FINITE ELEMENT

MODELING AND
COMPUTATIONS IN THE
FIELD OF CIVIL
ENGINEERING
This collaborative work elaborated by professionals confronted daily to the use of FE methods targets young engineers,
who recently graduated from university and often studied exclusively the fundamentals of the FE calculations. Its
objective is to answer some recurring questions (such as the mesh size, how to smoothen the peaks, …,) to avoid
some issues, and to specify what the FE cannot do. The task force favored to name the project Recommendations and
Advices rather than Guide because the document aims to share applications and discuss about finite elements and not
create an exhaustive guideline.

Table of contents
The AFGC Editorial
Finite Element (FE) calculations – a paradigm shift
Preface - Words from the Scientific and Technical Council

Foreword
Recommendations and Advice Content – The authors
Leave us a comment for us to improve the website

Introduction
A short and broad introduction of the Recommendations and Advice

PART 1 – THEORETICAL ELEMENTS

Chapter A. Introduction
A1. General formulation for linear elastic calculations
A2. The dimensionality of the model
A3. Choosing the finite elements
A4. Interaction between the structure and its environment
A5. Estimation of the quality of the approximated numerical solution
Chapter B. Structural Dynamics
B1. Analysis based on a modal search
B2. Analysis based on a direct temporal integration
B3. Considering damping
B4. Specificities of the seismic analysis
Chapter C. Static non-linear calculations
C1. Mechanical non-linear problems
C2. Why performing non-linear calculations
C3. Implementation
C4. Convergence issues? Symptoms and solutions
Chapter D. Civil Engineering
D1. Civil engineering materials
D2. Different categories of structural elements
D3. The different construction phases
Chapter E. Typical post-treatment in Civil Engineering
E1. Generalities
E2. Quantities in Dynamics
E3. The specific case of reinforced concrete
Chapter F – Geotechnical calculations
 F1. Geometric aspects
F2. Material non-linearities
F3. Soil-structure interactions
F4. Hydraulic effects
F5. Uncertainties and recommendations
F6. Normative aspects: Principles of the Eurocode 7
F7. Modeling in Dynamics
F8. Characteristic scales

PART 2 – USEFUL ELEMENTS

Chapter A. Understanding the finite elements
A1. What does the software do in a finite element calculation? The example of beam structures
A2. What is a finite element?
Chapter B. Computational objectives and necessary characteristics of the tool
B. Calculation objectives and necessary tool characteristics
B.7 Organization of the calculation
Chapter C. Good practices to create a model
C1. Input data and units
C2. Modeling of the main elements
C3. FE and meshing
C4. Modeling the non-structural elements or equipment
C5. Boundary Conditions
C6. Connections - links – assembly
C7. Offsets
C8. Composite Sections (Beams/Slabs)
C9. Materials
C10. Specific behavior in shear and torsion
C11. Modeling the loading
C12. More about solid elements
C13. More about non-linear calculations
C14. More about prestressed concrete
C15. More about phased calculations
C16. More about dynamic and seismic calculations
Chapter D. Analysis and processing of the results
D1. General information on numerical calculations
D2. Load combinations
D3. Data processing
D4. Normative verifications: the behavior of reinforced concrete elements
D5. Understanding and analyzing the peaks (case study about concrete)
D6. Understanding and analyzing the peaks (case study about steel assembly)
D7. Further information specific to dynamic calculations
Chapter E. How to ensure quality?
E1. Starting with a new software
E2. Model validation using self-checking
E3. Traceability and group work
Chapter F. How to properly present the finite element calculation note?
F. How to properly present the finite element calculation note?

PART 3 – EXAMPLES AND COMPLETE CASE STUDIES

EXAMPLES AND COMPLETE CASE STUDIES
Example A - Modeling a complex high-rise building
Example B - Mixed and Steel Girders
Example C - Beam Grillage Modeling
Example D - Simple example: modeling of a Br wheel
Example E - Transverse bending of a prestressed concrete box girder
Example F – Dynamic calculations of tanks
Example G – Cable-stayed bridges
Beam Grid Example
Contribution to the beam grillage analysis
Transverse bending of a multi-beam bridge
Example of a beam grillage calculation according to different methods - Part 1
Example of a beam grillage calculation according to different methods - Part 2
Example of a beam grillage calculation according to different methods - Part 3
Practical application of smoothing
BIBLIOGRAPHY
Bibliography

FOLLOW-UP OF ADDS-UP AND MODIFICATIONS

Additions and Modifications
$translationBooks

Table of contents
EDITORIAL
Finite Element (FE) calculations – a paradigm shift

Words from the Scientific and Technical Counsel

FOREWORD
Recommendations and Advice Content – The list of authors

Let us know about your comments and feedback to improve the website!

INTRODUCTION

A short and broad introduction – Structural analysis and finite elements

PART 1 – THEORETICAL ELEMENTS

Chapter A. General Observations

Chapter A - General Observations

Chapter B. Structural Dynamics

Chapter B - Structural Dynamics

Chapter C. Static non-linear calculations

Chapter C - Static non-linear calculations

Chapter D. Civil Engineering

Chapter D - Civil Engineering

Chapter E. Typical post-treatment of Civil Engineering

Chapter E - Typical post-treatment of Civil Engineering

Chapter F. Geotechnical calculations

Chapter F - Geotechnical calculations

PART 2 – USEFUL ELEMENTS

Chapter A. Understanding the finite elements

A.1. What does a finite element software do? Example of framed structures.

A.2. Explicitly, what is a finite element?

Annex 1 – File of the matrix calculation example

Chapter B. Computational objectives and necessary characteristics of the tool

Creating a finite element computational model includes several steps. The choice of the tool is critical
and depends on various criteria. A successful model requires good organization and preparation.
B.1 to B6 Criteria to be considered

B.7 Organization of the computational model

Chapter C. Good practices to create a model

The different levels of analysis and the complexity associated with the model have already been
defined in the previous chapter before the modeling stage. This chapter highlights the simplifications
that can be used to create a model that is structurally representative of the real-life structure and the
loads it is subjected to.

C.1 Input data and units

C.2 Modelling of the main elements

C.3 Finite elements and meshing

C.4 Modelling of the non-structural elements or the equipment

C.5 Boundary conditions

C.6 Connections – links – assembly

C.7 Eccentricity

C.8 Combined cross-sections (beam/deck)

C.9 Materials

C.10 Behaviors specific to shear and torsion

C.11 Loads modeling

C.12 Further information related to volumetric elements

C.13 Further information related to non-linear calculations

C.14 Further information related to prestress

C.15 Further information related to phase calculation

C.16 Further information related to structural dynamics and seismic calculations

Chapter D. Analysis and processing of the results

D.1 General information about numerical calculations

D.2 Load combinations

D.3 Results processing

D.4 Validation rules: the behavior of concrete elements

D.5 Understanding and analyzing the peaks (case study about concrete)

D.6 Understanding and analyzing the peaks (case study about steel assembly)

D.7 Further information specific to dynamic calculations

Chapter E. How to ensure quality?

Below are some simple advice to assess the quality of the finite element calculations. The principal
challenges are:

§ The proper use of the software

 § The appropriate modeling of the structural behavior

§ The traceability of the modeling hypotheses and results

The advice below covers the engineer’s or the team’s handling of the software, the verification controls
that must imperatively conduct any engineer at the end of its modeling, and the tracing of the
minimum items so that the work can be completed thereafter.

E.1 Starting with a new software

E.2 Model validation using self-checking

E.3 Traceability and group work

Chapter F. How to properly present the finite element calculation note?

This paragraph introduces the fundamental elements that must be present in a note to provide a clear
description of a FE computational model.

F. How to properly present the finite element calculation note?

PART 3 – EXAMPLES AND COMPLETE CASE STUDIES

This part contains modeling examples for simple and more complex objects. They are presented as a
complete or partial study or even a comparison of different models of the same structure.

If you happen to have an example that you wish to share with us of a complexity or paradox noticed on
a part of a model, please send it to the following address: [email protected]. (It could be a note
about a real project rendered anonymously.)

Example A – Modelling a complex high-rise building

Example A – Modelling a complex high-rise building

Example B – Modelling of composite bridges

Example B – Modelling of composite and steel bridges

Example C – Modelling of beam grids

Example C – Modelling of beam grids

Example D – Simple example: modeling of a Br wheel [CH1]

Example D – Modelling of a Br wheel

Example E – Transverse bending of a prestressed concrete box girder

Example E – Transverse bending of a prestressed concrete box girder

Example F – Dynamic calculations of tanks

Example F – Dynamic calculations of tanks

Example G – Cable-stayed bridges

Example G – Cable-stayed bridges

BIBLIOGRAPHY

Link to the bibliography

FOLLOW-UP OF ADDS-UP AND MODIFICATIONS

List of adds-up and modifications
The AFGC Editorial
Thierry Kretz was the president of the AFGC Scientific and Technical Counsel when the finite element task force was
launched. He shares his vision of this project.
$translationBooks

The AFGC Editorial

Finite Element (FE) calculations –

a paradigm shift
Finite Element (FE) calculations – a paradigm shift
It is no exaggeration to say that FE calculations applied to civil engineering structures are a paradigm shift for
structural analysis. Indeed, it is possible to transpose to structural engineering the concept of paradigm as defined by
Thomas Kuhn in The Structure of Scientific Revolutions. According to him, paradigms are “universally recognized
scientific discoveries, which provide to a community of researchers typical problems and their solutions for some
time.”

Despite its seemingly trivial aspect, the appearance of FE calculations truly is a revolution. The ancient world is one of
the classical Strength of Materials based on the hypotheses of Saint Venant and Navier-Bernoulli. The latter translate
into the beam and thin plate theories. Thus, the ancient world relies on a coherent set of hypotheses and resolution
methods, whose results are then transformed into design principles for structures and their reinforcements.

The new world that uses FE calculations is built on other foundations. The behaviors of the materials themselves are
not questioned, but the structural analysis hypotheses are. They affect, on the one hand, the meshing (the principle
and precision of discretization of the structures,) and on the other hand, the choice of the types of elements (the range
of displacements considered.) The results require new analytical methods to allow the accurate design of the
structures and their reinforcements.

The calculation rules, mainly the Eurocodes, are largely established in the logic of the classical Strength of Materials
theories. They propose simple rules mostly based on experience with well-recognized validity domains. Some
examples are the strut-and-tie method, the dimensioning of consoles, punching control, etc... These regulations allow
the FE computations of the structures but give shallow explanations concerning the methods used (meshing, choice of
elements) and the interpretation of the results. Various technics exist to transcribe results in terms that are compatible
with the regulations. However, the “finite elements calculations doctrine” is certainly still in the development phase.

The AFGC guide that you are reading aims to participate in establishing this new doctrine. It reflects the AFGC's
willingness to accompany the development and innovation in the field of civil engineering by representing a place of
sharing and transmission of knowledge and technological advancement.

I want to thank the redaction group and particularly the two facilitators Didier Guth and Claude Le Quéré for their
incredible work. I am convinced that this guide along with the website will remain for a long time a point of reference
to the engineers working in design offices.

Thierry Kretz – May 2020

$translationBooks

The AFGC Editorial

Preface - Words from the Scientific

and Technical Council
Emmanuel Ferrier is the president of the AFGC Scientific and Technical Council. He
“
shares with us his foreword.

Preface
Numerous students, engineers, scientists, and researchers use numerical methods to develop or use software that will
solve engineering problems related to construction. These numerical methods rely on FE calculations. The AFGC
wished to contribute to the field and has proposed in 2016 the creation of a task force focusing on the FE calculation
methods related to the field of construction. The main objective of this task force is to meet a need that is often
expressed: to have an educational document, like the former SETRA guidelines, discussing the FE computational
modeling (beams, plates, shells,…) of civil engineering structures. This project is aimed towards, among others, young
engineers in TP/GC design offices.

The FE analysis is a fundamental subject that all engineers working in major companies and design offices need to
understand to create essential designing tools.

This document introduces the FE analysis with an overview of the various concepts and applications. It brings forward
the primary concepts of the FE method along with analysis examples using consistent procedures. The FE concepts
involving 1D problems are discussed in detail so that the reader can make sense of the methodology and progressively
understand to better analyze 2D and 3D problems.

Since this field is in a constant state of advancement, the AFGC did not wish to freeze the document as a set of
recommendations like it usually does. Instead, it proposed for the first time a numerical format that is dynamic and
evolving with time. Those are, therefore, unlike many AFGC guidelines, not pre-normative recommendations but rather
documents to understand and practice FE calculations in the field of construction.

The CST of the AFGC would like to thank the facilitators of the task force, Claude Le Quéré and Didier Guth, and
all the other participants, who rendered it possible through their work to publish this document.

Emmanuel Ferrier – May 2020

Foreword
$translationBooks

Foreword

Recommendations and Advice

Content – The authors
Recommendations and Advice Content – The authors
This collaborative work elaborated by professionals confronted daily to the use of FE methods targets young
engineers, who recently graduated from university and often studied exclusively the fundamentals of the FE
calculations. Its objective is to answer some recurring questions (such as the mesh size, how to smoothen the
peaks, …,) to avoid some issues, and to specify what the FE cannot do. The task force favored to name the
project Recommendations and Advices rather than Guide because the document aims to share applications and
discuss about finite elements and not create an exhaustive guideline.

The content of the website is dynamic and evolving. The task force is open to suggestions and correction
propositions. We are especially interested in complementary examples for Part 3. If you have any
recommendations, please contact us on the comments page or by email at [email protected].

The three parts can be studied independently.

The first part discusses the theoretical elements.

The following topics will be considered:

Chapter 1 discusses general information about the method.

Chapter 2 explains the different concepts related to FE calculations in structural dynamics.
Chapter 3 describes non-linear static calculations.
Chapter 4 is about the modeling of civil engineering materials and the questions related to considering
phasing.
Chapter 5 explains why post-treatment is necessary to transcribe the peculiarity of “composite materials”
that constitute reinforced concrete.
Finally, chapter 6 highlights the possible use of the FE method applied to geotechnical problems. The rest of
the document has nothing to do with this specific topic.

The second part provides elements to incorporate into a structural analysis using
the FE method.
Two main reasons push the engineer to use iterative methods with various step verifications and specific treatments:

The first reason is that the FE calculation methods are deduced from the Strength of Materials hypotheses
and are not consistently compatible with the regulatory calculation methods. It is true for reinforced concrete
for which the regulatory deformation diagram is not agreeing with the one represented by the FE.
Additionally, the regulation stipulates calculations that not all the software take into consideration
(contributing widths, offset of the bending moment diagram in the strut-and-tie method, crack consideration,
…) It is also true for steel for which there are discrepancies between the regulatory assembly or instability
calculations and the results found using FE methods, which can be very precise.
The second reason is related to the theoretical calculation, which considers homogeneous elements and does
not directly handle materials such as reinforced concrete. In this case, data processing is necessary to
transcribe the peculiar behavior of concrete and its reinforcement. The focus must be specifically directed
towards steel netting (especially for shells) that can create the illusion that the software is performing the
regulatory calculation whereas it is not always the case.

The second part starts by advising to create a structural analysis using FE. Later, the first chapter presents the
objectives of the calculations to perform. Then, the second chapter describes the modeling of the structure. It focuses
on detailing the various materials, the treatment of the interfaces, (to complete,) for the different types of calculations
from the simplest to the most complex one. The third chapter discusses the treatment of the results. Finally, the two
last chapters highlight in detail the verifications that must be conducted to validate the results and the final formatting
of the computation report. The latter will attest to the work that was performed, the accurate inclusion of the
structure’s use, the objectives related to the construction phases, and of course the project’s compliance with the
regulation.

For each step, the authors worked on demonstrating, using examples, all the precautions one must take when using
simplifications, and the verifications to conduct to ensure the accuracy of the hypotheses. Those will allow to obtain a
rigorous model and results that will be as close as possible to the real behavior of the studied structure.

The authors
The Recommendations and Advice were written between 2016 and 2019 by the Task Force about Finite Elements of
the French Association of Civil Engineers (AFGC.) Thank you to all from the group who participated, editors, and
proofreaders. This paper is posted online so that the users can share with the team all their questions, comments,
observations, and suggestions.

Main facilitator of the guideline: C. Le Quéré (Egis) and D. Guth (Setec diadès)

Facilitators of part 1: G. Hervé-Secourgeon (EDF R&D) and P. Bressolette (UCA)

Main authors from the various chapters:

Chapter 1: P. Bressolette (UCA,) L. Adelaide (IFSTTAR,) L. Jouval (ARTELIA)

Chapter 2: G. Hervé-Secourgeon (EDF R&D)
Chapter 3: S. Michel-Ponnelle (EDF R&D)
Chapter 4: J.-J. Brioist (CEREMA,) H. Somja (INSA of Rennes,) M. Scalliet (CERIB)
Chapter 5: G. Hervé-Secourgeon (EDF R&D)
Chapter 6: Leopoldo Tesser (GDS,) S. Burlon (IFSTTAR)

Facilitator of part 2: S. Juster (CPA Experts,) D. Guth (Setec diadès) and C. Le Quéré (Egis)

Main authors of part 2:

Arnaud Bouard (SCE)

Franck Dubois (VINCI Construction France)
Christian Gallois (Areva)
Didier Guth (Setec diadès)
Landry Jouval (Artelia)
Sandrine Juster-Lermitte (CPA Experts)
Claude Le Quéré (Egis)
Emilie Leroux (Tractebel Engineering)
Pascal Mangin (CTICM)
Pierre Mazurelle (Arcadis)
Jean Michalewicz (ETPO)
Gildas Potin (Tractebel Engineering)
Vincent Rousseau (Bouygues TPRF)

Proofreading of part 2:

Valentina Bruno (Setec tpi)

Olivier Cheray (A.Aegerter & Dr.O.Bosshardt AG - Ingenieure und Planer)
Yavuz Demir (CTE Strasbourg)
Franck Dubois (VINCI Construction France)
Sylvie Ezran (Setec tpi)
Didier Guth (Arcadis)
Sébastien Miossec (VINCI Construction France)

*You? We are looking for volunteers!

Proofreaders of parts 1 and 2:

Fabien Coulon (VINCI Construction Grands Projets)

Thierry Kretz (IFSTTAR)
Michel Marchetti (Formule Informatique)
Stéphane Velin (VINCI Construction Grands Projets)

Facilitator of part 3: Christian Gallois (ex-Areva)

Main authors of part 3:

Valentina Bruno (Setec tpi)

Jacques Combescure (Artès)
Jean-Paul Deveaud (Cerema Centre-Est)
Franck Dubois (VINCI Construction France)
Sylvie Ezran (Setec tpi)
Samy Guezouli (INSA of Rennes)
Didier Guth (Setec diadès)
Pierre Perrin (Dir Est)
Albert Lepeltier (Setec tpi)
Aymeric Perret du Cray (Setec tpi)
Thierry Richard (VINCI Construction France)
Hughes Somja (INSA of Rennes)
Benjamin Tritschler (Arcadis)
$translationBooks

Foreword

Leave us a comment for us to

improve the website
Leave us a comment for us to improve the website
It could be:

a disagreement,
a proposal for a better phrasing,
a proposal for a bibliography reference,
a link to a relevant internet webpage,
a practical example illustrated or not,
a question.

Please leave your comments as shown below:

Date Name Contact information (optional)

Part 1 Chapter A3.3 My comment

Part 2 Chapter Ex My comment

Part 3 Example y My comment

Your turn

Date Name Contact information (optional)

Part 1 Chapter A3.3 My comment

Part 2 Chapter Ex My comment

Part 3 Example y My comment

Date Name Contact information (optional)

Part 1 Chapter A3.3 My comment

Part 2 Chapter Ex My comment

Part 3 Example y My comment

or you can send us an email at [email protected]

Introduction
$translationBooks

Introduction

A short and broad introduction of

the Recommendations and Advice
A short and broad introduction of the Recommendations and
Advice
Civil engineering construction projects require to justify the work regarding the risks identified upon technical
baselines, which are usually based on the current regulation or chosen by the construction manager. Then, the
engineers must conduct design and execution studies with a level of detail that depends on the complexity and
advancement of the project. Moreover, for civil engineering, it is a matter of studying the structural aspects
considering the many possible interactions according to their predominance.

Structural performance
Modeling and finite elements are tools for evaluating the performance of civil engineering structures either in the
design phase or verification of existing structures. These tools can be useful for analyzing new structures, but also for
analyzing strengthening or structural recovery in existing ones. Thus, computational models and finite elements are
first and foremost an answer to a contemporary need and not an end in itself.

“Structural performance” describes the capacity for a structure to comply with the requirements it was designed for,
which are divided into three categories:

structural safety, which ensures the structure’s resistance to the expected loads under normal circumstances
as well as exceptional ones,
serviceability, which ensures that the structure can remain in use,
durability, which describes the capacity of the structure to remain in a state of structural safety and
serviceability under given conditions of maintenance for a predefined period.

The structural performance evaluation of new or existing structures rely on the reliability theory and can be conducted
using different approaches:

deterministic (all the data are supposedly known,)

probabilistic (the uncertainties are intrinsic and represented by random variables or functions,)
semi-probabilistic (relying on the notions of limit state and partial safety factors.)

Structural analysis
In the field of civil engineering, the evaluation of the structural performance is handled using a semi-probabilistic
approach. It relies on the limit state method as proposed in the Eurocodes, which can help solve most of the
construction cases. One can then talk about structural analysis, which is a fundamental concept of the Eurocodes.

Structural analysis is conducted by creating models, which are the results of an engineer’s reflection and assumptions,
taking into consideration the available data (by default, stating the most probable hypotheses and verifying them
afterward): it is all about “well-stating the problem” to solve and defining the system to study.

Structural analysis is a process in which the engineer establishes a simplified representation of the
structure that replicates as effectively as possible its mechanical behavior.

Modeling
According to the Eurocode, structural modeling must imply calculations that are conducted using appropriate
computational models including the related variables. It is expected that structural models can predict the structural
behavior within an acceptable range of precision. Moreover, the models must be adequate to the considered limit-
states throughout all the steps of the project. The structural models used must be based on a theory and an
established practice and must be verified experimentally if needed.

Modeling is then an approach or even a procedure, that consists of establishing a model. According to its objectives
and the means used, modeling can be handled using distinct methods. It is all about representing an object, a real
phenomenon, or a phenomenon extracted from its environment and simplifying it using an isolated system that follows
a concept or a theory. It is developed starting from the acting physical phenomena, which means that it incorporates
physical laws. Thus, the constitutive laws are formulated under mathematical equations compatible with each material
and each structural element of the overall structure. This leads to solving complex continuum mechanics systems of
equations written as partial differential equations.

The main peculiarity of civil engineering calculations compared to other engineering fields is the need to consider all
the phases of construction such as the clearance and filling works, the concrete work of the foundations, attaching
braces, tensioning, or anchoring cables. Modeling transcribes these events in the static language in various ways: by
the continuation of computation with internal constraints (“initial constraints”,) by changing the stiffness of some
elements (go from a zero value to a positive one,) by getting rid of or modifying the nature of some supports, by
changing the points of application of loads, etc.

“Modern structures” in civil engineering (built after the second industrial revolution) present favorable features to their
structural analysis: homogeneous geometry, repeated patterns, qualitative construction materials (conformed to the
“product standards”.) These structures are however affected by multiple factors that are hardly or cannot be
controlled: environmental factors such as the type of soil, the quality of the construction itself, as the structure might
be built-in zones of limited access or under unfavorable meteorological conditions (e.g. the sensitivity of wooden
elements to humidity, which can impact the dimensions and the mechanical features of the element under normal
conditions by creating gaps or shrinkage, or steel elements that expand when subjected to high temperatures…,) and
the heterogeneous aging of their various elements.

To face the issues related to designing and the equilibrium verification of real structures, the structural analysis relies
on creating models, which means creating simplifications of the real structure: design of materials, ideal kinematic
assumptions (perfect connections, bilateral, independent from the reaction forces,) and even arbitrary choices about
what factors are involved in the equilibrium of the structure (by not considering some elements or by neglecting the
involvement of some components on the structural behavior.) These simplifications imply that there is only a
contingent relationship with the real structures in their environment. Consequently, the output results from the
modeling must always be analyzed critically while keeping in mind the explicit assumptions and the less explicit ones
that were made: firstly, isolating the system from its environment, then selecting a mathematical model and a solving
method…

The exact solutions are extremely rare (only a few textbook cases,) it becomes a necessity to solve real problems
using approximated methods:

either analytically: for instance, using Strength of Materials formulations for which geometrical considerations
lead to a simplification of the problem from 3D to 2D (average surface of plates and shells) or even 1D
(average fiber of beams) …
or numerically: FE methods, finite difference methods...

To validate the obtained results, one can compare them to experimental values. It is important to note that different
sources of error exist in this comparison such as:

modeling errors: coming from the modification of the real problem into a simplified one,
discretization errors: due to the numerical resolution of the problem,
measurement errors: related to the equipment and the measuring process.

The obtained results should, ideally, be compared to the experimental measurements that can also be filled with
errors… Moreover, it is possible to combine two families of methods: FE for beams, finite differences for plates…

The simplification should allow to solve the problem to evaluate numerically the effects (efforts,
stresses, displacements) of different load cases on the structure. However, it should fit as well as
possible the behavior of the real structure.

From the problem to its approximated solution

Finite Element (FE) Method

The FE method is a modeling tool that allows us to solve numerically problems that are set using partial differential
equations. The FE method can be applied to civil engineering design since modern construction materials such as steel
or concrete allow to work in the continuous framework, which is the underlying assumption of the FE formulations.
Thus, it is implied that before using this method, the engineer had a preliminary reflection about it and is aware of the
modeling procedure.

The FE method is a general numerical method allowing to handle real-life (“industrial”) problems from various fields
(solid mechanics, fluid mechanics, thermodynamics…) in continuous frameworks, static or dynamic, linear or non-
linear, as well as coupled problems: thermo-mechanical behaviors (structures subjected to fire,) fluid-structure
(vibrations of a tank containing liquids,) chemical-mechanical (considering corrosion of rebars in reinforced concrete)…
The surge for the FE method is closely related to the improvement of computer science since the 1980s and led to the
creation of numerous software. This guideline covers only the use of the FE method, however, this is not the only
structural analysis method: besides some traditional approaches such as the graphical statics, the use of graphs and
tables or analog techniques on models (recognized by the structural Eurocodes,) there exist much more tools (spectral
methods, Boundary Element Methods (BEM), finite difference, etc..) Some methods are still being researched, but the
FE approach is the most general and the most used in the field.

Comprehensively, the principle behind the FE method can be summarized as follows: the domain of analysis, originally
complex, is divided (discretized) into a wide range of subdomains with simple shapes (finite elements) from which
points are identified (nodes): the meshing is now defined. The approximated solution is calculated only at the nodes
(discrete solutions.) The solution at any other point can be recovered by interpolating the values at the nodes.

More specifically, the different steps of the FE method are:

1. Discretizing the structure in several nodes linked by elements that have well-known behaviors. Establishing
the column displacement vector D containing the N degrees of freedom of the structure. The degrees of
freedom are the possible displacements (translations and rotations according to the model) of the structure.
There are the unknowns of the problem that belongs to the RN space.
2. Establishing the behavior of the elements. It can be divided into two parts:
1. Determining the displacement forces fed: the forces applied by the nodes linked to the element on
this element to inflict its displacements de. The relation is established in a local coordinate system
specific to each element, then transposed to the global coordinate system: Fed=Ke.De. One can now
go back to the relation given in 1 p 8, but for an educational purpose, it might be preferable to
handle separately the issue of the combined behavior of the elements (which is addressed by
picking the solution in some kind of behavioral library.)
2. Determining the constraint forces feb: the forces applied by the nodes on an element directly loaded,
to prevent any displacement. The nodes, in this case, play a virtual support role for the element, and
one can calculate the reaction forces in a local coordinate system of the element feb, then in the
global coordinate system Feb.

3. Assembling. Establishing the structure’s stiffness matrix and the forces applied by the nodes onto the
elements according to their loading. This process is conducted by summation, in the RN space for vectors and
RN x RN space for matrices, of the elementary stiffness matrices and the constraint force matrices: K=∑Ke and
FB=∑FBe.
4. Writing and solving the equation of displacements, which is the equilibrium equation of the nodes. The nodes
are applying forces onto the elements, which leads to a displacement D (KD) and constraint forces, namely
reaction forces, as a consequence of the loads directly applied to the elements (FB.) The nodes can be
subjected to external forces that would be directly applied to the nodes (FN,) such as reaction forces at the
support for instance. The general equation writes KD=F (with F=FN-FB.)
5. Solving: displacement calculations: D=K-1F. In practice, one must distinguish the degrees of freedom with
known values (the supports) from the others. Let D1 denote the unknown degrees of freedom and D2 the
known degrees of freedom. One then organizes the working space in a way that the unknowns are first.

The general equation then reads:

In this equation, F1 and D2 are known. One must then calculate

then F2.
6. Knowing the displacements, it is possible to determine the efforts in the elements by using the fundamental
behavior equations of each element.

Geometry and meshing (triangular finite elements and nodes.)

This guideline describes the available tools often used in the computational FE software in statics and dynamics. To
save space and be as clear as possible, we have chosen not to introduce any problem related to interstitial flow
despite their great interest in geotechnical work. Regarding the difficulty of modeling, the geotechnical field (also
covered by the FE method) raises challenges, on the one hand, analogous to the issues met in dynamics of structures
(choosing the time discretization,) and on the other hand, elastoplastic issues (heterogeneous permeability, boundary
between saturated and non-saturated zones, etc.).

The FE method is a tool for numerical calculations. The modelling prior to solving can rely on all the
principles of the preexisting structural analysis.
PART 1 – THEORETICAL ELEMENTS
$translationBooks

PART 1 – THEORETICAL ELEMENTS

Chapter A. Introduction
Chapter A. Introduction
When performing a FE calculation of a structure, the modeling assumptions as well as the objectives must be defined.
Those goals generally consist of solving:

the displacements caused by a static loading (permanent) quasi-static (without inertial effects) or dynamic
(fast-loading,)
internal state, such as stresses, damages inflicted (cracking, etc.)

Among the modeling hypotheses, it is preferable to handle chronologically:

the choice of the general formulation,

the dimensionality of the model,
the choice of the elements,
the definition of the interactions with the environment.

A1. A general formulation for linear elastic calculations

A1. A general formulation for linear elastic calculations

A2. The dimensionality of the model

A2. The dimensionality of the model

A3. The choice of the FE

A3. The choice of the FE

A4. Interaction between the structure and its environment

A4. Interaction between the structure and its environment

A5. Estimation of the quality of the approximated numerical solution

A5. Estimation of the quality of the approximated numerical solution

$translationBooks

PART 1 – THEORETICAL ELEMENTS

A1. General formulation for linear

elastic calculations
A1. A general formulation for linear elastic calculations
The two main formulations for linear calculations are displacements-based and stress-based. Most of the software
nowadays rely on a displacement formulation because it can easily consider the boundary conditions, unlike a stress-
based formulation. Mixed formulations, using displacements and stresses also exist. There are other formulations, but
they will not be discussed in this chapter.

In this document, only the displacement formulation (the most commonly used) is introduced: in this context, the basic
unknowns are the displacements at the nodes, typically represented by q, from which all the other values are deduced
(strains ε, stresses σ ....)

The displacement at any point of an element, represented by u(x,y,z), are obtained by interpolation between the
values at the nodes using interpolation functions (known as shape functions) represented by N(x,y,z), which are low
order polynomials (generally of degree 1 to 3):

The strain-displacement relation is obtained by computing the derivative using the differential operator (matrix) D
(different according to the situation,) from which the matrix B appears, connecting strains and displacements to the
nodes:

To obtain the stresses, it is essential to introduce the constitutive law of the material. Thus, in linear elasticity, the
strain-stress relation is written (H representing Hooke’s matrix):

One can observe that all the interesting quantities (displacement at any point, strains, stresses) can be obtained from
the displacements at the nodes of the mesh.

Note: as the deformations (and consequently the strains) are obtained by computing the derivative of the
displacements at the nodes, there is a loss in accuracy when calculating those values.

For a static calculation in mechanics of structures, identifying the numerical solution of the problem by the Finite
Elements Method (FEM) can be summed up in 4 main steps:

Step 1. Determining the elementary matrices and vectors of each element of volume V by applying the
following equations (with BT representing the transposed of B):

elementary stiffness matrix

elementary vector of equivalent loads

with fV the volumetric load vector and fS the surface load vector. The vector of equivalent loads allows us to “bring
back to the nodes” the loads applied to the elements (in the volume or the surface,) according to the chosen
interpolation.

Be careful! Usually, it does not correspond to an even distribution at the nodes of the total load applied on an
element.

Step 2: Determining the stiffness matrix K and the force vector F with respect to the global coordinate
system by assembling the elementary quantities and introducing the boundary conditions.

Step 3: Determining the vector of nodal displacements q by solving the linear system (large scale):
The vector q gives the node displacements of the mesh, likewise, the vector F contains the values of the equivalent
loads.

Step 4: Resolution of the quantities of interest such as strains and stresses by post-processing the vector
of nodal displacements q.

Going back to the elements, the displacement, the strain, and the stress values at any point can be deducted from the
displacements at the nodes by using the previous equations. As opposed to the displacements, the stresses and the
strains are calculated element by element (and these quantities are not always continuous from one element to
another.)

Note: it is possible to make an analogy between solving, using finite elements, a mechanical problem, and a
thermal problem in a linear permanent regime: in this case, the temperatures T at the nodes are obtained by
solving a linear system: Λ. T = Φ, with Λ representing the conductivity matrix and Φ the flux vector at the
nodes.

The order of the shape functions N(x,y,z), which is deducted for the solid elements (or continuum mechanics – CM)
from the number of nodes in the edge of a finite element (2 nodes: linear element, 3 nodes: quadratic element) has an
impact on the degree of strains and stresses evaluated using finite elements. For example, if the differential operator
D corresponds to the first-order derivative (which is the case in CM,) the strains and stresses will be constant along
with the element for linear elements, and linear along with the element for quadratic elements… As a reminder: in
Strength of Materials (SM) to compute the strains we make use of the second-order derivatives (bending of beams,
plates….)

Shape functions – Triangular element with 3 nodes (linear functions)

Shape functions – Triangular element with 6 nodes (quadratic functions)

Be careful! It is not possible to mix in the same mesh finite element with shape functions of a different order (for
example linear and quadratic elements.) Actually, the continuity of strains cannot be ensured at the interface
between elements with shape functions of a different order.

In the software, most finite elements are defined with the same parameters (isoparametric,) meaning that the
different finite elements of a mesh are built as a geometric transformation of a reference element (which is itself
expressed in unitary coordinate system ξ,η,χ.)
Isoparametric element (quadratic): different transformations (2D)

These geometric transformations, which generally rely on the same nodes used for the interpolation of the
displacements (hence “isoparametric”, with the same parameters,) are characterized by their Jacobian matrix J (matrix
of the geometric transformation’s partial derivatives of the first order.)

Isoparametric element (quadratic): local and global coordinate systems (2D)

One of the advantages of this transformation is the possibility of approximating the boundaries of a given domain by
using polynomial geometries. However, the main problem is that the calculation of elementary matrices and vectors
can become more complex. Their evaluation is made by numerical integration using a quadrature scheme: weighted
summation of the values at a certain point located inside the elements (weights ω_i, points, ξ_i) particular to the
method (Gauss method, Hammer method….) In this case, the calculation of the elementary stiffness matrix and the
elementary vector of equivalent loads are written:

When the matrices (by applying the B matrix) are evaluated this way, it is obvious that one must calculate the strains
and stresses at the same points (also as a function of B.) The software provides these quantities at the integration
points (Gauss or Hammer points, according to the integration scheme.) Moreover, talking about the values at the
nodes of a mesh for strains and stresses does not make sense because there are as many values in a node as the
number of elements connected to it. Be careful, there is not necessarily a continuity between the elements! Wanting at
all costs a nodal value (of strain or stress) of the mesh implies some data processing: extrapolating from the values at
the integration points and the average values of different elements sharing the node, which will, at the very least,
smoothen the results (nonetheless, talking about values at the nodes in a particular element is justified.) When the
quantities are obtained by analytical integration (in the case of beam elements for instance,) they can be expressed at
the nodes, at the center of gravity… the previous observations about the continuity between elements being still valid.
Quadratic elements and integration points (Gauss points)

The way to consider the displacements at the boundary conditions (restraint, support…) might lead to different results
depending on the adopted numerical technique. There are two ways of considering the displacements at the
boundaries:

Penalization method: it is simple to implement but it has the limitation of being sensitive to the order of
magnitude of the stiffness matrix terms.
Lagrange multipliers method: it is not affected by the above-mentioned accuracy problems, but slightly
increases the size of the system to be solved.

Note: in dynamics, the displacements at any point x,y,z at any moment t is obtained, as in statics, by
interpolating the values at the nodes: ux,y,z=Nx,y,z.q(t). These nodes verify the equilibrium condition, given at
any time t by:

The mass matrix can be obtained by assembling the elementary mass matrices MV (volumetric mass) and C the
damping matrix (viscous). To compute the aforementioned matrix, a system of differential equations needs to
be solved using adequate numerical methods (see the chapter on dynamics.)
$translationBooks

PART 1 – THEORETICAL ELEMENTS

A2. The dimensionality of the

model
A2. The dimensionality of the model
It is very important to simplify the full-size problem to model the interaction between the structure and its
environment. To do so, there are 2 known techniques:

the first one consists of transforming the full-size problem into a lower-dimensional space by modeling it as
an axisymmetric structure or a 2D structure (plane strain or plane stress assumptions). This technique
reduces the used space (see section 2.2).
the second one consists of working in a 3D space while still reducing the model thanks to kinematic
hypotheses. For instance, it is possible to adopt simplifications such as beam, plate, or shell theories (see
section 2.1).

Those 2 techniques help reduce the computational cost. However, the user has to be very careful when adopting those
simplifications. For instance, the second technique uses hypotheses that are valid only in a limited domain. It is
important to make sure that the adopted simplification stays within the domain of the full-size problem to obtain
relevant results.

1) Case of finite elements applied to Strength of Materials

From a finite elements method perspective, the main difference between Continuum Mechanics and Strength of
Materials relies on the geometry, which is simplified through additional hypotheses: the initial 3D problem is then
transformed into a 2D problem (mean surface for plates and shells) or 1D (mean fiber for beams), but still represented
in a 3D space (except for 2D problems, as mentioned above). When the FE method is used to solve a Strength of
Materials problem, the finite elements have specific characteristics. It is necessary to assign to those elements
geometrical properties (the section, the inertia for beams, the thickness for plates and shells). Moreover, they combine
the effects of tension/compression (for beams) or membrane effects (for plates and shells) to the bending effects. The
traction/compression and membrane effects are treated as they are treated in Continuum Mechanics, while the
bending effects are handled separately.

For the computation of bending effects, simplifications in the geometry lead to a particular definition of deformations,
which results in different expressions of the differential operator D according to the situation. Another consequence is
due to the definitions of the unknowns at the nodes: while in Continuum Mechanics the unknowns are the
displacement components, in Strength of Materials the unknowns corresponding to the rotations are added. This
happens because it is no longer possible to evaluate directly the rotations as a consequence of the simplifications of
the geometry. The initial choice of treating the problem in the frame of Continuum Mechanics or Strength of Materials
implies choosing between different families of finite elements. It is, in principle, not possible to mix finite elements of
different families because at the interfaces between elements of different nature the rotations are not transmitted
unless they are adapted to take it into account. Moreover, the calculated stresses inside the element are usually
generalized stresses (or Strength of Materials efforts: axial, shear, torsion, bending). To obtain the stresses (in
Continuum Mechanics) at a given point, it is necessary to provide additional information (position inside the beam’s
cross-section, for example).

In this scenario, choosing a finite element brings on an additional difficulty related to the consideration or not of the
shear forces (Euler-Bernoulli or Timoshenko beam theories, Love-Kirchoff, or Reissner-Mindlin plate theories). This
choice is related to geometric considerations (slenderness of the cross-section or thickness of the plate). Moreover,
taking into account the shear forces might lead to numerical errors (shear locking), which renders the use of some
finite elements complicated.

The Euler-Bernoulli beam finite element allows to represent exactly the bending moments that vary linearly
along with the mean fiber of an element (shape functions of order 3 for the bending displacements and the
bending moments obtained by taking the second derivative of the displacements): it is therefore not useful to
introduce several elements between 2 nodes subjected to point loads (a dense mesh should be used when the
loading is distributed between these 2 points).

Finally, for plate and shell elements, the monotonic convergence is not guaranteed depending on the mesh geometry.
This behavior is connected to the formulation of the element itself. This kind of behavior is illustrated in the figure
below, showing the evolution of the relative error of the deflections ω and Mx as a function of the number of degrees
of freedom (in the case of a rectangular plate supported by its four sides submitted to uniform loading): in green, a
non-compliant element (COQ3) for which the results are much less precise than for a compliant element (DKT). Also,
for a non-compliant element (COQ3) the convergence is not monotonic (for the bending moments Mx in the figure):
more elements leads, paradoxically, to results that can be less precise!
Bending plate: convergence (deflection ω and bending moment Mx) as a function of the number of degrees of freedom

This domain is particularly important to civil engineering, but it presents specific difficulties that will be explored in the
next chapters.

2) Two-dimensional calculations
The studied problems are three-dimensional, however, it is faster to calculate two-dimensional problems. In some
cases, it is possible to transform a three-dimensional problem into a two-dimensional problem:

if the problem admits a revolution axis (for the geometry, the loading, and the boundary conditions): it is
possible to perform an axisymmetric calculation with no additional hypotheses needed. In the case where the
loading is not axisymmetric, it is possible to decompose it in the Fourier series and treat the initial problem as
a superposition of axisymmetric calculations.

Cylinder under axisymmetric pressure

if one adopts the hypothesis of neglecting the stresses or the strains that are out of the plane, for cases in
which the structure is either very slender or very thick, plane stress or plane strain assumptions can be made.
The obtained solution is then an approximation of the three-dimensional problem. It is important to keep in
mind that in-plane stress, the strains outside of the plane are equal to zero (similarly, in-plane strains, the
stresses outside of the plane are equal to zero).

Dam: important thickness of plane strain assumption

Assembly: low thickness/plane stress assumption

3) Taking into account the symmetries

Some problems present symmetries (axis of symmetry, plane of symmetry…) and it can be interesting to use them to
make the finite element calculations faster. Although, it is important to notice that solving symmetric problems will
provide only symmetric results (especially when computing eigenmodes).

For that matter, it is important to keep in mind that to use those symmetries in the calculations, not only the geometry
has to be symmetrical, but also the loading and the boundary conditions as mentioned above. The resolution of this
model will not represent the solution of the whole structure unless the symmetry conditions are added to the model. In
Continuum Mechanics, one should impose the displacement components perpendicular to the axis/plane of symmetry
equal to zero. In Strength of Materials, it is convenient to add no-rotation conditions around the axis/plane of
symmetry. Finally, taking into account symmetries might impact the loading: for instance, a punctual load should be
applied with half the intensity if applied to an axis of symmetry.
$translationBooks

PART 1 – THEORETICAL ELEMENTS

A3. Choosing the finite elements

A3. Choosing the finite elements
Choosing the element is an important step. The goal is to select the type of element (its shape, the order of the
shape functions associated), and the size of the element. The type and size of the elements define the form and the
precision of the displacement fields, and consequently the stresses and strains.

Moreover, besides the shape, the element has an aspect. It should be avoided to create decayed or altered
elements (flattened or elongated) because it degrades the precision in the resolution of the problem.

Generally, the generated mesh can be structured (a regular division of the elements) or unstructured. It is possible to
mix structured and unstructured parts in the same domain depending on the complexity of the geometry.

In the same context, the size of the element chosen depends on the geometry of the structure to model and the
loadings it is subjected to. The zones with high-stress variations (gradients) or high stresses (friction or cracking, for
example) determine the parts where the mesh needs to be more refined in comparison with the other parts to properly
observe the stresses and the strains.

It is important to perform the first simulation with an initial meshing to determine the sensitive zones and then refine
the mesh where needed.

The person in charge of modeling must analyze the meshing critically considering the geometry of the structure and
the important zones to observe.

To choose correctly the finite elements construct the mesh, it is imperative to think about the type of calculation
desired:

the whole geometry is going to be represented. It is a Continuum Mechanics problem, so the elements are of
the “solid” type and the problem must be predefined as three-dimensional or two-dimensional (plane stress,
plane strain assumptions, axisymmetry… see chapter 3). If one wishes to use linear or quadratic shape
functions, the precision of the calculations improves with a mesh of quadratic elements, but with an
important addition in the computational cost. Thus, a compromise has to be made. In any case, the degrees
of freedom are the displacement components (u, v in 2D, u, v, w 3D). The main elements are listed below,

the geometry is simplified, in which case one has a Strength of Materials problem (or structural analysis, see
Chapter 2) and the finite elements are:
bar/beam elements (a bar element can only transmit tension or compression, whilst the beam
element is also capable of transmitting bending moments. Be careful, some software use the term
“bar element” to designate any 1D element),
plate/shell elements (the difference between plate and shell is related to the curvature of the mean
surface and most software do not make any distinction between the two).

Besides the displacement degrees of freedom, the Strength of Material elements have also rotational degrees of
freedom (θx, θy, θz), allowing to take into account the non-meshed geometry (cross-section for beam elements,
thickness for plates and shells). Moreover, one should ask the question if the shear forces should be taken into account
or not (Navier-Bernoulli or Timoshenko beam elements, Love-Kirchoff or Mindlin-Reissner shell elements). In the case
of shell elements, as mentioned before, the question about the quality of elements comes to mind (compliant or non-
compliant elements). It is particularly difficult to choose between plate and shell finite elements, especially because
the literature is scarce in this area. It can be helpful to perform a calculation with known results to test the quality of
the available elements.

The different elements that are commonly found in Strength of Materials are described in the table below.
$translationBooks

PART 1 – THEORETICAL ELEMENTS

A4. Interaction between the

structure and its environment
A4. Interaction between the structure and its environment
Taking into account the interactions between the structure and its environment: this consideration allows us to
reduce the number of discrepancies between the model and the real-life structure. The effect of the environment on
the structure is determined from the nodal values such as the nodal displacements and nodal efforts. The first ones are
connected to the boundary conditions and the second ones to the external loading [1].

the displacement boundary conditions (nodal) make it possible to impose displacement values to a node
(zero or non-zero). The imposed displacements are often named kinematic constraints. They enable to
connect the displacements of certain nodes. Initially, it is preferable to determine if the symmetries should
be taken into account before imposing the boundary conditions. When symmetries are adopted the nodal
displacements perpendicular to the plane of symmetry in 3D or the axis of symmetry in 2D should be blocked
(figure 3) (see chapter 4).

Boundary conditions for a plate with a hole in the center, 2 axes of symmetry, submitted to tension

Afterward, it is essential to eliminate the rigid body movements. A correct finite element model has to restrain free
rotations. In 3D, the 6 rigid body movements previously introduced should be avoided (in 2D, there are 3 movements,
figure 4).

Possible rigid body movements in 2D: a) possible horizontal translation; b) possible vertical translation; c) rotation
around a hinge; d) all rigid body displacements restrained

Once these 2 steps are accomplished, it should be verified that every rigid body movement is correctly restrained and
that no rigid body movement was blocked when this mode has already been removed. In the first case, the problem
has no solution and in the second case, unexpected strains may appear.

the loading corresponds to exterior efforts exerted on certain parts of the mesh. Among the loadings, forces
such as gravity and inertia should be taken into account. They are in general modeled as volumetric forces
represented by nodal forces applied to all the nodes in the domain.

There are also contact forces such as pressure or any other force that needs contact with the structure. They can be
surface, linear, or punctual forces. Their application should also be transformed into nodal forces. Special attention
should be paid to the transformation of the contact forces when creating a model respecting its validity domain. Using
a point load can generate singularities such as stress concentration in the neighborhood of the node where the load
was applied. Thus, to avoid this kind of singularity, it is necessary to apply the point load as a surface load in the
neighborhood of the node. This technique consists of applying a surface load containing a relatively large area around
the application point. Then comes the question of mesh refinement in this area and its influence on the obtained
results.
Connection conditions
There are different types of finite elements, such as volumetric, plates, shells, beams, and bars. Thin plates and shells
are elements, which have a thickness smaller than their other 2 dimensions. A plate works only perpendicularly to its
plan (3 degrees of freedom (DOF) for a node: 1 translation and 2 rotations), whereas a shell element works along its
plan and perpendicularly to it (6 DOF: 3 translations, 3 rotations). The solid elements also have 6 DOF per node.
Chapter 2 presents some practical examples.
$translationBooks

PART 1 – THEORETICAL ELEMENTS

A5. Estimation of the quality of the

approximated numerical solution
A5. Estimation of the quality of the approximated numerical
solution
The difference between the exact solution of the problem and the approximate one obtained through the finite
element method allows us to assess the quality of the solution: this is called “discretization error”. Since in the general
case the exact solution is not known, the idea is to estimate this difference by computing an “error estimator”. The
error estimators can be categorized as:

global error estimators: to evaluate the quality of the solution on the whole domain (stress smoothing
method, equilibrium residuals, constitutive law errors…),
local error estimators: to evaluate the quality of a particular quantity, such as displacement on a certain point
or stresses in a particular zone (weighted residuals method…).

The “error estimator” tools (available in some software, with different computational costs in each one) can be used
for two different purposes:

to improve the quality of the results of a finite element calculation by refining automatically the mesh and or
the time discretization,
to obtain an interval of confidence (inferior and superior boundaries) associated with a global error or a
particular quantity of interest.

It is important to highlight that certain error estimators, such as the stress smoothing method, only estimate the error,
while others, such as the weighted residual method, guaranty the numerical results by computing the range of error.
$translationBooks

PART 1 – THEORETICAL ELEMENTS

Chapter B. Structural Dynamics

Chapter B. Dynamics
For many applications such as seismic calculations, collisions, vibrations… it is necessary to consider the dynamic
phenomena.

The dynamic charges applied to a civil engineering structure belong to two different categories:

phenomena that can be compared to static events: constant wind flow, swell, rotary machines.
transitory phenomena: collision, explosion, earthquakes.

Concerning seismic motions, if they are theoretically considered as transitory, it is admissible to assimilate them as
stationary phenomena while in their strong phase. For the cases in which one tries to model the structure with
geometric or material non-linearities, stationary assumptions cannot be considered anymore.

The means of representing the loading categories can then be distinguished as shown below:

Stationary:
- Complex Fourier Transform (FT),
- Power Spectral Density (PSD),
- Oscillator response spectrum (ORS).
Transitory:
- Load-displacement curve, speed or acceleration expressed as a function of time,
- Efforts or pressures are expressed as a function of time.

Two big families of analysis can be considered:

The modal analysis, which enables to identify the natural frequencies and the associated modes of a
structure. This data is useful to characterize:
- The stationary loading response applied using a method of spectral response,
- A temporal response using the Duhamel integration of each loading curve corresponding to the modal
responses.
- A transfer function convolved to the signal expressed in terms of the frequencies to deliver an FT or PSD
response.
- The time-dependent dynamics that enables to compute the structure’s transitory dynamic response to any
temporal vibration. The resolution can be conducted using schemes of time integration, which can be explicit
or implicit.

The explicit schemes dictate the choice of very small-time steps. Thus, they are the most used to solve problems with
small periods (like collision/impact problems.) On the contrary, implicit schemes allow us to use greater time steps and
are therefore favorable to study problems occurring on wider time ranges.

Examples of applications

Applications Loading representation Quantities available

FT FT
Vibrations analysis
PSD PSD
Modal

Tracking of natural Spectrum extrema of quantities

ORS
frequencies of various interests

Seismic Study
Implicit Accelerations, velocities, forces, pressures, or displacements Quantities of diverse interests
transitory as a function of time expressed throughout time
Weakening

Fall of on object Modeling of projectiles in contact, collisions

Explicit Quantities of diverse interests

transitory Accelerations, velocities, forces, pressures, or displacements expressed throughout time
Plane crash
as a function of time

Once the dynamic problem is discretized in finite elements, the resolution of the equilibrium equation can be written as
shown below (cf. chapter 1):

M: the mass matrix expressed at the nodes,

C: the damping matrix expressed at the nodes,
K: the stiffness matrix expressed at the nodes,
 q: the vector containing the nodal displacements,
q': the vector containing the nodal velocities,
q": the vector containing the nodal accelerations.

In a modal analysis, the natural frequencies ω_i and the associated modes φ_i are used.

In the time-dependent analysis, the displacements q(t), the velocities q'(t), and the accelerations q"(t) at the nodes are
calculated for each time-step t by direct integration of the equilibrium equations.

The second approach has an advantage as it allows us to handle nonstationary solicitations.

B.1 Analysis based on modal search

B.1 Analysis based on modal search

B.2 Analysis based on a direct temporal integration

B.2 Analysis based on a direct temporal integration

B.3 Considering the damping

B.3 Considering the damping

B.4 Particularities of a seismic analysis

B.4 Particularities of a seismic analysis

$translationBooks

PART 1 – THEORETICAL ELEMENTS

B1. Analysis based on a modal

search
B1. Analysis based on a modal search
A reminder of the concept of simple oscillator – Concept of the elastic response spectrum

In the following section, a lot of applied methods return the response of damped oscillators with a degree of freedom
often named simple oscillator.

Its general behavior is briefly reminded here.

Its behavior is explained starting from the dynamic equation applied to an oscillator with a degree of freedom
submitted to a time-dependent harmonic loading and building its transfer function with the following parameters: the
ratio between its natural frequency and the natural frequency of the harmonic loading, and its critical damping ratio.

The dynamic equilibrium equation applied to the mass-spring-damper with only one degree of freedom is reminded
below:

It can also be written in its canonical form by dividing all the terms by the mass m such that:

If the harmonic loading is applied xs=X.ejΩt, one can find the transfer function as an absolute response by establishing
the ratio between the input loading and the absolute output response below:

The norm of this question is the amplitude of the transfer function that allows highlighting the dynamic amplification
phenomenon illustrated for different values of the critical damping in figure 1. This transfer function’s independent
variable is the phase.
Transfer function

A response spectrum is different from a transfer function. The curve giving the acceleration (named spectral
acceleration) as a function of the period (or frequency) is called the spectrum of elastic response. The latter
corresponds to the maximum absolute acceleration seen by a simple oscillator throughout time as a function of its
natural period (or natural frequency) and its critical damping ratio. It dimensions the seismic motion. It is possible to
construct an approximate relation between the Fourier spectrum of acceleration and the spectrum of spectral
velocities for a damping ratio equal to zero.

Starting from the dynamic equation of the simple oscillator subjected to an arbitrary acceleration:

The maximum responses are deduced from its resolution according to time using the method of your choice (Duhamel
integration, direct calculation…) to then calculate the relative spectral values of displacement, relative velocity, and
absolute acceleration:

The concept of pseudo-displacement, pseudo-velocity, or pseudo-acceleration is often used, which means

approximating these quantities, starting from one of them, based on the fact that the damping ratio is low (less than
20%).

A spectrum can be presented as shown in figure 2.

Spectrum

Concept of vibration modes

The structure’s response is based on a sum of harmonic responses, which are part of the domain of solutions q(t) of
the equation:

They can therefore be written as follows:

If the Basile hypothesis is used (C is neglected), the algebraic solution becomes:

If the Basile hypothesis is not considered, one is now interested in finding complex modes such that:

and the eigenproblem can be resolved:

To get there without neglecting C, one had to postulate that:

With the matrices A and B being:

and

The vector ϒ is created such that:

The solutions presented as pairs of combined complex harmonics are calculated as follows:

with:

ψi: a complex modal deformation vector,

λi: a complex natural frequency.

In any case, real or complex, one must solve an equation in λi2 of the same degree N (2N for complex cases) than the
degree of the matrix N (2N for complex cases). The order of this matrix is equal to the discretized system’s number of
degrees of freedom (twice as much for complex cases). To solve the mentioned equation one must find the cases in
which the determinant of this matrix and its characteristic polynomials are equal to zero.

To simplify the problem, the case of real vibration modes assuming the Basile hypothesis is considered.

For each natural frequency ωi a modal shape vector is associated. The latter will be calculated by fixing one of the
components to 1, then solving a system with n-1 parameters.

Be careful! The modes are defined with a precision level of more or less a multiplying constant. Moreover, they
are later normalized using various procedures. The most common for FE software being the normalization
according to the mass matrix. This procedure is detailed in the following section. The modes can also be
normalized using their greatest modal displacement.

The vibration modes are orthogonal to the mass matrix, which implies that:

When there are no loadings, the modes have no physical meaning. Different methods can then be used to normalize
them to make their visualization more comprehensible. Thus, it is common in FE software to:

normalize the vibration modes according to the mass matrix, such that:
 normalize the vibration modes according to a particular vibration mode.

For the cases in which the software does not normalize this value, it can be denoted by the generalized mass term. It
can then be written that for the mode i:

The objective of this document is to explain case studies for which FE methods were applied, so it is important to
remember that the vibration modes are normalized according to the mass matrix, and thus the generalized mass is
always equal to 1.

The physical value called modal mass is also created, which will allow identifying the structure’s mass quantity that is
dragged by a mode in a given direction Δk:

Please note: modal mass is a very different quantity from generalized mass, as the above equation
shows.

Use of the modal basis

Time-dependent response by projection onto the modal basis:

A particular solution to the global equation:

can be broken down on the bases of orthogonal modal shape vectors into N independent problems that are describing
the response ri(t) of a simple oscillator:

In the case of earthquakes, the loading f(t) is an inertial loading applied to the entire structure:

Harmonic Analysis:

For this type of analysis, the goal is to construct a function describing the response of the structure according to the
frequencies at different nodes of the model. This response can be a quantity of arbitrary interest (displacement,
velocity, acceleration – absolute or relative -, efforts…) as a function of a variety of input harmonic loadings that can
also be described in various ways similar to the output quantity of interest.

In this case, the imposed loading on an arbitrary set of nodes are of the harmonic type:

The result of the analysis at all nodes of the structure is a complex transfer function whose norm (amplitude) and
argument (phase) are the most generally used. They enable us to identify the resonant frequencies of a structure of
equipment and to know the amplitude of its response according to different harmonic loadings.

Modal basis truncation – General case

Usually, because it is not possible to extract all the modes using calculations (too computationally costly, even
numerically impossible in some cases), one must focus on those that are most likely to respond to the loading of the
structure.

If the representative loading frequency saved is fc (Hz), one must extract the modes up to 2fc.

It is also important to make sure the modes inclined to contribute locally are well represented.

A search for vibration modes must not be accompanied by interpreting the frequency, but by analyzing the modal
shapes.
$translationBooks

PART 1 – THEORETICAL ELEMENTS

B2. Analysis based on a direct

temporal integration
B2. Analysis based on a direct temporal integration
Schemes of integration

Main Principles:

The principle of the different direct integration methods is to divide the studied interval into n smaller intervals of
length ∆t = T/n and to verify the equilibrium at the discrete times Ti = i ∆t = i T/n. The differences between the
methods (centered differences, Wilson, Newmark…) rely on the hypothesis that is considered about the variation of
the kinematic values on the interval ∆t.

Explicit scheme:

If the displacement value can be calculated directly from the value (or values) of the previous step, the method is said
explicit. In this case, the equilibrium is considered at the beginning of the increment.

Implicit scheme:

The implicit methods must consider the equilibrium at the end of the interval, which requires the resolution of a linear
system.

As stated earlier, there exist many schemes of integration. The objective here is not to make an exhaustive description
of all of them (see the specialized publications for that). Only two schemes will be presented in this guideline, the
centered differences method: an explicit and conditionally stable method, and the Newmark method: unconditionally
stable (for linear problems).

A scheme is unconditionally stable if, for any initial condition and time step ∆t, the solution is bounded, particularly
when ∆t/T is large. On the contrary, a scheme is conditionally stable if the obtained solution is bounded only if ∆t is
smaller than a critical value ∆tcrit.

Precision is a concept that is different from stability and it is extremely important for unconditionally stable schemes.
Besides the inevitable rounding errors, the precision has an impact on two sources of approximations: an (artificial)
increase of the period and a decrease of the amplitude. The influence of these two phenomena gains importance with
the increase of ∆t, but independently.

Centered Differences Method:

It is a scheme resembling finite differences. It relies on the approximation of the acceleration (Taylor series of the
second order):

To obtain the same order of error for the velocity, one uses:

The displacements at t+∆t are obtained considering the equilibrium at t:

Thus, introducing the approximations of the acceleration and the velocity:

Of the form:

This method requires a starting procedure to calculate q(-∆t) (from the equilibrium at t=0). The damping introduced by
the scheme is equal to zero (no decrease in amplitude).

Newmark method:
The scheme is based on the following approximations of the velocity and the displacement at the end of the intervals:

From the values of both parameters α and δ (ranging between 0 and 1) depends on the precision and the stability of
the method: the values (δ = 1/2 and α = 1/4) lead to an unconditionally stable scheme named the Newmark method: it
corresponds to considering the mean constant acceleration. Considering the equilibrium at the end of the studied
interval (at t+Δt), one obtains:

with

and

the coefficients being defined as follows:

The scheme also requires a starting procedure: the value of q¨(0) is obtained considering the equilibrium at t=0. Alike
the centered differences method, the Newmark scheme with basis (δ = 1/2 and α = 1/4) does not introduce numerical
damping. The values (α = 0, δ = ½) enable to get back to the centered differences method.

