions of formation of condensed zones (kernels) in it under the widening bottom
and the cylindrical footing bottom to be elucidated, the stress distribution character
on the contact of the bottom and the soil to be determined, quantitative relation-
ship between the loads on the widening and the footing to be found. However, the
experiments on the elucidation of the specific features of the condensed kernel for-
mation and the displacement distribution in the foundation model base give only the
qualitative description of the effect.
Quantitative estimations of the experimentally observed facts of the effect of the
widening bottom shape on the base deformation character are required. In particular,
it has been found that at equal displacements in the widening bottom boundary areas,
the shear areas are less developed for a more convex bottom. Besides, with the
increase of the convexity degree the development of the vertical deformation area
decreases. Therefore, data regarding the limiting diameter value for the foundations
with a convex bottom, are required. In case this limiting value being exceeded, the
relative bearing capacity (i.e. the ratio of the bearing capacity to the foundation
concrete volume) of the foundation decreases.
Finally, quantitative estimations of the available test data for the bored pile foun-
dations (of various diameter, with various widening depth, with footing excluded)
at various soil conditions are required, that will enable the distribution of the overall
load among the foundation elements to be specified.
The currently available calculation methods and schemes do not fully comply
with the specific features of bored pile foundation structures and do not take into
account the functioning of the footing together with the widening as well as the
geometrical characteristics of the widening and a number of other factors. There-
fore, a further development of the calculation method is required with the account
[email protected]
of a spatial formulation, suitable for different bored pile foundation types and sizes
with different geometry of the widening and its different location with respect to the
cylindrical footing.
442 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

A simplified method for the bored pile foundation calculation (only for the verti-
cal load), proposed in [227], is based on an assumption of the whole structure func-
tioning being divided into the functioning of two independent elements – a widen-
ing and a footing, interaction between them being formally excluded. A vertical
load N, applied to the foundation, is divided into the parts λN and (1–λ)N, accepted
by the widening (λN), considered as a ring, and by the footing, (1–λ)N. A coeffi-
cient of participation of the ring-shape widening λ is introduced, being taken into
account in the calculations in accordance with the values of the widening diameter
and the footing height, according to the Table 6 of [227]. The calculation method
is based on the choice of such value of the λ parameter, for which the settlements
of the ring-shaped widening S1 and the cylindrical footing S2 , rigidly assembled
into a unique solid structure, are equal. The settlement values S1 and S2 are found
in a rather approximate way using a known layer summation method. It is a rather
labour-consuming task to provide the main condition S1 = S2 to be fulfilled with an
accuracy of 5%. It is performed in several stages using a plenty of tabulated data.
For the determination of stress in the base of a bored pile foundation, a theoretical
solution by Egorov [34] under a flat punch with a ring-shaped bottom is used. The
roles of the conical and spherical parts of the widening are not being taken into
account. Besides, for the sake of simplification of the calculations, the contact stress
profile over the ring-shaped widening bottom and the nominal shaft toe footing is
taken rectangular.
Experimental studies of functioning of an off-centre loaded bored pile foundation
with a widening were performed in [211, 212], where the recommendation regard-
ing their calculation were also worked out. The effect of eccentricity on the average
settlement, slope, horizontal displacements, rotation centre position, etc. was found.
The proposed method for the calculation of bored pile foundations with the account
of the resistance of the soil above the bottom, is based on the method for the calcu-
lation of deep massive foundations [244]. The Fuss-Winkler base model is taken as
the calculation model. Note that usage of the coefficient of subgrade reaction for the
calculation of piles only for the horizontal load was earlier proposed by Tsymbal
[222].
Spatial functioning of the foundation was considered by Sorochan and Gruodis
[211, 212] only in the vertical plane of its symmetry under a horizontal force and a
moment. The effect of the vertical force is taken into account only for the determi-
nation of pressures under the footing of the foundation widening.
The crucial point of the calculation method, proposed in [211, 212], to a great
extent determining the reliability of the calculations, is the choice of the coefficient
of subgrade reaction C variation with depth. Horizontal coefficients of subgrade
reaction are calculated on the base of the experimental data and essentially depend
on the geometrical shape of the support widening. For comparison of the calculation
and the experimental data, two laws of the C variation with depth were used:
linear Cz = C · hzf ,
[email protected] 
nonlinear Cz = C · hzf
5.5 Spatial Contact Problem for a Bored Pile Foundation with a Widening 443

where hf is the footing bottom depth, C=k·C0 is the non-uniform compression coef-
ficient (a characteristic of the base rigidity), kis a coefficient, taking into account the
relationship between the uniform and non-uniform compression coefficients (for
round foundations k=2.68), p0 is the average additional pressure under the widen-
ing bottom, W is the foundation settlement, determined from the Construction Rules
and Regulations as for a nominal foundation on a natural base.
The application of both hypotheses gives a rather high discrepancy with the
experimental results of the slope determination (for the linear law – by factor of up
to 2.5, for the nonlinear law – by factor of 1.5). One can just make the calculations
according to the Construction Rules and Regulations more specified as well as trace
a general trend of the increase of slopes, settlements, and horizontal displacements
at a fixed average pressure under the bored pile foundation widening bottom.
The method for the calculation of bored pile foundations with widenings, elabo-
rated in [211, 212], due to its simplicity and visuality, was included into a designer’s
handbook [215]. However, one should take into account that it enables the calcula-
tions to be preformed according to the second group of limiting states with insuffi-
cient accuracy and only with a great amount of field test data being available.
Thus, the complicated character of functioning of bored pile foundations with
widenings at the interaction with the soil base requires additional detailed studies
of the soil reactive pressure over the lateral surface based on the formulation and
solution of the spatial contact problem (as a rule, bored pile foundations transfer
a spatial force system to the base, this force system being in the general case a
combination of a vertical, a horizontal, and a momental load).
Since bored pile foundations with support widenings, having rather small relative
depth, are characterized by high rigidity, for them a calculation scheme in the form
of a volumetric deepened punch seems natural to be applied. The base model will be
treated as a linearly deformable half-space, weakened by a cavity, whose boundaries
are in full correspondence with the contact surface of the foundation structure to be
calculated.
Below the examples of numerical boundary-element calculation of bored pile
foundations with spheroconical widenings under a vertical and an inclined load are
considered.

5.5.3 Calculation of Deformations of the Base of a Bored Pile

Foundation with a Spheroconical Widening Under a Central
Loading (Axisymmetric Contact Problem)
Consider a bored pile foundation in the form of a deepened punch with a sphero-
conical widening, fixed in an elastic half-space. In the simplest case it is assumed
to be loaded only by a vertical force along the symmetry axis. Due to the axial
symmetry, the deepened absolutely rigid punch in the shape of a rotation body with
[email protected]
a spherical and a conical widening will be displaced only vertically. The stressed
state of the elastic half-space (the soil base) is characterized by a radial and a tan-
gential component of the contact stress vector (tangential stress is zero due to the
444 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

axial symmetry) and depends only on the vertical coordinate. The system of inte-
gral equations, written on the base of the reciprocality theorem, including the punch
equilibrium equations, is considered above in Sect. 2.4.1.
For the numerical solution of the system of Eqs. (2.30) and (2.34), the contact
surface of the foundation and the soil is partitioned into meridional zones, each of
them being divided by horizontal planes into triangular and quadrangular bound-
ary elements. On each boundary element the contact stress functions are averaged.
The dimensionality of the boundary-element problem is reduced and equals (2M+1)
(i) (i)
with respect to the values pr , pz (i=1, 2,. . .M) and z where M is the total num-
ber of the boundary elements along the generatrix of the bored pile foundation.
The averaged contact force values are expressed only in terms of the punch rigid
axial displacement. As noted above in Sect. 2.4.1, in the cylindrical coordinate sys-
tem the problem is essentially simplified and becomes in fact a one-dimensional
one.
In the practical calculations, the generatrix of the bored pile foundation was parti-
tioned by horizontal planes into 25, and along the angular coordinate – into 16 equal
partitions. As above, regular integrals were computed using the cubature formulae
of the highest accuracy degree, singular integrals were calculated semi-analytically
with selection of the singularity. The calculation scheme for the general case and
the boundary-element grid, used for the numerical solution, are shown in Figs. 5.19
and 5.20.
Figure 5.21 presents the results of the contact problem solution in the form of
normal and tangential contact stress profiles for a meridional cross-section of a
short bored pile foundation (h/2a=3) in the fractions of the contact pressure aver-
age value σav = N/Fwhere F is the contact surface area, N is the vertical load on the
foundation. The Poisson ratio value was taken ν=0.3. The widening dimensions are
determined by the ratios b/a=2.5, rf /a=3, its depth hf /a=2.5, 4.0. A specific fea-
ture of the normal stress σ profile should be noted, consisting, as one should expect,

Fig. 5.19 Calculation

[email protected]
scheme for a bored pile
foundation with a support
widening under an inclined
load
5.5 Spatial Contact Problem for a Bored Pile Foundation with a Widening 445

Fig. 5.20 Discretization of

the contact surface of the
foundation and the soil using
boundary elements

in the presence of tensile stresses in the conical part of the deepened punch lateral
surface what is related to the lack of account of unilateral constraints in the contact
domain.
The calculation results for the bored pile settlements, using the boundary-element
method (requiring only the deformational characteristics of the soil E and ν to be
used), are listed in Table 5.5. The calculations were performed both with and without
the account of unilateral constraints.
The geometrical dimensions of the bored pile foundation with a support widening
and its loading conditions are presented in Fig. 5.22. The following physical and
mechanical characteristics of the soil were taken (sand-clays and clayeys of low-
plasticity and semihard consistency):

deformation modulus E=1270 t/m2 ,

Poisson ratio ν=0.3,
density =1.7 t/m3 ,
internal friction angle δ=16◦ ,
cohesion C=2.9 t/m2 .

As follows from Table 5.5, an increase of the contact surface discretization degree
above 400 elements by each 100 elements results in the increase of the numerical
solution accuracy by less than 1%. Assuming the calculation results to be linearly
dependent of the value n–1 (n being the number of the boundary elements), one can
easily find the limiting values of the foundation settlement: S=3.829 cm without

Table 5.5 Settlement (cm) of a bored pile foundation with a support widening, calculated by the
boundary-element method

Number of boundary elements

[email protected] n = 400 n = 500 n = 600

Without the account of unilateral constraints 3.881 3.870 3.864

With the account of unilateral constraints 4.127 4.113 4.105
446 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

(a)

(b)

[email protected]
Fig. 5.21 Profiles of dimensionless contact normal (σ ) and tangential (τ ) stresses in a meridional
cross-section of bored pile foundations with widenings. (a) hf /a = 2.5, (b) hf /a =4.0
5.5 Spatial Contact Problem for a Bored Pile Foundation with a Widening 447

Fig. 5.22 Scheme for the

example of calculation of a
bored pile foundation with a
support widening for a
vertical load

the account of unilateral constraints and S=4.060 cm with the account of unilateral
constraints. Thus, if unilateral constraints are not taken into account, under vertical
loading, an error of up to 6% in the determination of the settlement value can arise.
According to the method of calculation for the same foundation, quoted in [237],
in assumption of the building being without a basement, at first the condition of
applicability of the deformation-based calculation is checked: Pav = N/F≤R. Since
the average pressure over the foundation surface (in the widening plane) Pav =
43 t/m2 , and the calculated pressure on the base without the account of the cylindri-
cal footing (included into the safety factor), obtained on the base of the Puzyrevsky
solution (Construction Rules and Regulations 11–15–74) R=43.5 t/m2 , then the
check is performed. Then, using the method and the tables, quoted in [227], one
can show the required equality of the nominal settlements S = S1 = S2 to be pro-
vided if the coefficient of participation of the widening in the foundation functioning
λ=0.65. The distribution of load N = NB + NF between the ring-shaped widening
and the footing is the following:
ring-shaped widening NK = λN = 137t,
footing NF = (1 − λ) N = 73 t.
Finally, the calculation of the foundation settlement [227] gives a final result

S = S1 ≈ S2 = 2.94 cm.

Thus, an approximate calculation, performed according to the recommendations

for the design of bored pile foundations [227], has a discrepancy with our data,
[email protected]
obtained by the boundary-element method, by 23–27%. Such discrepancy, evi-
dently, is a consequence of a rather rough account of the widening geometry in
[227], of unilateral constraints being not taken into account as well as of using a
448 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

Fig. 5.23 Relative

settlements of bored pile
foundations of equal volume
at a fixed vertical load: (1)
foundation with a support
widening, without the
account of unilateral
constraints; (2) foundation
with a support widening,
without the account of
unilateral constraints; (3)
foundation without a support
widening, with a given
height; (4) foundation
without a widening, with a
given diameter

rather approximate tabulated data for references. Note that in the case of an inclined
load, the calculation according to [227] cannot be performed at all.
Figure 5.23 shows the results of boundary-element calculations, characterizing
the effect of the widening and its depth on the bored pile foundation settlements
under a vertical load. Relative settlements of the foundation with a widening are
compared to the settlements of cylindrical bored piles of the same volume as the
initial foundation with a widening, but of the same length (then, of a larger diam-
eter) or of the same diameter (then, of a larger length). The calculations are per-
formed in two variants – with and without the account of unilateral constraints on the
contact between the foundation and the soil and the values P/Ea2 = 1, ν=0.3. The
geometrical dimensions of the bored pile foundation are the following: b/a=2.5,
h1 /a=1, h2 /a=1.5, rf /a=3. The vertical load on the foundation was fixed. As one
could expect, the settlements of the bored pile foundation with a widening are prac-
tically always smaller than for cylindrical piles. Besides, with the increase of the
relative depth of the support widening hf /a the settlements decrease, varying prac-
tically linearly. Meanwhile, the account of unilateral constraints results in a consid-
erable increase of settlements what is related to the decrease of the contact domain
between the foundation and the soil.

5.5.4 Calculation of Displacements and Slopes of a Bored Pile

Foundation Under an Inclined Load

In our calculations, performed using the direct boundary-element method in a full

formulation, the boundary-element grid on the punch surface was generated accord-
[email protected]
ing to the same principle as for the problem in an axisymmetric formulation. The
total number of the boundary elements did not exceed 400. Surface integrals were
computed using a proposed numerical-and-analytical method. If the dimensionality
5.5 Spatial Contact Problem for a Bored Pile Foundation with a Widening 449

of the linear algebraic equation system in the axisymmetric formulation was n=51,
in the full formulation it was n=1206. Note that the dimensionality of the finite-
measure analog of the problem can be reduced in the presence of symmetry planes
in the foundation structure loading scheme.
Model experiments have shown that for the practical purposes of determination
of such integrated characteristics as displacements and slopes, the number of
boundary elements can be reduced almost twice without a loss of accuracy. On
the contrary, for the studies of the stress-strained state of the soil in the foundation
active zone one should apply a dense grid of boundary elements, corresponding to
the contact stress condensation degree. A developed Rostwerk software enables the
contact surface to be automatically partitioned into boundary elements, using the
given shape of the bored pile foundation axial cross-section. Axial cross-sections
of bored pile foundations with widenings of the most frequently used types are
shown in Fig. 5.24. The same figure shows the schemes of discretization of the
near-pile space, required for the postprocessor processing of the contact problem
solution results (construction of stress isolines, deformations and their invariants,
displacements, etc.).
The calculated dependences of settlements, horizontal displacements, and slopes
of a bored pile foundation with a support widening of a spheroconical shape under
an inclined load are presented in Figs. 5.25, 5.26, and 5.27.
An inclined force of a constant value was supposed to be applied in the centre
of the top edge of the foundation, its direction being determined by the angle ϕ of
its deviation from the horizontal plane. Due to the calculation scheme symmetry
with respect to the vertical axis of the foundation, the calculations were performed

(a) (b) (c)

[email protected]
Fig. 5.24 Discretization of the active area of bored pile foundations with support widenings of
various type: (a) toroidal, (b) hemispherical, (c) spheroconical
450 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

(a) (b)

(c)

Fig. 5.25 Relative settlements of a bored pile foundation with a support widening versus the
inclined load direction (1) without and (2) with the account of unilateral constraints; hf /a=2.5 (a);
3.25 (b); 4.0 (c)

for the external force inclination angles in the range 90◦ ≤ϕ≤180◦ . For plotting the
calculations were carried out mostly with a step of 6◦ . Within the ranges of non-
monotonous variation of displacements and slopes the step value was decreased
down to 3◦ , 1◦ , or even 0.5◦ .
The obtained dependences of the contact interaction characteristics without the
account of unilateral constraints are rather monotonous (Figs. 5.25, 5.26, and 5.27,
lines 1). With the increase of the angle of inclination of the force P the horizon-
tal component of the inclined load vector P·sinϕ increases and simultaneously its
[email protected]
vertical component P·cosϕ decreases. This results in the decrease and increase of
the vertical and horizontal displacements, respectively. Simultaneously, the slope
ψx increases, reaching its maximal value under horizontal loading.
5.5 Spatial Contact Problem for a Bored Pile Foundation with a Widening 451

(a) (b)

(c)

Fig. 5.26 Horizontal displacements of a bored pile foundation with a support widening versus the
inclined load direction (1) without and (2) with the account of unilateral constraints; hf /a=2.5 (a);
3.25 (b); 4.0 (c)

The account of unilateral constraints in the contact domain results in quantitative

and qualitative changes (Figs. 5.25, 5.26, and 5.27, lines 2). This is related to the
decrease of the contact area due to the formation of areas with tensile stress on the
foundation surface. Since the soil does not work in tension, in the corresponding
areas gaps are formed between the foundation and the soil. As shown by numer-
ical experiments, at vertical loading, the areas with negative stress emerge on the
conical part of the widening, expanding with the absolute load value Pas well as
with the parameter of the relative depth of the widening location hf /a. A similar
[email protected]
situation is also observed in the lower part of the cylindrical footing, in some sense
resembling the "wedging" effect, at which a tension area is formed in the lower
part of a cylindrical surface as well as a gap between the foundation and the soil.
452 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

(a) (b)

(c)

Fig. 5.27 Slopes of a bored pile foundation with a support widening versus the inclined load
direction (1) without and (2) with the account of unilateral constraints; hf /a=2.5 (a); 3.25 (b);
4.0 (c)

Note that the "wedging" effect is also observed in the case of a cylindrical pile with-
out a widening; however, it is revealed even more strongly. Under an inclined load,
the formation and development of tension areas is irregular and can rather hardly
be traced due to an essentially non-uniform stress-strained state in the foundation
active area. Finally this process is reflected in the values of the integrated charac-
teristics: displacements and rotation of the bored pile foundation as a rigid solid.
We proceed with a more detailed description of dependence of these values on the
inclined force direction.
[email protected]
For all external load inclination angles, the account of the foundation uplifting
from the soil always results in an increase of absolute values of horizontal displace-
ments Δz and slopes ψ x .
5.5 Spatial Contact Problem for a Bored Pile Foundation with a Widening 453

The influence of the "wedging" effect and underpressure in the conical part
of the widening affects the increase of the bored pile foundation settlements at
90◦ ≤ϕ≤120◦ , i.e. in this range of variation of the resultant force inclination angle,
the Δz values with the account of unilateral constraints exceed the corresponding
values without the account of unilateral constraints. Depending on the ratio of the
widening size and its depth, at a certain value of ϕ from the range 125◦ ≤ϕ≤135◦
both settlement curves intercept (Figs. 5.25a, 5.26, and 5.27a). Hence, there exists
such resultant direction, for which the account of unilateral constraints results in the
settlement values equal to those for the contact problem solution in a linear formu-
lation (without the account of unilateral constraints).
At the further increase of the ϕ angle, the effect of the increase of the horizontal
component of the external load vector goes on. Due to the uplifting of the structure
from the soil, the slope further increases and the absolute values of the foundation
horizontal displacements increase (Figs. 5.25b, c, 5.26, and 5.27b, c). Simultane-
ously, the vertical displacements Δz decrease (Figs. 5.25a, 5.26, and 5.27a).
As follows from the calculation data (Figs. 5.25, 5.26, and 5.27), there is a thresh-
old value of the angle ϕ ∗ , at which a qualitative change of the contact deformation
pattern occurs. The account of unilateral constraints results in the foundation slope
and the absolute values of the foundation horizontal displacements increase up to
the ϕ∗ value increasing strictly linearly along the tangents, built at ϕ=90◦ (vertical
loading) to the corresponding plots, obtained without the account of unilateral con-
straints. When the threshold value of the inclined force ϕ ∗ is reached, the calculated
curves become non-monotonous. After a certain decrease of the slope and horizon-
tal displacement values and a settlement increase, these characteristics continue to
vary nonlinearly. The threshold value ϕ∗ is most essentially affected by the depth hf
of the support widening location. In particular, for the series of calculation we have
performed, the influence of the depth on the ϕ∗ value can be seen from Table 5.6.

Table 5.6 Values of φ ∗ at different depth hf of the location of the support widening

hf /a 2.5 3.25 4.0

φ∗ 152◦ 138◦ 132◦

Hence a practically important result is obtained: in order to predict slope and hor-
izontal displacements of a foundation within the range of the inclined force angle
90◦ ≤ϕ≤ϕ ∗ , it is sufficient to have the results of only one calculation of these char-
acteristics.
At a near-horizontal load (ϕ≥175◦ ), at the foundation head vertical displace-
ments are practically zero, and the further increase of the load inclination value
even leads to the foundation uplifting (Δz < 0). In other words, at a horizontal (or a
near-horizontal) loading of the bored pile foundation its uplifting towards the base
[email protected]
surface occurs. Note that in case unilateral constraints not being taken into account,
this phenomenon is not observed, i.e. always Δz ≥0 and the bored pile foundation is
not uplifted.
454 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

Thus, from the performed analysis of deformations of bases of bored pile foun-
dations with support widenings under a complex spatial loading one can conclude
that the proposed numerical boundary-element approach enables one to take into
account the non-uniform stress-strained state of the base and formation of cavi-
ties between the soil and the widening in the course of the foundation displacement.
This approach is effective for the practical calculation of bored pile foundations with
various shapes of support widenings, without appealing to various regulating doc-
uments, where the specific features of bored pile foundation functioning are taken
into account rather roughly.

5.6 Calculation of Contact Interaction of Bases with Slotted

Foundations of Industrial and Civil Buildings

The experience of design and construction of objects of industrial and civil engi-
neering in Russia has shown [214] that in cohesive and low-watered soils of nat-
ural (as a rule, eluvial) origin, application of shallow monolithic foundations with
working lateral surface is justified. The most important feature of such foundations
is an effective technology of their construction, where the back fill is excluded.
This enables lateral friction over the walls to be actualized, that cannot be achieved
at the construction of traditional foundations in open pits. It is also advisable to
apply shallow (up to 6–8 m) foundations not only due to the possibility to use com-
mon excavating machines, but also because in non-watered soils one can refuse
from making trenches under the protection of clay mixture. The most efficient, in
comparison with different traditional (pile, strip, pier) structures of shallow mono-
lithic foundations with lateral working surface, are slotted foundations applied for
both industrial and civil engineering [136]. The role of slotted foundations increases
for low-rise engineering of residential buildings (cottages) and light skeleton-type
industrial buildings [21, 137, 211] in view of modern environmental requirements
to the foundation engineering, construction cost reduction, high bearing capacity,
simplification of the construction procedure.

5.6.1 Slotted Foundations of Various Structural Shapes

Slotted foundations are a sort of an engineering solution of foundations, built by
the diaphragm wall technique [47, 206]. They are made as a narrow wall of con-
crete or reinforced concrete in a soil base. Structures with a raftless joint of a
window-sill type or with a pile raft for the load transfer from the above-ground
structures to the walls are possible (Fig. 5.28). The load is transferred to the base
by lateral planes, wall ends, and the pile raft plate bottom. Therefore, the geomet-
[email protected]
rical shape of the slot, the pile raft, and the lateral surface is determined depending
on the above-foundation structure size, load direction and value, geological con-
ditions, depth of the soil freezing or weathering. In the recent years, based on the
5.6 Calculation of Contact Interaction of Bases 455

Fig. 5.28 Calculation (a) (b)

schemes for slotted
foundations: (a) under a
column of reinforced
concrete, (b) under a
three-hinge frame

experience of construction of foundations using the diaphragm wall technique [206],

Russian researchers and engineers have proposed and worked out new structures and
technologies of construction of slotted foundations, being successfully applied for
industrial and civil buildings [136, 214]: foundations with cavities, two-slot founda-
tions with a low pile cap, wedge-slotted foundations with monolithic plates, slotted
foundations with a widened footing, etc. (Fig. 5.29).

(a) (b)

(c) (d)

(e) (f)

Fig. 5.29 General view of

slotted foundation of various
structure: (a) with cavities;
[email protected]
(b) double-slotted; (c, d)
cross-shaped; (e) with a pile
raft; (f) with a bottom
widening
456 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

The technology of slotted foundation construction is highly mechanized. After

the site grading, slots are made in the soil using earth-moving excavator-based
machines with a special working equipment of backhoe, narrow grab bucket, or
rotor type. Narrow slots are often produced using a bar or a slot-profiling machine.
The residual sludge in the face is removed or compressed by impression or tamping.
Then reinforcing cages are installed and fixed in the slotted pit. While producing
the foundation, concrete is supplied directly into the foundation body. Vibrating
tampers are used for the concrete compaction. The concrete should have sufficiently
semifluid or pourable consistency. A subsequent concrete compaction by a platform
vibrator is useful. After the concreting being finished, maintenance of the placed
concrete is performed during the whole period of its strengthening.
Slotted foundations have the advantages of both shallow and pile foundations.
It is shown in [136, 214] that structural solutions of the slotted foundations as well
as their construction technology enable a number of shortcomings of traditional
foundation types to be avoided. In particular, they do not require deep pits under
the whole building to be made, the amount of formwork is strongly reduced, they
can be constructed at different depth without a surcharge of materials. The slotted
foundation structures having been proposed, enable technical and economic char-
acteristics of these foundations as well as their competitiveness to be considerably
increased in comparison with the known solutions. The experience of application of
slotted foundations for the construction of a number of industrial and civil objects
at various soil conditions has shown [137, 214] that, in comparison with the founda-
tions, produced in a pit or using the displacement piles, they enable the excavation
work volume to be reduced by 40–50% (in the case of specific structural solutions –
up to 70–80%), concrete consumption – by 5–6%, reinforcement consumption – by
15–20%, and formwork – by 70–100%.
In spite of the efficiency of the foundation structures under consideration, they
have not found proper wide application in engineering. One of the essential restric-
tive reasons is a low level of reliability of practically applied calculation methods,
based on separate empirical data and simplest calculation schemes [158, 159]. Until
now, no special methods for calculation of slotted foundations have been created,
which would take into account their structural features, friction resistance over the
developed lateral surface, and various combination of active forces and moments.
The calculations are performed similarly to the known methods for nominal and
pile foundations. A simplified elastic model of the soil medium based on the Win-
kler hypothesis is used. Calculation of slotted foundations for off-centre loads is
performed with the account of their fixation in the soil according to the scheme of a
fixed absolutely rigid rod [159].
It should be noted that for the calculation of the settlement of the base of a rigid
slotted foundation, application of the calculation scheme of a nominal foundation,
developed for pile foundations, is not quite correct. For shallow slotted foundations
one should take into account the effect of the lateral surface geometry on the foun-
[email protected]
dation displacements. As a rule, slotted foundations transfer to the base a system of
forces, represented, in the general case, by a vertical, a horizontal, and a momental
load. According to the design rules, adopted in Russia [115, 215], the stress-strained
5.6 Calculation of Contact Interaction of Bases 457

state of the base of a slotted foundation should be studied separately: for the vertical
load, using the model of a linearly deformable half-space, and for the horizontal and
momental loads, invoking the Winkler model. Dependence of the foundation calcu-
lation scheme and the base model on the external load direction is also a reason for
the low level of reliability of determination of displacements and slopes of slotted
foundation structures at the stage of their operation.
Thus, at efficient and rather well elaborated construction technology at various
soil conditions, the methods for calculation of settlements and slopes of slotted
foundations require further improvement on the base of theoretical studies with the
account of spatial formulations and detailed analysis of the processes of contact
interaction of the foundation with the soil.

5.6.2 Calculation of Slotted Foundations Based on the Base

Deformation
Slotted foundations are characterized by high rigidity, hence we assume a calcu-
lation scheme in the form of a volumetric punch, deepened in a soil base, treated
as a linearly deformable half-space, weakened by a cavity whose boundaries are
identical to the contact surface of the foundation structure to be calculated.
The developed Rostwerk software is a rather universal one, enabling numerical
experiments to be performed in order to evaluate the effect of various factors on the
slotted foundation displacements and the contact stress distribution over the lateral
and end surface and bottom. The input data preparation is simple, totally automated,
and suitable for discretization of the contact surface and active area of slotted foun-
dations of the known types. Typical examples are shown in Figs. 5.30 and 5.31.
Subsequent sections of the chapter are devoted to the systematic studies of the
processes of contact interaction of slotted foundations with soil bases. At the calcu-
lations, the load, acting on the slotted foundation, is considered known and given.
The effect of the geometrical shape of the foundation on its displacements and on
the stress-strained state of the base is evaluated. The influence of the inclination
angle and eccentricity of the resultant external load on the slotted foundation dis-
placement is investigated. The results of the settlement and slope calculation are
compared with the data of the field experiments as well as with the existing approx-
imate calculation methods. The analysis of this comparison has enabled the accuracy
of the approximate engineering methods to be estimated and the admissible range
of their application to be specified.
Effect of the degree of discretization of the contact surface of the foundation
with the soil. For numerical experiments a separate (or, using the terms of [136],
pier) slotted foundation with a bottom of a rectangular shape was considered (this
[email protected]
the simplest solution for a foundation under ferroconcrete and steel columns).
The foundation dimensions as well as the chosen coordinate system are shown in
Fig. 5.32a, b. The soil base had the following deformational characteristics: defor-
mation modulus E=20 MPa, Poisson ratio ν=0.4 what corresponds to firm-structure
458 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

(a) (b)

(c)

Fig. 5.30 Discretization of the contact surface of interaction with the soil for slotted foundations
of various types by a boundary-element grid: (a) pier foundation, (b) wedge-slotted foundation
with a widening, (c) pile raft foundation

(a) (b)

(c)

Fig. 5.31 Discretization of the active area of the base for slotted foundations: (a) with a pile raft,
(b) with a widening, (c) with an inclined bottom

low-moisture eluvial clayeys in a hard state. The foundation was subject to an essen-
tially spatial load by an inclined force P=103 kN which had an angle of 45◦ with
the vertical axis, acting in a plane, passing through the foundation centre and per-
pendicular to the slot plane (Fig. 5.32b). The results of calculations with sequential
condensation of the calculation grid at different discretization degree on the width
(n1 ), length (n2 ), and depth (n3 ) of the foundation block are given in Table 5.7.
Settlements z , displacements x , and slopes ψ x were calculated (Table 5.7).
For comparative evaluation and search for optimal (from the point of view of econ-
[email protected]
omy and accuracy) results, this table also contains the total number of boundary
elements K. Using the data of Table 5.7 and assuming the dependence of the numer-
ical solutions on K–1 to be linear, one can readily obtain the extrapolated values  ¯ z,
5.6 Calculation of Contact Interaction of Bases 459

(a) (b)

Fig. 5.32 Calculation schemes for a slotted foundation with a rectangular longitudinal cross-
section under an inclined force: (a) within the slot plane, (b) orthogonally to the slot plane

Table 5.7 Convergence of numerical solutions for a slotted foundation with a rectangular bottom
under an inclined load

n1 n2 n3 K –x ·102 ,m z ·103 ,m –ψ y ·103 ,rad

3 7 7 161 1.12439 7.35481 3.64699

3 8 8 200 1.12233 7.34523 3.63504
3 9 9 243 1.12079 7.33774 3.62609
3 10 10 290 1.11959 7.33171 3.61910
3 10 11 316 1.11892 7.32869 3.61489
3 10 12 342 1.11839 7.32616 3.61148
3 10 13 368 1.11795 7.32406 3.60863
3 10 14 394 1.11759 7.32227 3.60626
3 11 14 425 1.11726 7.32026 3.60467
3 12 14 456 1.11698 7.31859 3.60337
2 14 14 476 1.11828 7.32487 3.61159
3 13 14 487 1.11674 7.31719 3.60226
1 15 15 495 1.12108 7.34259 3.62836
2 15 14 506 1.11802 7.32352 3.61044
2 14 15 508 1.11793 7.32308 3.60911
Extrapolated values 1.11461 7.30782 3.58971

x , ψ y at K→∞. These values are also given in Table 5.7, they enable the accuracy
of the approximate solution to be estimated.
As follows from the calculations performed, the approximate numerical solu-
tions, obtained by the boundary-element method, rather soon stabilize with the
increase of the degree of discretization of the contact surface of the foundation and
the soil. The obtained data can serve to optimize the discretization degree for further
series of calculations. As one can see, the boundary-element grid (n1 ) × (n2 ) × (n3 )
= 3 × 10 × 14 with the total number of elements K=394 is suitable both in view
of the computation time and for obtaining the approximate values of the contact
deformation parameters with a high accuracy (0.3% regarding the displacements
[email protected]
and 0.5% regarding the slopes).
The calculations have also shown that a decrease of the discretization degree over
the end surface of the slotted foundation (n1 < 3) results in a decrease of the accuracy
460 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

of the numerical results in spite of the increase of the number of boundary element
on the lateral surface. Therefore, we have taken n1 = 3 as the minimal discretization
degree over the foundation end surface. Note that while calculating such integrated
characteristics of the contact interaction as displacements and slopes, one can prac-
tically without any loss in accuracy restrict oneself by the discretization degree over
the foundation lateral surface of the order of (n2 ) × (n3 ) = 7 × 7. However, this is
obviously insufficient for the studies of the two-dimensional contact stress field and
three-dimensional stress-strained state in the foundation active area.
Settlements and slopes under a complicated spatial loading. Calculation of slot-
ted foundations based on the base deformations is performed for the main load com-
binations. Settlements and slopes of a separate slotted foundation, as of any other
foundation structure, depend on its shape and size, mechanical properties of the soil
base, conditions at the surface of the contact of the foundation and the base, loading
type, influence of the neighbouring foundations and loads on adjacent areas, and a
number of other factors [101, 201] (the rigidity of the foundation itself, its depth,
size of plastic areas beneath it, depth of the compressed base, character and rate
of loading, conditions of work). Account of all these factors makes the problem of
calculation of a slotted foundation practically horizonless.
For deepened foundations of a block type, to which the slotted foundations
belong, the most important characteristics of the base deformations under a spatial
load system of the most general type will be the dependences of displacements ver-
sus size, shape and conditions on the contact of the foundation with the soil. Since
the slotted foundations are spatial structures with finite and different size in three
coordinate directions, and the above-foundation structures transfer to them vertical,
horizontal and momental loads, numerical studies of contact interaction of slotted
foundations for the following typical cases of spatial loading were performed:
– action of an inclined force in the plane of the foundation longitudinal cross-
section, applied in the centre of its top edge (Fig. 5.32a),
– action of an inclined force, applied in the centre of the foundation top edge in the
plane, orthogonal to the longitudinal cross-section of the foundation (Fig. 5.32b),
– vertical force, applied with an eccentricity y in the longitudinal cross-section of
the foundation.
The geometrical shape of the slotted foundation and the surrounding soil charac-
teristics were the same as considered above.
The dependences of the slotted foundation displacements and slopes for the load-
ing types under consideration are presented in Figs. 5.33, 5.34, and 5.35.
The calculation data, shown in Fig. 5.33, enable one to estimate the soil resis-
tance due to the development of tangential and normal forces developing mostly on
the lateral surface and partially on the bottom and end faces of the slotted founda-
tion, located parallelly or orthogonally to the acting force planes, respectively (at
different inclination angles). For the slotted foundation under consideration with a
[email protected]
relative depth H/b= 3.5 m/2.6 m = 0.74, the action of inclined forces in orthogonal
planes results at each inclination angle of the resultant α(β) to different horizon-
tal displacements and slopes of the slotted foundation. As one should expect, the
5.6 Calculation of Contact Interaction of Bases 461

Fig. 5.33 Displacements Δ (m) and slopes ψ(rad) of a slotted foundation under a longitudinal
(solid line) and transverse (dashed line) loading by an inclined force

Fig. 5.34 Displacements Δ (m) and slopes ψ (rad) of a slotted foundation versus the longitudinal
eccentricity of a vertical load with (dashed lines) and without (solid lines) the account of unilateral
constraints

biggest difference in these values is observed in the absence of the vertical compo-
nent of the inclined force (i.e. at horizontal loading, α(β) 180◦ ) and reaches 20%
for the displacements and 40% for the slopes. As follows from the calculations, with
the relative depth increase the noted discrepancies in the displacements and slopes
are also revealed, but to a lower extent. Since at each value of the angle α(β) of the
[email protected]
external load inclination the vertical component does not change under longitudi-
nal and transverse loading, this does not lead to any changes of the settlements z ,
which decrease with α(β) according to the sine law (Fig. 5.33).
462 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

Fig. 5.35 Displacements Δ

(m) and slopes ψ(rad) of a
(a)
slotted foundation under (a)
longitudinal and (b)
transverse loading by an
inclined force with (dashed
lines) and without (solid
lines) the account of
unilateral constraints

(b)

Figure 5.34 shows the results of calculation of displacements and slopes under a
vertical force with the eccentricity 0 ≤ y ≤ b/2 = 1.75 m acting on the top edge of
the slotted foundation along the longitudinal axis. Since this loading is equivalent to
a central vertical force P and a moment M = –P· y , with the increase of  y the values
ψ x and Δy should increase proportionally, and the settlement Δz should remain
unchanged what is confirmed with high accuracy by the calculation results.
Figures 5.34 and 5.35 enable one to judge upon the account of unilateral con-
straints in the contact domain of the slotted foundation with the soil. As seen from
the calculations, due to the contact area decrease the horizontal displacements have
an additional development. The account of the unilateral constraints provides maxi-
mal effect at horizontal loading when the vertical load component is zero. Quantita-
tively the account of unilateral constraints at the longitudinal (Fig. 5.35a) and trans-
verse (Fig. 5.35b) loading is similar, not exceeding 10%. This is evidently explained
by the fact that the detachment area on one lateral side under the transverse loading
[email protected]
comparable with the total area of symmetrical detachment zones on two lateral
sides under the longitudinal loading.
A somewhat different situation is observed in the settlement character with
the account of unilateral constraints under the longitudinal and transverse loading
5.6 Calculation of Contact Interaction of Bases 463

(Fig. 5.35). At such loading types the account of unilateral constraints can result
both in an additional settlement increase and in their decrease. For the calcula-
tion example under consideration the change of the increment sign is observed for
ᾱ ≈ 135◦ under the longitudinal loading and for β̄ ≈ 160◦ at the transverse load-
ing. Below the quoted values of the inclined force action ᾱ(β̄) its vertical component
results in a "wedging" effect being noticeably revealed (See Sect. 5.4) and, hence,
the settlement increment is positive. The maximal relative value of this increment is
observed for purely vertical loading (90◦ ) and for both loading types does not exceed
7.5%. For α (β) > ᾱ(β̄) the increment of the settlement for the centre of the slotted
foundation top edge with the account of unilateral constraints is negative, i.e. with
the approach of the force direction to the horizontal one, the settlements decrease
in comparison with the case of total cohesion of the slotted foundation with the
soil. At almost horizontal loading, the account of unilateral constraints results in the
absolute values of the settlement increments even exceeding the settlements for the
calculation in the linear formulation. As a consequence, at α(β)≈180◦ the slotted
foundation has negative settlements what, for the chosen coordinate system, corre-
sponds to the slotted foundation uplifting. In other words, the slotted foundation is
squeezed out of the soil. Note that the maximal value of such vertical displacement
is quite small – near 0.5 mm for both longitudinal and transverse loading.
The account of unilateral constraints at loading by a vertical force and a moment
(Fig. 5.34) increases both the slopes and the displacements in the whole range of the
y variation. The maximal relative increase of ψx and y (by 8.2%) is achieved at
the maximal value of the moment. The maximal relative increase of the settlement
z , similarly to the above examples with an inclined force, is obtained at central
vertical loading, i.e. at y = 0.
The results for convergence of iteration processes for the cases of spatial loading
under consideration are shown in Fig. 5.36. The highest computation resources were
required by the calculation of the transverse loading of a slotted foundation (the
average iteration number for each load direction angle was Nav =10, Nmax =12 at
β=135◦ , Nmin =7, at β=180◦ ). Under the longitudinal loading, the iteration process
characteristics were the following: Nav = 6 at Nmax = 9, Nmin = 3, for the greater part
of the inclined force angle variation interval (α≥120◦ ) N= 4. Under an eccentric
load with the increase of y the number of iterations monotonously decreased from
Nmax =10 to Nmin = 4, and Nav = 7.
Thus, it follows from the analysis of the iteration process convergence data
that though the account of the structure nonlinearity in the calculation of slotted
foundations with a rectangular shape of longitudinal cross-section in this specific
case results in a correction of the contact interaction parameters, it still requires
the computation time increase by factor of 6–10. For the practical purposes such
amount of computational work for the calculation result correction within not
more than 10% is not always justified. Therefore, for the calculation of slotted
foundations according to the second limiting state at a complex spatial loading with
[email protected]
a relative error not more than 10%, the calculation can be performed without the
account of unilateral constraint and, what is important for practical purposes, the
superposition principle can be applied. With the increase of the absolute values of
464 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

Fig. 5.36 Scheme of the

iteration process at the (a)
account of unilateral
constraints for a slotted
foundation under: (a) an
inclined force within the
transverse cross-section
plane; (b) an inclined force
within the longitudinal
cross-section plane; (c) a
vertical force with a
longitudinal eccentricity
(b)

(c)

forces and moments applied to the slotted foundation from the above-foundation
buildings and structures, the structure nonlinearity will be revealed to a higher
extent and a reliable prediction of the slotted foundation displacements and slopes
will require mandatory account of unilateral constraints in the contact domain.
Effect of the shape of the slotted foundation longitudinal cross-section. Depend-
ing on the trench excavation procedure due to the application of various types of
earth-moving machines [136], the longitudinal cross-section of the slotted founda-
tions can be rectangular, rectangular with a curved bottom, or curved. Since the
slotted foundations are subjected to high vertical and horizontal loads from the
above-foundation structures, the quality of the trench excavation and its geometrical
[email protected]
shape in the longitudinal cross-section essentially affect the technical and economic
parameters of the design solutions. As follows from the construction experience
[136], high requirements to the narrow trench excavation under slotted foundations
5.6 Calculation of Contact Interaction of Bases 465

are in general satisfied by using common dipper mechanisms with a rather high pro-
ductivity. Meanwhile, it is noted by the specialists that at short trench excavation
a constant check and correction of the geometrical size of the trench is required.
Therefore, the most promising is application of rod grab buckets with high produc-
tivity, capable of digging trenches of large depth (20–25 m), providing the trench
verticality and the required geometrical size in the longitudinal direction.
Hence, the spatial shape of the trench and, accordingly, of the slotted foundation,
is rather effectively checked in the course of excavation and production of the foun-
dation. However, at the design stage the calculation methods are used, operating, for
example, with the foundation depth parameters, its size, slot width. At the calcula-
tions of slotted foundations according to the scheme of an absolutely rigid rod, fixed
in the soil [136, 159] only the foundation bottom resistance moment and its size in
the plane, perpendicular to the load, are taken into account. Due to this inconsis-
tency, we have performed numerical studies in order to reveal the influence of the
shape of the slotted foundation longitudinal cross-section on its displacements and
slopes along each coordinate axis direction under all components of both force and
momental loads. In the calculations the following dimensions of the slotted founda-
tion were fixed: the foundation size (h× b= 0.6 × 3.5 m) and depth (H= 2.6 m).
The deformation parameters of the soil are taken the same as above (E= 20 MPa,
ν= 0.4).
The shape of the lateral surface for the case of curved longitudinal cross-section
was given using a power law (Fig. 5.37a)

2|y| m
z=H 1− . (5.5)
b

At such dependence, the increase of the m parameter results in the longitudinal

cross-section shape approaching the rectangular one (Fig. 5.37b).
The lateral surface of the slotted foundations of a rectangular shape with a curved
bottom was considered by a conjugation of a rectangle with the sides (2b) × (h–R)
and a semicircle with a radius R=b (Fig. 5.37c). The z-coordinate of the bottom
points was calculated as follows:

(a) (b) (c) (d)

[email protected]
Fig. 5.37 Longitudinal cross-sections of slotted foundations: (a) curved; (b) rectangular; (c, d)
rectangular with a curved bottom
466 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations


z=H−b+ b2 − y2 . (5.6)

Here one should note that the choice of formulae of the type of Eqs. (5.5) and
(5.6) is not crucial and they can be easily replaced for a specific practically applied
bottom contour according to the trench excavation procedure applied. It is also con-
venient to use the bottom contour in the form of, for example:

a circular arc with a diameter

 larger than the slotted foundation length,
a cosinusoid z = H − h 1 − cos πby (Fig. 5.37d),

a lower part of an ellipse z = H − h + hb b2 − y2 , etc.