Choice of the spatial and temporal discretization

Criteria about the element's sizes that satisfy the wavelengths

Different criteria can intervene. They are related to the precision of the expected results and the type of calculations
used.

For the transitory analysis, the general recommendation is to have a range of 8 to 10 elements per wavelength. The
stationary waves being composed of a sum of propagation waves, the stationary dynamics analysis (in the case of
earthquakes typically) will be using the same type of criteria.

As a reminder, the general equation relating wavelength λ, frequency f, wave celerity c, and natural frequency ω is:

According to the type of elements used and the type of waves of interest, the following formulations are to be
remembered:

Volumetric elements in an isotropic environment (typical case of the modeling of an elastic soil od with Young’s
modulus E, Poisson’s ratio , and density ρ):

Shell isotropic elements:

These formulations can be written in a more general form for anisotropic media with 2 principal directions 1 and 2:

One can find the size required to target a cutoff frequency of 40Hz for shells, plates, and volumetric elements. The size
required is given by the numbers below / λ for 10 elements.

Size coeff. of the

Types of modeling Target Frequency [Hz] Thickness [m] Type of wave
element

40 0.2 Flexural 0.59

40 0.25 Flexion 0.66

40 0.5 Flexion 0.93

Shells and blocks – GC
40 1 Flexion 1.32
Concrete
40 1.5 Flexion 1.61

40 - Shear 6.04

40 - Compression 6.75

40 0.01 Flexion 0.16

40 0.02 Flexion 0.22

Shells and blocks – Meca
40 0.1 Flexion 0.50
Steel
40 - Shear 8.02

40 - Compression 9.58

This criterion is not necessarily sufficient. In the cases where a temporal analysis by transformation to the modal
bases, or a spectral response is sought, a sensitivity analysis must be conducted to verify if the refining of the mesh
leads to significant variations in the modal participation factors (or the modal mass if this criterion is examined). These
elements are described later in the chapter, and their relations are expressed.

The criterion for the time steps

The conditionally stable temporal schemes, like the scheme of centered differences, must satisfy a condition
concerning the time step chosen. This condition is often called CFL (Courant-Friedrichs-Levy).

Example of application for a model of steel truss whose smallest element has a length L:

Let us consider a bar element with two nodes 1 and 2.

The bars have a length L, a section A, a density ρ, and a Young modulus E.

The mass matrix is considered as the total mass uniformly distributed at both ends.

The matrices K and M are expressed as follows:

The characteristic polynomial is expressed such as:

Which leads to:

Since the propagation velocity of compression waves in a continuous medium is expressed as follows:

One can find that:

In this model whose mesh is irregular, it is the smallest time step that controls the global time step.
$translationBooks

PART 1 – THEORETICAL ELEMENTS

B3. Considering damping

B3. Considering damping
The damping of a structure’s vibration is related to dissipation phenomena of various origins that can happen inside of
a structure. It can be due to viscous phenomena specific to fluid dampers, which can be found in spring boxes or
dampers (cable insulators, Jarret™ type dampers, viscous fluid dampers). It can also be caused by structural
dissipation phenomena related to cracks and friction at the interfaces between materials. The energy is consumed in
the behavioral hysteresis of the materials.

In most cases, the damping is considered in the dynamic calculations through a viscous damping matrix C. The
components of this matrix can be exact at the local scale in the cases where the damping is completely controlled
because the damping equipment used has well-known characteristics. However, most of the time, for civil engineering
structures, the damping components are unknown because of the heterogeneity of the structures and the non-viscous
characteristic of the materials.

It is preferable to build the damping matrix using critical damping ratios as defined in the regulation for various
materials. Those ratios are defined using a viscous behavior hypothesis proportional to the solicitation velocity. It
represents a fraction of the critical damping value that leads to, for a simple oscillator, a return to equilibrium without
going through an oscillatory regime.

For the cases in which the damping components are known, the matrix C is constructed explicitly, and it is necessary
to use a suitable temporal calculation method to respect its characteristics:

Either with a direct integration calculation: it requires discretizing the time and to solve the dynamic
equilibrium at each time-step.
Either with a calculation after transformation to the modal bases, using complex vibration modes.

The analysis of rapid dynamic phenomena (explosion, impact) requires specific considerations.

It is not necessary to consider damping for local resistance verifications or vibrations over short periods, not long
enough to “activate” harmonic responses of the structure. It is then not necessary for those cases to consider
damping. It is interesting to note that this kind of calculations, realized with dynamic algorithms applying explicit time-
dependent integration schemes (LS-Dyna, Radioss, PAM-Crash, Europlexus…), do not use the classical approach, which
consists of inversing the stiffness matrix. On the contrary, for vibration analysis induced on a period following the
impact duration, it is necessary to consider and characterize the damping.

Note: for rapid dynamic calculations (milliseconds), the damping component is often neglected because its order
of magnitude is much smaller than the one of inertial local stiffness terms during the period of analysis

For the more “classic” cases, the damping matrix is generally constructed from the critical damping ratios defined for
materials and a linear combination of the M and K matrices that were previously calculated by the FE computational
model. This process enables us to considerably simplify the stages of calculation. This approach is related to the
“Basile” hypothesis for modal analysis. This hypothesis was formulated to identify the structural damping (mainly in
aeronautics), and is explained as follows:

“Even when modal damping coupling is present, the modal equations of motion are dynamically decoupled, for
structures with a low damping ratio, if the separation of the modes in frequencies is satisfied”.

Be careful! Considering the Basile hypothesis leads to an underestimation of the dynamic response elsewhere
than at the edge of the materials or at the supports, which are subjected to a high damping ratio compared to a
time-dependent calculation by means of direct integration, or using more complex methods such as the complex
modes method. This justifies the need for direct integration calculations of the motion equations, or complex
modes.

Most of the time, the Basile hypothesis is considered and the matrix is constructed starting from the Rayleigh method:

The terms α and β must be introduced. To be pragmatic, the following steps must be considered:

f1 is a lower bound of the first significant vibration mode of the structure, fn, and the first mode encountered after the
cut-off of the elastic response spectrum characterizing the applied temporal signal. The value of the reduced damping
is calculated for all values of ω, applying the following formula:

Be careful! Since the damping is largely overestimated after the 2 frequencies of interest, it is necessary to
master these two values. If fn is chosen adequately, the overestimation of the damping has no consequences
because rigid body motions are found on these ranges of frequencies, which will be excited in the response, and
the modes with frequencies greater than the critical frequency are insensitive to damping. However, one must
be certain that there are no local vibration modes with frequencies worth considering.

The impact of an error in the choice of the first frequency can have the worst consequences if a significant mode
was forgotten because its response will then be neglected.

It is important to note that in between the two frequencies, the damping is underestimated. The evaluation of
the response of the structure is therefore conservative.
$translationBooks

PART 1 – THEORETICAL ELEMENTS

B4. Specificities of the seismic

analysis
B4. Specificities of the seismic analysis
Spectral response – Specific case of earthquakes

The principle of the method is, for a given seismic direction, to construct the maximum responses from the loading
spectrum at all points, mode by mode, then, to accumulate all of them using various methods.

When the seismic response is obtained for a given direction, the seismic directions are combined to get the global
response.

Let us be in a case where the modal vectors are normalized according to the mass matrix, which is the case for most
of the FE computational models.

One can write the expression for a degree of freedom in a canonical form:

The term pik is called the weighting coefficient. It is an important concept in the case of seismic analysis because it
restores the physical contribution of excitation in a given direction k.

It is determined for the cases in which the modal shape vectors are normalized according to the mass matrix:

Moreover, for a seismic motion q¨s(t) in a given direction Deltak, the mean solutions greater than the maximum values
of r¨ik(t), r°ik(t), and rik(t) in terms of pseudo-accelerations, pseudo-velocities, and displacements via the respective
spectra of this motion, are known:

This response is constructed for a modal damping ξi that can be evaluated as a proportion of the modal strain energy
(see §1.2.1 of Génie parasismique Tome 3, J. Betbeder-Matibet).

with:

EiT is the total strain energy of the structure for mode i such that: EiT=1/2 φiTK φi
Ei material j is the internal strain energy of the structure with the damping of the material being ξj.

Then, from the simple oscillator response, one can construct the response in the structure for each mode, for the
acceleration, the velocity, and the displacement respectively:

These modal values must then be combined for a given loading direction to obtain the global response of the
structure. There exist several cumulation methods (from now on, ak,p, for the sake of example, denotes the component
p of the ai,k vector, such as φi,p with φi) as shown below:

The “Square Root of the Sum of the Square” (SRSS) – At all nodes p:

The Total Quadratic Cumulation (TQC) – At all nodes p:

with Pij denoting the quadratic coupling coefficient of the modes i and j, such that:

It is important to note that, in this case, it concerns a cumulation of algebraic terms, but the signs are all positive,
placing it under the square root don’t cause any problem.

Let us consider the modal damping values ξi and ξj equal to a variable such that x = {0.2, 4, 7, 20 et and 30%}, one
can graph, for the natural frequency ratios ωi/ωj, the curve (figure 3) that highlights the cumulation of the neighboring
modes. The cumulation coefficient decreases rapidly when the frequency ratio increases, and even more when the
damping ratio is low.

Graphic of the coupling coefficient for different damping ratios varying as a function of the natural frequency ratios.

This last observation shows that the TQC method is more appropriate than the SRSS because the latter does not
consider the cumulation of modes close to one another.

On the contrary, there are no advantages of using the TQC method as a comparison to the SRSS method when the
nodes are far from one another.

Be careful! The cumulated modal responses must not be used to calculate other quantities of interest. For
instance, the efforts deduced from the total quadratic cumulation method cannot be calculated using:
: this calculation is incorrect in comparison to the evaluation by the cumulation of the modal
efforts.

One must, therefore, for this special case, calculate as follows:

1.
2. For each term p of the fk vector, it can be stated that:

It is also important to keep in mind that criteria allowing to evaluate the state of cracks based on the invariants
of the stress tensor, and thus, of the principal stresses, (cf. for instance the criterion proposed in the annex LL of
the EN1992-2) cannot be considered after using a quadratic cumulation of any kind.

The spectral responses of each mode, after being cumulated for one seismic direction, must be combined to obtain the
global response. In this case, it is called spatial cumulation.

The spatial cumulations can be handled using different methods:

By square root of the sum of the squares of the responses obtained in each direction. This method gets rid of
the sign and all logical relations between solicitations. It provides only one scalar quantity for each quantity of
interest:

By algebraic cumulation of “Newmark” type. This approach relies on the hypothesis of independence of all
 the spatial responses. It introduces a weighting coefficient μ, with values ranging according to the different
standards, which considers the two other unfavorable responses compared to the preferred direction. It takes
into account the sign variability of all the quantities of interest. Thus, it leads to not one accumulated
response, but a total of 24 as demonstrated in the following equations:

Truncation of modal bases – Seismic case

For the specific case of earthquakes, and normalizing the modes according to the mass matrix:

It corresponds to the mass driven by the mode i in the direction k. This mass is therefore related to a specific direction.
For a mode i, there will be 3 modal masses as a function of the different loading directions (if we are in a 3D space, 2 if
we are in 2D, and only one for a 1D problem). Figure 4 highlights a 2D example for a frame.

The modal displacements multiplied by the weighting coefficient represent the deformation of the structure
whose product at all points with the response of a simple oscillator gives the time-dependent response of a
given model. This concept is detailed in the following chapter.

Mode 1 and displacement in the ey direction Mode 1 and displacement in the ex

direction

Illustration of the mass displacements for the same nodes in 2 different directions

As explained earlier if the model contains N degrees of freedom, there will be at most N natural modes of vibration.

However, since the search for natural frequencies and vibration modes is conducted numerically, not all the modes are
extracted. In theory, one must have:

In practice, the algorithm will stop running after a threshold value of the frequency fixed by the user is reached. Then,
the user must verify that there are enough modes to restore the percentage of mass required by the engineering rules
followed. Usually, the criterion is to restore 90% of the mass.

Be careful! If at the threshold frequency, the percentage of mass targeted is not reached, it might be necessary
to include a pseudo-mode.

Conversely, if a significant percentage of mass is reached at a low frequency, way below the threshold
frequency, it is necessary to incorporate a pseudo-mode or to complete the modal base. Indeed, the
participating mass of a floor for a local flexural mode inside a sizable structure represents a very small
percentage of the total mass.

One should also be careful concerning the antisymmetric modes with a modal mass that can be equal to zero
because the masses displaced around one axis of the structure will balance one another.

Pseudo-mode or static correction

As formulated before, the selection criterion of the natural modes concerning the cumulation of the mass is equal to
90% for a frequency of the same order of magnitude as the seismic response’s threshold frequency i.e. 40Hz
maximum.
Whenever this value cannot be reached, the lowest percentage withheld will be completed with an additional mass
associated with a “pseudo-mode” of vibration.

Let us remember that for a seismic direction k, there is an accumulated response on n modes such that:

To be rigorous, if there are N degrees of freedom, the formulation becomes:

As seen previously, the modes after the threshold frequency are rigid body motion modes. The structure reacts in
phase with the seismic loading that it is subjected to with a relative displacement equal to zero.

To complete the modal base, one can construct the pseudo-modes considering that the total response is the sum of
the “dynamic” response, taking into account the modal base of the n first modes withheld, and the term proportional
to the seismic acceleration of the support q̈s(t). Let us remember the following equation:

with Pk the displacement of the structure in the direction k subjected to the static loading equivalent to the mass of the
accelerated structure at the acceleration of the last mode n extracted from the natural frequency:

Since the relative acceleration response for the structure is equal to zero after the threshold frequency, one can simply
write:

Reformulating the previous equation as a function of , it will be easier to link the pseudo-accelerations
spectrum, considering the frame of spectrum analysis:

Thus, the following formulation can also be considered:

After evaluating these vectors, they must be cumulated.

Indeed, the temporal solution for a seismic direction is given using the summation of the components on the modal
base:

However, this combination cannot be applied to the maximum values that were just calculated previously.
$translationBooks

PART 1 – THEORETICAL ELEMENTS

Chapter C. Static non-linear

calculations
Chapter C. Nonlinear Static calculations
Nonlinear calculations are generally time-demanding: a finite element software solves a nonlinear problem as a series
of linear problems (incremental process). To prepare this type of calculation the user has to make decisions such as
defining the increments, choosing the algorithm, etc. Therefore, the process requires a certain amount of experience
from the user.

Most of the nonlinear computational models can, in principle, run until one or multiple conditions are satisfied, or in the
case of contact problems, until a specified minimum number of connections (or supports) disappear. Therefore, before
starting a nonlinear calculation, it is important to make a preliminary estimation, which will enable us to know when to
stop the calculations.

C.1 Nonlinear mechanical problems

C.1 Nonlinear mechanical problems

C.2 Why performing nonlinear calculations?

C.2 Why performing nonlinear calculations?

C.3 Implementation

C.3 Implementation

C.4 Convergence problems? Symptoms and solutions

C.4 Convergence problems? Symptoms and solutions

$translationBooks

PART 1 – THEORETICAL ELEMENTS

C1. Mechanical non-linear

problems
C1. Mechanical non-linear problems
1. Description of possible non-linearities

The non-linearity of a mechanical problem comes from the fact that the coefficients of the equilibrium equation
depend on the displacements of the solid itself in equilibrium. In other words, the equilibrium equation is generally
implicit.

There are several non-linearity categories for mechanical static problems:

Material non-linearities: in cases where the constitutive law is not linear or that the response of the
material depends on the loading history. In other words, stress is not a linear function of strains. The most
common case in civil engineering is one of the material loaded beyond their elastic capacity, which then
develops elasto-plastic behaviors. This behavior is characterized by a dependency between the stiffness of
the material and its stress state.
Geometric nonlinearities: for cases where the structure is submitted to big displacements or big strains. In
the first case, one can no longer write the problem neglecting the changes in the geometry of the structure.
In the second case, one can no longer approach the strains simply as the gradient of the displacements.
Boundary condition non-linearities: in cases where a structure is progressively loaded and there is
potential contact between two bodies with follower forces. These types of non-linearities also appear when
the construction phasing or the assembly of a bridge’s deck are simulated, when digging a gallery,
constructing an embankment, etc.

All the above non-linearities can be coupled if the algorithm allows it, but the resolution of the problem becomes more
complex.

2. Principle of resolution of a non-linear problem: Newton method

When solving a finite element problem, one looks for the displacement field u, such that the inner forces Lint are equal
to the external forces Lext:

, which is a non-linear problem as a function of u.

Generally, to solve the non-linear static problem, an incremental algorithm is used. For that matter, the problem is
parameterized in terms of t (with t representing a pseudo-time, unlike the t parameter used in dynamics). This
parameter is used to index the successive load-steps applied on the structure. More precisely, it consists of searching
for the equilibrium state corresponding to the successive load-steps F1, F2, …

This separation leads to solving a series of quasi-linear problems as shown in the figure below and to determine the
state of the structure at the time-step t (displacements, strains, stresses) knowing the solution at the state t-1. The
greater the number of load-steps, the better the precision.

time
increment.

Principle of parametrisation in function of t

At each increment ti the discrete problem is Ki x qi = Fi where qi is the unknown displacement vector under the applied
imposed loading Fi. While in the linear case seen in chapter 1 the K matrix was explicit, when the problem is non-
linear, Ki is a matrix with its terms depending implicitly on the value of qi. So, qi cannot be determined directly by
computing the inverse of the matrix K.

The most used method to solve this non-linear equation is to use a Newton-type algorithm. The idea is to build a good
approximation for the equation’s solution

considering its first-order Taylor expansion

One must start from an initial point (close enough to the solution) and then compute by iterations

At each iteration, one should evaluate the residual vector F(qk) until it exceeds (in absolute value) the value arbitrarily
close to zero. This convergence criterion must be chosen with care, respecting the standard used by the calculation
code (see section 3.3 for more details).

Note: With the Newton method, at each iteration, one should compute the tangent matrix at the considered point:

The computational cost of this matrix can be time-consuming. If using this matrix allows having a quadratic
convergence (so, in fewer iterations), it is not essential to use this matrix. Other strategies can be adopted to estimate
this matrix, namely the quasi-Newton methods. It is conceivable to use the tangent matrix without updating it at each
iteration, but also to use the elastic matrix (figure b) or the secant matrix in the case of a damage model. An
illustration of the successive iterations according to the used matrix is shown below.

Illustration of the Newton or quasi-Newton method (elastic matrix)

In general, using the tangent matrix allows a faster convergence (in fewer iterations) but the alternatives might be
more effective or more robust according to the situation.

As the method is iterative, the process should be stopped when the stop criterion is reached, in other words, when it is
verified that a given value (or several values) becomes negligible. The global algorithm can be written as follows:
by defining the increments, i indexing the Newton iterations and ε being a positive value close to zero.

Note: The Newton algorithm is used to solve the equilibrium at each time step. It can also be used to find the stresses
in each Gauss point (at all iterations of the Newton problem on the global scale) when the constitutive law requires it.
$translationBooks

PART 1 – THEORETICAL ELEMENTS

C2. Why performing non-linear

calculations
C2. Why performing non-linear calculations
As in many areas of physics, it is only reasonable to undertake a non-linear calculation after having a good idea of the
"end of the story". In other words, how the structure will evolve until it becomes unstable. Below are presented the
good and more questionable reasons for undergoing non-linear calculations.

1. Good reasons

To study the effort redistribution. When some parts of a structure enter the plastic phase, the stress level
“freezes” to a determined value. The work of external forces can only be dissipated by increasing the stress
level somewhere else or by making matter flow from other zones that are already in the plastic phase. The
goal here is to verify in which order the structural elements “fail”, and to find the final failure mechanism.
To determine the evolution of support conditions of a structure, either by a contact computation or by
developing internal plastic joints.
Before reaching a state where the efforts are redistributed, to obtain an equilibrium state where, in the
plastic zones, the stresses are on average close to the yield stress of the material and the stresses are in the
elastic domain elsewhere. For geomaterials, to obtain an equilibrium state where the stresses are purely
compressive, except for some punctual zones (eliminate tensile stresses).

These procedures are reasonable, because:

Generally, they can be conducted in a small deformation framework,

They allow simplifying the model and the interpretation of the structure’s mechanics: since the plastic zones
have no structural purpose anymore, a weakening analysis of the structure via a succession of elastic
calculations can be proposed, which will “deactivate” the plastic zones at some stress levels. The analysis of
the plastic zones can also suggest a static approach to the plasticization (strut and tie model).

2. Questionable reasons

To determine the shape of the structure in response to large deformations (even while “exploding”),
eventually using an automatic re-meshing algorithm provided by the software,
To see if it is worth performing calculations with a behavior that seems very complete from a descriptive point
of view, or if elastic calculations performed previously give better results.

Those motivations often end up leading to non-convergence or “zero-pivot” problems. When the calculation is
interrupted at a given stage of the iterative process, the deformation of the structure (given by the deformation of the
mesh, or the strain field) is misleading:

The deformations are in general superior to 1%, so at this stage, the mechanical problem is no longer
expressed with correct variables,
The calculated deformation depends critically on the meshing. It means that for two different meshes initially
close, one can obtain very different local deformation values,
The calculation “follows”, by definition, one instability at a time; however, the large displacement problems
are by its nature multi-branched. In other words, in a certain state, there are several possible equilibrium
configurations for a given loading…
$translationBooks

PART 1 – THEORETICAL ELEMENTS

C3. Implementation
C3. Implementation
The correct method consists of always proceeding step by step so that non-linearities are not introduced all at once:

Performing an initial calculation with an elastic constitutive law before using non-linear ones,
Performing a calculation without contact before performing one with it, a calculation with no friction before
including it, …

Every step should be verified properly, and it is necessary to make sure that the solution converges when
the mesh and the time-step are refined.

Finally, it should be clear that in non-linear cases, the problem to be solved may have no unique solution (instability,
bifurcation, etc.).

1. Choosing the discretization

The discretization must be chosen focusing on the fact that it must be consistent with the loadings that one wishes to
apply, especially if those are not monotonic.

The more important the non-linearities are, the smaller the load-increments must be. Also, if the behavior of the model
is complex, the integral of the constitutive law is not completely implicit: in that case, it should be verified that the
increments are sufficiently small for the model to converge.

According to the standards, the management of the time-steps can be automated to a certain extend. This allows us to
reduce the time-steps if there is a convergence problem, or even to increase it if the non-linearity is weak and the
convergence is fast. For the sake of diligence, it is advised to verify that the convergence is reached by performing a
second calculation with smaller load increments.

2. Choosing the boundary conditions

It should be clear that in the case of softening-types constitutive laws or if a limit load exists, the loading with imposed
forces might become not admissible, as shown in the figure below.

In the first case, the closer the imposed forces get to the limit load, the harder it will be for the model to converge until
it becomes impossible. In the second case, it will be impossible to exceed the effort corresponding to the first peak and
obtain a solution after that.

Examples of response with limit load or loss of stiffness

There are other cases, where there is no unique solution for a given effort or displacement. It happens, for example, in
the case of buckling of thin shells or when the problem has bifurcations, for instance, dissipative branch and elastic
branch (see figure below).
Response (a) with multiple branches (b) non-monotonic loading and displacement

If the model requires imposed forces, the problem can be solved by continuation or arc length methods (see
Code_Aster [R5.03.80] documentation for instance).

For an algorithm to work effectively, it is then necessary to conceive its model and to prepare its data, so that one
looks for a unique solution by introducing multiple restrictions, such as:

The need for equilibrium stability (in a specific way to be explored!)

Equilibrium with a “minimization of plastic deformation energy”,
Equilibrium with a “minimization of average curvature”,
Equilibrium with a “minimization of the difference between reaction forces”

3. Evaluation of the convergence of the global non-linear algorithm

The stopping criterion for the Newton algorithm must be considered carefully. It is advised to verify what limit criterion
is programmed by default in the software. For example, certain algorithms test the convergence using the norm of the
residual (in that case, the criterion depends on the unit used for the forces in the calculation, the precision is not the
same if the required residual is 1N or 1MN). Other software use a criterion where the norm of the residual relates to
the norm of the loading (in this case, the convergence does not depend on the chosen physical unit). Special
precautions must be taken when different elements are mixed (solid, structural elements) or when compound
elements are used.

In general, one must be careful when the convergence criterion is modified from the recommended value to reach
certain results.
$translationBooks

PART 1 – THEORETICAL ELEMENTS

C4. Convergence issues?

Symptoms and solutions
C4. Convergence issues? Symptoms and solutions
As opposed to linear elastic calculations, non-linear calculations can perform iterations indefinitely, never verifying the
stopping criterion, but they can also be interrupted abruptly at a given load increment (after the first one), with the
warning “zero-pivot”.

Let us examine the causes of this type of errors.

1. Non-convergence of iterations

This phenomenon is almost always a direct consequence of the inadequacy of the used iterative method: the problem
to be solved certainly does not satisfy the mathematic conditions, which ensure convergence (non-contracting
operator, non-positive operator, non-convex operator, insufficient restrictions in the interval in which one looks for the
solution, etc.).

This phenomenon may be a consequence of accumulated round off numerical errors that can be attributed to the
algorithm itself; however, it does not happen often.

The evaluation of the residuals over the iterations may provide information about the nature of the problem:

The residuals tend to zero, but more slowly. This behavior indicates that the mesh is not adequate to
calculate the equilibrium: it is the case of the localized phenomenon, where the non-linearities are
concentrated in certain parts of the structure. In that case, it is better to restart the calculations with a mesh
that is better adapted to the structural mechanism.
The residuals become larger and larger. It means characteristic instability: there is no longer, for this
structure, a possible equilibrium configuration at this loading level.

After a given state, the residuals increase abruptly to a certain value and then start decreasing again, etc. This
behavior appears when the algorithm finds multiple possible solutions: it converges to one of them, then “bounces”
towards another one. In other words, there are multiple possible equilibrium configurations!

2. Abrupt stop

The abrupt stop of the algorithm in a non-linear calculation is often reported as a “zero-pivot” warning when solving a
linear system. This expression refers to the “pivot” in the Gaussian elimination method, which is the algorithm used
(with small variations) to solve linear systems. Most parts of iterative algorithms replace the problem with a linear
system.

This situation corresponds to the structure becoming unstable; in elastoplasticity or damage, it means that the
structure lost its cohesion. For large displacements or contact problems, it reflects the formation of a mechanism,
especially, the possibility of rigid body movement: all interface nodes slide, and the restraints are not enough to reach
equilibrium.

It may also happen that the algorithm stops for reasons not connected to the state of the structure whatsoever: disk
space overflow or programming bug. The first one is easy to solve.

3. When nothing works

If a non-linear calculation does not converge, and one cannot find in the explanations above the reason, it is
recommended to ask oneself the following questions:

Is the initial configuration of the structure in elastic equilibrium?

Is the meshing adapted to the problem one wants to solve? Is the mesh sufficiently refined in the zones where
the plastic deformations take place? Is the mesh not too deformed?
Does your model mix solid elements with structural elements (beams, plates, shells)? Do the zones with
plastic deformation/damage affect the nodes shared with those ‘Strength of Materials’ type elements?
Are the finite elements well-adapted to the problem? Are there stress oscillations?
$translationBooks

PART 1 – THEORETICAL ELEMENTS

Chapter D. Civil Engineering

Chapter D. Civil Engineering
D.1 Civil engineering materials

D.1 Civil engineering materials

D.2 The categories of structural elements

D.2 The categories of structural elements

D.3 The steps of construction

D.3 The steps of construction

$translationBooks

PART 1 – THEORETICAL ELEMENTS

D1. Civil engineering materials

D1. Civil engineering materials
The objective of this chapter is to highlight the peculiarities of civil engineering materials regarding FE modeling.
Among those, there are the particularities of the civil engineering materials themselves (concrete, timber, steel), and
their use within the structure.

1. Concrete

Concrete is a compound material composed of cement paste, and aggregates of various sizes. The cement paste that
contains water and additives acts as a binding agent (glue) between all the aggregates. Even though concrete is a
heterogeneous material, it revolutionized the world of construction because of its characteristics such as its versatility,
adaptability (offering a variety of aspects: form, color, texture), mechanical resistivity, fire resistivity, acoustic
insulation, and diversity (high-performance concrete, precast concrete, etc.). However, concrete is subjected to
degradation processes. The latter can be due to environmental factors such as humidity, rain, cold, freeze-thaw, heat,
wind, drought, aggressions from deicing rock salt, alkali-reactions, internal or external sulfate-reactions, corrosion,
mechanical solicitations due to collisions, and load increase. Depending on the degradation phenomenon, the
deterioration of the structural element is progressive and initiates as soon as the concrete tensile stresses are applied.
Degradation leads to the emergence of small cracks and localized deformations. The increase of the deformation leads
to the decrease of the material’s strength, which can then lead to softening behavior and the collapse of the structure.

Constitutive law – In the FE calculations, generally, only some aspects of the concrete behavior are considered
because of its complexity. To express the most relevant constitutive law, it is necessary to well understand the
characteristics of concrete. Thus, it is essential to know the behavior of the cement paste, as well as the behavior of
the aggregates, both responsible for the stiffness properties of the material. The paste is the cause of the mechanical
strength of the concrete and, because of its water content, is also the cause of the deficiencies such as the increase of
porosity, the decrease in mechanical strength, the delayed effects (shrinkage, creep), and the transfer of aggressive
agents.

Concrete is considered as a “quasi-fragile” material because of its complex behavior. Its main characteristics are
highlighted in the literature. The most significant one is its fragile behavior in tension and more ductile in compression
and when subjected to unilateral forces in a tension/compression cycle. Other properties must be considered such as
the shrinkage and creep according to the environment and the loading applied to the concrete.

In the literature, the two main approaches to model the behavior of concrete are:

Models relying on a continuous approach, meaning that concrete is considered as a continuous media
(Bazant, 1979). Using this approach, the calculations of the cracks are deduced from the displacement-force
relationship. There exist different models such as the elasto-plastic (Ottosen, Drucker-Prager), damage
(Mazars, Laborderie), smeared crack model, of gradient type, or energy type models (fictitious crack model
(Hillerborg, 1984), crack band model). In some cases, elasto-plastic models are used, but the user must be
very careful, once the elastic phase is overcome, it is possible to obtain deformations that do not reflect the
behavior of concrete at all.

Models based on a discrete approach. Indeed, concrete cracks generate geometrical discontinuities, which
are integrated between the boundaries of the connected elements. There exist discrete models of Ngo, (Ngo,
1967), Bazant (Bazant, 1979), Blaauwendraad (Blaauwendraad, 1999), and other lattice-type models.

However, in the literature, the two most used model categories are the ones based on mechanical damage and the
ones considering explicit discontinuities.

Although the elastic component is not exactly linear ([Baron] p.276; [Sargin 1971]), the calculations are generally
conducted using the isotropic linear elastic assumption (Hooke’s law) to represent the elastic phase of the concrete
behavior. The values of Young’s modulus and Poisson’s ratio are computed by characterization of the concrete. In the
eventuality that other mechanical characterizations of the concrete cannot be conducted, some properties such as the
compressive strength can be deducted using model laws that can be found in model-codes such as Eurocode 2 for
instance.

Note – Design offices use the model-codes except for study cases when they conduct detailed analyses on a
specific structural element on which particular mechanical phenomena occur: cracks, heave, or local
heterogeneity. Indeed, there exist finite elements of “cracked element” type, damage models, viscoplastic laws,
or even poro-mechanical coupling laws. However, to conduct these studies, data from on-site or sample analyses
of the material must be available to orient the engineer towards adequate constitutive laws, for which,
coefficients and moduli will have to be determined anyway.

Elastoplastic model – For the sake of simplicity, concrete is often modeled as an isotropic homogeneous
elastoplastic material, which is acceptable until the concrete experiences cracking.

Regarding an elastoplastic law, concrete cracking is not directly modeled: the cracking zones are characterized by
important anelastic deformations (>1-2%) and a fixed constraint state between tensile ft and compressive strength fc
of concrete.

As shown in chapter 3, an elastoplastic constitutive law for concrete is composed of an elastic law and a plastic flow
rule associated with a “criterion of plasticity”.

However, the post-cracking behavior can be approximated using a stress-strain curve going further than the tensile
strength. The latter can cover the effects of tensile softening (representing the necessary work to open the crack) and
tensile hardening (concrete contribution in between the cracks, stresses due to reinforcement adherence).

If the decreasing behavior of the material leads to a global decreasing behavior, one must pay attention to the
eventual local effects: the size of the finite elements will limit the size of the anelastic zones and the solution will
depend on the used mesh. Different numerical techniques enable us to solve or limit this issue.

Damage – A damage law is a law that considers one of the main macroscopic effects of concrete cracking: the loss of
material stiffness. The fundamental idea is to renounce following the eventual cracks (their apparition and
propagation) and consider that the concrete of a given structure deteriorates by multiplication of cracks in the
damaged areas. This type of law allows us to describe the decrease in material stiffness undergoing small cracks. This
stiffness loss is measured by an internal variable called damage, denoted D, that evolves from 0 (undamaged
material) to 1 (totally damaged material). This variable is generally a scalar value.

To best represent the behavior of concrete, the damage laws consider post-peak softening behavior. This enables
determining the stresses as follows:

with

The advantage of this method is to consider the concrete as a “continuous medium”, for which the FE methods are
perfectly adapted.

In the case of concrete behavior, the two main damage modeling families are the anisotropic and isotropic ones. The
isotropy characterizes the invariance of the physical properties of concrete regardless of the direction. On the other
hand, the anisotropy depends on the direction. An anisotropic law presents different responses due to loading
according to its orientation.

One of the most used damage models in the industrial and research world is the Mazars model [Mazars, 1984]. It is
certainly the first damage model for concrete that works accurately.

The main difficulties raised by the damage models are:

a dependence a priori of the meshing results: note that one should, in principle, demonstrate that the ruin
mechanism obtained with this type of model is independent of the refinement of the mesh, at least after a
certain threshold. Besides, this dependence led to the development of the regularization method.
the absence of an analytical solution in these simple cases.

Several regularization classes exist including the non-local regularization and the regularization by cracking energy
[Hillerborg, 1976] (that solves the problem only partially). Among those non-local methods, it is possible to cite the
integral methods [Pijaudier et al., 1987], [Giry et al., 2011] or gradient methods (deformation gradient or internal
variable gradient [De Borst et al., 1992], [Peerlings et al., 1996], [Nedjar, 2005], [Lorentz, 2017]. These methods
require the use of relatively thin meshes, which renders the analysis computationally expensive.

Delayed effects of the stress relocation – When analyzing the behavior of a concrete structure several weeks after
pouring, but especially for the long term, it is necessary to consider the delayed effects such as shrinkage and
creep.

These phenomena specific to concrete can, in principle, be modeled adopting a visco-elasto-plastic constitutive law
(Bingham law), or sometimes only visco-elastic (“scientific” creep, [Eymard]): this approach is generally applied by
research laboratories to analyze tests on materials. However, in the case of a refined model considering the delayed
effects, it is necessary to incorporate phenomena such as drying and cracking in addition to shrinkage and creep.
Indeed, all these phenomena interact with one another, so it is important to model them numerically.

The deformation caused by the shrinkage is induced by the drying of the concrete due to the environmental effects.
The shrinkage leads to a differential deformation, meaning that the stresses are more significant where the concrete
dries faster. This phenomenon causes tensile stresses at the surface, which then leads to cracks and compressive
stresses inside the body.

Concerning creep, its deformation is generally separated into two deformations, one is due to creep itself, and the
other one to shrinkage from the drying process. The latter can be explained by the strong dependence between
relative humidity and creep.

In Eurocode 2, it is possible to determine concrete deformations due to the delayed effects (without external loading).
For this purpose, one must compute the deformation due to the endogenous shrinkage (caused by the internal
humidity) and the desiccation deformation (caused by the drying process and the size of the structural element).
According to the §3.1.9 of the EN1992-1-1, the creep deformation under compressive stresses σc can be written as:

with Ec the tangent modulus (equal to 1.05 Ecm) and φ the creep coefficient.

The Eurocode 2 (EN 1992-1-1) describes a simplified calculation method of φ (∞, t0). Annex B gives a more complete
method enabling the estimation of φ(t, t0) as well as the shrinkage evolution (see the annex B of EN 1992-2).
Moreover, it is important to remember that the calculation of the delayed deformation is relative to the type of cement
paste.

2. Structural steel

The behavior of steel is much simpler than the behavior of concrete for several reasons: it is an isotropic material with
identical strength and moduli in tension and compression. Moreover, it is subjected to industrial quality control to
ensure its homogeneity.

Elastic models – Even though steel behaves essentially as anisotropic elastoplastic material (Eurocode 3, part 1.1,
§5.4.3), the typical models for steel structures, or compound containing steel, adopt a linear elastic behavior. One
must then check that the elastic constraint threshold fy was not reached. In the case of beam or column type
elements, the codes allow, if the sections have a size that guarantees a sufficient local ductility, to exceed in the
analysis the elastic strength and consider reaction moments based on the plastic distribution of the stresses.

Elastoplasticity and strain hardening – The elastoplastic theory was developed from the study of steel alloys,
mainly to predict the rolling and forging stresses ([Hill], [Nadai]). For steel, the usual model is composed of Hooke’s
law for the elastic deformation and the plasticity criterion denoted “Von Mises criterion”, as explained in the Eurocode
3 (part 2, §7.3; part 1.5 §10 and part 1.7 §5.2.3.2): for this criterion, only the elastic constraint threshold fy of steel
must be provided.

Considering the FE calculations, another issue is the strain hardening of the material, meaning the hardening in the
purely plastic phase. This aspect is notably explained in the annex C6, part 1.5 of the Eurocode 3. Since it is hard to
verify the model, it is preferable to conduct a preliminary computational model with no strain hardening: indeed, one
can then verify the quality of the obtained results in the plastic zones by observing the eigenvalues of the Von Mises
stresses (the plastic zones must be about monochrome). Nevertheless, it is important to note that the use of an elastic
perfectly plastic law with no strain hardening can lead to convergence issues of the linear analysis. Indeed, the plastic
zones have zero tangent modulus and, therefore, no stiffness. Considering strain hardening can stabilize the numerical
resolution.

In the case of strain hardening, until a stress fy+X is reached, the possible loading/unloading phases are elastic, with a
modulus equal to the initial one. Thus, fy+X becomes the new elastic yield stress of the material. Moreover, the
behavior of the steel is close to the kinematic strain hardening model. As a first approach, one can consider a tensile
strain hardening up to a certain stress value (fy+X) that will decrease the compressive elastic yield strength to fy-X,
and vice versa.

The most common steel model is the isotropic strain hardening, which depends on the accumulated plastic
deformation:

where and fy an increasing function.

In this case, there is no distinction between tension and compression, so a strain hardening in one direction leads to an
increase of the compressive elastic yield stress. This hypothesis is accurate only if the considered loading is monotone
and not cyclic.

In most numerical computational models, one can retrieve the eigenvalues of the cumulated plastic deformation. Thus,
it is possible to verify that the Von Mises constraint is distributed according to this deformation in the plastic zones.

In the specific case of stainless steel, the hypothesis of a linear behavior up the limit of elasticity fy corresponding to
plastic deformation of 0.2% is not respected, so it is recommended by the Eurocode 3-1-4, § 4.2. to consider the
effects of the nonlinear behavior on the force-displacement relationship in the deflection calculations. The material law
is of Ramberg-Osgood type and recommendations to implement this model can be found in annex C of this section of
the Eurocode.

3. Pre-stressing steel

Regarding their behavior, steel cables (braces, tension cables) and steel for pre-stressing differ from other structural
steel only by an elastic limit about three times greater (from 1680 to 2140MPa in France), which is justified by the
serviceability solicitations, generally greater than 1000MPa. One can then, firstly, model the pre-stressing
reinforcement as linear elastic elements or elastoplastic of Von Mises type. Steel elements for pre-stressing are all hot-
rolled, so the linear isotropic strain hardening law in the plastic phase should be taken into account. The strain
hardening coefficient h can be calculated using the characteristics guaranteed by the manufacturer:

with Rm the constraint to failure, fp0,1 the usual elastic limit, E the Young modulus, and Agt the elongation at the point
of failure.

A direct consequence of this high level of stress is the initiation of the relaxation mechanism. Relaxation of steel is a
non-elastic delayed effect that depends essentially on the time spent since the loading was applied. It leads to a
progressive decrease of the constraint for bars and cables subjected to a constant elongation. This mechanism
initiates at ambient temperature only for elongations exceeding about 60% of the elastic limit (≈1000MPa). Relaxation
increases slightly with temperature.

Alike concrete creep, it is possible to model the relaxation using a linear visco-elastic constitutive law. However, this
approach is mostly used in research work. It is not common to model steel for pre-stressing: generally, it is accounted
for by introducing a distributed force within the pre-stressed concrete. Nevertheless, for detailed analysis, in which the
interaction between the steel cables and the surrounding medium is of interest, relaxation can be considered using an
incremental time-dependent calculation with the relaxation losses introduced as initial stress conditions.

4. Passive steel

Normal mechanical actions of passive steels – For FE calculations, passive steel elements are frequently modeled
by linear 1D elements of “bar” type. According to the historical reinforced concrete methods, the reinforcement is
considered to work only in tension/compression along its own axis, which is exactly what FE models enable.
Connecting those bar elements to the nodes in the solid elements representing the concrete makes it possible to avoid
mechanisms specific to “bar-only” assemblies.
In most computational software, the bar elements are considered by default as elements with a linear elastic behavior.
Considering the 1D characteristic of these elements, the law is of the type N = E A u/L, with u the axial displacement,
and N the normal stress at the nodes. It must be verified afterward that the axial force N/A stays within ±fy.

Friction – What was said in the above paragraphs is valid only if there is a continuous friction force acting between
the steel reinforcement and the concrete.

5. Timber

Timber is a material that possesses some inherent peculiarities: it is not homogeneous (this is true at various scales), it
is not isotropic, it does not show symmetrical behaviors, and it can be subjected to ductile or fragile failures according
to the solicitation and its orientation. Timber is sensitive to humidity, which has a direct impact on its dimensional
characteristics, stiffness, and strength. The duration of the loading has an important impact on the strength and the
deformation of the wooden elements. The variations in humidity can amplify the deformations (mechano-sorption).
Relying on the previous observation, one can imagine a method to implement the FE analysis in this particular case.

Homogenization – The hypothesis must be validated for the Representative Elementary Volume (REV), or at least,
the volume of the smallest finite element. Knowing that according to the different timber species the growth ring can
exceed 1cm, it becomes difficult to assume this homogenization hypothesis close to the connections (peaks, pins,
bolts…) with a diameter of the same order of magnitude, or smaller.

The presence of nodes is rarely considered when modeling structures or structural elements.

Orthotropy –Timber has a structure and characteristics that are a function of three directions, the lengthwise
directions (the axis of the trunk, wood fibers), the radial directions, and the tangential directions perpendicular to the
lengthwise direction. The last two constitute a plan, often named the cross-sectional plane of the beams, on which the
growth rings can be visible. The orthotropy coordinate system is therefore quasi-cylindrical. On the contrary, the
coordinate system in which the elements are inscribed is cartesian. The representation of this orthotropy, when
considered, is in most cases limited by an isotropic transverse hypothesis (radial and tangential axes having identical
characteristics) in a cartesian frame. The slope of fibers associated with the presence of nodes is generally not
considered in the FE calculations. Rather, it is modeled in the assembly calculation especially for the assemblies by
contact.

The flexibility matrix (inverse of the stiffness matrix) can be defined as follows for an orthotropic plan:

Note: the orthotropy and symmetry hypotheses of Sij reduce the number of independent terms from 36 (most general
3D case) to 9 terms.

Timber – orthotropy hypothesis

Elastic modeling – One must simply know the isotropic transverse matrix of behavior for a 2D or 3D model. The
moduli between the lengthwise/longitudinal, radial, and tangential directions can present ratios of the order of 20.
They evolve as a function of the loading period (creep), timber humidity when constructing, and during its lifetime
(environment). The representation of the FE method will depend on the pertinence of the parameters considered.

Plastic modeling, failure criterion – For an isotropic transverse non-symmetric material, the Hill, Tsai… criterion
can be used knowing that it will be necessary to describe the perpendicular tensile and shear fragile failures. The
significant strength variability makes it harder to fix the criteria parameters. The sequence of growth rings or
assembled layers with different mechanical characteristics can render a homogeneous model difficult to fix in terms of
strength. Indeed, the strength and stiffness are highly correlated for timbers and “systemic” effects appear quickly in
terms of element strength. Consequently, there are strength limits in axial tension, axial compression, bending….

Finally, timber structures are particularly sensitive to the behavior of their assembly. Thus, the elements do not need
to be modeled in detail, only the assembly zones do. Nevertheless, problems can arise because of the contact zones,
the multitude of materials, some of them entering plasticity, the fragile failure of others, and the homogenization limits
reminded earlier.

It is clear that the modeling effort is related to the scale of the investigation or the advancement of the project.
Provided the above singularities are considered, the modeling of timber inside of a structure can be conducted like any
other material.

Modeling of the delayed effects and the interaction with the hydric/water phenomena – Timber is a
hygroscopic material sensitive to the relative variation of humidity of the air. Moreover, it is subjected to creep under
loading. If one wants to model these phenomena, a rheological viscoelastic model in a variable environment and
respecting the thermodynamic principles can be adopted. The generalized viscoelastic behavior of Kelvin-Voigt type,
characterized by rheologic parameters and dependent on the level and history of humidity can be associated with the
Ranta-Maunus non-viscoelastic behavior to characterize the shrinkage-swelling and the mechano-sorption.
$translationBooks

PART 1 – THEORETICAL ELEMENTS

D2. Different categories of

structural elements
D2. Different categories of structural elements
Here are presented the peculiarities of these elements regarding FE calculations.

1. Reinforced concrete elements

Considering the delayed phenomena – Usually, the project engineer is interested in the redistribution of stresses
and the delayed deformations that accompany the aging of the materials.

To achieve this, the static calculation by time increments is particularly well adapted and largely sufficient. The
shrinkage and the creep of concrete for instance depend only on the time that has elapsed since the pouring phase.
Using the incremental calculation, they will be introduced as imposed deformations at each node of the mesh. It is
possible to compute beforehand, for each time increment the shrinkage and creep related deformation map: the FE
software incorporates the initial state seeking an equilibrium of the elastoplastic material.

Nevertheless, attention should be paid to the interaction between the delayed effects and the construction phases, as
it will be explained later.

Consideration of cracking - When concrete cracks in tension, cracks develop towards the nearest face, as well as
along the steel-concrete interface ([Goto], figure below).

Internal cracking of the reinforced concrete

When considering a tensioned connecting rod of cracked reinforced concrete as a homogenized continuous medium,
the behavior law N(ε) (with N the normal tensile stress) follows the trend of the figure below.
Model of the tension stiffening effect

The Model Code 1990 of the CEB-FIP expresses an analytical formulation. This relationship can be used by modeling
the section of cracked concrete as a bar element whose stiffness would be updated in increments (depending on the
value of normal stress or elongation, since the relationship is invertible). It can also be used as a law for the behavior
of a fiber of a multi-fiber element.

Cast Concrete - For cast concrete areas (e.g. by passive reinforcement), it is possible to take into account the
residual post-crush strength, as indicated in EN1992-1, using Sargin's law:

with εc1 the peak value such that

with

For

The EN1992-1 §3.1.9 also suggests an increment of strength and deformation of concrete when it is subjected to a
confinement stress of σ2=σ3 such that:

However, this law is only one-dimensional and is only valid under monotonic loading. Therefore, it can only be used
with the finite element method when modeling concrete confined by a bar element (so the modulus would be
incrementally modified), or as a constitutive law of a fiber of a multi-fiber beam element.

2. Pre-stressed Concrete Elements in Pre-tensioning

Prestressing by pre-tensioning, which is characteristic of industrially prefabricated concrete products (beams, pre-
slabs, honeycomb slabs, girder, etc.), consists of tensioning cables (wires, strands or bars) on manufacturing benches,
then pouring the concrete before slackening the cables when a minimum concrete strength is reached (called concrete
sag resistance). The amount tension in the cable (which must not exceed the maximum prestressing force allowed by
design codes), the number of cables, and the strength of the concrete are adjusted according to the loads that the
prestressed floor or element must resist.

When the cables are loosened, instantaneous losses, of the order of 8% for a prestressed pre-slab for example, must
be taken into account (losses due to the entry of anchors, to the relaxation of the prestressing reinforcement during
the period between the tensioning of the cables and the transfer of the constraint, to the elastic shortening of the
concrete under the compressive stress imposed by the constraint) when designing in the provisional phase of
construction with a prestressing transmission length to be considered starting from the end of the precast concrete
element.

In the longer term, delayed losses due to shrinkage of the concrete, relaxation of the steel, or creep of the concrete,
which will finally reach a total loss of 20% for a prestressed pre-slab, must also be considered.

In many FE computational software, prestressing can be incorporated in beam-type finite elements representing the
cables connected to volumetric finite elements representing the concrete. Depending on the study area, it may be
necessary to consider the transmission length of the prestresses in the cables, for example in the end zones. The
distribution of the actual prestressing force (considering the instantaneous or delayed losses as a function of the
moment in the service life of the product where the calculation/verification is to be performed) is then variable along
this transfer length. A linear distribution of this prestressing force is allowed in most computational software and
remains safe for dimensioning purposes compared to the more realistic parabolic distribution.

For reasons of complexity and need (designs limited to elastic deformations), the modeling of these elements is
generally carried out with linear assumptions (constitutive laws of concrete and linear elastic steel, perfect contact
between concrete and steel...). However, for refined studies, non-linear hypotheses can be followed according to the
need such as damage type constitutive laws for concrete, elasto-plastic type constitutive laws for steel, and the
introduction of steel-concrete interface elements.

3. Pre-stressed Concrete Elements in Post-tensioning

Elastic losses - Post-tensioning of the prestressing cables is accompanied by instantaneous losses: friction, anchor
recoil, and loss due to elastic elongation.
Delayed effects: creep, shrinkage, and relaxation - Delayed effects are considered using an incremental
calculation. Creep and shrinkage can be introduced as incremental volume deformations given at each node of the
mesh.

4. Steel Structures

Choosing the type of analysis

Due to their high slenderness, steel structures are very deformable. As a result, the traditional assumption of reaching
stress equilibrium in the initial configuration is not always valid. It is then necessary to establish the internal stress
distribution in the deformed configuration.

The sensitivity to these non-linear effects, sometimes called second-order effects, is expressed through the critical
multiplier αcr: the load multiplier leading to the Eulerian instability of the structure. In the current version of Eurocode
3:

if αcr > 10, non-linear effects can be neglected. If the global structural analysis incorporates the plasticity of
the elements the limit value of αcr is increased up to 15,
if 4 < αcr < 10, they must be taken into account. This can however be done with a classical linear elastic
analysis by amplifying the transverse forces,
if αcr < 4, the non-linear analysis is mandatory.

In the last two cases, if the global geometrical irregularities (or imprecisions) and the element irregularities influence
the global behavior, they must be considered.

Element irregularities include:

geometrical irregularities: transverse and torsional,

material irregularities: rolled elements or welded elements have self-balanced residual stress distributions
created by their manufacturing process.

The latter can be represented by an equivalent geometric irregularity, whose value can be found in the current
standards.

These irregularities must be included in any analysis of a structural member incorporating non-linear effects.

In principle, structural or Strength of materials models (bars, beams, plates, and shells) are ideally suited for structural
steel calculations.

Wireframe structures

Elastic analysis

For modeling purposes, it is necessary to analyze precisely:

the nature of the connections between the different structural elements,

the load propagation from each part to the others.

Plastic analysis

When the elementary ductility and/or the assembly ductility is ensured, it is possible to carry out structural analyses
incorporating plasticity. Different methods can be used (EC3, §5.4.3):

elasto-plastic analysis, where the plasticized areas are modeled as plastic hinges,
non-linear plastic analysis, taking into account the partial plasticization of the bars along the plastic zones,
rigid-plastic analysis, in which the elastic behavior of the bars between hinges is neglected.

Torsion

Steel elements are rarely massive, they are composed of thin layers to form profiles, open or closed. Especially in the
first case, the response to torsional stresses is both in uniform torsion, known as Saint Venant torsion, and in non-
uniform torsion, leading to the buckling of the section. The latter phenomenon is usually not considered in commercial
software, even though it can have a significant effect on the response of structures. In this case, two solutions are
possible:

surface modeling of the element: this solution is not applicable if the analysis is performed on a structure with
more than a few elements,
if the situation is similar to that of an I-beam, where the non-uniform torsion can be represented by the
alternating bending of the flanges (sometimes called biflexion, Figure), one can proceed to bifilar modeling of
the element where the two footings are represented by two distinct elements, connected by transverse
elements representing the web. As a result, the specific bending of each flange, and thus the non-uniform
torsion, is represented. For instance, this is the case for double girder composite bridges. An application is
presented in chapter 3.
Decomposition of torsion into uniform and non-uniform parts: simplified biflexion hypothesis

2D or 3D elements

The linear elastic analysis of two- or three-dimensional steel elements does not present any specific problems, so the
general rules apply.

On the other hand, a non-linear analysis may be required to study instability phenomena. Indeed, structures composed
of steel plates, whether flat or curved, are subject to buckling phenomena.

In the case of flat elements, called plates, buckling is a relatively stable phenomenon: the initiation of buckling of the
plate does not lead to failure, the maximum load is reached after buckling. This is called post-critical behavior.

In the case of curved elements (shells), instability leads to the immediate and often brutal failure of the structure.
From a numerical point of view, in the non-linear static analysis, this leads to a rapid decrease of the load beyond the
maximum.

In both cases, the maximum load is strongly dependent on the initial deformation applied, in amplitude and shape. The
amplitude is fixed by standards. The shape is usually chosen affine to the first mode of instability. However, this choice
is not necessarily the most detrimental. It is advised to supplement it with local modes when the structure presents
panels of strongly different dimensions.

For example, in the case of an orthotropic deck, the deformation affine to the global buckling mode must be
supplemented by deformations affine to the buckling of the under panels.

In the case of shell-type structures, the problem is even more critical. It is advised, once a first calculation has been
performed according to the above assumptions, to adopt in a second calculation a shape of initial deformation affine to
the deformation obtained at failure.

5. Compound structures: steel-concrete

The alternative of a compound steel-concrete construction is sometimes preferred for certain types of industrial
buildings and bridges with small to medium spans (midspan < 100 m). The combination of these two materials by
making them "work" in their strength domains (concrete in compression and steel in tension) can result in a strong and
lightweight design. To achieve this result, the connection between these two materials must be correctly designed. A
distinction is made between:

mixed slabs: solid slab + reinforcement,

mixed beams: solid or mixed slab + steel profile + connectors,
mixed columns: steel profile + (concrete filling or concrete coating).

Overall structural analyses are usually carried out in the elastic domain by homogenizing the section, or by
representing the two materials separately. This second way of proceeding can lead to difficulties in the processing of
the results. Indeed, it requires recalculating the stresses on the globalized section to apply the current standards.

When more detailed analyses are needed, given the diversity of materials involved in this type of combination both
geometrically and in terms of non-linear behaviors, 3D finite element models are necessary to conduct local analyses,
which include the processing of the various interfaces (using contact finite elements for example). For studies at the
scale of the structure or the elements, relatively high-performance 2D models have been developed over the last
decade, such as those based on a fiber cross-section cut-out (fiber model) to enable the section stiffness to be
estimated by numerical integration.

The assembly of composite bridge girders is another relatively complex detail. Regardless of the assembly solution
chosen (butt-strap joint, joist trimmer, or diaphragm), 3D models using solid finite elements are preferred to simplified
2D models, but they are more computationally expensive.

Cracking - Composite girders are usually made of a steel section that is connected to a reinforced concrete floor or
deck by means of connectors. As a result, in areas undergoing a positive moment, where the slab is compressed,
strength and stiffness are particularly consequential, whereas in areas undergoing a negative moment, cracking
results in significantly lower mechanical characteristics. This cannot be neglected in the overall structural analysis.
Different levels of modeling are allowed in the codes:

Flat-rate approaches: for example, Eurocode 4 recommends considering a crack length equal to 15% of the
span, on either side of the supports. It also suggests a flat-rate stiffness for composite columns. Eurocode 8
recommends adopting an average stiffness over the whole length of the beam. These differences in approach
are justified by the different shapes of the moment diagrams under classical solicitations, mainly
gravitational, and seismic stresses,
Approaches defining a cracking zone by analysis of the stress envelopes: Eurocode 4 recommends to consider
 as cracked any section where the stress exceeds twice the average tensile strength under the envelope of
the characteristic stresses calculated, assuming that the structure is not cracked by adopting a long-term
concrete modulus.
It is also possible to use non-linear analyses.

Connection

Except in the case of non-linear analysis, it is not useful to model the connection. Current standards allow us to
consider the effect of a partial connection on the strength of the elements.

For modeling purposes, beam-type finite elements (2D or 3D) are generally used to model the point connection. If a
distributed connection hypothesis is considered, appropriate models are available in the literature.

Collaborative widths

Steel beams are connected to particularly wide members, and shear dragging can result in a non-uniform stress
distribution across the width of the slab.

In wireframe models, this phenomenon is usually taken into account by modeling a slab of reduced width, at constant
stress, since it is required for beam models.

Strictly speaking, since shear dragging is related to the transmission of shear forces through the connection to the
slab, it is therefore dependent on the shape of the moment diagram. The collaborating width should vary from one
combination to another. However, the standards allow a single width to be adopted for all calculations.

The variability of collaborative widths will be considered if the slab is modeled using surface type elements.

Delayed effects and shrinkage

Since the delayed effects of concrete influence the stiffnesses and stress distribution throughout the structure, they
must be taken into account. It should be noted that the Eurocode adjusts the value of the steel-concrete equivalence
coefficient according to the type of loading. The shrinkage solicitation induces a self-balanced stress distribution
throughout the section; thus it should also be taken into account.

Non-linear analyses

Non-linear computational models of composite structures assume the material and geometric irregularities hypotheses
used when modeling concrete and steel materials separately.

As mentioned earlier, the connection must be modeled with its own stiffness. However, it should be noted that its
numerical processing remains difficult. Usually, it is assumed that the concrete and steel parts cannot be separated
transversally and that only longitudinal sliding is possible. The finite element formulation of such a connection can lead
to the "locking" phenomenon when the sliding is blocked by the looseness of the transverse stiffness projection
preventing the uplift when the equilibrium is reached in the deformed configuration. Therefore, when this type of
element is used, it is advised to verify the consistency of the reaction forces in the connection with the
tensile/compressive forces of the steel elements.

6. Braces, tension cables, and suspension cables

Introduction to the calculations - Some cable elements can be modeled as bars of the equivalent section. This is
the case for vertical tension cables, or prestressing cables, guided in their duct. The curvature of these cables at
equilibrium is practically independent of their linear mass. In other cases, the cables tend to be bent by their own
weight: the vertical direction is a specific direction that plays a role in the stiffness of these elements.

When, at the scale of the cable's span, its curvature becomes important, it strongly influences the stresses and strains
it can transmit to the rest of the structure. As an example, for braces, the stiffness of the cable depends mainly on its
deflection and elongation. Deflection affects the stiffness of the cable; thus, the tension/deflection relationship of the
cable is inherently non-linear.

The cable elements present in most codes are based on the [Gimsing] model, which assumes that the cables deform
as a parabolic curve, an assumption that is valid if the deflection/span ratio is less than 1/12. This model decouples the
elastic elongation from the bending of the rope.

The catenary effect - When the deflection of a rope is greater than 1/12th of the span, it is no longer possible to
decouple the elastic elongation from the bending because the tension lever arm becomes the dominant factor in the
expression of the bending moment. Several computational models suggest catenary elements to account for this
geometrical effect. These elements connect two nodes of the mesh so that a single element is sufficient to model the
wire rope.
Nodal displacements of a catenary element

For this type of element, the nodal unknowns are the vertical displacement (in the direction of gravity) and the
horizontal displacement (see figure 4). The associated reactions are the variation of the horizontal component of the
cable tension, and the vertical reaction at the anchors. Since the relationship between nodal displacements and these
reactions depend on the cable tension, finding the equilibrium becomes a non-linear problem, although the structure is
globally elastic.
$translationBooks

PART 1 – THEORETICAL ELEMENTS

D3. The different construction

phases
D3. The different construction phases
Studying the phases of construction has two objectives:

to ensure the stability of the structure in the various transitional states leading to the final state,
to calculate the effects that have the assembly on the force distribution and the deformation of the structure.

There are various effects related to assembling. They are due to:

the evolution of the static state during construction. For example, a bridge span built by two cantilevered
bridge spans and finally clamped/keyed in its center will present, just after the clamping, a moment diagram
under permanent loads that cancels out in the middle of the span, which is quite different from the one that
would have been obtained without taking into account the assembly method (Figure 5),
the interaction of the delayed effects with the evolution of the static state. In the above example, after
keying, the moment in the middle of the span will increase due to creep,
the evolution of the sections over time. For example, in the case of composite steel-concrete structures, if the
structure does not rest on arches when the concrete is poured, the weight of the slab is supported by the
steel structure alone,
voluntary adjustments of the structure: support elevations, adjustments of the bracings, keystone actuators,
...

Taking these effects into account can be relatively complex and, in the most difficult cases, it may be essential to use
software that can model the evolution of the structure step by step.

However, it is often possible to proceed by superimposing various linear analyses.

Effect of the construction phases in the case of the keying of a cantilevered bridge span

The main issues are related to the effects of the delayed deformation of concrete. Indeed, how can one evaluate the
final state midspan moment in the above example? In the case of a single keying point, it is possible to use the
method known as the "coefficients method" (Figure 6). This approach is based on the following arguments:

Final state = (E(t0, t1)/E(t0, t∞)) x Final state not keyed + (1-(E(t0, t1)/E(t0, t∞))) x Final state without phasing.

with t0 the duration of load application, t1 the duration of the keying, t∞ the time considered for the final state, and
E(t0, t1) the concrete modulus for obtaining the deformation of the concrete at time t1 for a load applied at t0.
This method, in the case of a single keying point, outputs the exact theoretical final state. However, it is difficult to
extend it to the case of multiple keying points and can lead to absurd results.

The unitary case used for the coefficients method

It is best to externalize the effects of changes in the static state as follows (Figure 7) :

case 1: calculation of the final state, if the keying was not carried out ;
case 2: calculation just before keying, with the adequate concrete modulus;
case 3: calculation of the effect of keying: apply an imposed displacement to the structure at the keying point
returning the value of the discontinuity (for this example, in rotation) to the value fixed by the keying process,
the final state is the sum of cases 1 and 3.

This technique can be extended more easily to cases where the static configuration of the structures is modified many
times.

Unitary cases used for the superposition method

It is worth remembering that, under the effects of delayed deformation, concrete reacts with an apparent deformation
modulus called relaxation modulus, which is lower than the corresponding creep modulus. If the classical ratio between
the moduli of steel and concrete, including creep effects, is about 18, in the case of an imposed displacement, this
ratio increases up to 24. This tends to make the adjustments by support elevation and actuators less efficient when
the latter leads to imposing a deformation on the structure.

From the point of view of FE simulations, the large number of intermediate states that need to be processed multiplies
the risk of errors. The verifications must concern:

the respect of the boundary conditions in the intermediate phases;

the respect of the displacements fixed by the construction phases in the structure (for example, discontinuity
of slope at the keying point).

In construction studies, one must also remember that the actual creep of concrete can deviate strongly from the
theoretical formulations. The model must therefore be constructed such that it is easily adaptable to restore the
deformations appearing in the first phases, and thus improve the prediction of the following ones.
$translationBooks

PART 1 – THEORETICAL ELEMENTS

Chapter E. Typical post-treatment

in Civil Engineering
Chapter E. Typical post-treatment in Civil Engineering
Introduction

In civil engineering there is a great variety of quantities of interest because of the numerous subjects that can be
studied:

The nature of the construction considered and its functionalities (for example the sealing of dams, water
tanks, or nuclear power plants, crack opening, stress state, residual deformations, …)
The limit states considered (ultimate or serviceability, …)
The nature of loading (dynamic, static, delayed, …)
The constituent structural elements (reinforced concrete, prestressed concrete, steel, wood, masonry, …)

Moreover, the quantities of interest may be directly or indirectly accessible when performing FE calculations. They can
be obtained from primary results later processed to provide a given quantity of interest. The latter is then compared to
some criterion or used as data for the next steps (reinforcement area, rebar orientation, …).

The Eurocodes enables us to know the quantities that should be analyzed according to the various situations
mentioned above. However, the way to obtain these values is not established in the codes and standards. This chapter
highlights the maneuvers to access those quantities and the mistakes to avoid.

E.1 Generalities

E.1 Generalities

E.2 The quantities of interest in dynamics

E.2 Quantities of interest in dynamics

E.3 The particularities of reinforced concrete

E.3 Particularities of the reinforced concrete

$translationBooks

PART 1 – THEORETICAL ELEMENTS

E1. Generalities
E1. Generalities
The different post-treated quantities may change significantly according to the objective of the finite element analysis.

Usually the studied quantities in the pre-dimensioning phase (a sketch or preliminary draft) are different from those
searched in the pre-design and design phases. Other quantities of interest are needed when analyzing existing
structures to quantify structural risks.

It is tempting to consider:

Pre-dimensioning phases (sketch, preliminary draft, …): quantities such as displacements, deflection relative
to the span, …
Dimensioning phase: density of reinforcement for slabs, reinforcement area for beams and columns, crack
opening, …
Structural analysis: stress state (principal stresses in concrete: directions, signs, tri-axiality and its values,
comparison with criteria such as Ottosen, Rankine, Drucker-Prager if the analysis is elastic), principal strains:
directions, signs and its values, damage, stresses in the rebars, plastic strains, …

It is important to keep in mind that not all quantities of interest are accessible for every type of element and
constitutive models. The type of element determines the nature of the degrees of freedom: the nodes and moments on
solid elements are not accessible. Analogously, a model simulating plasticity cannot provide damage results. These
details may seem trivial, but they are often forgotten by the users, so it is important to mention it.

1. Stresses and strains

As exposed in the “Generalities” chapter, the stress and strain fields are not obtained directly when solving a
mechanical problem with a FE solver. They are obtained from the displacements by computing the derivative of the
interpolated displacement fields. Using the same notation as in the “Generalities” chapter:

In linear-elasticity, the strain-stress relation can be written as (H being Hooke’s matrix):

In this kind of procedure, the strain and stress fields evolve according to the interpolation functions of degree inferior
to one of the displacement functions.

For non-linear analyses, the stresses and strains cannot be obtained by simply multiplying Hooke’s matrix by the
derivative of the nodal displacements. They are based on the values of the element at the Gauss points, and the shape
functions are used to extrapolate the results to the rest of the element and nodes. This procedure allows the stresses
to evolve according to the same degree as the interpolation functions and not their derivatives.

As an example, on an element with n nodes, npg Gauss points located at the coordinates ξnpg with shape functions Ni,
we compute the least-square minimization between the interpolated field evaluated from the nodal values searched
and the known Gaussian values.

Building elementary nodal values starting from the values at the Gauss points in 1D

So, either it is a matter of minimizing the function:

Or for each node i among the n nodes of the element:

Which can be expressed in matrix form. The matrices are finally computed for all the isoparametric elements of
reference:

And so, it is possible to directly obtain the nodal stresses:

Thus, it is possible to have several nodal values of strains or stresses for a given node, shared by multiple elements.
The other nodal values can be deduced from those values.

Discontinuity of the elementary nodal values in 1D

Several methods can be used to recover the continuity of strains or stresses (if the material is the same in the 2
adjacent elements) to have a single nodal value of strain (or stress). Those methods consist of smoothening the stress
and strain fields by using certain algorithms on the totality of the structure (by least-square minimization). It can also
be done by computing the average of the values on a given node shared by multiple elements as shown in the figure
below.

Example of the calculation of an averaged nodal value starting from elementary nodal values in 1D

For convenience one might prefer obtaining smooth elementary values. However, when we look for precision, it is
convenient to give priority to the values obtained at the Gauss points.
 2. Going from continuum mechanics to structural mechanics stresses

The automatic methods to compute reinforcement rely on algorithms that take into account combined axial forces and
bending moments based on plate and shell torsors with 8 components, or deviated combined axial forces and bending
moments based on beam torsors with 6 components.

Some structures may need, for various reasons, to be modeled using solid finite elements (dam, structures such as
prestressed tanks: liquefied natural gas tanks, retaining structures). To avoid computations based on 8-component
torsors, it is necessary to rebuild the Structural Mechanics internal forces starting from the stress fields along the
segment as shown in the figure below.

Segment along which the stresses are used as a reference to rebuilding a shell type torsor

We consider the stress tensor expressed along the segment shown in the previous figure:

To be accurate, for thin plates satisfying the Kirchhoff-Love hypothesis, we consider the following torsor with 10
components that will be associated with the stresses according to the following equivalence principle:

This continuous integration is discrete along the segment and can typically be computed from the stress values
expressed at the Gauss points or the nodes.

Multiple integration techniques can be used (Trapezoidal rule, Gauss, Lobatto, etc …)

A continuous expression will be kept for the sake of simplicity, even if it is not rigorous in the discrete context of the
values obtained by the FE method.
Taking into consideration the Cauchy theorem equivalent to the shear stresses at the adjacent faces, this torsor is
reduced to 8 components:

Considering that this plate is an isotropic slab implies that the contribution of membrane efforts is neglected.

Then, it becomes a 5 component torsor (considering that the notation M_x represents the bending moment activating
the rebars in the direction e_x), which are the internal mechanical actions in a plate element:

(1) Shear efforts outside of the plane (z-direction) with x as the normal direction

(2) Shear efforts outside of the plane (z-direction) with y as the normal direction

(3) Bending moment activating the rebars in the x-direction

(4) Bending moment activating the rebars in the y-direction

(5) Torsion moment on the cross-section of the slab with x or y as the normal direction

3. Section method: elementary reduction elements (EF) to beam-type structural reduction elements

Depending on the stress state it might be convenient to build a model of the whole structure with plate elements
including the zones where these are not appropriate such as posts and lintels.

In these zones, one must restore a beam-type torsor to take into consideration more accurately the local behavior, in
particular, to estimate the required reinforcement.

A widely used technique to solve this problem consists of “cutting” the structure fictively and estimating the efforts at
the cross-section.

This “cut” must be chosen wisely to respect the Euler-Bernoulli hypothesis stating that cross-sections must remain
plane. It is usually the case for posts and lintels.

Example of cross-sections in a lintel or post passing or not through nodes

We then build in the local coordinate system of the cross-section the beam-type reduction elements substituting the
shell-type reduction elements. The latter can be the efforts at the nodes or evaluated at any point of the element
coinciding with the section line and analyzed utilizing elementary form functions.

For the example illustrated above, it can be written for the cutting line, where local and global coordinate systems
coincide, that:

(1) Axial forces at the cross-section of the associated beam, i.e. the “cut plane”

(2) Shear forces on the plane

(3) Shear forces off the plane

(4) Bending moment on the plane

(5) Bending moment off the plane

(6) Torsion moment

$translationBooks

PART 1 – THEORETICAL ELEMENTS

E2. Quantities in Dynamics

E2. Quantities in Dynamics
1. Time-dependent analyses

Post-processing of time-dependent quantities does not present any major difficulty as long as one is not interested in
targeting a precise instant or in characterizing a single representative value at all times of the analysis.

For cases where the loading is controlled and not random, a single analysis may be enough. If it is preferred to extract
a scalar quantity representative at all times of the analysis, the choice of the standard is the responsibility of the
engineer who analyses the results, and it must be justified. A statistical characterization may also be relevant (typically
expressed as a percentage).

If the engineer is interested in extracting a set of quantities constituting a vector, the problem of the concomitance of
the normalized quantities (absolute, maximum, minimum values, etc.) arises. The time t at which a bending moment is
maximal in a point for a torsor does not necessarily coincide with the point of maximal shear stress.

When the loading has a random character, it is important to multiply the number of calculation cases by integrating
the random character of the loading. As an example, if we consider N cases of calculations and we construct for a
quantity of interest gi(t) the corresponding dimensioning value Gi for a case i (it can be the maximum positive value
and also the maximum negative value to then incorporate them into an approach of actions concomitance). We
consider:

The used dimensioning value, which could be:

Its average:

Its standard deviation:

The estimation of the mean value of G for the N results:

With λ(N) calculated from the Student variable for N samples (calculation cases) for a 95% confidence level (one-
tailed) such that:

The table below provides estimations of the mean for various calculation cases.
 2. Seismic spectral analysis

The seismic spectral analyses provide representative quantities of the average of an extremum or period.

The purpose of this paragraph is to draw the reader's attention towards important precautions that must be taken to
avoid making significant errors in the calculation of the results coming from a quadratic accumulation.

As mentioned in the dynamic chapter, it is possible to quadratically cumulate simply or completely the spectral
responses of each mode. The result of these cumulations is a positive value.

It is important to consider that any operation that one wishes to perform on a quantity of interest that is the subject of
a quadratic cumulation (simple or complete) of the spectral responses of each mode must be performed before the
cumulation.

Let us take a simple illustration with the difference between the displacement of two nodes A and B cumulated simply
on modes 1 and 2 (an illustration with a CQC combination would just be more troublesome to write using cross-terms):

Strictly speaking, the difference evoked as an example can be written:

Therefore, it is obvious that estimating the difference afterward is largely erroneous:

The first value is always positive, but it integrates algebraic differences of quantities in the same mode. The second
expression might lead to a drastic underestimation of an extremum response. It can also be demonstrated by using
the Cauchy-Schwartz inequality that the first estimate is the upper bound of the absolute value of the erroneous
calculation presented in the second formula.

A more complex illustration can also be used with the estimation of a von Mises stress based on the main stresses for
2 modes :

An ‘a posteriori’ estimation of the accumulation of the von Mises stresses is largely erroneous:
Example of a specific application: estimation of the opening of a joint between two buildings

In seismic engineering, it is necessary to estimate the opening of a joint between two buildings under seismic
conditions to ensure that there is no risk of collision. The accepted practice is to independently calculate the extreme
values of displacement of the envelope of each building based on a quadratic accumulation and then to evaluate the
maximum opening of the joint by calculating the difference between these two positive values. Thus, an unfavorable
phase opposition of the maximum responses considered.
$translationBooks

PART 1 – THEORETICAL ELEMENTS

E3. The specific case of reinforced

concrete
E3. The specific case of reinforced concrete
1. Overview of the common methods for designing the reinforcement of plate elements

There are mainly 3 methodologies applicable to plates that can be consulted using the following references:

Johansen fracture line method:

Save M.A., Massonet C.E., De Saxcé G. - Plastic limit analysis of plates, shells and disks, Applied
Mathematics and Mechanics Vol. 43.
This method applies to weakly reinforced plates (reinforcement ratio less than 0.7%), allowing the
appearance of fracture lines to be evaluated and the use of the reinforcement opposing those
fracture lines.

Sandwich method:
Marti, P., Design of concrete slabs for transverse shear, ACI Structural Journal, Vol. 87, pages 180 to
190.
This method intrinsically considers the behavior of plates, it can be found in annexes LL and MM of
Eurocode 2 part 2,

Method of Capra and Maury:

Calcul automatique du ferraillage optimal des plaques et coques en béton armé, Annales de l'Institut
Technique du Bâtiment et des Travaux Publics, n°367, December 1978,
This method is based on the equilibrium of the cross-section in combined bending and axial forces, it
is detailed in the following section.

2. Example of the Capra-Maury method

A set of facets is defined, centered at the point of computation of the finite element code. This can be a node, a Gauss
point, or any point where forces are interpolated.

Its normal vector rotates from this point in the plane tangent to the mean sheet. The facet is marked by the angle θ
between the OX axis and the normal vector of the element in its local coordinate system (see figure 2.1-a). The angle θ
is discretized regularly from -90º to +90º (here with a step of 5º). The Ox and Oy axes are the axes of the
reinforcement layers.

Reference facet parallel to the beam’s cross-section in equilibrium subjected to combined bending and axial force

For each of these facets, the bending moment (M), the membrane force (N), and the shear force (V) are applied as a
function of the effort tensors using the following equations:

By using combined bending calculations, it is possible to determine the reinforcement area in upper and lower layers
AS(θ) and AI(θ) required in the θ direction to reach equilibrium in the chosen reinforcement concrete section.

The compressive strengths in the θ direction of the two reinforcement layers can be evaluated using the following
expressions:
where fyd is the maximum allowable stress of the steel (identical in both directions).
The strength is ensured if the resistive force is greater than the applied force, which is expressed as:

Thus, by considering an orthonormal coordinate system with AXS on the abscissa and AYS on the ordinate, one can
finally solve for the upper and lower reinforcement:

for all angles θ

for all angles θ

and

and minimum.

The resistance inequalities define for each θ a half-space limited by a straight line with negative slopes that reflects a
validity domain, as shown in the following figure.

Resistance domain for a facet θ

By browsing through all the values of θ, one can obtain the validity domain shown in the following figure, delimited by
the line ABCD.
 Resistance domain for all facets

For each point P in the validity domain, the total area of reinforcement can be obtained by projecting the point P in Q

onto the first bisector. The distance OQ then represents the value with AS = AXS + AYS.

Therefore, it can be noticed that the optimum reinforcement corresponds to one of the 36 points (considering the facet
rotation step chosen if one facet is taken every 10 degrees for example) of the boundary (illustrated by the 4 points of
the line ABCD ...) whose projection on the first bisector is closest to the origin of the axes. Researching this point can
be done using "dichotomy" type methods.

3. Connecting rod and tie-rod methods based on a finite element calculation result

In the presence of structural elements subjected to important point loads or presenting abrupt modifications of their
section and geometry, the classical plane section methods of analysis are not satisfactory. These locations are
generally reinforced using good practice rules based on experience or empirical guidelines. The strut and tie method
(STM) is a rational design procedure for complex structural elements. The procedure is based on mechanics but is
simple enough to be easily applied in design.

In general, the STM involves the idealization of a complex structural element into a simple structure capable of
representing the path of stresses within the element.

The truss model consists of struts representing the compression fields in the concrete, ties representing the elastic
steel reinforcement, and nodes representing the localized areas where the components interconnect or the areas
where the elastic steel is anchored in the concrete. The struts and ties can carry only uniaxial forces. This mechanism
must be stable and maintain the equilibrium with the applied loads.

The failure of the structure is dictated by the failure of one or more ties or by excessive compressive stresses within
the struts or nodes. Ideally, only the first mode of failure should occur.

Example of D region and system modeling to obtain stress fields and then loads at the ST truss supports

This method is applied to the so-called D regions. To characterize these regions, the distribution of deformations over
the depth of the element is considered to be non-linear.

Consequently, the underlying hypotheses about the cross-section procedure are not valid. The principle of equivalence
cannot be applied here.

According to Saint Venant's principle, an elastic stress analysis indicates that stresses due to axial forces and bending
moments come near a linear distribution at a distance approximately equal to the depth of the element, h, away from
the discontinuities.

In other words, a non-linear stress distribution exists in the depth of a member from the location where the
discontinuity is introduced. Then, it can be stated that the D regions are assumed to extend to a distance h from the
applied load and support reactions. In general, a region of a structural member is assumed to be dominated by a non-
linear profile, or a D region, when the ratio extent/depth (a / h) is smaller than 2 or 2.5. The shear extent (a) is defined
as the distance between the applied load and the nearest support in simple elements.

The approach consisting in defining a strut-and-tie truss can be summarized in Figure 10.
The general principle of design by STM

The verification methods are expressed in Eurocode 2 part 1-1.

The approach presented in Figure 10 is difficult to implement and depends strongly on the engineer who implements
it.

More and more approaches are being developed to automate this procedure. In France, the algorithm with code_aster
elements integrates a CALC_BT operator that renders the procedure semi-automatic based on:

an analysis of the local peak fields of the major and minor main stress fields,
a cutout of the D region modeled by Voronoi paving,
the projection of the mean directions of the main stresses in the Voronoi paving stones.
a set of optimization processes.

Reference example on the left - Solution obtained automatically by the CALC_BT operator of code_aster

This method requires a high level of experience and knowledge of the field.
$translationBooks

PART 1 – THEORETICAL ELEMENTS

Chapter F – Geotechnical
calculations
Chapter F – Geotechnical calculations
Introduction

Conventional methods for the design and dimensioning of geotechnical structures mainly aim to analyze the resistance
to failure of an isolated structure. These analytical or semi-analytical methods only consider very simple geometries
and provide little or no information on the ground deformations surrounding the structures.

The increasingly intensive use of urban spaces and soils, occupied by a variety of structures close to each other,
makes it necessary to control the interactions between structures. The designer of a structure must justify that the
displacements induced by its construction remain below the threshold set by the client. Traditional methods do not
meet this requirement, which explains the increasingly frequent use of numerical modeling and software dedicated to
geotechnics that are adapted to design offices. Specifically, digital modeling is used in two different situations:

during the design phase, to justify the dimensions when traditional methods are difficult or impossible to
implement,
as an expertise tool, to study the behavior of a damaged structure, to identify the phenomena responsible for
a specific issue, and to justify the use of a strengthening method.

Depending on the case, models with various levels of complexity can be used, considering the uncertainty of the
natural soil's behavior and their spatial variability. Consequently, it becomes difficult to choose a constitutive law and
determine its parameters. The user often has to choose between a robust model whose operation is well understood,
but which does not account for all the complexity of soil behavior and a model that is potentially more accurate to
describe the actual behavior of soils, but which includes many parameters whose roles are sometimes difficult to
identify and quantify.

The philosophy behind the design rules promotes the first approach, to use simple and robust models and ensure the
safety of the design by inputting appropriate factors and carrying out parametric studies to estimate the sensitivity of
the results with respect to the said parameters. However, this approach may lead to an oversimplification of the
problem and can lead to overly conservative and unnecessarily costly designs.

The other approach is to use constitutive models that attempt to better represent different aspects of soil behavior. A
very large number of models have been proposed, but their practical use remains difficult. Furthermore, the idea of
having one "universal" model that describes all phenomena is an illusion and should be abandoned. It is necessary to
define the limits of the model chosen: a perfectly plastic elastic model or one considering isotropic strain-hardening,
even with a complex load surface and strain-hardening law, cannot predict a progressive accumulation of deformation
in a soil mass.

Therefore, the use of finite elements requires the user to have a critical perspective on the soil constitutive model.
Though, this is not the only aspect to consider: the result of the calculations can depend essentially on the 3D
geometry of the structure and the geometry of the soil layers. It is sometimes (not always) justified to simplify the
representation of the soil behavior and to focus on the representation of the construction process. In any case, the
results of numerical simulations should be carefully verified.

Finally, it is clear that geotechnical structures are made of soil, in contact with soil, or totally or partially buried; thus,
numerical modeling must represent the mechanical interaction between the soil and the structure.

This chapter is organized as follows: it summarizes the main features of numerical calculations in geotechnics before
briefly presenting the verification principles adopted in Eurocode 7. Once these elements are presented,
recommendations are suggested for good practice using FE calculations in geotechnics. It also deals with the features
of dynamic calculations.

F.1 Geometrical aspects

F.1 Geometric aspects

F.2 Material non-linearities

F.2 Material non-linearities

F.3 Soil-structure interactions

F.3 Soil-structure interactions

F.4 Hydraulic effects

F.4 Hydraulic effects

F.5 Uncertainties and recommendations

F.5 Uncertainties and recommendations

F.6 Normative aspects: Principles of Eurocode 7

F.6 Normative aspects: Principles of Eurocode 7

F.7 Dynamic modeling

F.7 Dynamic modeling

F.8 Characteristic scales

F.8 Characteristic scales

Conclusions and references

This chapter presents an overview of the issues raised specifically by FE calculations in geotechnics, in statics or
dynamics. Specific precautions should be taken when performing the calculations concerning the meshing, boundary
conditions, phasing, the choice of constitutive models, and the setting of parameters.

With a certain amount of practice and the constant concern to control the obtained results, a wide range of problems
and structures can be dealt with. However, there are still situations that are difficult and/or costly to deal with
numerically.

Firstly, difficulties may arise due to the ambiguous mathematical structure of the studied problems. In dynamics,
choosing a time discretization coherent with all the input data remains difficult. In statics, managing several non-
linearities of different natures (unilateral contact, damage) may lead to numerical issues hard to overcome.

One should also remember that the physics of certain phenomena remain poorly understood and described, which is
reflected in the numerical analysis. It is the case for various problems such as the initiation and evolution of landslides,
implementation conditions of tasks such as the compaction of backfill materials, the initial state of stresses, water
content, and other parameters linked to the history of the structural elements, etc.
$translationBooks

PART 1 – THEORETICAL ELEMENTS

F1. Geometric aspects

F1. Geometric aspects
One of the advantages of the FE method is the possibility of describing the exact geometry of the structures even
during the various construction phases. CAD-like pre-processing tools make it easy to generate very complex
geometries. One of the features of geotechnical structures is that, generally, the earth mass that constitutes or
surrounds the structure must be considered within the mesh.

1. Boundaries of the studied domain

The first challenge is to identify the boundaries of the domain considered for the analysis. For a geotechnical structure,
the horizontal and the lower boundaries of the studied domain are rarely determined with precision: the extent of the
domain is then bounded by vertical planes whose position is generally fixed by using empirical rules.

Using the plane strain assumption, for example, the position at which the lower limit of the mesh is set has a direct
impact on the calculated settlement for a strip footing or above a tunnel. This relation is clear in the case of a
homogeneous linear elastic soil mass. It can be reduced by taking into account elastic moduli, which increases with
depth. Nevertheless, it is likely to induce a significant error in the calculated displacements. The ideal case is when a
rigid bedrock has been found at a given depth, which assumes that the research has been carried out to a sufficient
depth. To model a tunnel, for instance, it is advised to collect soil sampling way beyond the depth of the shaft, which is
generally not the case in real projects.

In the lateral directions, considering a domain that is too small can also significantly modify the response of the
numerical model. Fixed displacements lead to an overestimation of the stiffness of the solid, and "smooth contact"
type conditions lead to an overestimation of the displacements. The choice of the mesh dimensions adapted to a
structure remains a largely open problem even if some authors have suggested practical rules. However, those rules
should not be taken as absolute prescriptions (see Mestat and Prat, 1999).

Consequently, choosing the size of the domain considered for the mesh is an important step in the modeling of
geotechnical structures, even for relatively simple static analyses. In the case of dynamic calculations, the question of
the size of the meshing domain raises specific issues and is a crucial part of the modeling strategy. This will be
detailed in section 8.

2. Soil heterogeneities

In some regions, such as London or Frankfurt, geology allows the soil in the vicinity of the structure to be considered
homogeneous (in the sense that its mechanical and hydraulic behaviors can be represented by a single model).
However, in other contexts, particularly in the Paris region, it is common for the studied area to include several layers
of soil of very different natures and characteristics (particularly mechanical). Therefore, the elaboration of a
computational model begins, as for traditional methods, with a detailed study of the soil layers in the area of interest.
The goal is not to reproduce the exact geometry of the geological layers (which may be locally thin), but to define
geotechnically homogeneous sets.

3. Discontinuities

An important particularity of geotechnical calculations is the presence, within the solid, of failure planes that existed
before the construction of the structure or the implementation of the studied load. They produce a discontinuity of
displacement between the blocks located on either side of the failure plane. The FE method is well adapted to search
for continuous displacement fields, and the consideration of this type of discontinuities requires the implementation of
special techniques (specific elements are generally used) or even the use of another calculation method such as the
discrete element method.

Behaviors such as landslides are extremely complicated to predict. Nowadays, it is extremely difficult to predict the
occurrence and development of failure planes. Usually, the engineer is forced to consider an existing plane whose
position has been identified with more or less precision using the adapted equipment (for example, inclinometers that
follow the deformation of a slope).

Solid rocks often contain a large number of fractures with directions that are seemingly parallel to one or two main
directions (local diaclases). The fracture distribution is diffuse, and random, or at least impossible to characterize
completely at the scale of the solid. If the global behavior of the solid is of interest, it can be modeled as a continuous
medium employing a constitutive model, which incorporates, at the calculation scale, the effects of the discontinuities:
homogenization methods are generally used.

One may also have to consider large discontinuities in a solid bedrock (fractures), which can be treated as the failure
planes of landslides.

4. An "open" material system and construction techniques to be modeled

As stated in the introductory chapter, a peculiarity of civil engineering calculations is the need to consider construction
phases such as clearing or backfilling, implementing a concrete mass, tensioning cables, etc. Taking these
construction phases into account within the framework of the FE method can be a lengthy and complicated process:
the method consists of reducing the problem to the calculation of a stiffness matrix, a vector of nodal forces, then
solving the system obtained by considering the boundary conditions. To model the construction phases, one must
perform a series of calculations taking into account the change in stiffness of some elements, disappearance or
modification of some supports, changes in the point of application of the loads, etc.

Therefore, the issue is to propose simulation techniques that can take into account a large number of constructive
arrangements within the relatively narrow framework of the FE method. It is up to the user to decide whether the
modeling tools proposed by the software correctly reflect the phenomena involved.

In the case of tunnels dug by TBM, the stresses applied onto the ground during the various digging phases are
complex as the soil is closing in on the ring formed by the segments as the TBM advances. Pile driving is another
example, which is difficult to simplify to the FE method framework.
$translationBooks

PART 1 – THEORETICAL ELEMENTS

F2. Material non-linearities

F2. Material non-linearities
In geotechnics, it is very rare to be able to study a structure exclusively using linear behavior assumptions (except for
certain dynamic analyses). It may be useful to first perform a linear calculation, to check that the geometry and
boundary conditions are correct and get a preliminary idea of the deformation that the loading may cause.
Nevertheless, the results should be put in perspective as they can be completely wrong in some cases: for an
excavation in front of a cast wall for example, the kinematics calculated with linear behavior are clearly different from
the one observed.

1. Constitutive laws

Even when limited to cases that are perfectly saturated or perfectly dry, the behavior of soils is complex. In practice,
elastoplastic models are most often used, which give a force-displacement relationship that is non-linear but
independent of time. The effects of creep and viscosity can be considered for particular applications when required -
and if the corresponding parameters can be obtained experimentally - but the use of such models remains limited.

Among the elastoplastic models, perfectly plastic linear elastic models are still most often used (see the survey cited
by Gilleron, 2016). The use of non-linear elastic models associated with one or more plastic collapsible mechanisms is
gradually becoming widespread, particularly under the influence of software developers: in some cases, they give
results that are much more representative of reality (for example, for excavation in front of a molded wall), but the
influence of each parameter of the models is not necessarily well controlled. Generally, the choice of a constitutive
model for soils must consider the objectives set for the calculation, the type of structure (and the type of stress to
which the soil will be subjected), the level of precision of the available reconnaissance and laboratory tests.

2. Initial state

For non-linear models, the stiffness of the material depends on the initial state of the stresses in the studied soil mass
(and possibly other parameters, such as strain-hardening or damage). Therefore, the determination of the initial
stresses has a decisive influence on the results. Unfortunately, the initial stresses are generally evaluated in a very
simple way: they are either assimilated to a "geostatic" stress field (for a solid with a horizontal surface), or obtained
by applying gravity to the whole mesh from a state of zero stress. These processes are relatively poor compared to the
complexity of the rheological models used for soils. However, they remain unavoidable in practice, for lack of a better
way of estimating initial stresses in the soil.
$translationBooks

PART 1 – THEORETICAL ELEMENTS

F3. Soil-structure interactions

F3. Soil-structure interactions
Geotechnical structures often combine layers of soil with metal or concrete structures, which are generally much stiffer
than soil. The interaction may be limited to a few points where the structure rests on the ground, or it may be
continuous over a significant surface, such as the top surface of a tunnel, or retaining wall.

The interaction will be treated more or less precisely depending on the case.

In the case of tunnels, for example, it is very common to consider a perfect bond between the ground and the vault
because of the construction method: with traditional (sequential) methods, the retaining wall is made of concrete
sprayed directly on the surface of the ground uncovered by the excavation, which in principle ensures good continuity
of movement. During tunneling performed by a TBM, efforts are made to ensure good force transmission between the
ground and the segments by carrying out backfilling injections into the space between the arch and the ground.
However, in older tunnels, the pathologies observed (or the tests carried out in situ) may suggest that contact is locally
lost between the vault and the ground. For instance, it can be due to water seepage that may have washed out the
ground: the modeling must then describe more precisely the contact conditions between the ground and the vault.

Covered trench tunnels present a different problem, as the ground around the structure is backfilled. The modeling of
this operation may require explicit consideration, using specific elements, of the interface between the vault and the
ground.

It is also common to introduce explicit modeling of the interface between the soil and the structure for retaining
structures, when backfilling behind a wall (the phenomenon of soil sliding at the interface with the structure being
similar to that involved in covered trenches), or when excavating in front of a cast wall, for example, the mass of soil
supported may slide and present a vertical displacement greater than the height of the wall.

The question of modeling an interface between soils and structures must be considered on a case-by-case basis. One
can introduce contact or interface elements specifically intended to represent the mechanical interaction between the
two, but these elements introduce new parameters, which can be difficult to identify (such as the normal and
tangential stiffnesses of the interface). This modeling approach presents a risk: the interface elements tend to control
the behavior of the structure and to blur the role of the soil behavior, giving the impression that the response of the
structure hardly depends on the soil anymore.

Reinforced structures

In many cases, the soil is reinforced by inclusions with very high stiffness and strength characteristics. These inclusions
are discreetly distributed in the soil and very slender: piles, micro piles, tie rods, wall reinforcement in reinforced soil.
This geometrical particularity poses various difficulties. First of all, strictly speaking, a row of piles is not equivalent to
a continuous wall, and the use of plane strain assumption is not justified. In practice, one would adopt, for the planar
calculations of the wall, mechanical characteristics "equivalent" to those of a row of piles, using assumptions that can
be more or less difficult to justify. The same applies to the parameters of the mechanical interface between the soil
and the piles/wall. The difficulty is the same if the wall is represented by surface elements or by linear beam-type
elements.

To overcome this difficulty, 3D modeling can be used. However, because of the dimensions of the cross-section of the
inclusions, whenever there are more than a few units, it becomes impossible to represent in the mesh the real
geometry of the inclusions: for a reinforced earth wall, with reinforcements of 5 mm x 45 mm section, at a rate of 4 to
6 reinforcements per 0.75 m x 0.75 m, and for a volume whose dimensions are of the order of ten meters, the number
of nodes of a mesh that would respect the real geometry of the inclusions and that would give an acceptable
discretization exceeds the current calculation capacities. We can therefore propose to represent the inclusions by 1D
elements (with or without considering bending effects). This approach is questionable from a theoretical point of view
because the introduction of a linear density of force exerted by the inclusion in a 3D medium is not compatible with
the classical representation of internal forces by a stress tensor. It can be used, however, one must be careful in the
interpretation of the results, at least with respect to the stresses in the vicinity of the inclusions.

An alternative solution is to adopt homogenization-type approaches to take into account the influence of inclusions on
the overall behavior of the structure. More or less complex models have been developed and implemented in some
software.

Whatever the choices made (calculation in plane strains or in 3D condition, discretization of inclusions - by linear or
non-linear elements - or homogenized approach), it is necessary to represent the mechanical interaction between the
pile and the soil that occurs at the contact between the soil and the sidewall of the pile, and also between the soil and
the footing of the pile. Modeling the mechanical interaction at the footing of the pile is particularly difficult to master.
Modeling a single pile by volume elements, possibly with interface elements with the surrounding soil, gives results
that depend on the mesh and the constitutive model used for the soil. It is necessary to use a model that reproduces
the soil failure in compression if one is interested in the pile failure. For instance, models such as Mohr-Coulomb or
Drucker Prager are not suitable in this context.

In some 2D or 3D models where inclusions are represented by bar or beam elements, a fictitious end (e.g. a horizontal
beam element perpendicular to the pile) is associated with the inclusion, in an attempt to better represent the
interaction at the footing of the pile: performing studies to determine the sensitivity of the results to the dimension of
the added elements seems appropriate to verify the relevance of this approach.

Finally, other modeling techniques are available, which propose to explicitly integrate an interaction model for lateral
friction and another one for peak interaction via ad hoc elements.

Without going more into detail, let us highlight the fact that the user is free to choose between numerical simulation
techniques and models, which have a direct influence on the obtained results.
$translationBooks

PART 1 – THEORETICAL ELEMENTS

F4. Hydraulic effects

F4. Hydraulic effects
1. Hydromechanical coupling

Another peculiarity of geotechnical calculations concerns the role of water in soils. When a mechanical load is rapidly
applied to a saturated layer of soil, instantaneous deformation, and pressurization of the fluid in the vicinity of the
applied load occurs. Depending on the hydraulic boundary conditions, the gradient of the hydraulic load causes the
fluid to be set in motion, leading to pressure redistribution and delayed deformation of the soil.

Therefore, a problem of hydromechanical coupling must be dealt with. A solid theoretical framework was established
by Biot (1941) and developed by Coussy (1991). In terms of numerical resolution, the coupled problem is much more
difficult to deal with than a classical problem, for several reasons:

the problem involves, in addition to displacements, a new unknown field, the water pressure field,
it is necessary to specify boundary conditions specific to the hydraulic problem (define the parts of the
contour that are impermeable and those where pressure is imposed),
the solution (in displacement and pressure) is time-dependent,
the mathematical nature of the problem to be solved is different,
the permeability of the different soil layers must be described quantitatively.

Complete processing of the hydromechanical coupling is rarely performed. One generally tries to limit oneself to a
decoupled approach to the problem, in which one calculates the evolution of the pressure field while neglecting the
deformations of the solid. However, this decoupling has the consequence of strongly underestimating the duration
over which the pressure redistribution occurs.

2. Unsaturated soils

The treatment of unsaturated soils complicates even more the problem as the transition from near-saturated to
unsaturated zones introduces additional unknowns, non-linearities, and parameters. Again, the complete treatment of
unsaturated soils remains rare. It is preferred to propose simplified solutions, neglecting partially saturated zones for
example.

Defining the initial state in the case of unsaturated soils is extremely complex because of the lack of experimental
methods to characterize it in situ.
$translationBooks

PART 1 – THEORETICAL ELEMENTS

F5. Uncertainties and

recommendations
F5. Uncertainties and recommendations
1. Uncertainties

There are many sources of uncertainty in geotechnics. The first is the relative lack of knowledge of the geometry of the
soil layers that make up the solid being studied. In addition to the geological knowledge about the area where the
structure stands, the main sources of information are the boreholes drilled on the project site. The extrapolation
between boreholes, especially if they are far apart and not located at the exact location of the future structure (which
may be modified after reconnaissance, for example in the case of tunnels), does not necessarily give an accurate
representation of local variations in layer thickness.

In urban areas, the presence of heterogeneities (cellars, wells, foundations of previous structures) is often difficult to
detect.

The other source of uncertainty already mentioned concerns the initial state of stresses (and possibly pore pressures)
in the soil mass. It can have a major influence on the results of the calculation: this is particularly clear in the case of
tunnels, where the loads considered depend on the initial stresses.

Finally, the choice of constitutive models and the determination of the parameters of these models introduce a
significant uncertainty on the representativeness of the calculations: if the constitutive model does not capture a
phenomenon that controls the behavior of the structure, the result may be qualitatively and quantitatively very far
from reality.

2. Recommendations

In general, the user must be aware of the objectives of the calculation he is undertaking: the approach is different
depending on whether one is trying to justify dimensions or to evaluate the influence of certain constructive provisions
(the number and position of the struts, for example).

One must also be aware of the modeling choices on which the calculation is based (even if these choices are
sometimes partly imposed by the software). One must be able to identify the phenomena to be taken into account,
which leads to the choice of a quasi-static or dynamic analysis, the consideration or not of the hydromechanical
coupling, etc.

One must choose between a 2D or 3D calculation. Three-dimensional calculations remain rare for the moment because
of the time required to prepare the calculations. However, for some problems, it is clear that two-dimensional
calculations can only give a poor indication of the behavior of the studied structure regardless of how long the
engineer took in determining the soil parameters. For instance, the study of the stability of the tunnel face cannot be
considered outside a three-dimensional context. The same is true for the study of bolt reinforcements of the tunnel
face. The development of pre-processors specific to each application should simplify the use of 3D calculations and
improve the representativeness of many finite element analyses.

In geotechnics, special focus should be given to the choice of soil parameters: it could be the subject of a whole book.
Most advanced constitutive models do not come without a detailed and robust parameter identification procedure,
mainly because the model equations cannot be solved even for a simple problem such as triaxial compression. One
must therefore calibrate the parameters so that the modeling of triaxial tests, for instance, outputs results in
satisfactory agreement with the test results. A trial and error procedure is used, and since the agreement obtained is
evaluated subjectively (because we can choose to better reproduce one part or another of the experimental curves), it
is not guaranteed that two users will obtain the same parameter values from the same tests. Thus, all constitutive
models do not have the same qualities. Some have many parameters, each of which influences a particular aspect of
the soil response (but which does not necessarily appear in the available test results). Other models, on the contrary,
have a relatively small number of parameters, but each of them can simultaneously modify several aspects of soil
deflection. It makes the calibration much more complex.

The last recommendation that must be kept in mind is to check the calculations as much as possible. There are not yet
any general tools to measure the quality of a calculation: research work aims to provide error estimators, but their use
in geotechnics remains rare. It is therefore necessary to look in-depth at the results: some inconsistencies are
sometimes easy to detect. If there are any doubts, it is useful to have the results checked by someone else. In any
case, it is highly recommended to carry out parametric studies to get an idea of the influence of certain factors, in
particular soil parameters, if it is not certain that their influence on the results is moderate and that their value has
been determined with acceptable accuracy.
$translationBooks

PART 1 – THEORETICAL ELEMENTS

F6. Normative aspects: Principles

of the Eurocode 7
F6. Normative aspects: Principles of the Eurocode 7
Numerical modeling always had a particular relationship with calculation standards, which essentially focus on the
verification of ultimate limit states by calculating safety coefficients or the equilibrium of forces integrating partial
coefficients. Indeed, numerical modeling provides first and foremost values of displacements and deformations and is
therefore a tool very well suited to the verification of serviceability limit states (SLS) for which the partial coefficients
are equal to 1 Its use for the verification of ultimate limit states (ULS) therefore initially appeared to be limited. From
now on, particularly with procedures aiming to reduce the shear properties of soils (for example, the "c-phi reduction"
procedure), it is easy to calculate a safety coefficient. It is also possible, through the procedures suggested by certain
calculation standards, in particular Eurocode 7, to consider the results of numerical modeling both for the verification
of the ultimate limit states and for that of the serviceability limit states.

However, Eurocode 7 in its current version is not very clear about how the analysis and processing of the results using
numerical modeling should be conducted. Indeed, the three calculation approaches proposed by Eurocode 7, which
allow applying partial coefficients on the actions, the effects of the actions (bending moments and shear forces in a
retaining wall, axial force in a pile, etc.), the properties of the soils (c and or cU), and the geotechnical resistances,
have been designed to be used with limit equilibrium methods. The verifications related to the application of the
Eurocodes are mainly transcribed in the form of comparisons between actions or effects of actions and resistances.
Nevertheless, several publications (Potts and Zdravkovic, 2012, Tschuchnigg et al., 2015, etc.) and reports of different
working groups convened for the development of the second generation of the Eurocodes make it possible to outline
verification procedures. First of all, it is important to underline that the weighting "at the source" of the soil or rock
properties, i.e. the possibility of reducing the cohesion and the angle of friction before carrying out the calculation, is
not allowed because it leads to results that cannot be interpreted. Different procedures can however be used and are
summarized in the table below.

Synthesis of the different types of ULS verifications

4 - Structural (and
Type of ULS 2 - Geotechnical (and
1 - Structural ULS 3 - Geotechnical ULS geotechnical) ULS
verifications structural) ULS

Estimation of the
resistance centered
Multiplication of the Reduction of the shear around a specific Increase of the loads applied to
Type of procedures effects of the actions properties of the soils geotechnical structure the geotechnical structure
by 1.35
(pile, tie rod, etc.)

Must be combined with A priori, this approach is

Must be combined Must be combined with the
the verification of sufficient on its own
Comments with the verification of verification of structural
structural ULS
geotechnical ULS ULS

The approach that tends to be used is to carry out a calculation with characteristic values for both the loads and the
properties of the soil to reach the first state of equilibrium. Concerning this state of equilibrium, two types of
verifications must be carried out. The first verification, namely type 1, is relative to the structural ULS and consists of
multiplying the effects of the calculated actions by 1.35 (i.e. the coefficient usually considered in the Eurocodes for
unfavorable permanent loads). The second verification, type 2, is related to geotechnical ULS and consists of
progressively reducing the shear resistance properties of the modeled soil to reveal a failure mechanism. The
reduction factor applied can be considered as a safety factor, but its interpretation can be subjected to a discussion.
Indeed, during the reduction of shear properties, different elements may interact such as strain-hardening functions or
flow rules. In the case of calculations associating volumetric elements and structural elements such as "bar" or "beam"
type elements, this verification can also be used to check structural ULS. The displacements accumulated during the
procedure of reduction of the shear resistance properties generate forces in the structural elements whose
interpretation is a matter of debate among the engineering researcher’s community. Therefore, the user must be
extremely careful about the way to implement this type of procedure. Figure 7.1 shows the sequence of type 1 and 3
verifications in the case of a phased calculation.

Geotechnical ULS can also be specifically apprehended by considering the resistance centered around the geotechnical
structure to be dimensioned, a pile, an anchor, etc. and comparing it to the mobilized resistance provided by the
standards, for instance, the pressuremeter norms for the calculation of the bearing capacity of piles: this is the type 3
procedure and it must be associated with the type 1 for structural ULS.

In certain configurations, specifically for piles or footings, it is also possible to increase the actions applied to the
geotechnical structure to be dimensioned to directly obtain the maximum applicable force and to deduce the safety
coefficient: this is the object of the type 4 approach.

More generally, the methods used to justify geotechnical ULS must allow either to identify the failure mechanism and
verify that there is a sufficient margin to trigger the mechanism (this is the object of types 2 and 4 verifications) or to
compare the mobilized resistance with a mobilizable resistance that can be calculated elsewhere (this is the purpose of
type 3 verifications).
 Sequencing of the ULS verifications in a phased computation

When the failure mechanism is associated with the action of water special precautions must be taken. Indeed, the
procedures described above do not enable us to correctly compute the safety coefficient if the failure mechanism
results exclusively from the action of water. Firstly, it is necessary to define whether the failure mechanism is related
to a mechanical equilibrium defect (for example, water rising too high behind a retaining wall), or to a hydraulic
problem (for example, a hydraulic gradient being too high). In the case of a mechanical equilibrium defect, it is
necessary to carry out a parametric study relative to the level of the water table to estimate its effect by implementing
the ULS verifications presented in the previous table. In the case of a hydraulic problem, it is possible to compare the
calculated gradient with a critical gradient calculated elsewhere and to verify that the effective stresses remain
positive at any point of the solid (because the condition on the critical gradient is more conservative than the condition
on the effective stress).
$translationBooks

PART 1 – THEORETICAL ELEMENTS

F7. Modeling in Dynamics

F7. Modeling in Dynamics

Because of the continuous and unconfined nature of geomaterials, the treatment of dynamic problems in "soil" media
differs from the treatment of dynamic problems associated with classical civil engineering structures. While for most of
the latter, mechanical modeling through the assembly of masses and discrete sources of stiffness seems sufficient to
understand their dynamic behavior, geomaterials must be handled as continuous and unconfined media. Their
dynamic response must be studied as a wave propagation mechanical problem.

Concerning the numerical methods used in soil dynamics, it should first be noted that modeling wave propagation
requires to use of a range of modeling and calculation methods, which differ from conventional geotechnical
calculations using the FE method. Although the FE method has an important place in the inventory of methods
available to the engineer for soil dynamics problems, it is often coupled with, or replaced by, other numerical
computation methods that are better suited to model wave propagation in unbounded environments.

Additionally, the equations that describe dynamic problems involve a new variable (apart from spatial variables): time.
The computational algorithms are formulated in the time domain, and alternatively, in the frequency domain. The
methods used to solve soil dynamics problems are implemented in specific software that requires the user to have a
very good understanding of the underlined theory, and specific skills to determine the parameters, the calibration of
numerical models and constitutive laws, the optimization of calculations, etc.

Among the common soil dynamics problems, the following ones can be cited:

the study of vibrations in the soil due to construction (e.g. during pile driving),
the study of the soil’s response to the vibratory behavior of structures (rotating machines, vibrations due to
wind effects, insulation and reduction of vibratory effects of equipment, etc.),
vibrations generated by shocks, impacts, explosions,
vibrations due to moving loads, especially traffic loads,
the seismic response of geomaterials and the seismic propagation of waves in soil media, with the
characterization of seismic motion for the earthquake-proof design of structures,
soil-structure interaction and the consideration of settlement (if any) in the dynamic response of structures,
the calculation of impedance functions of foundation systems,
the characterization of the cyclic behavior of soils and the development of adapted constitutive laws,
verification of the seismic stability of geotechnical structures (foundations, earth structures, retaining
structures, underground structures),
modeling of the coupling with the fluid phase in the soil (liquefaction) etc.
$translationBooks

PART 1 – THEORETICAL ELEMENTS

F8. Characteristic scales

F8. Characteristic scales
Firstly, to frame the discussion about soil dynamics problems, it is necessary to distinguish the characteristic scales
relative to the definition of the physical phenomena studied in these problems. These scales are:

the geological scale,

the site scale,
the geotechnical scale,
the civil engineering structure scale.

The first issue one can encounter is that each problem in soil dynamics involves one or more of these scales. Indeed, it
is important to associate in the same model, domains with extremely different dimensions.

Moving from one scale to another corresponds to a specific type of problem, which requires its own solving methods.
Taking seismic problems as an example, the transition from geological to site-scale raises the issue of determining
local site effects. Moving to the geotechnical scale highlights the problem of characterizing seismic motion at a control
point (which may be located at the soil surface or a characteristic depth). Finally, scaling up the structure underlines
the problem of soil-structure interactions, i.e. the problem of understanding how the dynamic response of the soil is
modified by the presence of the structure and vice versa.

Depending on the problem studied, it is also possible that the scale of the structure interacts directly with the site
scale or the geological scale. This is especially true for large structures such as bridges (structure scale ~ site scale) or
tunnels (structure scale ~ geological scale).

In this section, the discussion on geotechnical and structural scales (characteristic scales for most civil engineering
structures) will be limited to the problems of Dynamic Soil-Structure Interaction (DSSI).

1. Classification of the methods for DSSI problems

Before proposing a method classification for DSSI problems, it should be noted that the dynamic behaviors of the
studied systems are made of two distinct domains:

1. the structural domain, including the studied structure and possibly an area of soil in the vicinity of this
structure. This domain highlights bounded dimensions, and is commonly called the generalized structure
domain or superstructure or simply structure,
2. the soil domain, surrounding this structure, whose dimensions are unbounded.

The two domains are separated by a boundary which is called the interaction horizon. This boundary is fictitious, and
its definition is, in a sense, arbitrary. For example, the interaction horizon can be defined as the real interface between
the structure and the soil, or it can be placed very far from the structure.

On the other hand, since DSSI problems are like wave propagation problems, the adopted formulation must satisfy a
radiation condition. For example, in the case of a load applied on the structure (i.e., the structure is identified as the
source of the waves), the radiation condition requires that there can be no waves converging towards the source.

The methods to solve DSSI problems are divided into three main families:

sub-structure methods,
direct methods,
hybrid methods.

Sub-structure methods are the most common for Civil Engineering DSSI studies. The interaction horizon in this case
is placed at the physical interface between the soil and the structure. This separation makes it possible to model the
"structure" domain using classical finite elements and the "soil" domain by other means, which are better suited for
the treatment of unbounded domains, such as the boundary element method (BEM or one of its variants). The
radiation condition is rigorously formulated to infinity. The sub-structure method acquires its practical interest if it is
combined with the linear hypothesis for the structure and soil domains. In this case, it can be shown, via Kausel's
superposition theorem, that the global soil-structure interaction problem can be decomposed into three sub-problems,
as shown in the figure below.

Kausel superposition theorem in the substructure method

The kinematic interaction concerns the diffraction of waves in the soil by the presence of the foundation and its
substantial loading. The inertial interaction concerns the exchange of energy between the soil and the structure which
can be characterized in terms of impedance. These first two problems provide respectively the driving motion and the
boundary conditions that can be used in the third sub-problem for the dynamic analysis of the bounded domain.
Consequently, the sub-structure method uses finite elements only to model the structure. One can then refer directly
to the chapters explaining the use of finite elements in conventional structural dynamics studies. One of the
disadvantages of Kausel’s superposition theorem is related to the nature of geomaterials, which have an intrinsic non-
linear mechanical behavior. Therefore, to call upon the hypothesis of linear behavior for the "soil" domain, it is
necessary to proceed to a calibration of the viscoelastic parameters of the geomaterials according to the anticipated
levels of deformations. Defining the effective viscoelastic parameters of the soil layers constitutes the first step to
implement the sub-structure method, it allows us to use the superposition theorem in systems where the non-linear
behavior remains moderate.

The direct methods are the second large family of approaches for the DSSI and can be applied for systems with
strong non-linear behaviors. The interaction horizon in this case is placed very far from the structure, encompassing a
large area of the "soil" domain, as shown in the figure below.

The general principle of direct methods

The dynamic equilibrium equation is formulated for the "structure + soil" assembly and involves an external loading
vector qf defined along the interaction horizon. The "structure + soil" set is modeled by the FE method and the
displacements in the "soil" and "structure" domains are computed simultaneously. The calculations are performed in
the time domain. It is then possible to incorporate in the model all the geometrical characteristics of the problem, the
material heterogeneities of the ground or the superstructure, and to introduce the constitutive laws necessary for the
description of the non-linearities of the system. Despite these important advantages, the direct methods present
several issues:

defining the optimal position for the interaction horizon and identifying the external loads applied onto it
(especially for seismic problems) are non-trivial tasks, often requiring substantial preparatory work,
in contrast to the sub-structure methods, where the radiation condition is rigorously formulated at infinity, in
the direct methods, the radiation condition is roughly formulated at the level of the interaction horizon. This
hypothesis allows us to neglect the unbounded domain in the model, but special elements must be
implemented along the interaction horizon (which is identified with the boundary of the modeled domain).
These elements are supposed to preserve the unbounded character of the modeled domain by ensuring that
there will be no spurious wave reflections along the boundary of the modeled domain,
the spatial and temporal discretization (mesh size, time step) depend on the mechanical parameters of the
modeled system, and often the mesh size or time step required for a problem can be expensive to compute,
the time integration scheme must satisfy the accuracy and stability requirements of the solution while
remaining optimized with respect to the computational cost,
the constitutive laws admitted that describe the non-linear behavior of soils require a calibration to represent
the cyclic behavior of soils,
finally, direct methods are still extremely expensive for 3D problems. Therefore, it is necessary to be able to
switch from 3D to 2D modeling while preserving the dynamic characteristics (mass/stiffness / modal
characteristics) of the studied system.

Finally, hybrid methods are the third large family of methods for DSSI processing. These methods are in-between
direct and sub-structure methods. The main idea is to separate the soil into two distinct domains: the first is a field
near the structure where all the non-linearities relevant to the DSSI problem are assumed to be developed. The second
is the far-field where the soil behavior is not affected by the interaction with the structure. Thus, the far-field can be
handled by techniques suitable for linear problems in unbounded domains (typically BEM with a resolution in the
frequency domain), while the near field is incorporated in the model of the superstructure and can be handled by a
direct method (FE modeling - non-linearity modeling - resolution in the time domain). The general principle of hybrid
methods is presented in the figure below.

The general principle of hybrid methods

The issue with hybrid methods is the definition of the boundary between the near field and the far-field (interaction
horizon) as well as the formulation of a scheme that restores rigorously and efficiently the coupling between the two
fields. Numerically, this operation results in a FEM-BEM coupling scheme, where the BEM component feeds the FEM
component with the histories of the interaction forces and impedance functions to be deployed along the interaction
horizon.

A special case of modeling using hybrid methods is the development of macro-elements in the nonlinear dynamic
behavior of different types of foundations. Macro-elements are connection elements, which are located at the
foundation level and have a constitutive law to describe the nonlinear mechanisms developed along the soil-structure
interfaces, such as soil sliding, debonding, and plasticizing.

What was said previously suggests that within the approaches to solve DSSI problems, the FE method is mainly used in
the families of direct and hybrid methods for near-field modeling.

In the following paragraphs, a methodology is given to subdivide the structure and some issues are discussed
regarding the implementation of direct methods for DSSI problems.

2. More about the sub-structuring method

The first step of the sub-structuring method is setting the viscoelastic parameters of the geomaterials according to the
level of deformation. The hysteretic non-linear behavior of soil under cyclic stresses can be approximated by an
equivalent shear modulus and viscous damping. Firstly, the curves of the reduction of the shear modulus and the
increase of the damping coefficient as a function of the deformation have to be defined according to the results of
laboratory tests specific to the project (part of the geotechnical investigation) or, in the absence of a dedicated
geotechnical campaign, based on scientific references. Then, a linear 1D analysis of the response of a soil column
sample must be conducted using transfer functions which, from the acceleration at the rock and the first set of soil
layer properties, gives the displacements, velocities, accelerations, stresses, and shear deformations in each layer.
Finally, an iterative procedure is required to ensure that the properties used in the analysis are compatible with the
maximum deformations calculated in the layers. The inverse analysis, i.e. the calculation of the rock motion from the
surface acceleration, is possible through the same transfer functions and is called deconvolution. The iterative
procedure does not always guarantee the uniqueness of the solution in the search for equivalent soil properties in case
of large deformations. These analyses can be carried out with dedicated software such as SHAKE 91 for example.

Diffraction and impedance problems can be solved in the frequency domain by adapted calculation tools such as
SASSI2010 or MISS3D codes. Alternatively, simplified methods valid under certain conditions can be used.

The study of kinematic interaction is of interest in the case of deep foundations but is neglected in the case of shallow
foundations. For a seismic analysis, the simplified method consists of the imposition of free-field soil deformations on
the foundation piles. This stress, of kinematic origin, must be added to the stresses resulting from the inertial
interaction for the dimensioning of the piles.

Using a dedicated software (e.g. SASSI2010, MISS3D), the study of the inertial interaction leads to the calculation of
the stiffness and damping at the monitoring point of the foundation as a function of frequency. For the calculation of
impedance functions, the foundation is by default considered a rigid substructure. For a shallow foundation, the
impedance terms relative to displacements and rotations are used, unlike for deep foundations, and the coupling
terms between the displacements and rotational stiffnesses must also be considered. The impedance values, retained
for the structural problem of dynamic analysis, are set based on the eigenmodes of soil-structure interaction with an
iterative procedure explained below. A first modal analysis is conducted with a set of test impedance values. The next
set of impedance values is calculated from the frequencies of the eigenmodes, identified as those of ground-structure
interaction for each degree of freedom of the (supposedly rigid) foundation. The second modal analysis is used to
correct the frequencies of the selected eigenmodes and so on. The procedure stops when the convergence of the
eigenfrequencies is reached.

Simplified methods for shallow foundations generally give formulae for the direct calculation of the impedance. The
best-known methods are those of Newmark-Rosenblueth, Deleuze, Eurocode 7, SETRA Guide, Veletsos, and Gazetas.

The main condition for the validity of the simplified methods is the assumption of homogeneous, single-layer soil,
except for the Gazetas method that allows the consideration of a double-layer soil for the cases of foundations of
approximate circular shape. Deleuze's method considers the frequency of the fundamental vibration mode of the
structure. The methods of Veletsos and Gazetas consider the embedding of the foundation in the soil.

For deep foundations, the Gazetas method and the Eurocode 8-5 method give formulas applicable in the following
three soil conditions:

soil whose Young's modulus varies linearly with depth,

soil whose Young's modulus varies with the square root of the depth,
soil whose Young's modulus remains constant with respect to depth.

Another simplified method for deep foundations is Winkler's method, which requires the modeling of a beam resting on
a system of springs and dampers with properties based on those of each layer and representing the frontal pressures
and frictional forces.

After solving the dynamic analysis problem, the Winkler method is also used to calculate the stresses along the piles
for their design. This method makes it possible to model an elasto-plastic behavior for each soil layer and to
incorporate a simplified analysis of the kinematic interaction by applying the soil displacements imposed on the pile.

3. Defining the mesh size and time step

The choice of mesh size is a very important step in soil dynamics modeling because the discretization influences the
characteristics of the waves that can be modeled with a given mesh. The usual criterion to model a wave propagating
in the soil with acceptable accuracy is to have in the mesh, 10 elements per wavelength. If we denote h the mesh size
and λ the wavelength, the criterion to be imposed on the precision of discretization is written as follows:
This criterion can also be formulated as a function of the minimum propagation speed Vmin that needs to be
considered in the model and the maximum characteristic frequency fmax of the waves to be represented in the mesh.
The criterion for the mesh size is then written as:

For seismic problems, the quantity Vmin usually corresponds to the minimum value of the effective velocities
(compatible with the seismic level) of propagation of S waves for the modeled soil layers. The quantity fmax
corresponds to the cut-off frequency of the dimensioning spectrum (~30 Hz) or a maximum frequency of interest for
the ground-structure interaction (~15 Hz).

Apart from the mesh size h, the choice of a time step Δt is also conditioned by the mechanical characteristics of the
modeled medium. Especially, by analogy with the wavelength, it is necessary to be able to describe the smallest
period that must be considered in the analyses by at least 10-time increments (the precision criterion is identical in the
spatial and temporal discretization). The criterion about the time step can be formulated as follows:

where Vmax is the velocity corresponding to the maximum value of the (effective) propagation velocities of the S
waves and h corresponds to the mesh size.

Note that for an applied problem where Vmin = 150 m/s, Vmax = 500 m/s and fmax = 15 Hz, the obtained values would
be h ≤ 1m and Δt ≤ 0.002s. One must pay attention to non-linear dynamic problems because the characteristics of
interest to define the mesh size (in particular Vmin) evolves with time (softening of the soil by passing in elastoplastic
regime).