Here b, H, h are the slotted foundation shape parameters in its longitudinal cross-
section.
In the cross-section of the slotted foundation of a rectangular shape the lower
corners were supposed to be rounded by arcs of a radius r (Fig. 5.37b). At r→0 the
slotted foundation has the shape of a rectangular paralellepiped. The examples of
dicretization of the lateral surface of the slotted foundations of the forms under con-
sideration using a program for automated construction of boundary-element grids
are shown in Fig. 5.38.
The results of calculation of the contact interaction parameters for slotted founda-
tions with the longitudinal cross-section of various shape under force and momental
loads with respect to three coordinate directions are presented in Tables 5.8 and 5.9.
For each of the slotted foundation structure types the upper row gives the absolute
values of slopes and displacements at Qx (Qy , Qz ) = 10 kN (Table 5.8) and Mx (My ,

(a) (b)

(c) (d)

[email protected]
Fig. 5.38 Discretization of the lateral surface of slotted foundations with longitudinal cross-
section of various type: (a) curved, m = 2; (b) curved, m = 5; (c) rectangular, (d) rectangular
with a curved bottom
 5.6

Table 5.8 Displacements and slopes of slotted foundations under force loads at different longitudinal cross-section shape ∗

Load Qx =103 kN Qy =103 kN Qz =103 kN

Displacements Δ, m and slopes ψ, rad Δx 102 Δz 104 ψ y 103 Δy 102 –Δz 104 −ψ x 103 Δz 102
Curved cross-section, m = 2; 1.71929 – 6.14466 1.59813 – 4.25920 1.25841
Slat = 16.05 m2 , V = 9.63 m3 1.91831 8.36191 7.25658 1.76358 6.30197 5.07059 1.30388
m = 4; 1.65872 – 5.67023 1.52879 – 3.88333 1.17008
Slat = 18.8 m2 , V = 11.28 m3 1.83744 7.18912 6.61700 1.72982 6.42275 4.75172 1.20826
m = 6; 1.63529 – 5.49618 1.49733 – 3.70465 1.13395
Slat = 20 m2 , 1.81236 6.75855 6.40984 1.69372 5.81801 4.49710 1.19123
V = 12 m3

[email protected]
m = 8; 1.62291 – 5.40697 1.47907 – 3.59981 1.11406
Slat = 20.7 m2 , 1.79198 6.34471 6.26931 1.67015 5.49375 4.34559 1.18714
Calculation of Contact Interaction of Bases

V = 12.42 m3
m = 12; 1.61015 – 5.31764 1.45861 – 3.48191 1.09271
Slat = 21.4 m2 , V = 12.84 m3 1.77479 6.03288 6.14541 1.64559 5.18912 4.18594 1.18252
Rectangular with curved bottom shape,
r = 1.75 m, 1.62335 – 5.62670 1.47109 – 3.59539 1.13768
Slat = 19.9 m2 , V = 11.94 m3 1.79531 7.23652 6.54792 1.68722 6.42676 4.47109 1.15162
r = 0.45 m, 1.58567 – 5.16954 1.41265 – 3.22471 1.04995
Slat = 23 m2 , 1.74983 5.66163 5.97709 1.58541 4.38824 3.84372 1.08510
V = 13.8 m3
Rectangular,
r = 0 m, 1.58051 – 5.10002 1.40291 – 3.14488 1.03552
Slat = 23.4 m2 , 1.73554 5.32605 5.86361 1.56926 4.32280 3.73618 1.12331
V = 14.04 m3

∗ The upper row presents the data without the account of unilateral constraints, the lower one – those with the account of unilateral constraints.
467
 468

Table 5.9 Displacements and slopes of slotted foundations under momental loads at different longitudinal cross-section shape ∗

Load Mx =103 kNm My =103 kNm Mz =103 kNm

5

Displacements Δ, m and slopes ψ, rad −Δy 103 −Δz 103 ψ x 103 Δx 103 −Δz 104 ψ y 103 ψ z 103
Curved cross-section, m = 2; 4.26069 – 3.48175 6.14423 – 4.87393 3.98424
Slat = 16.05 m2 , V = 9.63 m3 4.26069 – 3.48175 6.14423 – 4.87393 3.98424
m = 4; 3.88402 – 2.88659 5.66971 – 4.09107 3.12596
Slat = 18.8 m2 , V = 11.28 m3 7.62012 1.87499 6.86811 6.61246 3.21264 4.69717 3.41242
m = 6; 3.70510 – 2.64377 5.49563 – 3.82166 2.80387

[email protected]
Slat = 20 m2 , V = 12 m3 12.2946 11.3150 13.5728 6.36248 2.87255 4.36469 3.05412
m = 8; 3.60014 – 2.51124 5.40642 – 3.68641 2.63404
Slat = 20.7 m2 , V = 12.42 m3 12.1117 11.2701 13.2226 6.25191 2.69309 4.21045 2.86849
m = 12; 3.48212 – 2.37049 5.31707 – 3.55237 2.45789
Slat = 21.4 m2 , V = 12.84 m3 11.8629 11.3008 12.9047 6.13102 2.53637 4.04332 2.67473
Rectangular with curved bottom shape,
r = 1.75 m, 3.59569 – 2.70129 5.62619 – 4.10169 2.69669
Slat = 19.9 m2 , V = 11.94 m3 10.9798 7.06645 12.1986 6.51721 2.88868 4.69139 2.93986
r = 0.45 m, 3.22484 – 2.10258 5.16913 – 3.33392 2.11851
Slat = 23 m2 , V = 13.8 m3 4.90505 0.83356 3.05339 5.90579 1.90846 3.78202 2.29878
Rectangular,
r = 0 m, 3.14461 – 1.98897 5.09949 – 3.22714 2.03725
Slat = 23.4 m2 , V = 14.04 m3 4.61144 0.75369 2.73880 5.81562 2.02085 3.65644 2.21143

∗ The upper row presents the data without the account of unilateral constraints, the lower one – those with the account of unilateral constraints.
Calculation of Bases for Rigid Complex-Shaped Deepened Foundations
5.6 Calculation of Contact Interaction of Bases 469

Mz ) = 10 kN·m (Table 5.9). In the lower rows the values of Δz (m) and ψ (rad),
respectively, with the account of unilateral constraints are given.
As one should expect, the absolute values of the displacements and slopes
decrease with the increase of the contact surface area (the foundation volume). The
calculation data for the case of a moment Mx , rotating the slotted foundation in the
longitudinal cross-section plane, make an exception. A non-monotonous behaviour
of the contact interaction parameters observed here for the calculations with the
account of unilateral constraints is explained by an insufficiently uniform discretiza-
tion of the end surface.
For the foundations with almost equal contact surface area (equal volume) and
a close geometrical shape of the longitudinal cross-section (Tables 5.8 and 5.9) the
difference in the contact interaction parameters does not increase 5–7%. The results
for the foundations with a longitudinal cross-section of a rectangular shape are also
rather close. At it could be expected, the best parameters (in the sense of the smallest
slopes and displacements) are obtained for the foundations with a linear bottom
shape.
On the other hand, unit values of the displacements and slopes (for the founda-
tions with equal average values of forces and moments) have essential differences
(up to 30%) depending on the longitudinal cross-section shape. It especially con-
cerns the foundations with a curved bottom shape. With the increase of the dimen-
sionless shape parameter m these differences decrease.
Thus, the account of the longitudinal cross-section shape for the calculation of
slotted foundations based on the base deformations is quite important and can result
in the lower materials consumption and excavation volume for slotted foundations.
The noted trend to the decrease of slopes and displacements can be used to improve
the technical and economic parameters of design solutions. Namely, application of
slotted foundation with a curved bottom instead of those with a rectangular longitu-
dinal section is possible; according to our calculations, it will result in the decrease
of concrete consumption at the foundation production, the restrictions on the dis-
placements and slope, specified by the design assignments, being obeyed.
Besides, the obtained data provide the evidence for both the area and the shape
of the lateral surface essentially affecting the slopes and settlements of the slotted
foundations. In order to optimize the foundation parameters, while comparing the
calculation results one can also specify as characteristics [16] unit settlements z
(1/m2 ) and slopes ψx (rad/m3 ), given by

 = /V,ψ = ψ/V

where V is the foundation volume.

As follows from Tables 5.8 and 5.9, with the increase of Slat the unit slopes and
unit settlements will also decrease, even faster than the absolute values. The highest
values of the slopes and the settlements correspond to the foundations with a curved
[email protected]
bottom shape. Based on the calculation data obtained, it can be noted that the
geometrical parameters of slotted foundations with a curved bottom shape should
be assigned with high m index value. On the other hand, application of slotted
470 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

foundations with a rectangular bottom shape can result in an excessive (≈ 30%)

consumption of concrete for producing foundations without the corresponding (≈
10%) decrease of their base deformation parameters.
Thus, based on the analysis of the calculation data obtained, one can conclude
on the necessity of a substantiated assignment of the geometrical size of slotted
foundations with different longitudinal cross-section shape.
Comparison of calculated and experimental data under a vertical compressive
load acting on a slotted foundation. The numerical results, obtained using the pro-
posed boundary-element method, will be compared with the field test data. In order
to estimate the possibilities of the proposed calculation method we compare the plot-
ted dependences of the slotted foundation settlements on the loads with the static test
plots. As follows from the references [3, 29, 136], due to a rather labour-consuming
and expensive performance of field experiments, the experimental studies of interac-
tion of slotted foundations with soil bases were mostly carried out in laboratory with
models of various scale. In order to use the results of testing of the slotted foundation
models of various scale by a static load, for the determination of the characteristics
of their interaction with the soil bases of industrial foundations, empirical formulae
are required [221]. At present they are missing for the slotted foundations of various
shape and at various soil conditions. Taking into account these complications at the
transfer form the laboratory studies to the field experiments, in order to provide the
most complete and reliable study of the calculation and experimental data for the
construction sites, we will give a certain priority to the field experiments.
Among the rare field experiments available we will use the data [136] of static
tests by a vertical compressive load of a slotted foundation with the size of 1 × 0.6 m
and the depth of 4 m by a vertical compressive load. The field slotted foundation was
performed in a trench, worked out dry at a test site in Yekaterinburg.
Soil conditions. The base of the slotted foundation was formed by a loose satu-
rated alluvial clayey with liquidity index from low-plasticity to high-plasticity. The
physical and mechanical characteristics of the soil, determined from the drilling of
a 15 m deep well, were the following:

natural humidity w = 0.27,

compressive deformation modulus Ec = 4.8 MPa,
soil density γ = 19.8 kN/m3 ,
internal friction angle ϕ = 16◦ ,
unit cohesion C = 0.013 MPa,
liquidity index IL = 0.55.

Figure 5.39 shows an experimentally obtained [136] dependence of settlement

versus vertical compressive load for a field slotted foundation under consideration.
In the same figure the calculation of settlements of a slotted foundation using the
boundary-element method is shown both with and without the account of unilateral
[email protected]
constraints. A uniform discretization of the contact surface of the soil and the slotted
foundation with a rectangular bottom shape consisted of 394 quadrangular bound-
ary elements. The Poisson ratio value ν=0.445 and the total deformation modulus,
5.6 Calculation of Contact Interaction of Bases 471

Fig. 5.39 Dependence of the settlement of a test slotted foundation on load: (1) experiment, (2)
boundary-element method (mk = 4); (3) boundary-element method (mk = 2.5); (4) layered sum-
mation method (mk = 2.5); calculation with (dashed line) and without (solid line) the account of
unilateral constraints

corrected in accordance with [215], E = 4·Ek = 19.2 MPa, were used for the calcu-
lations. The foundation weight was GF = VF ·γb = 2.4 m × 25 kN/m = 60 kN. The
analysis of the data, presented in Fig. 5.39, has shown the results of the boundary-
element calculation of the base deformations of the slotted foundation are in a rather
good agreement with the field experiment at load values of 75–80% of the limiting
load. The settlement discrepancy did not exceed 3–5 mm.
Note that the agreement of the experimental and calculated data essentially
depends on the accuracy of assignment of the deformational characteristics of soils
(first of all, the calculated deformation modulus) which can become sufficient only
in the case of field methods of their determination. The lack of data for the punch
deformation modulus resulted in the necessity of a correction factor mc to be intro-
duced. Though this correction factor correctly reflects the necessity of the compres-
sive deformation modulus to be increased (the deformation modulus values obtained
from the compression tests practically always appear to be lowered), however, the
choice of its value is rather arbitrary. Note that the values of mc , quoted for clayeys
in [215, Table 1.16] with the intervals of 0.5 and 1.0, are rather essentially depen-
dent of the soil porosity coefficient for all liquidity indices IL ≤ 0.75. Application of
tabulated values of the deformation modulus E according to [215, Table 1.13] with a
more detailed account of the liquidity index in the interval 0.5<IL ≤0.75 enables in
the specific case under our consideration (IL = 0.55) the value of E =12 MPa to be
used for the calculations. Computations, performed for this value of the calculated
deformation modulus (i.e. for mc =2.5) enabled the calculation and experimental
data to be obtained practically identical (lines 3 in Fig. 5.39). In this case, the exper-
[email protected]
imentally obtained plot of settlement versus load in the load range, not exceeding the
calculation load N ≤300 kN, appeared exactly between the theoretical dependences
with and without the account of unilateral constraints.
472 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

For the calculation load, equal to N=300 kN, using the layer summation method,
we have determined the settlement of the slotted foundation under consideration
as for a nominal foundation according to Appendix B of the Construction Rules
and Regulations 2.02.01–83. The settlement value was 6.875 mm (Fig. 5.39). For
the same load value the linear boundary-element method calculation (i.e. without
the account of unilateral constraints) gives the settlement value of 8.035 mm. As
one can see, in the load range where the linear calculation can be performed, both
methods – layered summation and boundary-element – produce not only compa-
rable, but also practically hardly distinguishable results (the difference of settle-
ments is 1.16 mm, being within the measurement accuracy). It is important to
note that the account of the structure nonlinearity (unilateral constraints) in the
developed boundary-element approach results in the calculated settlement value of
11.572 mm what is by factor of 1.7 larger than according to the Construction Rules
and Regulations, and practically coincides with the experimental value of 12 mm
(Fig. 5.39).
Thus, the proposed numerical calculation method according to the second lim-
iting state enables the slotted foundation settlements in the phase of compression
and local shears to be reliably predicted and, hence, the boundary-element solutions
to be recommended for implementation in the calculation practice for mass engi-
neering when the application of the linearly deformable half-space model is most
justified.
Prediction of deformations of the slotted foundation base under an off-centre
loading. From the point of view of the method of production, slotted foundations
belong to the structures, made in cavities, formed inside the soil itself and, contrary
to the foundations, produced in open pits, for them, as for deep structures, the known
simplified methods of calculation of pile foundations and deep bored supports [51],
as fixation in the soil, can be applied. In spite of the greater preference of the elastic
half-space method, for practical calculations still a more simple calculation model
of an elastic medium, based on the Winkler hypothesis, is applied. A proportional
increase of the coefficient of subgrade reaction with depth is assumed.
The book by Pavlov [136] contains general considerations regarding the calcula-
tion of foundations with the account of their fixation in the soil as well as an example
of calculation of a slotted pier foundation for the action of an off-centre load. The
slotted foundation calculation method applied there (with the account of the fixa-
tion in the soil) does not take into account the soil resistance to friction at the lateral
faces, parallel to the action of the forces and moments. Since the slotted foundations
have a rather developed lateral surface, the actual stability of the slotted foundation
will always be higher than the calculated one what is noted in [136] at the analysis of
the experimental data. A lack of specific methods of calculation for slotted founda-
tions, in opinion of [136], essentially affects their economic efficiency. Therefore, it
is advisable to carry out an extended (with the account of contact tangential and nor-
mal stresses) calculation of deformation of the slotted foundation base for the elastic
[email protected]
half-space model and, based on its results, to evaluate the reliability overestimation,
obtained using the traditional calculation method [136, 159].
5.6 Calculation of Contact Interaction of Bases 473

The example of calculation for an off-centre load, quoted in [136], concerns the
design of the foundation of a one-storeyed industrial building with a ferroconcrete
frame and travelling cranes with rated load capacity of 50 tons. The bay width 24
m, the step of columns of the main frame 12 m. The column cross-section 600 ×
600 mm.
Input data. For the calculation the most loaded foundation of the external row is
taken, for which the calculated forces at the top edge level are

N = 1720.8 kN,
Mx = My = 458.3 kN·m,
Hx = Hy = 37.5 kN.

Soil conditions. From the soil surface to the depth of 5 m tough clay (E=15 MPa,
γ=18.3 kN/m, ϕ=14◦ , C=0.043 MPa, IL =0.37), below – to the depth of 10 m –
eluvial clayey (E=27 MPa, γ=21.2 kN/m, ϕ=25o , C=0.04 MPa, IL = 0).
The foundation dimensions are 3.5×0.6 m. The slot depth is 5 m. The slotted
foundation has a rectangular longitudinal cross-section with a curved bottom shape,
r=1.75 m (Fig. 5.40). The slotted foundation calculation is performed in [136]

[email protected]
Fig. 5.40 Scheme for the calculation of a slotted foundation with a rectangular longitudinal cross-
section with a curved bottom for an off-centre load
474 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

according to the scheme of an absolutely rigid rod. In order to simplify the calcu-
lations the curved contour of the foundation bottom is substituted by a rectangular
one with a ledge. The ledge value is determined from the condition of the equality of
volumes of the real and the equivalent foundations. The tests for the foundation bear-
ing capacity, based on the soil base resistance, as well as for the foundation bearing
capacity and stability in the plane perpendicular to the slot plane, were performed
with a considerable reserve. The calculated foundation weight is GF = VF ·γb = 291.6
kN (γb = 25 kN/m is the density of the concrete, class V15).
Based on the test for the foundation stability in the plane, perpendicular to the
slot plane, the angle of the foundation rotation around the rotation centre, located at
the depth of z = 3.42 m, was calculated:

ψ = 2.14 · 10−2 rad

The calculation using the layered summation method was performed as for a
nominal foundation according to the Appendix B to the Construction Rules and
Regulations 2.02.01–83. As a result, the settlement value was obtained:

S = 2.06 cm

For the calculation of the slotted foundation according to the proposed approach
using the boundary-element method, the point of application of external forces and
moments is considered to be the centre of the top edge of the foundation on the
level of the calculated soil surface (Fig. 5.40). The absolute values of momental
loads, applied to the slotted foundation, should be increased by a value Hx (Hy )·h
= 37.5 kN × 1.5 m = 56.25 kN·m where h=1.5 m is the height of the base of
the column. The calculation scheme for the slotted foundation under the specified
off-centre load and discretization of its lateral surface into the boundary elements
are shown in Fig. 5.40. The total number of the boundary elements on the contact
surface of the slotted foundation and the soil was 388.
Since two layers of soil are located above the foundation bottom, the calculations
are performed for three variants of given base parameters:

(1) E = E1 = 15 MPa, v = v1 = 0.46,

(2) E = E2 = 27 MPa, v = v2 = 0.38,
(3) E = E1 ·hh11 +E2 ·h2
+h2 = 17.8 MPa, v = v1 ·hh11 +v2 ·h2
+h2 = 0.44

being the weighted deformation parameter values, h1 = 5 m, h2 = 1.5 m being the

soil layer thicknesses.
The Poisson ratio values ν1 and ν2 were determined using the Wet [240] from the
tabulated internal friction values ϕ1 and ϕ2 of the upper and the lower soil layers,
[email protected]
respectively.
The results of calculation using all the three mentioned variants of the base defor-
mational parameters are listed in Table 5.10 where the contact interaction parame-
5.6 Calculation of Contact Interaction of Bases 475

Table 5.10 Parameters of a slotted foundation base deformation under off-centre loading

Displacements (cm) Slopes (rad)

x ·10 y ·10 z −ψ x ·104 ψ y ·104 −ψ z ·104

E = 15 MPa, ν = 0.46
2.3327 2.4824 2.1809 8.0619 7.5369 0
2.7854 2.8139 2.3489 8.2212 9.1031 1.0898
E = 27 MPa, ν = 0.38
1.2771 1.3973 1.2124 4.5279 4.0663 0
1.4362 1.6503 1.3211 4.9139 4.4970 0.0772
E = 17.8 MPa, ν = 0.44
1.9606 2.1011 1.8401 6.8229 6.3134 0
2.5239 2.3691 1.9884 7.0115 8.2887 1.1693

ters are presented both with (the upper row) and without (the lower row) the account
of unilateral constraints.
As one should expect, the account of unilateral constraints for all calculation vari-
ants leads to an increase of absolute values of the slotted foundation displacements
and slopes. For settlements this increase was 7–8%, for horizontal displacements –
10–21%, for slopes – 2–28%. Thus, for slotted foundations with a developed lat-
eral surface, the account of unilateral constraints in the calculations is necessary.
Besides, due to the action of forces and moments in orthogonal directions, a slight
rotation of the slotted foundation around the central symmetry axis appears. This
rotation is due to the account of the structure nonlinearity and in principle cannot be
obtained by any of the methods in linear formulation.
Calculations of the foundation settlements z according to the first and the third
calculation variants are in a rather good agreement with the calculation data accord-
ing to the absolutely rigid rod scheme [136]. A minimal discrepancy was obtained
for the third variant with the account of unilateral constraints, its value being only
4%. The calculation according to the second variant, when the whole base parame-
ters are associated with the deformational parameters of the soil layer in the bottom,
results in the discrepancies of the results both for the settlements and the slopes
more than by 40%. This, in particular, confirms that for a slotted foundation the lat-
eral surface bearing capacity value is much more important than that for the bottom.
If the calculation is based on the third variant, i.e. with the averaged soil
deformational parameters, then application of the numerical approach based on
the boundary-element method results in a decrease of the slotted foundation set-
tlements. This is even more applicable for slopes which for all the three vari-
ants appeared almost three times smaller than for the calculation using the coef-
ficient of subgrade reaction. Since according to the calculation of the foundation
as a fixation in the soil x = z0 ·tanψ = 3.42 m·2.14·10-3 ≈ 0.0073 m, for the
boundary-element
[email protected] a decrease of horizontal displacements is also noted
(Table 5.10).
Thus, the boundary-element approach enables a designer to obtain a full pat-
tern of the spatial strained state of the soil base of a slotted foundation subject to
476 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

an off-centre load of a general kind. The performed comparison of our numerical

calculations with the data of [136], obtained using the deformation-based method
of calculation of a slotted foundation as a nominal foundation on a natural base in
accordance with the requirements of the Construction Rules and Regulations, is the
evidence for the importance of the account of the soil back pressure forces for the
reliable determination of the slotted foundation displacements and slopes. In com-
parison with the design regulations being used [215], the approach proposed here
is sufficiently universal and results in a more economical design solutions since it
enables the pressure transferred to the soil to be increased and the foundation size
to be reduced. Besides, the results of the comparison performed have shown a pos-
sibility of effective application of the boundary-element method in practical design
of slotted foundations at various shape of the deepened part.

5.6.3 Contact Stress on the Lateral Surface of a Slotted Foundation

Slotted foundations are bearing structures which accept all types of loads from
the above-earth parts of buildings and structures. The loads are transferred to
the base by the bottom and lateral surfaces of the slotted foundations. The slot-
ted foundation thickness, corresponding to its width, is by an order of magnitude
lower than its other dimensions. Therefore, the studies of distribution of tangen-
tial and normal stress over the slotted foundation lateral surface, developed in
depth and in the longitudinal direction, are of the main interest. Here one should
note that the experimental studies of formation and transformation of the con-
tact stress field, required for better understanding of the processes of the slot-
ted foundation interaction with the soil medium at different loading conditions,
is rather difficult due to technical reasons, the most essential of which are errors
induced in the pressure cell readings in the course of their installation and con-
crete placement, structural features of the equipment, high measurement errors
(of the order of 20%), etc. [136]. Therefore, technical results for the determina-
tion of forces of contact interaction of slotted foundations with soil by means of
mathematical modelling are of great importance for soil mechanics and foundation
engineering.
The known mathematical difficulties and total lack of analytical solutions of spa-
tial contact problems for slotted foundations with all features of contact surfaces
lead to a necessity of the contact forces to be studied only approximately, using
numerical methods and computations.
Application of numerical solutions of spatial contact problems for deepened
punches, obtained using the boundary-element method, enables all components of
the contact stress vector for the slotted foundations to be estimated without invok-
ing expensive and labour-consuming field measurements. The proposed approach to
[email protected]
the contact stress studies enables the experiments to be much simplified and a suf-
ficiently complete information to be obtained on the contact stress field distribution
with an accuracy, a priori not worse than the experimental one.
5.6 Calculation of Contact Interaction of Bases 477

The most evident is consideration of the contact stress fields for a prismatic slot-
ted foundation whose all sides are perpendicular to the coordinate axes, since for
the corresponding faces one of the following relations will be fulfilled:

σn = ±px , σn = ±py , σn = ±pz .

On a slotted foundation lateral surface (x=±h/2), in a pre-chosen coordinate sys-

tem (Fig. 5.32), the normal stresses σn =±px , and the tangential stress vector will
always belong to the flat contact surface. Therefore, in order to indicate the tangen-
tial stress vector, a direction field is convenient, determined in each contact point by
the following angle:
⎧ 
⎪ pz
⎪
⎪
⎨
arctan py ,py > 0;
$ = π + arctan ppyz ,py < 0,pz ≥ 0; (5.7)
⎪
⎪ 
⎪
⎩ −π + arctan ppyz ,py < 0,pz < 0.

For the numerical modelling of the contact stress consider a typical pier slotted
foundation with a rectangular longitudinal cross-section and the dimensions of 3.5
× 2.6 × 0.6 m, subject to spatial loads of the following type:

(1) central vertical compressive load,

(2) eccentric vertical load,
(3) inclined force load in the plane of the longitudinal cross-section,
(4) momental load in the plane of the longitudinal cross-section,
(5) inclined force load, acting orthogonally to the plane of the longitudinal cross-
section,
(6) momental load, acting orthogonally to the plane of the longitudinal cross-
section,
(7) torque along the vertical axis.
The contact stress field analysis was carried out for the following fixed values
of the base deformational parameters: E=20 MPa, ν=0.4, corresponding to the
conditions of construction of slotted foundations in firm-structure soils of low-
moisture clayey type.

Knowing the direction cosines and stress vector components

 px , py , pz for each
boundary element, one can calculate tangential τ = ± p2y + p2z and normal σn
contact stresses for all faces of the slotted foundation.
A rather full pattern of contact stress distribution for slotted foundations with
a rectangular lateral surface is obtained from a numerical boundary-element solu-
tion of the spatial contact problem and plotted in Figs. G.2–G.15 of Appendix G,
[email protected]
using the isolines of dimensionless stresses σ̄n = ±p̄x , p̄y , p̄z , τ̄ . As a scale for
the contact stress we used the average pressure pav =102 kN/SF =42.69 kN/m2 where
SF =23.42 m2 is the surface of the contact of the foundation and the soil.
478 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

The vector contact stress fields, shown in Figs. 5.41–5.48, enable the soil resis-
tance to the friction on the slotted foundation lateral surfaces under spatial static
loads to be studied quite effectively and with an accuracy, sufficient for practical
purposes, avoiding complex and expensive field measurements. The information
about the tangential stress vector value and direction enables the tangential contact
stress field over the whole slotted foundation lateral surface to be visualized on a
boundary-element grid of even a moderate density. Thereby the mechanism of the
slotted foundation lateral surface resistance to the shear deformation can be studied
with the account of the three-dimensionality of the stress-strained state of the base.
Figures 5.49–5.53 contain isolines of the reduced direction angle θ̄ = 2θ/π for the
tangential stress vector as an additional geometrical characteristics of the contact
stress field.
As seen from Figs. 5.41–5.53 and Appendix G, the calculation using the
boundary-element method has enabled rather distinct patterns of the tangential
stress field characteristics on the lateral surface of the slotted foundations to be
obtained, what does not seem to be possible in the framework of traditional cal-
culation schemes.

(a) (b)

Fig. 5.41 Tangential contact stress vector fields on the lateral surface of a slotted foundation under
an overturning moment My =1.4·103 kN. m, (a):x=0.3 m, (b): x= –0.3 m

(a) (b)

[email protected]
Fig. 5.42 Tangential contact stress vector fields on the lateral surface of a slotted foundation under
a combined action of a force and a momental load Pz =103 kN, My =1.4·103 kN. m, (a):x=0.3 m,
(b): x= –0.3 m
5.6 Calculation of Contact Interaction of Bases 479

(a) (b)

Fig. 5.43 Tangential contact stress vector fields on the lateral surface of a slotted foundation under
a combined action of a force and a momental load Pz =103 kN, My =0.7·103 kN. m, (a)x=0.3 m,
(b) x= –0.3 m

(a) (b)

Fig. 5.44 Tangential contact stress vector fields on the lateral surface of a slotted foundation under
an inclined force R=103 kN, β=135◦ perpendicularly to the longitudinal cross-section plane,
(a)x=0.3 m, (b) x= –0.3 m

(a) (b)

Fig. 5.45 Tangential contact stress vector fields on the lateral surface of a slotted foundation under
a horizontal force R=103 kN, β=180◦ perpendicularly to the longitudinal cross-section plane,
(a)x=0.3 m, (b) x= –0.3 m

Consider how the contact stress field components, obtained from the calcula-
tions, help one to estimate the main features of the slotted foundation lateral surface
functioning at different spatial loading conditions.
First of all, from the value and the distribution character of normal and tangential
[email protected]
stresses one can study the resistance of the soil to the friction over the contact surface
under vertical, horizontal, and momental loads as well as their combinations, acting
on a slotted foundation.
480 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

(a) (b)

Fig. 5.46 Tangential contact stress vector fields on the lateral surface of a slotted foundation
(x= ±0.3 m) under an inclined force R=103 kN within the longitudinal cross-section plane,
(a)α=135◦ , (b) α=180◦

(a) (b)

Fig. 5.47 Tangential contact stress vector fields on the lateral surface of a slotted foundation
(x= ±0.3 m) under an eccentric vertical force Pz =103 kN, (a) y = –0.7 m, (b)  y = –1.4 m

Fig. 5.48 Tangential contact

stress vector field on the
lateral surface of a slotted
foundation (x= 0.3 m) under
a torque Mz = 0.5. 103 kN. m

The data on the contact stresses are important for the analysis of the stress-
strained state in the foundation active area. With an assumption, common for the
soil mechanics, that the stress components in case of a small development of plastic
deformation areas, can be determined from the solutions of theory of elasticity, one
can easily determine the calculated pressures on the soil base for slotted foundations
of different size and depth under a complex spatial loading.
[email protected]
The solution of the problem of increase of reliability and improvement of calcula-
tion methods for slotted foundations is impossible without the account of strength-
related characteristics of the soil base. As noted above, application of the contact
5.6 Calculation of Contact Interaction of Bases 481

(a)

(b)

Fig. 5.49 Isolines of reduced direction angles θ̄ = 2θ/π of the tangential contact stress vector on
the lateral surface of a slotted foundation (x= 0.3 m) under a central force R=103 kN: (a) vertical
force, (b) horizontal force

[email protected]
stress data enables the resistance to shear on the contact surface of the slotted foun-
dation and the soil to be estimated. Here we show, how the knowledge of the contact
stress field on the lateral surface of a slotted foundation and the knowledge of the
482 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

(a)

(b)

Fig. 5.50 Isolines of reduced direction angles θ̄ = 2θ/π of the tangential contact stress vector on
the lateral surface of a slotted foundation (x= 0.3 m) under: (a) a torque Mz =0.5·103 kN. m and (b)
an overturning moment My =1.4·103 kN. m

limiting state condition for the soil enables the contact strength of the foundation
[email protected]
structure binding to the base can be checked.
For the mechanical behaviour of soil, at each site the Coulomb friction law should
hold
5.6 Calculation of Contact Interaction of Bases 483

(a)

(b)

Fig. 5.51 Isolines of reduced direction angles θ̄ = 2θ/π of the tangential contact stress vector
on the lateral surface of a slotted foundation (x= 0.3 m) under a combined action of a force and a
momental load Pz =103 kN, My =0.7·103 kN. m: (a) x= 0.3 m, (b) x= –0.3 m

|τ | ≤ C + σn · tan ϕ
[email protected] (5.8)

where C is cohesion, ϕ is the internal friction angle. As known from [25], in the
points where the tangential component τ of the reactive soil pressure exceeds the
484 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

(a)

(b)

Fig. 5.52 Isolines of reduced direction angles θ̄ = 2θ/π of the tangential contact stress vector on
the lateral surface of a slotted foundation (x= ±0.3 m) under an inclined force R=103 kN within
the longitudinal cross-section plane: (a) α=135◦ , (b) α=180◦

[email protected]
value, determined by the strength condition of Eq. (5.8), slippage occurs. The con-
dition of Eq. (5.8) is applicable for the points inside a soil medium. On the contact
surface of the base and the foundation the following condition is fulfilled:
5.6 Calculation of Contact Interaction of Bases 485

(a)

(b)

Fig. 5.53 Isolines of reduced direction angles θ̄ = 2θ/π of the tangential contact stress vector on
the lateral surface of a slotted foundation (x= ±0.3 m) under an eccentric vertical force Pz =103
kN: (a)  y = –0.7 m, (b)  y = –1.4 m

|τ | ≤ a + σn · tan δ
[email protected] (5.9)

where |a|≤ C, τ and σn are the components of the stress vector in the point of the
soil contact with the foundation, δ (≤ϕ) is the angle of the soil friction across the
486 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

foundation (measure of roughness). It is known from the experiments [25] that m =

a/C = tanδ, 0 ≤ m ≤ 1. At τ= 0, i.e. when δ = a = 0, an absolutely smooth surface
is obtained, and at |τ|≤ C + σn tanϕ, i.e. when δ=ϕ and a=C one obtains an abso-
lutely rough surface. For the calculations of concrete gravity retaining walls with a
flat boundary, for the sake of simplicity δ=2ϕ/3 is usually accepted [25]. For drift
sand, obeying the Coulomb strength law and interacting with concrete foundation
structures, the value of m = a/C = 0.8 is suggested to be chosen [27]. The relation
of the internal friction angle ϕ and the angle δ of friction on the contact surface will
be hereinafter estimated using a theoretically substantiated formula

tan δ = sin ϕ (5.10)

which was obtained in [124] for a soil with plastic incompressibility.

Since for the soil masses the most dangerous is violation of cohesion due to shear,
then, knowing the parameters of resistance to shear for different parts of the contact
surface, one can, according to Eq. (5.9), from the tangential stresses on the boundary
elements, determine the contact slippage areas where the tangential component of
the reactive soil pressure exceeds the value, determined from the strength condition.
In the first approximation, in order to estimate the strength of the contact between
the foundation and the soil, one should built, based on the calculation results, the
lines of equal level of a function

 
f σij = |τ | − a − σn · tan δ.

The areas with f < 0 give the evidence for the lack of slippage, on a contact
surface with f = 0 a limiting state is achieved, at f > 0 no cohesion of the foundation
with the soil is provided. Building up areas with different cohesion level enables the
areas of most probable slippage to be identified what, in turn, enables the necessary
recommendation to be worked out for the rational design of slotted foundations.
One should note that while performing calculations for slotted foundations,
whose lateral surface can be oriented depending on the slot location and the exter-
nal force direction, in firm-structure soils, it is important to take onto account the
soil base anisotropy. The slippage areas in this case are revealed in a similar way,
using the method considered, invoking anisotropic strength conditions [18] as well
as experimental data on the studies of strength-related characteristics of soils and
rock masses. The effect of anisotropy on the resistance to shear is especially pro-
nounced for varved clays with a layered structure. For such clay the vertical cohe-
sion value can be several times higher than cohesion in the horizontal direction
[102]. The account of anisotropy as one of the essential and common features of
rocks and soils will enable the efficiency of design of slotted foundations to be
increased; in particular, it will enable the slot plane orientation on the construction
[email protected]
site to be purposefully chosen as well as the length-to-depth ratio of the slotted
foundations to be improve in order to reduce the possibility of formation of slippage
areas at their developed lateral surface.
5.6 Calculation of Contact Interaction of Bases 487

Consider the effect of the strength-related soil properties on the features of the
shape and size of the contact strength fault areas over the slotted foundation lateral
surface, in case the "elastic" field of the contact stress being known. The studies
of the contact strength fault areas is important from the practical point of view
for reliable assignment of the slotted foundation shape and loading parameters as
well as for the formulation of criteria for the safety factor assignment in the design
regulations.
Studies of the spread of the contact strength fault areas over the slotted founda-
tion lateral surface was carried out for five variants of the strength-related parame-
ters of the soil, presented in Table 5.11. The same table contains the strength param-
eters δ and a, characterizing the friction and cohesion on the contact surface of the
foundation and the soil and related to the soil strength parameters ϕ and C by Eq.
(5.10) and a ratio m = a/C = 0.8.
Using the calculated contact stress values (Appendix G), in Fig. 5.54 lines, for
which f (σij ) = |τ|–a–σn tanδ=0, are built. Consequently, the configuration of areas,
for which Eq. (5.9) is violated and slippage is possible, are indicated.