4. Cyclic behavior of soils - Damping

There is a growing interest in the implementation of direct methods for DSSI problems concerning the possibility of
rigorously modeling the non-linear behavior of geomaterials, at least within the domain encompassed by the
interaction horizon, which is modeled with finite elements. The designer can accurately model the geometry of the
ground layers and assign to each formation the most suitable constitutive law to reproduce the anticipated cyclic
behavior.

Calibration of non-linear models - The first step in this modeling procedure concerns the calibration of numerical
parameters involved in defining the chosen constitutive laws so that they accurately reproduce the cyclic behavior of
the soils. The latter is characterized, in most Civil Engineering applications, from the degradation curve of the shear
modulus G as a function of the distortion γ (G~γ curve). It should be remembered that the G~γ curve varies according
to the type of soil (sand/clay/silt/gravel), the effective vertical stress, the plasticity index, etc. The procedure for the
calibration of non-linear models includes the following steps:

organizing the mesh and showing the areas governed by the same G~γ curve,
defining the elastic parameters (low distortion parameters) for each zone,
adopting an adequate constitutive law for each zone (fracture surface, strain hardening law, flow law,
behavior in the deviation plane and the hydrostatic axis, etc.),
defining the G~γ curve characterizing each zone as a function -among others- of its mean effective stress,
calibrating the numerical parameters that characterize the non-linear constitutive law to reproduce the G~γ
curve of each zone.

The calibration procedure results in a table that provides for each zone, the type of constitutive law adopted, and the
set of numerical parameters necessary to define it.

Drained and undrained behaviors – When a water table is present in the soil domain, it is important to define
whether the response of the geomaterials during dynamic loading will be governed by drained or undrained behaviors.
This definition depends on the nature of the soils (cohesive materials; pulverulent materials) but also on the duration
of the studied phenomenon. For phenomena of limited duration (of the order of a few seconds, such as for
earthquakes) it is often difficult to prescribe whether the soil will respond in drained or undrained conditions. For these
situations, it is advised to carry out two separate calculations, both under the hypotheses of drained and undrained
behavior, and to define the overall response as the envelope of these two calculations.

Damping - When modeling the behavior of soil layers via an elastoplastic law, the anticipated damping of the material
is obtained directly via the dissipation deduced from the constitutive law. However, this definition of damping differs
from its conventional characterization (usually accompanying the definitions of the G-γ curves) via a β-γ curve, where
β represents the equivalent viscous damping rate. This difference comes from the fact that the type of damping is not
the same in both cases: on the one hand there is the viscous damping β, on the other hand, there is the hysteretic
damping directly derived from the elastoplastic law. Thus, it is inaccurate to calibrate the non-linear model according
to the β~γ curve. However, it is important to verify afterward that the damping rates generated by the calibration of
the elastoplastic model are not far from the values of β which come from the theoretical β-γ curves. In addition to the
hysteretic damping, it is necessary to define an "elastic" damping, i.e. a basic damping rate to consider as soon as the
load is applied. Defining this damping component (ranging between 1% - 3%) is useful for the proper conditioning of
the dynamic equilibrium equation and can be introduced via Rayleigh damping. Finally, it is possible to introduce a
third damping component, namely numerical damping, which results from choosing a dissipative integration scheme.
This third component should be used with care and avoided for common applications.

Liquefaction - Special calibration of soil models is required in the context of liquefaction risk studies. In this case, a
two-phase model is required for potentially liquefiable areas. As a result, it is possible to model the increase in pore
pressure that occurs in these low permeability materials during earthquakes, a phenomenon that can lead to
liquefaction. The calibration, in this case, consists of showing that a liquefied state is obtained on a soil sample that
was consolidated in the state of stress of the potentially liquefiable zone when it is solicited by the theoretical
earthquake leading to liquefaction. The most relevant parameters for calibration with respect to liquefaction are those
governing the response of the soil loaded along the hydrostatic axis (volumetric parameters), such as the angle of
expansion or the rule for defining plastic volumetric deformations, etc.

5. Boundary conditions: Absorbent boundaries - Floating models

Dimensions of the area to mesh - Another complexity concerning the implementation of a direct approach,
concerns the "dimensioning" of the mesh and the definition of boundary conditions. First, the dimensions of the mesh
must be sufficient to describe the phenomena involved. In the vertical direction, the mesh is usually extended to the
bedrock level, where a linear behavior can be assumed. In the horizontal direction, the mesh is extended at a
characteristic distance equal to 4 to 5B on either side of the structure, where B is the characteristic dimension of the
structure. It follows that the minimum dimension of a mesh for the implementation of a direct DSSI method is 10 times
the characteristic dimension of the structure. It is understood that this criterion, combined with the criteria for mesh
size, presented above, can make the cost of a DSSI calculation by the direct method prohibitive.

Boundary Conditions - For problems where external stresses come from the structure, defining the boundary
conditions consists of introducing absorbent boundary elements along the contour of the mesh whose role is to
annihilate the parasitic reflections of the waves that reach the edges of the mesh. The absorbent boundary elements
commonly used in practice are standard viscous dampers that are only accurate (as absorbent boundaries) for a given
direction of propagation and wave frequency. To overcome these limitations, several improved absorbent boundary
formulations are available. Nevertheless, their implementation remains relatively cumbersome and these formulations
are still little used in current studies.

For seismic problems, the role of absorbent boundaries is to reduce the parasitic reflections and to introduce the
incident seismic motion along the contour of the domain modeled in finite elements. To achieve this second objective,
it is often necessary to determine the motion at the bedrock level (since seismic standards generally provide the
dimensioning earthquake at the ground surface): this motion is used as the incident motion at the base of the model.
Additionally, the motion (under free-field conditions) on a soil column (from the bedrock to the surface) must be
determined, which will be used to establish boundary conditions along the sidewalls of the model. Furthermore, the
mesh needs to be sufficiently large in the horizontal direction because the free field movement at its lateral boundaries
must be expressed. For asymmetric meshes, the calculation of the soil column in the free field must be conducted
separately for the columns on the left and the right side of the mesh.

It must always be verified that the calculated motion at a node close to the lateral boundaries of the model is identical
or very close to the free field motion. This verification ensures that the mesh is correctly dimensioned for the
interaction problem under study.

Phasing - As with all geotechnical problems, it is also necessary for dynamic problems to establish the true state of
stress in the geomaterials before introducing dynamic loading. Usually, any DSSI studies are initiated by a gravity
loading phase. At the end of this phase, the system displacements are set to zero and the stress states are kept
throughout the entire mesh. At the end of the gravity phase, it is possible to introduce dynamic loading into the
system.

Floating Models - Still for seismic problems, it is important to adopt a model that allows simultaneous injection of the
horizontal and vertical components of the seismic motion. To consider the vertical component, it is often necessary to
use the so-called "floating model" method, which consists of the following steps:

blocking the nodes at the base of the model and initiating under self-weight,
identifying the reactions of the nodes blocked under self-weight,
continuing the calculations under self-weight by vertically releasing the nodes at the base and simultaneously
applying the gravity forces and the reactions calculated in the previous step: the model is not fixed at any
point in the vertical direction (hence the name "floating model"), the equilibrium is ensured by setting the
equality of the reaction forces and the gravity loads,
finally, after the gravity load, the earthquake is introduced. Moreover, having released the vertical nodes also
introduces the vertical component of the seismic motion.

6. Choice of the integration scheme

The implementation of the direct method for the DSSI is completed with the definition of the time integration scheme
of the dynamic equilibrium equation. The principles that apply to general non-linear dynamic problems are also
applicable in the case of DSSI problems. As these are non-linear problems, the time integration scheme is coupled with
a method for integrating non-linear constitutive laws, typically the Newton-Raphson method or the modified Newton-
Raphson method, etc. The time integration scheme is then used as a basis for the integration of the non-linear
constitutive laws.

Generally, problems of very short duration (impacts, explosions, etc.) are handled by explicit schemes. These schemes
are very efficient if the mass matrix can be triangulated and are conditionally stable (stability criteria often lead to
very small time-steps). For seismic problems, implicit schemes are used, the most used being the Newmark integration
scheme (which is unconditionally stable in a linear regime for a given set of integration parameters).

The choice of a dissipative scheme (leading to numerical damping) can also be proposed to facilitate the convergence
of the solution for highly non-linear problems.

In any case, the choice of an integration scheme, specific to a given problem, with the definition of all the associated
parameters (tolerances, number of iterations, strategy to manage the divergence, etc.) requires a very good
knowledge of the theory and also a significant experience in the typology of the problems treated (seismic, impacts,
fast dynamics, vibration problems, etc.). As a general principle, it is advised to start the analyses with non-dissipative
schemes and with relatively strict tolerances (of the order of 10-4). At the end of each analysis, the file containing the
detailed results should be carefully examined to verify that the residuals of each increment do not exceed the
tolerances and that convergence is obtained on all the meshes of the model.
PART 2 – USEFUL ELEMENTS
$translationBooks

PART 2 – USEFUL ELEMENTS

Chapter A. Understanding the

finite elements
Chapter A. Understanding the finite elements
A.1. What does a finite element software do? Example of framed structures.

A.2. Explicitly, what is a finite element?

Annex 1 – File of the matrix calculation example

$translationBooks

PART 2 – USEFUL ELEMENTS

A1. What does the software do in

a finite element calculation? The
example of beam structures
A1. What does the software do in a finite element calculation?
The example of beam structures
The purpose of this introduction is to show, in a simple example, what lies behind a static calculation carried out with a
finite element beam software.

In the wiki, by misuse of language, the term "bar" is often used to refer to "beam" type elements and not
exclusively to "bar" type elements in the strict sense (i.e. working only under normal stress).

A few reminders:

Computer calculation is based on a representation of the structure by a set of beams whose intersections are nodes.
Its purpose is to determine the displacements of the nodes of the structure, i.e. the displacements of the ends of the
beams, under the applied loads.

Figure 1 - Sign convention - forces and displacements

The forces at the ends of the beam depend linearly on the displacements at these points. It is shown that there is a
matrix relationship between forces and displacements for a beam connecting two nodes i and j, such that, for a given
coordinate system, [Kij][qij]=[Fij] (Figures 1 and 2), allowing the components of the vectors [Fij] to be expressed as a
function of those of the vectors [qij]. [Kij] is called the stiffness matrix of the beam, [Fij] the force vector at the nodes,
and [qij] the vector of nodal displacements.

Figure 2 - Stiffness matrix of a bi-clamped beam

The matrix [Kij*], associated with the global coordinate system, is deduced from the matrix [Kij] expressed in the local
coordinate system of the beam by applying a transformation [Λ]T [Kij] [Λ] (figure 3), making the displacements
compatible with those of the global coordinate system. [Kij*] = [Λ]T [Kij] [Λ] is then the stiffness matrix of the beam
expressed in the global coordinate system.
 Figure 3 – Rotation matrix

To calculate the displacements of all the nodes, the software:

"assembles" the matrices [Kij*] of each of the beams to form the overall stiffness matrix [K] of the structure
(figure 4),
inverses [K] → [K]-1,
multiplies the matrix [K]-1 by the matrix of the external forces [F], previously brought back to the nodes (if
they are applied to a beam other than at the ends), in the global coordinate system, to calculate the
displacement vector [q].

Figure 4 - Principle of assembly

Finally, to calculate the internal forces of the structure, the software multiplies the stiffness matrix [Kij] of each beam
by the vector of displacements [qij], in the local reference frame, at nodes i and j. Thus, the forces are calculated at
the nodes and added, if necessary, to the reactions of the bi-clamped beam (i.e. the forces introduced in the global
model, but here projected into the local reference frame of the beam) - see the second example below. The forces and
displacements along the beams can easily be deduced from those at the nodes using the Strength of Materials
formulae.

Application to a simple example:

 Figure 5 - Calculated structure

This example is processed using a spreadsheet, the source file is provided in Part 3 of this guide - it can be
downloaded by clicking on the link below. Source file in Excel format of the example above.

This spreadsheet covers two examples: the example developed below for a 2-beam structure (to remain easily
readable in paper format) and a second example with 4 beams. These examples allow you to visualize and understand
the sequence of tasks.

The required beam characteristics for a plane problem are the section S, inertia I, Young's modulus E, length L, and
orientation α about the horizontal for instance.

E = 36 000 MPa for all beams.

The numerical values of the stiffness matrix for each beam are easily determined from the literal values in figure 2.

Beam 1: Stiffness matrix

Beam 2: Stiffness matrix

The transformation is applied to these two matrices to make them compatible with the global coordinate system.

Rotation matrix for α = 1.571 Rad:

And its transpose:

Rotation matrix for α = 0.197 Rad:

And its transpose:

Beam 1 - Result of the operation [K*] = [Λ]T [K] [Λ]:

Beam 2 - Result of the operation [K*] = [Λ]T [K] [Λ]:

All these unit matrices are then assembled, i.e. all the matrix values relating to the same nodal displacement are
cumulated.

For our example, it means summing the values of the two matrices corresponding to the common nodes (in the
general case, the beams do not always connect two nodes whose numbering follow each other as shown below):

The blocked displacement components can be removed from the matrix since we already know that their value is
equal to zero:

This matrix is easily inverted using the spreadsheet:

Load 1: External forces applied on a node

Figure 6 - Load definition

The nodal displacements are calculated (in the global coordinate system):

The forces in the column (beam 1) can be determined after transforming the displacements to the local coordinate
system of the beam:

We have a compression of 12.68 kN, a shear effort of 1000 kN, and a bending moment that changes sign along the
beam (we check that 1000 kN * 8 m - 3654.89 = 4345.11 kN.m in the base of the column - ok).
 Figure 7 - Diagrams N, V, M - example 1

Loading 2: External forces applied to a beam (i.e. other than at the nodes).

Figure 8 – Loading definition

In this case, it is necessary to calculate the support reactions of the bi-clamped beam under these forces beforehand,
as they are the ones that have to be incorporated into the matrix of external loads (the software only knows the
nodes!). A bi-clamped beam reaction table is enough. Of course, the sign conventions must be observed.

Application to the horizontal distributed load p = -1000 kN/m on the column (figure 8). In this case, it is known that the
clamp moments are -pL²/12 and the shear forces at the restraints are +/-pL/2; therefore, with L=8 m and taking into
account the sign convention:

Multiplying [K*]-1 by these efforts provides the values of the displacements of all the degrees of freedom. It is then
enough to multiply the stiffness matrix of beam 1 by these displacements to recover the forces at the nodes.

The efforts in the column (beam 1) can be calculated, after converting the displacements from the global coordinate
system to the local coordinate system of the beam...

... and add them together with the forces of perfect clamping, with the appropriate sign convention:
We have a compression effort of 1298 kN, a shear effort that increases from 0 to 8000 kN (pL=1000 kN/m * 8 m =
8000 kN), and a bending moment at the base of the column which is much higher than the perfect clamping moment.

The same methodology applies to beam 2 to obtain the efforts at the nodes.

Figure 9 - Diagrams N, V, M - example 2

Cases where the beams present important differences in stiffness:

For example, the vertical beam is stiffened very strongly. We decide, keeping a length of 8 m, to increase its cross-
section to 1m * 1012m (b*ht). The matrix [K*] is shown below. The difference between the larger and smaller values
can be easily observed ... which can lead in some cases to numerical instabilities (the resolution leads to dividing the
matrix terms by one another).

Figure 10 - Matrix with contrasting values

$translationBooks

PART 2 – USEFUL ELEMENTS

A2. What is a finite element?

A2. What is a finite element?
The resolution of the efforts in the elements is carried out following the calculation of the node displacements. The
method is specific for each type of element and depends on the software used.

However, the main principle is common to all software, it consists of "isolating" an element to calculate the forces at
the Gauss points from the nodal displacements.

The position of the Gauss points is normally specified in the software documentation; in the case of a 4 node shell
element as below, they could be located at a distance from the edge of the element equal to about 1/5 of its width.

Example of a 4-node element

The efforts in the center of the element are calculated as the average of the forces at the Gauss points, the efforts at
the nodes are extrapolated from the Gauss points.

To summarize, the software calculates:

the efforts at the Gauss points G1 to G4 from the displacements at the nodes n1 to n4
the efforts at center C which are the average of the efforts at the Gauss points G1 to G4
the efforts at nodes n1 to n4 which are extrapolated from the efforts at the Gauss points G1 to G4
These calculations are carried out for all the elements. In the end, there are as many efforts at the nodes as there are
elements connected to this node (here 4 elements E1 to E4 connected to node n1).

We can then deduce:

either the maximum effort at the node (maximum of the efforts calculated from the elements E1 to E4)
or the average effort (average of the efforts calculated from E1 to E4)

Main remarks concerning common usage

Generally, quadrangular elements will lead to better accuracy of the results than triangular elements.
Results at Gauss points are the most accurate, but they are generally not accessible to users.
The results at the center of the elements are more reliable than the ones at the nodes because they are not
extrapolated.
It is up to the engineer to choose the type of result (maximum, average, smoothened, etc.) according to the
behavior of the structure. There are no predefined rules.

Example of a load on a bridge slab illustrating the differences in results during a FE calculation.
$translationBooks

PART 2 – USEFUL ELEMENTS

Chapter B. Computational
objectives and necessary
characteristics of the tool
Chapter B. Computational objectives and necessary
characteristics of the tool
Creating a finite element computational model includes several steps. The choice of the tool is critical and depends on
various criteria. A successful model requires good organization and preparation.

B.1 to B6 Criteria to be considered

B.7 Organization of the computational model

$translationBooks

PART 2 – USEFUL ELEMENTS

B. Calculation objectives and

necessary tool characteristics
B. Calculation objectives and necessary tool characteristics
Creating a finite element computational model includes several steps. The choice of the tool is essential and depends
on several criteria.

B.1 According to the object to model

First, computational software must be adapted to the object one wants to model.

a) The complexities are different for bridges (standard short-span structures or complex structures with several spans),
buildings, or geotechnical structures (retaining walls, tunnels, dams...).

Depending on the size of the object, one or more modeling scales can be defined:

a geological scale (which aims to process a structure in its environment according to geological data),
a global scale which aims to handle a structure as a whole (longitudinal calculations of bending for bridges
for example),
a semi-local scale with a more refined mesh for some elements of the structure (transverse calculations
under the actions of axles for bridges, for example),
a local scale, with very refined modeling and post-processing of the results (calculations of reinforced
concrete for punching, diffusion or bracing reinforcement, for example).

b) An object can result in several models that complement each other.

Scale GLOBAL SEMI-LOCAL LOCAL

Stress concentrations
Bridges Longitudinal bending Transverse bending (anchorage or deflection zone of
cables, concentrated support)

Load path / Bracing (wind, Local bending of horizontal Concentrated, accidental loads
Building earthquake) / Dynamic elements (floors, foundation (anchorage area, machine
calculation slabs) supports)

Faillure of anchored tendons,

Geotechnics Backfill, retaining wall
draining...

c) In the case of a structure whose construction kinematics have an impact on the final state, the software must be
able to authorize the complete simulation of the phasing, allowing, among other things, selective activation of the
elements (such as the tensioning of prestressing cables and braces, for example), which makes it possible to work on
the part of the structure under construction.

d) If a dynamic calculation is necessary (for example for a seismic calculation considering the effects of the soil or a
vibration calculation or even a fast dynamic problem such as an explosion) not all software can be used.

e) Non-linear calculations are not systematically possible (calculation with material non-linearities, elasto-plastic
supports, calculation of large second-order displacements for buckling verification, etc.).

f) Modelling of cable structures (whose transverse stiffness to bending and torsion is not very important compared to
the longitudinal stiffness) is also particular and only provided by some software.

g) The determination of a thermal field (volume loading expressed from a variation in temperature and the coefficient
of thermal expansion of the material) may require additional modules to the software.

h) Foundation modeling, for a foundation slab with uplifting, for instance, is generally linked to non-linear calculations
and thus agrees with point d).

B.2 Depending on the study phase

The design phase of a structure has an impact on the level of precision expected from the calculations.

In the Preliminary Study phase, the modeling must remain pragmatic, simple, and give approximated results.
In the Preliminary Design or Project phase, a pre-dimensioning of the work must be established as a basis for
a bid. The calculations are more thorough, even complex, but rarely exhaustive. Models must be fast, flexible,
and easy to correct so that they do not restrict the project and allow for easy testing of variants.
During the Execution phase, when the project is stopped, the calculations must be completed, they can be
complex, and must give concrete and exploitable results to guarantee a dimensioning that is both safe and
optimized.

The table below shows the level of detail generally expected for each phase. Depending on the specificities of the
project and the requirements of the customer, the content of the phases may vary:

Object \ Advancement Preliminary Study Pre-design PRO EXE

 Global model + semi-local
Ratios, Feedback, Simplified
Bridges, Industrial Building, verifications or even Global model + semi-
global model (a 2D model is Global model
Civil engineering structures verifications for critical local + generalized local
preferred)
points

B.3 Based on the verification objectives

For the same object and the same study phase, several models may have to be set up, each dealing with different
verifications.

One must try to ensure that the same model can meet as many verifications as possible, but the problems often have
to be decoupled.

B.4 Based on the expected results

The choices can be influenced by the results different software can handle.

a) Integrated post-processing modules make some software interesting, in particular those that allow the rebars of
reinforced concrete sections to be obtained from stresses, or those that incorporate the verification of steel profiles
when buckling or the verification of conventional connections. Beforehand, it is essential to check the conformity of the
post-treatment with the reference regulations of the project.

b) The types of output results can vary and be relatively adaptable (listings, stress diagrams, maps, combination
envelopes with or without conservation of concomitances).

B.5 Depending on time and resources

In terms of deadlines, and therefore budget, a distinction must be made between the modeling time (geometry, loads,
combinations, etc.), the calculation time, the time to set-up (debugging errors, verifying support conditions, etc.), and
finally the analysis time. These durations can vary significantly depending on the software because their user interface
is less intuitive (the design interface can be either graphical or programmable in a command window or using
spreadsheets). The extraction of results is not necessarily instantaneous and can be useful to identify critical points.

Moreover, the resources available within the design office guide the choice of the type of calculation: the hardware, on
which the software is installed, must be open, so does the user license, so that it can be easily handed over to another
user. The availability schedule of the tools can play an important role in deadlines and possibly lead to a change of
software.

If the design office has a type of hardware with reduced capacity, the degree of complexity of the calculations is
reduced accordingly. And if the complicated computational model is kept, the design office should not "tinker" a model
adapted to the capacity of its equipment, because the simplifications adopted can then lead to issues concerning the
accuracy of the results. It is then in the design office's best interest to subcontract the calculation.

For large structures, it may be preferable to implement two levels of modeling, using sub-models. Indeed, a single
model can quickly become disproportional in size, and therefore difficult to structure and manipulate. However, it is
necessary to be able to link the models together. Furthermore, the need for specific computing capabilities can also
lead to the splitting of models and the use of different software.

Another reason for splitting the models as described above can be linked to the study schedule (a consequence of the
organization of the projects): the calculation of the whole work sometimes precedes by several months, for important
projects, the calculation of certain parts of the work. There are no reasons not to compensate by post-processing for
certain shortcomings of the software, by extracting the results of a sub-model and to process them manually or using
another software.

An equally important resource is the staff, i.e. the engineer in charge of modeling. They must be trained to use
the software. If the engineer is new to the software, the learning curve should not be underestimated, as the
time required to design and develop the model can be greatly increased. Tutoring with a senior engineer is
strongly recommended, despite the time investment involved.

B.6 Depending on how user friendly the interface is

Finally, making the software user friendly is essential.

a) A software with a complete manual (installation, handling, and operating instructions) is always much appreciated.
The presence of a catalog of examples and applications, tutorials, and manipulations are extras but also valued.

b) The possibility of programming (creating and reading a text code in a programming language of one's own) allowing
intuitive and fast data entry is an asset. It can offer many possibilities to the user, for example, to automate the
modeling of simple and repetitive model structures or to adjust the layout of the results, by providing text or Excel
outputs adapted to the studied sub-structure.

c) Confidence in the software saves a significant amount of time by avoiding superfluous verifications. Thus, having
regular updates is an indicator, as is the existence of an available and reactive technical hotline, capable of providing
punctual assistance on a specific model. Of course, trust does not dispense the model verifications explained in this
guide.

d) The version of the software can also play a role, in the case where certain features have been added/removed or
where the stability and/or speed of a version is not satisfactory.

e) Some software have complete libraries (materials, profiles, bolts, assemblies...) that save a considerable amount of
time. Functionalities specific to civil engineering projects also exist, such as the application of regulatory automatic
loads (types A(l), Bc, LM1, LM2...).

f) Depending on the standards that apply to the project, the software can propose loads, combinations, and pre-
programmed verifications. This is a helpful feature but it must always be checked using simple cases.

g) A render function is an advantage because it allows visual verification of the type and orientation of the profiles or
bars. Besides, some software allow a 3D export which is a very useful communication support in meetings with
stakeholders (see also chapter E.3 for the BIM part).

h) A software that specifies the line of the data file that contains an error or the list of wrongly modeled objects
(overlays, ...) in the spatial model offers a real advantage. Error messages must be clear and precise (if possible, in the
numerical language of the engineer).

Experience feedback:

Feedback is important: meetings or feedback documents should allow drawing positive and negative lessons from
ongoing or completed projects. They must cover the methods used, but also the IT equipment used or even the
production level that was reached.
$translationBooks

PART 2 – USEFUL ELEMENTS

B.7 Organization of the calculation

B.7 Organization of the calculation
This is an essential step to be carried out at the beginning of the study.

1. Preliminary reflection on modeling

A preliminary step is essential for the modeler: the analysis of the overall behavior of the structure. Indeed, it would be
foolish to start modeling a structure before understanding its overall behavior. A first sketch of the structure reveals a
good understanding of the behavior and will be used as a framework for the creation of the model.

Mainly, this analysis of the overall structure allows distinguishing the main elements that reflect the behavior of the
structure. Among these main elements, one will distinguish for example :

for civil engineering structures: the load-bearing structure of the deck, the structure of the supports, the
bracings,
for reinforced and prestressed concrete buildings: columns, walls, and slabs,
for building structures: columns, beams, bracing elements.

The model must be based on input data, with at least:

a general assumption, which contains a description of the object, the standards applied and the loads applied,
sketches or general drawings of the object to be modeled,
a general construction principle,
an outline of the static and possibly dynamic behavior.

Modeling does not replace these elements of preliminary reflection.

BIM and modeling: The designer may be tempted to use the input data in an automated way to build his model.
This is a frequent argument of software publishers. In this case, he will have to be particularly vigilant about the
quality of the inputs provided (it is indeed not uncommon to find 3D models with geometric nonsense) and the
level of detail of the input data (quantity of hoppers for example).

In the case of automated processing, special focus should be given to the geometry construction process,
especially at the connections. Thus, the analysis phase of the overall functioning mentioned earlier is a way of
preventing any anomaly.

2. Input data validation / Input synthesis

All documents defining the geometrical assumptions, materials, loads must be referenced with their origin, index, and
date of issue.

It is necessary to validate the coherence of these different documents. For example, are the architectural plans and
the structural plans consistent with each other? Are there any geometric discrepancies between the different plan
files? Are the data exhaustive? Are infeasibilities already observed (e.g. complex load path, lack of bracing...)?

This synthesis work allows us to highlight the missing input data and/or likely to be modified and to define the
conservative measures taken to compensate for the missing data.

Based on this preliminary analysis, certain decisions will be immediately imposed on the designer: exchanges
with the client, revision of the geometry of the structure (design revision), input of parameterizable data...

This synthesis phase will ideally take the form of a "Modeling Note" that will evolve as the model progresses. The aim
is to have the hypotheses validated by all the participants in the study very quickly to avoid modifications, which are
often long and complex.
 3. Specifics of the study

All the specifics of the study must be listed at the beginning of the study:

the physical constraints of the project (important heaving, urban area, construction phasing, ...),
the study constraints (tight schedule, numerous interfaces, missing data, ...),
the particularities of the model (size of the model, non-linear, earthquake, ...),
the requirements and sensitive aspects of the project (slender structure, asymmetrical, heavy wind, etc.).

It is necessary to show that the modeling will consider all these points.

4. Planning of the study

It must clearly show:

the deadlines of the main modeling phases (geometry, materials, loads, interface or soil-structure interaction,
combinations, etc.),
the deadlines for receiving missing or modifiable input data,
the deadlines to send the deliverables used as input data to other study stakeholders (interface) and for other
deliverables,
the consistency of the study with respect to the dates of the phases.

5. Modelling Principles

The goal is to explain the calculation methods, making sure to:

clarify the principles of the structural model, the methods to consider loads, combinations, ...,
justify all approximation assumptions,
if necessary, present small test models validating the hypotheses,
present the sequence of calculations.
$translationBooks

PART 2 – USEFUL ELEMENTS

Chapter C. Good practices to

create a model
Chapter C. Good practices to create a model
The different levels of analysis and the complexity associated with the model have already been defined in the
previous chapter before the modeling stage. This chapter highlights the simplifications that can be used to create a
model that is structurally representative of the real-life structure and the loads it is subjected to.

The different levels of investigation and complexity associated with the model have already been defined before the
modeling phase in chapter B.

This chapter presents the possible simplifications one can adopt when creating a model that is structurally
representative of the design of the actual structure, its behavior, and the stresses it is subjected to.

C.1 Input data and units

C.2 Modelling of the main elements

C.3 Finite elements and meshing

C.4 Modelling of the non-structural elements or the equipment

C.5 Boundary conditions

C.6 Connections – links – assembly

C.7 Eccentricity

C.8 Combined cross-sections (beam/deck)

C.9 Materials

C.10 Behaviors specific to shear and torsion

C.11 Loads modeling

C.12 Further information related to volumetric elements

C.13 Further information related to non-linear calculations

C.14 Further information related to prestress

C.15 Further information related to phase calculation

C.16 Further information related to structural dynamics and seismic calculations

$translationBooks

PART 2 – USEFUL ELEMENTS

C1. Input data and units

C1. Input data and units
The input data must be consistent, whether these values are defined in the model itself or come from other files such
as a library of profiles or data from other software (*). Some elements of the methodology are provided in paragraph
B.7 Organization of the calculation.

(*) Special attention must be paid to the units and signs when introducing stiffness matrices that model the
foundations or other parts of the structure, especially if they come from another design office that uses different
software. Moreover, for soils, one should verify whether the characteristics are expressed for the long or short term.

Material characteristics, especially for concrete, must be consistent with the analysis conducted (see details in
paragraph C.9).

The unit system in which the data are expressed must be known because it will condition the units of the results. The
use of SI units is preferable.

One should ensure the consistency of the stresses, lengths, modulus, and stiffness units.
$translationBooks

PART 2 – USEFUL ELEMENTS

C2. Modeling of the main

elements
C2. Modeling of the main elements
C.2.1 Creating the geometry

The first stage of modeling consists of creating the geometry of the model by defining points, lines, boundaries, areas,
and volumes. The notions of nodes, elements, and meshes are associated with finite elements.

In some software, the geometry can be defined before creating nodes, elements, and meshes. In others, the geometry
is established outside the software, using diagrams or Excel spreadsheets, and the nodes, elements, and meshes are
then defined directly in the software.

In any case, the sign conventions used by the software must be known at the start of the modeling
(direction of gravity in the global coordinate system, sign convention for bending moments, forces, and
stresses).

Some general notions:

Nodes: their presence is essential at the supports, the joints, and the boundaries between geometries.
Depending on the software, nodes should also be present where the data will be extracted. Please refer to §
D.3.2 for more information. Depending on the software, a node must also be provided at each load
application point. Finally, depending on the phenomena one wants to quantify, for example for large
displacements or dynamic calculations, intermediate nodes must be defined on the beams to accurately
represent them. Defining the points of the geometry means incorporating all these node requirements.
However, the number of nodes must be limited to keep the model as light as possible.
Structural elements: in most non-solid models (bars, plates, shells ...), the elements will be modeled at
their average fiber. This is the safest method for the good transmission of forces between elements and the
consideration of secondary effects. In the case of beams for which the loads are located on a particular face
(bridge secondary beams, for example), it is possible to define the element at this face and to create an
offset, if the software does not do it automatically. The positioning of the average fiber of the elements with
offsets is discussed in § C.7 Offsets. In the case of beams of variable height, the mean fiber is no longer a
straight line, which leads to multiple local coordinate systems for the different inclinations and can complicate
the exploitation of the results. If the arch effect is not considered, the model can be simplified by keeping the
neutral fiber straight-lined.
See also the calculation examples for beam grids for which simplifications can be admitted (Link to the
examples).

C.2.2 Degree of simplification: shafts and openings

Depending on the study phase and the type of calculation carried out (for example stability or load reactions), not all
the openings will necessarily be modeled.

The case of buildings.

For buildings, when the openings are considered, it is advised to integrate into the geometry the shafts of non-
negligible sizes and the ones which may affect the behavior of the structure (at least in the bracing elements). The
latter will be cut according to the intersection wall/wall, wall/floor, shafts, to have a mesh as regular as possible.

In the case of modeling a building, the shafts are defined as a function of:

their size (any shaft in a wall or slab whose largest dimension is less than 1m is commonly neglected).
their spatial distribution - small but closely spaced openings can be considered as a single opening whose
dimensions correspond to the perimeter of the envelope.
their influence on the force transmission.

Particular attention must be paid to the lintels of doors in structural walls (and bracings). Indeed, these lintels might
have numerous openings, so they may no longer be able to fulfill their structural role. Consequently, the model must
be adapted.

When the openings are not known, the modeling of buildings subjected to horizontal forces (wind, earthquake) must
consider conservative measures concerning the large shafts (generally for ventilation). It is often necessary to reduce
the thickness of the lintels fictitiously or even to remove them from the model.

Example of how to model a group of openings:

Example of a structure with close openings... which clearly cannot be neglected.

Case of steel construction. The CNC2M recommendations for the dimensioning of steel beams with openings in the
web according to NF EN 1993 states that an isolated opening with a maximum dimension less than 10% of the height
of the web of the beam is not considered significant. When modeling these openings, the same rule can be applied.
Nevertheless, this opening must be considered when verifying the resistance of the cross-section according to NF EN
1993.

In the cases where there are wall collaborations with diaphragms made of ribbed plates, according to EN 1993-1-3 §
10.3.4, small regularly distributed openings whose cumulative surface area represents up to 3% of the total surface
area may be arranged without any calculation of the diaphragm. It is doable as long as the total number of
connections points of the panels constituting the diaphragm is respected. Thus, from a modeling point of view, such
openings may not be considered.

C.2.3 Degree of simplification: curvature, slope, ...

When modeling the geometry, at the structural element scale, the curved elements will have to be represented as
accurately as possible, knowing that the meshing phase will discretize these curvatures by a succession of straight
segments, depending on the mesh size chosen and the nodes already defined. The part of the Eurocodes NF EN 1993-
1-6 dealing with the strength and stability of shell structures gives some indications on how to take curvature into
account.

For bridges, the effects of slope, curvature, and skew angle must be considered, and their non-inclusion in the model
must be justified. For common straight structures, since the slope is normally limited at the design stage, it can
generally be neglected. On the other hand, for curved structures:

depending on the level of the supports and the hyperstatic degree of the structure, the slope cannot be
neglected.
whether the structure is curved over all or part of its length, the centrifugal force and the slope must be
considered. It should be noted that the standard NF EN 1991-2 indicates that the centrifugal force, including
dynamic effects, can be neglected if the radius of curvature of the pavement in the plan is greater than
1500m.
curvature and skew angle create non-negligible torsional moments in the structure even when the traffic is
 transversely centered on the structure.

To establish an order of magnitude, a structure can be considered as being of low sensitivity:

to the skew angle when it remains greater than or equal to 70 degrees.

to curvature when the angle between two adjacent supports is less than 0.3 rad.

However, it is difficult to establish general rules and the reader is invited to consult the design guides specific to each
type of work (PRAD, PIPO, PICF, ...).

Illustration of the angle between two supports

The example of a girder grid "Modelling the same structure using different approaches" (in Part 3) illustrates the
effects of skew angle and slope on an example of girder bridges.

C.2.4 Degree of simplification: Alignment of structural walls of variable thickness

In the case of a building, a vertical alignment of the elements is recommended to ensure a simple load transmission.
However, various requirements (sheltered equipment, available space, etc.) can lead to certain offsets from one level
to another. A simplification of the geometry during the modeling can however be made (mainly to avoid excessively
heterogeneous mesh) by aligning the vertical elements and even the horizontal elements.

This simplification results in a good representation of the overall functioning of the structure provided that the
recommended constructive arrangements are respected. However, it is necessary to make a local verification of the
proper functioning of the transfer of forces and to reintegrate the actual offsets in this local verification.

Similarly, for steel structures, in the presence of footings or tubes of variable thickness (ferrule for example), a single
average plan is usually used.

For example, a tank composed of shells of different heights and thicknesses will be modeled with cylindrical surfaces
with:

an identical radius equal to an equivalent average radius.

appropriate thicknesses according to the height (shell thickness).

The value of the equivalent mean radius can be defined according to the Seismic Guide for Storage Tanks DT108. This
guide brings examples on how to determine an equivalent uniform thickness, which allows defining the value of the
equivalent mean radius = Internal radius of the ferrule + Uniform equivalent half-thickness (see example below):
C.2.5 The use of symmetries

As discussed in Part 1 A2. The dimensionality of the model, when the structure presents a plane or planes of symmetry
in its geometry, it can be very interesting to limit the calculation time and the size of the model by using this
symmetry and model only a part of the structure. Appropriate boundary conditions must be applied on the plane of
symmetry.

However, it is important to highlight the fact that the loading must also be symmetrical, and that the solution obtained
will be symmetrical (for example, the antisymmetric natural frequencies will not appear).

Example 1: 4-span symmetrical bridge

The spans are 60/100/100/60m long. The bridge is symmetrical with respect to its midpoint.

One could be tempted to model half of the bridge, by placing a symmetry support condition at the center (on the right
side of the figure, vertical translation blocked, rotation blocked):

Symmetrical load case:

In this case, the results are identical for both structures.

Case of asymmetrical loading:

In this case, there is a significant difference in the results:

The examination of the bending moment influence lines on the second support in both configurations provides an
immediate explanation:

Example 2: Foundation mat

Square-shaped foundation slab modeled with shell elements.

The nodes have 6 degrees of freedom (ux, uy, uz, rx, ry, rz).
Because of its shape, it contains planes of symmetry. The following plane of symmetry was chosen as shown in the
figure below:

The foundation slab can then be subjected to a loading that is either symmetrical or antisymmetrical.

For example:

if a bending moment is applied around the Y-axis, the loading is symmetrical.

on the other hand, if a bending moment is applied around X in the forward direction, the foundation slab turns
upwards on the side with nodes Y>0 and downwards on the side with nodes Y<0. The loading is then
antisymmetrical.

The conditions that must be applied to the nodes on the plane vary.

In the first case, the nodes located on the plane of symmetry will be:

free for translations ux, uz, and rotations around y.

blocked for translations uy and rotations around x and z.

In the second case (antisymmetric loading), the nodes located on the plane will be:

blocked for translations ux, uz, and rotations around y.

free for translations uy and rotations around x and z.

It is important to note that considering the different types of loading, in this case, leads to creating two models
differing only by the boundary conditions associated with the loading, which is not prohibitive.

In the case of dynamic calculations of a soil volume, particular attention should be paid to the lateral boundary
conditions of the block to correctly translate the conditions of non-reflection of the waves (see Part 1, chapter F.8). The
definition of these spring-damper element systems is outside the scope of this guide.

© doc Plaxis

Even if computing resources are very powerful nowadays, the use of symmetry remains an approach that can be
very useful for complex calculations and/or for large models. It presents several delicate aspects that need to be
understood.

C.2.6 Modeling of the foundations

Most of the time, the soil is modeled by support conditions (simple supports or clamps).

Before modeling the foundations and the soil in detail, the sensitivity of the structure to the flexibility of its foundations
must be assessed.

If the structure is sensitive, the soil must be considered:

either indirectly through elastic supports or stiffness matrices, the parameters of which should be calibrated
elsewhere.
or directly by modeling a certain volume of soil (soil portion + boundary elements). Note that this type of
calculation requires special software.

In the case where the reliability of the soil parameters is low and/or their variability is high, it is recommended to
perform a range calculation.

In some cases, having to model the structure with its foundations is a regulatory obligation. Refer to the NF-EN-1998-5
§6 standard.

For more details on boundary conditions, refer to § C.5 Boundary conditions.

C.2.7 Modeling of bracing by bars

Beware! In the case of steel structures, some very slender elements (braces or cables) can only work in traction.
If the modeling does not take this into account, the strength and stiffness of the bracing are overestimated for
both static and modal calculations.

Example of a simple braced structure

“As-built" modeling of a braced frame, but without considering the fact that the bracing bars will buckle as soon as
they are put in compression:
In this case:

the horizontal deflection is 4.4 cm,

the maximum force in the diagonals is 321kN.

Because of the buckling of compressed diagonals, for the overall behavior, one diagonal out of two should be removed,
ideally those that are compressed, but this is not a requirement:

In this case, the displacement increases from 4.4 to 7.6cm. Thus, the stiffness of this pier is divided by 7.6/4.4=1.73,
which may have consequences on the verification of deformations and the calculation of proper natural frequencies for
the seismic calculation (Error of the order of 1.730.5 =1.31).
Consequently, the efforts in the diagonals increase from 321 to 641kN, which is logically about twice as much.

C.2.8 Structural Zoom - Local Model

For problems related to the local functioning of an element or an assembly of a structure, it may be interesting to build
a localized model by imposing boundary conditions that reproduce the interaction with the rest of the structure.

This is for example the case for spacers of mixed slabs, support zones of complex structures, or an arch/deck
embedding in a bowstring bridge.

Sometimes the entire structure is modeled using beam elements except for a part modeled using plate elements. In
this model, which incorporates beam and plate elements, it is necessary to check carefully that the transmission of
forces from one to the other is carried out correctly (for example by ensuring the sufficient rigidity of fictitious
connecting elements). See C.6.7.

Global modeling with beam elements (pseudo-volumetric view)

Local modeling with plate elements (view of average surfaces)
$translationBooks

PART 2 – USEFUL ELEMENTS

C3. FE and meshing

C3. FE and meshing
C.3.1 Types of finite elements

First and foremost, the user of FE software must ensure that he or she understands the vocabulary used by the
software: bar/beam, plate/shell, surface/panel, etc...

Part 1 § A.3 is a theoretical part devoted to finite elements. Most importantly, the different types of elements are
described.

The user should consider consulting the manual of his software concerning each finite element to check its degrees of
freedom, the stresses and deformations it considers, and if it contains the activate/deactivate option.

The questions to be asked are the following. Depending on the problem to be solved, do you want:

the element to work under normal stress, in flexion, or both?

them to consider shear and the associated deformations?
them to deform in-plane or out of plane?

C.3.2 Mesh shape

The first part of this guide (Part 1 § A.3) gives details about the different possible mesh shapes for surface elements
(triangles, quadrilaterals) and solid elements, as well as the conditions associated with these different shapes.

Here are described only the rules to be followed once the type of mesh is chosen.

Most software have automatic meshers with many options to improve and customize the mesh.

The main advice is to look for the most regular mesh possible but to refine it where necessary.

In some cases, and depending on the software used, it is more interesting to manually create the mesh. This way,
regular meshes are obtained, whose numbering can be controlled, which facilitates the application of loads and the
exploitation of the results.

There are rules on the slenderness of the elements (the ratio between the smallest and the largest dimension must
be greater than 1/3) and on the distortion (respect of the flatness of the elements). Distorted elements can affect
the relevance of the results. For instance, for a non-linear calculation, if an initially highly distorted element is located
in an area of high deformation, the distortion of this element may become more pronounced, causing the calculation to
be interrupted because the limit criteria have been exceeded. Some software points this out. Moreover, there are rules
to follow concerning the angles or aspect ratio of the elements. Some software can test the whole mesh according to
this criterion, if necessary, by weighting it according to the relative surface area of the finite element.

Example of detection of the FEs that do not meet a given ratio criterion

The aspect ratio of a triangle is the value 2Ri/Ro, where Ri is the radius of the circle inscribed on the triangle and Ro is
the radius of the circumscribed circle. The closer the value is to 1, the better is the quality of the triangle. This is the
case for equilateral triangles. Conversely, when the area of the triangle is zero, the aspect ratio is 0.
 Illustration of the definition of the aspect ratio

Remember that a triangle is said to degenerate when its area tends towards 0.

In any case, one must look at the shape and the appearance of the mesh.

If the mesh does not look good, it is always possible to test another mesh option, to create nodes, or to cut elements
to improve the mesh.

The ratio between the smallest and the largest dimension of an element should be greater than 1/3
and the aspect ratio should tend towards 1.

Example: a parallelepiped of 160x160x160mm³ with one side containing a circle. The average mesh size should be
approximately 40mm except in the center of the circle where the average size should be 2mm. 1st mesh: the
modeling of the surfaces is carried out in an elementary way. In the first one, the circular surface is meshed with a
mesh size of 2mm. The remaining surfaces are then meshed with an average mesh size of 40mm (in general, the
surfaces are meshed in the order in which they are created). 2nd meshing: the modeling of the surfaces is improved.
The meshing is controlled starting from the central surface.

Surface creation order 1st mesh

Elementary surface modelling = Non-regular mesh + degenerate elements

Adjustment of the geometry 2nd mesh

Improved surface modeling + controlled mesh = Meshing and satisfying elements

 Example of degenerate mesh

A good mesh is always "aesthetic", it should not be visually shocking.

C.3.3 Mesh size

The objectives of the calculations must be kept in mind when determining the mesh size.

Firstly, in a model, one must distinguish the elements for which results are expected and the ones that are there to
reproduce the rigidity and mass of the structure.

For surface elements and for elements from which results will be extracted, it is usual to respect a mesh size
between 1 and 2.5 times the thickness of the element.

Larger mesh sizes can be adopted for elements where no results are expected.

The areas of particular interest in the analysis of the results and those likely to have a strong gradient of stresses and
deformations must therefore have sufficient mesh refinement and very few degenerated elements.

Example of refinement of a mesh in the corners of the building via an emitting point (refinement of the mesh on a
concentric approach) to apprehend the problems of thermal gradient in the floors:

It is important to ensure that the evolution of the mesh from one point in the model to another is gradual. When
moving from one area to another, the mesh should not vary too abruptly.

The size of the mesh must also be adapted to the capabilities of the software and the available calculation time. Before
starting the real model, it may be useful to produce a model with a simplified geometry (parallel or orthogonal sails,
absence of beams and shafts...) and to launch the calculations, to check that the software does not contain errors,
outputs the results within a reasonable time and that it can process the results fluidly, especially if multimodal
calculations must be conducted.

A sensitivity analysis (by dividing or multiplying the mesh size by two and comparing the results - see the next
paragraph on fineness testing) makes it possible to set the optimal size without mobilizing superfluous resources.

For linear calculations of 1D elements, the problem is smaller because the finite element results are derived from the
beam theory and do not depend on the mesh size. On the other hand, the display of the results can be misleading. A
typical rule is to have a discretization of the order of 1/10th of the span.

For non-linear calculations, it is usual to refine the mesh near the plasticization areas.

For soil modeling in seismic calculations, a mesh size smaller than or equal to 1/10th of the excitation wavelength
should be applied (see Part 1 § F.8).

C.3.4 Mesh refinement test

A test that is often performed consists of making two identical calculations on the same model, one with the
refinement of the mesh improved by a ratio of one to two. The main results given by these two calculations are
compared in the areas of interest.

This exercise allows the refinement of the mesh to be adjusted to the objectives of the analysis. As the calculation time
varies exponentially with the number of degrees of freedom of the model, the reduction in the number of elements can
be appreciable in terms of computer downtime and memory required to store the results, if it does not lead to a loss in
the quality of the results.

Conversely, it may be necessary to refine the mesh so that the results are valid, but generally, this refinement will only
be carried out on the areas of interest.

The mesh quality indicators provided by the software is related to the shape and distortion of the elements, not
to the relevance of the mesh size. The refinement test is therefore always useful, especially for large models.

It should be noted that there are a few software packages that have an adaptive meshing capacity according to loads
and deformations (this option is rather useful for non-linear calculations).

Illustration

Example of the impact that changing the mesh size has on the results of a floor slab analysis - from the top to the
bottom, mesh sizes of 20, 40, and 80cm, respectively. The maximum shear which is equal to 0.92MPa with a mesh size
of 20cm increases to 1.49MPa with a mesh size of 40cm and 1.22MPa with a mesh size of 80cm.
C.3.5 Orientation of the local coordinate systems

The orientation of the elements has an important impact on the post-processing of the results.

Verifying the local coordinate systems should ideally be done before introducing the model loads, as these might be
referred to as the local axes of the plates.

In the case of 1D elements, the X-axis of the beam elements is usually directed from the "origin" point to the "end"
point, with the Y and Z-axes being in theory positioned in any way relative to this X-axis. However, the position of
these Y and Z-axes must be homogeneous for elements of the same family, on the one hand, to facilitate the
application of transverse loads (e.g. wind load), and on the other hand, to read the extreme fiber stresses which are
defined by the Y and Z translation of the neutral fiber.

In most software packages, the local axes of the elements are oriented by default either with respect to the global
coordinate system of the model (alignment of the local Z with the global Z) or with respect to the order in which the
entities are created. It is always possible to force a homogeneous orientation on a set of elements.

Similarly, for 2D elements:

the outgoing normal must be known when defining load cases (earth pressure, fluids, or temperature fields).
it may be advisable to follow the logic of determining the outgoing normal, both for the input of the concrete
covers in the case of a reinforcement calculation, but also to direct the element beforehand according to the
assumed direction of the reinforcement to be installed (or checked). One will try to follow the same logic for
the whole model (e.g. upward normal for all floors) so that errors are not induced in the exploitation of the
results.
a uniform orientation also helps avoiding discontinuities in the display of stresses for a given fiber in two
adjacent plates, for example.

Example: Plate and local coordinate system of elements

Subject: the direction of plate definition can, for some software, generate the orientation of the local coordinate
system of the elements.

Example: Plate 6×6 m² (modeled with 2 plates of 3×6 m²), supported on 4 sides, loaded with 3 T/m².
 View of the local coordinate systems of the elements

View of the bending moments

Then, there is a sudden discontinuity of moments on the connecting line between the two plates. This discontinuity,
which has no real origin, is solely due to the change in orientation of the local coordinate systems.

In particular, the change of orientation of the local coordinate systems as shown above will be a real problem if the
software is asked to calculate average forces in a given cut-off point...

Check that all local coordinate systems have the same orientation.

C.3.6 Model size

The calculation time is often a determining factor in the cost of the project. Therefore, it is always interesting to try to
optimize this calculation time.

The calculation time of a model depends on many parameters:

the number of degrees of freedom (number of nodes x DOF).

the performance of the machine.
the software performance (algorithm, parallelization, ...).
the amount of data saved (temporary non-linear calculation).
the type of calculation (linear - non-linear).

Depending on the software, it is often possible to optimize the amount of data that can be saved and the number of
degrees of freedom.

On an ordinary project, a model will run at least twenty times. Any gain in calculation time is appreciable.

There is no need to systematically save the result files, especially if the model runs in less than two to three minutes.
These files only clutter up large CO2-generating clouds.

C.3.7 "Merge" or "Combine" option

Most software have the option to merge nodes or geometric construction points that are very close to each other
within a tolerance set by default or by the user. This avoids mesh discontinuities.

This operation presents certain risks, particularly in the presence of expansion joints or the absence of welds that the
model could ignore.

In the presence of joints, the user might choose between:

 representing the joint with its width (modeled distance between the 2 lines defining the 2 edges of the joints).
This is easily visible when manipulating the model and less likely to be "merged" by mistake afterward, but
this may lead to elements with heterogeneous sizes (associated with the size of the joint) if the ends of the
lines do not meet,
placing the points and lines in the same position in the model but modeling them independently. It is then
difficult to check that the joint is well represented (unless the node numbers are displayed later) and node
"merging" operations must be carried carefully.
use the linear release features offered by some software.

The merge operation may also impact the node links. Therefore, the mechanical links between nodes must be defined
after merging.

C.3.8 Group of elements (for visualization and later processing)

Most software offers the ability to define groups of nodes or groups of elements.

This feature is very convenient and facilitates the assignment of materials and masses, the application of loads, or the
post-processing of results by elements of the same family.

C.3.9 Reading points for results and meshing

The points where the results are read are a consequence of the verifications to be performed on the structure. The
needs of the study may require several points for calculating stresses on the same section (for example for normal and
tangential stresses).

The calculation mesh (i.e. the set of nodes) and the points where the results are read (sometimes different from the
nodes) should not be confused.

Having many reading points does not make the mesh necessarily sufficiently precise.

In the example below, the multiple isolines of transverse moments, especially near the supports, could make it seem
that the calculation is accurate, whereas the mesh is too large to obtain reliable results.

Indeed, the reading points may give the illusion of a refined mesh even if it is not the case. The results on these
reading points are interpolated from the results at the nodes.

By plotting the bending moment and edge diagrams in a cross-section, this becomes clear (the slab is seen from
below):
By refining the mesh, the graphs become:

As soon as there is a singularity, in this case, the support line, the size of the mesh plays an important role in the
accuracy of the results. You only need to refine the mesh to see this:
The calculation of the integral of the efforts shows a strong impact (in this example) on the shear efforts (deviation of
22%) and a very weak impact on the bending moment between the coarse and the refined mesh area.

Integral of the shear efforts - plate with two single support lines

Integral of the bending moment - plate with two single support lines

It is enough to create a singularity for the moment, by clamping the edges, for a deviation from the moment to occur
(of the order of 17%)

Integral of the bending moment - bi-clamped plate

$translationBooks

PART 2 – USEFUL ELEMENTS

C4. Modeling the non-structural

elements or equipment
C4. Modeling the non-structural elements or equipment
Non-structural elements and equipment are elements that do not play any structural role in the behavior of the
structure. Three cases can be distinguished:

The element has a low mass and a low stiffness compared to the overall structure. In this case, it can be
neglected in the model,
The element has a low mass but its stiffness cannot be considered negligible (e.g. some facade cladding
restraining the movements of the supporting structure). In this case, its presence can modify the behavior of
the structure, so it is necessary to consider it in the model (at least by considering a fictitious rigidity),
The element has a non-negligible mass but its stiffness can be considered negligible. In this case, an
equivalent mass must be included in the model.

Equipment/Non-structural element Negligible mass Significant mass

Reliable stiffness Not considered Mass to include

Significant stiffness Stiffness to model Mass and stiffness to model

For seismic analyses, it is important to ensure that the element will not resonate with the supporting structure. The
following publications are useful references to identify when the element/structure interaction should be considered in
the model.

J. Betbeder-Matibet - Génie parasismique - volume 3 - Hermes Science Publications (2003).

FD P06-029 - Règles de construction parasismiques - Dimensionnement des ancrages en zone sismique
(décembre 2017)

AFPS proposal to know when to consider the dynamic interaction between non-structural elements/equipment and the
structure (Recommendations AFPS90, 1993).

Legend :

Me=mass of the equipment or non-structural elements

Mb= mass of the supporting structure
Te= period of the non-structural element
Tb= period of the supporting structure

Note: depending on the direction studied, the mass of the supporting structure may be limited to the support floor
only (for more details, refer to the documents mentioned above).

For seismic calculations of buildings, non-structural elements (such as partition walls) must be considered when
modeling the structure as they are likely to modify its transverse stiffness. These elements may be subjected to
justifications such as in EC8-1, Article 4.3.6 for framing with masonry infill.

It may also be necessary to remove small elements in a model intended for modal analysis.
$translationBooks

PART 2 – USEFUL ELEMENTS

C5. Boundary Conditions

C5. Boundary Conditions
C.5.1 General case

The boundary conditions are determined by the degrees of freedom that can be fixed or not at the nodes located at
the boundaries of the structure (supports, cuts ...).

The notions of reference frames and boundary conditions are closely related here. The orientation of a boundary
condition (forces, moments, imposed displacements, accelerations...) will depend on the orientation of the current
reference frame when it is implemented or on the entity (node, element) upon which it is applied.

In what concerns the coordinate systems, a distinction must be made between the ones linked to elements and the
ones linked to the nodes.

Usually, by default, the node coordinate systems are identical to the global coordinate system. However, a different
coordinate system can be assigned to one or several nodes in order to be able apply a boundary condition to the
node(s) in a particular orientation. Simple graphical controls of the orientation of nodal loads (forces, moments,
imposed displacements) enable visualizing the correct definition of these nodal coordinate systems.

For the elements, the local coordinate systems are generally linked to the main orientations defined during the
creation of the element. Simple graphical controls of the orientation of the loads applied to the elements (such as
pressure) allow visualizing the good definition of these element systems.

The boundary conditions must ensure the stability of the structure with respect to the loads applied and be
representative of its overall behavior. Stability must systematically be reached in the 6 global components X, Y, Z, RX,
RY, and RZ, even if the loads are applied only to a few directions.

Stability following a translation is always achieved by blocking at least one node in the direction of this translation.

Stability about an axis is achieved either by fixing one node in rotation (along the axis of rotation) or by blocking two
nodes in translation in a plane perpendicular to the rotation.

It is important to check that the reactions in the fixed directions are zero in order to have numerical stability.

Example:

Objective: loading of a slab supported by 4 columns ;

The slab is modeled with solid elements (elements with nodes having 3 degrees of freedom UX, UY, and UZ),
Boundary conditions: fixing the nodes at the supports in the vertical direction UZ = 0.0,
Problem: some software will not perform the calculation because of instabilities,
Additional boundary conditions: 2 fixed nodes in the transverse direction OY + 1 fixed node in the transverse
direction OX.

Summary:

4 nodes fixed in the OZ direction – the 4 supports are fixed vertically, which turns the slab stable in
rotation/OX and rotation/OY,
2 nodes fixed along OY – stability of the slab in translational/OY and rotational/OZ,
1 node fixed along OX - stability of the slab in translation/OX.

We will check that at the nodes following UX and UY, the RX and RY reactions are zero.
 Example of the minimum stability of a Z-loaded slab

Be careful: the slab is loaded only in Z, but to ensure the stability of the calculation the translational motion in X
and Y and the rotation around Z must be blocked. In this case, 2 points were blocked in Y and 1 point in X. (NB:
stability in Z requires a minimum of 3 non-aligned points blocked in Z).

In a 3D model, an embedding means that the 3 rotations and the 3 translations are blocked.

In practice, without giving any choice to the user, the software can proceed in two ways to block the displacements:

A method that is always numerically stable, which consists in suppressing the DOF in the solution of the
problem because its value is zero (see Part 2 - A.1 - in this case, a 5×5 matrix is used).
A second method consists in numerically placing a very stiff spring in the fixed direction so that the
displacement is numerically close to zero. (see Part 2 - A.1 - in this case, a 9×9 matrix is used by adding
terms of very different relative values).

It is important to understand how the software proceeds to be able to detect possible numerical instabilities that can
lead to a halt in the calculations or unbalanced results.

C.5.2 Modelling different types of supports

Generally, the supports can be considered as fixed, except when the ground intervenes in the behavior of the
structure, i.e. when there is a static or dynamic interaction between the soil and the structure (SSI). For instance, to
model a soft ground generating differential settlements at the supports, or to define a seismic behavior.

Structures are based on three types of foundations: shallow foundations (isolated or spread footings), foundation slab,
or deep foundations (piles, wells, bars, rigid inclusions, ...).

Illustration of shallow foundations, foundation slabs, and deep foundations.

C.5.2.1 Shallow foundations – For supports under columns

Shallow footings are modeled using punctual supports. They can be modeled in 3 different ways:

Illustration: clamped support, hinged support, and elastic support

Clamped support: the 6 degrees of freedom are blocked.

Hinged support: movements are blocked and rotations are free.

Elastic support: the support is defined by 6 elastic springs (one per degree of freedom), or stiffness or impedance
matrices.

Stiffness of the elastic supports

To calibrate the stiffness of the elastic supports, the stiffness at the interface between the footing and the soil is
required. This interface stiffness (usually denoted kv) is taken from the geotechnical report.

The term Kz (static stiffness of the spring against a vertical force) is obtained by the formula Kz = kv x S, with S the
surface of the footing.

The terms KR are deduced from the formula KRi (in N.m/rad) = Ii x kv with Ii the inertia (in plane) around the axis of
rotation i of the footing.

For a rectangle, we will find the terms Iy=b.a³/12 and Ix=a.b³/12, a and b being the dimensions in plane.

VIEW IN PLANE OF THE FOOTING

Particular attention will be paid to the notions of long-term, short-term and seismic stiffnesses provided by the
geophysicist, which differ significantly.

C.5.2.2 Shallow foundations – For supports under thin walls

Thin walls are generally modeled by plate or shell elements. Two types of models can be defined for the support.

Method 1: By defining linear supports distributed under the thin wall. This case is similar to that of the columns (see
above) by distributing the punctual stiffnesses linearly along the wall, or to that of foundation slabs (see §5.2.3 below),
but considering only one direction of calculation.

Method 2: By modeling a single central support to recover an overall torsor at the foot of the thin wall for designing
the foundations. A rigid bar at the base of the thin wall is then necessary to distribute the forces. The support
conditions are identical to those of the columns (embedded or elastic or with stiffnesses).

In both cases, the units for the introduced stiffness should be verified to remain consistent with the physical units
provided by the geophysicist: are they N/m, N/m², N/m³? Is it also N.m/rad or N.m/deg? (or derived units: kN, MN...)

C.5.2.3 Concrete slab foundations

The support of the foundation slab on the ground is modeled by placing springs under its various nodes. The spring
stiffnesses are then calculated according to the same principle as for isolated footings. They can be differentiated
according to the loading zones. Be careful, the stiffness must be proportional to the surface area of the node (a
possible issue in the case of an irregular meshing, if this assignment is not automatic in the software).

Some software also offer "surface" springs on plates. In any case, one must verify using a unitary case the good
concordance between forces and displacements.

Soil modeling using a spring system

In reality, the horizontal stability of the foundation slab is ensured by the friction between the concrete and the soil,
and a possible lateral abutment. For the model, either horizontal surface springs under the concrete base or springs on
the periphery (punctual or distributed) will be chosen, depending on the internal verifications that must be carried out.

Caution: for models with horizontal surface springs, the horizontal stiffness is related to the friction between the
foundation slab and the soil with possible slippage. In the case of a shrinkage study, an overestimation of this
horizontal stiffness will artificially constrain the invert and may generate significant and unrealistic tensile
stresses.

C.5.2.4 Pile Foundations

Pile foundations can be modeled using 3 methods.

Method 1: each pile is modeled using a beam on an elastic soil (or elastic linear supports). The horizontal soil
stiffnesses depending on the characteristics of the soil layers (Kx and Ky) are generally determined by the geophysicist
(pay attention to the mesh and the concordance of the units). A vertical support is positioned at the base of the pile to
represent the point stiffness.
Method 2: (intermediate but little used) each pile is modeled by an elastic support (or matrix) characterized by 6
stiffnesses that are generally calculated by the geophysicist.

Method 3: It is possible to replace a complex foundation by its stiffness or flexibility matrix, which integrates the
overall configuration of the footing with the set of piles.

The foundation will be modeled in the general model by an elastic support at node A, whose 6 stiffnesses will have
been calculated beforehand from a local foundation model. See the example below.

Because of the coupling between horizontal displacements and rotations in deep foundation systems, there is no
apparent reason to neglect the cross-terms (non-diagonal) of the stiffness matrix. However, very few software allow
considering the whole matrix, which is a problem. It must be demonstrated on a case-by-case basis that the
"diagonalization" of the matrix has no significant impact on the stresses and displacements of the structure.

Methods 2 and 3 allow the size of the model to be limited, especially in the case of many piles. On the other hand, the
first method gives the stresses in the piles directly.

More generally, in global models, piles or groups of piles are modeled by springs (method 3). It is only for designing
the piles themselves that one may want to model vertical beams with springs (method 1). Specific software can also
be used.

In any case, it is important to remember to include in the Kx or Ky coefficients all the normative requirements such as
the group effect.

Finally, if the piles are calculated by an entity other than the one in charge of the superstructure model, iterations may
be necessary to achieve the convergence of efforts.

Example of calculation of a stiffness matrix.

Data: 1.20m diameter piles, Concrete Young Modulus E= 30000MPa, Ksoil= 6495(kN/m)/m, footing height 2.00m.

Geometry

Unitary results (1kN or 1kN.m) introduced

Results:

Displacements under the loads = flexibility matrix [S]

Stiffness matrix [K]=[S]-1 - (units : KN, KN.m, m and rad)

(The matrices are easily inverted using a spreadsheet).

An indication that the cross terms were not considered: a pile of height 10m and section 5.00×1.00m², E=32000MPa is
added to the previous model. The same stack is modeled with a 6-component elastic embedding, which are the terms
of the diagonal of the matrix above. A force of 1000kN longitudinal (respectively transverse) is applied at the top:

In this case, a deviation of about 13% on the longitudinal displacement (respectively 17% on the transverse
displacement) is obtained - the flexibility of the pile comes into play but it is identical in both cases. It is up to the
engineer to judge the impact of this possible simplification.

One of the topics that must be addressed in the general hypothesis report is related to the simplifications that will be
accepted or not for the SSI calculation. On the one hand, they are related to the consideration of the short and long
term modules of the soil (ratio of 2 on K) and of the concrete constituting the structure (ratio of 3 on E). On the other
hand, they are related to the constitutive laws introduced for the soil (linear, with several slopes, with a possible
plasticization step). Depending on the software used, wanting to integrate all these factors can be very complex,
leading to the assembly of several differentiated models and the need to carry out certain verifications manually
(replacement of plasticized springs by an equivalent force, for example). In some cases, it may be interesting to
perform a range calculation, distinguishing between resistance and deformation verifications.
 Illustration of the complexity of the constitutive law of soil springs

C.5.3 Modeling of support devices

There are two ways to model the support devices of a structure:

either as a support, with the degrees of freedom and flexibility parameters of the device. In this case, the
support reactions are recovered,
or as a bar element with several modeling possibilities (described below).

In any case, the support devices must be placed at the real location (transverse and vertical offsets), otherwise, large
errors will occur.

It should be noted that the behavior of the support devices can be a source of non-linearity (sliding supports, for
example) and may require a de facto non-linear calculation.

Example of a bridge pile model including the elastomeric bearings.

If bar elements are used, there are several possible choices:

some software propose elements resembling connections that can reproduce the characteristics of the
bearings (it is not strictly speaking a bar),
one can also use "spring" elements between two bars, if the software allows it,
one can define a "shear beam", i.e. a bar that is very rigid in bending but with a calibrated deformability to
the shear force. The software must necessarily consider the shear deformation (option not activated by
default in some software). It is defined for this bar a weightless section, an elevated inertia, a strong straight
section, and a section reduced to the shear efforts, allowing to find an equivalence of the support device (for
the GS/h bar ⇔ G'S'/T for the support device),
finally, we can use a classic bar, working in flexion, embedded in the bottom and free at the top. It is defined
for this bar, a weightless section, a section reduced to the shear efforts, and a calibrated inertia is then
defined to have a global deformation equivalent to that of the apparatus (for the 3EI/h3 bar ⇔ G'S'/T for the
support apparatus ).
Regardless of the method, the element must have an overall behavior equivalent to the characteristics of
the supporting devices, as defined by the standards for supporting devices (NF EN 1337 series) and it must not
introduce any moment neither in the deck or in the pile (other than that related to the thickness of the
device).

The illustration below shows rigid extensions allowing to model the top of a pile supporting two isostatic spans, simply
connected by a flooring.

Modeling details of a pile supporting two isostatic spans

$translationBooks

PART 2 – USEFUL ELEMENTS

C6. Connections - links – assembly

C6. Connections - links – assembly
C.6.1 Releasing the bars/springs/degrees of freedom

In all models, the default connection between two beam elements is perfectly fixed. However, it is necessary to be
able to release degrees of freedom on some connection nodes between elements (beam-column, sliding plate). Each
software have different functionalities to do so.

It is important to rely on the software's manual and to check, with simple load cases and with static or modal
deformations, that the behavior of the connection coincides with what is expected.

Care must be taken into maintaining the stability at each node such that not all the bars arriving at a node are
released in rotation or displacement.

C.6.2 Mesh continuity

Sometimes the number of elements on either side of the connection line (or surface) is not the same. Thus, there is a
risk that only the common nodes (in green) are considered a connection. (figure below).

Bad connection between finite elements of the same type and DOF

More commonly, a transition zone can be created using elements containing the same DOF per node with suitable
geometries (figure below).

The meshing of the transition zone

C.6.3 Connecting different types of elements

Using elements of different nature in the same model introduces complexity and one should always question the
necessity of mixing the elements. This complexity arises at the connection between elements of different natures.
They can be beam/shell, shell/solid, or beam/solid connections.

Particular attention will be paid to the possible connection of different types of structural elements: element with 6
degrees of freedom (UX, UY, UZ, ROTX, ROTY, ROTZ) / element with 3 degrees of freedom (UX, UY, UZ). This type of
connection can cause the appearance of instabilities or unexpected joints.

Several software packages compensate for these difficulties with specific elements capable of handling these links and
degrees of freedom problems. This should be checked and the relevance of the local behavior of the model should be
verified.

C.6.4 Connection between a bar and a plate

There are three cases:

either the beam and plate elements are in the same plane,
or the beam is a rib of the plate,
or the beam and the plate are perpendicular.

C.6.4.1 Coplanar Beam and Plate

For a bar element connected to two plate elements, the transfer of moments should be ensured by means of additional
elements, or by the introduction of stress equations linking the degrees of freedom.

In the illustrations below, in case 1, there may not be an accurate transfer of moments and nothing forces the bar to
remain perpendicular to the plate (intrinsically the shell nodes cannot block axis moments perpendicular to the plane
of the FEs). Case 2 consists of imposing an equation that links the displacements of the plate edge with the bar. This is
a reliable method, but it is not proposed by all software. Cases 3 and 4 consist of adding rigid bars to reproduce the
displacement dependence between nodes. Special attention must be paid to the definition of the rigidity of these bars,
which can be a source of instability in the software.

Connection of elements of different nature - Moment transfer

C.6.4.2 Connection between a bar and an out-of-plane plate

The case where the beam acts as a stiffener associated with the slab as in the case of ribbed slabs is discussed in
detail in C.8 Composite sections (beam/slab).

C.6.4.3 Connection of a bar perpendicular to a plate

The last case is that of the column-plate connection. The big difficulty, in addition to the transmission of bending from
the bar to the plate, is the transmission of torsion from the bar to the plate. By default, the plate does not have a
rotational DOF around the axis perpendicular to its plane, so it cannot take up the torsional moment brought by the
column. Therefore, it is necessary to find the right kinematic connection conditions. To ensure that the bending and
torsional forces of the bar are taken up by the plate, it is necessary to have rigid connections at the junction (in red on
the diagrams below).
 Modeling of the bar-plate connection (deck supported by a column, seen from below)

Example of the column (1D element) embedded in a plate and subjected to torsion.

Illustration of the consideration of the different number of degrees of freedom between elements of different nature.

The model simulates a 20 cm thick concrete plate on which a concrete column of 1 m diameter is "embedded", simply
by connecting the lower end of the column to a node of the plate:

The horizontal translations are fixed at the corners of the plate to block this torsion. The torque introduced is 10 MN.m.

The results are as follows:

Unfortunately, the numerical computations converge, but ... several aspects of the results can and should attract the
attention of the modeler:

the value of the rotation, both at the head and at the foot of the posts (54.2 radians!)

the presence of Mz moments in the corners while the supports are released in Rz
the sum of the reactions is not zero
finally, the value of the reactions Fx and Fy seems low (order of magnitude to be found: 10000 kN.m/7 m
(lever arm)/4 points = 360 kN).

It is enough to create an embedment using (fictitious) bars at the foot of the column, in the slab ...

... to obtain accurate overall results. (The local efforts at the foot of the column are of course disturbed by these fictive
bars).
C.6.5 Plate/solid and beam/solid connections

In the case of plate-solid connections, it is necessary to establish a connection to recover the embedding moment. As
in C.6.4, you can either create a connecting plate on the surface of the solid (on one or both sides) or extend the shell
inside the solid.

Modeling the shell-solid connection

The same reasoning is applied in the case of a Beam-Solid connection.

C.6.6 Stiffness Values / Stiffness Deviations / "Rigid elements".

Many software offer "rigid beam" or "rigid link" elements.

This type of element is sometimes a kinematic dependency (mathematical relation) between two elements and
sometimes a fictitious bar whose stiffness is very high.

However, the presence in the global matrix of the system of elements with large differences in stiffness can cause
problems of convergence. See the final example on matrix calculation presented in paragraph A.1.

These instabilities or numerical errors do not necessarily appear with an error message.

In most cases, it is advisable to use elements whose stiffness is defined by the user and to test the influence of this
stiffness on the global behavior.

C.6.7 Linking different types of elements: Structural Zoom - Examples

Insertion of plate finite elements in a global model

To understand the specific behavior of a particular area of a structure modeled using 1D elements, and to avoid having
to manage a model that is too heavy, it may be necessary to insert plate elements instead of the initial bar elements.
The connection between these two parts of different nature is made by means of links or rigid bars in "cobwebs".

Examples of connections of a model composed of beam elements with parts modeled in plates:
 Structural zoom

One may also wish to model only a part of the structure with plate elements and impose at the boundaries of this part
the displacements or the efforts at the nodes resulting from the global beam model (principle of structural zoom).
These displacements or efforts are then transmitted to the plate elements by rigid links made of beam-type elements.
These areas of connections between beams and plates must be modeled far enough from the area to be studied to
ensure that the efforts introduced by the rigid connections are correctly diffused to the studied zone.

For example, as part of the analysis of a connection zone between two RWB (Reconstituted Welded Beams), the zone
was modeled in plate and shell elements (see figure below). At the ends of the RWB modeled over a certain length,
torsors are introduced through rigid connections, with the structure being supported at the level of the lower plate. The
view below shows that these rigid links are located sufficiently far (about 2 m) from the area to be studied. It should be
noted that the plate is sufficiently rigid so that no rigid links need to be created.
The following example represents the structure of the girder of the extremity of a relatively wide bridge. The two
support reactions are introduced (on the right) under this girder, which is considered perfectly embedded in the deck
(on the left). They come from a global beam/plate model. The self-weight and dead loads on the girder itself are
modeled, if necessary. This approach simplifies data entry, as it requires only a few support reactions rather than the
complex torsors to be obtained at the interface with the deck due to the nature of the global model.

The global model with simplified modeling of a girder in an extremity.

A detailed local model of the abutment box

$translationBooks

PART 2 – USEFUL ELEMENTS

C7. Offsets
C7. Offsets
Most finite element software offer several options to define a beam on a fiber other than its neutral fiber.

This option is very useful, for example:

to easily create links between elements of different or variable heights (or thicknesses), but with an aligned
face (see below),
to define the geometry of a variable-height deck (it is easier to introduce a fixed upper surface and to
introduce a vertical distance between the upper surface and mean fiber that may vary during design),
for the application of loads on one side (for example on the upper surface of a bridge, for the creation of “load
trains").

However, it is necessary to define the offset between this defining fiber and the neutral fiber, this offset being the
eccentricity that can vary along the beam.

In case of doubt about the operation of this option, you can check it with another model where each beam is defined at
its neutral fiber by creating the offset yourself with rigid links.

The subject of offsets is partially illustrated in the document Example of Prestressing and Offsets.

Illustration of the offset of a series of beam elements with respect to a horizontal upper surface.
$translationBooks

PART 2 – USEFUL ELEMENTS

C8. Composite Sections

(Beams/Slabs)
C8. Composite Sections (Beams/Slabs)
Composite sections are made up of the assembly, rigid or elastic, of elements of different nature (wood, steel,
concrete, ...) and/or at different dates.

We study here the most common cases encountered in modeling:

building floors (slabs + beams),

bridges beams (prestressed, precast beams),
girders of mixed steel/concrete bridges,
mixed building floors (steel beams + reinforced concrete slab),

These elements complicate the calculation with different approaches depending on the case studied.

C.8.1 Floor beams of buildings

This chapter concerns the floors of buildings calculated from a global model.

C.8.1.1 Calculation principle

The difficulty with this type of analysis is to reconcile the finite element calculations with the design regulations for
reinforced concrete.

Indeed, the reinforced concrete regulations (BAEL and EC2-1-1 §5.3.2.1) are based on precise rules on the effective
flange widths, on the offset of the bending moment curves (which correspond to the formation of the connecting
struts), and on the deformation diagram (consistency between the deformations of the slab and the beam).

However, the finite element models are based on Strength of Materials and not on these regulations.

In any case, the calculation of the reinforcement must be performed:

by considering the direction of the slab's span (in particular the prefabricated elements),
by using the efforts from the model,
by correcting them to account for the effective flange widths (non-participating zones of the slab weigh but
do not add to the strength),
by correcting for additional eccentricities not modeled (vertical or horizontal - effect P-Δ),
by redistributing the bending moments,
by performing a regulatory calculation with these post-processed efforts

C.8.1.2 How to model the beam/slab floor

The first aspect concerns the floor modeling method. Indeed, several options are available:

to model only the beams, the loads being directly applied to the beams,
to model beams and slabs on the same average fiber,
to model the beams with an eccentricity with respect to the slabs.
 Illustration of the last two approaches

The comparison between these cases is made by using the following example:

A structure with 2 spans of 8m each, beams with a cross-section of 25cm x 50cm spaced 2m, and slab with a thickness
of 15cm.

Plan view of the slab

Cross-section

We will study the central beam:

Load = dead weight + permanent load (cladding) of 3 kN/m² + accidental load of 5 kN/m².

We study the ULS case (1.35 PL + 1.5 AL).

There are 3 types of modeling:

 Case 1: the slab is not modeled, which means that the beam is calculated according to the usual methods for
reinforced concrete
Case 2: the slab is modeled on the same average fiber as the beam
Case 3: the slab is modeled with an eccentricity with respect to the beams

Case study 1: modeling the beams only - the slab is not modeled

Modeling scheme

Bending moment diagram (kN.m) in the beam

The bending moments are equal to -264kN.m on central support and 149kN.m in the spans; they are consistent with
the classical calculation methods; this requires reinforcement of 19cm²on central support and 10cm²for the spans.

Case Study 2: the slab and the beam are modeled on the same medium fiber
 Modeling scheme

Bending moment diagram (kN.m) in the beam

The bending moments in the beam are equal to -166kN.m on the central support and 94kN.m in the spans; these
forces are much lower (-37%!!) than those calculated in Case 1: the calculated reinforcement is only 11cm² at the
central support and 6cm² for the spans.

Longitudinal bending moments in the slab

The bending moments of the slab are equal to -21kN.m/m on the central supports and 12kN.m/m in the spans, which
results in a reinforcement on the central support of As = 5cm²/m, Ai = 0 and As = 0, Ai = 3cm²/m in the spans.

Case study 3: the beam is off-center with respect to the slab

 Modeling scheme

Bending moment diagram (kN.m) in the beam

Axial effort diagram (kN) in the beam

The bending moments in the beam are equal to -47kN.m on the central support and 21kN.m in the spans, but they are
accompanied by axial forces (tension on supports and compression in the spans); the calculated reinforcement is then
10cm² in the upper layer and 2cm² in the lower layer on supports, while there are no steels in the spans!!!
 Longitudinal bending moments in the slab

Longitudinal axial efforts

The bending moments in the slab are equal to -17kN.m/ml on the central supports and 9 kN.m/ml in the spans, they
happen simultaneously with normal efforts with peaks on the central supports.

The analysis shows that case 3, with the offsets, is unusable and incompatible with the normative verifications because
normal forces and peak forces appear in the slab. Indeed, how can the bending moments be redistributed considering
normal forces?

The tables below summarize the main results.

Required reinforcement in the central support and the spans for the 3 calculation methods
Cases 1 and 2 result in similar reinforcement areas, which seems to validate the modeling of the beam and the slab on
the same mean fiber, but the conclusions of this example should not be generalized. Indeed, as shown in the
deformation diagrams of the sections below, there is an inconsistency in the spans with tensioned rebars in the slab,
located at the level of the compressed zone of the beam.

This example shows that:

modeling only the beams alone gives good results, but this option is difficult to apply in a global model (how
to transmit horizontal forces for example?),
modeling an offset between beams and slabs allows good modeling of the floor stiffness, but it is not
compatible with the normative verifications (how to redistribute the bending moment diagram when part of
the bending moments appear as normal forces in the beams?),
the reinforcement of beams and slabs should not be calculated directly from the results of the global model.

C.8.1.3 General method for the design of a beam/slab floor

The calculations of slabs and beams must, on one hand, consider all the efforts calculated in the global model and, on
the other hand, respect the normative requirements.

Let us take the example of a building subjected to horizontal forces (wind, earthquake, thermal, etc...):

Step 1: Create a global model of the building. This global model allows the calculation of the forces in the diaphragms
formed by the floors, which results in the appearance of membrane efforts (normal and shear efforts) in the horizontal
elements. These are the forces that we will use for the rest of the calculation: Nxx, Nyy, Nxy in the slabs, and Nx in the
beams.

Step 2: Create a local model of the slab. Indeed, except in very particular cases, it is not possible to use the global
model of a building to justify the slabs for several reasons:

phasing is generally not modeled,

pre-slabs are generally not modeled,
the position of loads in a global model does not necessarily respect the zones of influence of beams and slabs
at the local scale,
from a regulatory point of view, punching, stress redistribution, stress discontinuities, etc., are not accounted
for.

In the local model of the slab, its geometry is extracted from the geometry of the global model. For the sake of
simplification, beams are generally replaced by linear supports, while slabs are modeled by shell elements (bending)
subjected to weighted loadings. It is this small and flat model that will be studied for the normative verifications,
possibly considering the phasing, pre-slabs, etc...

The bending moments (Mxx, Myy, Mxy) in the slabs, resulting from this local model, must be cumulated with the
normal forces (Nxx, Nyy, Nxy) of the global model to calculate the reinforcements and carry out the normative
verifications (pay attention to the combinations).

Step 3: Create a local model for the calculation of the beams. Indeed, for the same reasons as for the slabs, it is not
possible to use the global model to determine the totality of the stresses in the beams.

The geometry is identical to that of the local model of the slab, except that the beams are of course preserved.
In this model, the slabs do not have to take up bending forces, they play the role of load transmission to the beams,
therefore, they are modeled by distribution surfaces (refer to the documentation of the software used).

The resulting efforts in the beams must be added to the normal efforts of the global model, which allows proceeding
then to the normative verifications on the beams (either manually or using dedicated software).

C.8.2 Case of Bridge structures (ribbed slabs)

This approach applies to bridges such as PRAD, VIPP, ...

For the calculation of bridge structures, if the Guyon-Massonnet method is omitted, it consists of calculating the
structures:

in girder grids, i.e. crossing longitudinal bars, representing the section of the ribs + the effective flange width,
and of transverse bars, modeling the slabs: the advantage is that we directly have torsors that can be used in
the calculation of reinforced or prestressed concrete, the disadvantage may be the placement of loadings,
especially moving loads,

Beam grid model

in beam grids using ladder beams - can be advantageous for a phased calculation, especially if one wants to
model in detail creep or shrinkage effects
as in the third approach above (C.8.1.2), by modeling the ribs with beam elements and eccentric slab in the
form of a FE shell: the main advantage of this approach lies in the easy application of the loads, the
disadvantage is that one does not directly obtain torsors that can be used in reinforced concrete calculations.

It should be noted that the beams modeling the slab must be perpendicular (or almost perpendicular) to each other for
the model to be valid.

To illustrate this approach, in particular the reinforcement of the slab and the ribs, one must start from the example
discussed in C.8.1.
 View of the model - 25cm x 35cm drop-beams and 15cm thick slabs. Spans 2 x 8m - beam spacing 2m.

The calculation of reinforcement directly from a reinforcement module is not recommended if the assumptions used
for the design of reinforced concrete are to be considered. A small post-processor (a spreadsheet) is enough to
calculate the bending moment and the flange width assigned to the rib, as shown below.

Application to the central beam of the model - section on the support and section on the spans (note: both spans are
fully loaded, without considering the influence line):

Bending moments (only the drop-beams) - kN.m

Normal force (only the drop-beam) - kN

The methodology consists of applying the plane section remain plane assumption and calculating the (elastic)
equilibrium of the internal forces. 1. The stresses diagram is extended to obtain the stress on the upper fiber (slab top)
2. The normal stresses on the composite section are zero: the integration of the normal stresses must be equal to zero,
the effective flange width of the slabs is deduced 3. All the geometrical and stress parameters are determined, then all
that remains is to calculate the bending moment resulting from the stress diagram.

Application to the case of the supported section:

A reduced flange width (47cm) is observed, which is logical considering the shear drag effect. As=14.4cm² (ULS
calculation).

Application for the span section case:

A larger effective flange width (139cm) is observed, which logically is larger than on the support. As=7.20cm² (ULS
calculation).

If an automatic calculation is performed for the rib and the slab:

→ The software suggests reinforcement sections at locations that actually do not require them when performing a
"manual" reinforced concrete calculation (it has been verified in parallel that no compressed steel section was
necessary).

→ In the present case, the automatic approach leads to a slight reduction in reinforcement at the bottom layer in the
span and an increase of reinforcement at the upper fiber.
Mapping of the reinforcement in the upper fiber of the slab

Reinforcement in the upper fiber of the slab at the central support, central beam (18.36cm² along 2m)
Reinforcement in the lower fiber of the slab

The efforts calculated using the beam grid model are as follows and lead to 7.2cm² in the spans and 15.4cm²
on the central support:

Beam grid model (Mt = 140kN.m, Ma = -262kN.m) - without taking into account the effective flange widths

To conclude the above example, we realize that the automatic calculation of reinforcement is not satisfactory. Without
considering minimum reinforcement, it leads to placing reinforcement in zones where the normative verifications
would not require it and to under- and over-reinforcing some areas. Besides, let us recall once again that the automatic
calculation does not consider the redistribution of the bending moments diagram, punching, or connecting struts... It
is, once again, up to the engineer to analyze the results and to decide if they need to be considered or not. It should be
noted that the problem of stress or reinforcement smoothing arises also when using slab or shell type finite elements.

For a beam grid calculation, one should refer to the SETRA Guide "Advice for the use of beam grid programs" -
PRP 75 - a particular area of focus is how to consider the torsional inertias.

Other examples are given in Example C - Modeling of beam grids.

C.8.3 Mixed Steel-Concrete Beams and Slabs

Generally, the composite character of the sections is modeled. However, in some cases, the model may be limited to
the main beam alone, without considering phasing, such as for pre-dimensioning. After calculations, the stresses of the
steel beam are then used to dimension and verify the behavior of the composite beam according to the appropriate
normative reference frame. The model does not detect that it is a composite beam and there is a small error on the
stiffness, the acceptability of which must be evaluated.

When performing a normative verification of the beam, the mixed character and construction phasing should be
considered.

Modeling Approaches:

For a more rigorous calculation, it is possible to model the composite beam:

either as a beam whose mechanical characteristics consider the connection of steel and concrete. This way,
the difference in Young's moduli of the two materials is apprehended via an equivalence coefficient - in this
case, it is said that the materials are homogenized, generally by reducing the concrete to a metal equivalence
(a),
or as two superimposed beams, one, lower, metallic, the other, upper, made of concrete, at the altimetry of
their respective centers of gravity. These beams are connected at their ends by rigid links. This can make it
easier to consider differential shrinkage and creep. If the structure is modeled as a whole, the longitudinal
concrete bars described above are, besides, connected by transverse bars in a way that forms a beam "grid"
(b),
finally, one can also choose to replace the beam elements of the slabs by shell-type finite elements (c). In this
case, the calculation of the torsors requires post-processing, ideally automatic, following the method
described in C.8.2.

The approaches (b) and (c) should be applied for special cases. Indeed, although they may initially seem simpler, the
pre and post-processing are always much longer than with a model of type (a), especially if a software dedicated to
mixed calculations is used.

In order to account for phasing, one must, at each change of state, either modify the inertia of the homogenized beam
(a) or activate the beams of the slab (b) or the shells (c), once the concrete has poured and the formwork is removed
for example. Of course, the creep of the concrete must be considered, either by means of an equivalence coefficient or
a creep law and cracked zones.

Example of a mixed structure modeled according to approach (a):

Example of a mixed structure whose slab is modeled with eccentric shell elements - approach (c). The steel beams are
modeled in this case strictly according to the material distribution (metal only):

The use of software specifically developed for mixed calculations is always recommended whenever possible.

See in Part 3, example B - Mixed and Steel Beams.

C.8.4 Mixed floor (building)

A compound floor is composed of steel beams supporting reinforced concrete slabs (pre-slabs or not) or steel deck.

C.8.4.1 Weight and vertical loads: slab bearing direction

Stress calculations in a composite floor are performed by considering the resistance of steel beams alone. Concrete is
then considered as a non-resistant dead load. More generally, these floors are made of collaborating steel deck that
works in only one direction. These particularities require specific provisions in the models.

In the case of simple geometry, the concrete slab is not modeled and the loads are applied directly to the steel beams.
When the geometry is complex, the manual distribution of the loads on the profiles becomes too delicate, it is then
necessary to distribute the loads using distribution surfaces. The most common software have this type of element
which behaves like a very thin plate, without any bearing role but distributing the loads on the load-bearing beams.
Alternatives allow to consider the directions of distribution of the steel deck, but the modeler must pay attention to
respect the load-bearing directions, the verifications are essential. Let us take the example of the petals of the LUMA
foundation in Arles made up of mixed floor with steel deck, they are represented in blue in the scheme below ...

The blue distribution surfaces are meshed like slabs but do not contribute to the strength of the structure.
Plan view of the floor with the load-bearing direction of the steel deck

Visualization of beam loads calculated directly by the software

The calculations are then carried out classically.

The calculation of the slabs is then done by a specific slab calculation (orthotropic) between beams.

C.8.4.2 Horizontal loads (wind, earthquake)

In general cases, the concrete slab is not connected to the structure, so it does not participate in the bracing of the
floor. The floor is then braced by horizontal metal bracing.

However, in some complex cases, it may be necessary to brace the floors using the concrete slab. The modeling then
becomes very complex:

the slab must be modeled with an eccentricity with respect to the average fiber of the profiles
it is necessary to model the connectors between the slab and the profiles
the slab/column connection node is different from the column/beam connection node
Mixed floor without slab modeling - Bracing provided by the profiles

Bracing provided by the slab

$translationBooks

PART 2 – USEFUL ELEMENTS

C9. Materials
C9. Materials
Part 1 Chapter 4 is dedicated to civil engineering materials and their specificities.

Defining materials is a rather simple task when modeling because most software have predefined material laws that
follow one or more codes.

These laws correspond to a simplified "curve" of the material’s stress-strain behavior, which is considered linear elastic
and incorporates safety reduction coefficients (on the modulus and the limit strength). Some very specific problems
may require introducing a more complex curve (Sargin's law for example), which is allowed by most software.

When using the predefined laws for concrete, one should be aware that the Young modulus is generally by default the
short-term modulus. For long-term effects, for some thermal and seismic calculations, it will be necessary to correct
the modulus. This is also the case for phased calculations where the modulus varies according to the age of the
concrete.

Similarly, the software considers a default value of Poisson's ratio. Generally υ=0.2 for concrete and υ=0.3 for steel.
Some codes require that a coefficient of υ=0 is taken for concrete at ULS. In particular, see BAEL and §3.1.3 (4) of
Eurocode 2.

For all Strength of Material calculations that require considering the concrete cracking (seismic, second-order, mixed
bridge slabs ...), the moment-curvature law of the cracked section must be considered. It represents the weakening of
the section and the actual stiffness of the structure. Sometimes, the code provides simple rules for the adaptation of
inertias. This may require iterative calculations, first in uncracked inertia to determine the cracking zones, and then
taking into account the cracked inertia.

It should be noted that some software allows to directly consider cracked inertia.
$translationBooks

PART 2 – USEFUL ELEMENTS

C10. Specific behavior in shear

and torsion
C10. Specific behavior in shear and torsion
In general, it should be noted that beam element models do not systematically consider shear stress deformations, nor
do they adequately account for torsional deformations.

However, in the case of modeling a structure that is sensitive to shear and torsion, one must activate the option to
consider shear and torsional deformations and to clearly define the reduced cross-sections and torsional inertias.

It can also be noted that the phenomena with blocked torsion are impossible to model in beam-element structures
because the beam elements of Strength of Materials are built on the assumption of conservation of straight sections
(without distortion or buckling) and yet, their consideration leads to stress distributions different from those calculated
in "classical" Strength of Materials.

Considering the blocked torsion will generally require the separate modeling of all the plates constituting the thin
profile of the section.

Here are some examples of structures sensitive to these phenomena:

for shear: slender welded beams (mixed double girders, for example), console-type structure (bracing walls
with a low height/length ratio). Failure to consider the shear deformation will result either in an
underestimation of the deformations leading to an erroneous deflection or an overestimation of the
stiffnesses,
for torsion: structures not free to distort (at one or several points).

Comparing calculations of the angle of rotation of a cantilever I-beam

Data - cantilever beam:

Boundary conditions: fixed in x=0 (θ=0, θ'=0) and free in x=L (B=0, T=0)
Length: L = 1 m
Loading: torsional moment at the end x=L: Mx = 10 kN.m
Cross-section: Welded ht = 200 mm, bt = 200 mm, tf = 20 mm, tw = 10 mm

Plate element model:

Loading:
Reaction:

Displacement θ(L)=0.042 rad

Beam-element model

Loading:

Reaction:

Displacement θ(L)=0.1198 rad

Analytical calculation

The differential equation for the angle of rotation is given by:

With the boundary conditions given in the previous paragraph, the solution of this equation is:

With:

It: St Venant torsional inertia

Iω: sectoral inertia
Mx: torsional moment
L: length of the beam

Application:

L=1m
G = 80,770 MPa
E = 210,000 MPa

(calculated by the software)

(calculated by the software)
Mx = 10 kN.m

The analytical calculation and the surface element model give the same rotation result θ(L)=0.042 rad.

The beam element model calculation gives a result 2.85 times higher.

In the beam element model, the stiffness due to the buckling inertia is not taken into account for the calculation of the
rotation angle:

Conclusion

In general, for beam element models, the stiffnesses due to the torsion of an open section beam are not considered
properly in the calculations.

In case of any doubt, a shell-type element approach on a simplified, global, or local model can help identifying the
effects.
$translationBooks

PART 2 – USEFUL ELEMENTS

C11. Modeling the loading

C11. Modeling the loading
As explained in chapter E, it is important to always verify for each loading case (permanent, accidental, and thermal),
by computing manually the summation of the loads, the global torsor of the loads: reaction forces and moments. (Most
of the software enable outputting the global torsor).

C.11.1 Thinking about the test load cases

It is important, as soon as the modeling is completed, to plan test load cases that allow validating the good overall
behavior of the model.

For example, unit forces uniformly distributed in the 3 directions, unitary punctual loads, or even temperature
variations. For these load cases, the deformations (usual orders of magnitude in civil engineering, kinematics, or mesh
discontinuities, ...) and the reactions at the support (all forces supposedly applied must be found in the reactions) will
be examined.

Therefore, these tests enable verifying the stiffnesses, boundary conditions, and internal connections. They can also be
used to verify more complex load cases (order of magnitude of the effects).

C.11.2 The case of dead load

Most software automatically consider the self-weight deduced from the cross-sections of the elements and the
volumetric weight of the material.

This direct method must be systematically verified. Indeed, the geometrical simplifications necessary for modeling
may induce a self-weight that differs from the one calculated with the drawings. A rigorous manual approach of the
measurements should not lead to a discrepancy of more than 5%. In any case, any deviation must always be justified.

As soon as the net cross-sections deviate from the gross cross-sections of the structure (e.g. if cracking, shear drag, or
oversize are considered), the self-weight considered as dead load must be redefined, so the automatic option should
not be used. Another option would be to modify the characteristics of the materials that are being considered, element
by element.

C.11.3 Surface loads and linear loads

Surface loads are generally applied to the average fiber of the plate element. When considering the intensity and
perimeter of the surface load, the diffusion of the loads up to the average fiber, including possible diffusion through
the thickness of the concrete cover, must be taken into account.

Furthermore, one must verify whether the surface load is applied according to the normal of the element (local
reference) or according to the general reference of the model.

Some loads (e.g. snow) are defined according to a reference surface (the horizontal for snow), which must be
considered when applying to surfaces that are not parallel to this reference plane (e.g. a sloped roof for snow). Often in
software, it is necessary to explicitly specify whether a load is projected or not.

Other types of loads (wind, hydraulic pressure, ...) are always perpendicular to the surfaces.

Finally, the orientation of the loads should always be verified as well as the deformations and reactions at the
supports, just like for the self-weight case.

See the examples below.

Example of the nozzle with earth thrust loading

Problem: model a linear load on an inclined surface.

The software offer options when defining the load, which are sometimes not very explicit.

Case 1: Load defined in the user global reference frame

The load applied to the bar is defined as a horizontal load per linear meter of the element.

Case 2: Load defined in the projected reference frame

The load applied on the bar is defined as a horizontal load per linear meter measured perpendicular to the direction of
the load (vertically in this case).

Tip: Always check, on a simple example, that the option used corresponds to the desired load model.

Linear loads are also affected by these problems of diffusion and coordinate system.

Note: in the case of earth thrust, the best modeling technique is the second one, i.e. a projection of the loads on a
vertical plane.

C.11.4 Thermal loads

Thermal loads are made of two types of loads:

Linear variations: a very common special case is the uniform variation,

Temperature gradients that result from a temperature difference between the extreme surfaces of a
structure.

It is essential to use test cases to verify the correct consideration of thermal phenomena in coherence with the
clamping of the structure.

It should be noted that thermal loads only create efforts (or stresses) if the structure is not free to deform (clamping,
hyperstatic structures, ...).
Concrete cracking can play an important role in stress distribution (see 11.5 below).

C.11.5 Shrinkage and Creep Modeling

In the absence of a specific software option, concrete shrinkage modeling can be performed by applying equivalent
thermal load cases.

Creep modeling can be performed by applying thermal load cases or reduction of the elastic modulus of materials.

It is important to verify that the imposed deformations are consistent with the expected phenomenon.

If there is any doubt, it is always possible to perform range calculations, to frame the short and long term (case of
compound bridges, foundations, ...).

Detailed description: modeling of shrinkage in compound bridges

Link to shrinkage modeling in compound bridges

C.11.6 Live loads

Understanding the concept of influence lines is fundamental for a good apprehension of the positioning of convoys, it
avoids designing for too many load cases.

In the case of complex structures, the concept is not easily applicable, however, influence lines can always be
generated by placing unitary forces at different nodes of the structure. After post-processing the results, with a
spreadsheet, for example, both surface and live loads can be positioned to produce the most undesirable effect.

The codes frequently define load models that combine loads of different natures with concomitance rules and specific
geometric configurations. They should be read fully and carefully. It then allows, thanks to the influence line, to apply
the loads at the position that is most undesirable for the studied effect (deflections, forces, ...).

Loading according to the influence lines.

Case of distributed loads that can be broken up and of convoys with variable vehicle spacing.

To our knowledge, all the regulations require to load the structures along the influence lines. Common practices or the
phase in which the project is in (preliminary studies, pre-design, or even Executive design) can lead to simplifications:
loading of two adjacent spans, loading of complete spans alternately ("one out of two”) ...

Link to the ECAM website

In the case of engineering structures, more specifically for Execution design calculations, loading by influence lines is
mandatory, and it is not enough to load complete spans or to make vehicles drive uniformly close to one another.

It is therefore necessary to make sure that the software used is capable of performing influence line (IL) calculations,
i.e. adapting the loaded lengths or adapting the number and spacing of vehicles in a convoy to obtain the most
unfavorable situation for the desired effect, for example, the bending moment, the reaction of support or the
deformation, ...

1. Eurocode uniformly distributed loads. We are looking for the maximum shear force (i.e. positive or negative)
at mid-span of a bridge with two equal spans (2×25 m) and a constant section. We know that the influence
line of the shear force at mid-span has the following shape:

For lack of a better solution, one might be tempted to fully load one or both spans.

The diagram below shows, for a unit load of 10 kN/m, the shear envelope for the cases:

Span 1 loaded
 Span 2 loaded

Spans 1+2 loaded

At mid-span 1, one gets |V|max = 31.3 kN:

The diagram below shows the same shear diagram, but with a beam loaded according to the IL:

Zone IL upper curve

Zone IL lower curve

One gets |V|max = 53.7 kN, which is a significant difference.

The exercise could be repeated for all sections.

The latter is particularly true for the loadings in fascicule 61 title II, which we must still use when recalculating for
example:

Distributed charges A(L) similar to uniformly distributed loads, but which have the characteristic of varying in
intensity according to the loaded length L,
B or Mc convoys, whose spacing can vary, sometimes with a required minimum distance. Convoy Bc is
described below.
2) Illustration on the previous bridge for the case of convoy Bc for the reaction bending moment

The influence line of the reaction moment looks like this:

A refined study would have to be carried out to find the precise position of the trucks, but it can be easily observed
that (here for 25m spans - reminder) the trucks must be separated to obtain a maximum effect:

Application: we run two convoys on the bridge, the first one with the two trucks very close to each other, as drawn in
the codes, and the second one with a distance of about 28.80m (determined graphically).

The results:
Trucks very close to one another, envelope, and unfavorable position.

Separate trucks, envelope, and unfavorable position.

The difference concerning the reaction moments is about 13%. However, the load case used to obtain the maximum
reaction bending moment is not the one to be used for the spans.

Practically speaking, one quickly realizes that finding the loaded lengths and/or positions and the spacings of trucks,
for all sections and all values of interest, is incredibly difficult by hand. Thus, using a software becomes evident - again
for Executive desing level calculations. For the other phases, simplified calculations, with a certain margin on
justifications and quantities remains possible by studying certain judiciously chosen sections of the structure, but this
is beyond the scope of this document.

Set of load cases and results for distributed loads. Span 1, span 2, and spans 1+2:
Loading according to the IL:

C.11.7 Thrust modeling and land abutment

Generally, the actions generated by the soil (thrust for example), water pressures, or seismic actions are modeled by
loads. The reactions (pressures on the ground, which can go up to a plastic threshold, the abutment, ...) are
represented by linear or non-linear springs.

Linear seismic approaches are allowed if the foundation uplift is limited to 30% of the foundation surface. Be careful:
no reaction forces are applied on a face blocked by springs... we let the springs do the work.

Note: a displacement approach is also possible to model the thrust loads and can lead to a reduction of the overall
loading (see AFPS / AFTES Guide. GUIDE "Conception et protection parasismiques des ouvrages souterrains").
$translationBooks

PART 2 – USEFUL ELEMENTS

C12. More about solid elements

C12. More about solid elements
The principles stated for 2D modeling remain applicable in 3D, in particular, the use of isoparametric elements
(parallelepipeds rather than tetrahedrons) is preferable, which requires a heavy and rigorous preparation of the basic
geometry.

Automatic meshers should be used with caution.

Example of a bridge deck modeled using solid FE:

(Taken from Part 3 - Example C - Modeling of girder grids).

$translationBooks

PART 2 – USEFUL ELEMENTS

C13. More about non-linear

calculations
C13. More about non-linear calculations
This paragraph deals with non-linearities related to the laws of materials and the so-called geometric non-linearities.

It is common to associate nonlinear calculations with complex structures such as cable-stayed and suspended bridges,
yet this subject appears in the daily life of any structural engineer, for example in the case of:

a partial detachment of foundation footings,

the inability of certain bars to work in compression (see § C.2.7 about bracing),
a buckling calculation in reinforced concrete,
cases beyond the field of beam theory (for example, the calculation of stresses in a bridge bracing under its
own weight).

In general, for all non-linear calculations, it is important to perform a linear calculation before accounting for non-
linearity to understand the behavior of the structure and the specific effect of non-linearity.

C.13.1 Theoretical Geometry and Imperfections

Most standards require that non-linear calculations incorporate an initial imperfection in the geometry of the structure
or the elements implantation. Some software can directly integrate this imperfection. For others, it will be necessary
either to apply a load case that creates the initial imperfection, or to define the geometry with the defect.

It can be observed that in a finite element calculation, the use of triangular elements always allows us to consider the
pre-deformation of a flat surface.

C.13.2 Ropes and cables

The ropes and cables are essentially non-linear elements since they operate only in traction and because of the chain
effect, an "apparent" Young's modulus must be considered. This modulus is a function of the tension, length, density,
and gross Young's modulus of the cable.

In earlier phases of the project, it is not imperative to take these effects into account. It is possible to model the ropes
using a bar, ideally bi-articulated, making sure to neglect the dead weight of the bar or to apply it manually directly to
the edges. It should then be verified in the analysis that these bars are not compressed.

C.13.3 Zones with material non-linearity

A first linear calculation is used to identify the areas where non-linear behavior will appear. The calculation will
continue by successive iterations, progressively integrating the non-linearities.

C.13.4 Buckling and large displacement calculations

Buckling - calculation of critical coefficients

Most of the software can determine the critical buckling loads of the compressed bars (i.e. the buckling lengths of the
bars composing a structure) from a modal calculation, in small or even large displacements. The calculations must be
performed for each combination. Many software also allow to carry out the normative verifications from this calculation
of critical loads (or simply by manually entering the buckling lengths).

The calculation of critical coefficients is based on the search for αi values such as the determinant Det([Ko]+αi
[Kσ])=0, where Ko is the stiffness matrix associated with small displacements and [Kσ] the stiffness matrix associated
with initial stresses.

The objective here is not to develop all the possibilities offered by the software, but to insist (once again) on the fact
that the modeler must understand what a given software does and what is the impact of the modification of the
calculation parameters. A simple parameter may be the required subdivision of the bars to obtain the right results, as
shown in the example below.

Illustration on the braced gantry in Chapter C.2. Link to the example of the calculation of critical buckling coefficients.

This small example confirms that it is necessary to master what the tool does. Moreover, when more challenging
calculations are being performed (not linear elastic, or first-order), one should always refer to simple examples from
the literature.
Modal displacements

Calculations in large displacements:

These calculations require updating the stiffness matrices at each iteration, whether in reinforced concrete or steel
structures. What was highlighted here before for the calculation of critical buckling coefficients with respect to the
control of the software parameters, remains perfectly applicable.

We refer to two interesting articles on the subject:

- Calcul au flambement des arcs - Comparaison entre un calcul approché et un calcul en grands déplacements du
Bulletin Ouvrages d'art n°32“ - Lien vers l'article.

- Instabilité par flambement des arcs (CTICM) - Lien vers l'article.

$translationBooks

PART 2 – USEFUL ELEMENTS

C14. More about prestressed

concrete
C14. More about prestressed concrete
The proper modeling of prestressing requires using specific software that can manage the cable layouts, the
calculation of tensions (calculation of losses), and consider the phasing and creep laws.

Example of complex cabling

However, it is always possible to perform a prestressing calculation or to verify a complex calculation to model the
prestressing in a simplified way.

Beams (and shells) must be described at their center of gravity to ensure the correct positioning of the ropes in the
cross-section.

The next two sub-sections present respectively the simplified modeling of a rope within a concrete section and a rope
outside of a concrete section. It assumes that the rope path is known and in constant tension (after instantaneous or
long-term losses, for example). It is useful to specify that the modeling of prestressing losses would follow the same
logic but with a sign opposite to the action of the initial prestressing.

C.14.1 Cable inside of a concrete section

The external forces method makes it possible to understand the effects of ropes by modeling them as concentrated
forces at the ends and as pressures (thrusts) along the rope.

End anchors or embedded anchors, we will have:

 a horizontal force HA = P.cos(α),
a vertical force VA = P.sin(α),
a bending moment MA = HA.e.

(with the sign convention adapted to the software)

Along the beam, a cable exerts radial thrusts of pi≈P/Ri, which in the general case can be assumed to be vertical. This
is the thrust exerted by the totality of the cables. They are applied as classical distributed loads.

Straight segments do not produce thrust (R=∞).

HA and P are frequently confused, as the cos(α) is often close to 1.00.

Example of a simplified manual definition of a rope.

We analyze in this example a two-span beam of 25m each. The section is symmetrical and 1.25m tall (to set the limits
of the cable).

Input data of the cable.

One must ensure that the cable remains inside the beam (in this case the limit is set as 10cm away from the upper
and lower faces) and that the connections at the inflection points have the same slope.
Converted to loading on the beam:

The resulting shear efforts, i.e. the shear efforts considering the hyperstaticity of the system:

The bending moment:

For control purposes, we can:

always return to an isostatic system (hereby removing the central support),

calculate the isostatic bending moment at the middle of the span (or in any section), by summing the forces
on the left or the right. The differences cannot be greater than the %,
in this case, verify that the results are symmetrical since the structure and the prestressing are symmetrical.

“Isostatic" shear effort diagram:

“Isostatic" bending moment diagram (the curve divided by HA is the rope path):

It should be ensured that the support reactions in the "isostatic" prestressed load case are equal to zero:

Calculation using specific software:

For this example, a software capable of modeling directly the prestressing ropes was used. The comparison of the
results is available in the document. Example of prestressing and eccentricity.

C.14.2 Cable outside the concrete section: forces at anchors and deflectors

Just as before, assuming a uniform tension for the entire cable, the external forces method allows apprehending the
effects of a prestressing rope by modeling it as a sequence of concentrated forces.

At the anchor A, the rope applies to node n1 of the model the loads (HA, VA, MA), MA being the moment produced by HA
at node 1. At each deflection point, the rope applies the force FS to the bar n1-n2. This is done for all the deflections
along the rope up to the last anchor.

C.14.3 Modelling prestressed slabs

The study of prestressed slabs is carried out according to the same principles used for beams but applied to shell
elements.

The use of specific software is desirable, maybe even essential. It will be necessary to make sure that the elements are
modeled at their center of gravity, and that the sum of the support reactions of the prestressed load case is zero.

C.14.4 Tensioning cables (side, order)

Be aware that the forces brought by the prestressing, after losses due to friction and anchor recoil, are strongly
dependent on the tensioning method (from one side only, from both sides). For exceedingly long ropes, the error can
be significant.

Also, for highly prestressed structures, the order of tensioning might have an impact and it is important to analyze the
structure at certain intermediate phases of tensioning.
$translationBooks

PART 2 – USEFUL ELEMENTS

C15. More about phased

calculations
C15. More about phased calculations
The reader may also refer to Part 1 - D.3 Construction Phases.

Structural phasing can lead to modification:

resistant sections,
support conditions,
internal continuity conditions of the structure.

It can concern both elevation and linear structures, longitudinally or transversely, and of course, a combination of all
these cases is possible.

C.15.1 How to make a phased calculation with non-phased software?

Case A - Modification of the net cross-section

This is the case for the implementation of self-supporting collaborative pre-slabs.

During the implementation phase, it is the pre-slab that must resist the weight of the structure (pre-slab weight +
compression slab). Then, it is the complex [pre-slab + compression slab] which will take up the loads later
implemented (superstructure, overloads, ...).

At the SLS (Service Limit State), there is an accumulation of stresses in the tensioned reinforcement, but there is no
direct accumulation of compressive stresses in the concrete.

At the ULS (Unlimited Limit State), the verification must be conducted for the concreting phase and the service phase,
but without considering the phasing.

Case B - Modification of support conditions

This is the case for a structure with temporary supports.

It can be associated with a modification of the resisting structure (example: supported collaborative pre-slabs).
Accumulation of the loadings

Phase 1: Loading the structure with a temporary support

Phase 2: Removal of the temporary support

Phase 1 + Phase 2: The final stress is identical to a non-phased structure.

This method enables dealing with the installation and removal of temporary supports.

It is always necessary to pay attention to the conditions of deformation of the structures during the installation of
temporary supports (placing the temporary support in contact with a deformed structure).

Case C - Modification of structural continuity

This is the case for a structure that is clamped during construction.

The stresses generated by the loads associated with a static diagram are calculated, then the stresses are summed up
(if there has been no evolution of the net cross-section) or the stresses are accumulated (if there has been an
evolution of the net cross-section).

Accumulation of the loadings

Example:

Phase 1: Self-weight sustained by isostatic spans

Phase 2: Accidental loads sustained by a continuous structure

Particular attention should be paid to the evolution of materials over time. In the case of reinforced or prestressed
concrete or mixed concrete-steel structures, creep (quantifiable by the difference between instantaneous and delayed
deformation) must be considered.

In the above example, before clamping, the deformation of the structure corresponds to a quasi-instantaneous
deformation. After clamping, the concrete creeps and tries to increase its deformation under long-term load, but the
structure is now continuous. The clamping of the creep deformations will here generate a moment of continuity on the
support that stretches the upper fiber.
Creep behavior can be considered approximately (see the CEREMA documents on the subject) or with the help of an FE
calculation with so-called scientific creep.

C.15.2 Pushing of a concrete bridge and launching of a steel bridge

The one thing these two models have in common is that during their installation phases, they will witness a shift in the
position of the support nodes according to the advancement of the pushing or launching phenomena. Potentially, any
node of the structure can be, at a given moment, a support node. Software that accept a pseudo-language of
programming may, in this case, have an advantage in creating incremental loops to simulate the advancement (by
incrementing the numbers of support nodes). Whenever possible, having bars of the same length facilitates the regular
motion of the supports.

Modeling the cutwaters, in both cases, does not pose any problem: they are steel bars, usually I-beams, embedded at
the extremity of the final structure.

Pushed concrete bridges: The calculation is almost a classical phased calculation. The sections casted over
concrete beams behind the bridge are modeled using bars resting on non-linear Z-shaped narrowed supports (possible
lifting). The bars with their casting dates and the prestressing, temporary or definitive are activated throughout time.
Finally, the cutwater and part of the prestressing are deactivated.

Launched steel frame: models can represent classical bi or multi-beams, but also boxes. The main differences with
the pushed concrete bridge model lie in the facts:

that the structure is very flexible,

that in the design phase, the added sections (of the order of 30m in length) rest on punctual supports,
generally two supports per section, instead of a continuous ground beam,
that the structure has a camber, determined beforehand by the calculation of the framework on its final
supports. The junction of the steel sections must be made by taking the necessary measures to guarantee the
continuity of the rotations of the edges of the sections. The two possible types of modeling to describe the
phenomenon are detailed below.

During modeling, when a section is added to the rear of the already reassembled construction frame, the set of bars
must be deactivated and reactivated after the new section is added. Otherwise, there is no continuity of rotation at the
connection (figure below) and the structure would not be compatible with the 3rd bullet point above.

Modeling can also be done by classical phasing, provided that a prior "presentation" of the joints is carried out, which
consists of finding the altimetric offset of the two supports 1 and 2 that enables having the same rotation and the
same altimetry at each end E1 and E2, schematically (following figures) :

Vertical translation for the Z-correspondence of the lips

Displacement of supports 1 and 2 to generate a rotation of the section

Once these operations are performed in the model, continuity is ensured.

For the launching, practice consists of modeling the neutral axes of the framework and the cutwater according to a
geometry algebraically accumulating the shape of the intrados (rectilinear or parabolic, for example), the longitudinal
profile, and the counter-axis, at an arbitrarily chosen altitude. During the assembling of the structure, especially during
the launching process, the nodes located in front of the provisional supports are given a difference in elevation
corresponding to the altimetric offset between the geometry described above and the altitude of the provisional
supports. It will be verified that the structure is in contact with the launching supports thanks to the sign of the support
reaction. A support in tension means that the structure is no longer in contact and that the support must be released.
Finally, for landings, there are always two cases to be studied, one right before and the other just after.

C.15.3 Phasing Affecting the Straight Sections

Since the construction phasing of a structure has an impact on the stress distribution on the straight sections of the
structure, it must be considered.

This is the case for structures built with transverse phasing, where only certain parts of the structure see the first
loads: such as for composite bridges with coated girders, or ribbed girders, or cast slabs in a second phase, and for
compound slabs ...

C.15.4 Expanding a Structure - Delayed Connections

To expand a structure, when a new structure (metallic or concrete) is connected to an older structure, the modeling of
the transverse phasing and the apprehension of the relative stiffness of the different elements is essential to correctly
determine the deformations of the structure and especially the connecting forces between the structures.

The case of delayed connections between several new structures is similar: the consideration of creep and shrinkage is
essential for a good dimensioning of the forces developing in the elements.

C.15.5 Cast-in-place or prefabricated structure - Deflection - Effect on the calculation

Please refer to § 2.1 and 2.2 of the Cerema Guide "Conception des ponts à haubans".
$translationBooks

PART 2 – USEFUL ELEMENTS

C16. More about dynamic and

seismic calculations
C16. More about dynamic and seismic calculations
The dynamic solicitations generate inertial and kinematic forces in the structure.

If Eurocode 8 is used for the design, a detailed explanation of the permissible simplifications is provided for the
modeling and calculations of the structure, according to its uniformity. This notion of uniformity is explained in §4.2.3
of EN1998-1. For bridges, the guide "Ponts courants en zone sismique" also provides uniformity criteria and admissible
simplifications for the calculations.

Depending on the refinement of the model and the objectives sought, several points should be considered.

C.16.1 Defining the general X and Y axes

Defining the general X and Y axes warrants particular attention. Indeed, seismic results can be erroneous if these axes
are not close to the main axes of inertia of the structure.

2 examples illustrating the subject can be visualized below.

Example: Corner block of a stadium

Overview of the corner block model

It is necessary to adopt for the corner blocks different reference axes (X, Y) from those of the general project (XG, YG).
The X-axis must be radial in the direction of the 1st vibration mode.

The study of the eigenmodes then shows that the fundamental modes of a classical building with well-defined modes
are retrieved according to the X and Y directions as well as a torsion mode. This would not be the case if the general
axes XG and YG were adopted because each mode would activate masses in both directions, which would disturb the
Complete quadratic combinations (CQC) and Newmark combinations.
Example: Earthquake on a curved bridge

In the example below, two seismic calculations were performed for this TGV viaduct with an in-plan curvature.

The first computation depends on the general X and Y axes oriented according to the left abutment and the second
with the axes oriented according to the right abutment.

It is possible to switch from one model to another by changing the coordinates of the model nodes. The comparison of
the efforts at the base of the supports highlights very different results between the 2 models.

Graph n° foot support/effort (kN)

Therefore, it is advised to carry out seismic calculations for bridges on straight planar alignments if the curvature
allows it (refer to the CEREMA guides). Otherwise, several calculations must be conducted by varying the axes for each
support studied, it is a complex solution that should be avoided when possible.
C.16.2 Modeling of non-structural or secondary elements

See C.4 Modeling of non-structural elements or equipment.

Non-structural elements and equipment do not need to be modeled if their mass is small enough relative to the mass
of the building. However, one must ensure that their stiffness do not influence the behavior of the structure.
Otherwise, they must be considered. For more information, please refer to §2.4.5.2 of the ASN guide and the AFPS
technical book 36 of the AFPS concerning the dimensioning of non-structural elements under seismic conditions.

When the mass of these non-structural elements is not negligible, it must be correctly defined in space. For example,
the offset of the bridges’ lateral elements’ masses with respect to the center of the sections should be well defined so
that the torsion modes are considered.

The distribution of the masses on the floors can generate many local modes during a modal analysis and make the
model difficult to operate. In this case, it is recommended to apply punctual masses and to redo a local study if
necessary.

C.16.3 Transformation of loads into masses

Most software calculate the weight of the structure directly or have an option to do so based on the density or
volumetric weight of the materials.

For seismic and dynamic calculations, the weights, forces, and pressures must be converted into masses following the
normative ponderation factors (it is the case for equipment and superstructures or part of the operational loads).

To reduce the number of eigenmodes that are not useful for a global calculation, one can focus on options that
transform the distributed masses into masses at the nodes or introduce manually the masses at the nodes.

C.16.4 Pay attention to the units

The accuracy of seismic or dynamic calculations is particularly sensitive to the coherence of the units. Inertial forces
involve the acceleration of gravity g, whose unit (generally defined by default) must be consistent with all others.

Trivial, but it is always useful to remember that the unit of mass is ... the kilogram.

Be careful when using old standards or regulations because they may use units such as kgf (kilogram-force). In
general, it is advised to use exclusively the international system units, at the very least to control the
results.

Taking the simple example of the self-weight calculation. For most software, the action of gravity is defined by the
application of vertical acceleration applied to the whole structure. Internally, the software will calculate the mass of the
structure by first calculating for each element its volume multiplied by the density of its material. If one wishes to
obtain the self-weight in N and the geometrical dimensions and density have been defined respectively in mm and
kg/mm3, the acceleration will have to be defined in ... m/s²:

For example, the mass equivalent to an operational accidental load of 20KN is equal to 20,000(N)/ 9.81(m/s2) = 2038
Kg or 2,038 tons.

C.16.5 Materials

The material laws as well as the partial coefficients depend on the type of analysis performed.

Taking concrete as an example, the instantaneous Young's modulus will be preferred.

To consider the cracking state of the elements, the EI module can be modified:

either by a minor coefficient applied to Young's modulus E,

or by modifying the net cross-section or the inertia of the section directly.

The applied reference frame can specify the Poisson's ratio to apply according to the type of calculation. it can be
modified to consider the state of cracking, for exemple, equal to zero in the case of a cracked element or under ALS
(Accidental Limit State) earthquake.

C.16.6 Modeling of bracing elements of steel structures

Bracing elements ensure the lateral stability of the structure. It is important to transcribe their actual behavior. For St.
Andrew's Crosses, for instance, the bars only work in tension because they buckle instantaneously in compression.
Therefore, they should not be modeled in their entirety if a linear calculation is planned. Otherwise, the capacity of the
bracing would be overestimated by a factor of about 2. See C.2 Modelling of the main elements

C.16.7 Boundary Conditions

Depending on the models, the dynamic soil-structure interaction may need to be considered. If the springs are
modeled this way, it is necessary to ensure that the structure do not uplift excessively to remain within the validity
range of a linear study.

For earthquake studies, the engineer can calculate stiffnesses by referring to the following documents:

"Ponts en zone sismique" published by CEREMA, which proposes in chapter 4.3.3.2 fairly simple calculation
formulas,
Seismic Design-Building - V. Davidovici §4.3.4.4 Modelling of the ground by a system of damped springs -
Eurocode Collection - Afnor Editions,
Gazetas formulas: which can be consulted in Appendix D of "Fondations et procédés d’amélioration du sol de
Davidovici" (or other references).