Table 5.11 Parameters of internal and contact friction and cohesion

Strength Variants
parameters 1 2 3 4 5

φ, deg 17 27 37 37 37
δ, deg 16.3 24.4 31 31 31
C, kPa 6.25 6.25 6.25 11.25 18.75
a, kPa 5 5 5 9 15

Analyzing the presented calculation data, one can note the following qualitative
and quantitative features of the contact interaction of slotted foundations with soil.
First, the presence of slippage areas on the slotted foundation lateral surface is
revealed for all five characteristic types of spatial loading (Fig. 5.54, Table 5.12).
Numerical modelling of the contact interaction has shown that, depending on the
loading type and the soil strength-related properties, the slippage can occur over
rather extensive areas on the foundation lateral surface, comparable with the full
cohesion areas.
It is seen from the calculation data, presented in Fig. 5.54 that the slippage area
configurations are essentially dependent of the contact strength characteristics as
well as the load value and direction on the top edge of the foundation. The slippage
areas can be simply or multiply connected, usually, originating from the ribs, in
the vicinity of which the contact stress field is characterized by the highest nonho-
mogeneity degree. Under a force or a momental load across the plane of the slotted
foundation longitudinal cross-section, the slippage areas arise in the range of its bot-
tom on the lateral side where tensile normal stress is observed. On the opposite side
[email protected]
of the lateral surface, where the foundation is pressed to the soil (i.e. compressive
normal stress is developed), the tangential stress field is rather uniform, no areas of
shear contact strength faults are observed.
488 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

(a) (b)

(c) (d)

(e) (f)

Fig. 5.54 Boundaries of contact strength fault areas on the lateral surface of a slotted founda-
tion under different loading conditions (a–f) at the friction and soil cohesion parameters (1–5),
determined according to Table 5.6

The effect of the cohesion parameters (C, a) on the development of the slippage
areas is more noticeable than that of the internal and contact friction angles (ϕ,δ).
A double increase of δ under different type of loading in the plane of the slotted
foundation longitudinal cross-section results in an increase of the slippage areas
[email protected]
only by 3–5% (See the areas restricted by curves 1, 2, and 3 in Figs. 5.54b–f). This
is the evidence for the resistance to shear on the slotted foundation lateral surface
being determined mostly by cohesive forces. A somewhat different situation occurs
5.6 Calculation of Contact Interaction of Bases 489

Table 5.12 Maximal values of the function f (σij ) = |τ|−a−σn tanδ≥0 (kPa) on the lateral surface
of a slotted foundation with different parameters of contact friction and cohesion under spatial
loading

Strength parameters
1 2 3 4 5
a = 5 kPa, a = 5 kPa, a = 5 kPa, a = 9 kPa, a = 15 kPa,
◦ ◦ ◦ ◦ ◦
Loading type δ = 16.3 δ = 24.4 δ = 31 δ = 31 δ = 31

Transverse load 4.869 7.100 9.141 5.141 –

(a) R = 1000 kN, α = 135◦
Longitudinal load
◦
(b) R=1000 kN, α = 135 13.918 14.789 15.587 11.587 05.587
◦
(c) R = 1500 kN, α = 135 23.376 24.683 25.879 21.879 15.879
◦
(d) R=1000 kN, α = 90 13.945 14.838 15.653 11.653 05.653
Eccentric vertical load
R=1000 kN,
(e)  y = 0.7 m 16.259 16.904 17.495 13.495 7.495
(f)  y = 1.4 m 23.292 24.032 24.709 20.709 14.709

at the foundation being loaded across the slot plane. For such loading type, the
presence of a component, orthogonal to the foundation lateral surface, increases the
role of the internal friction forces in the soil, whose growth results in a practically
proportional expansion of the slippage areas (Fig. 5.54a).
As follows from the calculations, at a fixed value of the contact friction δ,
the increase of the soil contact cohesion parameter a rather essentially affects the
decrease of the fault areas for the strength of the contact between the foundation
and the soil (Fig. 5.54, curves 3, 4, and 5). In particular, at "transverse" loading
of the slotted foundation by an inclined force the increase of the a parameter from
5 kPa to 15 kPa (i.e. by factor of 3) at δ=31◦ has led not only to a sharp decrease of
the slippage areas, but even to their total absence (Fig. 5.54a).
The increase of the load on the foundation, as one could expect, results in an
increase of areas where the contact cohesion of the foundation with soil is violated.
It is seen, for instance, from the comparison of Figs. 5.54b and 5.54c where the
shear areas are shown on the lateral surface of the slotted foundation loaded by an
inclined force acting in the plane of its longitudinal cross-section. The load increase
by factor of 1.5 has led to a clearly revealed expansion of the slippage areas. The
discussed expansion (or appearance) of the slippage areas is revealed for all variants
of the strength-related parameters under consideration (Table 5.11). A general trend
to the expansion of the contact strength fault areas from the boundaries towards
the central part of the lateral surface along the inclined external force direction is
observed.
The effect of the resultant inclination angle on the configuration of the contact
strength fault areas can be revealed from the analysis of Figs. 5.54b and 5.54d. As
[email protected]
noted above, the inclined force (α=135◦ ), acting in the longitudinal cross-section
of the foundation (Fig. 5.32a), forms slippage areas on its lateral surface, adja-
cent to the opposite angles in accordance with the inclined force direction. Such
490 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

configuration of the slippage areas is naturally determined by the presence of a

vertical (impressing) and a horizontal (shearing) components of the inclined force.
Contrary to the above case, at α=180◦ , i.e. under a horizontal force, the mecha-
nism of formation of the slippage areas on the slotted foundation lateral surface is
qualitatively different. In this case the highest increase of tangential stress occurs on
the zones, adjacent to the vertical faces of the slotted foundation. A rather essential
irregularity of the slippage area boundaries is noted, due to the slotted foundation
size, loading level, and contact strength parameters.
Numerous calculations have shown that under a vertical load on a slotted founda-
tion, applied to its centre, the areas with highest tangential stress are localized near
its bottom. As follows from the data illustrated by Figs. 5.54e and 5.54f, the con-
tact strength fault areas under an eccentric loading are transformed and, besides the
areas, adjacent to the bottom, include also a more loaded vertical face. As a result,
the slippage area if formed at the loaded rib and the face which form it. An increase
of the vertical load eccentricity (double, for the case under consideration) results
in hardly noticeable deformations of the slippage area boundary contours. The slip-
page area boundaries are more sensitive to the variation of the contact cohesion
parameter a than to the variation of the contact friction angle δ.
The presented method of delineation of slippage areas on the lateral surfaces of
slotted foundations enables one, in case of a considerable growth of their fraction,
exceeding a certain threshold value (which for better reliability should be set, invok-
ing the empirical data), to conclude on the inapplicability of the linear elastic state
model and on a need for application of more complex calculation schemes of non-
linear deformation. On the other hand, a considerable complexity and high labour
consumption of solving nonlinear (elastoplastic) problems for a spatial stress-
strained state of the base essentially encumbers implementation of such solutions
into the practice of foundation engineering even at the modern level of mathematical
modelling. Therefore, in view of the slippage areas of different size, arising under
loading on the developed lateral surface of slotted foundations, for calculations it is
convenient to introduce a safety factor whose value for soils with known strength-
related characteristics can be assigned depending on the slotted foundation size,
required calculation accuracy, and the category of the building to be constructed,
based on multivariant boundary-element solutions, similar to those discussed above.
The calculations presented also enable one to make a conclusion, important for
practical purposes, that for soils with higher values of the C parameter the slip-
page areas on the slotted foundation lateral surface will shrink and, consequently,
the possibility of application of the elastic half-space model for base deformation-
based calculation of slotted foundations will be more justified. this will enable the
calculation loads to be noticeably increased and the slotted foundation economical
parameters to be improved.
In order to elucidate the contact deformation pattern in more detail one should
take into account that the stress-strained state of the slotted foundation + base sys-
[email protected]
tem, determined in the first cycle, will correspond to the real one only to a certain
extent: due to the soil slippage over the foundation the contact stress field will be
transformed what will result in the slippage area variation. Further improvement of
5.6 Calculation of Contact Interaction of Bases 491

the contact interaction results will require a series of calculations to be performed

and a special stepwise computation procedure to be organized. however, the trend to
the slippage development or stabilization will enable one to conclude on the change
of the calculation scheme, assignment of the foundation size and load values on its
top edge. The most important for practical purposes is to set the optimal size of the
slotted foundation. On one hand, the desire to decrease the foundation volume is
quite reasonable; however, it is related to an inadmissible increase of its settlements
and slopes. On the other hand, formation of contact strength fault areas on the lateral
surface of slotted foundations, even designed in an optimal way, but without a check
for sufficient cohesion over the contact surface, can reduce the quality of the design
solution, leading to an additional unpredicted increase of displacements and slopes.
Thus, theoretical studies of variation of contact stress fields enables the results
of base deformation-based calculations and field studies of slotted foundations to
be treated more exactly. Determination of contact stress fields on the slotted foun-
dation lateral surfaces by mathematical modelling and obtaining their qualitative
and quantitative dependences on the force, geometrical, physical and mechanical,
strength-related and other characteristics of the foundation+soil system will enable
purposeful control of the slotted foundation displacement parameters and check of
their stability to external spatial loads. As a result, the efficiency of application of
slotted foundations in the industrial and civil engineering will increase due to the
practical recommendations regarding the choice of the calculation schemes, opti-
mization of the shape parameters and loading as well as formation of substantiated
methods of calculation applicable for their design and functioning.

5.6.4 Slotted Foundations with Lateral Widenings

Based on the calculation experience, we propose new structures of slotted foun-
dations with lateral widenings in the form of longitudinal ribs of different cross-
section configuration (Fig. 5.55a). The horizontal support ribs result in additional
support area, increase the area of the contact surface with the soil. The geometrical
characteristics of a support widening are determined depending on the value and
direction of the external load, applied to the foundation, as well as on the physical
and mechanical properties of the soil. The presence of the support widenings in the
structure of the slotted foundations enables the settlements under vertical load to be
essentially (≈12%) reduced in comparison with the foundations without widenings.
The support widenings are the stress concentrators in the foundation active area.
Under inclined or momental loads the widening ribs prevent the foundation slopes in
both longitudinal and transverse directions, i.e. prevent the foundation overturning
effect under eccentric loading. At a given slotted foundation depth this is achieved
by a slight increase of the concrete volume in comparison with traditional slotted
[email protected]
foundations.
The analysis of the calculations performed has shown that the settlements and
slopes of the slotted foundations with lateral widening ribs are essentially affected
492 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

a b c

e f

Fig. 5.55 Technological scheme of production of a slotted foundation with support widenings:
(a) ready slotted foundation with support widenings; (b) trench digging; (c) making a support
widening; (d) reinforcement with reinforcement cages; (e) concreting

by such factors as the widening rib relative depth, cross-section shape and size. By
variation of these factors the settlement-vs-load characteristic (the main calculated
characteristic of a foundation) can be effectively controlled.
From the point of view of technology of production the slotted foundations
with support widenings can be attributed to the fast type. When a working is
made for a lateral widening of a foundation in a slot, a longitudinal cutting
[email protected]
(Figs. 5.55b, c) is performed by a basic machine, equipped with a special widener
of a pantograph or a tab type [47, 206]. As a result, a working with the required
configuration of the support rib is formed in the trench for the foundation. The
References 493

construction of a rotor widener presumes the presence of hard cutting elements

providing the efficient functioning of the widener in the soil mass with the account
of the speed of its forward motion. At the slotted foundation depth above 3.5 m a
second pass of the machine is possible in order to form the second widening rib
(Fig. 5.55c). Additional operations for the slot grading practically do not increase
the total labour consumption, do not make the cycle more complicated, do not slow
down the general pace of work. After the excavation work for the trench digging
and formation of a working for the support widening is finished, the foundation
reinforcement is performed by means of volume block reinforcement cages
(Fig. 5.55d). The reinforcement cages are joined into a single structure by vertical
groove elements with an additional function of restriction of the slotted foundation
length. Then the concreting stage follows (Fig. 5.55e) when the concrete mixture is
conveyed and placed by a force concrete pump, providing the required concreting
pace. The mixture distribution and compression is recommended to be performed
using a deep vibrator with a pulse effect on the concrete mixture. This provides
full air removal and formation of the concrete structure with high reliability and
guaranteed absence of structural defects of the foundation structure.

References
1. Aleksandrov A V, Lashchenikov B Ya, Shaposhnikov N N et al. (1976) Computer-aided
methods of calculation of rod systems, plates and shells. Stroyizdat, Moscow (in Russian)
2. Aleksandrovich V F, Fedorovskii V G (1982) Finite-element calculation of interaction of
a deepened foundation with an elastoplastic base. In: Engineering properties of soil and
calculation of the bearing capacity and settlements of foundations. Tr NIIOSP 78: 106–113
(in Russian)
3. Alekseyev A I, Nikiforov Yu N, Derevyanko V M (1985) Experimental studies of function-
ing of slotted foundations. In: Bases and foundations in the geological conditions of the
Urals. PPI, Perm, pp. 31–35 (in Russian)
4. Alekseyev V M, Lipson G A, Mitrenko Yu A (1980) Study of the bearing capacity of pyra-
midal piles on a vertical load. In: Soil mechanics, bases and foundations. Voronezh State
University, Voronezh, pp. 72–81 (in Russian)
5. Alekseyev V M, Lipson G A, Mitrenko Yu A (1984) Studies of bearing capacity of pyramidal
piles under an inclined force. In: Studies of rational structures of foundations. Voronezh State
University, Voronezh, pp. 17–24 (in Russian)
6. Amusin B Z, Fadeyev A B (1975) Finite-element method at the solution of problems of
mining geomechanics. Nauka, Moscow (in Russian)
7. Bakenov Kh Z, Bizhanov K S, Repina P I (1984) On the calculation of settlements and bear-
ing capacity of pyramidal piles using the finite-element method. In: Bases and foundations
in weak and heaving soils. LISI, Leningrad, pp. 32–42 (in Russian)
8. Bakholdin B V, Kolesnikov L I, Shikalovich N S (1989) Effect of frost-heave forces of soils
on the bearing capacity of pyramidal piles. Soil Mech Found Eng 26: 243–248
9. Baranov V S, Romanov D A, Romanov K D, Sidorchuk V F (1973) Investigation of the
pressure distribution in the soil around piles with spherical enlarged bases. Soil Mech Found
[email protected]
Eng 10: 96–99
10. Bartolomei A A (1994). Modern state and problems of pile foundation engineering. In: Proc
4th Int Conf Pile Foundation Engineering Problems. Pt 1. Improvement of methods of cal-
culation and technology of pile foundations. PGTU, Perm, pp. 38–43 (in Russian)
494 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

11. Bartolomei A A, Chikishev V M, Yushkov B S, Malyugin V P (1994) Experimental and

theoretical studies of interaction of foundations of piles with blades with the soil base. In:
Bases and foundations at the geological conditions of the Urals. PPI, Perm, pp. 3–13 (in
Russian)
12. Bartolomei A A, Lavrentyev V A (1988) Pile foundations of increased bearing capacity.
In: Bases and foundations in geological conditions of the Urals. PPI, Perm, pp. 8–13 (in
Russian)
13. Bartolomei A A, Pilyagin A V (1988) Stress-strain state of the beds of pyramidal pile foun-
dations. Soil Mech Found Eng 25: 136–140
14. Bartolomei A A, Ponomaryov A B, Yushkov B S (1994) Foundations of hollow conical piles.
In: Bases and foundations in geological conditions of the Urals. PPI, Perm, pp. 13–17 (in
Russian)
15. Bartolomei A A, Yushkov B S, Ponomaryov A B (1990) Experimental studies of the stress-
atrained state of the active zone of a hollow conical pile. In: Bases and foundations in geo-
logical conditions of the Urals. PPI, Perm, pp. 60–66 (in Russian)
16. Bekbasarov I I, Shilibekov S K, Kushekbayev B Z (1989) On the effect of some geometrical
parameters of foundations in tamped pits on their resistivity to static vertical load. In: Bases
and foundations in geological conditions of the Urals. PPI, Perm, pp. 88–93 (in Russian)
17. Budanov V G (1975) Functioning of a base at cross-beam fixation of horizontally loaded
power line support pillars. Energet Stroit (issue 7): 52–54 (in Russian)
18. Bugrov A K, Golubev A I (1993) Anisotropic soils and bases of buildings. Nedra, St. Peters-
burg (in Russian)
19. Bukharin Ye M, Gabliya Yu A, Levin L E (1971) Design of foundations for power line
supports. Energiya, Moscow (in Russian)
20. Bykov V I (1975) Experimental investigations of performance of horizontally loaded pile
foundations. Soil Mech Found Eng 12: 102–104
21. Bykov V I (1995) Foundations for individual residential buildings. Soil Mech Found Eng
32: 66–68
22. Chikishev V M, Bay V F, Dolgov V N, Shmidt V G (1994) Studies of functioning of foun-
dations of piles with enhanced bearing capacity in clay soils. In: Bases and foundations at
the geological conditions of the Urals. PPI, Perm, pp. 18–26 (in Russian)
23. Chikishev V M, Bay V F, Malyshkin A P (1990) Experimental studies of the stress-strained
state of soils in the base of group of piles with self-unfolding blades. In: Bases and founda-
tions at the geological conditions of the Urals. PPI, Perm, pp. 53–60 (in Russian)
24. Collins W D (1962) Some axially symmetric stress distributions in elastic solids containing
penny-shaped cracks. I. Cracks in an infinite solid and a thick plate. Proc Roy Soc Ser A
203: 359–386
25. Costet J, Sanglerat D (1975) Cours practique de mechanique des sols. Bordas, Paris
26. Dalmatov B I (1981) Mechanics of soils, foundations and bases. Stroyizdat, Moscow (in
Russian)
27. De Buhan P, Siad L (1989) Influence of a soil-strip interface failure condition on the yield-
strength of reinforced earth. Computers Geotech 7: 3–18
28. Dembicki E, Sieradzki M (1988) Analysis of the behaviour of piles in high compress-
ible soil subjected to torsion. In: Baltic Conf Soil Mech Found Eng. vol. 2: Con-
struction on heats and deformation of structures on highly compressible soils. Tallinn,
pp. 160–163
29. Derevyanko V M (1990) Experimental studies of the slotted foundation + soil system. In:
Bases and foundations in the geological conditions of the Urals. PPI, Perm, pp. 100–104 (in
Russian)
[email protected]
30. Drucker D C, Prager W (1952) Soil mechanics and plastic analysis or limit design. Quart
Appl Math 10:157–165
31. Druzhinin G A, Komzina A A (1988) Rational construction of shallow foundations for rural
buildings on a natural base. Kuibyshev State University, Kuibyshev (in Russian)
References 495

32. Dubov K A (1992) Calculation of piles with cross-shaped cross-section with diagonal rein-
forcement for raftless pile foundations of large-panel residential buildings. In: Increase of
efficiency of application of equipment and improvement of technology of engineering. VPI,
Vladimir, pp. 54–58 (in Russian)
33. Egorov K E (1948) Deformation of the base of a round rigid foundation under an eccen-
tric load. In: Bases and foundations. Issues of soil mechanics. Tr NIIOSP 11:119–138 (in
Russian)
34. Egorov K E (1958) On the issue of the base calculation under a foundation with a ring-shaped
footing. In: Bases and foundations. Soil mechanics. Tr NIIOSP 34:34–57 (in Russian)
35. Fadeev A B (1987) Finite-element method in geomechanics. Nedra, Moscow (in Russian)
36. Fadeev A B, Bakenov Kh Z, Repina P I (1988) Finite-element calculation of settlements for
round punches and single piles with the account of strength-related and rheological proper-
ties of soils. LISI, Leningrad (in Russian)
37. Fadeev A B, Bakenov Kh Z, Repina P I et al. (1987) Geomekhanika CREEP-3 software for
finite-element calculation of stress-strained state of round foundations with the account of
deformational, strength-related and rheological properties of medium. Gos FAP SSSR No.
50870000956, Moscow (in Russian)
38. Fadeev A B, Matveyenko G A (1983) Semianalytical finite-element method for prediction of
stress-strained state of bases for axisymmetrical foundations. In: Nonlinear soil mechanics:
Proc 6th Rus Conf St Petersburg, vol. 1, pp. 146–149 (in Russian)
39. Fadeev A B, Matveyenko G A (1988) Semianalytical finite-element method for solving spa-
tial problems of foundation engineering in elastic and elastoplastic formulation. Izv Vuzov
Stroit (issue 12): 113–116 (in Russian)
40. Fadeev A B, Matveyenko G A (1990) Semianalytical finite-element method for solving spa-
tial problems of geomechanics. In: Studies and developments on computer-aided design of
foundations and bases. NPI, Novocherkassk, pp. 28–35 (in Russian)
41. Fadeev A B, Preger A L (1984) Solution of the combined axisymmetric problem of the
theory of elasticity and plasticity by the finite-element method. Soil Mech Found Eng 21:
181–184
42. Fadeev A B, Repina P I, Abdyldayev E K (1982) Finite-element method for solving geotech-
nical problems and Geomekhanika software. LISI, Leningrad (in Russian)
43. Fedorovskii V G (1973) Algorithm of calculation of settlements of deep supports with
the account of gliding. In: Engineering constructions and theory of structures: Bases,
foundations and soil mechanics. Vysheyshaya Shkola, Minsk (issue 2): 123–128 (in
Russian)
44. Fedorovskii V G (1975) Settlements of piles in homogeneous and multilayer bases. In: Proc
1st Baltic Conf Mech Soil Found Eng, Gdansk, pp. 50–59 (in Russian)
45. Fedorovskii V G (1985) Modern methods of description of mechanical properties of soils.
VNIIS, Moscow (in Russian)
46. Fedorovskii V G, Kurillo S V, Kulakov N A (1988) Computing the lateral load on piles and
pile groups from the model of a linearly deformable half space. Soil Mech Found Eng 25:
166–172
47. Filakhtov A L, Lubenets G K, Pisanko N V et al. (1981) Experience of construction using
the "wall in ground" method. Budivelnyk, Kyiv (in Russian)
48. Gabibov F G (1995) Technology of production of a bored pile with a cross-section in the
shape of an equilateral triangle, In: Proc 2nd Ukr Sci Eng Conf Soil Mech Found Eng. Pt 2.
Efficient foundations produced with soil excavation, Poltava, pp. 114–117 (in Russian)
49. Gersevanov N M (1948) Experience of application of theory of elasticity to the determina-
tion of admissible loads on the soil based on experimental data. In: Selected works, vol. 1.
[email protected]
Pile bases and calculation of foundations. Stroyvoyenmorizdat, Leningrad, pp. 236–260 (in
Russian)
50. Glotov N M, Luga A A, Silin K S et al. (1975) Pile foundation. Transport, Moscow (in
Russian)
496 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

51. Glushkov G I (1977) Calculation of structures, deepened into soil. Stroyizdat, Moscow (in
Russian)
52. Glushkov V Ye, Mamayev N G, Kislitsyn S A (1994) Elastoplastic calculation of piramidal
pile bases under inclined and off-centre loads, In: Proc Int Conf Pile Found Eng Problems.
Pt 1. Improvement of methods of calculation and technology of pile foundation production.
PGTU, Perm, pp. 86–88 (in Russian)
53. Golubkov V N, Morgulis N L, Nikitin V F (1977) Tapered-pile foundations with intermedi-
ate cushion. Soil Mech Found Eng 14: 331–334
54. Gorbunov-Posadov M I, Malikova T A, Solomin V I (1984) Calculation of structures on an
elastic base. Stroyizdat, Moscow (in Russian)
55. Gorbunov-Posadov M I, Sivtsova Ye P (1966) Check of a pile for sliding. In: Bases and
foundations. Tr NIIOSP 56: 36–41 (in Russian)
56. Gorodetskiy A S, Zavoritskiy V I., Lantukh-Lyashchenko A I, Rasskazov A O (1989)
Automation of calculations of transport buildings. Transport, Moscow (in Russian)
57. Gotman A L (1987) Analysis of tapered piles under combined action of vertical, horizontal,
and flexural loads. Soil Mech Found Eng 24: 7–12
58. Gotman A L (1996) On the calculation of variable cross-section piles on the horizontal load
with the account of nonlinearity of deformation of soil and the pile material. In: Proc 5th
Intern Conf Pile Found Eng Problems. RNKMGF, Moscow, vol. 1, pp. 41–46 (in Russian)
59. Grigoryan A A (1984) Pile foundations of buildings and structures on subsiding soils. Stroy-
izdat, Moscow (in Russian)
60. Grilitskii D V, Kizyma Ya M (1967) Concomitant torsion of a rod and half-space. Intern
Appl Mech 3(issue 2): 54–59
61. Gwizdala K (1996) Analiza osiodan pali przy wykorzystaniu funkcji transformacyjnych.
Zeszyty Naukowe Politechniki Gdanskiej. Nr. 532. Budownictwo Wodne, vol. XLI.
62. Hunter S C, Gamblen D (1975) The theory of a rigid circular disc ground anchor buried in
an elastic soil either with adhesion or without adhesion. J Mech Phys Solids 22: 371–399
63. Ikonin S V (1988) Application of an elastic half-space model for the calculation of short
bored piles under a horizontal loading. In: Voronezh State University, Voronezh, pp. 103–
107 (in Russian)
64. Ikonin S V (1992) Contact interaction of a deepened punch with a base at arbitrary static
loading. In: Calculation and design of bases and foundations at complicated geotechnical
conditions, VISI, Voronezh, pp. 82–86 (in Russian)
65. Ilyichev V A, Mongolov Yu V, Shayevich V M (1983) Pile foundations in seismic areas.
Stroyizdat, Moscow (in Russian)
66. Ilyin V P, Karpov V V, Maslennikov A M (1990) Numerical methods of solving problems
of engineering mechanics. Handbook. Vysheyshya shkola, Minsk (in Russian)
67. Instructions on design of foundations for machines subject to dynamic loads (1982) Gerse-
vanov NIIOSP, Stroyizdat, Moscow (in Russian)
68. Instructions on design of pile foundations (1980) Stroyizdat, Moscow (in Russian)
69. Instructions on design of pyramidal pile foundations (1983) Glavn Upr Project Organiz,
Moscow (in Russian)
70. Ivanov P L (1991) Soils and bases of hydrotechnical structures. Vysshaya shkola, Moscow
(in Russian)
71. Kananyan A S, Kurnosov A I, Gorelov A V (1977) Designing the foundation bed of free-
standing reinforced-concrete standards from deformations. Soil Mech Found Eng 14: 23–27
72. Kananyan A S, Nikitenko M I (1973) Experimental studies of functioning of bases of hori-
zontally loaded rigid pillars. In: Engineering structures and theory of buildings: Bases, foun-
dations, and soil mechanics. Vysheyshaya shkola, Minsk 2: 185–194 (in Russian)
[email protected]
73. Kandaurov I I (1988) Mechanics of grain media and its application in engineering. Stroyiz-
dat, Leningrad (in Russian)
74. Kanwal R P, Sharma D L (1976) Singularity methods for elasto-statics. J Elasticity 6:
405–418
References 497

75. Karasev O V, Talanov G P, Benda S F (1977) Investigation of the work of single situ-cast
piles under different load combinations. Soil Mech Found Eng 14: 173–177
76. Karasudhi P, Rajapakse R K N D, Hwang B Y (1984) Torsion of a long cylindrical elastic
bar partially embedded in a layered elastic half space. Intern J Solids Struct 20: 1–11
77. Keer L M (1975) Mixed boundary value problems for a penny-shaped cut. J Elasticity 5:
89–98.
78. Khazin V I (1984) Foundation structures for rural buildings on short pyramidal piles. Budi-
velnyk, Kyiv (in Russian)
79. Khazin V I (1990) Study of functioning of new compaction piles for rural buildings. In:
Bases and foundations at geological conditions of the Urals. PPI, Perm, pp. 23–24 (in Rus-
sian)
80. Khazin V I, Khazin S V (1995) Study of functioning of anchor piles for pulling load. In:
Proc 2nd Ukr Sci Conf Soil Mech Effect Eng. Pt 1. Effective foundations erected without
soil excavation. PTU, Poltava, pp. 56–59 (in Russian)
81. Kholmyanskii M L (1991) Consideration of the effect of embedment depth and lateral back-
filling of machine foundations on the dynamic properties of their beds. Soil Mech Found
Eng 28: 223–228
82. Kholmyanskii M L (1993) Effect of the complex shape of the underground part on the
dynamical properties of deepened foundations for machines. Izv Vuzov Stroit (issue 2): 101–
104 (in Russian)
83. Khozhmetov G Kh, Shakasymov Sh (1986) Interaction of complex foundations with bases.
Izv AN UzSSR Ser Tekhn (issue 2): 44–46 (in Russian)
84. Kirilin V M (1975) Calculation of piles for horizontal and momental loads using the possible
displacement method. Izv AN KazSSR Ser Fiz-Mat (issue 3): 53–58 (in Russian)
85. Kirillov V M (1983) Plastic flow of soils. Izv Vuzov Stroit Arkhit (issue 5): 20–25 (in Rus-
sian)
86. Kirillov V M (1986) Nonlinear pile settlement. Izv Vuzov Stroit Arkhit (issue 6): 26–31 (in
Russian)
87. Kirillov V S (1980) Bases and foundations. Transport, Moscow (in Russian)
88. Kiselev V A (1976) Engineering mechanics. Stroyizdat, Moscow (in Russian)
89. Klein G K (1977) Engineering mechanics of granular bodies. Stroyizdat, Moscow (in Rus-
sian)
90. Klein G K, Cherkasov I I (1985) Foundations of urban transport structures. Stroyizdat,
Moscow (in Russian)
91. Klimanov V I, Litvinenko A G, Kavayeva V P (1988) Conical shell foundations. Stroyizdat,
Moscow (in Russian)
92. Kofman M M, Gumerov A G, Bikbayev A Z, Molodtsov G I (1993) Erection of tubular
piles in frozen soils using explosion energy. In: Functioning of oil industry equipment and
pipelines. Institute of Problems of Power Resources for Transport, Ufa, pp. 100–108 (in
Russian)
93. Kogan B M, Zinchenko V D (1960) Stressed state of a nonhomogeneous layer, resting on an
elastic half-space. Izv Vuzov Stroit Arkhit (issue 3): 8–18 (in Russian)
94. Komzina A A (1984) Foundations of agricultural buildings. Kuibyshev State University,
Kuibyshev (in Russian)
95. Konovalov P A (1988) Soil bases and foundations of buildings under reconstruction. Stroy-
izdat, Moscow (in Russian)
96. Korzh I V, Buslov A S (1972) Experience of application of short bored piles. Transport Stroit
(issue 9): 49 (in Russian)
97. Kosorukov I I (ed) (1974) Pile works. Vysshaya Shkola, Moscow (in Russian)
[email protected]
98. Kosterin E V (1978) Bases and foundations. Vysshaya Shkola, Moscow (in Russian)
99. Kravtsov V N (1990) Experimental studies of driven pyramidal piles with T-shaped section.
In: Bases and foundations in the geological conditions of the Urals. PPI, Perm, pp. 14–16
(in Russian)
498 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

100. Kryukov K P, Kurnosov A I, Novgorodtsev B P (1975) Construction and design of metal

and ferroconcrete power line supports. Energiya, Leningrad (in Russian)
101. Kushner S G (1990) Calculation of settlements of bases under buildings and structures. Budi-
velnyk, Kyiv (in Russian)
102. Lapshin F K (1979) Calculation of piles according to limiting conditions. Saratov State
University, Saratov (in Russian)
103. Lazarev G B (1977) Use of pyramidal piles for greenhouse foundations. Soil Mech Found
Eng 14: 20–22
104. Ledenev V V (1973) Study of interaction of a pier foundation, fixed in the soil, with the base
under an off-centre load. Ph D thesis, Novocherkassk (in Russian)
105. Ledenev V V (1978) Experience of implementation of bored pile foundations at the construc-
tion of buildings with a spacer-type scheme. In: Mechanics of soils, bases and foundations.
Voronezh State University, Voronezh, pp. 103–109 (in Russian)
106. Ledenev V V, Shelyapin R S (1973) Approximate determination of the settlement of a
rigid rectangular deep foundation by their corner point method. Soil Mech Found Eng 10:
22–24
107. Levachev S N, Fedorovskiy V G, Kurillo S V, Kolesnikov Yu M (2002) Piles in hydrotech-
nical engineering. Balkema, Rotterdam
108. Levenstam V V (1973) Study of interaction of short pile foundations with soil under a hori-
zontal load. Ph D thesis, Rostov-na-Donu (in Russian)
109. Likhovtsev V M (1976) Displacement and contact pressures for a rigid sunken plate. Soil
Mech Found Eng 13: 414–419
110. Likhovtsev V M (1978) Deformation of an elastic half-space with a cylindrical cutout. In:
Bases, foundations and underground structures. Tr NIIOSP 69: 73–78 (in Russian)
111. Luchkovskii I Ya, Lekumovich G S (1971) Analysis of laterally loaded piles in cohesive
soils. Soil Mech Found Eng 8: 184–187
112. Luco J E (1976) Torsion of a rigid cylinder embedded in an elastic half space. Trans ASME
J Appl Mech 43: 419–423
113. Malatsidze E G, Slyusarenko S A (1977) Design of foundations of short piles operating at
horizontal loads. Bases and foundations. Budivelnyk, Kyiv 10: 69–72 (in Russian)
114. Mal’tsev A T, Sazhin V S (1980) Behavior of short piles in collapsible soils under the action
of inclined forces. Soil Mech Found Eng 17: 235–239
115. Manual for the design of bases for buildings and structures (for the Construction Rules and
Regulations 2.02.01-83) (1986) Gersevanov NIIOSP, Stroyizdat, Moscow (in Russian)
116. Milligan G W E, Houlsby G T, Onisi Yu et al. (1991) Software for soil mechanics and
foundation engineering. Likhovtsev V M (ed). Stroyizdat, Moscow (in Russian)
117. Mironov V S (1955) Practical method for calculation of piles for the action of horizontal
loads. Izv Vuzov Stroit Arkhit (issue 5): 24–32 (in Russian)
118. Mironov V S, Krovyakov V N (1978) Testing of batter piles under the combined effect of
vertical and horizontal loads. Soil Mech Found Eng 15: 299–303
119. Mironov V S, Krovyakov V N (1980) Experimental studies of resistance of vertical and
inclined piles to the action of inclined loads. Izv Vuzov Stroit Arkhit (issue 8): 123–126 (in
Russian)
120. Molenkamp F (1981) Elasto-plastic double hardening model MONOT. LGM ReportCO-
218595. Geotechnics, Delft
121. Mura T (1982) Micromechanics of defects in solids. Nijhoff, the Hague/Boston
122. Mura T (1988) Inclusion problems. Appl Mech Rev 41: 15–20
123. Murzenko Yu N (1989) Computer-aided calculation of bases fir buildings and structures at
the elastoplastic stage of functioning. Stroyizdat, Leningrad (in Russian)
[email protected]
124. Nikolaevsky V N, Kuznetsov A S, Bellendir E N (1991) Mathematical dilatancy theory and
conditions at strong discontinuities. Intern J Eng Sci 29: 1375–1389.
125. Oding B S (1968) Investigation of the stressed state and deformations of soil at the transfer
of load by means of a pile. Izv Vuzov Stroit Arkhit (issue 10): 38–42 (in Russian)
References 499

126. Oding B S (1984) On the interaction of a conical pile with soil. In: Investigation of rational
foundation structures. Voronezh State University, Voronezh, pp. 67–74 (in Russian)
127. Oding B S, Shelyapin R S (1964) Application of Mindlin’s solution for the determination
of stress around piles. In: Theory of buildings and structures, No. 10(issue 1): 134–153 (in
Russian)
128. Ofrikhter V G (1996) Numerical modelling of interaction of pile foundations with the sur-
rounding soil. In: Proc 5th Inter Conf Problems Pile Found Eng, RNKMGF, Moscow, vol.
1, pp. 124–126 (in Russian)
129. Ogranovich A B (1989) Calculation of a rigid pyramidal pile for a horizontal load in an
elastic half-space. Stroit Mekh Raschet Sooruzh (issue 4): 7–9 (in Russian)
130. Ogranovich A B (1990) Calculation of settlement for a variable cross-section pile. Izv Vuzov
Stroit Arkhit (issue 12): 97–100 (in Russian)
131. Ogranovich A B (1991) Consideration of soil discontinuities in the analysis of tapered piles
under horizontal loading. Soil Mech Found Eng 28: 40–43
132. Ogranovich A B (1992) Calculation of a deepened beam on an elastic base (spatial problem).
Izv Vuzov Stroit (issue 9–10): 25–28 (in Russian)
133. Ogranovich A B (1992) Numerical method of calculation of settlement and slope of a rigid
foundation with a rectangular bottom, deepened into an elastic half-space. Izv Vuzov Stroit
(issue 1): 114–117 (in Russian)
134. Ogranovich A B (1992) Settlement of a round rigid foundation, deepened into an elastic
half-space. Izv Vuzov Stroit (issue 11–12): 29–33 (in Russian)
135. Pak R Y S, Gobert A T (1989) Computational aspects in a mathematical analysis of an
inclusion problem. Proc Can Congr Appl Mech 2: 542–543
136. Pavlov V V (1992) Slotted foundation of buildings. Stroyizdat, Krasnoyarsk (in Russian)
137. Pavlov V V, Katsov K P, Smorodinov M I (1992) Slit foundations of industrial and civic
buildings in the middle Urals. Soil Mech Found Eng 29: 140–143
138. Perley Ye M (1969) Tubular ferroconcrete piles for industrial and civil engineering. Stroyiz-
dat, Leningrad (in Russian)
139. Petrenko G M, Orobchenko P A, Tsymbal S I (1972) Data for the calculation of piles from
the base deformations. In: Osnovaniya i fundamenty. Budivelnyk, Kyiv (issue 5): 78–85 (in
Russian)
140. Petrenko G M, Vasilenko A Yu (1972) On the methods of determination of stress in the
base of a hanging pile. In: Bases and foundations. Budivelnyk, Kyiv (issue 5): pp. 20–26 (in
Russian)
141. Pilyagin A V (1992) Nonlinear methods deformation-based of calculation of bases of foun-
dations of various type. In: Efficiency of foundation design solutions, Mariy Polytechnical
Institute, Yoshkar-Ola, pp. 94–99 (in Russian)
142. Pilyagin A V, Kazantsev S V, Kazantseva N N (1985) Elastoplastic calculation of a soil base
at the conditions of a spatial stress-strained state. Mariy Polytechnical Institute, Yoshkar-Ola,
VNIIIS No. 5713 (in Russian)
143. PLAXIS – Finite Element Code for Soil and Rock Analyses. Ver. 7. General Information
and Tutorial Manual (1998) Balkema, Rotterdam
144. Ponomarev A B (1988) Study of bearing capacity of hollow conical piles. In: Bases and
foundations in the geological conditions of the Urals. PPI, Perm, pp. 29–33 (in Russian)
145. Popovich A P, Koldyrev V I (1991) Spatial foundations of structural type. In: Spatial con-
structions in Krasnoyarsk region. KISI, Krasnoyarsk, pp. 107–116 (in Russian)
146. Poulos H G (1971) The behaviour of laterally loaded piles: I. Single piles. J Soil Mech
Founds Divn ASCE 97: 711–731
147. Poulos H G (1971) The behaviour of laterally loaded piles: II. Pile groups. J Soil Mech
[email protected]
Founds Divn ASCE 97: 733–751
148. Poulos H G (1975) Torsional response of piles. J Geotech Eng ASCE 101: 1014–
1035
149. Poulos H G, Davis E H (1980) Pile foundation analysis and design. Wiley, New York
500 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