All these documents determine the stiffnesses from the shear moduli and Poisson's coefficients of the soils, but also
from the characteristic dimensions of the foundation. These stiffnesses depend on the vibration frequencies of the
studied structure.

The case of foundation slabs:

Modeling a foundation slab under dynamic loading is more complex because the springs will have to represent at the
same time the vertical, horizontal, and rotational stiffnesses, as determined by the soil-structure interaction study.

One can refer to more specific documents for this type of study.

Several forms of modeling are possible:

using a punctual spring in the center of the invert affected by the 6 stiffnesses (and 6 dampings), with rigid
connections on all the nodes of the foundation slab,

Advantage: an accurate representation of the SSI in dynamic calculations.

Disadvantage: it is not possible to determine the stresses in the foundation slab because of the presence of the rigid
connections that artificially stiffen it. It is then necessary to carry out a local calculation of the foundation slab
subjected to the soil pressures deduced from the forces in the central spring. More specifically, in the case of a
foundation slab with large dimensions compared to its thickness, this method is not adapted.

using springs placed uniformly under the foundation slab (as for the static study);

In this case, each node of the invert is connected to 3 springs, one in each direction X, Y, Z.

The horizontal springs according to X or Y will be deducted directly from the global translational stiffnesses, whereas
the stiffnesses of the vertical springs will represent either the global vertical stiffness or the global rotational stiffness
in a given direction. This approach implies 3 computational models to analyze the 3 directions of the earthquake.

Advantage: simpler model, allowing to calculate the forces in the foundation slab.

Disadvantage: one of the 2 vertical or rotational stiffnesses is not represented in each of the calculation models. The
torsional stiffness is not incorporated.

putting in place a spring mattress,

This type of modeling is mainly used in complex structures, a mattress of springs assigned to each node of the
foundation slab allows to represent all the global stiffnesses.

Advantage: the SSI is modeled in detail.

Disadvantage: the modeling is complex and can only be applied using specific software with a complete understanding
of the causes.
 Figure: Diagram of the spring mattress, Tractebel image

C.16.8 Spectral modal analysis

Truncation - number of modes

The theoretical concept of truncation is defined in the 1st part of this guide. In practice, concerning the number of
modes to be used for the calculation, we advise:

not to exceed 100 modes for classical works,

to go up to the cut-off frequency (generally 33 Hz),
to use a pseudo mode for the participating mass that is not being considered (EN 1998-2/§4.2.1.2),
not to be limited to the modes with the most participating masses, because the antisymmetric modes have
classically a very low number of participating mass but induce non-zero forces,
to reflect on the relevance of retaining or not local modes in the analysis.

Behavioral coefficients

Since the coefficient, or rather the behavioral coefficients, can be different in each direction, they are incorporated in
the definition of the calculation spectra. It is important to check that the calculated displacements are indeed re-
multiplied by this same coefficient.

Modes signature

At the end of the combination of spectral responses, the sign of the efforts is lost (all values are positive). This can
generate operating difficulties when one wishes to calculate a torsor or when one wishes to study concomitant forces
(see D 7.4.5).

To reallocate a sign to the different calculated quantities, there are several possible approaches, including those
described below:

1. Attribution of the sign of one of the modes. For structures having a predominant mode in each direction, it is
possible to allocate the sign of the predominant mode to the calculated quantities. This is interesting for the
overall behavior of the structure and is very efficient as long as the participation of this mode is greater than
60% of the modal mass of the structure. On the other hand, for elements responding on higher local modes,
this may not be appropriate (see the example of thick floors in industrial sites),
2. Attribution of the sign following a uniform acceleration analysis. For each direction, a unit acceleration is
applied, and the sign obtained is kept,
3. Ellipsis method analysis. When the justification of a structural element must consider several stress
components, it is possible to establish the range of concomitance of these quantities, so as not to introduce
conservatism in the calculation.

C.16.9 Damping

Within the framework of a structural study with a calculation spectrum including a behavioral coefficient, the latter
already considers damping. Thus, it is not useful to introduce another one.

For dimensioning with an elastic response spectrum, the damping of the materials must be taken into account.

When setting up the data, it is important to ensure that the material damping included or taken by default by the
software is consistent with the codes and the analyses carried out. For example, it is necessary to distinguish
reinforced concrete from prestressed concrete, or welded structures from bolted structures in the definition of material
damping.

C.16.10 Time discretization and integration scheme

The resolution of a dynamic solicitation requires the implementation of a specific integration method. Part 1 Chapter 2
provides details on these methods and guidance for choosing the time step and mesh size according to the problem. It
is recommended to consult engineers specialized in this type of study.
$translationBooks

PART 2 – USEFUL ELEMENTS

Chapter D. Analysis and

processing of the results
Chapter D. Analysis and processing of the results
D.1 General information about numerical calculations

D.2 Load combinations

D.3 Results processing

D.4 Validation rules: the behavior of concrete elements

D.5 Understanding and analyzing the peaks (case study about concrete)

D.6 Understanding and analyzing the peaks (case study about steel assembly)

D.7 Further information specific to dynamic calculations

$translationBooks

PART 2 – USEFUL ELEMENTS

D1. General information on

numerical calculations
D1. General information on numerical calculations
D.1.1 Calculation time

Computation time (in the broadest sense: solving the system of equations and storing the results) is an essential point
of reflection to be associated with Finite Element modeling.

It must remain compatible with the budget and timeframe associated with the project. The search for an optimized
computation time is even one of the key elements when considering the economy of a project.

The calculation time is influenced by different factors.

The finite element model:

Number of degrees of freedom,
Formulation of the elements, number of integration points.

One can optimize the computation time by using simplified models, sub-models, sub-structures, symmetries, or locally
adapted mesh sizes. However, be careful with explicit calculations because the time step is set based on the
dimension of the smallest element.

The computing capacity:

RAM: random access memory (temporary storage),
the operating system (32 or 64 bits),
is the calculation done locally (on the user's workstation)? on a local server? on an external server?
Model and results copy times, or the time it takes to launch other calculations on the same server
can be long,
the capacity and number of processors: the "CPU time" (Central Processing Unit) usually provided by
the software should be checked.

Optimizing a given computing system can consist of performing the calculation in batch (launching the calculation in
command lines without the interface), or parallelizing the calculations (on several processor cores), and/or performing
further calculations if the software allows it.

The type of analysis:

Linear/non-linear and the associated algorithms,
Numerical integration scheme (direct integration/modal-based projection, implicit/explicit scheme)
and the choice of the solver for dynamic calculations.

The type of analysis is specific to the problem and the desired accuracy of the results. Therefore, the modeler oversees
the optimization. However, it is emphasized that in the case of a transient calculation, the duration of the calculation
(in terms of loading and response of the structure) should not be overestimated so as not to add unnecessary
calculation time (machine time).

The anticipation of "post-processing":

for some software, it is possible to select (and keep) only the quantities of interest chosen by the
user, as well as the instants (in the case of transient or phase calculations): the machine time to
write the results is then reduced,
the number of saved/reused modes in the case of a modal analysis can generally be modified
(definition of variable filters depending on the software),
the real-time visualization of results can be used to stop a calculation in progress (but it generally
leads to an increase in calculation times),
the analysis of the results can be performed using post-processors independent of the Finite Element
Calculation software to reduce the operating time.

The automation of the calculations and post-processing (after validation of the first calculation round and using
appropriate verifications ) is interesting when the user has to perform the same calculation several times on different
models or similar calculations on the same model.

The acceptability of the calculation time must be assessed because the model will run many times and
will become more complex as the study phases progress.

D.1.2 Convergence of the software – The case of direct elastic calculations

In direct elastic calculations (linear static calculation and spectral modal calculation), there are very few reasons for
the software not to converge:

either the structure is unstable,

or the inversion of the stiffness matrix is impossible because there are stiffness differences between the
elements that are too large.

All error messages, at this stage, are related to these two cases.
Due to the boundary conditions in place, the displacements, translation(s), and/or rotation(s) of certain nodes of the
structure are imposed (a blockage is a zero imposed displacement). However, the set of these blocked displacements
may appear insufficient to prevent an overall movement of the structure.

It is advised, by choosing an appropriate reference frame, to analyze the effects of all the blockages on the overall
movements of the structure and to add one or more restraints to ensure the stability of the structure. At the end of the
calculations, it must be verified that the reactions due to these additional blockages are null or negligible: effort for a
blockage in displacement, moment for a blockage in rotation.

Some software, for specific geometries of structures, ignore the problem of instability. In this case, the stresses and
deformations are correct, but some displacements are somewhat unreasonable. Another instability issue often
encountered is related to the connection between different types of structural elements. This is specified in paragraph
C.6.3.

It should be noted that software error messages are sometimes not very explicit about the global, local, and/or
numerical nature of the instabilities.

Furthermore, some software programs allow the calculation to be completed despite a warning or error
message. This can help understanding or visualizing where the problem comes from, but the results are not
satisfactory in this context.

In the end, it is necessary to have a model that runs without errors.

These problems of global and local instabilities can be detected by performing simple calculations:

either linear static calculations by applying a global acceleration (10 m/s² for example) to the structure. Three
load cases can be created according to the 3 main directions,
or a modal calculation (determination of the first eigenfrequencies).

If the structure presents global instabilities, the static calculation is unlikely to succeed whereas the modal calculation
will present rigid modes.

If the structure presents discontinuities, they will be highlighted by the shape of the deformation or the eigenmodes.
Be careful, taking into account the shear stress deformations leads to failures in the deformed shape.

D.1.3 Convergence of the software - Case of iterative calculations

For non-linear calculations, the convergence of a Finite Elements calculation is obtained with a function interpolating
the solution for each of the finite elements (principle of discretization).

The convergence criteria are either given by the user or taken as default values by the software. The finite element
calculation consists of a series of iterations and stops either when the maximum number of iterations has been
reached or when certain deviation measures between two successive iterations are below the predefined thresholds
(this is called the convergence criteria).

On the other hand, a calculation that has stopped does not mean that it did not converge towards a valid solution. For
example, deformation criteria that are too loose compared to reality, or a maximum number of iterations that are too
low, can output a result that is not realistic.

D.1.4 Model Convergence

A model is considered to have converged when a small change in the size and geometry of the meshes does not
significantly disturb the results.

To be rigorous, it is advised to carry out a sensitivity study of the mesh to the post-processed quantities. Thus,
carrying out the same analysis on different meshes (reduction of the mesh size, for example), should - if not provide
similar results (acceptable tolerance) - converge towards an acceptable solution to the user. It is important not to focus
on peak values and to bear in mind that several successive refinements of meshes can lead to divergent results.

It is also advised to analyze the software's warning messages ("Warning" on mesh size or shape) to assess whether
they are likely to alter the results.
$translationBooks

PART 2 – USEFUL ELEMENTS

D2. Load combinations

D2. Load combinations
D.2.1 Combinations and envelopes

Recall that using a load combination consists of accumulating the structural effects of different loadings by assigning
weighting coefficients to the different loadings as defined in the standards.

Strictly speaking, the codes require all combinations to be verified. In the case of a building, the number of these
combinations remains small, so the software can calculate all of them. However, for more complex cases, and
particularly in the case of rolling loads, the theoretical number of combinations becomes unmanageable. In such cases,
force envelopes are used.

An envelope contains several load cases and only records the extreme values of the individual components (with the
concomitant components).

According to the codes, some loadings applied to the structure should not be accumulated because both occurring at
once is not a reasonable assumption. It is then useful to incorporate these non-cumulative loadings into an envelope
that will highlight, for each effect studied, the most unfavorable loading among a group of non-cumulative loadings.
Thus, for bridges, the envelopes of thermal, wind, and road loadings are generally considered.

Remember that accumulating unit cases calculated by a non-linear approach (NL) has no physical meaning. However,
some software allow non-linear calculations on "combinations". In this case, the software will create a new case (the
"combination") from the unit load cases and perform the NL calculation on this sum of loads. If the software does not
allow it, it will be necessary to create combinations by manually grouping the unit loads. In this case, it is, once again,
fundamental to understand what the software does.

Illustration of the above text using an example

A secondary gallery of a tunnel is considered. The structure is fully supported by non-linear springs – since the ground
does not take up traction, the springs are blocked by the software if there is a soil-structure separation.

Illustration of the structure

The following two unit cases are defined:

self-weight + earth weight and thrust,

hydrostatic pressure.
 Gravity loads (left) and water pressure loads (right)

The following results show that although the software can conduct non-linear calculations using all the unit cases, the
combination of the two cases is recalculated integrally and independently...

Results of the unit cases: bending moments - gravity loads (left) and water pressure loads (right)

Results of the combination of the two cases: bending moments - software combination (left) and cumulative loads in a
new case created manually from the unit cases (right) - identical results

... since the accumulation of the unit cases does not lead to the results of the combination.

The diagram of the pressures on the soil speaks for itself: the water pressure forces the vault to rest on the ground
"upwards" at the top (right figure) ...
 Soil pressures under gravity loads (left) and water pressure loads (right)

... but once accumulated in the case of gravity loads, the top of the vault no longer pushes upwards, which can be
seen on the results of the combination made by the software:

Soil pressures from the combination of gravity loads and pressures

Note: without springs on the arch, the second case would not converge.

D.2.2 Be careful when using the envelope results

When using envelopes, one must record in a database the displacements, extreme stress values, strains, or reaction
forces at the supports.

Most calculation software offer the possibility of storing the extreme values of stresses and strains, either alone or
together with the values of concomitant stresses and strains.

Before using the results of the envelopes for further post-processing, one should clearly understand whether the
stresses and strains are concomitant.

For example, if one wants to reconstruct the most unfavorable stress state of a section, it should be verified that the
extreme stresses selected for the upper and lower fibers are indeed concomitant.

Also, it is important to base the analysis of the results not only on the most unfavorable stresses and their
concomitances but also on the concomitant stresses that generate the most unfavorable stress states. Thus, a
maximum normal stress associated with a small concomitant moment can generate less unfavorable effects than a
slightly smaller normal stress associated with a larger moment.

To verify a section, it is acceptable in upstream phases to perform the verification with all the extreme stresses in the
same torsor. However, in the execution phase, for optimization reasons, it is advised to recover the torsors of
concomitant stresses.

D.2.3 Beware of automatic combinations!

The combinations of loadings are used differently for building or other civil engineering structures.

For buildings, elementary loads induce a very large number of possible positions, all of which must be explored to
determine the maximum effects on each structural element. This multiplicity of loads and configurations leads very
naturally to the use of automatic combination modules.

In general, the automatic combination modules proposed by most software should be used with great precautions
because it is a frequent source of errors. Some modules are black-box modules and not all software are allowed to
know what are the elementary load cases used for the envelope combinations.

Moreover, the verification and coding of the combinations is a tedious exercise for which it is difficult to detect the
error.

One of the most effective methods to prevent errors related to combinations and envelopes is to dissect the design
stresses and strains. For some key stresses or strains (maximum bending moment, extreme stresses), it is a matter of
finding the participation of each design elementary load in the overall stress or strain. Thus, one can verify that there
are no errors in the accumulated values and coefficients and that the "logical" load cases are design loads.

For buildings, the same exercise can be done with the reaction forces at the supports.
$translationBooks

PART 2 – USEFUL ELEMENTS

D3. Data processing

D3. Data processing
D.3.1 Stresses/Deformations or Internal forces

It is important to define what you are looking for before starting the calculations, and this depends on the type of
study:

for structures modeled as beam elements, we will prefer processing internal forces,
for civil engineering buildings, the stress approach allows us to understand the overall behavior of the
structure by identifying different zones. On the other hand, when the objective is to obtain reinforcement
results (quantity or internal forces), it is necessary to process the internal forces and to carry out the
corresponding calculations in a second step.

D.3.2 Values at nodes, Gauss point values or values at the center

See also A.2 What is a finite element?

Generally, the software calculates the values of stresses and strains at the Gauss points, whose position is defined in
the software's finite element manual. It can then infer the value at the center of the element. It extrapolates to each
node and, since a node is usually linked to several elements, it computes the average of the values obtained for each
element.

In regular zones, and with a proper mesh such as shown in paragraph C.3, the values at the center, at the nodes, or
the Gauss points are very close.

The issue arises close to the peaks, where it is necessary to reflect on the most representative values.

Types of values displayed

Most software calculates the stresses at the integration points (or Gauss points). But there are two ways to visualize
them:

at the elements – This method consists of directly plotting the average value on each element; an average of
the values calculated at the integration points of the element. The plotted stresses then present
discontinuities, which are accentuated when the discretization is not well adapted to the zone of interest. The
results displayed in the center of the elements are reliable.
or at the nodes - This second method consists of displaying the mean values at the nodes. For each node, the
calculated value is the weighted average of the stresses from the selected elements adjacent to the node.
This treatment, called smoothing (*), is intended to display a continuous field, which may seem more
relevant.

Stress field - Principle of calculation of the displayed values

(*) Do not confuse this smoothing action with the smoothing described in paragraphs D.5 and D.6, which consists of
computing average forces over a given length.

In the figure below, the stresses calculated at the integration points have been extrapolated to the nodes before
averaging. This is what some software packages do by default.

We can illustrate it using the example of the Br wheel (Example D - Simple case: modeling of a Br wheel). The
software used does not give access to the values at the Gauss points.
Mapping of transverse bending moments - values at the center of the slabs - overview (28.52kN.m/m is an extreme
calculation value)

Values at node 3 - they are different depending on the finite element

 Values at the center of the elements - Zoom

Each software documentation should provide information on how it displays the solicitations and the options available.
By default, software can propose that the linear part of the result of the integration point is extrapolated to the node
while the non-linear part (plastic deformation for example) is copied. And an option in the same software allows the
linear and non-linear parts to be copied to the nodes.

We will see later in paragraph D.5 the important variations that can be linked to smoothing the results at the nodes
or elements.

D.3.3 Stress Analysis - Identification of Sensitive Zones

To follow up on the above details, displaying the stresses at the elements allows better visualization of areas of high
discontinuity. For steel structures, software usually display the equivalent Von Mises stress, which gives an idea about
the zones of high stress and/or with a high-stress gradient. One can also display the internal energy of deformation of
each element. In reinforced concrete structures, either the stresses are displayed, or the reinforcement mapping,
which allows visualizing the highly stressed areas.
$translationBooks

PART 2 – USEFUL ELEMENTS

D4. Normative verifications: the

behavior of reinforced concrete
elements
D4. Normative verifications: the behavior of reinforced concrete
elements
D.4.1 Reinforcement mapping – reinforced concrete normative
verification / connecting struts
Calculations of reinforcement mapping are carried out by the current software generally by using the method of Capra
Maury (Annals of ITCBTP - Institut des Techniques de la Construction du bâtiment et des Travaux Publics - of
December 1978) or the method of Wood and Armer (“The reinforcement of slabs according to a predetermined field of
moments" Concrete February 1968, August 1968).

These methods render it possible to determine the necessity of reinforcement in the 4 directions Axi, Ayi, Axs, Ays on
the lower and upper faces of an element according to the 2 directions of reinforcement considered orthogonal, X and
Y.

A good approach to the calculation of these mappings can be made from Wood's method by simplifying:

Either an element subjected to the following stress components:

Nxx, Nyy, Nxy membrane stresses (positive if traction)

Mxx, Myy, Mxy bending stresses

The first step is to calculate the following intermediate stresses:

Nwx = Nxx + |Nxy|

Nwy = Nyy + |Nxy|
Mwx = Mxx + |Mxy| if Mxx is positive, Mwx = Mxx - |Mxy| otherwise
Mwy = Myy+ |Mxy| if Myy is positive, Mwy = Myy- |Mxy| otherwise

D.4.2 Bending elements: slabs

The slabs work mainly subjected to the bending moments Mxx, Myy, Mxy.

The membrane efforts Nxx, Nyy, Nxy are often negligible.

The Mxy bending moments can be important, they are null in the zones where the principal bending moments are
aligned with the principal axes (often confused with the directions of the reinforcement) and therefore in the middle of
the span and on continuous supports.

They should not be neglected especially in the corners of the slabs and in the case of concentrated loads.

The calculations of passive reinforcement carried out in the determination of mappings are according to the rules
brought by the Eurocode, because they are adapted to the study of a rectangular section subjected to combined
loading (N, M), they can thus be carried out in SLS and the ULS.

The calculation of the reinforcement in SLS considering the crack openings is more delicate and requires the use of
well-tested software. Indeed, the presence of MXY bending moments requires steel calculations on several facets
because cracking does not necessarily occur according to the direction of the reinforcement.

As the behavior of slabs in bending is similar to that of beams, it is necessary to redistribute the bending moment
diagrams to have a truss-like behavior.

The reinforcement areas Ax are calculated directly from the efforts (Nwx, Mwx) and the reinforcements Ay from (Nwy,
Mwy).

Example: 5m square slab

The slab is hinged on 3 sides and clamped on the 4th, it is subjected to a uniform permanent load of 50kN/m², a
thickness of 20cm, and a concrete cover of 3cm.
 Horizontal bending moments MYY | Bending moments MXY

Ay reinforcement is calculated manually with fyd=200MPa (SLS).

Ay manual = 10 x (MYY + MXY)/(0.9 x 0.17 x 200) cm²

Values that can be compared with the reinforcement areas calculated by the software:
Reinforcement Ays software / Reinforcement Ayi software

There is a good agreement between the reinforcement calculated manually and the one determined by the software
(CAPRA MAURY method).

It can be deduced that the Mxy bending moments are cumulated with the Mxx and Myy bending moments.

D.4.3 Elements submitted to shear forces in their plane: concrete walls

The bracing walls of a building are subjected to normal stresses and shear forces in their plane.

We have the examples of "large dimension walls of lightly reinforced concrete" studied in EC8-1 chapter 5.4.3.5. or
transfer slabs of buildings subjected to horizontal forces (inclined columns).

For these elements, the components Mxx, Myy, Mxy, Vxz, and Vyz are very low or even null.

They are only subjected to membrane efforts Nxx, Nyy, Nxy:

Nxx and Nyy being the tensile/compression stresses along x and y axes
Nxy the shear in the plane of the wall.

Taking into consideration the previous calculations of reinforcement distribution, this gives:

Nwx = Nxx + |Nxy|

Nwy= Nyy + | Nxy|

Hence the reinforcement:

Ax = (Nxx + |Nxy|)/fyd
Ay = (Nyy + |Nxy|)/fyd
Ax and Ay being the sum of the reinforcement in X and Y (2 faces included)
and fyd is the design stress of the rebars

These formulas lead to the observation that if the normal stresses are zero (Nxx=Nyy=0), then the reinforcements Ax
and Ay are shear reinforcements and their values are equal to Ax= |Nxy|/fyd and Ay= |Nxy|/fyd.

Hence, shear efforts require reinforcement in both directions, as opposed to the classical shear verifications performed
according to EC2. This is because the strength of the concrete is not accounted for.

Therefore, the shear reinforcements calculated by the mappings are "higher" than those usually calculated using EC2.

Article 5.4.3.5.2 of EC8-1 concerning the study of "Poorly reinforced large dimension concrete walls" was consulted so
that no shear reinforcement is required Ved is lower than Vrdc.

Thus, it is recommended in these cases to use reinforcement maps only to consider local effects, and to perform
calculations of the main reinforcement by taking cross-sections at the base of the walls and to determine the
reinforcement from the resulting torsors at the level of these cross-sections.

Example of a sail:

Let us study an isolated concrete wall, 5m high, 4m wide, and 20cm thick.
It is clamped at the bottom and subjected to a seismic horizontal load of 2000kN at the top.

To avoid peak efforts, the horizontal load is linearized over the width of the concrete wall (500kN/m).

The resulting membrane forces Nxx, Nyy, and Nxy are illustrated below.

Stresses Nxx (horizontal) (kN/m) – Stresses Nyy (vertical) (kN/m)

Stresses Nxy (shear) - Reinforcement mapping (cm²/m²) for one layer

The steel sections are calculated manually and compared with the map values.

This table shows that, on one hand, the manual calculation provides a good approximation of the required
reinforcement section and, on the other hand, that the shear efforts Nxy are added to the two membrane efforts Nxx
and Nyy, which does not reflect the real behavior of reinforced concrete walls.

Normative verifications of a reinforced concrete wall:

The reinforced concrete wall is recalculated according to EC2.

At a cross-section at the base of the wall, the resulting efforts are equal to:

M-flection = 2000 x 5 = 10 000kN.m

Vu = 2000kN (ALS)

Hence the bending reinforcement: A-tension = 10 000 /(0.9 x 3.9 x 50) = 57 cm2 or 29cm2 per layer.
Vrdc = 997kN is lower than Vu = 2000kN so, 11cm²/m of shear reinforcement must be distributed (6cm²/m per layer)
considering cot(θ)=1 or 5 cm²/m with cot(θ)=2.5.

Comparing the 2 methods:

The bending reinforcement is more important using reinforcement mappings, (36+23+15) cm²/m x 0.5 m= 37cm²
compared to 29cm² using the normative verifications (30% greater).

For the shear efforts, 7cm²/m must be distributed using the reinforcement mappings whereas the normative
verifications allow distributing only 2.5cm²/m considering cot(θ) = 2.5.

Summary of the comparison between the calculation of a reinforced concrete wall either by the reinforcement
mappings or by the normative verifications for reinforced concrete (EC2).

D.4.4 Cross-section method

Most software allow to “cut” the elements to calculate the resulting torsors in their centers.

Let us do a study case on the base of a reinforced concrete wall:

The horizontal cross-section at point 0 at the base of the wall allows obtaining the resulting torsor consisting of the
normal efforts, the shear efforts, and the bending moments (in-plane or out-of-plane), by the integration of the
stresses.

The reinforcement can then be determined by using a beam type calculation, for which one must be sure to stay within
the definition range of a beam element.

The use of cross-sections is above all very useful for the engineer to quantify the stress paths in a structure.

D.4.5 Scope of validity of steel mapping

D.4.5.1 Mapping and cuts

Reinforcement mapping is the result of a numerical calculation carried out for each element independently, therefore
without considering the overall reinforced concrete behavior of the structure.

Note: The cross-section method is the only one that respects the behavior of the reinforced concrete walls and is
considered normative. The usual additional verifications must be applied: redistribution of bending moment diagram,
verification of the connecting struts, lapping, and minimal reinforcement ... is still to be done.

Therefore, the engineer needs to validate the obtained results with other normative methods.

Example of a wall-beam calculated with the reinforcement mapping:

The example of a beam on two supports is studied here. The beam is modeled as a tall wall-beam, to show on a simple
case the inconsistencies of the reinforcement mappings.

Let the isostatic beam with a 10 m span, 3 m high, and a uniform load of 200 kN/m². This beam is modeled in plate
elements working in their plane with a mesh size of 0.5 x 0.5 m².

Visualization of horizontal efforts Fxx

Calculation of reinforcement with reinforcement mappings:

The bending capacity is equal to F0 x d0 = 2 layers x 0.5 m x (24 x 2.75 + 16 x 2.25 + 8 x 1.75) = 116cm².m.

The resulting reinforcement is equal to Ax = 116 cm².m / 2.9 m = 40 cm².

Calculation of the reinforcement with the cross-section method:

This second method consists of “cutting” the beam in the middle (A-A).

The software integrates the efforts Fxx along the height of the cross-section to deduce the resulting torsor in the
middle of the cross-section.

The bending moment at the center is equal to 3358kN.m, (we then find the classical bending moment using Strength
of Materials, Mu=1.35ql2/8 = 3375 kN.m), hence the reinforcement calculated according to the rules of reinforced
concrete: A = 3358/(0.9 x 2.9 x 43) = 30 cm².

Conclusion and comparison between the two methods:

Reinforcement resulting from the reinforcement mappings / Reinforcement resulting from the cross-section method
(Reinforced concrete calculation).

Conclusion: This example shows the limitations of steel mapping since it does not respect the deformation rule
for reinforced concrete. The cut-off method makes it possible to optimize the reinforcement.

D.4.5.2 Strut-and-tie method: Finite element contribution

Case of a wall-beam

The Eurocode 2 strongly encourages the use of the strut-and-tie method, in this case, finite elements can help the
engineer to define the operating scheme of the connecting struts as well as their inclinations. Let us consider the
example of the following wall-beam:
As the span of 3m50 is smaller than 3 times the height of the beam, the classical rules for beams do not apply (EC2-
5.3.1), this wall-beam is calculated using the strut-and-tie method.

As this example is quite simple, we can manually define the strut behavior:

The usual rules estimate the height of the connecting strut Z in 1.90m, we obtain tg(Θ) = 2.18, i.e. a tie with
H=515KN, or a theoretical reinforcement area A=11.8cm² (ULS calculation).

In more complex cases, the engineer will have to define a strut-and-tie behavior which may be difficult. The finite
elements then bring precious help for the engineer, we propose to follow the following method:

Representative isolated model:

It consists of creating an isolated representative model of the problem.

 Isolated representative model

The principal stresses:

We will also refer to part 1, chapter E.3.3.

What to remember: There are 2 principal stresses, the min S1, the max S2, they are represented as perpendicular
arrows, the length of each arrow depends on the intensity of the stress. The S1 stresses show the negative
compressions and S2 positive tensions.

Principal stresses S2 (compressions)

 Principal stresses S1 (tractions)

In our case, the long blue arrows show compressed zones, the red ones show tensioned zones.

Nota bene: when the 2 arrows S1 and S2 are almost equal, i.e. when the representation is a cross, then the
zone is in pure shear.

Definition of strut-and-tie behavior:

The visualization of the compressive stresses at the base allows to visualize the direction of the connecting struts, its
angle is of the order tg(Θ)= 2. Hence, a tie with 562kN from which a reinforcement area A = 13cm² with some
uncertainty due to the graphical method.

Reinforcement mapping:

The reinforcement mapping directly reinforces the lower tie:

The reinforcement area along 1m summed results in 5.07cm² per layer, i.e. 10cm² in total. This more precise value can
be retained.

Balance sheet

This approach makes it possible to identify the stress path and to set up a model of a suitable connecting strut
(compatible with the stress path).

Load path in a wall-beam with multiple openings, S1 stresses

Load path in a wall-beam with multiple openings, S2 stresses

The reader might refer to §8 "Strut-and-tie modeling" of the FIB Bulletin nº45 for details on modeling this type of
approach and to SETRA Bulletin Ouvrage d'Art nº14, pages 23 to 32.
$translationBooks

PART 2 – USEFUL ELEMENTS

D5. Understanding and analyzing

the peaks (case study about
concrete)
D5. Understanding and analyzing the peaks (case study about
concrete)
The smoothing referred to in this chapter concerns the clipping of stresses or strains or the distribution of
reinforcement, by averaging it, over a certain length (or width) of the calculated element.

D.5.1 Stress concentration and peak stress (forces) - Different types of

stress visualization
D.5.1.1 Stress and strain peaks

Contrary to common ideas, stress and strain peaks did not appear with finite element software, but have always been
part of the stress of structural engineers, they are inherent to the inclusion of point forces in the calculation of slabs.

Let us consider, for example, the study of an infinite isostatic slab subjected to a load concentrated in its center.

A possible approach to calculate the forces in the center of a slab is the use of Pücher’s charts.

The chart above (surfaces of influence) shows a peak of moments in the center: the line of influence increases rapidly
as the load approaches the middle of the span: the values 3 / 4 / 5 then 6 and 7 narrow to an value which is infinite,
but corresponds to a surface tending toward 0. Pücher arbitrarily truncated the representation to the value of 8.

Let us consider a slab with a span of 3 m articulated on its sides, subjected to a concentrated load of 10 KN in its
center.

According to the chart, the theoretical moment at the load is infinite, which is not satisfactory for the engineer who has
to dimension the reinforcement. In reality, point loads do not exist, especially since the common practice is to diffuse
the loads to the layer of a slab. We are then led to calculate integrals to obtain the value of the moment in the slab (for
more details, please refer to Pücher’s original publications).

By distributing the load over a 20×20 cm² square, the integration of the surfaces shows that the maximum moment is
equal to 3.0 KN.m.

Another example of the use of Pücher’s charts is given in the example of modeling a Br wheel.

It should not be forgotten that the calculations are generally carried out with raw sections (formwork) and elastic
materials (linear behavior). In reality, the cracking of reinforced concrete will lead to a redistribution of forces that
tends to reduce these peak effects. This type of calculation is not (to date) current practice.

It is therefore necessary to know how to correct simply, often manually, the results of a linear calculation.

Examples of correcting the results of a linear calculation are given below.

D.5.1.2 Finite Element Peak Studies

What happens when this same slab is calculated using finite elements?

Slab of 3 m articulated span subjected to a concentrated load of 10 KN at its center (node stresses)

Same in 3D

We note the appearance of a peak whose maximum value is not infinite, but equal to 2.96 KN.m.

We saw that the software calculates the efforts at the elements' integration points, then extrapolates the results to the
center and then to the nodes of the elements. A node is generally common to 4 elements, so there are 4 results per
node (one for each element). What will the chosen result be? The maximum value? The average value? Indeed, finite
element software does not propose a single result for each calculation, but several results depending on the options
chosen by the engineer: the software is able to draw up maps of results from the forces on the nodes, or from the
forces at the center of the elements, they can be smoothed, not smoothed, etc.

The engineer must choose the visualization options carefully, as the results vary greatly depending on the option
chosen. This is what we propose to show with the example below.

The previous figure shows the moments calculated on the nodes of the elements whose maximum value is 2.96 KN.m,
it is very close to the value of 3.0 KN.m calculated manually.

We take the same example by displaying the calculated moments in the center of the elements (instead of the nodes);
the central peak is smaller: 1.92 KN.m for 2.96 KN.m previously. This result is also far from the manually calculated
value of 3.0 KN.m.
Slab of 3 m articulated span subjected to a concentrated load of 10 KN distributed over 20×20 cm² in its center (forces
in the center of the elements).

The representation below of the curve of moments on a section in the middle of the slab allows to understand these
differences.

1. Node efforts

Below is the curve of Myy moments calculated by smoothing on the nodes.

The forces are calculated in the integration points of the elements and then extrapolated to the nodes.

The maximum moment is 2.96 KN.m in the center of the slab according to the manual calculation.

2. Unsmoothed efforts on the elements

The forces are calculated in the integration points and then averaged to obtain the force at the center of the element.

The maximum value is 1.92 kN.m , average of the central elements.

We do not find here the manually calculated value of 3.0 kN.m, but a "smoothed" value on the elements surrounding
the peak. From this example, we will note that the forces at the nodes give results that are consistent with those
calculated manually, which is not the case for the forces calculated at the center of the elements.

3. Smoothed efforts on the elements

The forces are calculated in the center of the elements and then smoothed between them.

This curve gives the impression of a curve which is extrapolated on the nodes, whereas the extrapolations are carried
out only on the results at the centers of the elements, the resulting curve is without physical meaning and therefore
"false".

On the other hand, in another case, this option of smoothing the efforts at the center could be useful if we want to
know the efforts in the plane of the walls.

To be valid, the width of the mesh should be equal to the thickness of the wall.

The visualizations of the finite element results at the peaks give very different results depending on the options
chosen by the engineer. These results cannot be taken as they are, but must be analyzed and interpreted by the
engineer.

The figure below shows the values obtained with a point force of 10 kN at the center of the slab.

The moment value increases to 5.2 kN.m and the pace of the moment curve shows a clear peak.

Later in this chapter, we will see that point forces (which have no physical meaning) lead to effort peaks and that it is
better to avoid using them in order to obtain accurate local results.

D.5.2 Peak analysis method

The reinforcement specifications often show peaks of steel which have very important consequences in the
dimensioning of the reinforcement.

The user is often unaware of these peaks: should they be taken into account by considering that they are structural, or
should they be ignored by assimilating them to numerical calculation problems?
 Example: the peaks shown above at the fixed end of this cantilever beam are of course structural.

The answer to this problem lies in understanding the functioning of the structure and the path of forces at the peak
level, an analysis that is indispensable to solve peaks (limit the maximum demandclipping, linearization of
reinforcement).

This can be difficult in the case of complex models, but is always essential.

Three types of analyses are possible:

Analyses Objective

Identifying the singularities of the modeling at the peak to determine

1: Geometrical analysis
its geometric origin.

Making a first distinction between membrane, shear and bending

2: Steel sections analysis Axi, Axs, Ayi, Ays
forces

3: Analysis of the efforts generating the peak Detecting the "faulty" component(s) and quantifying efforts

D.5.2.1 First analysis: geometrical

Experience shows that 90% of the peaks are located at the level of geometric singularities (columns, supports,
wall/wall or wall/slab intersections, etc.). This is due to the fact that the modeling, carried out from middle planes or
middle fibers, does not represent the elements with their real volumetric geometry (for example, a slab is represented
by a plane element and the columns on which it rests, by wire elements). In a pictorial way, this leads, as explained in
D.5.1, to taking into account forces applied to null surfaces, thus inevitably leading to numerical problems. It is
therefore essential to identify these singularities on the model.

1. Examples of peaks related to geometry

Below we share an example of a slab on a column/beam array grid for which moment peaks appear at the supports
formed by the columns.

These peaks must be interpreted in detail, it is possible to reduce the values of the dimensioning moments by limiting
the maximum demandclipping at the beams.

Example of a slab on a grid n array of columns and beams

 Visualization of bending moments in the slab

Peaks appears; they can be limited clipped at the beams:

In the above case, conventional clipping is performed in the plane of the edges of the beams or column in the case of a
flat slab on headed columns.

Some types of software allow to define supports with plan dimensions to directly obtain the efforts in the plane.
Example of a 10×10 m² slab, 30 cm thick, subjected to its self-weight, supported linearly on an edge and on two point
supports in line with two posts 1 m from each corner. The left support is a classic point support, the right support is a
column type support of 50×50 cm².

The representation of efforts with and without reduction of effort shows a significant difference in values.
Note: It is always advisable to check the methodology used by the software and to ensure that it is compatible with the
regulatory justifications to be carried out.

2. Peaks caused by point forces

The peak treatment at concentrated loads is similar to the peaks caused by point links. (A force or torsor can represent
the effects of point support; there is a strict equivalence).

Example:

We take the trivial example of a load arriving via a column on a wall (30 cm thick wall, 55×30 cm² column).

Depending on what we want to calculate, the approach will not be the same. If we are looking for a global vertical load
calculation, the approaches on the left, either via a point load or via a wire bar are perfectly suitable. However, if we
are interested in local effects, it is absolutely necessary to use a distributed load to minimize the stress peak, which
does not facilitate automated calculations (and does not anticipate other manual calculations to be performed:
diffusion, punching ...).

Peak related to the way of modeling of a load on a slab: case of a wheel load on a bridge slab.

A concentrated load which is perpendicular to a slab generates a moment peak (see D.5.1). This peak must be
processed to calculate the reinforcement. It can be reduced by replacing the concentrated load by a loading block that
takes into account the diffusion of the load in the slab (diffusion of a wheel on a bridge slab).

Example:

The longitudinal moment for the same 100 kN load is compared using four approaches:

a point force in the center of a mesh;

a point force on a node;
a pressure corresponding to a force of 100 kN on a surface of 0.25×0.25 m² (the mesh);
the force of 100 kN/4 = 25 kN distributed to the 4 nodes framing a mesh of 0.25×0.25 m².

100 kN force in the center of a mesh → Mx=26.62 kN.m/m

100 kN force on a node → Mx=48.12 kN.m/m

Distributed pressure on a mesh → Mx=26.62 kN.m/m

1/4 force on 4 nodes → Mx=26.62 kN.m/m

In conclusion, we realize that modeling a distributed load by a point force can be very detrimental, especially if this
load is applied to a node of the mesh: it is better to use pressure (knowing that the software distributes the loads to
the nodes), or alternatively to split the force into several loads to avoid the potential peak force.

Tip: Avoid modeling distributed forces by a point resultant for the calculation of forces in slabs or one-way joist
slab!

Make sure that the mesh size is in correct proportions with both the thickness of the plate and the
impact area of the smallest load.

In addition, to illustrate the different results that can be obtained for an extremely simple case, the Working Group
calculated the forces generated by a Br wheel of installment 62 Title II, using several software programs and several
smoothing approaches. The study is available in part 3 or directly below: example of modeling a Br wheel.

Punching

It has already been evoked, several times, the fact that finite elements did not deal with certain subjects such as the
shifting of moment curves, the limiting of maximum demandclipping of shear forces close to the supports or the
punching checks.

Except in special cases, therefore, the checks related to punching still have to be done manually. We will illustrate this
on the example of the load on the one-way joist slab bridge shown above.

The modeler might be tempted to reason about the shear stresses averaged over the punched surface from the FE
calculation. We refer to the Limit state design (LSD), also known as Load And Resistance Factor Design (LRFD), which
has the advantage of being simple. The comparison stress is simply the load divided by the perimeter of the impact
which is diffused at the average layer and by the thickness of the slab.

Let τ=100 kN/(4*0.25 m)/0.25 m=400 kPa (the slab is 25 cm thick).

The FE calculation, on the other hand, results in smoothed stresses (here in the most unfavorable zone, not strictly on
the load diffusion perimeter!) of the order of:

τ=5.33 kN (integral over 0.125 m, read below)/0.125 m/0.25 m= 171 KPa;

τ=5.42 kN (integral over 0.125 m, read below)/0.125 m/0.25 m= 173 KPa.

We are far from the 400 KPa that the regulatory approach to the BAEL gives - taking into account FE values would not
be safe. But these are two different approaches: shear in the slab is used to calculate the shear reinforcement and
punching is another type of verification.
 For the calculation of reinforced concrete structures, it should be noted that the peaks are caused by point
forces or supports.

3. Peaks caused by mesh size problems

The geometrical marking of the peak often makes it possible to detect the causes.

Example of an inconsistency of results related to modeling: a graphical construction can make us think that an
edge is common to two shells, whereas in reality there is an extremely small shift that leads to aberrant results. To
illustrate this case, we extend the slab of an example above by another rectangular element that we consider to be
strictly in the same plane as the previous one. The results should alert us that some peaks appear:
By zooming, we can see, on the right side, that the connection is only made on a few points (corresponding to a
geometrical tolerance of the software) while on the left side there a very small shift (and which is caused in this case!)
but which prevents the connection.

You can see in the figure below that not all the nodes on the common edge are connected: this should alert us.

The observation of the deformation should also alert us:

In case of difficulties, the user can himself or herself create beforehand the nodes of the joined edge, if necessary with
bar elements that will be deleted once the mesh is reliable and fixed. Although this approach may seem tedious at first
glance, the time saved can be well worth it.

We must remain cautious about the use of automatic corrections proposed by some software (in this case the linking
of nodes through kinematic links) which can lead to local stress peaks. In this case, it is better to erase some shells
and start again some parts of the model. See also § C.3.7.
D.5.2.2 Second Analysis: Study of steel sections perpendicular to the peak.

To help us locate the origin of the peak, we can analyze the steels details on each face, in both reinforcement
directions. Indeed, the analysis of the steel sections of Axi, Axs, Ayi and Ays perpendicular to the peak quickly provides
important information:

high NXX and NYY membrane stresses are detected by large and equal steel sections on both sides. If all 4
steel sections are equal, the element works in shear in the NXY plane;
high bending moments MXX and MYY can be detected by important steel sections on one fiber and very weak
ones on the other one.

To illustrate this, the table below summarizes the consequences on the reinforcement of each stress component:

The code X indicates an important value of the steel section, the code 0 a low value.

From there, the analysis described below is carried out.

D.5.2.3 Third Analysis: Effort analysis perpendicular to the peak

If the two previous analyses are not enough to explain the peak, it is then necessary to study the effort components in
detail to determine their intensities and identify cases of dimensional loads.

This analysis, which is often long and delicate in the case of complex modeling, can be greatly simplified for simple
modeling by carrying out effort mapping or local cuts.

D.5.3 Peak Resolution: determination of final reinforcement

After having understood the functioning of the structure and mastered the effort paths, the engineer has all the
elements to solve the peak and deduce the strictly necessary reinforcement.

D.5.3.1 Cases where maximum limitingclipping or smoothing are not possible

In some cases, the study of the effort path shows that the peaks are real and cannot be limitedclipped. This is the case
of lintels which show very important peaks at their ends, which is logical because they are fixed end beams that must
be calculated according to the mechanics of materials rules and reinforced concrete (fixed end beam subjected to
constant shear).

D.5.3.2 Limiting Clipping moments on supports

Article 9.5.3.2.2 of EC2-1 allows to limitingclipping of the moments in the plane of the supports (walls).

1. Beam resting on a column - example 1

Consider the example of a slab with 2 spans of 6 m which are uniformly loaded by 25 kN/m².

Theoretical moments are -11.2 kN.m on supports and 6.3 kN.m in span.
The FE software shows a peak of -12.1 kN.m on the support, which is real.

Limiting Clipping at the plane of the wall:

To obtain the moment at the plane of the wall (thickness = 20 cm), the user has several solutions:

either making cuts at the support level;

or not visualizing the support area (but this requires an adaptation of the mesh);

Moment Clipping at the plane of the supports = -10.6 kN.m

or adapting the mesh in order to make the width of the supported mesh equal to the thickness of the wall, as
shown below.

The moment in the center of the mesh will then represent the moment at the plane of the support.

The same type of reflection and approach is to be carried out in order to reduce the transverse shear effort, if
necessary.

2. Beam resting on a column - example 2

The example below shows two ways of modeling a simple column-beam structure to highlight the effects of the real
dimensions of the structures.

The principle is as follows, for an out-of-plane width of 1 m:

Firstly, a wireframe model is used to calculate the moments at the central support (truncated view):

The moment on the central support is 82.09 kN.m, giving an extreme upper and lower fiber stress of 6M/(bh²)=
(6*82.09)/(1.0*0.5²) = 1970 kPa = 1.97 MPa. The stress in the plane is (6*70.00)/(1.0*0.5²) = 1680 kPa = 1.68 MPa.

Below is the result (in stresses) of the same calculation performed on a model with plates, locked in out-of-plane
displacement. This model allows to represent the behavior in section, with the real thickness of the elements.
(truncated view).

The stress obtained in the upper fiber is 1.58 MPa. (The lower fiber stress is not representative because it is calculated
in a singular area that constitutes the right angle).

This model highlights the legitimacy of limiting clipping in practice.

D.5.4 Reinforcement smoothing

The following method, derived from a common practice in the field of nuclear civil engineering, provides simple
rules for smoothing the reinforcement.

Its application is nevertheless subject to the engineer's judgment. In particular, in the case of floors of ordinary
buildings or bridge slabs, these values are probably too favorable and it seems possible to reduce the size of the
smoothing to half of the values below, therefore limiting this size to 2h (h being the thickness of the slab).

In this context, the smoothing of the longitudinal and shear reinforcement sections must be carried out:

between adjacent elements (and not successive in relation to the reinforcement direction); the smoothing is
done perpendicularly to the reinforcement direction;
over a reasonable distance (engineer's judgement) and less than a value that is correlated to the plate
thickness and the plate span.

The current common practice is to average the results of three elements: the element for which the maximum is
observed and the two adjacent elements, limiting the width over which redistribution takes place to 4 times the slab
thickness (see table below).

E.g.: if the size of the elements is 1 m, averaging over three elements is the same as averaging over a width of 3 m.
For a 0.5 m slab, this width is limited to 4 times the thickness of the slab (2m) which leads to averaging over only two
elements.

The distribution width should also be limited according to the slab span or the wall height, because the smaller the
span (or the height), the smaller the distribution width should be.

Proposed rules for smoothing the longitudinal reinforcement peaks of slabs (resulting from N, M forces)

They are presented in the following table. They are valid for a slab subjected essentially to membrane forces and
bending moments due to out-of-plane distributed loads, and with a sufficiently fine mesh size that has:

an odd number of elements in both span directions;

at least 5 elements according to the small span (7 elements if possible);
a mesh size equal to the thickness of the loadbearing elements.

Width over which longitudinal and

Limitation of the distribution width as a Limitation of the distribution width
transverse reinforcement sections can
function of the plate thickness h according to the plate span L
be smoothened

Zone where efforts can be redistributed in

4h 0.5L
both directions

Zone where the redistribution of efforts can

only be done in one direction (at the edge of 2h 0.25L
the shaft)

For out-of-plane concentrated loads, the distance of the load from the support and its diffusion must also be taken into
account.

Reinforcement peaks are frequently located at the edge of the shaft, in which case the redistribution of forces can only
be in one direction and therefore over limited widths:

if additional reinforcement is required, it is always better to place it as close as possible to the shaft edges.
if, after smoothing the reinforcement as specified above, the current reinforcement is sufficient, it is however
recommended to place additional reinforcement at the edge of the shaft if more than one current
reinforcement is cut by the shaft in one of the two directions.

In the case of small openings (sleeves in particular) that fit into the reinforcement mesh or cause the interruption of a
single reinforcement, it is possible to dispense with additional reinforcement.

Proposed rule for smoothing longitudinal reinforcement peaks in membrane elements (walls)

For elements subjected to tensile membrane stresses, redistribution can only take place in one direction.

Width over which longitudinal and

Limitation of the distribution width Limitation of the distribution width
transverse reinforcement sections can
according to the plate thickness h according to the plate span L
be smoothened

Effort can only be redistributed in one

2h 0.25 L
direction (at the shaft edge).

In bracing walls subjected to an axial bending moment perpendicular to the wall plane, the tensile membrane stress
varies linearly.
When smoothing the reinforcement peaks, the smoothed reinforcement must be extended over a length large enough
to maintain the bending capacity: F1 x d1 > F0 x d0, where:

F1 = Resulting stress taken up by the reinforcement after smoothing over the length L;
d1 = distance between the resultant F1 and the zero moment point;
above parameters with index 0 = before smoothing.

For elements subjected to membrane shear, it is possible to transfer part of the required section AX to section AY, if it
is overabundant and vice versa. The Capra Maury method, which optimizes the sum of the reinforcement sections AX
+ AY, is used in common software.

The AY cross-section can sometimes be determined by the minimum reinforcement condition and the expected AY
cross-section is then bigger than the required AY cross-section from the strength calculation, thus allowing a
redistribution of reinforcement cross-sections. The strength of the section should then be checked with the new
reinforcement sections.

Conclusion concerning the smoothing of longitudinal reinforcement peaks

In all cases, it is necessary to take into account the origin of the steel requirements by analyzing the stresses (Nxx,
Nyy, Nxy, Mxx, Myy, Mxy) and to interpret the results case-by-case, with a concrete approach.

In general, the wall reinforcement comes mainly from membrane forces (Nxx,Nyy,Nxy), the bending forces
(Mxx,Myy,Mxy) being then negligible. On the other hand, floor reinforcement is mainly due to bending forces
(Mxx,Myy,Mxy), and in some cases by membrane forces (Nxx,Nyy,Nxy) when the building is subjected to horizontal
forces (wind, earthquake) or irregularities (wall beams).

A wall beam is a good example of an irregularity producing horizontal forces in floors: the following example is based
on the wall beam resting on 2 columns studied in chapter D.4.5, but with a lower floor; the tensile stresses in red show
that the bottom tie is formed not only at the base of the wall but also in the lower floor:

A wall beam is a good example of an irregularity producing horizontal forces in floors: the following example is based
on the wall beam resting on 2 columns studied in chapter D.4.5, but with a lower floor; the tensile stresses shown in
red show that the bottom tie rod is formed not only at the base of the wall but also in the lower floor:

Tractions are even more important (here +50%) in the case of openings at the wall base:
Proposed rule for smoothing shear reinforcement peaks (the shear being perpendicular to the elements)

Generally, peaks in shear forces which are perpendicular to the elements occur at the intersection of several plates.

Reinforcements shear stress peaks often occur when shear stress is concomitant with high traction.

As a reminder, for the justification of reinforced concrete, the concomitance of a shear force with a traction requires
particular attention because it means that there is no compression strut formation in the concrete and therefore a risk
of breakage.

The resolution of this type of peak requires to look back at the stresses, to average the shear and normal stresses and
to recalculate the reinforcement.

Approach illustrations:

Example 1:

Example 2:
The reinforcement peaks read on the charts are smoothed according to the following principle:

Smoothing principle of the local peaks of longitudinal reinforcement

Smoothing principle of the local peaks of longitudinal reinforcement

Smoothing principle of the local peaks of transverse reinforcement

D.5.5 Stress distribution in beams and slabs

It is important to remember that, with few exceptions, the calculation of the internal forces in the elements is a linear
elastic calculation with a constant concrete modulus.

Peak stresses often determine a cracking of the reinforced concrete section and thus a local reduction of the peak
stresses and a redistribution of the stresses.

It is sometimes useful - and it is accepted - to redistribute bending moments. A distinction must be made between:

the SLS (limited redistribution possibility, taking into account the weakening of the section’s stiffness due to
concrete cracking in the highly stressed area of the peak);
the ULS (possibility of wider redistribution: same phenomenon as the one taken into account for the SLS with
the addition of the wider redistribution possibilities indicated in Eurocode 2 in paragraphs 5.5 "Linear-elastic
analysis with limited redistribution of moments" and 5.6 "Plastic analysis".

However, it should be noted that the professional recommendations for the application of standard NF EN 1992-1-1 (NF
P 18-711-1) authorize, for buildings, to use in the SLS, a moment redistribution with the same redistribution
coefficients as in the ULS.
Within the Eurocodes, the ratio δ of the moment after redistribution to the elastic bending moment must be greater
than or equal to 0.7 for class A longitudinal reinforcement and 0.8 for class B or C.

This redistribution of moments in a continuous beam is possible if the beam does not participate in the bracing. It is
more delicate if the beam belongs to a portal frame (beware of the elastic moments coming from the portal frame
effect).

We must not forget to take into account the impact of the redistribution of bending moments on shear forces.
$translationBooks

PART 2 – USEFUL ELEMENTS

D6. Understanding and analyzing

the peaks (case study about steel
assembly)
D6. Understanding and analyzing the peaks (case study about
steel assembly)
The results of a calculation performed on a surface finite element model (plate or shell type elements) may present
stress peaks (or singularities). The presence of stress peaks is not annoying if they are located in areas out of interest
with respect to the objectives of the study and if their presence is justified.

It should be kept in mind that a stress peak, i.e. a very high stress, must concern two or three nodes, or even a single
node at most (but the structure may nevertheless present several stress peaks).

When dimensioning an assembly of plates which are welded together (metal bridge box at a support for example), the
realization of a finite element model is necessary to correctly take into account the diffusion of stresses. And of course,
from the same model, the stresses in the plates can be extracted to proceed to the dimensioning of the assembly of
these plates.

Nevertheless, the general idea will remain to always keep a critical eye on the results obtained. They will have to be
analyzed properly before they can be considered reliable with regard to the objectives of the study carried out.

This analysis of the reliability of the results often comes up against the interpretation of the possible presence of stress
peaks.

The purpose of this paragraph is to give clues to help the modeler who is responsible for assessing whether or not a
stress peak is a problem.

The appearance of stress peaks can have different origins, of which the main ones are:

the presence or omission of details such as connecting fillet;

meshing inconsistencies;
setting up loads and boundary conditions.

D.6.1 Modeling – Details

Before embarking on the surface modeling of a metal plate assembly, it is essential to know and master the software's
CAD tools and its meshing capabilities.

It is customary when creating geometry to neglect certain details. Two cases among others:

Connecting fillets - if the connecting fillets of an opening in the web of a beam are not shown, stress
concentrations will inevitably be all the higher that the web loads are important and the angle is acute (case
of non-perpendicular opening edges);

Note 1: The modeling of a connecting fillet will no longer create a stress peak, but a simple stress concentration. Stress
concentration charts are based on this type of modeling.

Note 2: In fracture mechanics, the crack model characterized by a zero angle is a special case used to model crack
propagation which is based on the notion of stress intensity factor.

Transverse stiffeners - the main beams of a bridge require transverse stiffeners at the support zones. The modeling
of these stiffeners is often necessary in view of the forces that can transit in this zone and locally stress the web.
According to the bridge’s rules, these stiffeners have a particular geometry and too often this geometry is
oversimplified during the modeling process;

Real geometry / Simplified geometry

The stress transition may not be affected, but the mesh of the stiffener will have to be controlled so as not to end up
with degenerated elements that can create singularities. And at the tip of the stiffener, a stress peak may appear
which in case of misinterpretation may lead to false conclusions. See also § C.3;

Details such as connecting fillets, mouse holes, chamfers, etc. may therefore not be modeled if they are located
outside an area of interest. The time spent modeling details may not be negligible, but if the objective of the
simulation is to calculate a stress or a deformation in an area with such details, it is essential to model them and,
moreover, to associate a representative mesh to them.

Non-representative mesh / Representative mesh

In the absence of detailed blueprints of the constructive details to be modeled, we will rely on the constructive
provisions and bridges rules, which, in terms of metal assemblies (connecting fillets, local plates, stiffeners, etc.),
provide a basis generally leading to a satisfactory diffusion of efforts. For example, the Commission for the
standardization of metallic and mixed construction (CNC2M) recommendations for the dimensioning of steel beams
with openings in the web according to NF EN 1993 specify that, for a rectangular opening, the radius of the connecting
fillets must be bigger or equal to twice the web thickness without being smaller than 15 mm, or that an isolated
opening with a maximum dimension less than 10% of the web height of the beam is not considered significant (this
opening must nevertheless be taken into account when checking the section).

In some cases, the consideration of details cannot be neglected (see also § C.2.2). For example, during a fatigue
verification of the connections of an orthotropic slab, simplified calculation models described in the design regulations
may be used. However, these models are safe and, if the fatigue strength is not justified with these simplified
approaches, two possibilities are available for the designer:

either modify the geometry of the assembled parts, which is often difficult in the execution phase;
or perform a more refined calculation on a plate finite element model.

In order to take into account the vehicle traffic effects (longitudinal and transverse effects, influence lines), it may be
necessary to model a fairly substantial length of the structure. On the other hand, the detail modeling zone may be
limited.
D.6.2 Modeling - Surface model creation and meshing

It is also important to have a good understanding of the software's meshing possibilities. Most types of software can
automatically mesh a plate assembly.

The user's intervention can then be limited to the definition of a few parameters such as the average size of the
elements or their shape (quadrangular, triangular). It is better, however, to be able to control the mesh by choosing
the order of the surfaces to be meshed, by imposing finer mesh sizes in some zones, in order to avoid the presence of
degenerated elements.

Example of degenerated mesh

In a complex assembly, due to the shape and number of parts assembled, the risk of mesh inconsistencies is high. To
reduce this risk, several points should be considered, as a reminder:

The analysis and the perception of the mechanics of materials operation of the assembly by the engineer
prior to the modeling are strongly recommended. Some reinforcements and details that have only a very
localized effect on the overall behavior of the assembly can be omitted, thus reducing the number of
interfaces between plates;
To avoid mesh discontinuities, all plate intersections must appear. For example, in the case of modeling an I-
beam, each of the flanges will appear composed of two identical surfaces on both sides of the web. At the
intersection of each flange with the web, a single line common to all three surfaces (the two surfaces
constituting the flange and the web surface) must be present;

Elementary surface decomposition of an I-beam

do not hesitate to decompose the extended surfaces into several quadrangular shaped surfaces, the meshing
will be facilitated as well as for example the application of loads like as pressures;
 The work structure can present panels with a deformed geometry. The software can allow the creation of
deformed surfaces, within a certain limit. Therefore, depending on the deformation, the initial panel will have
to be decomposed into several sub-panels (the common boundary between two sub-panels being made up of
the same entities). During the meshing process, elements resting on a deformed surface will also show
warping. Depending on the software, the warping limit allowed for a CAD surface (on which the finite element
mesh will be supported) may be different from that allowed for finite elements. In any case, it is always
possible to approximate warped elements using successive triangular flat surfaces.

Example of panels with deformed surfaces

The assembly to be modeled may have cylindrical tube intersections. If during meshing elements without a
middle node (4 node elements) are used, the mesh will present facets; the larger the size of the elements, the
larger the size of these facets. This "facetization" can therefore be reduced by increasing the number of
elements. The use of mid-knot elements can also be a good solution, provided that the software used allows
the mid-node elements to follow the curvature of the CAD surface (or CAD line).
if a tube is connected to a plate, pay attention at the connection, to the compatibility of the types of elements
used (for example connection of elements with 4 and 8 nodes). In some cases, discontinuities in the results
may occur (see § C.6.3);

D.6.3 Modeling - Loading and boundary conditions

One of the frequent causes of the appearance of stress peaks is the presence of point loads or point blockages.

If a point load (respectively a point support) put in place is outside the area of interest, the presence of a stress peak is
not annoying. On the other hand, if it is in the area of interest, rather than a point force, it is better to apply the force
on several nodes (respectively block several nodes).

As described in the previous chapters, in reality, point forces and point supports do not make physical sense (note,
however, that in a wire beam model, point supports do not create peaks).

Practical application:

Let us consider a HEB beam modeled in plate elements. A stress peak would appear, if the stress torsor in a section
was applied at a node. It is better to get closer to reality, by transforming the stress torsor into normal and tangent
stresses (as linear forces, if any) and proceed as follows:

the normal effort will be distributed over all the nodes of the end section;
the shear force will be distributed over all the nodes of the web of the end section;
the moment will be decomposed into two forces distributed over all the nodes of the two flanges of the end
section.
 Normal force / Shear force / Bending moment

The same will be done to model the beam supports. Let us consider the two following cases:

bi-articulated floor beam (end web angle connection) subjected to a distributed load. We will consider two
cases differing by the boundary conditions; `

1st case: at each of the ends, installation of a point vertical blocking (the transversal blocking of two nodes is
carried out to stabilize the beam in rotation around its axis).

1st case - boundary conditions/longitudinal iso-stresses (in MPa) - the peak stress area is circled

2nd case: in each extremity the vertical and transversal blocking of the nodes of a web part is performed in order to
get closer to reality.

2nd case - boundary conditions / longitudinal iso-stresses (in MPa)

The stress peaks appearing in the 1st case are thus due to the presence of a point nodal reaction. However, by
observing the results only on the central part of the beam, we find identical results to those of the 2nd case.

1st case - central part - longitudinal iso-stresses (in MPa)

bi-supported beam (supports under the bottom flange) subjected to a distributed load. First, only the beam (type H)
is modeled and boundary conditions are applied such that, near each of the two ends, the nodes of the bottom flange
are blocked in the vertical direction.

Case without stiffeners - boundary conditions / Longitudinal Iso-stresses (in MPa)

Case without stiffeners - vertical iso-stresses (in MPa)

Stress peaks appear at the lower flange/web intersection despite the fact that vertical blocking was performed on
several nodes. The problem here is more of a design problem. It is common practice to install transverse stiffeners
perpendicularly to the supports (see design regulations for the transverse force resistance of the beam web and
checks for local buckling). If we take the model by adding transverse stiffeners perpendicularly the supports, the peak
stresses perpendicular to the supports disappear.
Case with stiffeners - boundary conditions / longitudinal iso-stresses (in MPa)

The stress peaks appearing in the absence of stiffeners are therefore due to a design problem. However, if we observe
the results only on the central part of the beam, we find similar results to the case with stiffeners.

Cases without stiffeners - central part - longitudinal iso-stresses (in MPa)

D.6.4 Results analysis - Model validation

Once the modeling is complete, having best followed the advice given above, the calculation is executed. It is
necessary to carefully analyze all the messages that can be generated by the software. Usually, an error message will
block the solver, unlike warning messages, which must be carefully analyzed and interpreted. It is recommended not
to ignore these messages without having carefully evaluated the possible consequences, even if it means calling the
editor's hotline.

It is important to understand the possibilities offered by the software for the display and the types of results. The
following is a non-exhaustive list of parameters that should be considered:

the coordinate system in which efforts and moments, displacements or constraints are displayed: is it the
global coordinate system of the model, local coordinate system of the elements, nodal coordinate system?
Sign conventions for efforts and stresses;
the convention used by the software for naming forces and moments: for example, some software calls My
the moment around the Y axis, while for others it corresponds to the moment parallel to the Y direction. This
is particularly true when using torsors supplied by a third party;
the different stresses: directional stresses, principal stresses, Von Mises equivalent stresses;
the type of stresses displayed and how they are calculated: node or element stresses, mean surface or
facesheet stresses;
the possibilities of stress extraction at Gauss points;
the relevance of the stress display scale, which can distort the results interpretation.

In the example below, stresses well above the elastic limit appear on localized peaks.

Example of a scale not adapted to a stress analysis

An upper bounding of the scale, possible with a lot of software, makes it possible to show the areas impacted by
overruns:
 Scale bounding

The software's operating parameters being well understood, before any exploitation of the results and analysis of
possible remaining stress peaks, it is advisable to carry out some simple checks to validate the calculation model:

control the vertical load calculation (support reactions) in order to ensure that all the introduced loads are at
the level of the declared model supports;
control the configuration of the global deformation of the structure;
Control the magnitude of the displacements and their compatibility with the boundary conditions imposed on
the structure and the applied load(s).

Finally, these are the fundamental checks that must be done for any finite element modeling (see § E.2 Self-checking).

D.6.5 Results analysis - Acceptance of residual peak stresses

As mentioned above, it is possible to use a model with stress peaks if we are certain that these peaks do not disturb
the outcome of the objectives of the study. In any case, it is up to the engineer to evaluate whether or not a stress
peak is disturbing or not, using his or her experience and critical thinking.

Modeled zone without consideration of design details

The implementation of connecting fillets or local stiffeners is intended, among other things, to ensure a better diffusion
of efforts. The constructive provisions and the bridges regulations in terms of metal assemblies are in accordance with
this.

Therefore, the non-modeling of a constructive detail will probably result in a stress peak.

However, the modeling of the constructive detail does not exempt us from the presence of a stress concentration all
the stronger that the mesh of the local area of the detail is thick. Sensitivity tests on the size of the elements can be
performed in order to get a good understanding of the displayed results. Stress concentration charts can also be
useful.
 Examples of stress concentrations at singularity level

Stress concentrations peaking at values above the elastic limit of the material may be acceptable at the ULS if they are
very localized and in facesheets. However, their justification may sometimes require an elasto-plastic calculation,
especially if the area of stress exceeding the elastic limit is transverse (risk of plastic hinges). If the software allows it,
this elasto-plastic calculation is performed considering a bi-linear behavior law of the material. Appendix C
(informative) of EN 1993-1-5 authorizes a limit value of the maximum deformation of 5% for the zones in tension. For
areas in compression, it is advisable to remain vigilant with regard to local warping phenomena.

Zone with a mesh singularity

The meshing of the areas of interest must be adapted to the fields of expected stresses or deformations.

If the mesh is poorly adapted with, for example, abrupt changes in size, stress jumps from one element to another are
to be feared. Faulty discretization will result in significant discrepancies between simulation results and reality. These
discrepancies can be reduced by carrying out a sensitivity study of the mesh: the mesh is considered satisfactory
when the refinement leads to a small variation of the result; for example, a variation of less than 5% for element sizes
divided by 2). But be careful, it is not because a result seems to have stabilized at 5% that we approach the reality at
5%.

It is not necessary to refine the mesh over the entire model. Most of the software allows to display error maps to locate
areas with high stress jump. Some software even allows to automatically correct the mesh to reduce these deviations
(adaptive mesh).

Zone with punctual effort or punctual blocking

We have previously presented the effects of the implementation of a punctual effort or blockage.

However, even if care has been taken to distribute the load over several nodes, stress concentrations may occur with
maximum values above the elastic limit of the material. This situation may nevertheless be quite acceptable
depending on the purpose of the study and modeling. Take the case, for example, of the modeling of the attachment
zone of a suspender cable. The stress from the suspender is distributed sinusoidally over the nodes of half the
circumference of the bore (hole).
We obtain a stress concentration around the axis bores with a maximum value of 558 MPa, a value well above the
elastic limit of 400 MPa. However, this maximum value is not to be compared with the elastic limit. The stress
concentration here is due to the diametral pressure and the regulations indicate that the diametral pressure resistance
of the plate is equal to 1.5.t.d.fy. Since the thickness t of the coping and the bore diameter d are taken into account in
the modeling, 558 MPa should be compared to 1.5 x 400 = 600 MPa.

D.6.6 Synthesis

If the stress peaks are due to singularities (reflex angle, interface between two different materials, point loads), they
can be neglected ... if the state of stress in the proximity of the singularity is not part of the objectives of the FE study.
Otherwise, it is necessary to improve the modeling (by replacing the reflex angle by a connecting fillet, replacing the
zone of discontinuity between behavior laws of different materials by a transition zone in which the parameters vary
continuously, replacing a point force by a contact pressure on a non-zero surface).

If these stress peaks appear in areas outside the proximity of a singularity, we should do a successive mesh
refinement of these areas to know the more realistic stress level.

As far as smoothing is concerned, there is no simple and direct method. The reader will be able to look at what is
practiced for fatigue calculations, in the case of a large stress gradient in the weld proximity in connections, with the
application of the hot-spot geometrical stress method (see bibliography (ref. Hobbacher)).

Concerning the value of the maximum stress obtained, it must be compared to the limit value specified by the
Calculation Standards. For steels, the Eurocodes standards define the value of the elastic limit according to the grade
of steel and the thickness of the plate or tube; for example, for a tube in S355 steel with a thickness of 35 mm →
elastic limit = 345 MPa according to EN 10210.

Any exceeding of this limit:

must therefore be justified. And this may be paradoxical, but it is easier to justify a stress overstepping in the
case of a singularity than in the case of a current zone;
must also be acceptable. If the elastic limit is exceeded, plastic deformation will occur. An elasto-plastic
calculation will give information on the rate of deformation (for structural steels, the Eurocode EN 1993-1-5
specifies a maximum principal deformation limit of 5%). The standard specifies the ULS criteria that can be
used. For zones in tension, it is a question of reaching a limit value of the main membrane deformation (limit
of 5% recommended) and for structures sensitive to warping phenomena, reaching the maximum load.

Finally, it should be remembered that an FE calculation (unless certain very specific calculation options are activated)
does not take into account phenomena other than shear lag, for example instability phenomena such as warping and
plate buckling.

Numerical application: Illustration of chapter D.6 - Peaks in angles, mesh incidence and fillet incidence. Link to
the file.
Illustration of Chapter D.6 - Peaks in angles, mesh incidence and fillet
incidence.
Authors: D. Guth (Arcadis) - P. Mangin (CTICM) - v0 – Jun 25th ,2020

1. Input data

The modeled structure is a simple flat angle bracket made of 10 mm thick steel (dimensions in m on the view):

We apply a load of 100 kN/m to the upper horizontal edge.

We do not take into account warping phenomena - this is a "simple" mechanics of materials model.

We start with a mesh size of 10 x 10 mm² and refine the mesh in the peak areas, as shown in D.6.

"If the stress peaks are due to singularities (reflex angle, interface between two different materials, point
forces), they can be neglected ... if the state of stress near the singularity is not part of the objectives of the
FE study. Otherwise, it is necessary to improve the modeling (replace the reflex angle by a connecting fillet,
replace the zone of discontinuity between behavior laws of different materials by a transition zone in which
the parameters vary continuously, replace a point force by a contact pressure on a non-zero surface).
If these stress peaks appear in areas outside the proximity of a singularity, we should successively refine the
mesh of these areas to know the most realistic level of stress.”

2. Results

We will also find that the more refined the mesh is, the more the extreme stress will increase.

2.1) Calculation without fillets (right angle) - successive mesh refinements

Mesh =10x10 mm² to start - each successive view represents a x2 mesh refinement, at the angle where the peak
appears.

Zoom - Von Mises stress in the angle:

Refinement x2:

Refinement x2:
Refinement x2:

Refinement x2:
Let us set an arbitrary elastic limit of the material at 460 MPa: we present below the extent of the zone having a stress
above this elastic limit - about 10 mm long

The curve below gives the evolution of the stress peak according to the mesh thinness.

This curve reflects the fact that at a singularity, the stress peaks will increase singularly (literally)with the mesh
thinness.

If this zone is of no interest with respect to the study's objectives, we admit the presence of a stress peak at the
singularity. It should be noted that if this zone is not of interest, it is not necessary to refine the mesh of this zone.
Nevertheless, it is always disturbing to display iso-stress with a stress peak at 753 MPa.

If the recipients of the calculation note have some experience, they will admit this singularity. Some may ask for an
elasto-plastic calculation, if the value of the stress peak is too excessive (of the order of 2x the elastic limit of the
material).

For information, for an S460 steel, an elastic stress peak at 750 MPa should not produce excessive plastic deformation
(we would remain below 5%, the limit value recommended by Eurocode EN 1993-1-5).

If this area is relevant to the objectives, it is necessary to model the connecting fillet.

2.2) Fillets Implementation r=50 mm

The Wiki indicates that the minimum fillet must have a radius greater than or equal to 2 times the thickness of the part
- here r=50 mm> 2x10 mm - ok.

Note.

The fillet consists of a succession of straight segments.

The mesh at both ends of the fillet shows FE that do not have an optimal aspect ratio, due to the connection
angle that tends towards 0.

Base mesh 10x10 mm²:

Refinement x2:

Refinement x2:

It can be seen that the increase rate of the peak value is lower at each refinement than in the previous paragraph.

Arbitrary filtering at 460 MPa does not show an overflow area:

Let us assume that we have an elastic limit at 440 MPa, filtering at 440 MPa:

(dimensions in mm)

If the material’s elastic limit is 440 MPa, a stress peak at 459 MPa is perfectly acceptable for an SLS analysis (for a
Fatigue analysis this would not be the case but this is another approach).

At this stage, there is no need to further refine the mesh. We can see that the value of the peak stress tends towards a
realistic value.
 3. Results

The curve below gives the evolution of the stress peak as a function of the mesh thinness and according to whether or
not the connecting fillet is taken into account in the modeling.
$translationBooks

PART 2 – USEFUL ELEMENTS

D7. Further information specific to

dynamic calculations
D7. Further information specific to dynamic calculations
D.7.1 Total mass verification

In the case of dynamic studies, one of the fundamental parameters is the mass of the structure which is used to
determine its eigenfrequencies.

It is therefore very important to make sure that the entire mass of the structure is actually entered into the model.
Indeed, in the case of using a model that has already been used for static calculations, it may happen that some
permanent or variable loads, such as equipment, have been entered as loads (point loads, linear loads, surface loads,
etc.) and not as a mass. Therefore, it can happen that the software doesn’t consider these loads as masses but only as
overloads, and does not take them into account in its mass calculation. This may result in a reduction of the seismic
forces.

It is therefore always necessary to make sure that the total mass of the structure is indeed the desired one. This
information is generally available in the results of modal analysis or, even better, can be obtained by performing three
static calculations, by applying a unit acceleration field in the 3 directions (X,Y,Z): only the elements with mass will
therefore be taken into account, and the sum of the reactions will therefore make it possible to know the mass actually
taken into account in the model, in each direction.

D.7.2 Verification of the participating masses

It should be verified that the modal analysis carried out takes into account enough eigenmodes. For this, it must be
verified that the participating modal masses in the studied direction and cumulated for the different calculated modes,
represent at least 90% of the total mass that can be set in motion, calculated from the unit cases of acceleration,
otherwise the standards authorize the taking into account of a pseudo mode (per direction).

Trap: Some software indicates cumulative modal mass % which may be based on a wrong hypothesis of
mobilized total mass: in fact, the parts of masses blocked in movement by supports will not be counted by the
software, which will therefore overestimate the mobilized modal mass %. A trick to overcome this is to define
elastic supports with high stiffness rather than fixed supports: the total mass will then be exact.

In general, it is preferable not to model mass associated with fixed supports.

Example of the study of a skewer model with 5 degrees of freedom:

Three cases are studied:

Case 1: Similar masses and stiffnesses at all levels;

Case 2: Case 1 but with a stiffness 100 times higher in the height of the first floor;
Case 3: Case 1 but with a stiffness 100 times higher in the height of the first 2 floors.
Assuming that all periods correspond to the spectrum plateau (identical spectral value for all periods):

Case 1 Case 2 Case 3

Mode number Base effort Base effort Base effort

Difference if we choose 1 mode

99.5 % V 96.0 % V 83.2 % V
instead of 5

It is therefore important to take into account all significant modes otherwise the calculation efforts could be
significantly underestimated.

Trap: Symmetrical and anti-symmetrical modes.

Example of beam vibration

Depending on the type of calculation being carried out, modes that do not provide a % modal mass supplement in a
given direction should not necessarily be considered irrelevant.

Simple case of the beam on two supports - the masses are mobilized only vertically. The table of modal results shows
that all even modes do not add any additional % modal mass.

Looking at the modal deformations, we realize that these are modes with anti-symmetric deformations:

Modal deformation - mode 1

Modal deformation - mode 2

In the case of a spectral seismic calculation, these modes do not actually add anything new, whereas in the case of a
beam or bridge vibration calculation, these modes have all their interest. It is indeed admitted that pedestrians, in
their movements, can have actions in opposition and in the direction of the modal deformation. A harmonic calculation
is indeed carried out from the loads positioned as below:
We will usefully refer to the SETRA (operating society for transport and automobile repairs)/CEREMA (center for studies
and expertise on risks, environment, mobility and development) guide on pedestrian footbridges for more information.

Spectral analysis: finally, we give below the forces at the middle node of this beam, calculated by a spectral seismic
analysis - it can be seen that even modes do not actually make any contribution.

Trap: Torsional modes

Example of a building in torsion

Generally, common buildings have a torsional mode. On the example below, we can observe:

The 1st mode: with a preponderant mode according to UY (longitudinal),

The 2nd mode: with a preponderant mode according to UX (transverse),

 The 3rd mode: with little modal participation while it significantly affects the structure. It is a torsional mode.

Modes Modal deformation Modal mass following UX (%) Modal mass following UY (%)

1 Following UY 0 77.36

2 Following UX 74.11 0

3 Torsion 0.45 0

D.7.3 Verification of the main eigenfrequencies

In order to verify the effective behavior of the structure in dynamics, it must be ensured that the main
eigenfrequencies of the structure have a coherent rough estimate.

These frequencies correspond to the main modes of bending, torsion, shear, and they generally correspond to modes
with important participation factors and participating masses.

It should nonetheless be noted that limiting to modes with a significant participation factor is not an exhaustive
guarantee of the modes that may be problematic under dynamic loading. Indeed, participation factors can be
calculated by software based on signed modal displacements. Thus, it may happen that the participation factors taking
into account the accumulation of values return low values when the mode is important.

This can happen, for example, for a continuous beam on three identical supports with two identical spans. The main
mode of bending of this beam is the bending of one span in one direction and the bending of the other span in the
other direction (wave form). The participation factor of this mode can be very small, as the displacements of one span
counterbalance the displacements of the other span in the calculation, whereas this mode is the main bending mode of
the structure, and can be the one giving the highest acceleration response if the structure is subjected to periodic
excitation.

When the structure is complex, the dynamics formulas given in the literature do not allow to find precisely the
eigenfrequency values obtained, since these formulas concern simple structures (eigenfrequency of an isostatic beam
on two supports, of a cantilever beam, of a fixed end beam or an oscillator with a few degrees of freedom). However,
these classical formulas allow to estimate the rough estimate of the main eigenfrequencies by estimating in a
simplified way the behavior of the structure to reduce it to simple functioning for verification.

In the case of a beam type structure, we can thus estimate the eigenfrequencies of bending from the classical
formulas. For example, and in a very general way, the eigenfrequency of bending of a beam on two supports with
rotational flexibility will be between the eigenfrequency of bending of the same beam but on hinged supports and the
eigenfrequency of bending of the same beam on fixed supports.

D.7.4 Modal/spectral dynamic seismic calculations

D.7.4.1 Verification of the first modes' relevance (instabilities, displacements)

The first 2 or 3 global modes visualize the functioning of the structure, which allows on the one hand to understand
how it works, and on the other hand to identify modeling problems.

For a well-dimensioned building, the first 2 modes are always according to X and Y, the third mode is a torsional mode.

For common buildings, the fundamental period is of the order of 1/25 to 1/16 of the number of floors.

D.7.4.2 Verification of the global X and Y axes with respect to the first modes

It is necessary to verify that the seismic directions studied X and Y are aligned according to the first important modes.
If not, the complete quadratic combination (CQC) calculations will be wrong.

The solutions are:

in the case of a building, to rotate the model according to the main axes
in the case of a curved bridge, to either calculate the earthquake on a straight model, or make several
calculations by varying the axes according to the skew of the piles.

D.7.4.3 Validation of the number of modes - Complementary mode

Eurocode 8 specifies a minimum percentage of mass to be of interest in the spectral calculation.

If 90% of the mass is not interested, the software allows a complementary mode to be taken into account. It is a
fictitious mode affected by the mass not excited by the studied modes and affected by the spectral acceleration
associated with the last studied mode.

It has been shown that neglecting modes distorts the results (see 1st example in D.7.2).

D.7.4.4 Spectral Calculation and CQC Combinations

The spectral calculation allows to obtain the structural effects (efforts, displacements...) of each mode. Then, according
to the statistical distribution of the earthquake according to the frequencies (defined by the regulatory spectrum), the
effects are combined in order to obtain the statistical response of the structure to an earthquake.

The combination of the different unit modal calculations is done according to the CQC or SRSS mode, the theoretical
definition of which is provided in Part 1 - Chapter 2.

It is important to differentiate between the regulatory spectra used in construction (which are generally dimensioning
spectra) and those used for bridges (which are elastic spectra).

We go from the second to the first by dividing by a behavior coefficient (equal to or higher than 1.5) which takes into
account the plasticity of the structure. The values of the behavior coefficients are defined by Eurocode 8.

In all spectral calculations, it is important to be sure that the damping of the structure is well defined. Depending on
the software, the damping is assigned to modes or materials. In the second case, the mode damping will depend on
the participation of each material in the deformation of the considered mode.

Caution, if we want to attribute a damping in the springs modeling the soil, this corresponds to a study of type soil-
structure dynamic interaction, and it is not possible to use a dimensioning spectrum, only an elastic spectrum.

Finally, as indicated in C16.8, a different behavior coefficient can be assigned to each direction.

After the CQC or SRSS combinations (which combine the modes), the Newmark combinations must be made (seismic
direction combinations), and then the action combinations.

D.7.4.5 Verification of support reactions under elementary cases

First of all, we evaluate the support reactions of the elementary seismic cases EX EY EZ and we compare it to the total
masses.

Verification of support reactions can only be done by signing the modes.

For a building based on a base slab or a strip footing, it is important to limit the detachment of the supports. Indeed,
the elastic calculation shows tensions in the supports that do not exist in reality because the foundations heave.

Negative support reactions (in red) cannot exist because in reality the foundations heave, so the actual stress
distribution on the ground is different from the calculated one (cf. C16.8.3).
 Support reactions of a building under superficial foundations

It is allowed to consider representative "elastic" models if the heave is limited: the limit is taken equal to 30% in the
general case (10% in the case of nuclear buildings).

When detachment is important, much more complex non-linear seismic calculations are required. This verification of
non-heave of the building must be done with care:

it is necessary to give to the modes according to the main modes because CQC support reactions are always
positive.

CQC Support reactions / Signed CQC Support reactions

Seismic combinations (CP + E) make no sense if the seismic E efforts are all positive, while the CP dead loads are
either positive or negative.

It is necessary to study all Newmark combinations separately.

Example of the 24 Newmark combinations for the current frame:

CP +/- 1.0 EX +/- 0.3 EY +/- 0.3 EZ

CP +/- 0.3 EX +/- 1.0 EY +/- 0.3 EZ
CP +/- 0.3 EX +/- 0.3 EY +/- 1.0 EZ

D.7.5 Dynamic calculations other than seismic calculations

This paragraph concerns dynamic calculations other than spectral modal calculations (e.g. vibration of a railway
structure or a pedestrian bridge in order to verify comfort criteria), and corresponds to calculations in which the load
and the structure are calculated with an evolution over time.
D.7.5.1 Verification of free vibration or resonance behavior

In the case of a comfort study of a railway structure or a footbridge, it is advisable to look for loads cases that can
cause the structure to resonate and the consequences of such resonance. For this, it is therefore necessary to apply
periodic loads that can cause these resonances.

As a reminder, the resonance of the structure occurs when the periodic loading is at a frequency identical to one of the
structure’s eigenfrequencies.

To verify that the frequencies of the applied periodic loads are consistent with those of the structure, we can easily
find them on the time graphs obtained a posteriori. This method is applied to the temporal evolution of the
acceleration, velocity or displacement of a representative node (for example the middle of a beam).

To do this, we count the number of periods between two distant points in time and divide this number by the time
separating these two points. This gives a good approximation of the vibration frequency of the node in question:

Structure’s first eigenmode temporal acceleration of a representative node

6.36 Hz/6.25 Hz =1.018; the approximation giving 6.25 Hz gives the right rough estimate
=> the node is excited according to the first eigenmode of the structure.

If the curve shows a very marked periodic variation over the time period in which the loading is applied, this
corresponds to a forced vibration of the structure, and the method described above ensures that the frequency of the
loading is indeed that expected.

If the curve shows this periodic variation over a time period after loading (the calculation was continued after the
loading was stopped), this corresponds to the free vibrations of the structure. The method described above allows, in
this case, to estimate the main eigenfrequency of the structure and thus to make sure that the excited eigenmode is
the right one.
$translationBooks

PART 2 – USEFUL ELEMENTS

Chapter E. How to ensure quality?

Chapter E. How to ensure quality?
Below are some simple advice to assess the quality of the finite element calculations. The principal challenges are:

The proper use of the software

The appropriate modeling of the structural behavior
The traceability of the modeling hypotheses and results

The advice below covers the engineer’s or the team’s handling of the software, the verification controls that must
imperatively conduct any engineer at the end of its modeling, and the tracing of the minimum items so that the work
can be completed thereafter.

E.1 Starting with a new software

E.2 Model validation using self-checking

E.3 Traceability and group work

$translationBooks

PART 2 – USEFUL ELEMENTS

E1. Starting with a new software

We propose below some simple advice to deploy a quality approach in finite element calculations. The main issues are:

The correct use of the software

Appropriate modeling of the structures' behavior
Traceability of modeling assumptions and results.

The advice below deals with the good handling of a software by an engineer or a team, the self-checking tests that
each engineer must imperatively carry out at the end of his or her modeling, and finally the minimal elements to be
traced to allow the work of several engineers or the later resumption of a model.

E1. Starting with a new software

When a new software is acquired in a design office, or when new engineers arrive, there is a very important, and
unfortunately often overlooked step: the appropriation of the software and the validation of the user-software
couple.

This validation, under the responsibility of the management, is at the heart of the quality approach of the studies that
will be produced. We have seen, in the previous paragraphs, all the possible errors linked to a lack of knowledge of the
software's operation, during the modeling process and when the results are used.

Tools exist to help during this validation phase. In particular, we can quote the "Validation guide for structural
calculation software” published by AFNOR (French Standardization Association) 1990 (ISBN 2-12-486611-7). This
guide, initially established for software validation for developers, provides a database of test data and simple modeling
examples, accompanied by the correct results.

It is advisable to choose a few tests from this database, distribute them as an exercise to the team and share the
results, right or wrong, of these tests so that everyone understands how the software works, the options taken by
default and the mistakes to avoid.
$translationBooks

PART 2 – USEFUL ELEMENTS

E2. Model validation using self-

checking
E2. Model validation using self-checking
Before exploiting the model results, several verifications should be carried out on:

input data;
the behavior of the model under elementary stresses and kinematic conditions;
the potential of the model and the software to provide satisfactory and exploitable results.

If some of these points prove to be redundant for the person who carried out the modeling himself or herself (see
previous paragraph), they are essential in the case of the exploitation of a modeling carried out by a third person.

If the final objective of the global study is to perform non-linear calculations (geometrical non-linearities, non-linear
material behavior law...), this validation is essential because it can avoid the unnecessary execution of long
calculations. This is really important because obtaining an acceptable solution for a complex problem to be solved is
rarely immediate.

In general, the realization of a small simplified model using bars allows to quickly validate some rough estimates for
values from an FE model. See the example of the High-rise building tower (skewer model): Example A - High-rise
building (part C).

It is also necessary to verify the model as the model progresses, especially in the case of dead weight load. All too
often, we see engineers starting a model for several weeks and then stumbling over multiple error messages. It is also
possible to save intermediate data files throughout the day, by incrementing the file name, so that you can return to a
previous version very easily (the one that worked before the last updates).

This approach will also make it possible to evaluate the time required to generate the model, calculate and display the
results.

Some important checkpoints are presented below. This list is neither limitative nor exhaustive. For specific problems,
other checkpoints may be considered.

E.2.1 Initial model verifications

Before any calculations are carried out, it is necessary to carry out some basic verifications.

These verifications may seem tedious, but the detection of errors, often basic, can save a lot of time in the end.

a) Model geometry control

Simple graphical controls usually reveal inconsistencies in the geometric features of the elements.

Some software allows you to visualize the elements with their real section. This possibility is particularly interesting for
beam type elements because it allows to control visually the correct orientation of the main axes and the adequate
position of the longitudinal axis which will be used as reference for the loading for example.

The connection of the elements, the correct connection of the shells and the boundary conditions are an integral part
of the geometric verification. The visual inspection of local coordinate systems before applying local loads is also part
of the geometric verification.

b) Element features

For beam type elements, some software can propose predefined profiles. The very first time you use one of those
elements, you should verify that the features displayed by the software correspond to the geometry; compare these
features with those provided by the catalogs in the case of standard sections, or make approximate manual
calculations of the features for non-standard sections.

For a section defined by its contour, the surfaces and inertias calculated by the software must be verified.

For curved problems or with eccentric loads, it is important to verify the position of the torsional center of the beams
and to verify whether or not the software takes into account a possible offset between the center of gravity and the
torsional center.

c) For material features, a proofreading of the data is necessary.

In the case of a mono-material model, an error on the Young's modulus can affect the results of the deformations
without having an impact on the stresses. Whereas in the case of a multi-material model, an error on one of the
modules will influence all the results.

d) Comparison with previous versions

When we have a simplified model from a previous phase, or when we make a modification on a model, we must
systematically recheck certain main measures to detect possible errors.

e) Load features

For loads, a proofreading of the data is also necessary.

It is a question of visualizing all the loads applied to the model and this for each of the defined load cases.

The load values must be correctly identified; characteristic values or pondered values. The orientation and direction of
these loads must also be verified.

If the model contains several load cases, it should be observed whether they are independent or successive load
cases.

For dynamic studies, a verification of the masses of the model in all directions is essential.

E.2.2 Basic verification of calculation results

This step is based on simplified linear static approaches.

For each calculation performed, basic verifications must be made. These verifications, in addition to participating in the
model validation, will also allow a beginner to become familiar with the post-processor of the software used and to
verify that the different options of the elements and/or calculations have been activated correctly.

a) Deformation – Rough estimate of results

The general pattern of the deformation is very explicit because it gives an immediate response on the structure's
behavior to a given load or stress. It allows to validate the hypotheses on the static scheme (simple support, fixed, ...)
and on the modeling of the assemblies.

Beware of graphic amplification factors, which can be misleading on local displacements (impose a factor of 1 to
verify possible inconsistencies).

In linear elasticity, the displacement rough estimates must be satisfactory. Their amplitude must be small in relation to
the dimensions of the structure.

b) Vertical loads calculation

Static equilibrium must be verified. The results of the loads applied in the model can be calculated manually and
compared with the components of the sum of the support reactions displayed by the software.

The distribution and direction of the reactions on the different supports must be analyzed in relation to the blocked
degrees of freedom.

The presence of a null reaction for a blocked degree of freedom must be analyzed. This will generally be a symmetry
effect.

E.2.3 Tests on connections and assemblies

a) Null or non-zero support reactions

The support reactions must correspond to the static diagram.

The sign must also be verified and allows the detection of referential errors for load cases.

b) Modeling an assembly

The general calculation does not eliminate the need for a local (and manual) analysis, for example with a load close to
an assembly to verify that the force transfers are made in the expected way.

E.2.4 Sensitivity tests on specific modeling points

We must question when modeling produces an effect (global or local) that varies a lot as input data changes. This
would be a case of model instability.
$translationBooks

PART 2 – USEFUL ELEMENTS

E3. Traceability and group work

E3. Traceability and group work
While there are projects where a single engineer will do all the calculations, from start to finish and at all phases, most
of the time the work is divided among several people. This requires the implementation of a particular methodology,
described below. And even when the engineer is alone, these elements are part of the global quality approach and
allow to remember the work done.

E.3.1 Transmission / traceability / archiving

For the project success, it is essential that any model can be passed on to another person with the skills required to
operate it without wasting time trying to understand what has been carried out.

To allow the transmission of the FE model, and without redefining the classic notions of quality (files storage in a well-
defined folder, explicit filename in relation to the valid version of the model), it is necessary to trace precisely what has
been modeled (in a modeling note if possible, but failing that in a simplified text that can easily be found), which will
list at least the following data:

the software used;

the type of elements used;
the material properties considered;
the support modeling principle;
the representation by sketch of the main principles of geometric modeling (in particular simplifications made);
the principle for numbering nodes and elements;
applied loads;
the calculations performed;
the combinations made.

If the models are modified as the projects progress, it is necessary to make a point of noting the changes made at
each stage so that the effects of this or that modification on the results obtained can be traced and the intermediate
models used can be saved. These should be stored and identified. We should be wary of names which at time t seem
to mean something and which 6 months later are no longer understandable ("test_support_2_z_flexibility", for
example).

In case of software operating by lines of code, see the following paragraph.

E.3.2 Operating and commented code

In the case of a model imported directly in the interface, the user does not have access to the entire process of
creation of the model.

On the other hand, in the case of FE programs operating by lines of code, it will be advantageous to take advantage of
the options that outline the entire model construction process. This makes it possible to retrieve all the reasoning and
to understand how the software created the different parts of the structure and loads step by step.

It is necessary to make maximum use of the commenting options to explain each line of code or group of lines of code,
to quickly find the specific data you are looking for in the model, and for a person who is not familiar with the language
of the software to at least be able to identify the main modeling principles.

E.3.3 Reflection about BIM

BIM (Building Information Modeling) is currently a new work method applied to the various construction trades. This
approach aims to interconnect the different trades to create a single model from multiple files.

The connection of calculation models to geometric models is beginning to be developed by software companies.

Nevertheless, it is important to be vigilant and not to take for granted all the options presented as automatic. Indeed,
FE modeling is based on an engineer's approach, to model only what is useful and necessary, whereas geometric
modeling aims to provide additional information that is useless for mechanical dimensioning.

The geometric model is created to present 3D plans and views of a structure, as well as the interfaces between the
structure and the equipment, and to detect possible volume conflicts. It is not created to ensure that element
connections are mechanically correct. It does not distinguish between main structural elements and secondary (non-
structural) elements that should not be taken into account. It does not integrate any of the simplifications described
above which are important to understand the functioning of the structure (for example the shifting of the neutral axis
of successive elements or the reprocessing to obtain the nodes).

Thus, the use of FE models derived automatically from geometric models does not necessarily save time compared to
the conventional method, given the need for exhaustive control of the EF model, on the one hand, and the time
required to rework the FE model to make it conform to the desired objectives, on the other hand.

This is true at the time of writing, but companies are making improvements to their products every day, which could
render the previous text obsolete.

The BIM implementation, whose objective is to facilitate exchanges with other professions, should in no way cause us
to lose sight of the fact that FE modeling is another profession, based on the added value of the engineer's viewpoint.
In any case, the use of BIM to obtain a computationally compatible model forces to rethink the traditional modeling
sequence (engineer/projector), to redefine the responsibilities towards the information ... which leads to define specific
processes for the project.

Example of a structure whose calculation model comes directly from the BIM model:

BIM model

FE model from the BIM model

$translationBooks

PART 2 – USEFUL ELEMENTS

Chapter F. How to properly

present the finite element
calculation note?
Chapter F. How to properly present the finite element
calculation note?
This paragraph introduces the fundamental elements that must be present in a note to provide a clear description of a
FE computational model.

F. How to properly present the finite element calculation note?

$translationBooks

PART 2 – USEFUL ELEMENTS

F. How to properly present the

finite element calculation note?
F. How to properly present the finite element calculation note?
This paragraph provides the minimum elements that must be included in a note in order to provide a clear description
of a finite element calculation model.

First of all, the modeling note cannot be the first note of the project. It is either preceded or accompanied by the
general assumptions note. The general assumptions note provides essential information about materials, foundation
stiffnesses, load cases and combinations, and all the justifications that will have to be carried out on the structure.
Ideally, the two notes are written in parallel.

Often, points that would have their rightful place in the hypothesis note or in the modeling note are deferred to a later
note. This way of doing things is harmful insofar as the clarifications arrive afterwards with a very long note containing
the results and sometimes even the justifications ... generating tensions with the Project Manager in charge of the
verifications (or the controller) when the latter will require to complete the work or even redo it and, above all, wasting
time for all the parties involved.

It is fundamental to understand that there is a real leverage effect involved and that it is better to spend some time
detailing and fixing the elements at the beginning and have them validated. This will greatly improve the process.

Let us not forget that in a basic verification mission, the general hypothesis note is the only one that will be subject to
an in-depth examination by the Project Manager.

F.1 Note introduction - Description of the object of the calculation

a) The EF calculation note must begin with a brief reminder of the object under study. Excerpts from blueprints are
always welcome.

b) The study phase should also be mentioned. If the study phase is advanced, it is interesting to remind how the
subject has been treated in the previous phases. Sometimes a simpler model was developed in the previous phase,
sometimes a manual calculation was done. In both cases, the note should compare the results of the simplified model
and the more complete model.

c) It is advisable to specify the calculation objectives, i.e. the justifications that one intends to carry out with this
model: global stability, internal efforts, deformations... The model is not an objective in itself, it is merely a tool to
obtain a result.

d) It is not mandatory to use a single model for all project justifications. Specify what will not be covered by the current
model but by another sub-model.

e) The modeling note must declare all documentary references used: blueprints with their index, market parts,
calculation or geotechnical notes.

f) In case of a model update, the changes made must be explicitly traced.

g) The note must describe the principle of the results exploitation, the direct software outputs and the possible post-
processing that are contemplated.

F.2 Geometrical description of the model

a) The physical boundaries of the study should be very clearly defined: which elements are modeled and which are
not. Some secondary structures often do not need to be modeled (stairs or walkways, equipment). Some main
structures can be simplified such as bridge abutments or piles, which can be represented by supports with their
flexibility. In case of a succession of structures, the modeling boundaries must be described with the way to take into
account the interaction with adjacent non-modeled structures.

b) Hypotheses for geometric simplification, choice of nodes and sections must be set out in detail. The provision of a
complete listing of the calculation file cannot satisfy this request. Drawings are required. Hand sketches, which the
engineer uses when coding, can advantageously be provided. They help understanding the modeling logic.

c) If there is an exchange between the drawing software and the finite elements software, it is interesting to indicate
it.

d) All the units used must be explained: distances, forces, stresses and masses. By default, the SI system is preferred.

e) Define the global coordinate system for the model and recall the efforts sign convention for all support reactions.

f) The same applies to the finite elements: the local coordinate system and the sign conventions adopted by the
software for stresses and strains should be indicated.

g) Images and graphic outputs are interesting to visualize the model, but they should be accompanied by the
corresponding descriptions. A modeling note consisting of a series of screenshots is not acceptable.
F.3 Finite elements description

a) As discussed in Chapter B, the software choice depends on many criteria. It is necessary to explain, even briefly,
why the software used is appropriate. If it has computational limitations, do not hesitate to write it down and explain
how these limitations will be overcome.

b) Description of FE properties: this part is often missing in the description, yet some software has a wide variety of
element types that do not have the same functionality. In particular, for plate models, the elements take or do not take
into account membrane effects, which may change the results; this is also the case in 1D for shear strain in beams.

c) Describe the number of nodes, the size of the elements, the mesh type. If a mesh refinement test has been
performed (as recommended in paragraph d), report it.

d) For a bar model, a table of the mechanical properties of the bars must be provided.

e) The link between the global coordinate system and the local coordinate system should be illustrated with
screenshots. Most software programs have quite explicit ways of displaying the coordinate systems. Note that it is
often possible and useful to force the coordinate systems in order to facilitate the results analysis.

f) It is interesting to give the number of elements as well as the number of degrees of freedom of the model.

F.4 Mechanical description of the model

a) The properties of the materials must be fully explained: density, Young's modulus, Poisson's ratio, static behavior
law, shrinkage, creep or relaxation laws...

b) Boundary conditions must be correctly described. The static diagram must be recalled, along with how the supports
are modeled. If a stiffness matrix is introduced, explain how it has been calibrated from geotechnical parameters.

c) If supporting devices are present, specify whether they are modeled by bars with special characteristics or by
special connections in the model.

d) Non-structural elements (equipment) must be listed and it must be specified which ones are modeled or,
conversely, taken into account as loads. As indicated in c), this choice depends on the stiffness and mass of these non-
structural elements.

e) The way in which efforts and loads are introduced is not evident. It must be specified whether the software does
automatic load placement, incremental load positioning or whether load cases are entered manually.

f) In the case of a force to be introduced on a cut, it is necessary to illustrate how this force diffuses in the model
(spider connecting the edges of the cut).

g) In general, and in particular for seismic or dynamic calculations, it is necessary to detail how the masses are
introduced into the model. If the mass of the elements is generated automatically by the software, in the assembly
areas, there are volumes counted twice; it is therefore necessary to be able to correct the densities. Non-structural
elements modeled by load cases are not recognized as masses and have to be added. A manual verification of the
global mass of the model is always useful and reassuring.

F.5 Demonstration of the self-monitoring approach

Self-monitoring is a fundamental element of the modeling quality.

This self-monitoring process must be visible to the person who is going to verify the calculation note.

a) Tests and elements of verification of the mesh validity must be mentioned.

b) All the model validation tests that have been conducted contribute to win the controller's trust. It is not a question
of providing a large amount of information and data, but just indicating the tests that have been conducted.

c) On the other hand, the verification tests for the vertical load calculations are absolutely essential. They must at least
include the structural self weight, the self weight of the equipment, a uniformly distributed load and cases of thermal
loads.

d) This also includes global mass verification for dynamic and seismic models.

F.6 Description of effort recovery and post-processing

a) It is necessary to describe in which form the efforts or displacements of the model are recovered (listings, graph and
chart reading, screen display).

b) In cases where the results are expressed in a local coordinate system and a change of coordinate system is
necessary afterward, the risk of error is frequent. The validity of the baseline change must be demonstrated.

c) In general, the process of post-processing the efforts, with the associated tests, should be described.

d) For combinations and envelopes, it is necessary to say whether they are made by post-processing or directly by the
software. In the second case, it is necessary to indicate if the combinations are formed manually or if they are
automatic (source of error). In any case, it must be specified whether the combinations and envelopes generate
concomitant forces or not.

F.7 Results report

The results are often presented in the form of tables, sometimes cumbersome to understand.
a) As said before, the system of units must be defined and the units must be systematically indicated in the table
columns.

b) It is necessary to recall the vertical loads calculation for elementary load cases.

c) The dimensioning values in the tables must be highlighted (highlighted, circled or put in bold or red, etc.).

d) Result listings should not be in the body of the text. They make the document more difficult to understand and lead
to unnecessary printing. They will be placed in an appendix.

F.8 Specific complements for volume elements

a) The choice of cross-sections must be consistent with the expected results and must be consistent with the critical
plans of the structure.

b) In the same way as for surface elements, the software offers a wide variety of volume elements, with different
codes. Some elements are very specific to certain materials and certain types of calculation. It is necessary to refer to
the software's manual to choose the "simplest" element, unless there is a very particular need.

c) It is advisable to give priority to results in the form of steel mapping highlighting the dimensioning points and
specifying whether the values are smoothed or at nodes, for example.

d) In the case of forces integration on a cross-section, it is useful to explain the method chosen.

F.9 Specific complements for non-linear calculations

a) It is necessary to provide the behavior law used, which may be different from the standard law of the software.

b) It is interesting to present, in the calculation note, the evolution of a remarkable magnitude (displacement of a
point, specific effort, etc.) during the increase of the load, to visualize the plasticity.

F.10 Specific complements for dynamic calculations

a) If a spectrum automatically provided by the software is used, it must be demonstrated that it has been verified to
be consistent with the expected spectrum.

b) It is necessary to define the selected damping (which is not the same in dynamics and seismic) and/or the behavior
coefficients for seismic calculations.

c) The participating masses and mode participation coefficients must be given mode by mode, and modal
combinations and modal sign conventions must be specified, if applicable.

d) Modal deformations should be presented for the most representative modes. The modes shape is an important
element in verifying the global structural behavior.

e) For a calculation by time steps, as for the non-linear calculation, it is interesting to present the temporal evolution of
the representative quantities (displacements, accelerations, etc.).
PART 3 – EXAMPLES AND
COMPLETE CASE STUDIES
$translationBooks

PART 3 – EXAMPLES AND COMPLETE CASE STUDIES

EXAMPLES AND COMPLETE CASE

STUDIES
This part contains modeling examples for simple and more complex objects. They are presented as a complete or
partial study or even a comparison of different models of the same structure.

If you happen to have an example that you wish to share with us of a complexity or paradox noticed on a part of a
model, please send it to the following address: [email protected]. (It could be a note about a real project
rendered anonymously.)

Example A – Modelling a complex high-rise building

Example A – Modelling a complex high-rise building

Example B – Modelling of composite bridges

Example B – Modelling of composite and steel bridges

Example C – Modelling of beam grids

Example C – Modelling of beam grids

Example D – Simple example: modeling of a Br wheel

Example D – Modelling of a Br wheel

Example E – Transverse bending of a prestressed concrete box girder

Example E – Transverse bending of a prestressed concrete box girder

Example F – Dynamic calculations of tanks

Example F – Dynamic calculations of tanks

Example G – Cable-stayed bridges

Example G – Cable-stayed bridges

$translationBooks

PART 3 – EXAMPLES AND COMPLETE CASE STUDIES

Example A - Modeling a complex

high-rise building
Example A - Modeling a complex high-rise building
Franck DUBOIS - Structures Engineering
Thierry RICHARD - Structures Engineering

Part A: Presentation of the example

Purpose of the example

This example focuses on the calculation of the general forces of a high-rise building using finite element global
modeling.

This calculation takes place at the beginning of the execution studies, its objective is firstly to determine the
reinforcement of the foundations (diaphragm walls and drilled shafts) of the base in order to start the drilling of the
diaphragm walls and drilled shafts.

Then, in a second step, the modeling will be used to calculate the general forces in the walls and floors necessary to
calculate the reinforcement of the different elements (walls, columns, floors).

This example studies in particular the sensitive points of the modeling of this building:

soil-structure interaction;
seismic calculations;
non-linear calculations;
as well as phased calculations.

Particular attention is paid to the management of the modeling, detailing its preparation, its integration in the general
study, its organization and its validation.

Presentation of the project

This building is located in Monaco on a major urban site, in the middle of sloped soils. It is made up of 2 major areas:

a base buried in its rear area on 13 levels consisting of rear parking lots and dwellings including a school on
the sea side;
a 25-storey tower sitting on part of the base.
Figure 2: general view of the model (3D view).

Part B: Preparation and organization of the modeling

This is a very important step, because it will be very difficult to modify the model later on when it is advanced, so it is
essential to have clarified all the characteristics of the calculation model before starting it.

A.1 Particularities and constraints of the project

They allow to identify the sensitive points of the modeling:

The large excavation area, the school area and the tower form a single monolithic block, there are no
expansion joints.
The excavation is integrated in sloped terrains, its dimensions are significant (109m x 169m on 13 levels), its
support consists of a peripheral diaphragm wall reinforced by buttresses and inner drilled shafts.
It is moreover asymmetrical with strong active earth pressure on the upstream zone and no passive earth
pressure on the downstream zone.
Diaphragm wall panels rest on buttresses (no embedding)
The project is on an urban site in the middle of existing structures which are very close to the site
Presence of the high-rise building on part of the base
Taking into account a second tower (possibly built in the future) for the dimensioning of the base
Very tight work schedule

A.2 PRO file studies

The study of the calculation notes of the PRO file allows to quantify a rough estimate of the forces, and to identify the
sensitive points.
In our case, it clearly appears that the foundations are dimensioned by the earthquake cumulated to the earth
pressure, in fact the foundations’ notes show very important shear rates in the diaphragm wall (5 Mpa) with
reinforcement rates exceeding the usual uses.

A resizing of the foundations will certainly be necessary.

A.3 Main modeling features

The first findings for the modeling are:

The size of the model will be very important because there is no expansion joint.
Taking into account the active earth pressure as well as seismic calculations are the key elements of the
modeling.
The precision on deformations required due to the proximity of existing structures cannot be obtained
through conventional structural modeling.

It is therefore necessary to carry out simultaneously a "3D Geotechnical” finite element modeling which will be used
solely to calculate the deformations. It will take into account on the one hand all the soils around the project and on
the other hand the structure of the project itself.

The active earth pressure applied to modeling will be derived from the "3D Geotechnical” model.

Asymmetric active earth pressures and their high intensities require consideration of passive earth pressure
zones, and modeling of horizontal wall friction on the ground where possible.
The up and down excavation construction method without the use of tie rods, associated with active earth
pressure asymmetry, has very important consequences on modeling. Indeed, during construction, the active
earth pressures are taken up on the one hand by the peripheral diaphragm wall (which is classic), but also by
the inner drilled shafts, which make the classical 2D calculations obsolete. It is therefore necessary to
integrate the construction phases in the determination of the forces.

A.4 Input data:

It is necessary to list and validate all the input data that form the basis of the calculation:

The PRO architectural file

The PRO structural work file carried out by the project management (plans and calculation notes)
The overview of the reservations which appear in the PRO files and which unfortunately will evolve
throughout the studies
The "2D" active earth pressure calculated by BET FONDATION in a first step, which will be used to develop the
modeling
The "3D" active earth pressure that results from the 3D geotechnical modeling that takes place in parallel
with the structural modeling. They will make it possible to correct the classic "2D" active earth pressure
caused by the geometry of both the soils and structure (vault effects in particular)
Seismic active earth pressures that will be added to inertial seismic effects
Ground stiffness and friction in statics
Ground stiffness and friction in dynamics
Construction phases
Construction site methods

A.5 Output Data

They are quite simple as they are relative to the definition of the reinforcement of the diaphragm walls and drilled
shafts, which are to be transmitted to BET Fondation so that it can carry out the reinforcement plans.

A.6 Interfaces between participants

The complexity of the input data and the ties between participants requires an interface flowchart to ensure
consistency between the modeling and all other stakeholders. (see next page).

A.7 Assumptions Note and Modeling Methodology Note

The Assumptions Note describing the assumptions taken into account in the calculations (materials, definition of loads,
etc.) must be accompanied by a Modeling Methodology Note describing how these assumptions are taken into account
in the modeling (application of active earth pressure, calculation method, etc.).

The Methodology Note sets out the principles implemented in the modeling and allows all project participants to
validate them before modeling them. This avoids the long and laborious modifications that often occur when the model
is finished. To simplify, this is equivalent to writing the modeling note before it is completed.
Part C: Modeling
A.1 Geometric modeling principle

A.1.1 Breakdown into three independent parts

The modeling has been broken down into three independent parts, assembled during the final phase:
 Partial modeling

Complete model

A.1.2 Coordinate system

The general reference frame is taken in the direction of the rear wall, which corresponds to the axes of the tower and
overall to the directions of the active earth pressures, in order to make the seismic calculation coherent as it will be
carried out according to the main axes of the tower.

Global reference frame

Tower axes

A.1.3 Using the BIM model

It could have been interesting to use the PRO file including a BIM model to generate the geometry of the model, but it
soon appeared it was much simpler and more rigorous to study each level separately from the AutoCAD files to define
the average lines of the panels.

This helps understanding how the structure works and ensures the proper transfer of loads between levels or between
the elements themselves.

This manual step is important because it simplifies the modeling by removing local details that have no effect on the
overall distribution of the forces, which will then be the subject of local studies. It mainly deals with the alignment of
walls, slabs, the removal of small reservations, the removal of secondary walls, etc.

A.2 Modeling of the rear zone of the "large excavation" base

A.2.1 Model features

Model view of the rear area of the base: large excavation

It is the most delicate zone because it is subjected to earth pressure, and it is done from the top to the bottom with the
following phasing:

Diaphragm walls and drilled shafts

Upper slab (tower support)
First level earthworks levelling
1st floor concreting
Lower levels according to the same principle

The active earth pressures asymmetry between upstream and downstream requires a partial active earth pressures
resumption by the central drilled shafts during the levelling, such resumption not being calculable by conventional 2D
methods – hence the need to proceed to a 3D phased calculation.

The main features of this modeling are:

Modeling of the diaphragm wall in shell elements with the installation of springs at the junction between
panels
Modeling of the buttresses into shell elements, with springs modeling the support of the diaphragm wall on
these buttresses
Taking into account passive earth pressure zones in areas that do not affect the surrounding areas
Taking lateral earth pressure into account in areas that don’t affect the surrounding areas
Making 4 seismic calculations (+X -X +Y +Y directions)
Phased calculation of earthworks phases

View of the large excavation without the upper slabs

A.2.2 Soil modeling by springs

The connection of the diaphragm wall to the soil depends on the direction of the force:

Either the wall presses on the soil, the spring is then in passive earth pressure, and its stiffness is in passive
earth pressure
Either the wall pulls on the soil, there is then a detachment, the stiffness of the spring is thus null (the wall is
in fact in an active earth pressure state)

Active and passive earth pressures of walls framing a floor

There are therefore active and passive earth pressure zones for each load case.

We have considered 2 methods:

 Either define manually by iterations the passive earth pressure zones for each load combination.
Or use non-linear springs working only under passive earth pressure.

This non-linear second method was adopted, as the calculation times proved to be acceptable.

Two laws of physics are used for soil modeling springs:

Springs in passive earth pressure. Friction springs (passive earth pressure at the bottom of the excavation).

The friction springs are horizontal: there is no friction in the vertical direction.

These springs are relative to frontal stiffnesses and friction stiffnesses, and they vary according to the soil layers.

Example of soil spring location

A.2.3 Connections between panels of molded walls

The connection between diaphragm wall panels and drilled shafts or between diaphragm wall and buttress is modeled
by a gap between the elements: 0.20m between panels and 0.5m between wall and buttress. The 2 panels are then
connected by a non-linear spring connection with the following behavior law:

Connection between panels of the diaphragm wall

A friction coefficient of 0.7 is added to model friction in the panel contact plane with a limit of 0.7 x normal stress to
the surface.

A.2.4 Connections between the buttresses and the diaphragm wall panels
They are carried out according to the same principle than for the connection between diaphragm wall panels.

Connection between the buttresses and diaphragm wall panels

A.2.5 Loading

A.2.5.1 Weight loads

They are simple to describe: dead weight, additional permanent loads, and overloads.

A.2.5.2 Static active earth force

They are derived from 3D geotechnical modeling

A.3 Modeling of the front area of the base (school and dwelling)

This modeling is much more conventional and traditional.

It is made up of shell elements (floors and sails) and wire elements (posts and beams).

The foundations are modeled entirely as for the rear area.

This area is only subjected to weight loads.

 General view of the school zone model

It should be noted that the lengths of the lower levels exceed 100 ml, thus exceeding the regulatory lengths for taking
into account the shrinkage effect.

A calculation of this zone under the shrinkage effect will have to be carried out during the second phase of the study,
in order to determine the longitudinal reinforcement of the walls and floors.

For homogeneity with the large excavation, the piles of the foundations are modeled entirely with the installation of
horizontal and vertical springs according to the soil layers.

A.4 Modeling of the tower

The modeling of the tower is conventional and traditional, consisting of shell elements for the sails and floors and wire
elements for the beams and posts.

The main problem is the large size of the number of elements due to the large number of levels.

In the first step of this foundation calculation, the tower only interacts by its effects at its base (upper base level). The
size of the meshes does not interact much, it will thus be large meshes.

View of a typical level of the tower

One of the peculiarities of the tower is the design of the slab edge which is very complex and variable at each level.
There are no two identical slab edges.

The layout of the slab edge had to be worked on in order to remove many points from the architect's original DWG file,
which had very small distances between two points – even down to the millimeter.

Meshing of the slab before purge the useless points on the slab edge

This example shows the problems encountered during the automatic recovery of DWG or REVIT files.

Although it does not initially study the effects of the wind, the modeling includes at each floor the definition of a
"WIND" node positioned at the center of gravity of the floor and connected to the core walls by rigid links.

These nodes will then be used to introduce the wind load torsors calculated in the wind report for each level of the
tower.

A.5 Modeling of the optional second tower

This second tower is identical to the first one and may be built in the future.

Whereas for the first tower the need to model it entirely by shell elements quickly appeared, for this optional second
tower, we studied solutions limiting the model size.

A.5.1 Solution 1: modeling of the 2nd tower by its forces torsor at its base

This is the simplest solution which consists in calculating the resulting torsors at the base (O2 point below) for each
load case and applying them directly to the general model.
The resulting seismic torsors of the 2nd tower are then calculated on a model of the tower alone which is embedded at
its base.

This method could not be implemented because it directly adds up the seismic forces of the two towers, thus
generating inadmissible forces in the diaphragm wall. Indeed, it appeared during the PRO file that there was no seismic
interference between the 2 towers.

A.5.2 Solution 2: modeling of the 2nd tower by a skewer model

We realized a skewer model of the isolated tower.

This model consists of a vertical bar modeling the core whose characteristics are defined from test cases of the tower
model by arranging horizontal loads at the head (according to X and Y), then by studying the equivalent inertias
producing the same deformations.

Given the asymmetries of the floors, it soon became clear that the conventional skewer modeling consisting in
assigning masses at each level is not suitable, because torsion is then not taken into account.

Each floor has been cut into 4 parts to which the corresponding masses are assigned.

Visualization of the skewer model

We then compared and validated the resulting torsors between the 2 calculation models under the CQC seismic cases,
and compared the main modes.

Mode 1: transverse according to X:

f= 0.68 Hz with 56% of the mass f=0.63 Hz with 62 % of the mass

Mode 2: Longitudinal Y:
f= 0.77 Hz with 66% of the mass f=0.70 Hz with 62 % of the mass

Torsion mode:

f=2.2 Hz f=1.3 Hz

Vertical seismic mode:

f= 5.0 Hz with 73% of the mass f=5.5 Hz with 82 % of the mass

The validated skewer model was implemented in the general modeling by having rigid joints between the base of the
skewer, and the diaphragm walls and bearing drilled shafts of the large excavation.

Compared to the first solution, a decrease in the forces in the diaphragm walls appeared. On the other hand, the rigid
connection has generated very significant not admissible forces in the diaphragm wall, which do not appear in the
connection of the first tower.

Therefore, the first level of the second tower should be modeled in shells to obtain consistent results.

This solution has therefore been abandoned in favor of the third solution.

A.5.3 Solution 3: modeling of the 2nd tower entirely by shells

It is the solution that we were trying to avoid that was used!!, the calculation times have increased while being
reasonable.
 Rear view of the model showing the anchoring of the 2nd tower in the diaphragm wall

Part D: Global modeling calculations

A.1 Global model features

The global model is meshed with elements of 1.50m size, except for the foundations (diaphragm wall, drilled shafts)
with smaller meshes of 1m.

In the end, the model includes 168,000 nodes.

The total calculation with phasing and seismic calculations takes one night.

A.2 Phasing calculation

The calculation takes into account the 18 construction phases (earthworks) of the large excavation.

Then the school zone and the tower are activated.

Operating overloads can thus be applied to all floors of the model.

A.3 Seismic calculation

A.3.1 Modal spectral calculation

4 seismic calculations are performed in the directions +X, -X, +Y, and -Y, neutralizing the linear springs in tension in
each case. These cases are then studied twice; with or without the 2nd tower. There are therefore 8 modal spectral
analyses.

Modal analyses are carried out on 100 modes, which allows to interest at least 70% of the participative mass. The
residual mode is then added to reach 100% of the mass.

The 2 calculations with or without the second tower are quite close, the 1st mode is 0.471Hz with 2 towers, and
0.582Hz with only one tower.

First modes visualization (with 2 towers)

A.3.2 Seismic active earth pressure

Dynamic increments are applied in the 3 directions +X, -X, and -Y.

They are added to the static active earth pressure.

Cumulative dynamic increments with static active earth pressure

The seismic active earth pressures are then added to the inertial seismic forces from modal/spectral studies.

A.4 A few results

Cumulative phasing deformations

Deformations of dynamic increments

Seismic CQC deformations +X direction

A.5 Iron framework of the foundations: diaphragm walls, drilled shafts and buttresses

The global forces are calculated for the SLS and ULS limit states, and ULS seismic.

Cuts are made over the entire height of each panel to deduce the global resulting forces.

We visualize below the graphs of the forces in a buttress in the SLS state.

Normal stress in a buttress

Bending moment on high inertia

Bending moment on low inertia

The iron framework is then calculated by applying the usual and regulatory rules related to reinforced concrete.

Part E: Modeling validation

This is the fundamental question of complex modeling: how can we show the validity of the results?

Several types of validation were carried out.

A.1 Comparison with PRO file studies

The main results were compared with those in the PRO file:

Mass balance
Torsor of support reactions for elementary load cases
Deformations under permanent loads and earthquakes
Specific modes
Etc.

A.2 Internal validations during modeling

They are carried out during the modeling and during the verification of the main results

Moreover, they relate to:

Geometry visualization, local frames, thicknesses, etc.

Graphical visualization of the loads applied to the model,
Visualization of steel mapping to show any problematic zones identified by large steel sections
Visualization of the deformations of elementary load cases

A.3 Internal validations by a partner not involved in the modeling team

A person external to the modeling team checked several points:

Activation and deactivation of groups of elements (surface, wire, or spring elements) in the calculation phase
Behavioral laws according to input data
Behavior law assignment for all passive earth pressure or friction springs, depending on panel location and
level.
Orientation of non-linear springs which only work in compression
Respect of the springs’ plastic zone during the different phases
Loading of elements

A.4 Modeling validation meeting

Outside the modeling team, it is absolutely impossible for other project participants (project managers, technical
controllers, construction site, other technical design offices, etc.) to understand the details of this "black box" and to
be able to validate the results of the model.

A general meeting was therefore scheduled "live" in front of the modeling computer. Everyone was then able to
request data visualizations, request additional results, understand the model structure, see all the parameters included
in the calculation data, access intermediate results, etc.

The aim of such a meeting is to answer all the questions raised by the participants directly with the modeling
computer.

Part F: Calculations of the structure’s iron framework

The modeling will be completed in a second phase by the calculation of wind and shrinkage forces.

The mesh will be refined for the tower’s walls by adopting 3 meshes on the height (i.e. a 1m mesh).