150. Poulos H G, Davis E N (1968) The settlement behaviour of single axially-loaded incom-
pressible piles and piers. Geotechnique 18: 351–371.
151. Rabotnikov A, Zats S (1970) Experience of application of short bored piles. Promysh Stroit
Inzh Sooruzh (issue 6): 6–7 (in Russian)
152. Rajapakse R K N D (1988) The interaction between a circular elastic plate and a transversely
isotropic elastic halfspace. Intern J Num Anal Method Geomech 12: 419–436.
153. Rajapakse R K N D, Selvadurai A P S (1985) Torsional stiffness of non-uniform and hollow
rigid piers embedded in isotropic elastic media. Intern J Num Anal Method Geomech 9:
1213–1229.
154. Rajapakse R K N D, Selvadurai A P S (1989) Torsion of foundations embedded in a non-
homogeneous soil with a weathered crust. Geotechnique 39: 485–496
155. Rajapakse R K N D, Selvadurai A P S (1991) Response of circular footings and anchor
plates in non-homogeneous elastic soils. Intern J Num Anal Method Geomech 15: 457–470
156. Randolph M F (1981) Piles subjected to torsion. J Geotech Eng ASCE 107: 1095–1111.
157. Rausch E (1959) Maschinenfundamente und Andere Dynamisch Beanspruchte Baukon-
struktionen. VDI-Verlag, Düsseldorf
158. Recommendations on design and construction of slotted foundations (1982) Gersevanov
NIIOSP, Moscow (in Russian)
159. Recommendations on design and construction of slotted foundations in the conditions of the
Central Urals (1990) UPI, Sverdlovsk (in Russian)
160. Recommendations on the calculation of bases under foundations with an inclined bottom
(1983) Gersevanov NIIOSP, Moscow (in Russian)
161. Recommendations on the calculation of nonlinear settlements of deep supports using a com-
puter (1974) Gersevanov NIIOSP, Moscow (in Russian)
162. Recommendations on the rational usage of piles of various type in engineering (1982)
Gersevanov NIIOSP, Moscow (in Russian)
163. Redkov V I (1982) Calculation of the settlement of a round rigid punch, deepende into an
elastic half-space. In: Bases and foundations. PPI, Perm, pp. 44–51 (in Russian)
164. Reisner E, Sagoci H (1944) Forced torsional oscillation of an elastic half-space. J Appl Phys
15: 652–662.
165. Rower R K, Booker J R (1979) A method of analysis of horizontally embedded anchors in
an elastic soil. Intern J Num Anal Method Geomech 3: 187–203.
166. Rybalko Yu Ya (1968) Foundations of short bored post piles. Promysh Stroit (issue 6): 46
(in Russian)
167. Ryzhenko A P (1973) Work of a pile raft of bored piles for the action of horizontal forces.
Transport Stroit (issue 4): 43–44 (in Russian)
168. Sakharov A S, Boyko I P, Kozak A L (1983) Summary of the Rosinka software. Osnovaniya,
Fundamenty Mekh Gruntov (issue 5): 7 (in Russian)
169. Saprykina N M (1977) Application of short bored piles in rural engineering. Promysh Stroit
(issue 9): 25 (in Russian)
170. Savchenko F M, Oding B S (1988) Economically efficient foundations under strutted
structures of agricultural buildings. In: Pile foundation studies. Voronezh State University,
Voronezh, pp. 162–167 (in Russian)
171. Savinov O A (1979) Modern design of foundations under machines and their calculation.
Stroyizdat, Leningrad (in Russian)
172. Sazhin V S, Shishkin V A, Zhanalinov B N (1991) Design and technology of erection of
foundations on heaving soils. Gylym, Almaty (in Russian)
173. Schofield A, Wroth P (1968) Critical state soil mechanics. McGraw-Hill, London.
174. Segerlind L J (1976) Applied finite element analysis. John Wiley & Sons, New York
[email protected]
175. Selvadurai A P S (1976) The load-deflexion characteristics of a deep rigid anchor in an
elastic medium. Geotechnique 26: 603–612
176. Selvadurai A P S (1979) An energy estimate of the flexural behaviour of a deep rigid anchor
embedded in an isotropic elastic medium. Intern J Num Anal Method Geomech 3: 285–292.
References 501

177. Selvadurai A P S (1979) On the displacement of a penny-shaped rigid inclusion embedded

in a transversely isotropic elastic medium. Solid Mech Arch 4: 163–172.
178. Selvadurai A P S (1980) The eccentric loading of a rigid circular foundation embedded in
an isotropic elastic medium. Intern J Num Anal Method Geomech 4: 121–129
179. Selvadurai A P S (1984) Elastostatic bounds for the stiffness of an elliptical disk inclusion
embedded at a transversely isotropic bi-material elastic interface. J Appl Math Phys (ZAMP)
35: 64–77.
180. Selvadurai A P S (1984) Torsional stiffness of rigid piers embedded in isotropic elastic
soils. Laterally loaded deep foundations: analysis and performance. ASTM STP 835: 49–
55
181. Selvadurai A P S (1989) The influence of a boundary fracture on the elastic stiffness
of a deeply embedded anchor plate. Intern J Num Anal Method Geomech 13: 159–
170.
182. Selvadurai A P S (1993) The axial loading of a rigid circular anchor plate embedded in an
elastic half-space. Intern J Num Anal Method Geomech 17: 343–353.
183. Selvadurai A P S (1994) A unilateral contact problem for a rigid disk inclusion embedded
between two dissimilar elastic half-spaces. Q J Mech Appl Math 47: 493–510.
184. Selvadurai A P S, Au M C (1986) Generalized displacements of a rigid elliptical anchor
embedded at a bi-material geological interface. Intern J Num Anal Method Geomech 10:
633–652
185. Selvadurai A P S, Bauer G E, Nicholas T J (1980) Screw plate testing of a soft clay. Can
Geotechn J 17: 465–472
186. Selvadurai A P S, Nicholas T J (1979) A theoretical assessment of the screw plate test. In:
Wittke W (ed) Proc 3rd Intern Conf Numerical Methods Geomech 3: 1245–1252
187. Selvadurai A P S, Rajapakse R K N D (1987) Variational scheme for analysis of torsion of
embedded nonuniform elastic bars. J Eng Mech ASCE 113: 1534–1550.
188. Selvadurai A P S, Singh B M (1986) The axial displacement of a disc inclusion embedded
in a penny-shaped crack. J Appl Math Phys (ZAMP) 37: 64–77
189. Selvadurai A P S, Singh B M, Au M C (1990) The in-plane loading of a rigid disk inclusion
embedded in an elastic halfspace. Trans ASME J Appl Mech 58: 362–369
190. Shadunts E G (1973) Calculation of one-row pile foundations with a pile raft, resting on the
soil. Izv Vuzov Stroit Arkhit (issue 9): 22–27 (in Russian)
191. Shadunts K Sh, Podetlkov V V (1990) Arch foundations and their calculation with the
account of sliding on the base. In: Studies and developments on computer-aided design of
foundation and bases. NPI, Novocherkassk, pp. 87–95 (in Russian)
192. Shapiro D M (1985) Practical method of analysis of bases and earth structures in nonlinear
formulation. Soil Mech Found Eng 22: 182–187
193. Shapiro D M (1993) Mathematical modelling of limiting states of bored piles. In: Nonlinear
soil mechanics. Proc 4th Russ Conf St Petersburg 1: 140–145 (in Russian)
194. Shapiro D M, Poltorak G V (1989) Program for elastoplastic numerical calculation of soil
structures and structures, interacting with soil. Osnovaniya Fundamenty Mekh Gruntov
(issue 5): 20 (in Russian)
195. Shapiro D M, Poltorak G V (1990) Implementation of nonlinear calculation method at the
design of bases and soil structures. In: Nonlinear methods of calculation of bases and foun-
dation. Mariy Polytechnical Institute, Yoshkar-Ola, pp. 24–27 (in Russian)
196. Shapiro D M, Zotsenko N L, Beda S V (1996) Elastoplastic calculation of bearing capacity
of piles. Izv Vuzov Stroit Arkhit (issue 6): 34–39 (in Russian)
197. Shekhter O Ya (1956) On the calculation of a deepened rigid foundation. In: Mechanics of
soils. Tr NIIOSP 30: 45–60 (in Russian)
[email protected]
198. Shelyapin R S (1965) Approximate determination of settlement of a rigid round deepened
foundation. Izv Vuzov Stroit Arkhit (issue 6): 11–18 (in Russian)
199. Shvets N S, Sedin V L, Kirichek Yu A (1987) Structural methods of reduction of vibrations
of foundations for machines with dynamical loads. Stroyizdat, Mocsow (in Russian)
502 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

200. Shvets N S, Zakhvatkin M P (1995) Study of functioning of piles of non-traditional shape

at horizontal dynamic load. In: Proc 2nd Ukr Sci Conf Soil Mech Found Eng. Pt 1: Efficient
foundations, erected without soil excavation. PTU, Poltava, pp. 69–71 (in Russian)
201. Shvets V B, Ginzburg L L, Goldshtein V M et al. (1987) Handbook on mechanics and
dynamics of soils. Budivelnyk, Kyiv (in Russian)
202. Silin K S, Glotov N M, Zavriev K S (1981) Design of deep foundations. Transport, Moscow
(in Russian)
203. Simvulidi I A (1978) Calculation of engineering structures on an elastic base. Vysshaya
Shkola, Moscow (in Russian)
204. Sivtsova Ye P (1963) Calculation of settlement of a single pile with the account of the tip
functioning. Tr NIIOSP 53: 47–66 (in Russian)
205. Smorodinov M I (1983) Anchors in engineering. Stroyizdat, Moscow (in Russian)
206. Smorodinov M I, Fedorov B S (1986) Construction of structures and foundations using a
"wall in the soil" method. Stroyizdat, Moscow (in Russian)
207. SNIP (Construction Rules and Regulations) 2.02.01-83. Bases for buildings and structures
(1985) Stroyizdat, Moscow (in Russian)
208. SNIP (Construction Rules and Regulations) 2.02.03-85. Pile foundations (1986) Stroyizdat,
Moscow (in Russian)
209. Snitko N K, Snitko A N (1967) Calculation of rigid and flexible supports embedded in the
ground under the simultaneous action of horizontal and vertical forces. Soil Mech Found
Eng 4: 153–156
210. Solovyov Yu I, Vlasov Yu V (1974) Investigations and engineering calculation method of
short piles in subsidence soil on the action of horizontal loads. In: Engineering and geolog-
ical conditions and specific features of foundation engineering in Siberia. Proc NIIZHTA,
Novosibirsk 152: 108–119 (in Russian)
211. Sorochan E A (1986) Foundations of industrial buildings. Stroyizdat, Moscow (in Russian)
212. Sorochan E A, Gruodis R Yu (1983) Performance of cast-in-place concrete foundations in
interaction with the soil. Soil Mech Found Eng 20: 204–208
213. Sorochan E A, Li E A (1993) Investigation of the operation of pyramidal piles in swelling
soils. Soil Mech Found Eng 30: 42–46
214. Sorochan E A, Piven V G, Rybnikov A M (1991) Monolithic foundations with an effective
lateral surface. Soil Mech Found Eng 28: 99–102
215. Sorochan E A, Trofimenkov Yu G (ed) (1985) Bases, foundations and underground struc-
tures: Designer’s handbook. Stroyizdat, Moscow (in Russian)
216. Sotnikov S N (1992) On the evaluation of reliability of results of calculation of final settle-
ment of bases under buildings and structures. In: Erection and reconstruction of foundations
on weak soils. SPBISI, St. Peterburg, pp. 5–13 (in Russian)
217. Ten I A (1964) Effect of friction forces at the calculation of foundations with the account of
fixation. Transport Stroit (issue 10): 39–41 (in Russian)
218. Tetior A N, Felkin V I, Surguchev V G (1981) Design of foundations. Budivelnyk, Kyiv (in
Russian)
219. Timoshenko S P, Goodier J N (1970) Theory of elasticity, 3rd edn. McGraw-Hill, New York
220. Tishin V G (1985) Bases and foundations of objects of oil and gas industry. Stroyizdat,
Moscow (in Russian)
221. Trofimenkov Yu G, Vorobkov L N (1981) Field methods for the studies of constructive
properties of soils. Stroyizdat, Moscow (in Russian)
222. Tsiprianovich I V (1971) On the calculation of piles with widenings for horizontal load
from the base deformations. In: Bases and foundations. Budivelnyk. Kyiv 4: 122–127 (in
Russian)
[email protected]
223. Ukhov S B (1973) Calculation of structures and bases using the finite-element method. MISI,
Moscow (in Russian)
224. Ukhov S B, Semenov V V, Znamenskii V V et al. (1994) Soil mechanics, bases and founda-
tions. Association of Engineering Institutions, Moscow (in Russian)
References 503

225. Ulitskiy V M, Shashkin A G (1999) Supervision of urban reconstruction (investigation, cal-

culations, works implementation, monitoring). ASV, Moscow (in Russian)
226. Urban I V (1939) Calculation of thin walls with the account of elastic properties of the soil
and the wall. In: Issues of Engineering Mechanics. MIIT Works 55: 43–69 (in Russian)
227. Vershinin V P, Kovalev I P, Chelnokov Ye L (1978) Bored pile foundations with support
widenings. Stroyizdat, Leningrad (in Russian)
228. Veryuzhskiy Yu V (1978) Numerical methods of potential in some problems of applied
mechanics. Vyshcha Shkola, Kyiv (in Russian)
229. Vyalov S S (1978) Rheological principles of soil mechanics. Vysshaya shkola, Moscow (in
Russian)
230. Wet J (1966) Application of energy hypothesis in soil mechanics. In: Soil mechanics and
foundation engineering. Proc 5th Intern Conf Stroyizdat, Moscow, pp. 105–114 (in Russian)
231. Williams I, Hiecks M A (1992) Finite-Elemente-Prognose für ein schrag belastetes Funda-
ment. Geotechnik 15: 66–72.
232. Wilson E L (1965) Calculation of strength of axially symmetrical bodies. Raket Tekh Kos-
monavt (issue 12): 124–131 (in Russian)
233. Winnicki L A, Zienkiewicz O C (1979) Plastic (or visco-plastic) behaviour of axisymmetric
bodies subjected to nonsymmetric loading – semi-analytical finite element solution. Intern J
Num Method Eng 14: 1399–1412
234. Yagudin A M, Druzhinin G A (1968) Foundations for buildings of driven piles with a radial
widened footing. Promyshlennoye Stroitelstvo (issue 11): 19–22 (in Russian)
235. Yarutin V M (1974) Flow-line construction of cast-in-place pile foundations of farm build-
ings. Soil Mech Found Eng 11: 375–380
236. Zabylin M I, Linovskii S V, Nuzhdin L V (1991) Design of pile foundations under machines.
NISI, Novosibirsk (in Russian)
237. Zakhvatkin M P, Shvets N S (1989) Studies of functioning of piles of various shape under
dynamic loads. In: Foundations under machines with dynamic loads. Proc Intern Symp
Stroyizdat, Leningrad, pp. 195–198 (in Russian)
238. Zaretskii Yu K, Karabaev M I (1985) Limit-state design of cast-in-place piles. Soil Mech
Found Eng 22: 169–175
239. Zaretskii Yu K, Karabaev M I (1987) Analysis of settlements of bored-cast-in-place piles in
collapsible soils. Soil Mech Found Eng 24: 13–19
240. Zaretskii Yu K, Lombardo V N (1983) Statics and dynamics of earth-fill dams. Energoat-
omizdat, Moscow (in Russian)
241. Zaretskii Yu K, Orekhov V V, Karabayev M I (1984) Application of the finite-element
method to the calculation of bored pile foundations. In: Modern computer-based methods
of calculation of hydropower and nuclear power plants. Tr Gidroproekta im S Zhuka 100:
3–10 (in Russian)
242. Zavriev K S (1976) Approximate method of designing piles for horizontal loading and deter-
mining their flexibility. Soil Mech Found Eng 13: 152–156
243. Zavriev K S, Kryukov Ye P, Shpiro G S (1960) Study of the bearing capacity of foun-
dations under contact line suports. Tr VNIITS. Transzheldorizdat, Moscow, 39: 1–216 (in
Russian)
244. Zavriev K S, Shpiro G S (1970) Calculation of foundations under deep bridge piers. Trans-
port, Moscow (in Russian)
245. Zhemochkin B N (1948) Calculation of the elastic fixation of a rod (rod bending in an elastic
half-space). Stroyizdat, Moscow (in Russian)
246. Zhemochkin B N, Sinitsyn A P (1962) Practical methods of calculation for beams and plates
on an elastic base without Winkler hypothesis. Gosstroyizdat, Moscow (in Russian)
[email protected]
247. Zhukov N V, Dondysh A M, Pogosyan Z G et al. (1972) Short piles in rural engineering.
ONTI CNIIEPSelstroy, Moscow (in Russian)
248. Zienkiewicz O C (1971) The finite element method in engineering science. McGraw-Hill,
London
504 5 Calculation of Bases for Rigid Complex-Shaped Deepened Foundations

249. Ziyazov Ya Sh (1977) On the calculation of pyramidal piles in a nonhomogeneous base

under a vertical and a horizontal load. In: Foundation engineering issues. Niipromstroy, Ufa
21: 40–45 (in Russian)
250. Zotsenko N L (1988) On the calculation of pyramidal piles under a combined action of ver-
tical and horizontal loads. In: Pile foundation studies, Voronezh State University, Voronezh,
pp. 94–102 (in Russian)
251. Zureick A H (1988) Transversely isotropic elastic medium with a rigid spheroidal inclusion
under an axial pull. Trans ASME J Appl Mech 55: 495–497

[email protected]
Conclusions

In modern geotechnical engineering, field and laboratory experiments have become

complicated, durable and expensive, and the simplified formulae of soil mechan-
ics, based on analytical solutions of separate problems, no longer correspond to
the increased requirements of design. In various fields of science and engineering
computational experiments have been widely introduced, being a powerful tool for
studies of engineering problems. Computational experiment plays an important role
in soil mechanics and foundation engineering, hence once can claim a new field of
knowledge, computational geotechnics, to have been formed. The aim of compu-
tational geotechnics is modelling of stress-strained states of soil bases and founda-
tions by numerical solutions of equations of mathematical physics. For the design
of engineering constructions, the computational experiment enables real situations
to be predicted and practical economical effect to be achieved.
Boundary-element method, being developed in this study, due to its visual char-
acter and indifference to the object geometry, is undoubtedly an efficient numerical
method for solving many spatial problems in geotechnical engineering.
Evident advantages of the boundary-element method are revealed first of all at
the studies of stress-strained states in unlimited domains (half-space, constant- and
variable-thickness layers, etc.) which serve as models for soil bases in geotechnical
problems.
Using flat boundary elements of triangular or quadrangular shape, one can rather
accurately reproduce the surface of contact with soil practically for all existing
foundation structures. Besides, in comparison with other popular numerical meth-
ods (e.g., finite-difference or finite-element methods, requiring discretization of not
only the boundaries, but also of the interior of the calculation domains), the base free
surface does not need discretization into boundary elements and, consequently, the
input data amount is considerably reduced and their preparation is much simplified.
The studies performed have shown that the universality of the boundary-element
method can be compared with that of the finite-element method, which is generally
considered to be the main method of structural analysis in various fields of science
and engineering. Meanwhile, in spatial problems of a foundation structure contact
[email protected]
with a linearly deformable soil base, application of the boundary-element method
is more reasonable and more efficient for practical purposes since the boundary-
element algorithms do not require large computation time and memory size.

S.M. Aleynikov, Spatial Contact Problems in Geotechnics, Foundations of 537

Engineering Mechanics, DOI 10.1007/b11479_BM2,  C Springer-Verlag Berlin Heidelberg 2011
538 Conclusions

Test and model calculations, performed using the Rostwerk software described
here, have been compared for the problems having an analytical solution or for those
which had been also solved by other approximate methods.
Extensive methodological calculations show a good convergence of the numer-
ical results and high accuracy of the boundary-element method, sufficient for the
practical purposes, even at piecewise constant approximation of the contact stress
field. On the other hand, the obtained estimation of the boundary-element method
calculation accuracy is of great importance for practical engineering, especially due
to the necessity of setting the applicability limits for various empirical formulae,
recommended in handbooks.
One should note that an important feature of the developed boundary-element
algorithms is their applicability for a wide range of spatial contact models (influ-
ence functions) – both the existing ones and those to be developed. The results
of calculations, using the available methods, enable the suitability of the influence
functions to be estimated, their features to be revealed at the modelling of spatial
contact interaction processes.
Due to their sufficient universality, the developed boundary-element method
without any essential additional processing can be applied for the calculation of base
deformations for practically all existing foundation structures, used in industrial,
civil, and transport engineering, including non-traditional complex-shaped rigid
foundation structures. This enables the foundation base calculations to be unified
and development of special calculation methods for each foundation structure type
and loading conditions to be avoided. From the boundary-element calculations the
designer can see the whole pattern of the spatial stress-strained state of the soil base
for foundations of various depth, subject to a spatial load of a general type. In com-
parison with the engineer-oriented methods and regulations being currently in use,
the boundary-element calculation methods enable a discrepancy between the theo-
retical and experimental data to be considerably reduced; they lead to more econom-
ical design solutions enabling the pressure, transmitted to the soil, to be increased
and the foundation size to be decreased. All this enables the methods developed to
be recommended for application at the calculations of base deformations for both
high-priority foundation structures and large-scale housing construction. A consid-
erable increase of computer speed and memory parameters and application of an
efficient numerical method are the reasons for an optimistic view in case a large
amount of calculation work is to be carried out in order to determine stress-strained
states in three-dimensional active areas of bases for complex-shaped foundation
structures.
The presented analysis of numerous calculation examples shows rather con-
vincingly wide opportunities being open while solving spatial contact problems by
boundary-element method in engineering mechanics, soil mechanics, mathematical
physics, etc. Simultaneously, the ideas and methods of numerical solution of integral
equations of mechanics of deformable solids, presented in the book, set a number of
[email protected]
problems which have not been fully solved yet and still require further studies. Note
some of the promising trends that, in our opinion, need further theoretical develop-
ment and implementation into practical engineering.
Conclusions 539

1. Further development and analysis of contact models of soil bases of anisotropic

and nonhomogeneous half-space and layer type are required, what will enable, using
the BEM, the effect of the soil base mechanical properties on the processes of spatial
contact interaction to be studied.
2. Undoubtedly important is the improvement of calculation schemes for spa-
tial contact deformation (account of the contact friction, effect of the soil intrinsic
weight, presence of initial stresses, finite rigidity of foundation structures at their
interaction with soils, etc.) in order to have a distinct idea about the limits of ratio-
nal application of boundary-element methods in geotechnical engineering. It is espe-
cially helpful to determine the conditions, for which more complicated calculation
schemes are required in order to take into account the essential features of contact
deformation processes.
3. The progress in computer industry along with the application of the boundary-
element method opens the possibility for a transition from simple calculation of
deformation of bases of complex foundation structures to the search of optimal solu-
tions, i.e. to the application of optimization methods in design process. Mathemati-
cal optimization methods are rather effectively used in various fields of science and
engineering; however, their implementation for optimization of foundation struc-
tures has just begun. The reason for this is the fact that solution of the optimization
problem for foundations of various depth is very cumbersome: even a single calcu-
lation of contact interaction characteristics is related to considerable computational
expenses which are multiplied by any iterative process. In the present study, the
search for optimal solutions was performed only for shallow foundation structures,
resting on non-uniformly compressible bases. The calculation models and efficient
boundary-element algorithms proposed here, will further become a reliable basis for
the development of mathematical means for computer-aided design of foundation
structures.
4. Application of the developed boundary-element methods seems promising for
the solution of essentially spatial contact problems of inelastic deformation (with
the account of elastoplastic properties, viscoplasticity, creep, etc.) which have been
studied quite insufficiently yet. Elastoplastic solutions in the problems of engi-
neering mechanics and mechanics of deformable solids can be at small increments
implemented by iterative processes on the basis of a well-known method of “elastic
solutions”. Solutions of spatial dynamic problems, problems of viscoplasticity and
creep can be obtained according to a stepwise procedure of integration over time.
Some approaches to the solution of inelastic deformation problems by boundary-
element method (as a rule, in a two-dimensional formulation or for finite-size bod-
ies) have been already carried out in a number of publications. In future, while
solving spatial contact problems, using the boundary-element method for rheolog-
ically complicated bases, one can expect application of a formalism with a ficti-
tious matrix of “elastic equivalent”, reducing the decisive equations in increments
to the relations of linear theory of elasticity (in the general case, anisotropic and
[email protected]
non-uniform).
5. In the future studies of the base+foundation+structure system, based on the
proposed boundary-element methods, it is natural to perform the three-dimensional
540 Conclusions

analysis of rigidity coefficients for complex-shaped punches, deepened into

the elastic base.
Elastic properties of soil bases are known to be quite effectively described by
a rigidity coefficient matrix. This is a symmetrical, positively determined square
matrix of the sixth order. Its elements are reactions due to unit progressive displace-
ments or rotations of a rigid punch. Application of the rigidity coefficient matrix
enables the number of degrees of freedom due to the elastic base reaction to be
reduced to six and, consequently, essentially simplifies calculation of foundation
structures from the base deformations. In the cases when undular processes in the
soil can be neglected, application of the rigidity coefficients is possible not only for
static, but also for dynamical problems. Application of rigidity coefficients of soil
bases is efficient for static calculations as well as the calculations of vibration of
machine foundations when the complex shape of their below-grade part should be
taken into account. The problems of determination of the rigidity coefficients are
rather important at the account of soil-mediated interaction of parts of foundations
made by different techniques and at different depths (piles and a pile raft; foun-
dations on natural bases, reinforced by piles; deepened foundations, reinforced by
plates; slotted foundations, combined by a plate resting on the soil, etc.). Often the
deepened part of foundations is intentionally made more complicated in order to
reduce vibrations and to improve the contact with the soil.
In practical calculations, due to the absence of exact solutions for spatial contact
problems of theory of elasticity for complex-shaped punches with different depth,
rigidity coefficients are used only for foundations of simple shape resting on the
elastic base surface or, with the account of depth, at the flat problem conditions.
In the framework of the developed boundary-element method the rigidity coef-
ficients for complex-shaped punches, deepened into an elastic base, can be conve-
niently calculated, by setting unit loads and moments in the coordinate axes direc-
tions and subsequent inversion of the obtained matrix of displacements and rota-
tions.
6. For specific types of foundations, the results of boundary-element calcula-
tions for the determination of settlement-vs-load dependences and characteristics
of the stress-strained states in the active areas of the foundation structure bases
should be accumulated and generalized. After a detailed experimental testing at var-
ious engineering-and-geological conditions, the reliable limits of applicability of the
method can be determined, and the method can be recommended for correction and
improvement of design regulations.
7. The developed methods of numerical modelling of spatial contact interaction
are sufficiently general, applicable not only for calculation and design of founda-
tion structures, but also in other areas where methods of engineering mechanics
and mechanics of deformable solids are applied (strength calculations in machine-
building, calculation and design of road coverage, studies of mechanical properties
of materials by impression of indentors, mining, etc.). Therefore, it is important to
[email protected]
continue further improvement of the known three-dimensional calculation schemes
of the boundary-element method as well as development of new ones in order to
extend their application area, based on (i) use of boundary elements of higher order,
Conclusions 541

(ii) efficient discretization of the boundary (contact) domains, (iii) elaboration of

special methods of numerical and analytical integration of fundamental solutions,
(iv) application of the boundary-element method in combination with other numer-
ical methods of solution of problems of mathematical physics.

[email protected]
[email protected]
Appendix A
Fundamental Solutions of Spatial Theory of
Elasticity for a Homogeneous Isotropic
Half-Space

Equations (A.1–A.9)
( ( give an extended form of the Green’s displacement tensor
( (k) (
components (i ( = {U(k) , V(k) , W(k) } for a rectangular Cartesian coordinate sys-
tem (Fig. A.1)

⎧
⎪
⎪  
⎪
⎨ 3 − 4ν  1 
1 1 x12 1 1 R22 − 3x12
U (K, N) =
(1)
+ + + 3 + zζ
8π G(1 − ν) ⎪⎪ 2 R1 R2 2 3
R1 R2 R52
⎪
⎩
⎡   ⎤⎫
z2 ⎪
R2 (R2 + z2 ) − x1 2 +
2 ⎪
⎬
⎢ R2 R22 − x12 ⎥
⎥ ,
−(1 − 2ν) ⎢ 2(1 − ν)z − (1 − 2ν)
⎣ R32 ⎦ ⎪
2
R22 (R2 + z2 )2 ⎪
⎭
(A.1)

⎧
⎪
⎪  
⎪
⎨ 1 1
x1 y1 1
(1)
V (K,N) = − − + 3
8π G(1 − ν) ⎪
⎪ 2 R31 R2
⎪
⎩
⎫ (A.2)
⎡ ⎤ ⎪
2+
z2 ⎪
⎪
⎢ 1 − 2ν R2 ⎥ z2 ⎬
+(1 − 2ν) ⎣ −2(1 − ν)z2 2 ⎦ + 3zζ ,
R32 R2 (R2 + z2 )2 R52 ⎪
⎪
⎪
⎭

  
x1 1 z1 z2 2(1 − ν)(1 − 2ν)
W (K,N) =
(1)
[email protected] + 3 +
8π G(1 − ν) 2 R1 3 R2 R2 (R2 + z2 )
 (A.3)
2(1 − ν)ζ − (1 − 2ν)z z2
− − 3zζ 5 ,
R32 R2

543
544 Appendix A

Fig. A.1 Geometric

representations at the
formulation of fundamental
solutions of the spatial theory
of elasticity for a
homogeneous isotropic
half-space

⎧
⎪
⎪  
⎪
⎨ 1 1
x y
1 1 1
U (K,N) = −
(2)
− + 3
8π G(1 − ν) ⎪
⎪ 2 R31 R2
⎪
⎩
⎡ ⎫ (A.4)
⎤ ⎪
2+
z2 ⎪
⎪
⎢ 1 − 2ν ⎬
⎢ R2 ⎥ 3zζ
+ (1 − 2ν) ⎢ −2(1 − ν)z2 ⎦ + ,
⎣ R32 R22 (R2 + z2 )2 R52 ⎪
⎪
⎪
⎭
⎧
⎪
⎪  
⎪
⎨  
1 3 − 4ν 1 1 y2 1 1 R22 − 3y21
V (2)
(K,N) = + + 1 + + zζ −
8π G(1 − ν) ⎪
⎪ 2 R1 R2 2 R31 R32 R
⎪
⎩
⎫
⎡ ⎤⎪
R2 (R2 + z2 ) − y21 (2 +
z 2 ⎪
⎪
R22 − y21 ⎥⎬
)
⎢ R2
−(1 − 2ν) ⎣2(1 − ν)z2 − (1 − 2ν) ⎦
R22 (R2 + z2 )2 R32 ⎪
⎪
⎪
⎭
(A.5)
  
1 z1
y1 z2 2 (1 − ν) (1 − 2ν)
W (2)
(K,N) = + 3 +
8π G (1 − ν)
3
2 R1 R2 R2 (R2 + z2 )
 (A.6)
2 (1 − ν) ζ − (1 − 2ν) z z2
− − 3zζ 5 ,
R32 R2
 
[email protected]
x1 z1 z1 z2 2(1 − ν)(1 − 2ν)
U (K,N) =
(3)
+ (3 − 4ν) 3 + 3zζ 5 − ,
8π G(1 − ν) 2R31 2R2 R2 R2 (R2 + z2 )
(A.7)
Appendix A 545
 
(3) y1 z1 z1 z2 2(1 − ν)(1 − 2ν)
V (K,N) = + (3 − 4ν) 3 + 3zζ 5 − ,
8π G(1 − ν) 2R31 2R2 R2 R2 (R2 + z2 )
(A.8)


1 3 − 4ν z2 1 − ν 2 + (2 − 3ν)2
W (3) (K,N) = + 13 +
8π G(1 − ν) 2R1 2R1 2R2
 (A.9)
z2 R2 − 3z2
+(3 − 4ν) 23 − zζ 2 5 2 ,
2R2 R2

where

x1 = x − ξ , y1 = y − η, z1 = z − ζ , z2 = zζ ,

 
R1 = x12 + y21 + z21 , R2 = x12 + y21 + z22 ,

ξ , η, ζ are the coordinates of the point K where a unit concentrated force is applied;
x, y, z are the coordinates of the observation point N.
Using Eqs. (A.1–A.9) and Hooke’s law relations one can obtain expressions for
(k)
the stress σij (A.10–A.17), arising in an elastic half-space from the action of a unit
concentrated force at the point K(ξ , η, ζ ) in the direction of the Xk axis:

⎧
⎪
⎪
⎪
⎨
(1) 1 y1 y1 x2 y1
τxy = −(1 − 2ν) 3 − 3 51 − (8ν 2 − 10ν + 3) 3 +
8π (1 − ν) ⎪
⎪ R1 R1 R2
⎪
⎩

3y1 x12 6zζ y1 30y1 zζ x12

+ (8ν 2 − 8ν + 1) − + +
R52 R52 R72
 ⎡   ⎤⎫
z2 x12 ⎪
⎪
⎢ y1 z2 2 + 1−3 2 ⎥⎪⎪
⎬
⎢ R2 R2 2y z x2 ⎥
+ 4(1 − 2ν)(1 − ν) ⎢ −
1 2 1 ⎥ ,
[email protected] ⎢
⎣ R22 (R2 + z2 )2 R32 (R2 + z2 )3 ⎥
⎦⎪⎪
⎪
⎪
⎭
(A.10)
546 Appendix A
⎧
⎪
⎪
⎪
⎨
(2) 1 x1 y2 x1 x1
τxy = −(1 − 2ν) 3 − 3 1 5 − (8ν 2 − 10ν + 3) 3 +
8π (1 − ν) ⎪
⎪ R1 R1 R2
⎪
⎩

3y21 x1 6zζ x1 30x1 zζ y21

+ (8ν 2 − 8ν + 1) − + +
R52 R52 R72
⎡    ⎤⎫
z2 y21 ⎪
⎪
⎢ x1 z2 2 + 1−3 2 ⎥⎪⎪
⎬
⎢ R2 R2 2x z y2 ⎥
⎢
+ 4(1 − 2ν)(1 − ν) ⎢ −
1 2 1 ⎥ ,
⎣ R22 (R2 + z2 )2 R32 (R2 + z2 )3 ⎥
⎦⎪⎪
⎪
⎪
⎭
(A.11)

(3) 1 3x1 y1 z1 x1 y1 z1 30x1 y1 zz2 ζ
τxy = − − 3(3 − 4ν) 5 − +
8π (1 − ν) R1 5
R2 R72
 ⎤
z2 (A.12)
x1 y1 2 +
R2 ⎥ ⎥,
+ 4(1 − 2ν)(1 − ν) 2
R2 (R2 + z2 )2 ⎦

* 1
1 3x1 y1 z1 3y1 x1 30x1 y1 zz2 ζ
τyz
(1)
= − − [2(1 − ν)z2 + (1 − 2ν)z1 ] + ,
8π (1 − ν) R51 R52 R72
(A.13)
*
(2) 1 z1 3z1 y21 z1 6zz2 ζ 30zz2 ζ y21
τyz = −(1 − 2ν) 3 − + (1 − 2ν) 3 − + −
8π (1 − ν) R1 R1 5 R2 R25 R72
1
3y21
− [(1 − 2ν)z1 + 2(1 − ν)z2 ] 5 , (A.14)
R2
*
(3) 1 y1 3z2 y1 y1 30y1 zz22 ζ
τyz = −(1 − 2ν) 3 − 15 + (1 − 2ν) 3 − −
8π (1 − ν) R1 R1 R2 R72
1 (A.15)
 3y 1
− 2(1 − 2ν)z2 z − z22 + 2z2 ,
R52
*
(1) 1 z1 3z1 x12 z1 6zz2 ζ 30zz2 ζ x12
τzx = −(1 − 2ν) 3 − + (1 − 2ν) 3 − + −
[email protected]
8π (1 − ν) R1 R1 5
R2 R25 R72
1
3x12
− [(1 − 2ν)z1 + 2(1 − ν)z2 ] 5 , (A.16)
R2
Appendix A 547
* 1
1 3x1 y1 z1 3x1 y1 30x1 y1 zz2 ζ
(2)
τzx = − − [(1 − 2ν)z1 + 2(1 − ν)z2 ] + ,
8π (1 − ν) R51 R52 R72
(A.17)

*
(3) 1 x1 3x1 z21 x1 30x1 ζ zz22
τzx = −(1 − 2ν) 3 − + (1 − 2ν) − −
8π (1 − ν) R1 R51 R32 R72
1 (A.18)
 3x 1
− 2(1 − 2ν)z2 z − z22 + 2z2 ,
R52

⎧
⎪
⎪
⎪
⎨
1 1 x1 3 x13 x1
σx(1) = − (1 − 2ν) 3 − − (8ν 2 − 11ν + 3.5) 3 + (4ν 2 −
4π (1 − ν) ⎪
⎪ 2 R1 2 R15
R2
⎪
⎩

3x13 9zζ x1 6x1 ζ z2 15zζ x13

− 4ν + 0,5) − +ν + +
R52 R52 R52 R72
⎡   ⎤⎫
z2 x12 ⎪
⎪
⎢ 2 1
3z x 2 + 1 − ⎥⎪⎪
⎬
⎢ R2 R22 3
2x1 z2 ⎥
⎢
+ 2(1 − 2ν)(1 − ν) ⎢ − ⎥ ,
R22 (R2 + z2 )2 ⎥
R32 (R2 + z2 )3 ⎦⎪
⎣ ⎪
⎪
⎪
⎭
(A.19)

⎧
⎪
⎪
⎪
⎨1
(2) 1 y1 3 x12 y1 1 y1
σx = (1 − 2ν) 3 − − (1 − 2ν) 3 +
4π (1 − ν) ⎪
⎪ 2 R1 2 R1 5 2 R2
⎪
⎩

y1 x12 3y1 zζ 6y1 ζ z2 15y1 zζ x12

+(12ν 2 − 12ν + 1.5) − +ν + + ,
R52 R52 R52 R72
 
⎡  ⎤⎫
z2 x12 ⎪
⎪
⎢ y1 z2 2 + 1−3 2 ⎥⎪⎪
⎬
⎢ R2 R2 2y z x2 ⎥
+2(1 − 2ν)(1 − ν) ⎢ −
1 2 1 ⎥
[email protected] ⎢
⎣ R22 (R2 + z2 )2 R32 (R2 + z2 )3 ⎥
⎦⎪⎪
⎪
⎪
⎭
(A.20)
548 Appendix A
⎧
⎪
⎪
⎪
⎨
(3) 1 1 z1 3 x12 z1 z2
σx = (1 − 2ν) 3 − − (1 − 2ν)(1.5 − 2ν) 3 −
4π (1 − ν) ⎪
⎪ 2 R1 2 R1 5
R2
⎪
⎩

ζ 3 z1 x12 6ζ z22 15zz2 ζ x12 3zz2 ζ

− 3(1 − 2ν) − (3 − 4ν) − ν − + −,
R32 2 R52 R52 R72 R52
⎡   ⎤⎫
z2 ⎪
⎪
x1 2 +
2 ⎪
⎬
⎢ 1 R 2 ⎥⎥
⎢
−2(1 − 2ν)(1 − ν) ⎣ − 2
R2 (R2 + z2 ) R2 (R2 + z2 )2 ⎦⎪⎪
⎪
⎭
(A.21)

⎧
⎪
⎪
⎪
⎨1
(1) 1 x1 3 y21 x1 1 x1
σy = (1 − 2ν) 3 − − (1 − 2ν) 3 +
4π (1 − ν) ⎪
⎪ 2 R1 2 R1 5 2 R2
⎪
⎩

x1 y21 3x1 zζ 6x1 ζ z2 15x1 zζ y21

+ (12ν 2 − 12ν + 1.5) − +ν + +
R52 R52 R52 R72
⎡    ⎤⎫
z2 y21 ⎪
⎪
⎢ 1 z2
x 2 + 1 − 3 ⎥⎪⎪
⎬
⎢ R 2 R 2
2x z y 2 ⎥
⎢
+2(1 − 2ν)(1 − ν) ⎢
2
−
1 2 1 ⎥ ,
⎣ R22 (R2 + z2 )2 R32 (R2 + z2 )3 ⎥
⎦⎪⎪
⎪
⎪
⎭
(A.22)

⎧
⎪
⎪
⎪
⎨
(2) 1 1 y1 3 y31 y1
σy = − (1 − 2ν) 3 − − (8ν 2 − 11ν + 3,5) 3 +
4π (1 − ν) ⎪
⎪ 2 R1 2 R15 R2
⎪
⎩

y31 9zζ y1 6y1 ζ z2 15zζ y31

+(12ν 2 − 12ν + 1.5) − +ν
+ +
R52 R52 R52 R72
 ⎡   ⎤⎫
z2 y21 ⎪
⎪
⎢ 3z2 y1 2 + 1− 2 ⎥⎪⎪
⎬
⎢ R2 R2 2y 3
z ⎥
+ 2(1 − 2ν)(1 − ν) ⎢ − 1 2 ⎥ ,
[email protected] ⎢
⎣ R22 (R2 + z2 )2 R32 (R2 + z2 )3 ⎥
⎦⎪⎪
⎪
⎪
⎭
(A.23)
Appendix A 549
⎧
⎪
⎪
⎪
⎨
(3) 1 1 z1 3 y21 z1 z2
σy = (1 − 2ν) 3 − − (1 − 2ν)(1.5 − 2ν) 3 −
4π (1 − ν) ⎪
⎪ 2 R1 2 R1 5
R2
⎪
⎩

ζ3 z1 y21 6ζ z22 15z2 zζ y21 3zz2 ζ

−3(1 − 2ν) − (3 − 4ν) − ν − + −
R32 2 R52 R52 R72 R52
⎡   ⎤⎫
z2 ⎪
⎪
y1 2 +
2 ⎪
⎬
⎢ 1 R 2 ⎥ ⎥
⎢
−2(1 − 2ν)(1 − ν) ⎣ − 2 ,
R2 (R2 + z2 ) R2 (R2 + z2 )2 ⎦⎪⎪
⎪
⎭
(A.24)
6
(1) 11 x1 3 z2 x1 1 x1
σz = (1 − 2ν) 3 − 1 5 − (1 − 2ν) 3 +
4π (1 − ν)
2 R1 2 R1 2 R2
1 (A.25)
3x1 15z22 zζ x1
+[(ν − 0.5)z1 z2 − (1 − ν)z2 + ζ ] 5 +
2 2 ,
R2 R72
6
1 1 y1 3 z2 y1 1 y1
σz(2) = (1 − 2ν) 3 − 1 5 − (1 − 2ν) 3 +
4π (1 − ν) 2 R1 2 R1 2 R2
1 , (A.26)
3y1 15z22 zζ y1
+ [(ν − 0.5)z1 z2 − (1 − ν)z2 + ζ ] 5 +
2 2
R2 R72
6
(3) 1 1 z1 3 z31 1 z1
σz = − (1 − 2ν) 3 − + (1 − 2ν) 3 +
4π (1 − ν) 2 R1 2 R1 5 2 R2
1 (A.27)
3 2 3z2 15ζ zz32
+[(ν − 0.5)z1 z2 + (ν − 0.25)z2 − z1 ] 5 −
2 .
4 R2 R72

At the action of the concentrated force at the half-space surface, i.e. when ζ = 0,
R 1 = R 2 = R, Eqs. (A.10)–(A.27) lead to the Boussinesq and Cerruti solutions pre-
sented in the following combined form:
*  1
1 1 x12 1 x12
U (1) = + + (1 − 2ν) − , (A.28)
4π G R R3 R + z R(R + z)2

1 x1 y 1 x1 y1
V (1) =
− (1 − 2ν) , (A.29)
4π G R3
[email protected] R(R + z)2

1 x1 z x1
W (1) = + (1 − 2ν) , (A.30)
4π G R3 R(R + z)
550 Appendix A

1 x1 y1 x1 y1
U =
(2)
− (1 − 2ν) , (A.31)
4π G R3 R(R + z)2
*  1
1 1 y21 1 y21
V (2) = + + (1 − 2ν) − , (A.32)
4π G R R3 R + z R(R + z)2

1 y1 z y1
W (2) = + (1 − 2ν) , (A.33)
4π G R3 R(R + z)

1 x1 z x1
U =
(3)
− (1 − 2ν) , (A.34)
4π G R3 R(R + z)

(3) 1 y1 z y1
V = − (1 − 2ν) , (A.35)
4π G R3 R(R + z)
2

1 z 2(1 − ν)
W =(3)
+ , (A.36)
4π G R3 R
  
y1 3x12 (1 − 2ν) 2Rx12
τxy
(1)
= − 2 − R −
2
− x1
2
, (A.37)
2π R3 R (R + z)2 R+z
  
x1 3y21 (1 − 2ν) 2Ry21
τxy
(2)
= − 2 − R −
2
− y1
2
, (A.38)
2π R3 R (R + z)2 R+z

x1 y1 3z 2R + z
τxy
(3)
=− − (1 − 2ν) , (A.39)
2π R3 R2 (R + z)2
3x1 y1 z
τyz
(1)
=− , (A.40)
2π R5

3x1 y1 z
τyz
(2)
=− , (A.41)
2π R5

3y1 z2
τyz
(3)
= , (A.42)
2π R5

(1) 3x12 z
τzx =− , (A.43)
2π R5

3y21 z
(2)
τzx =− , (A.44)
2π R5
[email protected]
3x1 z2
τzx
(3)
= , (A.45)
2π R5
Appendix A 551

  
x1 3x12 1 − 2ν 2Ry2
σx(1) = − 2 + R2 − 1
− y21 , (A.46)
2π R3 R (R + z)2 R+z
  
y1 3y21 1 − 2ν 2Rx12
σx(2) = − 2 − 2
R − − x1
2
, (A.47)
2π R3 R (R + z)2 R+z
*  1
1 x12 z x12 (2R + z) R2 − Rz − z2
σx(3) = −3 2 + (1 − 2ν) − , (A.48)
2π R3 R (R + z)2 R+z
  
x1 3y21 1 − 2ν 2Rx12
σy(1) = − 2 + 3R −
2
− x1
2
, (A.49)
2π R3 R (R + z)2 R+z
  
y1 3x12 1 − 2ν 2Ry2
σy(2) = − 2 + 3R2 − 1
− y21 , (A.50)
2π R3 R (R + z)2 R+z
*  1
1 y21 z y21 (2R + z) R2 − Rz − z2
σy(3) = −3 2 + (1 − 2ν) − , (A.51)
2π R3 R (R + z)2 R+z

3x1 z2
σz(1) = − , (A.52)
2π R5

3y1 z2
σz(2) = − , (A.53)
2π R5

3z3
σz(3) = − , (A.54)
2π R5

(1) 1 − 2ν (1) (1 − 2ν)(1 + ν) x1

εV = θ =− , (A.55)
E πE R3

(2) 1 − 2ν (2) (1 − 2ν)(1 + ν) y1

εV = θ =− , (A.56)
E πE R3

(3) 1 − 2ν (3) (1 − 2ν)(1 + ν) z

εV = θ =− , (A.57)
E πE R3

Equations (A.55)–(A.57) give the dilatation functions of the stressed state

(i)
v , i = 1,2,3.
ε[email protected]
Further simplification of Eqs. (A.28)–(A.54) corresponds to the case when the
components of the displacement and stress tensors are determined only at the bound-
ary plane of the half-space.
552 Appendix A

Taking into account that at z = 0


R =r= (x1 − ξ )2 + (x2 − η)2 ,

the combined fundamental Boussinesq–Cerruti solution is given in the simplest

form
 
1 1 x12
U (1)
= (1 − ν) + ν 3 , (A.58)
2π G r r
ν x1 y1
V (1) = , (A.59)
2π G r3

1 − 2ν x1
W (1) = , (A.60)
4π G r2
ν x1 y 1
U (2) = , (A.61)
2π G r 3
 
1 1 y21
V (2)
= (1 − ν) + ν 3 , (A.62)
2π G r r

1 − 2ν y1
W (2) = , (A.63)
4π G r2

1 − 2ν x1
U (3) = − , (A.64)
4π G r2

1 − 2ν y1
V (3) = − , (A.65)
4π G r2

1−ν 1
W (3) = , (A.66)
2π G r
 
−y1 6νx12
τxy
(1)
= + (1 − 2ν) , (A.67)
2π r3 r2
 
−x1 6νy21
τxy
(2)
= + (1 − 2ν) , (A.68)
2π r3 r2
x1 y1
τxy
(3)
= (1 − 2ν) , (A.69)
[email protected] π r5

τyz
(1)
= τyz
(2)
= τyz
(3)
= τzx
(1)
= τzx
(2)
= τzx
(3)
= 0, (A.70)
Appendix A 553

 
x1 3νy21
σx(1) = −(1 + ν) + 2 , (A.71)
π r3 r
 
y1 3νx12
σx =
(2)
−(1 + ν) + 2 , (A.72)
π r3 r
 
(1 − 2ν) x12 − y21
σx =
(3)
, (A.73)
2π r3 r

3νx1 y21
σy(1) = − , (A.74)
π r5

3νx12 y1
σy(2) = − , (A.75)
π r5
 
(1 − 2ν) y21 − x12
σy(3) = , (A.76)
2π r3 r

σz(1) = σz(2) = σz(3) = 0. (A.77)

[email protected]
[email protected]
Appendix B
Numerical Schemes for Surface Integral
Calculations

B.1 Parametric Representation of Surface Integrals

While calculating surface integrals of the form


f (x, y, z)d (B.1)


we assume the surface G, restricted by a contour L, to be given parametrically by

equations

x = x(u,v) , y = y(u,v) , z = z(u,v) (B.2)

where the parameters u and v vary within a domain ω, restricted by a piecewise-

smooth contour l on the (u,v) plane (Fig. B.1). Let one-to-one correspondence be
set between the points of the surface G and the domain ω, and the contour L be
transformed into l. Consider a matrix
(  (
(x yu zu (
(
D=( u
( x
v yv zv (
and denote its determinants as A, B, and C:
+  + +  + +  +
+y zu ++ +z xu ++ +x yu ++
A = ++ u , B = ++ u , C = ++ u .
yv zv + zv xv + xv yv +

The surface area element in the parametric coordinates (B.2) is given by [20]

d = 2 + B2 + C2 du dv .
[email protected] (B.3)

Then for any function f(x,y,z) limited at the surface G the following formula is
valid:

555
556 Appendix B

Fig. B.1 Surface parametrization

  
f (x, y, z)d = f (x(u, v), y(u, v), z(u, v)) A2 + B2 + C2 dudv
 ω
Thus, in order to reduce the surface integral under consideration (B.1) to a double
integral one should substitute coordinates x, y, z by the expressions using the u
and v parameters while the surface area element dG should be substituted by the
expression in accordance with Eq. (B.3).
The contact surface of the 3-dimensional volumetric deepened punches can prac-
tically always be approximated with sufficient accuracy by a set of plane boundary
elements (see B.4). Usually only elements of the two types – triangular and quadran-
gular – are used. For each element a local coordinate system is introduced, linked to
the global system (surface parametrization). In the local coordinate system (within
the parameter plane) the surface integrals under consideration are reduced to double
integrals over the areas of simplex and standard square type.
Triangular boundary elements. A separate triangular element and its local coor-
dinate system ξ 1 , ξ 2 are shown in Fig. B.2. Global Cartesian coordinates of the
triangular element points can be given by [11]
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
x x1 x2 x3 ξ1
⎣ y ⎦ = ⎣ y1 y2 y3 ⎦ ⎣ ξ2 ⎦ (B.4)
z z1 z2 z3 ξ3

[email protected]

Fig. B.2 A flat triangular boundary element and its parametrization

Appendix B 557

where ξ 1 , ξ 2 , ξ 3 aretriangular coordinates, related by a formula

ξ1 + ξ2 + ξ3 = 1 .

Applied to a triangular integration area, the calculations result in

+  + + +
+y zξ1 ++ ++ (y1 − y3 ) (z1 − z3 ) ++
A = ++ ξ 1 = ,
yξ2 zξ2 + + (y2 − y3 ) (z2 − z3 ) +
+  + + +
+z xξ 1 ++ ++ (z1 − z3 ) (x1 − x3 ) ++
B = ++ ξ 1 = ,
zξ2 xξ 2 + + (z2 − z3 ) (x2 − x3 ) +
+  + + +
+x yξ1 ++ ++ (x1 − x3 ) (y1 − y3 ) ++
C = ++ ξ 1 = ,
xξ2 yξ2 + + (x2 − x3 ) (y2 − y3 ) +
√
A2 + B2 + C2 = 2S
where S is the area of the triangular element.
For the calculation of a surface integral over a triangular boundary element the
following calculation formula is obtained:
 
f (x, y, z)d = 2S f (x(ξ1 , ξ2 ), y(ξ1 , ξ2 ), z(ξ1 , ξ2 ))dξ1 , ξ2 =
 
1 1−ξ
 1 (B.5)
= 2S f̃ (ξ1 ,ξ2 )dξ1 dξ2 .
0 0

Quadrangular boundary elements. A separate quadrangular element and its local

coordinate system ξ 1 , ξ 2 are shown in Fig. B.3. Global Cartesian coordinates of the
quadrangular element can be found from the following equations [11]:
⎧
⎨ x = ϕ1 x1 + ϕ2 x2 + ϕ3 x3 + ϕ4 x4 ,
y = ϕ1 y1 + ϕ2 y2 + ϕ3 y3 + ϕ4 y4 , (B.6)
⎩
z = ϕ1 z1 + ϕ2 z2 + ϕ3 z3 + ϕ4 z4

where x i , y i , z i , i = 1,4 are the coordinates of the quadrangular cell vertices,

⎧
⎪
⎪
1 1
⎨ ϕ1 = (1 − ξ1 ) · (1 − ξ2 ) , ϕ2 = (1 + ξ1 ) · (1 − ξ2 )
4 4
(B.7)
⎪
⎪ 1 1
⎩ ϕ3 = (1 + ξ1 ) · (1 − ξ2 ) , ϕ4 = (1 − ξ1 ) · (1 + ξ2 ) ,
[email protected]
4 4
are interpolating functions; ξ 1 , ξ 2 are point coordinates of a standard square |ξ1 | ≤
1 , |ξ2 | ≤ 1 .
558 Appendix B

Fig. B.3 A flat quadrangular boundary element and its parametrization

Using Eqs. (B.7), we find ∂ϕi /∂ξj , i = 1,4, j = 1,2, and then A, B, and Cvalues.
For the calculation of surface integrals over a flat quadrangular boundary element
the following calculation formula is obtained:


f (x, y, z)d

1 1  (B.8)
= f (x(ξ1 ,ξ2 ), y(ξ1 ,ξ2 ), z(ξ1 ,ξ2 )) A2 + B2 + C2 dξ1 dξ2
−1 −1

where

1
A= (a1 ϕ1 + a2 ϕ2 + a3 ϕ3 + a4 ϕ4
4
1
B = (b1 ϕ1 + b2 ϕ2 + b3 ϕ3 + b4 ϕ4 ) ,
4
1
C = (c1 ϕ1 + c2 ϕ2 + c3 ϕ3 + c4 ϕ4 ) ,
4
+ + + + + + + +
+ y21 z21 + + y21 z21 + +y z34 ++ +y z34 ++
a1 = ++ + ,a = + + , a = + 34 , a = + 34 ,
y41 z41 + 2 + y32 z32 + 3 + y32 z32 + 4 + y41 z41 +
+ + + + + + + +
+x z + +x z + +x z32 ++ + x41 z41 ++
b1 = ++ 41 41 ++ , b2 = ++ 32 32 ++ , b3 = ++ 32 + , b = + ,
x21 z21 x21 z21 x34 z34 4 + x34 z34 +
+ + + + + + + +
+x y + +x y + +x y34 ++ +x y34 ++
c1 = ++ 21 21 ++ , c2 = ++ 21 21 ++ , c3 = ++ 34 , c = + 34 ,
x41 z41 x32 z32 x32 z32 + 4 + x41 z41 +
x21 = x2 − x1 , y21 = y2 − y1 , z21 = z2 − z1 ,
[email protected]
x32 = x3 − x2 , y32 = y3 − y2 , z32 = z3 − z2 ,
x41 = x4 − x1 , y41 = y4 − y1 , z41 = z4 − z1 ,
x34 = x3 − x4 , y34 = y3 − y4 , z34 = z3 − z4 .
Appendix B 559

B.2 Gauss Cubature Formulae for a Standard Simplex and a

Standard Square
Surface integrals, contained in discretized integral equations, require different
schemes of numerical integration, depending on where the application point of the
concentrated force is located. When the point K(ξ , η, ζ ) is outside the cell of the
boundary element under consideration, the integrand function remains bounded (a
regular integral) and, hence, can be integrated numerically by some cubature for-
mulae of different orders. Since the calculation of such integrals requires the major
part of the processing time, this procedure should be optimized. Such optimizing
is achieved by setting a given maximal upper limit of uncertainty. This means that
the order of the cubature formulae should flexibly vary depending on the distance
between the boundary element and the K(ξ , η, ζ ) point. Below the cubature for-
mulae for a simplex and a standard square are given, constructed using the Gauss
quadrature formula. Using the Gauss quadratures has an advantage over conven-
tional methods (trapezoid rule, Simpson etc.): the given accuracy of the result can
be achieved by using a twice smaller number of summands. This is related to the
optimal choice of both weights and nodes at the approximation of an integral by a
sum (see also B.4).
Numerical integration over triangular boundary elements is performed using the
cubature formulae for a simplex 0 ≤ ξ1 ≤ 1, 0 ≤ ξ2 ≤ 1 − ξ1 (Fig. B.4a). Node
coordinates and weights of the corresponding cubature formulae of the Gaussian
type [13]

1 1−ξ
 1 
n
f̂ (ξ1 ,ξ2 ,ξ3 ) dξ1 dξ2 = wi · f̂ (ς1i ,ς2i ,ς3i ) (B.9)
0 0 i=1

are listed in Table B.1. Contrary to the formulae given in the known books on numer-
ical methods [8, 10, 12, 14, 16, 26, 31], this formula at high accuracy is completely
symmetrical with respect to the triangle vertices. The symmetry means that if a

[email protected]
Fig. B.4 Location of the nodes of cubature formulae for a standard simplex (a) and a standard
square (b)
560 Appendix B

Table B.1 Weights w i and nodes ζ i of cubature formulae for a simplex [13]

2w i ζ1 ζ2 ζ2 Multiplicity

3-point formula (degree of accuracy 2)

0.333333 0.666667 0.166667 0.166667 3
4-point formula (degree of accuracy 3)
−0.562500 0.333333 0.333333 0.333333 1
0.520833 0.600000 0.200000 0.200000 3
6-point formula (degree of accuracy 4)
0.109952 0.816847 0.091576 0.091576 3
0.223381 0.108103 0.445948 0.445948 3
7-point formula (degree of accuracy 5)
0.225000 0.333333 0.333333 0.333333 1
0.125939 0.797426 0.101286 0.101286 3
0.132394 0.470142 0.470142 0.059716 3
12-point formula (degree of accuracy 6)
0.050845 0.873821 0.063089 0.063089 3
0.116786 0.501427 0.249286 0.249286 3
0.082851 0.636502 0.310352 0.053145 6
13-point formula (degree of accuracy 7)
−0.149570 0.333333 0.333333 0.333333 1
0.175615 0.479308 0.260346 0.260346 3
0.053347 0.869739 0.065130 0.065130 3
0.077114 0.638444 0.312865 0.486903 6

cubature node (ζ 1 , ζ 2 , ζ 3 ) occurs, all its permutations will also necessarily occur. If
all ζ i are different, then there are six such nodes in the cubature; if two values of ζ i
coincide, then there are three such nodes. In case of one term (linear interpolation)
the central point (1/3, 1/3, 1/3) is used with the weight of 1/2. The nodes and weights
of Eq. (B.9) are determined from the systems of strongly nonlinear equations. The
requirement of symmetry of nodes with respect to the triangle vertices enables the
number of the unknown values (and, hence, the number of equations) to be reduced
to a minimum.
Numerical integration over quadrangular boundary elements is performed using
the cubature formulae for a standard square (Fig. B.4b)

1 1 
n 
m
j
f̂ (ξ1 ,ξ2 ) dξ1 dξ2 ≈ wi wj · f̂ (ς1i ,ς2 ) . (B.10)
−1 −1 i=1 j=1

Such formulae are obtained by generalization (direct product) of one-

dimensional Gaussian quadrature formulae over each variable
[email protected]
1  
n  ∂ 2n f
f̂ (ξ ) dξ = wi · f̂ (ςi ) + εn , εn = O . (B.11)
∂ς 2n
−1 i=1
Appendix B 561

Table B.2 Nodes and weights of the Gauss cubature formula [10]

n ±ζ i wi

1 0.000000 2.000000
2 0.577350 1.000000
3 0.000000 0.555556
0.774596 0.888889
4 0.339981 0.652145
0.861136 0.347855
5 0.000000 0.568889
0.538469 0.478629
0.906179 0.236927
6 0.238619 0.467914
0.661209 0.360762
0.932469 0.171324
7 0.000000 0.417959
0.405845 0.381830
0.741531 0.279705
0.949108 0.129485
8 0.183435 0.362684
0.525532 0.313707
0.796666 0.222381
0.960289 0.101229

Here ε n is the uncertainty of the quadrature formula of the n-th order. The
weights w i and nodes ζ i of the Gaussian quadrature formulae for n≤ 8 are listed
in Table B.2. For n> 8 the corresponding values can be found in a special literature
(see, e.g. [1, 8, 10, 12, 14, 21, 22, 26, 31]).

B.3 Transformation of Coordinates, Reducing the Order of the

Integrand Function Singularities
Integration over a triangular element, arbitrarily oriented in space, according to the
cubature formula (B.9), is performed by transformation to a simplex using a local
triangular coordinate system. In this case the presence of a singularity in one of the
element nodes will by no means be taken into account. Evidently, direct usage of
such cubature formulae is not applicable for the calculation of surface improper and
singular integrals. In this case an additional subdivision of the boundary element
(adaptive algorithm of integration [29]) or usage of cubature formulae of a special
type [12] is required. The latter can be applied only for the integrals with weak
singularities, their accuracy being not high enough.
Consider a coordinate transformation at which the order of the integral function
[email protected]
singularity is reduced by a unit. Without narrowing the generality, one can assume
that the point of application of the unit force K(ξ , η, ζ ) is located in one of the
triangle vertices. Such triangles represent the subdivision of the boundary element
562 Appendix B

into subelements for which the mass center of the boundary element is the common
vertex K(ξ , η, ζ ) (see Sect. B.3, Fig. 2.4).
According to [23, 29], the coordinates of the triangular element points are given
by
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
x x1 x2 x3 1 − σ1
⎣ y ⎦ = ⎣ y1 y2 y3 ⎦ ⎣ σ1 · (1 − σ2 ) ⎦ . (B.12)
z z1 z2 z3 σ1 σ2

Under such a transformation a unit square [0,1] x [0,1] in the (σ 1 , σ 2 ) plane

is transformed into a spatial triangular element (Fig. B.5a, b, c). Comparing Eqs.
(B.4) and (B.12) one can see that the new local coordinates σ 1 and σ 2 are related to
conventional triangular coordinates ξ 1 , ξ 2 , ξ 3 by equations

ξ1 = 1 − σ2 , ξ2 = σ1 · (1 − σ2 ) , ξ3 = σ1 · σ2 . (B.13)

It is easily seen that the condition ξ1 +√ξ2 + ξ3 = 1 holds identically. In order

to find the differential surface area d = A2 + B2 + C2 dσ1 dσ2 in the new local
coordinates (σ 1 , σ 2 ) define
+  + + +
+y zσ1 ++ + (y1 − y3 ) (z1 − z3 ) +
A = ++ σ 1 = σ + + ,
yσ2 zσ2 + 1+
(y2 − y3 ) (z2 − z3 ) +

[email protected]
Fig. B.5 A flat triangular boundary element with an integrand function singularity in the first node
(a) and consecutive transformations of the local coordinates to a standard square (b–d)
Appendix B 563
+  + + +
+z xσ 1 ++ + (z1 − z3 ) (x1 − x3 ) +
B = ++ σ 1 +
= σ1 + + ,
zσ2 xσ 2 + (z2 − z3 ) (x2 − x3 ) +
+  + + +
+x yσ1 ++ + (x1 − x3 ) (y1 − y3 ) +
C = ++ σ 1 +
= σ1 + +
xσ2 yσ2 + (x2 − x3 ) (y2 − y3 ) +
Now an equation for the calculation of a surface integral over a triangular element
with a singularity of the integrand function in the first node is obtained:

1  1 
f (x(σ1 ,σ2 ), y(σ1 ,σ2 ), z(σ1 ,σ2 )) A2 + B2 + C2 dσ1 dσ2 =
−1 −1 (B.14)
1 1
= 2S f̃ (σ1 ,σ2 ) σ1 dσ1 dσ2 .
0 0

Here, as above, S is the triangular boundary element area; f̃ (σ1 ,σ2 ) =

f (x(σ1 ,σ2 ), y(σ1 ,σ2 ), z(σ1 ,σ2 )) . Note that the fact conditions (B.13) being fulfilled
and the form of Eq. (B.14) enabled the authors of [23] to call the variables (σ 1 , σ 2 )
triangular polar coordinates.
All the quadrature and cubature formulae quoted in handbooks use nodes defined
at the standard interval [a, b] = [–1,1]. Therefore, for the convenience of handbook
data application we perform one more linear substitution of variables

1 1
σ1 = (1 + η1 ) ,σ2 = (1 + η2 ) .
2 2
Now in the new local variables (η1 ,η2 ), { |η1 | ≤ 1 , |η2 | ≤ 1 } the integration
will be performed for a standard square ω (Fig. B.5d) according to a formula

  1 1
1
f (x,y,z)d = S f̃ (η1 ,η2 )(1 + η1 dη1 dη2 (B.15)
4
 −1 −1

It follows from Eq. (B.15) that along the side η1 = –1 of the square ω the integrand
function equals to zero. According to the considered train of transformations, this
side corresponds to the first node of the triangular element. Therefore, the either
present singularity of the integrand function in one of the triangular element vertices
can be eliminated, or its degree can be reduced by unit.
Thus, after the transformation
⎡ ⎤ ⎡ ⎤ ⎡ 1 ⎤
x x1 x2 x3 · (1 − η1 )
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢ ⎥ ⎢
[email protected] ⎥ ⎢1 2 ⎥
⎢ y ⎥ = ⎢ y1 y2 y3 ⎥ ⎢ (1 + η1 ) · (1 − η2 ) ⎥ (B.16)
⎢ ⎥ ⎢ ⎥ ⎢4 ⎥
⎣ ⎦ ⎣ ⎦ ⎣ 1 ⎦
z z1 z2 z3 (1 − η2 )2
4
564 Appendix B

having been applied, numerical integration over triangular boundary elements with
an integrand function singularity in one of the nodes can be carried out uniformly
and efficiently on the base of cubature formulae for a standard square (B.10).

B.4 Highest Algebraic Order of Accuracy Cubature Formulae of

Interpolation-Orthogonal Type Based on Chebyshev Polynomials
A formula is at present considered the best from the point of view of accuracy at a
given number of points [12]. However, in case the Gaussian quadrature being used, a
certain practical inconvenience is related to the need of the nodes and weights corre-
sponding to a great number of integration point to be input in the computer memory
by typing. Using of additional software capable of relatively fast calculations of
quadrature nodes and weights for a given order of accuracy, seems a more rational
way. We have elaborated such software based on the first-order Chebyshev polyno-
mials, following [19, 24, 25, 30]. The developed algorithms and software on their
base can be readily introduced into the existing boundary-element or finite-element
software.
In the theory of quadrature formulae for a given n a problem is set for such a
choice of nodes xk and weights Ak (k = 1, 2,...,n), at which Eq. (B.11) will be exact
for all polynomials of the highest degree. Since each polynomial of the degree N =
2n−1 has also 2n parameters to be determined, once can choose the parameters xk
and Ak in such a way that the quadrature formulae will be exact. The quadratures
possessing this property belong to the quadratures of the highest algebraic degree
of accuracy [24, 27]. Besides, it is proven in [27] that there is no choice of nodes
and weights at which a n-th order quadrature formula can possess the algebraic
degree of accuracy higher than N = 2n−1. Therefore, in order to calculate surface
(two-dimensional) improper and singular integrals it is natural to use the quadra-
ture formulae with the highest algebraic degree of accuracy. If zeros of classical
orthogonal polynomials [27] are used as nodes of the quadrature formulae given by
Eq. (B.11), then the quadrature formulae of interpolation-orthogonal type with the
highest degree of accuracy are obtained.
(n)
If the zeros of x = xk , (k = 1, 2,...,n) of Legendre polynomials Pn (x) are cho-
sen as nodes, one obtains the above mentioned widely known Gaussian quadrature
formulae. The weights of these formulae are determined from the relation [27]

(n) 1 2
Ak = (n)
·  . (B.17)
1 − (xk )2 (n) 2
Pn (xk )

For instance, in the handbook [1] one of the most complete tables of nodes and
weights of the Gaussian quadratures is given, but it contains the data only for some
[email protected]
values of n = 2. . .12,16(4)24(8)48(16)96. Direct use of this table as well as other
tables in regular computations of surface integrals over triangular and quadrangular
boundary elements with check for accuracy is not convenient enough. The weights
Appendix B 565

can be calculated using Eq. (B.17) after the zeros of the Legendre polynomials hav-
ing been determined with the required accuracy. However, this is an iterative process
requiring initial approximations to be given and, besides, recurrent formulae to be
used both for the Legendre polynomials and for their derivatives. Evidently, such a
way will result in a considerable growth of the machine time.
A great convenience is application of zeros of first-order Chebyshev polynomials
Tn [1]

2k − 1
xk(n) = cos θk , θk = π (B.18)
2n
(n)
as integration nodes x =9xk , (k = 1, 2,...,n). Then calculations with the application
√
of the weight function 1 1 − x2 result in a quadrature with coefficients [24]

(n)
Bk = π/n , k = 1,2,...,n . (B.19)

As a result, the following convenient and practical quadrature formula of the

highest algebraic degree (Hermit formula) is obtained:

1 n 
π (n) (n)
f̂ (ξ ) dξ ≈ 1 − (xk )2 · f̂ (xk ) . (B.20)
n
−1 k=1

The node coordinates and weight factors of the quadrature formula (B.20) are
easily calculated at arbitrary n and input in a computer’s random-access memory or
a disk prior to the operation of the boundary-element software main routines. As an
illustration, the corresponding quadrature parameters, computed for n ≤ 8, are listed
in Table B.3.
In [19] a problem of calculation of definite integrals by approximation of the
integrand by an interpolation Chebyshev polynomial [25] is considered. It can be
shown [9] that interpolation of functions over nodes being the roots of first-order
Chebyshev polynomials (B.18) possesses the property of optimality. After using the
integration formula for Chebyshev polynomials [19] a quadrature formula of the
type of Eq. (B.11) with the nodes of Eq. (B.18) and the weights of

(n) 2 
n 
1 − ( − 1)j
Ak = ( − 1)k−1 sin θk cos (n − j)θk , k = 1,n (B.21)
k j
j=1

is finally obtained [30].

Explicit formulae for the approximation of a function of two variables F(x, y)
in a standard square (x, y) ∈ [−1,1] × [−1,1] by a two-dimensional interpolation
[email protected]
Chebyshev polynomial built in [25], appeared very helpful for practical calculation
of surface integrals in the boundary-element method. If the zeros of the first-order
Chebyshev polynomials
566 Appendix B

Table B.3 Nodes and weights of the Hermit cubature formula (B.20)

n ±ζ i wi

1 0.000000 2.000000
2 0.707107 1.000000
3 0.000000 1.111111
0.866025 0.444444
4 0.382683 0.735702
0.923880 0.264298
5 0.000000 0.613333
0.587785 0.525552
0.951056 0.167781
6 0.258819 0.503561
0.707107 0.377778
0.965926 0.118661
7 0.000000 0.454422
0.433884 0.398242
0.781832 0.287831
0.974928 0.086716
8 0.195090 0.385877
0.555570 0.324152
0.831477 0.222988
0.980785 0.066983

(n) 2i − 1 (m) 2k − 1
xi = cos π, yk = cos π, i = 1,n ; k = 1,m (B.22)
2n 2m
are again taken as the interpolation nodes, then termwise integration if a two-
dimensional interpolation polynomial results in a cubature formula [19]

1 1 
n 
m
F(ξ1 ,ξ2 ) dξ1 dξ2 ≈ Aik · F(ξi ,ξk ) (B.23)
−1 −1 i=1 k=1

with the coefficients

'
16( − 1)i+k n−1 1 (n − r)(2i − 1)π
Aik = sin ×
m·n r=0(2) 1 − r
2 2n
. (B.24)
' 1
m−1 (m − s)(2 k − 1)π
× sin
s=0(2) 1 − s
2 2m

If a case when the nodes being the extremal points of the first-order Chebyshev
polynomials

[email protected]
(n)
x = cos
iπ (m)
, y = cos
kπ
, i = 0,n − 1 ; k = 0,m − 1 (B.25)
i
n−1 k m−1
is considered, then a quadrature formula
Appendix B 567

1  1  
n−1 m−1
F(ξ1 ,ξ2 ) dξ1 dξ2 ≈ Bik · F(ξi ,ξk ) (B.26)
−1 −1 i=0 k=0

holds [19] where

16   1
n−1
irπ   1
m−1
isπ
Bik = cos × cos .
(n − 1) · (m − 1) 1−r 2 n−1 1−s2 m−1
r=0(2) s=0(2)
(B.27)
In Eqs. (B.21), (B.24), and (B.27) a prime near the summation symbol means
that the first term of the sum should be taken with a factor of 1/2, two primes mean
that the first and the last terms of the sum are taken with the coefficient 1/2. Three
primes indicate that only the last term has the factor of 1/2.
Numerical calculations have shown that the cubatures (B.20), (B.21), (B.24), and
(B.27), obtained on the base of the Chebyshev polynomials, are easily realized on
a computer by explicit formulae, being convenient from the point of view of pro-
gramming (the use of tabulated data is not required), provide high accuracy of cal-
culations (very close to the accuracy of the Gaussian formula) and enable, contrary
to the Gaussian formulae, the order to be easily changed without additional parti-
tioning of boundary elements into subelements, what is especially convenient at the
calculation of both regular and improper surface integrals.

[email protected]
[email protected]
Appendix C
Round Punch on an Elastic Layer of Variable
Thickness at Central and Off-Centre Load

This appendix contains the results of numerical boundary-element solution of a spa-

tial contact problem for a rigid round punch with a flat bottom, located on a z = 0
surface of an elastic layer x ≥ 0, |y| < ∞, 0 ≤ z ≤ h whose thickness varies as h
= x·tanα, α<π/2. The lower boundary of the layer is rigidly fixed. The influence
function for such a base is described in detail in Sect. 1.4.2. The punch is suppose
to undergo a spatial static load reduced to a vertical resulting force P, eccentric in
general case. There are neither friction forces between the punch and the base, nor
any load outside the punch. The calculation scheme for the problem is shown in
Fig. C.1.
The calculation data are obtained using STAMP-C software, composed for
a EC-1060 computer (FORTRAN-IV programming language), implementing the
boundary-element algorithm, described in v Sect. 2.5.1. The lack of tensile stress in
the soil is achieved by imposing unilateral constraints in the contact area.
For the numerical solution of the contact problem the contact area is discretized
into triangular and quadrangular elements by means of radii and concentric circles,
getting more condensed towards the boundary near which contact pressures have
higher gradients. The radii of the concentric circles are calculated using the follow-
ing quasiuniform relation:

eβ tj − 1 j
rj = α ,tj = j = 1, . . . , L.
eβ − 1 L

Fig. C.1 Calculation scheme

[email protected]
for a round punch on an
elastic layer of variable
thickness at off-centre
loading

569
570 Appendix C

The condensation degree is controlled by the choice of the parameter β. The con-
tact pressure values of required accuracy were achieved at β = −1 and discretiza-
tion of the circular area using up to 400 boundary elements. In order to calculate
integrated characteristics of the contact interaction in most cases it is sufficient to
perform the discretization of a circle using 96 boundary elements, formed by six
concentric circles and sixteen rays (see the discretization diagram in Fig. C.2).
The results for the calculated parameters of the contact interaction of the punch
on an elastic half-space and an elastic layer of variable thickness are shown in Figs.
C.3–C.17 and in Table C.1. In the calculations we used E = 10 MPa, P = (10a)
kN, a being the punch radius. A constant vertical load was applied at the point with
coordinates x = xc + εx , y = yc + εy where εx , εy is the eccentricity of the force P
along the OX, OY axes with respect to the punch centre (xc , yc ), yc = 0.
Figures C.3, C.4, and C.5 show the plots of the contact pressures under a centrally
loaded rigid punch at certain values of the tilt angle α of the seat layer and the fixed
values of ν = 0.25 and xc = 2a. For comparison the same figures show the contact
pressures corresponding to the exact solution of the axially symmetrical contact
problem for the round punch on a half-space

P(1 − ν 2 ) P
Wc = , p(r) = √ .
2Ea 2π a a2 − r2

The pressure values pj in the centres of gravity of the elements located along
the radius, calculated from the solution of the contact problem, were interpolated
by cubic splines. As follows from the comparison of the calculated dependences,

[email protected]

Fig. C.2 Discretization of the contact domain into boundary elements

Appendix C 571

Table C.1 Characteristics of linear and angular displacements of a round punch at ν = 0.25, xc =
2a, εy = 0

Off-centre load, Off-centre load,

Foundation Central load, ε x = 0 εx = 0.5a ψy = 0
type εx = 0 εx = 0 εx = 0 εx = 0 εx = 0

Elastic layer
0.147 0.262
α = 15◦ 0.124 0.454 1.048
0.106 0.379
0.247 0.368
α = 30◦ 0.221 0.513 0.809
0.213 0.458
0.312 0.400
α = 45◦ 0.290 0.441 0.619
0.285 0.470
0.361 0.421
α = 60◦ 0.346 0.355 0.460
0.334 0.489
0.387 0.408
α = 75◦ 0.373 0.281 0.370
0.363 0.466
0.485 0.390
Elastic half-space 0.485 0.0 0.0
0.461 0.449

the pressure under the punch on a layer of variable thickness is essentially non-
monotonous. With the increase of α values the difference between the exact solution
for a half-space and the calculation results for a compressed wedge decreases what
is explained by the increase of the compressed layer thickness under the punch. For
the values α<15◦ the pressure profiles at the interval 0 < r < 3a/4 are practically con-
stant what corresponds to a rather uniform distribution of forces in the central part
of the punch bottom. In order to visualize the asymmetric character of the pressure
field in the contact area, the dependences, corresponding to the diameter sections
of the punch at the angles of ϕ = 0, ±π /2, π with the X axis, are plotted (Figs.
C.3, C.4, and C.5). The comparison of the obtained results shows an asymmetry,

[email protected]

Fig. C.3 Contact pressures (ϕ = 0◦ , ν = 0.25; xc = 2a, εx = εy = 0) at α = 5◦ , 15◦ , 45◦ , 75◦

(1–4), 5 – half-space
572 Appendix C

Fig. C.4 Contact pressures (ϕ = 90◦ , ν = 0.25; xc = 2a, ε x = ε y = 0) at α = 5◦ , 15◦ , 45◦ , 75◦
(1–4), 5 – half-space

the most pronounced in the 0 − π direction. A more detailed view of the asymmet-
ric character of the contact phenomena can be seen from the angular dependences
shown in Fig. C.6. The tilt angle of the lower face of the elastic seat layer imposes
the most essential effect on the distribution of the contact pressures for the values α
≤ 30◦ . For example, variation of α from 5 to 15◦ results in a decrease of the con-
tact pressure in the punch centre almost by 20%. As one should expect, the contact
pressures unlimitedly increase with the approach to the domain boundary. The non-
axially symmetrical character of the contact pressure distribution is seen also from
the fact that near the circular punch boundary for the variable-thickness elastic layer
they can be both larger and smaller than for the half-space, depending on the chosen
direction ϕ. Note that with the decrease of α the contact pressures are concentrated
in the central part of the punch.

[email protected]
Fig. C.5 Contact pressures (ϕ = 180◦ , ν = 0.25; xc = 2a, εx = ε y = 0) at α = 5◦ , 15◦ , 45◦ , 75◦
(1–4), 5 – half-space
Appendix C 573

Fig. C.6 Contact pressures for r/a = 0.16 (a), 0.69 (b), 0.82 (c), 0.93 (d), ν = 0.25; xc = 2a, ε x
= εy = 0 at α = 15◦ , 45◦ , 75◦ (1–3), 4 – half-space

Figures C.7, C.8, and C.9 show the variation of the contact pressures in the punch
centre pc , displacements Wc , and punch slopes ψ x with the distance from the elastic
wedge rib x = 0. As seen from Fig. C.7, with the increase of α the pressures in
the punch centre grow rapidly. With the increase of pc /a the values of pc equalize
independently of the tilt angle α. In this case the difference from the correspond-
ing value of the contact pressure for the half-space pc = P/2π a 2 does not exceed
10%. Settlements of the punch centre increase both with its distance from the rib
of the elastic layer of variable thickness and with the angle α of tilt of the bottom
plane of the distorted foundation (Fig. C.8, Table C.1). With the increase of xc /a the
settlement values asymptotically approach the corresponding values of the punch
settlements on the half-space.

Fig. C.7 Pressures in the

punch centre versus the
[email protected]
distance from the wedge rib
for ν = 0.25; ε x = εy = 0 at
α = 5◦ , 30◦ , 45◦ , 60◦ (1–4), 5
– half-space
574 Appendix C

Fig. C.8 Settlements of the

punch centre versus the
distance from the wedge rib
for ν = 0.25;εx = ε y = 0 at
α = 5◦ , 30◦ , 45◦ , 60◦ (1–4), 5
– half-space

Figure C.9 illustrates the effect of relative distance xc /a and angle α on the punch
slopes at central load. The calculations show that for α = 5◦ a characteristic max-
imum of the slope ψ x occurs, observed at xc ≈ 2a, what possibly results from the
approximate character of the fundamental solution being used or the specific stress-
strained state features of the thin layer. With the increase of the relative distance xc /a
the punch slopes rapidly decrease to zero, the greater α value the higher the slope
variation rate.
Plots in Figs. C.10, C.11, and C.12 show the effect of the base Poisson’s ratio on
the contact pressures, punch centre displacements and punch slopes with distance
from the x = 0 rib. As seen from extraneous dependences (Fig. C.10), the increase
of ν results in the increase of pc values. Already for xc > 10a the contact pressure
values in the punch centre are practically the same for the Poisson’s ratio ν values
in the whole range of its possible variation. Variation of the Poisson’s ratio most
essentially affects the punch centre settlement value. With the punch distance from

Fig. C.9 Punch slopes

[email protected]
versus the distance from the
wedge rib forα = 5◦ , 30◦ ,
45◦ , 60◦ (1–4),ν = 0.25; ε x
= εy = 0
Appendix C 575

Fig. C.10 Pressures in the

punch centre versus the
distance from the wedge rib
for α = 30◦ , ε x = εy = 0 at
ν = 0, 0.25; 0.35, 0.5 (1–4), 5
– half-space (ν = 0.25)

the wedge rib the relative settlements achieve the values corresponding to the punch
settlements on the half-space (Fig. C.11). Dependence of the punch rotation angles
on the Poisson’s ratio is illustrated by Fig. C.12. As follows from the quoted calcu-
lations, at xc > 4a the punch slopes cannot be noticed, and already at xc > 5a there
is practically no punch slope (ψ x < 10−2 ) independently of ν y values.
Numerical studies show that at off-centre punch load negative-pressure zones
arise in the contact area. Therefore, for a correct description of the contact interac-
tion of the punch with the base when the soil cannot bear the tensile stress, unilat-
eral character of constraints in the contact area should be introduced. Thus adjusted
problem qualitatively changes the contact interaction pattern. Figures C.13, C.14,
C.15, and C.16 and Table C.1 show the calculation results for the centre displace-
ments and the punch slopes for various cases of the off-centre load with the account
of the unilateral constraints. The punch distance from the x = 0 wedge rib and the

Fig. C.11 Settlements of the

[email protected]
punch centre versus the
distance from the wedge rib
forα = 30◦ , εx = ε y = 0 at ν
= 0, 0.25, 0.35, 0.5 (1–4)
576 Appendix C

Fig. C.12 Punch slopes

versus the distance from the
wedge rib forα = 30◦ , εx =
ε y = 0 at ν = 0; 0.25, 0.35,
0.5 (1–4)

Poisson’s ratio for the distorted base were fixed: xc = 2a, ν = 0.25. The plots in
Figs. C.13 and C.16a show that the dependences of the punch centre settlements
on eccentricities ε x and ε y without the account of unilateral constraints are strictly
linear. In the case of ε y = 0 (Fig. C.13) the angular coefficients of the corresponding
straight lines are evidently nonzero and practically independent of the angle α. At
εx = 0 (Fig. C.16a) the punch centre settlement does not depend on εy .
Evidently, the approach without the account of unilateral constraints cannot show
the separation of parts of the punch from the base and the essential punch slope
at high eccentricities of the applied load. This shortcoming is overcome for the
solution with the account of the unilateral character of constraints in the contact

Fig. C.13 Punch centre

settlements for ν = 0.25; εx
= 0, xc = 2aat α = 15◦ , 30◦ ,
45◦ , 60◦ , 75◦ (1–5), 6 –
[email protected]
half-space with the account of
the punch separation (solid
curve) and without account of
separation (chain curve)
Appendix C 577

Fig. C.14 Punch slopes

versus εx for ν = = 0.25; ε y
= 0; xc = 2a at α = 15◦ , 30◦ ,
45◦ , 60◦ , 75◦ (1–5), 6–
half-space

area. At partial separation of the punch from the layer surface the plots (Figs. C.13,
C.16a) begin to show the nonlinear law of settlement variation. For ε x (εy ) > 0.75a
a sharp increase of displacements Wc is observed what corresponds to even stronger
shrinkage of the contact area and the punch turnover. The analysis of numerous
calculations enables one to conclude that in the intervals −0.4a < εx < 0.3a (ε y =
0, Fig. C.13) and |ε y | < 0.35a (εx = 0, Fig. C.16a) the contact interaction of the
punch and the distorted layer goes without separation. Besides, independently of
the eccentricity of the resultant force application, the increment of the settlements
of the punch centre applied to a half-space over the corresponding values for the
wedge bases is preserved. The plots in Fig. C.15 clearly show that the dependences
of the punch centre settlement on the angle α for the fixed eccentricity values are
similar and their character is monotonous. The punch slopes (Figs. C.14, C.16b) do

Fig. C.15 Punch centre

settlements and punch slopes
[email protected]
versus the angle α for ν =
0.25, εy = 0, xc = 2a atεx /a
= –0.4 (1); 0 (2); 0.3 (3); 4 –
|ψ x | at εx /a = –0.4
578 Appendix C

(a) (b)

Fig. C.16 Punch centre settlements (a) and punch slopes (b) versusε y for ν = 0.25, ε x = 0, xc =
2a at α = 15◦ , 30◦ , 45◦ , 60◦ , 75◦ (1–5), 6 – half-space with the account of the punch separation
(solid curve) and without account of separation (chain curve)

not possess such a pronounced dependence on the angle α, as the settlements. At the
intervals of coincidence of the solutions, obtained with and without account of the
unilateral character of constraints, the dependences of the slopes and the settlements
on ε x (εy ) are rather close to linear. At high eccentricities the slope absolute values
infinitely grow what corresponds to the punch turnover. The solution for the half-
space, as seen from Figs. C.14, C.15, and C.16b, is intermediate, i.e. for different
tilt angles α the punch slopes can be both higher or lower than the corresponding
values for the half-space.
Figures C.14 and C.15 indicate some specific features of the off-centre punch
loading when ε y = 0. At such loading the slope ψ y = 0. Depending on the values of
ε x , the punch slopes ψ x can acquire values of different signs (Fig. C.14). Table C.1
gives the calculated values for the eccentricities εx for the case, the most important
in practice when the punch has no slope (ψ x = 0). In the same table for εx = 0.5a
the punch displacement parameters without the account of unilateral constraints are
shown in the numerator, and those obtained with the account of unilateral constraints
– in the denominator. The dependences of the punch slopes ψ x on the tilt angle of
the rigid seat base do not have such pronounced monotonous character as settlement
curves (Fig. C.15, Table C.1), their character being essentially dependent of the sign
and value of the eccentricity εx . At ε x = 0 the slope ψ x is practically constant, rather
close to zero.
Figure C.17 enables one to compare the calculation results for the contact pres-
sures under a round punch interacting with the layers of constant and variable thick-
ness. In the last case the known Egorov influence function for an elastic layer of
finite thickness with a smooth contact with the rigid base was used. The depth of
both layers Hc under the punch centre was fixed, the tilt angles of the bottom plane
of the elastically compressible layer were varied. The contact pressure distributions
[email protected]
at Hc /a = 2 for α = 60◦ , 45◦ are shown in Fig. C.17.
The calculation data, given in Appendix C, give the evidence for the efficiency of
the proposed method to be applied for design of foundations under tower-type struc-
Appendix C 579

Fig. C.17 Contact pressures along the punch diameter (0–π ) at ν = 0.25; Hc = 2a; 1, 3 – wedge
base, α = 45◦ (1), 60◦ (3), 2 – layer of constant thickness

tures to be constructed at complicated geotechnical conditions (with the account of

the actual character of occurrence of weakly compressible soils). This enables the
method to be recommended for practical application for search of rational project
solutions, calculation of base deformations and determination of preliminary size of
foundations, round in plane.

[email protected]
[email protected]
Appendix D
Foundation Under a Tower-Type Structure on a
Wedge Base

Under tower-type structures (compact in plan high public and dwelling buildings,
chimneys, water towers, TV towers, masts, etc.), constructed on a wedge founda-
tion, as a rule, rigid foundation plates are used with flat bottom of simple geometric
shape: a square, a rectangle, a circle, or a ring. The wedge base is understood as a
layer of compressible soil of non-uniform thickness restricted by a horizontal flat
surface from above and by an oblique plane resting on an uncompressible massive
rock. The most important aspect of design of the foundations in question is restric-
tion of non-uniformity of the wedge base settlements and, hence, the slope of the
structure in general.
It follows from the experience of practical applications of traditional calculation
methods that foundation plates with a symmetrical bottom have high consumption
of material since, due to the non-uniform compressibility of the wedge base, in order
the condition (iν +im ) ≤ iu to be fulfilled, the plate part of the foundation should be
essentially extended in plan against the calculated values obtained from the condi-
tions pmin ≥ 0, pmax ≤ 1.2R, p ≤ R. Here iν , im are the components of the total slope
of the foundation i, resulting from vertical V and moment M loads, respectively,
transmitted by the foundation to the base, iu is the maximal admissible slope for
the given class of structures, pmin and pmax are the minimal and maximal pressures
on the soil under the fundament plate bottom edges, respectively, p is the average
pressure on the soil over the plate bottom, R is the calculated soil resistance.
Thus, the non-uniform compressibility of the wedge base is the reason for the
underutilization of the soil strength capacity and, hence, excessive consumption of
reinforced concrete for foundation plates.
We propose a foundation for tower-type structures in the shape of a round plate
with a hole in the bottom whose longitudinal axis is displaced with respect to the
plate centre towards the wedge base rib by the value of eccentricity determined by
the wedge angle, the distance to the rib, the geometrical size of the foundation and
the deformational characteristics of the base. Due to the eccentric location of the
hole the pressure is essentially concentrated under the plate edge where the wedge
thickness is minimal what, in turn, contributes to the foundation slope removal.
[email protected]
Figure D.1 shows the in-plan view and section of a foundation intended for the
construction of tower-type structures on a wedge base. The considered foundation
is a round in plan plate 1 with a hole 2 whose longitudinal axis is displaced by the

581
582 Appendix D

(a) (b)

Fig. D.1 Plan (a) and vertical section (b) of the foundation under a tower-type structure on a
wedge base

value ε0 with respect to the plate centre (point C) towards the rib 3 of the wedge base
4 resting on a massive rock 5. The hole in the foundation bottom can be either blind,
or through. The hole depth should exceed the expected base settlement under the
plate. The eccentricity value ε 0 , by which the hole axis is displaced with respect
to the plate centre, is determined depending on the wedge angle, distance from
the foundation to the wedge rib, geometrical size of the foundation and the defor-
mational characteristics of the base. The proposed construction of the foundation
plate enables the pressure over the bottom on the wedge base to be redistributed in
such a way that the foundation slope component due to the vertical load resultant
be excluded, and under the variable-sign moment loads the wedge base becomes
equally compliant (the foundation slope depends only on the moment value, but not
on its direction).
The calculation scheme explaining the character of interaction of the proposed
foundation construction with the wedge shape at vertical loading is shown in
Fig. D.2a, while Fig. D.2b shows a similar situation with the account of the moment
loading. The vertical load resultant on the base V causes translational (without slope)
downward vertical shift of the proposed foundation. It happens so due to the pres-
ence of an eccentrically displaced hole in the bottom that enables the adjustment
of the distribution of the contact pressures on the wedge base, and thus affects
the foundation slope. The increase of the eccentricity ε0 results in a considerable

(a) (b)

[email protected]
Fig. D.2 Calculation scheme of interaction of the foundation with the wedge base under central
(a) and off-centre (b) vertical load
Appendix D 583

concentration of pressures p1 under the plate edge where the wedge thickness is the
smallest and, on the contrary, at the opposite side of the plate the pressures p on the
soil become minimal, as shown in Fig. D.2a. As a result, with the increase of ε0
the foundation slope in the direction of the wedge base thickness increase, induced
by the vertical load resultant V, will decrease and at certain values of ε0 can even
change its sign to the opposite. The eccentricity value ε0 required in order to pro-
vide the foundation settlement uniformity is found from the solution of the known
integral equation of the soil mechanics contact problem

W(x,y) = p(ξ ,η) · ω(x,y,ξ ,η)dξ dη (D.1)
F

where W(x, y) is the base settlement in the point (x, y), p(ξ , η) is the contact pressure
value in the point (ξ , η), ω(x,y, ξ , η) is the base surface settlement in the point (ξ ,
η) due to the vertical unit concentrated force applied to the base in the point (x,y), F
is the area of the foundation contact with the base. The contact pressures p(ξ , η) in
Eq. (D.1) should obey the equilibrium conditions
  
p(ξ ,η)dξ dη = V, p(ξ ,η)dξ dη = V · L + My , p(ξ ,η)dξ dη = Mx (D.2)
F F F

where L is the distance from the point of application of the vertical load resultant V
to the wedge rib, Mx and My are the moment load components with respect to OX
and OY axes, respectively.
In the case under consideration the contact area F has the shape of an eccentric
ring shown in Fig. D.3. By substitution in an explicit form of the integration lim-
its in the polar coordinate system (the pole being in the centre of the outer circle)
and taking into account the main condition of the problem W(x,y) = W = const
(corresponding to a vertical translational shift of the foundation without any slope),

[email protected]
Fig. D.3 Geometrical
scheme of the contact area of
the foundation and the base
584 Appendix D

Eq. (D.1) is given by

π r2
2 dϕ p(ρ,ϕ) · ω(x,y,ρ,ϕ)ρdρ − w = 0. (D.3)

0
(1−cos2 ϕ)ε02 +r12 −ε0 cos ϕ

The search parameter ε 0 , settlement W and contact stresses p(ρ,ϕ) can be found
from the joint solution of Eq. (D.3) and the conditions of Eq. (D.2) from numerical
computations. We assume Mx = 0 and My = 0 in Eq. (D.2). The chosen bottom
shape in the form of an eccentric ring and its size its afterwards used to determine
the foundation plate slopes induced by variable-sign moment loads Mx and My . For
this purpose we jointly solve numerically Eq. (D.1) with the conditions of Eq. (D.2).
The base settlement function in the left-hand side of Eq. (D.1) is given by

W(x,y) = W + ψx · (x − xc ) + ψy · (y − yc )

where ψ x and ψ y are the foundation plate slopes around the OX and OY axes due to
the moments Mx and My , respectively.
As an example we use the relative eccentricity values ε 0 /r2 from Table D.1, cal-
culated at the radii ratio r1 /r2 = 0.6 and Poisson’s ratio of the soil ν = 0.35 (see
also Sect. 4.3.3). The function ω(x,y, ξ , η) is determined according to the formulae
of Sect. 1.4.2. The soil deformation modulus E and the vertical force value V were
assumed dimensionless E = 1, V = 1, since they do not affect the ε0 values what fol-
lows from the structure of Eq. (D.1). Numerical computer experiments were carried
out using boundary-element method (Sect. 2.5.1).
The data quoted in Table D.1 give the evidence for the necessity of the eccentric-
ity ε 0 increase with the decrease of the wedge angle α and distance L.
As noted above, the specific feature of the proposed foundation plate is the depen-
dence of its slope only on the absolute value of the moment load |M| and does
not depend on its direction. This was shown by the numerical computations, some
of the results being given in Table D.2. The parameter values in this case were
the following:
 L/r = 8, r1 /r2 = 0.6, α = 30◦ , E = 1, ν = 0.35, ε 0 /r2 = 0.2088,
|M| = Mx2 + My2 = 1.
The quoted results of the numerical experiments confirm that a hole made in
the bottom of a round foundation plate with a longitudinal axis shifted with respect

Table D.1 Relative eccentricity values ε 0 /r2 at which the centrally loaded foundation with the
bottom in the shape of an asymmetric ring possesses uniform settlement on a wedge base

ε 0 /r2 α = 30◦ α = 45◦ α = 60◦

[email protected]
4 – 0.3398 0.3071
8 0.2088 0.1667 0.1413
12 0.1274 0.09828 0.08247
Appendix D 585

Table D.2 Slope values ψ x ,

ψ y at moment loads of My Mx ψx ψy
various direction on a
1 0 0.73464 0
foundation in the shape of an
−1 0 −0.73573 0
asymmetric ring
0 ±1 0 ±0.74105

to the foundation centre towards the wedge base rib by the calculated value of ε0 ,
provides equal compliance of the wedge base both to the vertical load Vand the
variable-sign moment load M. That is, the character of the plate displacements under
external loading is the same as if it rested on the surface of a homogeneous half-
space or a constant-thickness layer.
The most unfavourable case of the combined action of the vertical and moment
loads is shown in Fig. D.2b. In this case the edge stresses under the foundation
bottom achieve extreme values pmin and pmax . The dotted line shows the contact
pressure profile without the moment load. The scheme shown in Fig. D.2b should
be taken as the calculation one and serve as the basis for the choice of the foundation
size. The calculation procedure can be illustrated by an example.
The foundation plate bottom size should be chosen for the following initial data:
load at the foundation edge N = 100000 kN, M = 60000 kNm, Q = 700 kN, foun-
dation depth d = 2.5 m, the soil specific weight above the foundation bottom γII =
15 kN/m3 , the base is formed by plastic sand loam IL = 0.4 with the following char-
acteristics: ϕ II = 24◦ , CII = 15 kPa, γI I = 19 kN/m3 , E = 16000 kPa, ν = 0.35, the
inclination angle of the massive rock seat to the horizon α = 30◦ , the distance from
the load application point to the wedge rib L = 120 m, averaged specific weight of
the reinforced concrete and the soil γmt = 20 kN/m3 , the maximal admissible slope
iu = 0.0025.
Assume the radii ratio β = r1 /r2 = 0.6. In the first approximation consider the
bottom to be in the shape of an axially symmetrical ring.
Put r2 = 12 m. Then r1 = βr2 = 0.6·12 = 7.2 m.

The ring width b = r2 (1 − β) = 12 · (1 − 0.6) = 4.8 m .

The ring area F = π · r22 (1 − β 2 ) = π · 122 · (1 − 0.62 ) = 289.5 m2 .
The ring moment of inertia

Ic = π · r24 (1 − β 4 )/4 = π · 124 (1 − 0.64 )/4 = 14175.4 m4 .

The average pressure over the bottom

p = N/F + γm t d = 100000/289.5 + 20 · 2.5 = 395 kPa.

The maximal pressure over the bottom

[email protected]
(M + Q · d)r2
pmax = p + = 395 + (60000 + 700 · 2.5) · 12/14175.4 = 448 kPa.
Ic
586 Appendix D

The minimal pressure over the bottom

(M + Q · d)r2
pmin = p − = 395 − (60000 + 700 · 2.5) · 12/14175.4 = 343 kPa.
Ic
The calculated soil resistance of the soil under the foundation bottom
γc 1 · γc 2
R= (Mγ · kz · b · γI I + Mq · d · γI I + Mc · CI I ) =
K
= 1.2 · 1.1 · (0.72 · 0.53 · 4.8 · 19 + 3.87 · 2.5 · 18 + 6.45 · 15) = 404 kPa
where K is a coefficient depending on the way of determination of physical and
mechanical characteristics of the soil, kz is a coefficient depending on the foundation
width: kz = z/(2r2 ) + 0.2 = 8/24 + 0.2 = 0.53, 1.2R = 485 kPa. Check: p ≤ R
(395 kPa < 404 kPa), pmax ≤ 1.2R (448 kPa < 485 kPa), pmin ≥ 0 (343 kPa > 0).
The check holds.
The ratio L/r2 = 120/12 = 10. At this ratio and the wedge angle α = 30◦ Eq. (D.3)
should be solved jointly with the conditions (D.2), having set Mx = 0 and My = 0. In
our case one can use the results available from Table D.1. According to the tabulated
data, the relative displacement of the hole axis in the foundation bottom towards the
wedge rib should be ε0 /r2 = (0.2088 + 0.1274)/2 = 0.1681. The absolute value
ε0 = 0.1681 · 12 = 2.02 m.
The coordinate x0 of the gravity centre of the foundation bottom in the shape of
an asymmetric ring is calculated as

Sy π · r12 ε0 β 2 ε0 0.62 · 2.02

x0 = = = = = 1.136 m
F π · r22 (1 − β 2 ) 1 − β2 1 − 0.62
where Sy is the static moment of inertia with respect to OY axis.
The moment of inertia of the foundation bottom in the shape of an asymmetric
ring with respect to the main axis of symmetry is as follows:

π 
I = Ic + π · r22 x02 − π r12 (x0 + ε0 )2 = r24 (1 − β 4 ) + π r22 x02 − β 2 · (x0 + ε0 )2 =
π 4 
= 124 (1 − 0.64 ) + π · 122 1.1362 − 0.62 (1.136 + 2.02)2 = 13137 m4 .
4
The ring width in the narrowest part

b = r2 − r1 − ε0 = 12 − 7.2 − 2.02 = 2.78 m .

The maximal pressure under the foundation bottom edge

(M + Qd + Nx0 )(r2 + x0 )
pmax =p +
[email protected] = 395
I
(6000 + 700 · 2.5 + 100000 · 1.136) · (12 + 1.136)
+ = 570 kPa .
13137
Appendix D 587

Check: pmax ≤ 1.2R (570 kPa > 485 kPa).

The check does not hold. We increase the foundation bottom size.
Assume r2 = 15 m. Then r1 = 0.6·15 = 9 m, L/r2 = 120/15 = 8.

According to Table D.1 ε0 /r2 = 0.2088, ε0 = 0.2088·15 = 3.132 m, x0 = 0.62 ·

3.132/(1 − 0.62 ) = 1.762 m, F = π · 152 · (1 − 0.62 ) = 452.4 m2 , Ic = π · 154 · (1
−
0.64 )/4 = 34607.8 m4 , I = 34607.8+π ·15· 1.7622 − 0.62 · (1.762 + 3.132)2 =
30707.5 m4 .
The pressures over the foundation bottom

100000
p= + 20 · 2.5 = 271 kPa
452.4

(60000 + 700 · 2.5 + 100000 · 1.762) · (15 + 1.762)

pmax = 271 + = 401 kPa
30707.5

(60000 + 700 · 2.5 + 100000 · 1.762) · (15 − 1.762)

pmin = 271 − = 168 kPa
30707.5

The soil resistance is calculated as

1.2 · 1.1
R= · (0.72 · 0.467 · 2.78 · 19 + 3.78 · 2.5 · 18 + 6.45 · 15) = 381 kPa
1.0
where kz = 8/30 + 0.2 = 0.467. 1.2R = 1.2 · 381 = 457 kPa.

Check: p ≤ R (271 kPa < 381 kPa), pmax ≤ 1.2R (401 kPa < 457 kPa), pmin ≥
0 (168 kPa > 0). The check holds.

Thus, the size is determined, for which the foundation plate under vertical load
will uniformly settle on a wedge base (iν = 0).
Now the foundation slope from the moment load should be determined; for this
purpose Eq. (D.1) should be solved jointly with the conditions of Eq. (D.2) at the
value ε 0 /r2 = 0,2088. In this example the data from Table D.2 can be used as well
as a formula for the transition from the tabulated data to the foundations of real
dimensions:

(M + Q · d)ψxtab (60000 + 700 · 2.5)

im = = · 0.73573 = 0.00084.
E r23 16000 · 153

[email protected]
The total foundation slope i = iv + im = 0.00084.
Check: i ≤ iu (0.00084 < 0.0025). The check holds.
The foundation plate settlement in this case is W = 0.105 m .
588 Appendix D

(a) (b)

Fig. D.4 Plan (a) and sectional view (b) of the designed foundation

The plan of the designed foundation is shown in Fig. D.4a and its sectional view
together with the wedge base is shown in Fig. D.4b. The hole in the foundation is
set through. If the foundation plate had been designed as a symmetric ring, then,
in order the condition i ≤ iu to be satisfied, its size should be essentially increased,
since in this case the vertical load component V results in a considerable foundation
slope.
Thus, due to a hole made in a round foundation plate from the bottom side with
the depth h > W and the longitudinal axis, shifted from the foundation centre
towards the wedge rib by the calculated value ε 0 , the following results are achieved:

1. Uniformity of the settlement of a tower-type structure on a wedge base due to

the vertical load V.
2. Essential (by factor of up to 1.5) saving of reinforced concrete for the foundation,
in comparison with traditional structures.

The considered foundation design is covered by Russian Federation patent

No. 043462 of September 10, 1995.
A solution, close to the above foundation design, is a ring-shaped foundation
whose central hole is performed in the shape of a truncated circle [2, 3]. The chord
in this case is displaced from the circumference of the part of the circle, cut off as
a segment, at a distance D, corresponding to the segment height, directed towards
the increase of the base thickness (Fig. D.5). The value D is determined, depending

[email protected]
Fig. D.5 Contact area for a
ring-shaped foundation with a
truncated internal circle
Appendix D 589

on the angle of the elastically compressible wedge, the distance from the foundation
centre to the wedge rib, the geometrical size of the foundation bottom and the defor-
mational characteristics of the base, from the results of the spatial contact problem
for a ring-shaped punch located on an elastic wedge shape.
Similarly to the case of the eccentric inner circle, this foundation design enables
a uniform settlement of the structure under vertical load and a uniform compliance
of the base with respect to a variable-sign moment load to be provided. Simulta-
neously, construction of such foundation is technologically convenient since tradi-
tional symmetric ring-shaped foundations can be used with a minor constructional
finishing. Similar minor constructional changes of standard ring-shaped foundations
(with Dvalue correction) can be recommended for avoiding overcritical slope val-
ues at reconstruction and reinforcement of tower-type structures subject to moment
loads constantly applied in a given direction.

[email protected]
[email protected]
Appendix E
Finite-Difference Equations of Cylindrical Bend
of Orthotropic Slabs Located on an Elastic
Foundation

The differential equation (4.19) of bend of an orthotropic slab at given boundary

conditions of Eqs. (4.21)–(4.23) for deflections and forces on the rectangular slab
contour (fixing, hinged-supporting, or free edges) is solved by the finite difference
method, partial derivatives of the deflection function W(x,y) being substituted by the
corresponding finite-difference expressions. As a result, the boundary-value prob-
lem is reduced to a system of linear algebraic equations with a matrix whose order
equals to the number of unknown deflections of the slab in the chosen discrete pop-
ulation of points.
At first a discrete model of the rectangular elastic slab is built. The origin is put
into the lower left corner of the slab. On a domain 0 ≤ x ≤ a, 0 ≤ y ≤ b, occupied
by the slab, a mesh is put along the X an Y axes with the array pitch x along the
OX axis and y along the OY axis (Fig. E.1). The coordinates of the mesh nodes are
determined as {x·(i−1),y·(j−1)}, i = 1, 2,...,M+1; j = 1, 2,...,N+1; x = a/M,
y = b/N; M and N denote the number of intervals of the slab sides being partitioned
by the mesh along the X and Y axes, respectively. The mesh density is characterized
by the number M·N, the total number of nodes on the slab is (M+1)×(N+1).
Consider Eq. (4.19) in an internal node of the finite-difference mesh (i, j)
     
∂4 W ∂4 W ∂4 W
D1 + 2D3 + D2 = qi,j − pi,j (E.1)
∂x4 i,j ∂x2 ∂y2 i,j ∂y4 i,j

i = 3, 4,...,M−1; j = 3, 4,...,N−1.

[email protected]

Fig. E.1 Finite-difference

mesh plotted on the slab

591
592 Appendix E

Partial derivatives in each summand of the left-hand side of Eq. (E.1) are approxi-
mately represented in the extended form, using the known finite-difference relations
for derivatives
+
∂ 2 W ++ Wi+1,j − 2Wi,j + Wi−1,j
+ =   , (E.2)
∂x i,j
2 x2
+
∂ 2 W ++ Wi,j+1 − 2Wi,j + Wi,j−1
=   , (E.3)
∂y2 +i,j y2
+
∂ 2 W ++ Wi+1,j+1 − Wi−1,j+1 − Wi+1,j−1 − Wi−1,j−1
+ = , (E.4)
∂x∂y i,j 4xy
+
∂ 3 W ++ Wi+2,j − 2Wi+1,j + 2Wi−1,j − Wi−2,j
+ = , (E.5)
∂x i,j
3
2 (x)3
+
∂ 3 W ++ Wi,j+2 − 2Wi,j+1 + 2Wi,j−1 − Wi,j−2
= , (E.6)
∂y3 +i,j 2 (y)3
+
∂ 3 W ++ Wi+1,j+1 − 2Wi+1,j + 2Wi−1,j − Wi−1,j+1 + Wi+1,j−1 − Wi−1,j−1
+ = ,
∂x∂y i,j
2
2 (x) (y)2
(E.7)

+
∂ 3 W ++ Wi+1,j+1 − 2Wi,j+1 + 2Wi,j−1 − Wi+1,j−1 + Wi−1,j+1 − Wi−1,j−1
+ = ,
∂x ∂y i,j
2
2 (x)2 (y)
(E.8)

+
∂ 4 W ++ Wi+2,j − 4Wi+1,j + 6Wi,j − 4Wi−1,j + Wi−2,j
+ = , (E.9)
∂x i,j
4
(x)4
+
∂ 4 W ++ Wi,j+2 − 4Wi,j+1 + 6Wi,j − 4Wi,j−1 + Wi,j−2
= , (E.10)
∂y4 +i,j (y)4
+
∂ 4 W ++ Wi+1,j+1 + Wi+1,j−1 + Wi−1,j+1 − Wi−1,j−1 − 2Wi,j+1 −
+ =
∂x ∂y i,j
2 2
(x)2 (y)2
(E.11)
[email protected]
−2Wi,j−1 − 2Wi+1,j − 2Wi−1,j + 4Wi,j
.
(x)2 (y)2
Appendix E 593

Equation (E.1) in the finite-difference form for the internal node (i, j) are given
by

  
6D1 λ2 + 8D3 λ + 6D2 Wi,j − 4D1 λ2 + 4λD 3 Wi+1,j + Wi−1,j
 +
+D1 λ2 Wi−2,j + Wi+2,j − [4λD3 + 4D2 ] Wi,j+1 +Wi,j−1 + 
+2D3 λ Wi+1,j+1 + Wi−1,j−1 + Wi+1,j−1 + Wi−1,j+1 + D2 Wi,j+2 + Wi,j−2 =
   
= qi,j y4 − pi,j y4 ,
(E.12)
i = 3, 4,...,M−1; j = 3, 4,...,N−1.
Hereinafter a dimensionless geometrical parameter of the finite-difference mesh
λ = (y/x)2 is used, being the ratio of the squared array pitches of the
finite-difference mesh along the OY and OX axes, respectively. Equations (E.12)
relate the deflections of the slab middle plane in thirteen nodes (Fig. E.2a). The
finite-difference equations can be conveniently composed using a 13-point pattern
(Fig. E.2b), by making its centre to coincide with the mesh nodes and noting the
coefficients Ck (k = 1, 2,...,13) at each variable.
While setting up the difference equations for (E.1) in the precontour and the
contour nodes using the 13-point pattern one should know not only the deflections
in the nodes of the mesh plotted on the slab, i.e. inside the contour and on it, but
also the deflections of two series of points outside the slab contour (Fig. E.3). It
is convenient to give such deflections at these outside-contour mesh nodes that at
the nearest node edge the boundary conditions would be satisfied. For this purpose
Eqs. (4.21−4.23) are subjected to difference discretization. Below, as an example,
consider the main and the most complicated type of boundary conditions for a free
contour of Eqs. (4.24–4.26) which are most often used at the calculation of slabs on
an elastic foundation and which for the contour mesh nodes are given by
at i = 1, j = 1, 2,...,N+1; i = M+1, j = 1, 2,...,N+1,

(a) (b)

[email protected]
Fig. E.2 Part of the finite-difference mesh for the internal node (i, j) with the accepted numbering
notations (a) and schematic representation of the finite-difference equation structure using a 13-
point pattern (b)
594 Appendix E

Fig. E.3 Extension of the finite-difference mesh beyond the slab contour

       
∂2 W ∂2 W ∂3 W ∂3 W
+ ν2 = 0, + ε2 = 0;, (E.13)
∂x2 i,j ∂y2 i,j ∂x3 i,j ∂x∂y2 i,j

at j = 1, i = 1, 2,...,M+1; j = N+1, i = 1, 2,...,M+1,

       
∂2 W ∂2 W ∂3 W ∂3 W
+ ν1 = 0, + ε1 = 0; (E.14)
∂y2 i,j ∂x2 i,j ∂y3 i,j ∂x2 ∂y i,j

at i = 1, j = 1; i = M+1, j = 1; i = 1, j = N+1; i = M+1, j = N+1,

 
∂2 W
= 0. (E.15)
∂x∂y i,j

For the finite-difference approximation of the boundary conditions, as before, we

use the formulae for approximate calculation of derivatives of different order from
the function of two variables (E.2–E.11). After substitution of all partial derivatives
by expressions in finite differences the boundary conditions of Eqs. (E.13)–(E.15)
are given by
   
2 (λ + ν2 ) Wi,j − ν2 Wi,j+1 + Wi,j−1 − λ Wi+1,j + Wi−1,j = 0 ,
√   √  
2 λ (λ + ε2 ) Wi+1,j − Wi−1,j − ε2 λ Wi+1,j+1 + Wi−1,j+1 −Wi+1,j−1 −Wi−1,j−1
√  
−λ λ Wi+2,j − Wi−2,j = 0
(E.16)
at i = 1, j = 1,...,N+1, i = M+1; j = 1,...,N+1;
 
2 (1 + λν1 ) Wi,j − Wi,j+1 − W
 i,j−1 −λν1 Wi+1,j + Wi−1,j = 0, 
[email protected]
2 (1 + λε1 ) Wi,j+1 − Wi,j−1 − λε1 Wi+1,j+1 + Wi+1,j−1 − Wi−1,j+1 − Wi−1,j−1
−Wi,j−2 + Wi,j+2 = 0
(E.17)
at i = 1,...,M+1,j = 1; i = 1,...,M+1, j = N+1;
Appendix E 595

Wi−1,j−1 + Wi+1,j+1 − Wi+1,j−1 − Wi−1,j+1 = 0 (E.18)

at i = 1, j = 1; i = 1, j = N+1; i = M+1, j = 1; i = M+1, j = N+1.

The finite-difference approximation of the boundary conditions (E.16)–(E.18)
enables the unknown outside-contour deflections to be included into the overall
equation system and to be expressed in terms of deflections of the nearest inside-
contour points and finally to be excluded from the overall equation system. Then
the difference equations will contain the number of unknowns equal to the number
of the mesh nodes plotted on the slab.
Thus, after the outside-contour deflections having been excluded, in any node of
the slab (i, j) the difference equations are written in the following general form:

C1 (i,j) Wi,j + C2 (i,j) Wi,j−1 + C3 (i,j) Wi,j+1 + C4 (i,j) Wi+1,j +

+C5 (i,j) Wi−1,j + C6 (i,j) Wi+1,j−1 + C7 (i,j) Wi−1,j−1 +
+C8 (i,j) Wi+1,j+1 + C9 (i,j) Wi−1,j+1 + C10 (i,j) Wi,j−2 + (E.19)
+C11 (i,j) Wi,j+2 + C12 (i,j) Wi+2,j + C13 (i,j) Wi−2,j =
=(y)4 αqi,j − pi,j ,

i = 1, 2,...,M+1; j = 1, 2,...,N+1
where Ck (i,j), k = 1,...,13 are the difference coefficients; the values of pi,j are set
according to the model of the elastic foundation being used, qi,j are external load
intensity values in the finite-difference mesh nodes;
⎧
⎨ 1 − for an internal node,
α = 0.5 − for a node located on the side,
⎩
0.25 − for a node in an corner point of the slab.

Difference equations (E.19), corresponding to each node depending on its loca-

tion at the difference mesh (Fig. E.1), can be built using the patterns given for each
type of the boundary conditions, respectively. There are totally 25 such patterns,
corresponding to the number of subdomains shown in Fig. E.4. For a partial (and
the most complicated) case of the slab with free edges, the corresponding patterns
are given in the end of this appendix.
Patterns No. 1, 5, 21, 25 correspond to the corner points of the difference mesh.
Using patterns No. 2, 3, 4 and No. 22, 23, 24 one can build difference equations
for horizontal boundary points. In order to build difference equations for vertical
boundary points one should use patterns No. 6, 11, 16 and No. 15, 10, 20.
Patterns No. 7, 8, 9 and No. 17, 18, 19 correspond to precontour horizontal points,
while those No. 7, 12, 17 and No. 14, 9, 19 correspond to precontour vertical points.
The most often used is pattern No. 13 intended for building up difference equations
[email protected]
in the internal nodes of the finite-difference mesh being more than two node rows
distant from the slab edges.
596 Appendix E

Fig. E.4 Subdomains of the slab (1–25) in which the finite-difference equations are built up
according to the same patterns

The values Ai (i = 1, 2,...,36), quoted in patterns No. 1–25, are determined as

follows:

A1 = D2 ,
A2 = 2λD3 ,
A3 = −4 (λD3 + D2 ) ,
A4 = λ2 D1 ,
A5 = −4λ (λD1 + D3 ) ,
A6 = 6λ2 D1 + 8λD3 + 6D2 ,
A7 = A6 + 2 (A3 + A5 ) + 2A2 (a2 + a4 ) + 2D1 (a4 a5 + a6 ) + 2A1 (a2 a7 + a8 ) ,
A8 = A6 + A3 a1 + A2 a2 − A4 + A1 a2 a7 + A1 a1 a8 ,
A9 = A6 + a1 (A3 + a8 A1 ) + 2a2 (A2 + a7 A1 ) ,
A10 = A6 + a3 (A5 + a6 D1 ) + a4 (A2 + a5 D1 ) − A1 ,
A11 = A6 − A1 − A4 ,
A12 = A6 − A1 ,
A13 = A6 + a3 (A5 + a8 D1 ) + 2a4 (A2 + a5 D1 ) ,
A14 = A6 − A4 ,
A15 = A5 + a2 (A3 + a8 A1 ) + 2A2 + 2A4 + 2a7 A1 ,
A16 = A5 + a2 (A3 + a8 A1 ) + A1 (A2 + a7 A1 ) ,
A17 = 2 [a1 (A2 + a7 A1 ) − a6 D1 ] ,
A18 = A5 + a3 A4 ,
A19 = −2a6 D1 ,
A20 = A4 + a2 (A2 + a7 A1 ) ,
A21 = 2 (a2 A2 + a2 a7 A1 ) + D1 ,
[email protected]
A22 = 2A4 ,
A23 = −2a5 D1 ,
Appendix E 597

A24 = A2 + a4 A4 ,A25 = A2 + a4 A4 + a2 A1 ,A26 = −4 (a5 D1 + a7 A1 + A2 ) ,

A27 = A3 + 2 (A1 + A2 ) + a4 (A5 + a6 D1 ) + 2a5 D1 ,
A28 = A3 + a1 A1 ,
A29 = A3 + a4 (A5 + a6 D1 ) + a3 (A2 + a5 D1 ) ,
A30 = −2a8 A1 ,
A31 = 2 [a3 (A2 + a5 D1 ) − a8 A1 ] ,
A32 = A1 + a4 (A2 + a5 D1 ) ,
A33 = 2 [a 4 (A2 + a5 D1 ) + A1 ] ,
A34 = 2A1 ,
A35 = −2a7 A1 ,
A36 = A2 + a2 A1

where ai (i = 1, 2,...8) are introduced to make the notation shorter and are given by

a1 = 2 (1 + λν1 ) , a2 = −λν1 , a3 = 2 (1 + ν2 /λ) , a4 = −ν2 /λ ,

a5 = −λε2 , a6 = 2λ (λ + ε2 ) , a7 = −λε1 , a8 = 2 (1 + λε1 ) .
Comparison of the common 13-point pattern (Fig. E.2b) with patterns No. 1–
25 for the known values of Ai (i = 1, 2,...36) enables one to identify the difference
coefficients Ck (i, j) (k = 1, 2,...,13), i.e. to develop from the known formulae a global
rigidity matrix solving the linear algebraic equation system

[#] {W} + {p} = {q} .

After solving the algebraic equation system (E.19) and finding the node deflec-
tions Wi,j (i = 1,...,M+1; j = 1,...,N+1) bending moments Mx , My and torques Mxy
as well as lateral forces Qx , Qy are calculated using the known formulae [15, 18, 28]
   
∂2 W ∂2 W ∂2 W ∂2 W
Mx = −D1 + ν 2 , My = −D2 + ν 1 , (E.20)
∂x2 ∂y2 ∂y2 ∂x2

∂2 W
Mxy = Myx = −2DKP, (E.21)
∂x∂y
 3   3 
∂ W D3 ∂ 3 W ∂ W D3 ∂ 3 W
Qx = −D1 + , Qy = −D2 + . (E.22)
∂x3 D1 ∂x∂y2 ∂y3 D2 ∂x2 ∂y

Having presented Eqs. (E.20)–(E.22) in the finite-difference form, one obtains

D1     
(Mx )i,j = − λ Wi+1,j + Wi−1,j + ν2 Wi,j+1 + Wi,j−1 − 2 (λ + ν2 ) Wi,j ,
(y) 2
 
D1   D3  
[email protected]
(Qx )i,j = − λ W i+2,j − W i−2,j − 2 λ + Wi+1,j − Wi−1,j +
2x (y) 2 D1

D3  
+ Wi+1,j+1 + Wi−1,j+1 − Wi+1,j−1 − Wi−1,j−1
D1
598 Appendix E

for

j = 1; i = 2,...,M;
j = 2; i = M+3,...,2(M+1)−1;
j = N+1; i = N(M+1)+2,...,(N+1)(M+1)−1;

  D2   
My i,j = − λν1 Wi+1,j + Wi−1,j + Wi,j−1 + Wi,j+1 − 2 (1 + λν1 ) Wi,j ,
(y) 2
(E.25)
 
  D2 λD3  
Qy i,j = − W i,j−2 − W i,j+2 − 2 1 + Wi,j+1 − Wi,j−1 +
2 (y) 3 D2

λD3  
+ Wi+1,j+1 + Wi+1,j−1 − Wi−1,j+1 − Wi−1,j−1
D2
(E.26)
for

j = 2; i = M+2,...,2(M+1);
j = 3; i = 2(M+1)+1,...,3(M+1);
j = N; i = (N−1)(M+1)+1,...,N(M+1);

  DKP 
Mxy i,j = − Wi+1,j+1 + Wi−1,j−1 − Wi+1,j−1 − Wi−1,j+1 (E.27)
xy

for

j = 1; i = 2,...,M; j = 2;
i = (M+1)+1,...,2(M+1);
j = N+1; i = N(M+1)+2,...,(N+1)(M+1)−1.

Based on the algorithm described in this appendix, an “ORTOPLIT” software

was elaborated for studies of contact bending of rectangular slabs located on elastic
nonclassical foundations. The results obtained have shown the efficiency of applica-
tion of the finite-difference method for solving contact problems of statical bending
of orthotropic slabs when a half-space and layers of constant and variable thick-
ness are used as models for the elastic foundations (see Sects. Section 4.7.3, Section
4.7.4, [5–7]). Using an elastic foundation model with rigidity coefficient variable
over the foundation area, a project of a potable water basin reinforcement by piles
foundations was performed [4]. From the values of the foundation rigidity function,
variable over the foundation area, at the finite-difference mesh nodes, found from
the solution of the contact problem, the rigidities of the supplied pile foundations
[email protected]
were calculated what enabled the basin bottom deflections to be adjusted. From the
rigidity values found, the loads on each piles foundation were determined and the
size of the piles was chosen.
Appendix E 599

Patterns for Building up Finite-Difference Equation of Cylindrical

Bend of an Orthotropic Slab
Pattern No. 1: i = 1, j = 1.
Pattern No. 2: i = 2, j = 1.
Pattern No. 3: i = 3, 4,. . ., M−1, j = 1.
Pattern No. 4: i = 1, j = M−1.
Pattern No. 5: i = 1, j = M.
Pattern No. 6: i = 2, j = 1.
Pattern No. 7: i = 2, j = 2.
Pattern No. 8: i = 2, j = 3, 4,..., M−2.
Pattern No. 9: i = 2, j = M−1.
Pattern No. 10: i = 2, j = M.
Pattern No. 11: i = 3, 4,..., N−1, j = 1.
Pattern No. 12: i = 3, 4,..., N−2, j = 2.
Pattern No. 13: i = 3, 4,..., N−2, j = 3, 4..., M−2.
Pattern No. 14: i = 3, 4,..., N−2, j = M−1.
Pattern No. 15: i = 3, 4,..., N−2, j = M.
Pattern No. 16: i = N, j = 1.
Pattern No. 17: i = N−1, j = 2.
Pattern No. 18: i = N−1, j = 3, 4,..., M−2.
Pattern No. 19: i = N−1, J = M−1.
Pattern No. 20: i = N−1, j = M.
Pattern No. 21: i = N, j = 1.
Pattern No. 22: i = N, j = 2.
Pattern No. 23: i = N, j = 3, 4,..., M−2.
Pattern No. 24: i = N, j = M−1.
Pattern No. 25: i = N, j = M.

[email protected]
600 Appendix E

[email protected]
Appendix E 601

[email protected]
[email protected]
Appendix F
Calculation of the Base for a Pyramidal Pile
Under Vertical Load According to the
“Instructions Manual for Design of Foundations
Made of Pyramidal Piles”

0.7 × 0.7
The aim is to build a graph of settlements of the base of a 5-m long
0.1 × 0.1
m pyramidal pile, submerged in a high-plasticity clayey soil with the following
physico-mechanical properties: internal friction angle ϕ = 10◦ , specific resistance
C = 15 kPa, specific weight of the soil γ = 16.3 kN/m3 , modulus of deformation of
the soil, determined from the punch test data Ep = 2750 kPa.

F.1 General
Settlement of the base of a single pile S (Fig. F.1) due to the transfer of pressure on
the soil by the pile lateral surface according to [17] is given by
⎡ ⎤
  1 + sin ϕ
2 (1 + μ) · (1 − 2μ) ⎢ PF,6n + C · cot ϕ sin ϕ − P ⎥ (F.1)
S= · ρ · ⎣PH F,6n ⎦
E PH + C · cot ϕ

where μ is the soil transverse expansion coefficient, determined for clayeys and
clays from the formula [17]

μ = 0.1 · (1 + 3JL ‘) (F.2)

JL is the soil consistency index, ρ is a geometric characteristic of the pile (m),

given by
 

l 2
ρ = 2 dH + 2 l · tan α dH + l · tan α
2
(F.3)
dB 3

where dH , dB , l, α are the notations shown in Fig. F.1: dH is the diameter (cross-
[email protected]
section side) of the bottom end of the pile, dB is the diameter (cross-section side)
of the upper part of the pile in the level of the calculated soil surface, l is the pile
submergence depth, α is the pile convergence angle, E is the calculated modulus of

603
604 Appendix F

Fig. F.1 Scheme for calculation of the pyramidal pile settlements under a vertical load

deformation of the soil (kPa), determined from durable pressiometric or punch tests,
[email protected]
using a relationship

E = η · Ep , (F.4)
Appendix F 605

η is an adjustment factor, taking into account the type of the soil, PH is the initial
squeezing pressure (kPa), given by

PH = P0 (1 + sin ϕ) + C · cos ϕ (F.5)

P0 is the natural lateral soil pressure (kPa), given by

μ
P0 = · γ · h, (F.6)
1−μ

γ is the natural weighted average specific weight of the soil within the pile sub-
mergence depth (kN/m), h is the distance from the grading level to the middle of the
pile part under consideration (m), PF,lat is the squeezing pressure (kPa), on the pile
from the load F, determined as

Flat −  · cos α · C
PF,lat = (F.7)
 · cos α · ( tan α + tan ϕ)

where Flat is the calculated load on the soil, transferred by the pile lateral surface
(kN),  is the pile lateral surface area (m2 ).
The settlement of the base under the pile’s bottom end S0 is given by
⎡ ⎤
  1+sin ϕ
0.3(1 + μ) · (1 − 2μ) P + C · cot ϕ sin ϕ
dH · D · ⎣PH − PF,0 ⎦ (F.8)
F,0
S0 =
E PH + C · cot ϕ

where PF,0 is the soil resistance (kPa) under the bottom end of the pile, given by

F0
PF,0 = A · 2
− B · C, (F.9)
dH

where F0 is the calculated load on the soil, transferred by the bottom end of the pile
(kN), dH is the diameter (side) of the bottom section of the pile (m), A, B, and D are
the coefficients, taken from Table 2.1 in the instructions manual [17] depending on
the calculated value of the internal friction angle ϕ.
While calculating pyramidal piles from the base deformations, one should, in
accordance with the instructions manual [17], partition soil seams (along the pile
length) with respect to the base deformations into homogeneous layers (parts) not
more than 2-m thick (long). Then the calculated load on the soil, transferred by the
whole pile surface, is determined as the sum of loads transferred by the bottom end
F and the lateral surfaces of the pile partitions Fi (i = 1,...n) at a given settlement S
[email protected]

n
F = F0 + Fi . (F.10)
i=1
606 Appendix F

At the final stage the calculated loads F, corresponding to the given pile settle-
ments S, are determined from the S = f(F) plots and, vice versa, the settlements S,
corresponding to the given loads F are determined from the F = g(S) plots.

F.2 Calculation Procedure

For clayeys JL = η = 1, hence

E = ηEp = 2750 kPa,

μ = 0.1(1 + 3JL ) = 0.1(1 + 3.1) = 0.4

In the calculation example under consideration for ϕ = 10◦

sin10◦ = 0.1736, cos10◦ = 0.9848, tan10◦ = 0.1763, cot10◦ = 5.6714,

1 + sin 10◦
= 6.759.
sin 10◦
According to Table 2.1 of the instructions manual [17], we imply

A = 0.445, B = 0.851, D = 2.428.

Now we calculate the geometrical characteristics of the pile and its three marked
partitions 1 m, 2 m, and 2 m long, respectively (Fig. F.1):

side of the pile lower section dH = dH1 = 0.1 m,

side of the 1st section of the pile dB1 = dH2 = 0.22 m,
side of the 2nd section of the pile dB2 = dH3 = 0.46 m,
side of the pile head dB = dB3 = 0.7 m,
length of the lower partition of the pile l1 = 1 m,
length of the medium partition of the pile l2 = 2 m,
length of the upper partition of the pile l3 = 2 m,
overall pile length l = l1 + l2 + l3 = 5 m,
depth of the centre of the 1st partition h1 = l1 /2 + l2 + l3 = 4.5 m,
depth of the centre of the 2nd partition h2 = l2 /2 + l3 = 3 m,
depth of the centre of the 3rd partition h3 = l3 /2 = 1 m,
tangent of the pile convergence angle tanα = (dB −dH )/2/l = 0,06,
cosine of the pile convergence angle cosα = 0,9982,
lateral surface area of the 1st partition of the pile
[email protected]
(dH1 + dB1 ) l1
1 = 4 · · = 0.6412 m2 ,
2 cos α
Appendix F 607

lateral surface area of the 2nd partition of the pile

(dH2 + dB2 ) l2
2 = 4 · · = 2.7249 m2 ,
2 cos α
lateral surface area of the 3rd partition of the pile

(dH3 + dB3 ) l3
3 = 4 · · = 4.6484 m2 ,
2 cos α
geometrical characteristics of the 1st partition of the pile
 

l1 2
ρ1 = 2
d 2
H1 + 2l1 · tan α d H1 + l1 · tan α = 0.5537 m,
dB1 3

geometrical characteristics of the 2nd partition of the pile

 

l2 2
ρ2 = 2
d 2
H2 + 2l2 · tan α d H2 + l2 · tan α = 1.138 m,
dB2 3

geometrical characteristics of the 3rd partition of the pile

 

l3 2
ρ3 = 2 dH3 + 2l3 · tan α dH3 + l3 · tan α
2
= 1.3927 m.
dB3 3

F.3 Determination of the Base Settlements S0 Under the Bottom

End of the Pile
At dH = dH1 = 0.1 m and l = 5 m using Eq. (F.9) we calculate the soil resistance at
the given value of F = 3.3 kN:

F0 3.3
PF,0 = A · 2
− B · C = 0.445 · − 0.851 · 15 = 134.085 kPa
dH 0.01
and, having calculated

μ 0.4
P0 = ·γ ·h= · 16.3 · 5 = 54.33 kPa,
1−μ 0.6

 ◦
 ◦
H = P0 · 1 + sin 10 + 15 · cos 10 = 54.33 · 1.1736 + 15 · 0.9848 = 78.54 Kpa,
[email protected]

according to Eq. (F.8), we determine the base settlements under the bottom end of
the pile
608 Appendix F

Table F.1 The load under the

bottom end of the pile versus F0 (kN) PF,0 (kPa) S0 (cm)
the loads on the soil
3.3 134.085 0.3206
3.6 147.435 0.5170
3.9 160.785 0.7943
4.2 174.135 1.1768
4.5 187.485 1.6947
4.8 200.835 2.3845
5.1 214.185 3.2899
5.4 227.535 4.4634
5.7 240.885 5.9665
6.0 254.235 7.8719

   
0.3 · 1.4 · 0.2 134.085 + 15 · 5.67 6.759
S0 = · 0.1 · 2.428 · 78.54 − 134.085
2750 78.54 + 15 · 5.67
= 0.3206 cm.

Then we perform calculations, varying the load within the range, corresponding
to the settlement variation within the operating interval 0.3−8.0 cm under the bot-
tom end of the pile versus the loads on the soil are given in Table F.1. The data
obtained are approximated using the least-square method by a logarithmic depen-
dence

F0 = 0.84743 ln (10−2 · S0 ) + 8.0383, (F.11)

its plot being shown in Fig. F.2.

Fig. F.2 Plots of calculated

loads on the soil for separate
partitions of the pile, resulting
in the same given base
settlements: F0 is the load on
the pile’s lower end, F1 ,F2 ,F3
are the loads on the lateral
surfaces of the pile partitions,
F is the sum of the loads
[email protected]
transferred by the lower end
and the lateral surface of the
pyramidal pile at a given
settlement
Appendix F 609

F.4 Determination of the Base Settlements S1 of the Lower Part of

the Pile (Partition 1)
At dH1 = 0.1 m, dB1 = 0.22 m, h1 = 4.5 m, l1 = 1 m, F1 ·cos10◦ = 0.64 m2 ,
ρ 1 = 0.5537 m, using Eq. (F.7), at the given value of F1 = 22 kN we determine the
squeezing pressure, acting on the 1st partition of the pile:

F1 − 1 · cos α · C 22 − 0.64 · 15
PF,1 = = = 81.985 kPa.
1 · cos α · ( tan α + tan ϕ) 0.64 · (0.06 + 0.1763)
Then, having calculated

μ 0.4
P0 = · γ · h1 = · 16.3 · 4.5 = 48.9 kPa,
1−μ 0.6

 
PH = P0 · 1 + sin 10◦ + 15 · cos 10◦ = 32.6 · 1.1736 + 15 · 0.9848 = 53.03 kpa,

using Eq. (F.1), we determine the settlements of the base of the pile’s lower partition
S1 :
   
2 · 1.4 · 0.2 81.985 + 15 · 5.67 6.759
S1 = · 0.5537 · 72.16 − 81.985
2750 72.16 + 15 · 5.67
= 0.3011 cm.

Then we perform calculations, varying the load within the range, corresponding
to the settlement variation within the interval 0.3/8.0 cm. The values of the base
settlements S1 of the lower partition of the pile versus the loads on the soil are given
in Table F.2.The data obtained are approximated using the least-square method by a
logarithmic dependence

F1 = 2.79232 ln (10−2 · S1 ) + 37.3238, (F.12)

its plot being shown in Fig. F.2.

Table F.2 The values of the

base settlements of the lower F1 (kN) PF,1 (kPa) S1 (cm)
partition of the pile versus the
22.0 81.985 0.3011
loads on the soil
23.0 88.597 0.5942
24.0 95.209 0.9773
25.0 101.820 1.4681
26.0 108.432 2.0864
[email protected] 27.0
28.0
115.044
121.655
2.8555
3.8013
29.0 128.267 4.9533
30.0 134.879 6.3459
31.0 141.491 8.0137
610 Appendix F

F.5 Determination of the Base Settlements S2 of the Medium Part

of the Pile (Partition 2)
At dH2 = 0.22 m, dB2 = 0.46 m, h2 = 3 m, l2 = 2 m, F2 ·cos10◦ = 2.72 m2 ,
ρ 2 = 1.138 m, using Eq. (F.7) at the given value of F2 = 79.5 kN we determine the
squeezing pressure, acting on the 2nd partition of the pile:

F2 − 2 · cos α · C 79.5 − 2.72 · 15

PF,2 = = = 60.205 kPa.
2 · cos α · ( tan α + tan ϕ) 2.72 · (0.06 + 0.1763)
Then, having calculated

μ 0.4
P0 = · γ · h2 = · 16.3 · 3.0 = 32.6 kPa,
1−μ 0.6

 
PH = P0 · 1 + sin 10◦ + 15 · cos 10◦ = 32.6 · 1.1736 + 15 · 0.9848 = 53.03, kPa

using Eq. (F.1), we determine the settlements of the base of the pile’s medium par-
tition S2 :
   
2 · 1.4 · 0.2 60.285 + 15 · 5.67 6.759
S2 = ·1.138· 53.03 −60.205 = 0.3528 cm.
2750 53.03 + 15 · 5.67

Then we perform calculations, varying the load within the range, correspond-
ing to the settlement variation within the interval 0.3 ÷ 8.0 cm. The values of the
base settlements S2 of the medium partition of the pile versus the loads on the soil
are given in Table F.3. The data obtained are approximated using the least-square
method by a logarithmic dependence

F2 = 8.9314 ln (10−2 · S2 ) + 127.66, (F.13)

its plot being shown in Fig. F.2.

F.6 Determination of the Base Settlements S3 of the Upper Part

of the Pile (Partition 3)
At dH3 = 0.46 m, dB3 = 0.7 m, h3 = 3 m, l3 = 2 m, F3 ·cos10◦ = 4.64 m2 ,
ρ 3 = 1.393 m, using Eq. (F.7), at the given value of F3 = 110.5 kN we determine
the squeezing pressure, acting on the 3rd partition of the pile:

[email protected]
F3 − 3 · cos α · C 110.5 − 4.64 · 15
PF,3 = = = 37.3 kPa,
3 · cos α · ( tan α + tan ϕ) 4.64 · (0.06 + 0.1763)
Then, having calculated
Appendix F 611

μ 0.4
P0 = · γ · h3 = · 16.3 · 1.0 = 10.87 kPa,
1−μ 0.6

PH = P0 (1 + sin 10◦ ) + 15 cos 10◦ = 10.87 · 1.1736 + 15 · 0.9848 = 27.53. kPa,

using Eq. (F.1), we determine the settlements of the base of the pile’s upper part S3 :
   
2 · 1.4 · 0.2 37.3 + 15 · 5.67 6.759
S3 = · 1.393 · 27.53 − 37.3 = 0.3124 cm.
2750 27.53 + 15 · 5.67

Then we perform calculations, varying the load within the range, corresponding
to the settlement variation within the interval 0.3/8.0 cm. The values of the base
settlements S3 of the upper part of the pile versus the loads on the soil are given in
Table F.4. The data obtained are approximated using the least-square method by a
logarithmic dependence

F3 = 14.3802 ln (10−2 · S3 ) + 189.57, (F.14)

its plot being shown in Fig. F.2.

Table F.3 The values of the

base settlements of the F2 (kN) PF,2 (kPa) S2 (cm)
medium partition of the pile
79.5 60.205 0.3353
versus the loads on the soil
82.5 64.873 0.6395
85.5 69.540 1.0245
88.5 74.207 1.5033
91.5 78.874 2.0898
94.5 83.541 2.8001
97.5 88.208 3.6518
100.5 92.875 4.6644
103.5 97.542 5.8598
106.5 102.209 7.2619
109.5 106.876 8.8969

Table F.4 The values of the

base settlements of the upper F3 (kN) PF,3 (kPa) S3 (cm)
partition of the pile versus the
110.5 37.299 0.3124
loads on the soil
115.0 41.403 0.5383
119.5 45.507 0.8343
124.0 49.611 1.2122
128.5 53.714 1.6849
133.0 57.818 2.2669
137.5 61.922 2.9746
142.0 66.026 3.8259
[email protected] 146.5 70.129 4.8409
151.0 74.234 6.0421
155.5 78.337 7.4539
612 Appendix F

Fig. F.3 Generalized plot of load versus settlement for the pyramidal pile under vertical loading

The calculated load on the pile F at the given settlement S is obtained as the sum
of loads, transferred by the bottom end and the whole lateral surface of the pile

F = F0 + F1 + F2 + F3 . (F.15)

The corresponding functional dependence after the summation of Eqs. (F.11)–

(F.14) is given by

F = 26.951 ln (10−2 · S) + 362.591, (F.16)

its plot also being shown in Fig. F.2. Figure F.3 shows the sought plot of the pyrami-
dal pile’s base settlements at vertical load built in a convenient practical form using
Eq. (F.16).

[email protected]
Appendix G
Isolines of Contact Stress on a Lateral Surface of
a Slotted Foundation

The appendix contains the results of boundary-element calculations of contact stress

at the lateral faces of a slotted foundation with a rectangular shape of the longi-
tudinal cross-section (the simplest construction of a foundation under columns of
reinforced concrete and steel). The calculation scheme for the corresponding spatial
problem is shown in Fig. G.1.
The deformation parameters of the soil base are taken as follows: deformation
modulus E = 20 MPa, Poisson coefficient ν = 0.4 (cohesive slightly wet eluvial
clayey in a solid state).
The following spatial actions on the slotted foundation are considered:

(1) a central vertical compressive load;

(2) a combined action of a central vertical force and a tilting moment, acting within
the longitudinal cross-section plane (a vertical load with an eccentricity in the
longitudinal direction);
(3) a tilted force load within the longitudinal cross-section plane;
(4) a torque around the vertical axis;
(5) a tilted force load acting orthogonally to the longitudinal cross-section plane;
(6) a torque load acting orthogonally to the longitudinal cross-section plane;

[email protected]
Fig. G.1 Calculation scheme
at spatial loads of a slotted
foundation with a rectangular
longitudinal cross-section

613
614 Appendix G

(7) a combined action of a central vertical force and a tilting moment, acting orthog-
onally to the longitudinal cross-section plane (a vertical load with an eccentric-
ity in the lateral direction)

which are typical for calculations of bases for deformations under the main com-
binations of loads when the above-foundation constructions transmit, in general
case, tilted forces and torques onto the slotted foundations.
At the chosen coordinate system on the lateral surface of the slotted foundation
the normal stress are σ n = ±px , and the tangential stress vector lies within the
contact surface (Fig. G.1).
Figures G.2–G.15 represent isolines of dimensionless contact stress:

(a) normal σ n = σn /pav ,

(b) tangential τ = τ/pav ,
(c) horizontal py = py /pav ,
(d) vertical pz = pz /pav .

As a scale factor for the contact stress the average pressure pav = 10 kN/Sf =
42.69 kN/m was used where Sf = 23.42 m2 is the area of the foundation and soil
contact surface.
Taking into account the fact that in the case of the above listed (1)–(4) type loads
the medial longitudinal cross-section is a symmetry plane in the calculation scheme,
the isolines of the contact stress on the lateral faces will be symmetric or skew
symmetric, and, hence, for these cases only the calculation results for the slotted
foundation face x = 0.3 m are given (Figs. G.2, G.3, G.10–G.15).
For the case of (5)–(7) type loads the medial longitudinal cross-section is not
a symmetry plane in the calculation scheme, and the spatial stressed states of the
base are symmetric only with respect to the medial lateral cross-section; hence, the
isolines of the contact stress are given for both lateral faces of the slotted foundation
at x = ± 0.3 m (Figs. G.4–G.9).

[email protected]
Appendix G 615

(a) (b)

(c) (d)

Fig. G.2 Contact normal σ n /ppv (a), tangential τ /pav (b), horizontal py /pav (c), and vertical pz /pav
(d) stresses at the lateral surface of a slotted foundation (x = 0.3 m) under a central vertical force
Pz = 103 kN

[email protected]
616 Appendix G

(a) (b)

(c) (d)

Fig. G.3 Contact normal σ n /ppv (a), tangential τ /pav (b), horizontal py /pav (c), and vertical pz /pav
(d) stresses at the lateral surface of a slotted foundation (x = 0.3 m) under the action of a tilting
moment My = 1.4·103 kN. m

[email protected]
Appendix G 617

(a) (b)

(c) (d)

Fig. G.4 Contact normal σ n /ppv (a), tangential τ/pav (b), horizontal py /pav (c), and vertical pz /pav
(d) stresses at the lateral surface of a slotted foundation (x = 0.3 m) under the combined action of
a force and a torque load, Pz = 103 kN, My = 0.7·103 kN. m

[email protected]
618 Appendix G

(a) (b)

(c) (d)

Fig. G.5 Contact normal σ n /ppv (a), tangential τ/pav (b), horizontal py /pav (c), and vertical pz /pav
(d) stresses at the lateral surface of a slotted foundation (x = –0.3 m) under the combined action
of a force and a torque load, Pz = 103 kN, My = 0.7·103 kN. m

[email protected]
Appendix G 619

(a) (b)

(c) (d)

Fig. G.6 Contact normal σ n /ppv (a), tangential τ/pav (b), horizontal py /pav (c), and vertical pz /pav
(d) stresses at the lateral surface of a slotted foundation (x = 0.3 m) under the combined action of
a force and a torque load, Pz = 103 kN, My = 1.4·103 kN. m

[email protected]
620 Appendix G

(a) (b)

(c) (d)

Fig. G.7 Contact normal σ n /ppv (a), tangential τ/pav (b), horizontal py /pav (c), and vertical pz /pav
(d) stresses at the lateral surface of a slotted foundation (x = –0.3 m) under the combined action
of a force and a torque load, Pz = 103 kN, My = 1.4·103 kN. m

[email protected]
Appendix G 621

(a) (b)

(c) (d)

Fig. G.8 Contact normal σ n /ppv (a), tangential τ/pav (b), horizontal py /pav (c), and vertical
pz /pav (d) stresses at the lateral surface of a slotted foundation (x = 0.3 m) under a tilting force
R = 103 kN, β = 135◦ normally to the longitudinal cross-section plane

[email protected]
622 Appendix G

(a) (b)

(c) (d)

Fig. G.9 Contact normal σ n /ppv (a), tangential τ/pav (b), horizontal py /pav (c), and vertical pz /pav
(d) stresses at the lateral surface of a slotted foundation (x = –0.3 m) under a tilting force
R = 103 kN, β = 135◦ normally to the longitudinal cross-section plane

[email protected]
Appendix G 623

(a) (b)

(c) (d)

Fig. G.10 Contact normal σ n /ppv (a), tangential τ/pav (b), horizontal py /pav (c), and vertical pz /pav
(d) stresses at the lateral surface of a slotted foundation (x = 0.3 m) under a horizontal force R =
103 kN, β = 180◦ normally to the longitudinal cross-section plane

[email protected]
624 Appendix G

(a) (b)

(c) (d)

Fig. G.11 Contact normal σ n /ppv (a), tangential τ/pav (b), horizontal py /pav (c), and vertical pz /pav
(d) stresses at the lateral surface of a slotted foundation (x = 0.3 m) under a tilting force R = 103 kN
within the longitudinal cross-section plane; α = 135◦

[email protected]
Appendix G 625

(a) (b)

(c) (d)

Fig. G.12 Contact normal σ n /ppv (a), tangential τ/pav (b), horizontal py /pav (c), and vertical pz /pav
(d) stresses at the lateral surface of a slotted foundation (x = 0.3 m) under a tilting force R = 103 kN
within the longitudinal cross-section plane; α = 180◦

[email protected]
626 Appendix G

(a) (b)

(c) (d)

Fig. G.13 Contact normal σ n /ppv (a), tangential τ/pav (b), horizontal py /pav (c), and vertical pz /pav
(d) stresses at the lateral surface of a slotted foundation (x = 0.3 m) under an eccentric vertical force
Pz = 103 kN; ε y = –0.7 m

[email protected]
Appendix G 627

(a) (b)

(c) (d)

Fig. G.14 Contact normal σ n /ppv (a), tangential τ/pav (b), horizontal py /pav (c), and vertical pz /pav
(d) stresses at the lateral surface of a slotted foundation (x = 0.3 m) under an eccentric vertical force
Pz = 103 kN; ε y = −1.4 m

[email protected]
628 Appendix G

(a) (b)

(c) (d)

Fig. G.15 Contact normal σ n /ppv (a), tangential τ/pav (b), horizontal py /pav (c), and vertical pz /pav
(d) stresses at the lateral surface of a slotted foundation (x = 0.3 m) under the action of a torque
Mz = 0.5·103 kN. m

[email protected]
Appendix H
Numeric Schemes of Volume Integration

The volume integrals contained in the integral representations of Chap. 6 are finally
reduced to the integrals of the form

 
P(x,y,z)V (x,y,z)dV = f (x1 ,x2 ,x3 )dV (H.1)
V V

which, as a rule, cannot be calculated analytically due to a complicated form of

the integrands and the intervals of integration. Since in this case the integrand
functions remain limited in the corresponding ranges, i.e. all the volume inte-
grals under consideration do not contain singular parts, one does not need to
use any special regularization techniques, and the integration can be performed
numerically.
In order to calculate the integrals of the form of Eq. (H.1) in the point K(ξ ,
η, ζ )∈V
/ we perform a discretization of the interval V to elementary volumes
Vi ,i = 1,n. Further we assume the global interval V of a sufficiently arbitrary
shape to be presented with any degree of accuracy as a combination of a finite num-
ber of pyramids, triangular and quadrangular prisms. Some details of such repre-
sentation were considered in Sect. 6.4. Besides, we consider that in the vertices
of the elementary volumes (the nodes of the spatial network) the integrand func-
tion values are known or can be rather easily calculated. Each of the elementary
volumes under consideration admits transformation into canonical elements with
standard integration limits. Figure H.1 presents three types of spatial basic ele-
ments with the introduced system of local coordinates &1 ,&2 ,&3 , used at the numer-
ical integration. Equation (H.1) is transformed in the local coordinates to the form
[8, 10]

  
f (x1 ,x2 ,x3 )dV = f̃ &1 ,&2 ,&3 )fracD(x1 ,x2 ,x3 )D(&1 ,&2 ,&3 )d&1 d&2 d&
Vj j
[email protected]
V
(H.2)

629
630 Appendix H

(a)

(b)

(c)

Fig. H.1 Three types of canonical volume elements

+ +
+ ∂x ∂x ∂x +
+ +
+ ∂&1 ∂&2 ∂&3 + ++ +
G G2 G3 ++
+
D(x1 ,x2 ,x3 ) + ∂y ∂y ∂y ++ ++ 1
J(x,&) = =+ + = G4 G5 G6 + =
D(&1 ,&2 ,&3 ) + ∂&1 ∂&2 ∂&3 + ++ G G G ++
+ ∂z
+ ∂z ∂z ++ 7 8 9
+ ∂& +
1 ∂&2 ∂&3
= G1 (G5 G9 − G8 G6 ) − G2 (G4 G9 − G7 G6 ) + G3 (G4 G8 − G7 G5 )

where ΔVi is the standard volume, J(x,&) is the transformation Jacobian.

Since the procedure of the volume integral calculation should be optimized,
[email protected]
we assume the integrand function to vary linearly at each element. The use of a
quadratic or cubic approximation results in a considerable increase of the number
of nodes at each element. This, in turn, due to the complicated form of the integrand
Appendix H 631

function and a great number of calls to them, leads to a considerable increase of

the computation time. Then the accuracy, required for the practical purposes, will
be achieved for the expense of the increase of the integration interval discretization
degree, as a rule, by an adaptive increase of the calculation network density at the
sections with higher gradients.
Thus, we restrict ourselves to the linear isoparametric volume elements for which
the geometrical and functional nodes coincide. Global Cartesian coordinates of any
internal points of the element and integrand functions are presented by an expansion
over the same basis functions N α (&i )

ρ
'
xi (&i ) = N α (&i ) · Xiα ,
α=1
ρ
'
f (&i ) = N α (&i ) · fα , i = 1,2,3
α=1

where Xiα are the geometric node coordinates, fα are the functional values in the
nodes, ρ is the number of nodes of the isoparametric element.
Here we give the detailed expressions for the basic functions and Jacobians for
the three types of volume elements most often used in practice (Fig. H.1) as well as
present the formulae for the transition from the corresponding triple integrals to the
iterated integrals included into the software elaborated.
Tetrahedral element, ρ = 4 (Fig. H.1a)

Nα = &α (α = 1,2,3,4)

where 0 ≤ &α ≤ 1 are the space coordinates,

&1 + &2 + &3 + &4 = 1 (or &4 = 1 − &1 − &2 − &3 ) ,

x = &1 X1 + &2 X2 + &3 X3 + (1 − &1 − &2 − &3 )X 4 ,
y = &1 Y 1 + &2 Y 2 + &3 Y 3 + (1 − &1 − &2 − &3 )Y 4 ,
z = &1 Z 1 + &2 Z 2 + &3 Z 3 + (1 − &1 − &2 − &3 )Z 4 ,
∂Nα ∂N4
= δαi ; = −1 ,
∂&i ∂&i
G1 = X 1 − X 4 , G2 = X 2 − X 4 , G 3 = X 3 − X 4 ,
G4 = Y 1 − Y 4 , G5 = Y 2 − Y 4 , G6 = Y 3 − Y 4 ,
G7 = Z 1 − Z 4 , G8 = Z 2 − Z 4 , G9 = Z 3 − Z 4 ,
+ 1 +
+ X − X4 X2 − X4 X3 − X4 +
+ +
D(x1 ,x2 ,x3 ) + 1
[email protected] +
= + Y − Y4 Y2 − Y4 Y 3 − Y 4 + = mes Vj ,
D(&1 ,&2 ,&3 ) + +
+ Z1 − Z4 Z2 − Z4 Z3 − Z4 +
632 Appendix H

 1  1
1−&  1 −&2
1−&

= g(&1 ,&2 ,&3 )d&1 d&2 d&3 = d&1 d&2 g(&1 ,&2 ,&3 )d&3 =
Vj 0 0 0

1 1 1
(1 − &1 )2 d&1 (1 − &2 )d&2 g(&1 ,&2 ,&3 )d&3 =
(H.3)
0
 6 1 0 0
1  1 1
= (1 − &1 )2 d&1 (1 − &2 )d&2 g(&1 ,&2 ,&3 )d&3 =
2
 6 −1
 −1 −1
1
= (1 − &1 )2 (1 − &2 )g(&1 ,&2 ,&3 )d&1 d&2 d&3
2

where is a standard (unit) cube,

1 1
g(&1 ,&2 ,&3 ) = Vj · f̄ (&1 ,&2 ,&3 ) ,&1 = (1 + &1 ) , &2 = (1 − &1 )(1 + &2 ) ,
2 4
1
&3 = (1 − &1 )(1 − &2 ) .
2

8
Triangular-prismatic element, ρ = 6 (Fig. H.1b),

1
N1 = (1 − &1 )&2 ,
2
1
N2 = (1 − &1 )&3 ,
2
1
N3 = (1 − &1 )(1 − &2 − &3 ) ,
2
1
N4 = (1 + &1 )&2 ,
2
1
N5 = (1 + &1 )&3 ,
2
1
N6 = (1 + &1 )(1 − &2 − &3 ) ,
2
(H.4)
−1 ≤ &1 ≤ 1 , 0 ≤ &2 , &3 ≤ 1 .
x = N1 X 1 + N2 X 2 + N3 X 3 + N4 X 4 + N5 X 5 + N6 X 6 ,
y = N1 Y 1 + N2 Y 2 + N3 Y 3 + N4 Y 4 + N5 Y 5 + N6 Y 6 ,
z = N1 Z 1 + N2 Z 2 + N3 Z 3 + N4 Z 4 + N5 Z 5 + N6 Z 6 ,
∂N1 ∂N4 1
= = &2 ,
∂&1 ∂&1 2
∂N1 ∂N4 ∂N3 ∂N3 1
=
[email protected] =− =− = (1 − &1 ) ,
∂&2 ∂&3 ∂&2 ∂&3 2
∂N1 ∂N2 ∂N4 ∂N5
= =− =− =0,
∂&3 ∂&2 ∂&3 ∂&2
Appendix H 633

∂N2 ∂N5 1
=− = − &3 ,
∂&1 ∂&1 2
∂N3 ∂N6 1
=− = − (1 − &2 − &3 ) ,
∂&1 ∂&1 2
∂N4 ∂N5 ∂N6 ∂N6 1
= =− =− = (1 + &1 ) ,
∂&2 ∂&3 ∂&2 ∂&3 2
 
D(x1 ,x2 ,x3 )
f̃ (&1 ,&2 ,&3 ) d&1 d&2 d&3 = g(&1 ,&2 ,&3 )d&1 d&2 d&3 =
D(&1 ,&2 ,&3 )
Vj Vj
1 1  2
1−& 1 1 1 1
= d&1 d&2 g(&1 ,&2 ,&3 )d&3 = d&1 (1 − &2 )d&2 g(&1 ,&2 ,&3 )d&3 =
−1 −1 −1 8 −1 −1 −1

1
(1 − &2 )g(&1 ,&2 ,&3 )d&1 d&2 d&3
8
(H.4)
1 1
where &2 = (1 + &2 ) ; &3 = (1 + &3 )(1 − &2 ) .
2 4
Hexahedral element, ρ = 8 (Fig. H.1c)

1
Nα = (1 + Sα1 &1 )(1 + Sα2 &2 )(1 + Sα3 &3 ) , α = 1,2,...,8.
8

⎡ ⎤
−1 , −1, −1
⎢ −1 , −1 ⎥
⎢ 1, ⎥
⎢ ⎥
⎢ −1 , 1, 1 ⎥
⎢ ⎥
⎢ ⎥
⎢ −1 , −1, 1 ⎥
Sαi  = ⎢
⎢ 1,
⎥ arethei − th uniform coordinates of the nodeα,
⎢ −1, −1⎥⎥
⎢ ⎥
⎢ 1, 1, −1 ⎥
⎢ ⎥
⎢ ⎥
⎣ 1, 1, 1 ⎦
1, −1, 1

−1 ≤ &1 ,&2 ,&3 ≤ 1 ,

x = N1 X 1 + N2 X 2 + ... + N8 X 8 ,
y = N1 Y 1 + N2 Y 2 + ... + N8 Y 8 ,
z = N1 Z 1 + N2 Z 2 + . . . + N8 Z 8 ,
∂Nα Sα1
= (1 + Sα2 &2 )(1 + Sα3 &3 ) ,
∂&1 8
[email protected]
∂Nα Sα2
= (1 + Sα1 &1 )(1 + Sα3 &3 ) ,
∂&2 8
634 Appendix H

∂Nα Sα3
= (1 + Sα1 &1 )(1 + Sα2 &2 ) ,
∂&3 8
 
D(x1 ,x2 ,x3 )
f̄ (&1 ,&2 ,&3 ) d&1 d&2 d&3 = g(&1 ,&2 ,&3 )d&1 d&2 d&3 =
D(&1 ,&2 ,&3 )
Vj Vj

1 1 1
= d&1 d&2 g(&1 ,&2 ,&3 )d&3 = g(&1 ,&2 ,&3 )d&1 d&2 d&3
−1 −1 −1
ω

The integrals of Eq. (H.3)–(H.5) are reduced to the same canonical form with the
unit integration limits admitting multiplicative approximation [8, 10]

1 1  1 
n 
m 
p
(n) (m) (p) (n) (m) (p)
F(q1 ,q2 ,q3 )dq1 dq2 dq3 = wi wj wk F(q1i ,q2j ,q3 k )
−1 −1 −1 i=1 j=1 k=1

where (q1i , q2j , q3k ) are coordinates of the nodes whose location is determined by
the type of the integration formula applied, and wi , wj , wk are the corresponding
weight factors. As known from [10], the greatest effect is achieved by the application
of the Gauss formulae, enabling the number of integration points to be essentially
reduced. This gives the required achievement of the given calculation accuracy with
a simultaneous reduction of the machine time. Due to the regularity of the integrals
under consideration we used in our calculations the nodes and weights of the two-
point Gaussian quadrature formula q (1) = −q (2) = 0.5773503, w (1) = w (2) =
1.0, providing exact integration of the third-order polynomials. The number of the
integration points over each coordinate is assumed to be the same (n = m = p = 2)
and the integrand should be calculated in eight points. Besides, note that at large-
scale calculations of the volume integrals the presence of the unit weight factors
in the quadrature used enables the number of arithmetic actions to be to a certain
extent reduced and the total calculation uncertainty to be lowered.

References
1. Abramowitz M, Stegun I A (1972) Handbook of mathematical functions with formulas,
graphs, and mathematical tables, 10th edn. Dover, New York
2. Aleynikov S M (1996) Ring-shaped foundation with an internal cut-off circle. In: Calculation
and design of bases and foundations at complex geotechnical conditions. VGASA, Voronezh,
pp. 9−16 (in Russian)
3. Aleynikov S M, Belozerov V A (1999) Ring-shaped foundation under tower-type construc-
tions on a wedge-shaped base. Patent 2135694 RU, MPK6 E 02 D 27/42.No. 98102534/03
(Russia)
4. Aleynikov S M, Ikonin S V (1998) Reinforcement of a potable water reservoir by means of
piles. In: Proc. 6th Intern Conf Pile Found Eng, vol. 4. Reinforcement of bases and foundations
[email protected]
under the reconstruction of buildings. RNKMGIF, Moscow, pp. 5−10 (in Russian)
5. Aleynikov S M, Nekrasova N N (1992) Numerical calculations of orthotropic plates on a
variable-thickness elastic base. In: Calculation and design of bases and foundations at complex
geotechnical conditions. VISI, Voronezh, pp. 22–31 (in Russian)
Appendix H 635

6. Aleynikov S M, Nekrasova N N (1996) Bending of otrhotropic foundation plates resting on

elastic nonclassical bases. Izv Vuzov Stroit (issue 9): 65–71 (in Russian)
7. Aleynikov S M, Nekrasova N N (1996) Contact problem for orthotropic foundation slabs
with consideration of deformation peculiarities of spatial and nonhomogeneous bases. Studia
Geotechnica et Mechanica 20: 63–104
8. Altenbach J, Sacharov A S (eds) (1982) Die Methode der Finiten Elemente in der Festkörper-
mechanik. Fachbuchverlag, Leipzig
9. Bakhvalov N S (1973) Numerical methods, vol. 1. Nauka, Moscow (in Russian)
10. Banerjee P K, Butterfield R (1981) Boundary element methods in engineering science.
McGraw-Hill, New York
11. Brebbia C A, Walker S (1980) Boundary element techniques in engineering. Newnes-
Butterworths, London
12. Brebbia C, Telles J, Wrobel L (1984) Boundary element techniques. Springer-Verlag, Berlin
13. Cowper G R (1973) Gaussian quadrature formulas for triangles. Intern J Num Method Eng 7:
405–408
14. Engels H (1980) Numerical quadrature and cubature. Academic Press, London/New York
15. Gorbunov-Posadov M I, Malikova T A, Solomin V I (1984) Calculation of structures on an
elastic base. Stroyizdat, Moscow (in Russian)
16. Hammer P C, Marlowe O P, Stroud A H (1956) Numerical integration over simplexes and
cones. Math Tables Aids Comp 10: 130–137
17. Instructions on design of pyramidal pile foundations (1983) Glavn Upr Project Organiz,
Moscow (in Russian)
18. Kiselev V A (1973) Calculation of plates. Stroyizdat, Moscow (in Russian)
19. Klokov Yu A, Shkerstena A Ya (1986) Numerical integration of functions using interpola-
tional polynomials of Chebyshev type. Latvian Math Annual 30: 207–217 (in Russian)
20. Korn G A, Korn T M (1961) Mathematical handbook for scientists and engineers. McGraw-
Hill, New York
21. Kronrod A S (1965) Nodes and weights of quadrature formulas. Consultants Bureau, New
York
22. Krylov V I, Shulgina L T (1966) Handbook on numerical analysis. Nauka, Moscow (in
Russian)
23. Li H B, Han G M, Mang H A (1986) Quadrature of singular integral in the direct BEM applied
to tree-dimensional elasto-plasticity. Z Angew Math Mech 66: 314–316
24. Paszkowski S (1975) Zastosowania numerycne wielomianów i szeregów Czebyszewa. PWN,
Warszawa
25. Shkerstena A Ya (1984) Interpolated Lagrange formulae, constructed from Chebyshev poly-
nomial node. Latv Mathem Ann 28: 130–142 (in Russian)
26. Stroud A H, Secrest D (1966) Gaussian quadrature formulas. Prentice-Hall, Englewood Cliffs
27. Suyetin P K (1976) Classical orthogonal polynomials. Nauka, Moscow (in Russian)
28. Timoshenko S, Wojnowski-Krieger S (1959) Theory of plates and shells. McGraw-Hill, New
York
29. Ugodchikov A G, Khutoryanskiy N M (1986) Boundary-element method in mechanics of
deformable solids. Kazan State University, Kazan (in Russian)
30. Vasilyev N I, Klokov Yu A, Shkerstena A Ya (1984) Application of Chebyshev polynomials
in numerical analysis. Zinatne, Riga (in Russian)
31. Zienkiewicz O C (1971) The finite element method in engineering science. McGraw-Hill,
London

[email protected]
[email protected]
Index

A structure, design, and specific features,

Abramov effect, 257 425–428
Accuracy cubature formula, 562 vertical cylindrical piles, 428–435
Adaptive integration schemes, special, 75 Base settlements, determination of, 601,
Alpha piles, 387 603–604
Analytical-and-numerical integration method, BEM-based numerical modeling, 91
106 Bending moments, 216, 252–253, 271, 312,
Analytical solutions, 109, 138, 197, 216, 242, 315, 355–357, 359, 364, 367, 369,
255, 257–259, 268, 334, 476, 512 371–372, 400, 425, 592
Analytic language, 107 Berezantsev’s formulae, 348
Anchor axial rigidity, 401 Bessel function, 14, 32–33, 52, 71, 204, 345
Anchor foundation plates, 224 first-order, 344
Anchor structures, 387 Betti’s theorem, 92–93, 273, 518
Anisotropic properties of the soil, 238 Biharmonic Love function, 26
Anisotropic strength conditions, 486 Block diagonalization, 146
Anisotropy, 4, 263, 266, 277, 356, 398, 486, Bored pile foundation, 187, 385, 389, 438–445,
507, 524 447–454
Approximate analytical solutions, 258 Bored pile foundation with widening, 438–454
Artesian water, Japan, 506 deformations calculation, 443–448
Asymmetrical foundation structure, 435 displacements calculation of and slopes,
Asymptotic methods of ‘large and small λ’, 448–454
261 engineering methods for calculation,
Automatic boundary-element grid formation, 441–443
143 production and structures, 439–441
Axial rigidity, 400
Botkin criterion, 411
Axisymmetric
Botkin dependence, 265
calculation scheme, 410
Botkin model, 271
contact problem, 125
Boundary-element
algorithm, 109, 131, 136, 139, 197, 216,
B 229, 233, 242, 253, 300, 538–539,
Back fill, 426, 454 566
Backhoe, 456 approach, 205, 223–224, 228, 243, 252,
Back pressure forces, 476 276, 419, 454, 472, 475
Bangkok, 506 discretization, 98, 186, 192, 426
[email protected]
Base deformations, 390, 394 grid, 171, 174, 176–177, 179–180,
Bases and rigid bored foundations, 424–437 183–184, 190, 195, 198, 203, 208,
inclined piles and rectangular pile raft, 222–223, 229, 232, 273, 279, 303,
435–437 305, 307, 444, 448, 458–459, 478

637
638 Index

method (BEM), 1, 27, 52, 54, 82–93, 96, Circular foundation bases, 269
98–101, 113, 118–119, 121, 123, Circular (or ring-shaped) punch, indentation
128, 131, 135–137, 151, 162–163, of, 125
167–168, 171, 174, 176, 192, Circular punches, 128, 255, 259, 272
195–196, 202, 204–206, 216–217, slopes, 286
219, 225, 228–229, 233, 242–243, smooth punches, 276
251, 264, 272–277, 279, 311, 323, Clayey sands, 236
334, 338, 354, 361, 372, 385, 389, Coalinga oil field, 506
395, 399, 405, 417, 420, 422, 437, Coefficient matrix, 139, 144, 265, 418, 540
445, 447, 459, 470–472, 474–476, Cohesion parameters, 488
478, 511, 519, 534, 538–541 Collapsing, 357, 390–392, 410, 415
node, 106, 138, 190 Collocation method, 97
plane, 106
Complex-shaped foundation, 188, 252, 294,
polyhedron surface, 228
327, 538
solution, characteristics of the, 194, 202
rigid, 389
Boundary integral equations, 64, 91–96, 99,
Complex-shaped punches, 91, 123, 163, 216,
106, 111, 116, 119–120, 146, 151,
251, 255, 295, 324, 339, 540
162, 217, 272, 405, 522
Compliance functions, 37
numerical algorithms, 334–351
Compressed soil mass thickness, 286, 371
finite-measure analogue, 339–341
Compressibility effect, 50
nonlinear contact equations, 337–338
nonlinear deformation effects, 348–351 Compression tests, 471
nonlinearly deformable base, 335–337 Computer-aided design, 146, 355, 539
round punch contact problem, 341–348 Concrete mixture, 426, 493
Boundary macroelement, 176–180, 184–186, Condensation subdomain, 155
188–190, 192 Condensed zones (kernels) formation of, 441
Boundary node, 155, 163–164, 168 Conditionality estimations, 363
Boundary-superelement method, 146 Cone generatrix, 229–230, 232
Boussinesq-Cerruti solution, 12, 120, 131, 551 Conical punches, 233, 235
Boussinesq equation, 60, 274 Constant-width foundations (CWFs), 311,
Boussinesq fundamental solutions, 8 312–315, 320
Boussinesq law, 399 Construction Rules and Regulations
Boussinesq solution, 7, 14, 16, 59, 65–66, 68, 2.02.01–83, 388–390, 393–394, 405, 472,
76, 272–273, 342, 345, 399, 512, 474
523, 548 2.02.03–85, 403, 411, 425
Bubnov-Galerkin type method, 262 Contact deformation, semi-empirical models,
Bulk deformation modulus, 265 253
Burmister solution, 17, 275 Contact friction, 277, 487–490, 539
parameters of, 406
C Contact interaction characteristics, 279, 289,
Cam-Clay model, 412 371, 432, 450, 539
Canonical equations, 302, 362, 397, 399 Contact models, development and analysis of,
Capacity-based calculations, 390 539
Carrying capacity, analysis of, 408 Contact pressure
Cartesian coordinate system, 12, 28, 62, 92, curve, 282
102, 105, 107, 110, 137, 176, 218, function, 121–122, 195, 257, 260, 262,
542, 625 327, 337–338, 354
Cauchy principal value, 94 isolines, 203, 313
Cauchy relations, 94 Contact problems for anchors, 401
[email protected]
Cerruti displacement functions, 127 Contact stress, determination of, 92, 258
Cerruti solution, 7–8, 127, 129 Contact stress functions, 120, 274, 444, 519
Chebyshev polynomials, 204, 562–565 Contact tangential stress functions, 127
Cinematic loading scheme, 271 Contact zones formation, 429
Index 639

Correction and improvement of design Detachment zones, 251, 462

regulations, 540 3-D finite-element, 166, 174
Coulomb fluidity, 408 Diaphragm wall technique, 454–455
Coulomb friction law, 406, 482 Differential equation, 29–30, 264, 270, 277,
Coulomb strength law, 486 353, 358, 360–361, 393, 510, 586
Coupled elastic half-spaces, 513 Dilatation, 30, 32, 36, 39, 62, 270, 511–513,
Coupled half-spaces, 8–12 517, 523–524, 532, 550
Crack-proofness, 386, 392 equations, 513
Crank mechanisms, 393 function, 30, 32, 36, 39, 511–513, 517,
Cross-shaped foundation, 328 523, 550
tests, 312, 315 Dinite-difference equation, 264, 361, 586, 588,
Crushers, 180, 393 591
Cyclic discretization, 113, 118, 125, 130 Dipper mechanisms, 465
Cylindrical coordinate system, 59, 61–62, 108, Dirac delta function, 2
110–111, 115, 184, 444 Dirac δ-function, 66
Cylindrical foundations, 186–187 Direct boundary-element
Cylindrical punch, 224, 428 formulation, 118
axisymmetrical contact problem, 225 method, 96, 106, 112, 116–117, 218, 301,
418, 448
D Dirichlet cell, 167
2-D and 3-D images, visualization of, 141 Dirichlet problem, 5
Darcy’s equation, 509, 511 Dirichlet-Voronoy cells, 171, 193
Deepening effect, 400, 409 polygons, 173, 195, 197–198
Deep pile settlements, 406 Discretization
Deflection isolines, 367 circular domain, 161
Deflection nonuniformity, 367 complex-shaped surface, 175
Deformable reservoirs, 508 contact domain, 121, 124, 128, 204, 216,
Deformation-based calculation, 390, 392, 394, 339, 428
403, 415–416, 447, 490–491 contact surface, 97, 177, 229, 445, 458
Deformation characteristics, 8, 56, 234, 242, 2-D domains, 150–174
271–272, 278, 403, 405, 439, 445, algorithm of triangulation, 151–162
457, 471, 577–578, 585 dual grids applications, 162–174
Deformation components, 270 degree, 102, 142, 175, 195, 265, 292, 313,
Deformation modulus, 1, 55–58, 60, 64–70, 327, 364, 457, 459–460, 625
72–75, 78–82, 135, 201, 203, domain, 162
228, 234, 238–239, 242, 253, 261, foundations with pile rafts, 190
265, 275, 292–294, 305, 309, 337, numerical modeling, 151
353, 356, 398, 401–402, 405, 408, procedure, 154, 159, 401
419–420, 445, 457, 471, 607 rigid inclusion contact surface, 218
parabolic law of, 65 rotation surface, 185
square law of, 67 scheme of, 179, 191
Deformation of soil, spatial processes of, 255 spherical surface, 220
Deformation parameters, 12, 43, 238, 243, 459, triangular boundary, 180
465, 474–475, 477, 533, 607 Disc-shaped anchor, 400
Deformation properties, 1, 4, 55, 234, 242, Displacement
251, 265, 269–270, 276, 292, 313, components, 256
334, 358, 367, 414 decay, 67
variation of, 14, 74 matrix, 3
Deformations due to soil compression, 390 piles, 181, 386, 399, 410, 438–439, 456
[email protected]
Deformation stage spatial contact models, 335 tensor, 3, 8–9, 542
Deformation tensor, 351, 510 vector components, 3, 9, 26, 93, 121
Delaunay triangulation, 165 Double-precision calculations, 363
Desalination degree, 275 Drop caissons, calculation of, 55
640 Index

Dual boundary-element grids, efficiency Elliptical contact domains, 123

characteristics, 217 Elliptical punches, 128, 259
Dual grid, 151, 162–166, 168–173, 194, 201, Elliptical-shaped contact domains, 123
212, 216 Elliptic centrosymmetric, 526
application, 216 Eluvial clayeys, 458
cell, 169 Engineering-and-geological conditions, 25, 56,
examples, 165 79, 242, 252, 412, 415, 424, 525,
polygonal, 197, 205–216 534, 540
rectangular, 166 Engineering theory of beam bending, 263

F
E
Fatigue crack nucleation, 279
Earth-moving excavator-based machines, 456
FEM, see Finite-element method
Earthquakes, 403, 506–507
Ferroconcrete, 356, 438, 457, 473
Eccentric ring-shaped punch, 210, 213,
Fictitious force method, 22
309–310
Field experiments, 56, 410, 425, 457, 470
Egorov influence function, 275, 575
Filtration
Egorov solution, 15, 17, 290
coefficient, 514
δ2 -Eitken transformation, 339 consolidation, 505
Elastic base rigidity matrix, 362 Finite-difference approximation, 264, 362,
Elastic bases with deformation modulus, 55–82 589–590
nonhomogeneous half-space surface, Finite-difference grid, 174, 265, 354, 361–362,
63–82 364–365, 368
normal concentrated force, 58–63 nodes, 361
variation with depth, 55–58 Finite-difference method (FDM), 136, 138,
Elastic compressible wedge, 285, 290, 304, 145, 253, 264–266, 270, 354–356,
363, 365–368 365, 372, 395, 402, 404, 593
Elastic foundation model, 593 Finite-difference relations, 587
Elastic half-space, 93, 292, 397, 512 Finite-difference software, 354
boundary, 400 Finite-element algorithm, 354, 357
homogeneous, 91, 130, 192, 200, 203–205, Finite-element formulation, 355
212, 219, 228, 273, 292, 406, 526 Finite-element method, 146, 175–176, 218,
model, 397 242, 264, 266–272, 277–278,
torsion, 116, 130 354–355, 364–365, 395, 407,
Elastic homogeneous isotropic half-space, 5–8 409–412, 416–417, 420
Boussinesq and Cerruti solutions, 6–8 computer implementation, 176
Mindlin’s solution, 5–6 matrix, 146
Elastic layered bases, 12–55 Finite-measure analogue, 96–101
constant-width layer, 12–17 Finite-thickness elastic layer, 27, 43, 50, 275,
multilayer half-space, 25–55 345, 354
variable-thickness layer, 17–25 First-order model, see Mohr-Coulomb model
Elastic modulus, 77, 91, 258, 313 Fisher criterion, 349
Elastic non-classical models, 253 Flat-base punch settlement, 232
Elastic spatial wedge, 17 Flat boundary macroelements, 177
Elastic wedge, 17, 22, 281, 284–287, 290, 296, Flat graph theory, 165
306–307, 366, 368, 585 Flat ring-shaped punch, 131, 261
rib, 281, 284–287, 296 Flexible punch, calculation scheme for, 252
Elastoplastic deformation law, 414 Forging hammers, 180, 393
Elastoplastic deformation of soil bases, 395 FORTRAN, 137, 145, 159, 173, 178, 241, 327,
Elastoplastic problem, 410 338, 363, 566
[email protected]
Elastoplastic solution, 408–409, 411–412, 539 FORTRAN-77, 137, 159, 173, 327
Elastostatic problem, 401 FORTRAN-IV programming language, 566
Elastoviscoplastic problems, 408 Foundation bases with account of depth factor,
Electric modelling, 255 396–415
Index 641

Foundation engineering, 252, 276, 300, 385, Geotechnics, 1, 25–26, 97, 224, 234
415, 438, 454, 490 Geotextiles, 54
Foundation models, 254, 272, 403, 470 Gilbert boundary problem, 18
Foundation slope, 300–301, 391, 425, Gorbunov-Posadov solution, 364–365
577–579, 583–584 GOST standard, 320
Foundation structure, calculation of, 17, 64, Green’s displacement functions, 3
389, 540 Green’s displacement tensor, 8, 542
methods, 136 Green’s formulae, 18
principles for, 385 Green’s function, 3, 5, 404
theory, 388 Green’s tensor, 512–513
Foundation structures at reduced pore pressure, Grid
517–524 approximations, 192
finite-dimensional algebraic analogue, condensation, 152, 194–195
519–520 nodes, 174, 198, 211, 268
integral equations, 517–519
numerical algorithm, 520–522 H
shallow foundations, 522–524 Half-round abutting ends, 212
Foundation structures from deformations, Half-space
calculation of bases, 390–396 model, 255, 257, 300, 345, 353, 396–398,
Fourier method, 257, 403 401–403, 414, 424, 472, 490
Fourier series, 27, 352, 357 surface settlements, 64, 71–72, 78
Fourier transformation, 5, 26, 30–31, 33, 52, theory, 91
513 Hankel integrals, 27, 39, 52, 54
two-dimensional, 1, 27, 29, 54 Hankel operator, 258
Fourth-order differential equation, 397 Hankel transformation, 26–27, 259, 276, 401,
Fredholm equation, 263, 345, 401 403
Fredholm integral equations, 259 Hard cutting elements, 493
Friction resistance, 388, 456 Heaviside function, 241
Functional coefficient, Dependence of, 46–48 Hermit cubature formula, 563
Functioning horizontal wells, 514–516 nodes and weights of, 564
finite radius of well, 516 Heterogeneous stress-strained states, 385
predetermined intensity sources, 514–516 High λ method, 22
Hill-Tresk hypotheses, 407
Fuss-Winkler model, 350, 416, 442
Hollow conical pile, 386
Homogeneous half-space model, 64
G Homogeneous isotropic half-space, 542
Galerkin method, 261, 357 Homogeneous layer, 15, 599
Galerkin vector, 9, 26 Hooke equations, 94
Gauss Cubature formulae, 557 Hooke’s law, 4, 30, 60, 270, 402, 510, 512, 544
nodes and weights of, 559 Horizontal displacements, 15, 183, 256,
Gauss elimination, 101, 119, 139, 147–148, 258–259, 263, 390–391, 403, 415,
363 425, 430–432, 435–436, 442–443,
Gauss formulae, 628 449, 452–453, 462, 475, 508, 524,
Gaussian function, 262 527–528, 530–531
Gaussian surface equations, 174 Hydrocarbon fields, 507
Gauss-Legendre quadrature formula, 52, 77, 79 Hydrotechnical engineering, 54
Gauss method, 52, 147–148, 150, 363 Hypergeometrical function, 52
GAUSS program, 139
Gauss type methods, 521 I
Gazli gas field, Uzbekistan, 506, 508 Ilyushin elastic solution, 270
[email protected]
Geogrids, 54 Inadmissible deformations, 505
Geomechanics ADA code, 412 Inclination angle, 428, 430, 433, 435, 437, 450,
Geomechanics Creep–3, 408 452, 457, 460, 489
Geometrical scheme, 6, 124, 579 Inclined bottom foundations, 183, 387
642 Index

Industrial engineering, 181 Layer permeability, 507

Inelastic deformation, 539 Layer summation method, 395, 472
development of, 371 Least-square method, 27, 171, 255, 275, 308,
Infinite-measure matrix, 257 345, 602–605
Infinite stripe punch, 257 Legendre polynomials, 562–563
Influence matrix, 106–107, 115, 119, 121, Limiting transition, 12, 47, 81, 94, 203, 232
129–131, 141, 265, 273–274, 276, Linear algebraic equation system, 101, 139,
362–363, 403, 405 146–151, 218, 274, 364, 449, 592
coefficients, 119, 130, 273–274, 403 Linear algebra method, 302
Integral-differential operators, 13, 256 Linear and angular displacements,
Integral equation, two-dimensional, 128, 263 characteristics of, 568
Integral transformations, 26, 257–258 Linear law of deformation modulus, 64
Integrand function singularities, 559 Linearly deformable base theory, 350
Integration procedure, 102, 121, 521–522 Liquidity index, 470–471
Internal graphic software tool, 162 Load-carrying capacity, 183, 187–188
Inverse Fourier transformation, 32, 38 Logarithmic law, 516, 533
Inverse trigonometric law, 80 Lommel function, 52
Irregular grids of boundary elements, 197 Long Beach, USA, 506
Isobars, 198, 209, 212, 293, 313 Longitudinal-to-transverse dimension ratio,
Isolines of contact stress, 607 397
Isomorphism, 163 Love formula, 316, 319
Isotropic media, 400 L-shaped punch, 304–306, 330, 333
Iteration processes, 463
M
J
Macdonald function, 19
Jacobian transformation, 624
Jacobi or Seidel methods, 340 Macrofragmental soils, 236
Maizel’s method for thermoelasticity, 511
K Man-triggered damage, 506–507
Kantorovich-Lebedev transformation, 21 Maple, 107
Kelvin displacements, 4 Masts, 186, 577
Kelvin influence matrix, 107 Matcad, 107
Kelvin’s kernels, 138 Material nonhomogeneity, 277
Kelvin’s solution, 3–6, 12, 26, 102, 106–107, Mathematica, 107
119 Mathieu functions, 257
Kelvin transformation, 257 Mechanical behaviour of soil, 482
Kernel boundary point, 324 Mechanical properties of soils, determination
Kirchhof–Love hypotheses, 358 of, 234–235
Koltso software, 276 Meridional zone, 113–114, 444
Kronecker symbol, 2 Mesh condensation, 154
Metal-cutting machines, 393
L Mexico Engineering Research Institute, 506
Labour-consuming Mindlin displacement, 5–6, 102
calculations, 388 Mindlin equation, 93, 102, 401
solution, 398 Mindlin’s solution, 5–6, 8, 12, 27, 47, 91,
Lamé equations, 2, 29 96–97, 101, 106–107, 110, 119,
Landau order symbol, 67 121, 218–219, 273, 398–399,
Land surface 401–402, 405, 417, 517
deformation, 524, 528, 532, 534 Mises plasticity, 270
subsidence, 505–506 Mises–Schleicher–Botkin hypotheses, 407,
[email protected]
Laplace operator, 509 409, 411
three-dimensional, 3 Mises–Schleicher–Botkin strength condition,
Lateral expansion coefficient, 401 267
Lateral friction, 454 Mohr-Coulomb model, 412
Index 643

Mohr-Coulomb strength condition, 267 flat punches with a smooth base, 192–217
Moment of inertia, 216, 582 Numeric schemes of volume integration, 623
Monolithic
blocks, 180 O
foundation structures, 385 Obtuse-angle conical punches, 231–233
plates, 455 Obtuse-angled cones, dimensionless
MONOT model of elastoplastic type, 413 settlements, 233
Multiparametric method, 149 Oozes, 396
Multivariant calculations, 300, 313 Optimal solutions, 539
Orthogonal polynomials, 262
Orthotropic foundation plates, contact
N
problem, 351–372
Narrow grab bucket, 456
integro-differential equations, 358–361
Navier’s equations, 27
numerical modelling, 364–372
NBE array, 167
rectangular orthotropic plates, 361–363
NBK array, 167, 173
static calculations, 352–358
Negative (tensile) contact stress, 202
Orthotropic plate with free edges, 365
Neumann series, 18, 20
ORTOPLIT software, 363, 372, 593
convergence of, 22
Osaka, Japan, 506
Newton method of nonlinear equations, 340
Niigata, Honshu Island, Japan, 506 P
NNBK arrays, 173 Paired integral equations method, 258
Non-classical model, 258 PALOS software, 413
Nondeformable rod, 398 P-analytic functions, 109
Nonhomogeneous half-space settlements, 82 Papkovic-Neuber representation, 21
Nonlinear (elastoplastic) displacements, 407 Parabola, 204, 324
Nonlinear law of settlement variation, 574 Partial differential equations, 14, 255, 355
Nonlinearly deformable Peaty sand-clays, 396
finite-thickness layer, 343 Physical nonlinearities, 277
half-space, 342 Pile calculations, 224, 416–417
Nonlinear programming methods, 349 Pile drivers, 393
Nonlinear settlement-versus-load dependences, Pile foundation, 181, 183, 187, 190, 276,
343 386–387, 393–394, 397, 409,
Nonlinear soil base models, 404 411–413, 418, 422, 424–426,
Non-linear stage of deformation, 351 432–434, 437–441, 443, 454, 456,
Non-uniform base compressibility, 300, 577 517, 593
Non-uniform compression coefficient, 443 calculation of, 398, 401, 411, 472
Non-uniform deformations, 301, 391 construction, 181
Nonuniformity function, dimensionless, 79 deformation of, 393
Nonuniformity parameter, 73, 78, 80 displacement, 394, 424, 432, 437
Non-uniform settlements, 392 settlement, 393, 412
North Stavropol oil field, 508 Pile raft foundations, 189–190, 425, 432
Numerical algorithm testing, tool for, 131 Pile rigidity, 408
Numerical-and-analytical method, 52, 54, 75, Pile settlement
82, 91, 101–107, 115, 131, 148, behaviour, 431
218, 238, 239, 273, 403, 448, 524 curve, 420
Numerical calculations examples, 524–534 Pile shaft compressibility, 406
settlement and slopes of rigid foundation Pile sliding, 405
plates, 533–534 Piles with self-unfolding blades, 386
spatial deformation of land surface, Piles with triangular crosssection, 387
[email protected]
526–532 Pipe casing failures and crushing, 507
surface deformations of layer, 532–533 Planar graph theory, duality in, 151
Numerical modeling test examples, 192–243 Plastic deformation, 234, 278, 390, 395,
axisymmetric punches, factor for, 218–243 408–409, 411, 413, 480
644 Index

Plastic flow theory, 270, 407, 409–410 Puzyrevsky solution, 447

Plastic fluidity laws, 417 Pyramidal displacement pile, 386
Plate flexibility index, 364–365 Pyramidal piles, 415–424
Plate foundations, orthotropic properties of, application, 416
356 horizontal load action, 420–422
PLAXIS software, 412–413 inclined force and moment, 423–424
Plevako solution, 9, 12, 513, 517 inclined load action, 422–423
Poisson coefficient, 607 piles with variable cross-section, 416–420
Poisson ratio, 5, 8, 14–15, 17–18, 22–23, 50, settlements, 422
54–55, 57–58, 70–71, 91, 143, vertical load, 420
192, 195, 222–223, 228–232, 234, Pyramidal-prismatic foundations, 388
236, 256, 258–259, 263, 265, 274,
281, 290, 296, 303, 307, 309, 313, Q
353, 356, 359, 364, 399, 408, 417, Quadrangular boundary elements, 555
419, 445, 457, 470, 474, 526, 533, Quadratic approximation, 240
571–573, 580 Quasiuniform dependence, 193, 341
Polar coordinate systems, 104 Quasiuniform discretization of a square
Polish engineering regulations, 414 domain, 281
Polygonal cross-section punches, 208–209 Quasi-uniform grids, 180
Polygon-type boundary elements, 129 Q–Z curves, 413
Polynomial
R
approximation, 263
Radial deformations, 58
coefficients, 275
Reciprocality theorem, 260, 444
Pore liquid squeezing, 507
Rectangular punch, 212
Pore pressure, 505–506, 508, 510–512, 514,
Reference vertices, 154, 156, 160
516–517, 519–522, 524–525,
Reinforced concrete forming machines,
532–534
precast, 393
Pores and cracks, 505
Reisner square plate, 355
Porous elastic massif, 512
Relative punch displacement, 230
Postprocessor algorithm, 171 Relative settlement difference, 391–393
Power law, 56, 69–70, 73, 356, 465 Relative torsion angle, 391
Prandtl-Reiss equation, 270 Reverse arches, 387
Pre-conditions resulting, 388 R-function, 212, 216, 262, 324
Preprocessor algorithm, 151, 167, 169, 173 Rheological calculation models, 351
Prismatic foundations, 181 Rheological properties, 267, 277, 408, 413
Pulse effect, 493 Richardson method, 292, 313
Punch boundary, 97, 194–195, 297, 323 Riemann convolution, 511
Punch centre settlement, 285, 571, 574 Rigid disc, axisymmetric problem for a, 401
Punch contour, 204, 279, 325–326, 328–330 Rigid displacements, 95
Punch displacement, 91, 100, 115, 122, 147, Rigid flat-bottom punches, spatial contact
149, 202, 204, 275, 323, 337, 345 problems, 253–278
Punch equilibrium equations, 225, 259, 444 Rigid foundation plates, 300–311
Punch foundation, 255, 272 external load control, 303–307
Punch generatrix, 113, 115, 118 kernel boundary for, 323–334
Punch indentation, 121, 125–126 problem formulation and numerical
Punch settlement values, 273, 286 implementation, 301–303
Punch in the shape of a rotation body, 108–119 shape parameter control, 307–311
axisymmetric contact problem, 110–115 structural analysis, 243
axisymmetric punch torsion, 115–119 Rigidity coefficients, 352, 540
[email protected]
Punch test, 68, 234, 238, 251, 341–342, Rigidity matrix, 267, 271, 408, 412, 592
348–349, 413, 420, 598 Rigidity-related characteristics, 400
Punch uplifting zone, 296 Rigid plates, 180
Punch vertical displacement, 260 Rigid punch displacement, 274
Index 645

Rigid punches contact problems, 119–131 S

elastic half-space torsion, 126–131 Schleicher formula, 273, 276
punch indentation, 121–126 Second-order Bessel function, see Macdonald
Rigid rectangular punches, contact problems, function
278–300 Second-order model, see Cam-Clay model
contact interaction at central loading, Section kernel boundary configurations, 331
282–294 Seidel method, 265
contact interaction at off-centre loading, Semihard and hard consistency, 396
295–300 Semi-inverse method, 261–262, 399
Rigid ring-shaped punches, 276 Settlements and slopes, 281, 289, 460
Rigid rod scheme, 475 Settlement-vs-load curve, 408–410
Rigid sphere, displacements of, 223 Shallow foundations, 82, 91, 138, 141, 190,
Rigid spherical inclusion, 135, 219 251–253, 255, 300, 324, 329, 396
Rigid strip variable-width foundation, design of, 252
311–322 Shallow monolithic foundations, 454
contact pressure distribution, 317–322 Shallow slotted foundations, 456
strip foundation base, 315–317 Shandru problem, 12
variable-width strip foundation, 312–315 Shearing force, 252–253, 359
Ring-shape Shear modulus, 5, 14, 55, 265, 270, 404
belts, 224–225 Shebelynka gas field area, 508
centrally loaded punch, 268 Shekhter method, 26
centrosymmetric, 525 Shrine Museum, 506
conductor, 260 Shvedov–Bingham model, 408
foundation, 269, 276, 310, 584–585 Similarity function, approximation of, 241–242
punch, 128, 130–131, 211, 213–215, 255, Simpson rule, 16
257, 260–261, 272, 276, 308, 310, Single-precision calculations, 272, 363
585 Singularities and irregularities, 155
problem, 131 Slippage, 16, 268, 274, 277–278, 402,
rotation, 130 405–406, 410, 429, 484, 486–491
Roben elastostatic problem, 223 delineation of, 490
Roben equation, 239 Slotted foundations, 181, 189, 454, 456–457,
Rock stratification, 508 476
Rolling equipment, 393 calculation of, 456, 460, 474
Romberg integration scheme, 53 construction technology, 456
Rosinka software, 409 contact interaction calculation of industrial
Rostwerk software, 135–137, 139–140, and civil buildings, 454–493
143–146, 150, 242, 389, 419, 449, base deformation calculation, 457–476
457, 538 contact stress on the lateral surface,
block structural scheme of, 139 476–491
deep foundations in, 142 lateral widenings, 491–493
shallow foundations in the, 141 structural shapes, 454–457
Rotating furnaces, 393 geometrical shape, 460
Roughness measure, 486 Soatial contact problems for, 241
Round punch, 130–131, 194, 197, 200, Sobolev δ-shaped averaging functions, 274
202–204, 256, 258–259, 266, 270, Soil back pressure, 386, 397, 399, 416, 476
273, 287, 301, 306–307, 325, 331, Soil compression, 392, 394, 439
342–344, 346–347, 399, 566, 568, Soil conditions, 470, 473
575 Soil consistency index, 597
Round punch on an elastic wedge, 306 Soil deformability, 56
[email protected]
Russia, 424–425, 454, 456, 506–508 Soil deformation modulus, 1, 64, 143, 147,
Russian Federation patent No. 043462, 238, 242, 356, 398, 526
584 Soil deformation parameters, 320
Russian handbooks and regulations, 258 Soil deformation properties, 76, 141
646 Index

Soil drilling technology, 187 Spatial theory of elasticity, fundamental

Soil grain composition, 238 solutions, 1–4
Soil linear displacement phase, 238 Concentrated forces, 1–2
Soil mass deformation, 509–513 Green’s displacement tensor, 2–3
dilatation relations, 512–513 Kelvin’s tensor, 3–4
porous elastic medium, 509–511 Spatial triangular element, 560
Soil mechanics, 64, 82, 252, 254, 267–268, Spatial wedge, 21–22, 330
272, 277, 334, 395–396, 407, 476, Spherical harmonic functions, 400
480, 508, 538, 579 Square punch, 173, 195–196, 208–210, 213,
Soil nonhomogeneity, 82, 352, 391, 404, 416 216, 264, 273, 282–283, 285–286,
Soil properties, characterization of, 234 288–290, 292, 294, 296–300, 304,
Soil resistance, 224, 268, 415–416, 440, 460, 307, 326–327, 332, 533
472, 478, 582–583, 599, 601 orientation scheme, 290
Soil skeleton deformation, bulk modulus of, Squeezing pressure, 603
509 STAMP-C software, 566
Soil tests, 56, 234, 410 Strength-related characteristics, 407, 480, 486,
Soil transverse expansion coefficient, 597 490
Somigliana formula, 93, 273, 278 Strength-related soil properties, 238, 414, 487
Somigliana identity, 93 Stress components, 256, 351
Southern California, 506 Stress distribution, 109, 136, 143, 188, 219,
Spacer-type scheme, 435 231, 252, 262, 273, 312, 315,
Spatial contact interaction processes, 254, 263, 322–323, 402, 441, 457, 477
517, 539 Stress extrapolation, 229
Spatial contact models, 1, 131, 207, 538 STRESS program, 139
Spatial contact problem, 21, 82, 91, 95–97, Stress-strained state, 7–8, 12, 25–26, 31, 36,
113, 118–123, 128, 130, 135–136, 39–40, 55, 58, 60–61, 70–71, 92,
147–148, 163, 173, 175, 192, 97, 109, 135–136, 218–219, 251,
204–205, 207–208, 211–213, 253–254, 256, 264–268, 270–271,
215, 218–219, 228, 234, 242, 276, 311, 334, 346, 351, 357–358,
251, 253–259, 263–264, 271–273, 363, 369, 371–372, 386, 394,
277–278, 292, 300, 311, 313, 324, 403–406, 408–409, 411–412, 415,
327, 337–339, 341, 359, 361, 389, 417, 449, 452, 454, 457, 460, 478,
398–399, 411, 415, 418, 420, 424, 480, 490, 508, 534, 538
426, 443, 476–477, 517, 519, 523, characteristics of, 30, 39, 50, 118, 311, 404,
538–540, 566, 585 410–411, 540
equations of, 91, 99, 110–111 Stress tensors, 6, 550
numerical solution of, 192 components, 30, 94
rigid punches, 175, 277 Stress vector, 108–111, 116, 139, 141, 150,
solutions of, 385 219, 443, 476–480
theory of elasticity, 21, 118, 258, 327, 540 components of the, 518
Spatial deformation, 505, 508 Strip discretization, algorithm of, 152
Spatial discretization, 135, 186, 511, 525 Stripe-shaped punches, 257
Spatial elastic layer, 12 Strip foundation, 251, 260, 263, 312–315,
Spatial elastoplastic problem, 266 318–319, 322, 415
Spatial formulations, 264, 457 Strip punch, 209
Spatial foundations of structural type, 388 Structural nonlinearity, 323
Spatial functioning, 442 Structure-type nonlinearity, 297
Spatial grids of boundary elements, 174–192 Subsidence-01 software, 524, 534
Spatial loading, 121, 162, 196–197, 202, 228, Suffusion-related
[email protected]
278, 345, 385, 389, 392–393, 401, compression, 275
418, 454, 460, 463, 479–480, 487 settlement, 392
Spatial nonhomogeneity, 137, 300, 363, 370 Superposition principle, 7, 277, 296, 298–299,
Spatial problems solving software, 136–146 323, 397, 463
Index 647

Surface approximation, 175 Travelling cranes, 473

Surface discretization, 175, 190, 224, 228, 312, Trench, 238, 464–466, 470, 492–493
445, 519 Triangular boundary
Surface integrals, parametric representation of, elements, 554
553 macroelement, 179–180
Surface parametrization, 554 Triangular discretization, 195
Surface subsidence, 505–508 algorithm of, 152
Swelling, 390, 392, 415, 439–440 Triangular grid, 216
Triangular polar coordinates, 561
T Triangular-prismatic element, 626
Tangential stress, 9, 13, 15, 61, 110, 121, Triangulation of a 2-D domain, algorithm of,
126–128, 130, 137, 144–145, 231, 151
256, 264, 279, 347, 402, 410, 443, Triangulation algorithm, 162
478–479, 486–487, 490, 508 Tridiagonalization, 146
determination of, 272 Trough formation, 505
Taylor polynomial, first-order, 171 T-shaped cross-section, 386
Tectonic anomalies, 506 Two-dimensional integral Fourier
Telescope-shaped cylindrical foundations, 187 transformation theory, 14
Telescopic element method, 419 Two-layer deformable system, 42
Tensile forces, 54 Two-slot foundations, 455
Tensile functioning of soil, 136 T–Z curves, 413
Tensile stress, 141, 202, 216, 419, 445, 451,
505, 566, 572 U
Tetrahedral element, 625 Ukraine, 506, 508
Theater Center, 506 Underground mining, 505
Theoretical substantiation, 301, 352 Underground water level decline, 506
Theory of elasticity, 1–4, 25–26, 52, 55–56, Unilateral constraints, 202–204, 218, 281,
64, 72, 98, 108–109, 118–120, 123, 296, 309, 323, 339, 355, 388–389,
128, 147, 224–225, 258, 272, 277, 402, 418, 428–437, 445, 447–448,
335, 349, 351, 405, 416–418, 480 450–452, 462–464, 469–472, 475,
axisymmetric problem of, 112 566, 572–573
classical, 63 Uplifting zone, 202, 296
edge effects in, 143
equations of, 73 V
inverse symmetrical problem of, 116 Variable-sign moment load, 578, 581, 585
linear, 2, 234, 257, 268, 539 Variable-thickness elastic layer, 17–18, 25,
spatial problems of, 4, 12, 27, 54, 71, 253, 279, 282, 287–289, 291,
106–107, 109, 171, 257, 417, 512 296–299, 310, 331–333, 358, 365
three-dimensional, 138, 175, 261 Variable-thickness elastic layer, 291
Thermoelasticity equation system, 510 Variable-thickness elastic wedge, 23
6th order polynomial, 399 Variable-width foundations (VWFs), 311–315,
9th order polynomial approximation, 240 318–320, 322
Torque load, 136, 259, 403, 607 Variation-difference problem, 270
Torsional rigidities, 404 Varved clays, 396, 486
Total (mean square) error, 195 Vasilyev solution, 12
Tower-type structures, 213, 256, 258, 276, 310, Vertical displacements, 61, 279, 337, 522
577, 585 Vertical resultant force, 279
Transformational functions, 412–413 Vertical rod constraints, 353
Translational displacements, 95 Vibroimmersion, 438
[email protected]
Transverse loading, 462–463 Viscosity-related (fluidity-related) properties,
Transverse-and-longitudinal bending, 508 407
Trapezoidal punches, 307, 328 Volumetric punch, 92, 416, 457
Trapezoid rule, 557 Voronoy polygons, 151, 164–165, 195
648 Index

W Winkler model, 55, 257, 352, 396–397, 403,

Wedge rib, 23, 287, 366, 570–573, 578–579, 416, 424, 457
584–585 Winkler-type elastic bases, 352–353
Wedge-slotted foundation, 183, 190,
455, 458 Y
Yekaterinburg test site, 470
Wedging, 429–431, 433, 437, 451–452,
463
Z
Wet formula, 234
Zero deformation modulus, 261
Winkler base, 336, 352, 357, 414 Zeros of classical orthogonal polynomials, 562
Winkler constraints, 55 Zeroth order model, 412
Winkler hypothesis, 389, 456, 472 Zhemochkin method, 353, 398–399

[email protected]