The walls will then be calculated directly from the results of the modeling by making cuts at their base.

The bending forces of the floors will be calculated "manually", i.e. with local modeling, which will be added to the
membrane forces (N and FXY) determined by the modeling in order to determine their iron frameworks.
FXY membrane forces to be taken into account in the slab calculation
$translationBooks

PART 3 – EXAMPLES AND COMPLETE CASE STUDIES

Example B - Mixed and Steel

Girders
Example B. Mixed and Steel Girders

The subgroup responsible for these examples is still working on the subject. Some patience will be required.

Mixed Structure

This subgroup has studied the same structure in three different ways:

A 2D wireframe approach (link) - Valentina Bruno (Setec tpi).

A 2D beam grillage model - Hugues Somja (INSA de Rennes).
A 3D model - Aymeric Perret du Cray (Setec tpi).

Orthotropic Deck Steel Structure

Setec pdf : 3 models;

Arcadis: 1 models.
$translationBooks

PART 3 – EXAMPLES AND COMPLETE CASE STUDIES

Example C - Beam Grillage

Modeling
Example C - Beam Grillage Modeling

Benjamin Tritschler (Arcadis) - Transverse Bending of a Multi-Beam Bridge

Link: https://cloud.afgc.asso.fr/s/9NK3MYEiYpGwsQ7

Jacques Combescure (Artes) - Beam Grillage

Lien: https://cloud.afgc.asso.fr/s/DcWCg2KWRLj2FMm

Pierre Perrin (Dir Est) - Contribution to the Study of Beam Grilles

Lien: https://cloud.afgc.asso.fr/s/yP33KYxSFjJPqmW

Didier Guth (diadès) - Modeling the Same Structure Using Various Approaches

Lien : https://cloud.afgc.asso.fr/s/KLJsLyZiC8pmmqY
Fichiers ST1 : https://wiki-gtef.frama.wiki/_media/fichiers_st1.zip
$translationBooks

PART 3 – EXAMPLES AND COMPLETE CASE STUDIES

Example D - Simple example:

modeling of a Br wheel
Example D - Simple example: modeling of a Br wheel
The purpose of this example is to show the differences that can be obtained with several different software for the
same calculation, which is very simple a priori, and to confirm the propositions of chapter D.5.2.

Authors: Franck Dubois, Valentina Bruno and Didier Guth.

Link: https://cloud.afgc.asso.fr/s/58c5AX359ePN92K
$translationBooks

PART 3 – EXAMPLES AND COMPLETE CASE STUDIES

Example E - Transverse bending of

a prestressed concrete box girder
Example E - Transverse bending of a prestressed concrete box
girder

Author : Jean-Paul Deveaud - Cerema est center

Link to the file: https://cloud.afgc.asso.fr/s/apaRyaTMbJpNtJg

Link to the file: https://cloud.afgc.asso.fr/s/XxmzPsxQjkpDTE7

$translationBooks

PART 3 – EXAMPLES AND COMPLETE CASE STUDIES

Example F – Dynamic calculations

of tanks
Example F – Dynamic calculations of tanks
Coming soon... a little patience required.

Author: Gildas Potin & Co.

$translationBooks

PART 3 – EXAMPLES AND COMPLETE CASE STUDIES

Example G – Cable-stayed bridges

Example G – Cable-stayed bridges
Coming soon... a little patience required.
$translationBooks

PART 3 – EXAMPLES AND COMPLETE CASE STUDIES

Beam Grid Example

Beam Grid Example
1. Beam grid
The beam grid is a generally horizontal structure consisting of longitudinal beams, called "main beams", associated
with transverse beams. Beam grid modeling allows the calculation of a large number of structures in civil engineering,
it is mainly used for the following works:

the bridge deck with prefabricated beams made of prestressed concrete with bonded wires, or reinforced
concrete girders;
the mixed-frame deck when it has more than two beams;
the ribbed slab deck;
by extrapolation, the slab type deck.

This modeling has the advantage of imposing the main directions of the reduction elements. Indeed, the plate
calculation produces principal directions of moments that often deviate from the directions of the reinforcement, and
the complex system of deviatoric moments or mxy torsional moments must be taken into account in the strength
calculation.

Whereas, for a wireframe model, when the directions of the reinforcement are fixed, according to the geometry of the
structure and the recommendations, we will be able to deduce the directions of the beams system in our model. The
beam grid remains a simple model to model and above all simple to use. Consequently, we are often led to model a
concrete slab (bridge one-way joist slab) by cross girders (transverse beams).

1.1 The GUYON-MASSONNET method

In the 1950s, the construction of multi-beam bridges required calculations tools. The method developed by Yves Guyon
[Calculation of Wide Bridges with Multiple Girders Joints with Cross Girders], then improved by Charles Massonnet
[Contribution and Complements to the Calculation of Multi Girder Bridges] is the most widely used, it has been
compared to model tests or measurements on real bridges and has always been found to be in good or very good
agreement with experimental measurements. The method uses two fundamental parameters whose data can be used
in a FEM.

The data:

1.- the flexural stiffnesses of the beams and cross girders:

BP=E.IP

BE=E.IE

2.- the torsional stiffnesses of the beams and cross girders:

CP=E.KP

CE=E.KE

The method assumes that all the beams are identical. However, in order to make building easier, the edge beams can
be different!
The deck jacking operations require the construction of an end cross girder:

Moreover, this is an isostatic bridge, and even if the method was extended to hyperstatic continuous bridges by the
rule of thumb which consists in calculating a fictitious span with the same elastic deflection under a concentrated load
applied in the middle of the span, what about the stresses on supports? Can the moments on the piles be calculated
using a method based on the deformation calculation?

Modern calculation devices have favored geometrical and architectural criteria in construction and have taken many
works out of the method's application field. The FEMs seem to abolish the perimeter of calculation of the structures,
and the difficulty often lies in the choice of the data and the model and the exploitation of the results.

Note. - In reinforced concrete it is preferable to use reducing elements rather than stress tensors.

1.2 Purpose of this chapter

The simplicity of exploiting the results makes the beam grid a widely used calculation model. In this chapter, we will
discuss data elaboration and the transformation of the results for the effects of the skew and the implementation of
hyperstatic spans. These computational difficulties are discussed through the following examples:

1. Straight isostatic bridge deck with a 100 degree skew.

2. Bridge deck with 2 hyperstatic straight spans with a 100 degree skew.
3. Straight isostatic bridge deck with a 70 degree skew.
4. Straight isostatic bridge deck with a 50 degree skew.
5. Straight 2 span hyperstatic bridge deck with a 50 degree skew.

1.3 Beam characteristics

As for many statically indeterminate structures, the results are subject to the relative stiffness of each beam grid. The
load distribution on the main beams is favored when the cross girders are stiff in relation to the main beams.

1.3.1 Flexural beam stiffness

The main beams are usually surmounted by a continuous one-way joist slab and the beam grid is made up of a single
plan, therefore the sections of each structural element should be well defined.

The geometry of the compression flanges is defined in EN 1992-1-1 art. 5.3.2.1.

with:

1.3.2 Torsional beam stiffness

Due to the lack of discontinuity between the beams, we cannot strictly speak of torsion, it is rather a plate bending.
However, in order to remain consistent with the beam logic, a torsional inertia is determined for each element. By
considering the beam free to rotate around its center of torsion, the stiffness is underestimated. Therefore, we will use
for rectangular elements a shape coefficient K = 1/3, whatever the slenderness of the part is: J=ab33(*).

Note. - K is a shape coefficient studied by Mr. Caquot. He indicates that K can be evaluated with a very good
approximation by the formula 1k=1+1m23.560-0.56m-1m+12 where m is the ratio ab with a≥b… according to
“Reinforced Concrete Form” by R. Chambaud et P. Lebelle.

(*) Important: The SETRA (operating society for transport and automobile repairs)/CEREMA (center for studies and
expertise on risks, environment, mobility and development) guides, the PRP 75, the guide relating to independent-
span and post-tensioned prestressed beam viaducts and the guide relating to viaducts with prefabricated beams made
of prestressed concrete with bonded wires all offer a more sophisticated approach - it will be necessary to refer to
them in the context of a real project.

1.4 Model geometry

1.4.1 Model characteristics

The model characteristics are:

Span: 15.0m;
Width: 9.4m;
9 main T-beams with a rib of 0.80x0.40m;
A 25cm thick one-way joist slab.

Cross section of the deck

Longitudinal section of the isostatic deck

1.4.2 Geometrical characteristics of the sections

The characteristics of the sections are:

Common main beams:

Section:

Flexural inertia:

Torsional inertia:

Common cross girders:

The model is made with a cross girder every 150cm:

Section:

Flexural inertia:

Torsional inertia:

The FE model presented below is made with ROBOT v2019 -Autodesk:

1.4.3 Note on torsional inertia

By default, the software calculates the torsional inertia. The table below compares the different calculations:

J[m4] Software J=∑ab3/3 J=∑ J_Caquot

Beam 0.0191936 0.0261760 0.0199764

Cross girder 0.0069919 0.0078125 0.0069646

In order to be able to compare the finite element method to the Guyon-Massonnet method we will use the torsional
inertias from the software.

1.5 Loading of the model

Each beam of the different geometries is loaded with:

A mid-span point load of 100kN:

A 10kN/m distributed load along each beam:

1.6 Model results

1.6.1 Comparison of simple geometry with the Guyon-Massonnet method

The mid-span moments per beam for a 100-degree isostatic deck are (for e=y):

Guyon-Massonnet method FEM model

 Beams y (m)
Point load Distributed load Point load Distributed load
K
Miso=P∙l4 Miso=q∙l28

Edge 0.405 150 kN.m 112 kN.m 169 kN.m 99 kN.m

-3.375 0.244 91 kN.m 68 kN.m 121 kN.m 65 kN.m

-2.250 0.188 70 kN.m 52 kN.m 107 kN.m 52 kN.m

-1.125 0.170 63 kN.m 47 kN.m 102 kN.m 48 kN.m

Center 0.166 62 kN.m 46 kN.m 101 kN.m 47 kN.m

It should be noted that the application of the Guyon-Massonnet method results in greater stresses on the edges than
the FE model.

This can be explained by the fact that the torsional inertia of the beams joined by the one-way joist slab is
underestimated. Indeed, if the torsional inertia of the main beams is higher, the strain of the edge beam will be taken
off, the load being better distributed over the entire deck.

Below is a study in which the torsional inertia of the main beams Kp is affected by a coefficient (noted f_Kp) varying
from 0.10 to 3.00. The graph presents the comparison of the moments obtained by applying the Guyon-Massonnet
method to the two end beams with e=y:

The different torsional inertias used are listed below:

J=Σ FEM = G-M FEM = G-M

[m4] J = Σab3/3
J_Caquot Software [y=b; e=b] [y=3b/4; e=3b/4]

Kp 0.0261760 0.0199764 0.0191936 0.0282102 0.0258026

These results indicate the importance of the evaluation of torsional stiffness in the calculation of moments by the
Guyon-Massonnet method. The presence of the one-way joist slab tends to increase the main beams’ torsional inertia.

1.6.2 Consideration of the geometry of the edge beams and end cross girders

The mid-span moments per beam for a 100-degree isostatic deck are (for e=y):

Moment - Bar model: Moment - Bar model: variable inertia beams (edge) and end cross
Beams beams of identical inertia = basic girders
case

Point load Distributed load

y(m) Point load Distributed load Impact Impact

Edge 169 kN.m 99 kN.m 154 kN.m -9.81% 86 kN.m -15.24%

-3.375 121 kN.m 65 kN.m 123 kN.m 1.31% 66 kN.m 1.62%

-2.250 107 kN.m 52 kN.m 107 kN.m 0.60% 53 kN.m 1.15%

-1.125 102 kN.m 48 kN.m 102 kN.m 0.01% 48 kN.m 0.02%

Center 101 kN.m 47 kN.m 100 kN.m -0.23% 46 kN.m -0.47%

These results, which are far from calling into question the method’s validity, indicate the importance of the decrease in
the stiffness of the edge beams and the influence of the end cross girder.

Note: the torsional inertias retained are given below (software column):

J[m4] J = Σab3/3 J=ΣJ_Caquot

Software

Beam 0.0191936 0.0261760 0.0199764

Cross girder 0.0069919 0.0078125 0.0069646

End cross girder 0.0232905 0.0318938 0.0232925

Edge beam J=ΣJ_Caquot 0.0242880 0.0181062

1.6.3 Complex model results

Beams – y(m)
Max. span moment
edge -3.375 -2.250 -1.125 Center

Point load 147 kN.m 120 kN.m 107 kN.m 101 kN.m 99 kN.m

51 kN.m 44 kN.m 43 kN.m

Isostatic deck with a 70 Distributed load 77 kN.m 61 kN.m
degree skew
The influence of the skew is very small and the common cross girders (hollow core
Note
elements) are arranged skewed.

Point load 138 kN.m 105 kN.m 98 kN.m 88 kN.m 93 kN.m

72 kN.m 54 kN.m 43 kN.m 38 kN.m 37 kN.m

Distributed load
Isostatic deck with a 50
degree skew
The common cross girders are laid out straight. The end cross girder, on the obtuse angle
Note side, is in flexion rather than torsion, therefore it reduces the span moments to its
detriment.

Point load 130 kN.m 105 kN.m 93 kN.m 90 kN.m 88 kN.m

Statically indeterminate 65 kN.m 51 kN.m 41 kN.m 37 kN.m 38 kN.m

Distributed load
deck (2 spans) with a
100 degree skew
There is no variation on the load distribution between beams caused by the deck being
Note
statically indeterminate

Point load 132 kN.m 97 kN.m 91 kN.m 82 kN.m 87 kN.m

Statically indeterminate 66 kN.m 48 kN.m 36 kN.m 33 kN.m 31 kN.m

Distributed load
deck (2 spans) with a
50 degree skew
The fact that the deck is statically indeterminate causes little variation in the span
Note
moments.

Beams – y(m)
Max. pile moment
edge -3.375 -2.250 -1.125 Center

Point load -63 kN.m -40 kN.m -33 kN.m -31 kN.m -30 kN.m

Statically indeterminate -70 kN.m -46 kN.m -41 kN.m 39 kN.m 38 kN.m
Distributed load
deck (2 spans) with a
100 degree skew
No variation on the load distribution between beams caused by the deck being statically
Note
indeterminate

Point load -84 kN.m -53 kN.m -31 kN.m -28 kN.m -27 kN.m
Statically indeterminate
-92 kN.m
deck (2 spans) with a Distributed load -46 kN.m -37 kN.m -36 kN.m -36 kN.m
50 degree skew
Note There is a concentration of moments close to the obtuse angle of the load.

1.7 Conclusion

For a classical bridge deck, i.e. a high stiffness of the beams compared to the cross girders, the difference of the finite
element model beam grid compared to other methods is mainly expressed through the consideration of a lower
stiffness of the edge beams and the consideration of the end cross girders.

When the skew of the deck is pronounced, only a finite element model can accurately value the reduction elements of
the beams.
$translationBooks

PART 3 – EXAMPLES AND COMPLETE CASE STUDIES

Contribution to the beam grillage

analysis
Contribution to the beam grillage analysis
Author: Pierre PERRIN – Assistant director of the bridges department - east department

Date: Wednesday, April 29th, 2020

1) Introduction
The modeling of structures by a grillage model is frequently used in the field of structural engineering. This technique
consists in establishing systems of 2 or more beam families, most often parallel, which are assembled at nodes and
able to make various angles between them. It has the advantage of being declined to many typologies of works, and
allows to perform the modeling of a work with a simple bar calculation software, while retaining the possibility of local
refinement. The exploitation of the results is also advantageous in that the forces are directly recovered in the bars.
On the other hand, the implementation of a grillage model, depending on the type of work and the load to be studied,
can become tedious. Indeed, the multiplicity of beam elements to be defined, the precautions to be taken in the
definition of the mechanical characteristics, or the need to discretize the applied loads can considerably lengthen the
time required for the model. Moreover, if this work is not carried out with sufficient precautions and checks, the
accuracy of the obtained results can be affected.

Grillage modeling competes with other methods or tools to evaluate the mechanical behavior of structures. Thus, we
find in the literature several analytical methods or charts for the cross-sectional distribution calculation. In addition, the
appearance of high-performance finite element calculation software, with the possibility of modeling in beam
elements, but also in plate elements or in 3D, offers multiple alternatives to the classic grillage modeling.

This chapter details through examples the different structures that can be modeled, as well as the multiplicity of finite
element models that can be considered, for a given structure, as an alternative to a conventional grillage model. We
are particularly interested in the problem of structures such as bridges with girders under pavement, including
intermediate spacers or not, and possibly with a skew. The examples are intended to present the advantages and
disadvantages in the development or operation of the results of each type of model.

2) Modeling
2.1 Models in classic beam grillage

2.1.1 Types of bridges which can be modeled by grillage models

In a classic way, the principle of a grillage model consists in modeling a structure in the 2 dimensions of space, the
structure being able to receive loads perpendicularly to its plane. By extension, a grillage model can also be realized in
the 3 dimensions of space to take into account offsets between the elements. A grillage is generally used to model
structures or parts of structures having a behavior similar to that of a slab, with possibly a preferred direction
(orthotropy, as is the case for girder bridges under pavement).

The structures concerned are numerous, and we can list in a non-exhaustive way:

orthotropic bridges, reinforced concrete or prestressed concrete bridges, possibly of variable inertia
longitudinally, and with or without cantilevers.
ribbed slab bridges, with wide or narrow ribs
bridges with girders under pavement, with reinforced concrete girders, or prestressed concrete girders, by
post tension or by pretensioning. These structures may possibly include cross girders on supports or
intermediate cross girders (intermediate spacers are excluded for pretensioned).
bridges with coated girders and possibly with open-webbed and transversely prestressed girders.
Lightweight concrete bridges
etc.

Structures can have several spans and may present a skew.

2.1.2 Classic grill examples

We give here a few examples to illustrate the diversity of structures that can be modeled using classic grills. We draw
the attention to the fact that this type of model requires being careful about the characteristics to be assigned to the
bars in order to avoid redundancy of bending and torsional stiffness (see on this subject the recommendations of the
SETRA (operating society for transport and automobile repairs) guide PRP 75 [2]).

Figure 1 below shows a beam grid made with the ST1 software for the recalculation of a pretensioned type structure.
This structure has been reinforced at the edge by a coated beam structure that appears in the model with a negative
offset. The structure has a skew of 80 grades, a width of 16.20m and has 2 isostatic spans of 11.36m each.

Grillage model of reinforced Cross section

pretensioned type structure Detail of the area reinforced with coated girders
Figure 1 – Grillage model of a pretensioned type structure reinforced by an adjoining structure

The modeling is based on nodes defined at the center of gravity of the hollow-core element. Transversally, the hollow-
core element is represented by bars connecting these nodes. Longitudinally, bars representing each beam and its
associated hollow-core element are defined by leaning on the same nodes through rigid offsets: the effective distance
between the centers of gravity of the elements is therefore taken into account. The specific offset of the reinforced
zone in coated beams is taken into account with the same principle.

Figure 2 below shows a grillage model generated by Cerema's (center for studies and expertise on risks, the
environment, mobility and development) CHAMOA-P bridge design software. This model is made specifically for the
calculation of stresses in the joist trimmer beams of a reinforced concrete or prestressed concrete type slab bridge.
The structure here comprises 3 spans of 11.5m - 15m - 11.5m for a width of 17.50m and has a skew of 76 degrees.
The model makes it possible to reproduce the behavior of the 3D structure in particular at the supports to dimension
the reinforcement in these areas. It takes into account the positioning and elasticity of each support and includes a
refined discretization on supports to allow the precise evaluation of longitudinal and transverse local moments.

Figure 2 – Grillage model of a reinforced concrete type slab bridge

Figure 3 shows a project for the widening of a concrete ribbed bridge which is prestressed by a mixed box girder type
structure. The initial structure has 2 continuous spans of 33.40m and 36.30m respectively, for a total width of 12.70m.
The 2 ribs, the box girder and their associated hollow-core elements are represented by offset longitudinal bars. The
intermediate and cantilever hollow-core elements are represented by longitudinal and transverse bars.

Cross section of existing deck

 Figure 3 – Grillage model for the widening of a 2-ribbed bridge which is pre-stressed by connection with a mixed box
girder type structure.

The examples in Figure 4 and Figure 5 below are taken from the SETRA (operating society for transport and
automobile repairs) Guide PRP 75 [2]. Figure 4 concerns the beam grid modeling of a 2 wide rib structure with a span
of 39.50m, consisting of an isostatic span with a skew of 51 degrees.

Figure 4 – Grillage model of a ribbed slab bridge

Figure 5 shows a slab with cut-outs, whose morphology is similar to that of a multi-cellular box girder bridge. The
layout of the supports leads to an atypical trapezoidal geometry that requires a thickening of the webs on the high
span side.

Grillage modelling

Cross section

Figure 5 – Grillage model of a hollow core slab

The examples presented illustrate the versatility of beam grid modeling. However, these models present some
difficulties. On the one hand, the discretization into beam elements can be tedious to implement and make the
application of moving loads on the structure more complex. On the other hand, the recovery of results in forces or
constraints may require a reconstruction for justification. Depending on the case studied and the tools available, it may
be appropriate to use a model comprising plates or shells elements to represent the slabs or the portions of the hollow
core deck.

2.2 Finite element, plate and 3D modeling

2.2.1Beams and plates modeling

Plate (or shell) type finite elements allow to represent either directly the slab or only the hollow core part of a girder
bridge. The use of plates and shells has a double advantage: on the one hand the implementation of the model and
the loads is easy, and on the other hand the representation of the behavior of the structure is more faithful to reality
(plate effect). The constitution of the plate element mesh is managed by the software’s algorithm, which simplifies the
construction compared to a grillage model. The geometry of the mesh can be performed in a single plane by defining if
necessary, and if the software allows it, an offset of the plates or beams with respect to the mesh. This offset doesn't
appear in the mesh but simply as a characteristic of the elements: it is therefore advisable to check that the software
actually takes into account the offsets. This verification can be done at least by comparing the average deflection
calculated with the finite element model under uniform loading and that calculated with the beam model affected by
the characteristics of the total deck width.

Figure 6 - Modeling beams and plates, plane geometry

If the software used doesn't offer the possibility to offset plates or beams, it is necessary to build a mesh with rigid
offsets (Vierendeel girder modeling). The realization of the model is then more complex but the facilities provided by
the plates are kept.

Figure 7 - Modeling beams and plates, geometry with rigid offset bars

The presence of intermediate cross girders can also be taken into account by adding crossbeam elements. The use,
from a plane geometry, of plate or beam offsets then remains possible. The implementation of the cross girders can
also be carried out by integrating offset bars into the mesh (Figure 8).

Figure 8 - Beams and plates modeling, geometry with rigid offset bars and intermediate cross girders

Based on these main modeling principles, several variants can be established to be adapted according to the
morphology of the deck. For example, the rigid offset bars can also be defined as skewed as in the example defined in
Figure 3.

The support conditions must include a fixed support for the torsional effect on beams. This assumption is valid
whatever the model used (beam grid, or volumetric finite elements), and is also found in the basic assumptions of the
analytical methods for calculating the transverse distribution described below.

To apply forces to the model, it is necessary to provide either a geometry with groups of meshes dedicated to loading,
or a sufficiently fine and regular mesh of the slab (generally quadrangles) to allow the description of the impact zones
a posteriori by selection of meshes.

The exploitation of the results is carried out from the forces coming from the different elements, which must be
recombined for the justification of the composite beam (or cross girders) sections and hollow core elements.

2.2.2 3D modeling

The realization of a complete 3D model of the structure can be relevant in certain cases, for example in order to check
local stresses, or when the geometry has a particular configuration that is difficult to model using only plate and beam
elements. The construction of the geometry in 3D, if it appears complex at first glance, can be done efficiently with
current 3D design tools that offer advanced extrusion functions. The 3D approach is valuable as it allows to avoid the
calculation of the mechanical characteristics of the beam elements. In addition, the location of the material is precise,
allowing in particular to define the transverse span of the hollow core elements between beam ends and not between
axes, or to avoid material redundancies at the intersections between elements that are found in other types of
modeling. Finally, this type of model allows a direct graphic visualization of the geometry and deformations.

Figure 9 - 3D volumetric modeling of a post tensioned type structure with cross girders

The definition of the study sections, in order to be justified, must be thought out from the beginning if we want to
obtain clean cuts with nodes located in the cutting plane. In the previous example, since the model was built by
successive extrusions between cross girders, additional study sections were provided to facilitate the exploitation of
the results (Figure 10) while reducing the size of the output files.

Figure 10 - 3D volumetric modeling: study sections

The volume model must also include a fixed support for the torsional effect on beams. It is possible to create this
embedding using boundary conditions, or by modeling the volumes of the supported beams as in Figure 9.

The application of forces on the model must, as with plate models, be anticipated, and can also be done in 2 different
ways. In the case of considering groups of meshes dedicated to load impacts, it is possible to de-correlate the load
impacts of the intermediate study sections (longitudinal or transverse) by providing a horizontal separation of the
upper part of the hollow core element, as shown in Figure 11. This arrangement allows multiple loading configurations
to be managed without interfering with the deck construction elements.

Figure 11 - Volumetric modeling: definition of loading impacts

The results from a volumetric model are given in constraints: depending on the justifications to be carried out it is
necessary to integrate these constraints to deduce the torsor resulting in forces.

2.3 Analytical methods for calculating transverse distribution

Independently of finite element methods, the search for alternatives to classic grillage modeling has already been the
subject of numerous works. As early as the 1920s, conservative methods based on simplified hypotheses made it
possible to treat the simplest grillage systems using charts. Subsequently, the three main analytical methods still used
today are those of Guyon-Massonet, Courbon and CartFauchart. We recall here the principles and the fields of validity.
Recent improvements have been made to these methods, such as the analytical formulation of Millan which
generalizes the Guyon-Massonet and Cart-Fauchart methods.

Guyon-Massonet method
A major evolution in transverse distribution calculations was provided in the 1960s by the theories of Guyon, Massonet
and Barès, who developed a calculation method [3] [4] taking into account the torsional rigidity of the elements. A
novelty of this method is the assimilation of the structure to an orthotropic slab, governed by an equation with partial
derivatives of the form;

with:

P Longitudinal bending stiffness per unit length

E Transverse bending stiffness per unit length

P Torsional rigidity around the longitudinal axis per unit length

E Torsional rigidity around the transverse axis per unit length

This so-called "Guyon-Massonet" method is based on the calculation of the transverse distribution coefficients which
simplify the resolution of equation (1) in order to obtain the deformation and then the forces in the structure. It was
then improved and extended by other authors, for example to extend it to the calculation of structures with edge
girders different from standard girders.

This method is recommended for the calculation of slab or girder bridge type decks with a high number of girders and
with or without intermediate cross girders. Due to the slab analogy, the method is well adapted to decks with more
than four girders. For a 3 or 4 girder structure, it is preferable to use the Cart-Fauchart method if there are no
intermediate cross girders.

Courbon Method

The principle of this method was set out in the Annales des Ponts et Chaussées (Bridges and roads annals) of
November-December 1941 [5]. 5] It applies to structures with intermediate cross girders which are considered as
infinitely rigid. In practice, it is used when the span of the structure is about twice its width, and the height of the
spacers is close to that of the beams. Figure 12 illustrates the principle of the method: an infinitely rigid section of
deck transversely rests on springs whose stiffness is proportional to the deflection at half span of the beams. The
reactions on each spring give the transverse distribution of the applied load.

Figure 12 - Courbon method principle

Cart-Fauchart method

This method was described in the technical institute of public works (ITBTP) annals N° 271-272 of July-August 1970 [6].
It is applicable to decks without cross girders (except on supports) and made of beams, ribs of constant section, or
multiple box girders if the deformation of the box girders can be disregarded. The hollow core element is assimilated
to a series of infinitely thin cross girders fixed on the beams. The resolution of the equation system of deformation is
carried out by development in Fourier series.

This method can be used for bridges with girders or ribs without cross girders, as well as for multiple box girders, if, as
a first approximation, the inherent deformation of the box girders is disregarded.

Innovative Millan method

This method is also based on the theory of orthotropic slabs and was presented in the newsletter Construction
Métallique (steel construction) n°2 in 2004 [7]. It brings improvements to the Guyon-Massonnet method: a more
efficient analytical formulation, the possibility to take into account a non-zero Poisson’s ratio for orthotropic slabs, or to
impose edge conditions on slab elements. This last possibility allows the study of the local bending of the hollow core
element by fixing the free edges or the modeling of several contiguous slab elements. It is thus possible to finely
model girder or ribbed bridges - which is a generalization of the Cart-Fauchart method - without necessarily having all
these elements being identical. Practical elements concerning the application of the Millan method to the modeling of
girder bridges of any geometry are given in article [9] of the Bulletin Ouvrages d'Art (Bridges newsletter) n°71.

2.4 Criteria for choosing a model or calculation method

2.4.1 Geometrical specificities of the work

The analytical methods presented above are only valid if they remain within their scope. The criteria related to the
cross girders’ stiffness have been recalled for each method and make it possible to orient oneself towards the method
that seems best adapted to the structure, subject to the condition that the structure presents a relatively regular
geometry. On the other hand, if the studied structure is more atypical - with asymmetries, thickness variations, strong
skew (> 70 degrees), local reinforcements... - it becomes necessary to build a specific calculation model such as a
grillage model or finite element model.

2.4.2 Loading
Loading the model can be more or less complex. In particular, the discretization of loads on a conventional grillage
model is often tedious and requires approximations. From this point of view, finite element modeling with plate or 3D
elements is preferable in terms of accuracy of load impacts.

The choice of the model can also be determined by the automatic load displacement facilities that may be provided by
the software used.

Whichever model is chosen, it may be judicious to go through the construction of the lines or influence surfaces of the
studied effect to perform the resolution of the equilibrium of the model only once, and to quickly deduce the forces
obtained under any loading position.

2.4.3 Global or local justification

Depending on the element to be justified, the model can be global or local. For example, to study the local stress path
in assemblies, or in parts of specific geometry, a 2D or 3D finite element modeling is the most appropriate. On the
other hand, other local effects do not systematically require the use of 2D or 3D finite element modeling. For example,
the calculation of local forces in the joist trimmer beam of slab bridges, as shown in Figure 2, is carried out with a
simple refinement of the model in a grillage model at the support.

2.4.4 Post processing

Depending on the nature of the justifications to be conducted, the results from the model may require post-processing
which can be managed either by the software or by an external component. The most frequent problems are those of
combinations or envelopes of forces. This functionality should not be neglected for studies with multiple load cases.

3) Specific points and examples

3.1 Transverse moment distribution

Results obtained with different models

The different models or calculation methods presented above give similar results, but with generally safe values for
the Guyon-Massonet method. Below is an example of a pretensioned structure with 26 beams, a span of 25m, a width
of 21m, with identical beams, subjected to the centered loading of an regulatory E3F1 type exceptional convoy (400
tons). The structure is studied with several different models: Guyon Massonet model, Millan multi-plate model
(contiguous slabs), volumetric model (Figure 13). and model plates and beams.

Figure 13 - Volumetric model: displacements under E3F1 convoy

The curves of the transverse distribution coefficients of moments at mid-span (Figure 14) are consistent between all
the models, except for the Guyon Massonet method, which remains safe (16% increase in bending forces in the most
stressed beam).
Figure 14 - Comparison of distribution coefficients on the different models

Case of structures with reinforced edge beams

The Guyon-Massonet method, which comes down to an equivalent orthotropic slab, does not allow the direct
processing of the excess stress generated on the edge beams when these are more solid or twinned (a layout
commonly adopted to deal with the problem of off-road vehicle impacts). An adaptation of the method to this specific
case was therefore proposed in the European Journal of Civil Engineering [8]. Otherwise, if no adaptation is carried out,
the Guyon-Massonet method becomes false for structures with solid or twin edge beams: the example of a structure
with twin edge beams (Figure 15) shows that the forces obtained in the edge beams are largely underestimated and
those in the central beams overestimated compared to other models.

Transversal distribution of moments. Structure with twin edge beams

Figure 15 - Comparison of the distribution coefficients on a structure with twin edge beams (source [9] Bulletin
Ouvrages d'Art (Bridges newsletter) n°71)

Evolution of the transverse distribution according to the load position

No matter which model is used, it is customary to dimension the structures by considering the transverse distribution
of moments as identical all along the length of the structure, by safely using the distribution coefficients obtained at
mid-span. The figure below illustrates this hypothesis on the case of the previous 26-beam structure subjected to a
uniform surface load of 3m wide over the entire deck length. The transverse distribution of moments on each beam is
given at different study sections. It can be seen that the maximum coefficients decrease as the study section deviates
from the mid-span section. However, these coefficients apply at lower moments, which limits the oversizing of sections
away from mid-span.

Figure 16 - Distribution coefficients according to longitudinal study section

Using the mid-span as the study section for the distribution of moments, a maximum coefficient of 2.55 is obtained in
the most stressed beam at mid-span. The influence of the load position on this distribution is then studied. The 3m
wide load, previously applied over the whole length, is reduced to a length of 1m and then positioned at different
distances from the support. Figure 17 shows the decrease in the coefficients when we move away from mid-span.

Figure 17 - Mid-span moment distribution coefficients for different load positions

3.2 Transverse distribution of shear force

The study of the transverse distribution of the shear force near the supports is often carried out by simplification using
the coefficients of transverse distribution of the moments at mid-span. It should be noted that this hypothesis can lead
to deviations of around 20% from the actual distribution of the shear force. As an example, Figure 18 shows the values
of the transverse distribution coefficients under the effect of an E3F1 convoy running centrally. For the most stressed
beam, the ratio between the distribution coefficients of the shear force on support (Kq, blue curve) and the moment at
mid-span (Km, orange curve) is 18%.

Figure 18 - Cross-sectional distribution of mid-span moment and shear force on support

The previous curve of shear force distribution on support is obtained for the E3F1 uniform load which is applied over
the entire length of the structure and over a 5.15m width. Actually, the transverse distribution depends on the
longitudinal positioning of the applied load, for a given transverse width of the load. The study of the effect of the
longitudinal position of the load on the distribution is carried out by assuming a 3m wide and 1m long load, positioned
at different distances from the support (Figure 19). It can be seen that near the support the loads don't have the effect
of distribution: only the beams located immediately under the load take up the loads. This underlines the fact that the
given order of magnitude of 20% difference between Kq and Km applies to a longitudinally uniformly distributed load:
if the load is punctual and depending on its position on the deck, the deviations can be more significant.

Figure 19 - Distribution of shear force on support for different load positions

Furthermore, it should be noted that near the support, the shear forces are actually reduced because the efforts are
transferred directly to the support. This favorable effect can only be taken into account in a 3D model in which a
diffusion of forces (direct transmission connecting rod) takes place in the material near the support.

4) References
[1] - Guide technique CHAMOA P CHaîne Algorithmique Modulaire Ouvrages d’Art – Annexes
http://www.setra.fr/html/logicielsOA/Ponts_Types/CHAMOA-P/chamoa-p.html
[2] - Guide pour l’utilisation des programmes de réseaux de poutres - PRP 75 - SETRA -1975

[3] - Compléments à la méthode de calcul des ponts à poutres multiples - C.Massonnet - Annales de l’ITBTP - janvier
1962.

[4] - Le calcul des grillages de poutres et dalles orthotropes selon la méthode GuyonMassonnet-Barès - R.Barès et
C.Massonnet - Dunod – 1966

[5] - Calcul des ponts à poutres multiples solidarisées par des entretoises – J.Courbon Annales des ponts et chaussées -
Novembre-Décembre 1941

[6] - Méthode de calcul des ponts nervurés sans entretoise intermédiaire - Annales de l'ITBTP -Juillet-Aout 1970

[7] - Nouvelle formulation analytique de la flexion transversale d'une dalle orthotrope A.L.Millan - Construction
Métallique n°2 – 2004

[8] - Méthode de Guyon Massonnet Barès appliquée aux ouvrages à poutres d'inertie distincte -G.Bondonet et P.Corfdir
- Revue Européenne de Génie Civil - Volume 9, n°9-10 – 2005

[9] – Calcul analytique de flexion des ponts à poutres de géométrie quelconque, calage des inerties de torsion
transversale par comparaison à des calculs aux éléments finis – P.Perrin et G.Bondonet – Bulletin Ouvrages d’Art - n°71
- 2015
$translationBooks

PART 3 – EXAMPLES AND COMPLETE CASE STUDIES

Transverse bending of a multi-

beam bridge
Transverse bending of a multi-beam bridge
Author: B. Tritschler (Arcadis) – April 29th, 2020

1. Preamble
This document presents a solution for modeling a multi-beam deck with high longitudinal bending stiffnesses (pre‐
tensioned or post-tensioned prestressed concrete beam types).

The purpose of this model is, more particularly, to study the transverse bending (study of the one-way joists and the
intermediate cross girders) and to conclude on the relevance or not to model the one-way joist slab in an offset way,
according to the efforts that we want to recover.

1.1 Need for a global study

The mechanical behavior of the hollow core element of the deck slab with intermediate cross girders is strongly
conditioned by the transverse stiffness provided by the cross girders. In this case, the study of the slab is inseparable
from the study of the intermediate cross girders.

In order to get as close as possible to the real state of stress of the slab and the cross girders, it is necessary to carry
out a global study of the structure through a model using the stiffness of all the deck elements (beams - cross girders
and slab).

The forces applied to the cross girders and the slab will be determined using a common finite element software to
model the entire deck.

This model is the closest to reality because it takes into account the flexibility of the beams and the stiffness of the
slab (when they exist).

The forces provided by the model directly cumulate the local forces and the general transverse bending of the deck.

In practice, there are two possible models to choose from:

3D modeling, with hanging beams and cross girders offset with respect to the plane (or level) of the slab.
“Pseudo-plane" modeling, with beams and cross girders in the same plane (or level) as the slab.

Notes:

The different models will be set up with the interaxis distances of the beams and cross girders.

Transverse prestressing is not presented in this study.

1.2. PURPOSE OF THE NOTE

The purpose of this note is to compare two models of a same deck;

a model taking into account the offset of the "bar" elements, i.e. the hanging beams and the cross girders.
a model considering the "bar" elements that are the beams and the cross girders, without offset with respect
to the plane of the hollow core element.

In both cases, the one-way joist slabs are modeled as shell elements of varying thickness.

The aim is to compare the results according to two criteria:

validity of the results - direct comparison of the results provided by the two models / validation of the results.
ease of exploitation of the results / analysis of the functioning of the models.

2. MODELING
2.1. REAL GEOMETRY

The geometry of the structure is detailed below.

2.1.1. Longitudinal section

Longitudinal section (cf. A009_0729-3_a1966_PL 05d_OUVRAGE_coffrage)

2.1.2. Cross section

2.1.3. Beam

2.1.4. Cross girders

2.2. GEOMETRY OF MODEL N°1

The first model consists of taking into account the offsets of the hanging beams and cross girders. For this model the
geometry is visually very close to the real geometry.

2.2.1 Beams

Beam compression flanges are modeled in shell model as part of the slab. Only the hanging beams are modelled in a
"bar" model, which is the most suitable hypothesis given their "linear" geometry.

The "complex" geometry of the hanging beams does not allow us to directly enter their dimensions in Robot; we will
provide the software directly with the mechanical characteristics to be taken into account.

SECTION ANALYSIS
The variation in web thickness near the supports is neglected (little influence for the calculation of the bearing and the
slab).

(hbeam = 1,90 m – hcompression flanges = 0,22 m – vpz = 0,642 m) => offset = 190 – 64,2 – 22/2 = 114,8 cm
rounded to 1.15 m.

2.2.2 Cross girders

Cross girders compression flanges are modeled in shell model as part of the slab. Only the hanging beams are
modelled in a "bar" model, which is the most suitable hypothesis given their "linear" geometry.

The geometry of the hanging beams is entered directly into the software.

=> offset. = 122/2 + 16/2 = 69cm.

2.2.3 One-way joist slab

The slab is made up of the elements between compression flanges of beams “hollow bricks of the joist slab " of
constant thickness = 16cm, and the compression flanges, which will be considered variable from 16cm at the edge, to
22cm at the axis.
2.2.4 Span

The span is modeled on single supports (1 support/beam).

2.3. GEOMETRY OF MODEL N°2

For the second model, we consider the linear elements (beams and cross girders), in the middle plane of the slab. In
order to model the longitudinal and transverse bending stiffnesses as accurately as possible, we will consider the real
sections of the beams and cross girders.

2.3.1 Beams
The "complex" geometry of the hanging beams does not allow us to directly enter their dimensions in Robot; we will
provide the software directly with the mechanical characteristics to be taken into account.

SECTION ANALYSIS

2.3.2 Cross girders

In accordance with the requirements of the standard post-tensioned prestressed beam file, the width of the slab taken
into account on either side of the cross girders rib is equal to one tenth of the distance between the axes of the edge
beams, i.e. 1.80m.

Flange total width = 1.80 + 0.25 + 1.80 = 3.85

2.3.3 Slab
The modeling of the slab is strictly identical to the one used for model n°1.

2.3.4 Span

The span is modeled on single supports (1 support/beam).

2.4. LOADING

2.4.1 Principle

We will compare the effect of identical loads on the two models.

To do so, we will consider cases of arbitrary charges:

a distributed load over the entire slab (equivalent to loads brought by superstructures or to the loads of
system A of instalment 61 part II)
point loads, or system of point loads (corresponding to concentrated loads, reconciliation of the systems of
loads of system B, of instalment 61 part II).

2.4.2 Distributed load

A distributed load of 1 t/m² (order of magnitude value of A(1)) will be considered.

2.4.3 Bc system (2 trucks of 30Ton per lane)

We will consider two Bc trucks, arranged side-by-side in order to obtain a maximum moment in the slab (resulting from
the rear axles being centered on the plate).

Note: the central slab is not loaded (presence of a DBA).

2.4.4 Point load

We consider a point load of 10 Tons (equivalent to the Br wheel) centered on the plate.
2.5 SUPPORT CONDITIONS

Valid for both models: a simple support under each beam (except for one beam → pinned).

3. RESULTS
3.1 Central cross girders

3.1.1 Model n°1

3.1.2 Model n°2
3.1.3 One-way joist slab summary

The results provided by model n°1 are not directly usable: the effects only concern the hanging beams (normal effort
+ moment). It is necessary to recombine these effects with those of the one-way joist slab which correspond to the
solicitations of the cross girder’s compression flanges (normal force and moment).

Reconstitution of the forces in the cross girder:

Hypotheses:

The normal effort of the compression flanges balances the one present in the hanging beams. The moment present in
the one-way joist slab is negligible compared to the moment of the hanging beams (global effect).

MTotal=N∙Mhanging beam

For information purposes, we will evaluate the width of the one-way joist slab which is necessary to balance the normal
effort under a load case of type A(l).

Case 1 (distributed load)

(symmetrical load case)

We study several sections and integrate the normal stress. The integration is carried out according to the trapezoidal
rule.

The integration of the normal stress on the sections studied above makes it possible to balance the normal effort
existing in the hanging beams of the cross girder.

The required "real" width of the one-way joist slab is: (16.27 - 14.02)x2 = 4.50m.

That is to say a participating width of 4.50m (instead of the 3.85 recommended by the standard).

The participating width prescribed by the typical post-tensioned prestressed concrete beam standard is close to the
one highlighted in this case: the geometry of the compression flange of the cross girders (T-beam) is validated.

M hanging beams
N (t) e (m) M model n°1 M model n°2 Difference %
(Ton*m)

Distributed load -32.30 -0.69 3.35 25.64 23.71 8.1%

Bc system -25.17 -0.69 6.93 24.30 23.20 4.7%

Point load -6.49 -0.69 1.51 5.99 5.86 2.2%

The visible difference comes from the approximation made as to the width of the compression flange to be considered.
The mapping on the following page shows the variations of the compression state in the slab: the participating width is
variable: increasing from the edge towards the center. This difference remains small and thus allows the validation of
both models.

The model which considers the offsets provides, after reconstitution of the efforts, moments in the cross girders which
are very close to those provided by the model that does not consider offsets.

The increasing complexity of the model does not seem to bring any precision or visible advantage compared to a
simpler model, brought back to the middle plane of the slab.

We recommend using a model similar to the model n°2 presented in this note.

3.2 One-way joist slab

3.2.1 Model n°1

3.2.2 Model n°2
3.2.3 One-way joist slab summary

Model n°1 Model n°2

MYY max (Ton*m/ml) MYY min (Ton*m/ml) MYY max (Ton*m/ml) MYY min (Ton*m/ml)

Case n°1 distributed

0.91 -0.40 0.92 -0.43
load

Case n°2 Bc system 2.75 -1.89 2.82 -1.91

Case n°3 point load 1.11 -1.59 1.14 -1.62

Model No. 2, which does not consider offsets, provides very close moments in the one-way joist slab (within 3%) to
those of the model considering offsets.
The increasing complexity of the model does not seem to bring any precision or visible advantage compared to a
simpler model, brought back to the middle plane of the slab.

We recommend using a model similar to the model n°2 presented in this note.

4. Conclusion
A model with offset beams and cross girders does not provide directly usable results for the study of the beams and
cross girders, contrary to a model brought back to the middle plane of the one-way joist slab.

Both models provide similar results with respect to the bending stresses of the slab.

We select a model similar to the model n°2 presented in this note, such model allowing us to obtain coherent results
that can be directly exploited for the purpose of this study.

Note: The user must verify whether the approach presented in this note is applicable to his or her particular project.

5. APPENDIX - INFLUENCE OF BEAMS’ TORSIONAL INERTIA

For the selected model, we considered the non-cracked torsional inertia of the beams, a hypothesis which is not in
conformity with the recommendations of the SETRA (operating society for transport and automobile repairs) guide PRP
75.

In order to assess the influence of the torsional inertia of the beams on the behavior of the intermediate cross girders
and the hollow core element, we reran a calculation while considerably reducing the torsional inertia (so as to see an
exaggerated effect).

We have divided by 10 the value of Ix retained in the previous model.

5.1 Cross girders

5.2 One-way joist slab

 Cross girders In non-cracked Ix/10 Difference %

Distributed load 23.71 25.94 9.4%

BC system 23.20 23.98 3.4%

Point load 5.86 5.91 0.9%

Hollow core element In non-cracked Ix/10 Difference %

Distributed load max 0.92 0.97 5.4%

Distributed load min -0.43 -0.52 20.9%

BC system max 2.82 2.88 2.1%

BC system min -1.91 -2.01 5.2%

Point load max 1.14 1.21 6.1%

Point load min -1.64 -1.75 6.7%

The moments show little variation relative to the torsional inertia retained.

Taking into account the torsional inertia in the cracked section should reduce the elastic value by about 30%, such a
reduction will have no visible effect on the results.

It should be noted that we are, in almost all cases, in the presence of prestressed cross girders and hollow core
elements which are therefore rarely cracked overall.

We suggest keeping the torsional inertia determined by an elastic calculation in uncracked section.

Note: The study of the influence of the torsional inertia explained above is only valid for structures with intermediate
cross girders.
$translationBooks

PART 3 – EXAMPLES AND COMPLETE CASE STUDIES

Example of a beam grillage

calculation according to different
methods - Part 1
Example of a beam grillage calculation according to different
methods
Comparison of the results - multi-criteria analysis

Editor : Didier GUTH - Arcadis - June 14th, 2020

1) Introduction
1.1 Objectives

In this document, we will model the same multi-beam deck using several approaches:

the Guyon-Massonnet method,

a plane grid approach,
a 3D grillage (modeled as a ladder beam)
a 3D model associating bars and shells, with two approaches.

We will compare:

Support reactions,
Efforts, moments and shear.

In addition, we will perform two transverse bending calculations to highlight the differences and limitations of the
methods.

We will test the incidence of a moderate skew (70 degrees), a more consequent skew (50 degrees) and the presence
of cross girders.

A table, by way of conclusion, will attempt to give the reader some clues as to the advantages and disadvantages of
each of the models.

The dimensions and applied loads are plausible for a structure deemed to be made of reinforced concrete.

We insist on the fact that each work is particular and that we find ourselves, in the context of this example, in a given
configuration of flexural and torsional rigidity, and that consequently the conclusions cannot be generalized as such.

1.2 Bibliography

We invite the reader to refer to the following references:

[1] - Guide technique CHAMOA P CHaîne Algorithmique Modulaire Ouvrages d’Art – Apendix
http://www.setra.fr/html/logicielsOA/Ponts_Types/CHAMOA-P/chamoa-p.html

[2] - Guide pour l’utilisation des programmes de réseaux de poutres - PRP 75 - SETRA -1975

[3] - Compléments à la méthode de calcul des ponts à poutres multiples - C. Massonnet – ITBTP annals - January 1962

[4] - Le calcul des grillages de poutres et dalles orthotropes selon la méthode Guyon-MassonnetBarès - R. Barès et C.
Massonnet - Dunod – 1966

[5] - Calcul des ponts à poutres multiples solidarisées par des entretoises – J. Courbon - Annales des ponts et
chaussées - November-December 1941

[6] - Méthode de calcul des ponts nervurés sans entretoise intermédiaire – ITBTP annals – July-August 1970
[7] - Nouvelle formulation analytique de la flexion transversale d'une dalle orthotrope - A.L. Millan - Construction
Métallique n°2 – 2004

[8] - Méthode de Guyon Massonnet Barès appliquée aux ouvrages à poutres d'inertie distincte - G. Bondonet et P.
Corfdir - Revue Européenne de Génie Civil - Volume 9, n°9-10 – 2005

[9] – Calcul analytique de flexion des ponts à poutres de géométrie quelconque, calage des inerties de torsion
transversale par comparaison à des calculs aux éléments finis – P. Perrin et G. Bondonet – Bulletin Ouvrages d’Art -
n°71 – 2015

[10]- Emploi des éléments finis en génie civil (Tome 1) : La modélisation des ouvrages – sous la direction de Michel
Prat

[11] - Contribution à l’étude des grillages de poutres – Pierre Perrin – Dir Est – sur le wiki de l’AFGC [https://wiki-
gtef.frama.wiki/accueil-gtef:partie-3:exemple-c]

[12] - Flexion transversale d'un pont multipoutre – Benjamin Tritschler – Arcadis - – sur le wiki de l’AFGC [https://wiki-
gtef.frama.wiki/accueil-gtef:partie-3:exemple-c]

[13] Guide pour l’évaluation structurale et la réparation des Viaducs à travées Indépendantes à Poutres Préfabriquées
précontraintes par post-tension (VIPP) – CEREMA – (à paraître)

[14] Dossier PRAD 73 – SETRA

[15] Dossier VIPP 67 – SETRA

1.3 Possible complements

To complete the study, in a non-exhaustive way, in the end we could add:

Modeling with 3D elements (see reference [11])

The study of a longitudinal and/or transverse phasing, taking into account the creep shrinkage, either as at a
fixed rate or by using a calculation with behavioral laws,
A study using an improved "Guyon Massonnet" approach
Tests to find the optimal width of the cross bands
How to take into account geometric or material non-linearities,

And extend the study to cases of structures with flexible connections.

2) Description of the structure and loads

2.1 Geometry

It includes:

Eleven ribs of 40 cm x 100 cm ht, spaced at 0.90 m intervals.

Two 25 m spans,
A 25 cm thick hollow-core element,

It is made of C35 concrete, E=36000 MPa, ν=0.2 and rests on simple supports.

2.1 Studied load cases

The applied loads are:

Self-weight:
A fictitious superstructure load of 3.00 kN/m² on the entire surface of the hollow-core element and 5 kN/ml on
the edges at the end of the cantilever:

Multiple operating loads

A linear load of 9 kN/m in span 1, on beams 1 to 3:

A linear load of 9 kN/m in span 1+2, on beams 1 to 3:

A distributed load from the abscissa 10.00 to 11.25 m, astride beams 6 and 7
(will be used for the transverse bending study)

A point load on the abscissa 1.25 m from an abutment on beam n°4:

(will be used to study support reactions)

A linear load of 100 kN/ml at the edge of the deck:

(will be used for the transverse bending study)

3) Different modelling approaches

3.1 Approach using the Guyon Massonnet method

We will not come back to the method as such, it is largely explained in the texts cited in the bibliography of this
document.

In practice, to apply the method, first we must calculate the efforts, case by case, on a 2D beam representing the
whole width of the deck (if we are not interested in the deformations, this beam can have ordinary characteristics, as
long as they are constant).

The cases of charges to be applied have been described in the previous §.

(For cases modeling a load distributed over two spans, we could have modeled a unit load and multiplied the results by
the effective load).

Summary of charges:

Self-weight (only one rib in our case) - case 1.

Superstructure - distributed load - case 2:
 Cross beam - case 3 (covers both cases in the Guyon-Massonnet approach):

Case span 1 and span 1+2 - cases 11 and 12

And the specific case for reactions:

Results:

Self weight:

Superstructures, 3 kN/m²:
 Superstructures, cross beams:

Envelope for cases span 1 and spans 1+2:

We use a program that automatically determines the parameters and calculates the lines of influence, also called
"Guyon Massonnet coefficient" (KGM):

Guyon-Massonnet coefficients -> to divide by 11

Beam 1 2 3 4 5 6 7 8 9 10 11

Y= -4.5 -3.6 -2.7 -1.8 -0.9 0 0.9 1.8 2.7 3.6 4.5

Super
1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000
distributed

Left cross
3.287 2.523 1.860 1.317 0.893 0.577 0.348 0.186 0.072 -0.013 -0.081
beam
 Right cross
-0.081 -0.013 0.072 0.186 0.348 0.577 0.893 1.317 1.860 2.523 3.287
beam

∑ cross beams 3.206 2.510 1.932 1.503 1.241 1.154 1.241 1.503 1.932 2.510 3.206

q span 2.407 2.089 1.751 1.407 1.087 0.812 0.586 0.406 0.262 0.144 0.042

support 1.347 1.412 1.459 1.445 1.322 1.136 0.933 0.738 0.560 0.401 0.257

As an example, graphically, the KGM coefficients:

for the left cross beam - (e.g. beam 2 takes up 2.50/11 of the load)

for distributed load q (spans 1 or 1+2):

The second column contains the moment M, the shear V or the reaction Rmax, for the unit cases, for the whole deck,
which are distributed according to the KGM of each beam:

Beam 1 2 3 4 5 6 7 8 9 10 11

Moment on support

-1198. -1198. -1198. -1198. -1198. -1198. -1198. -1198. -1198. -1198. -1198.
Self weight -1198.0
0 0 0 0 0 0 0 0 0 0 0

Superstructure -2320.0 -210.9 -210.9 -210.9 -210.9 -210.9 -210.9 -210.9 -210.9 -210.9 -210.9 -210.9

Cross beam -391.0 -114.0 -89.2 -68.7 -53.4 -44.1 -41.0 -44.1 -53.4 -68.7 -89.2 -114.0

Span -2109.0 -461.5 -400.5 -335.7 -269.8 -208.4 -155.7 -112.4 -77.8 -50.2 -27.6 -8.1

-18389. -1984. -1898. -1813. -1732. -1661. -1605. -1565. -1540. -1527. -1525. -1530.
M=
0 4 6 3 1 4 6 4 2 8 7 9

Moment Span 1

Self weight 674.0 674.0 674.0 674.0 674.0 674.0 674.0 674.0 674.0 674.0 674.0 674.0

Superstructure 1305.0 118.6 118.6 118.6 118.6 118.6 118.6 118.6 118.6 118.6 118.6 118.6

Cross beam 220.0 64.1 50.2 38.6 30.1 24.8 23.1 24.8 30.1 38.6 50.2 64.1
 Span 1615.0 353.4 306.7 257.1 206.6 159.6 119.2 86.0 59.6 38.5 21.1 6.2

M= 10774.0 1210.1 1149.5 1088.4 1029.3 977.0 934.9 903.5 882.3 869.7 864.0 862.9

Shear C0

Self weight 144.0 144.0 144.0 144.0 144.0 144.0 144.0 144.0 144.0 144.0 144.0 144.0

Superstructure 278.0 25.3 25.3 25.3 25.3 25.3 25.3 25.3 25.3 25.3 25.3 25.3

Cross beam 47.0 13.7 10.7 8.3 6.4 5.3 4.9 5.3 6.4 8.3 10.7 13.7

Span 295.0 64.6 56.6 47.0 37.7 29.2 21.8 15.7 10.9 7.0 3.9 1.1

V= 2251.0 247.5 236.0 224.5 213.4 203.7 196.0 190.3 186.6 184.6 183.9 184.1

Shear P1, left

Self weight -240.0 -240.0 -240.0 -240.0 -240.0 -240.0 -240.0 -240.0 -240.0 -240.0 -240.0 -240.0

Superstructure -464.0 -42.2 -42.2 -42.2 -42.2 -42.2 -42.2 -42.2 -42.2 -42.2 -42.2 -42.2

Cross beam -78.8 -22.7 -17.8 -13.7 -10.7 -8.8 -8.2 -8.8 -10.7 -13.7 -17.8 -22.7

Span -422.0 -92.3 -80.1 -67.2 -54.0 -41.7 -31.2 -22.5 -15.6 -10.1 -5.5 -1.6

V= -3682.0 -397.3 -380.1 -363.1 -346.8 -332.7 -321.5 -313.5 -308.4 -305.9 -305.5 -306.5

Abutment reaction C0

Self weight 144.0 144.0 144.0 144.0 144.0 144.0 144.0 144.0 144.0 144.0 144.0 144.0

Superstructure 278.0 25.3 25.3 25.3 25.3 25.3 25.3 25.3 25.3 25.3 25.3 25.3

Cross beam 47.0 13.7 10.7 8.3 6.4 5.3 4.9 5.3 6.4 8.3 10.7 13.7

Span 295.0 64.6 56.0 47.0 37.7 29.2 21.8 15.7 10.9 7.0 3.9 1.1

R= 2251.0 247.5 236.0 224.5 213.4 203.7 196.0 190.3 186.6 184.6 183.9 184.1

Reaction pile P1

Self weight 479.0 479.0 479.0 479.0 479.0 479.0 479.0 479.0 479.0 479.0 479.0 479.0

Superstructure 928.0 84.4 84.4 84.4 84.4 84.4 84.4 84.4 84.4 84.4 84.4 84.4

Cross beam 156.0 45.5 35.6 27.4 21.3 17.6 16.4 17.6 21.3 27.4 35.6 45.5

Span 844.0 184.7 160.3 134.3 108.0 83.4 62.3 45.0 31.2 20.1 11.0 3.2

R= 7353.0 793.5 959.2 725.1 692.6 664.4 642.0 625.9 6158 610.9 610.0 612.1

Effort calculation according to the Guyon-Massonnet method

We ensure that the sum of the variable of interest of the complete deck (orange columns) is equal to the sum of the
variable of interest of the 11 beams.

3.2 Plane grid approach - Model Grill1

This model is constituted of longitudinal bars representing the rib + the hollow core element (whose characteristics are
defined in Appendix 1) and coplanar transversal bars of 2.50 m length.

Care is taken to apply the calculation method recommended by the SETRA (operating society for transport and
automobile repairs)/CEREMA (center for studies and expertise on risks, environment, mobility and development)
documents for the calculation of torsional inertias, in particular for the hollow core element, the torsional inertia is
be3/6 - ref [13], [14] and [15].

Model view in plane grid

ULS envelope of the moment (1.35G+1.35Qspan), edge beam

ULS envelope of the shear (1.35G+1.35Qspan), edge beam

3.3 Ladder-beam approach - Model Grill2

The model is similar to the previous one, but the longitudinal beam is subdivided into two bars. A lower one which
represents the rib alone (0.40*1.00 m²), an upper one which represents the hollow core element (0.90*0.25 m²). The
characteristics are given in the appendix.

The two beams are connected by very rigid bars (S=100 m² and I=100 m4) to ensure mechanical continuity.

Special care is taken to make duplicate bars and rigid bars non-weighty.

It is not possible to read the moments directly (see also section C.8 of the wiki).

We have to start from a duplicate (N,M) in the rib, for example, and recalculate the moment from a stress line
diagram).

Envelope (ULS N,M) (1.35G+1.35Q span), edge rib (beam 1)

Envelope (ULS N,M) (1.35G+1.35Qbay), second rib (beam n°2)

An advantage of this type of model is the possibility to define a construction phasing and elements with different
materials without going through equivalence coefficients, if any.

3.4 Bar + shell approach

3.4.1 Model with offset hollow-core element- model EF1

This model is quite comparable to the model presented in the previous §, but the grillage analysis that models the
hollow core element is replaced by shell-type finite elements.

3.4.2 Model with non-offset hollow-core element- model EF2

Reference is made to the "Guide pour l’évaluation structurale et la réparation des Viaducs à travées Indépendantes à
Poutres Préfabriquées précontraintes par post-tension (VIPP)" - ref [13]: one possibility is to let the hollow core element
joint with the neutral axis of longitudinal beams, whose cross section is composed of the hanging beams and the
hollow core element (which brings this model close to the first grillage analysis).

It is evident that this approach slightly overestimates the overall longitudinal stiffness but offers a simplification in the
results exploitation - The influence of this simplification remains to be verified on a case by case basis by comparing
Ibeam(*) + Ihollow core element to Ibeam - see references [12] and [13].

In this case:

Ihollow core element=0.90*0.253/12=0.00117 m4

Ibeam=0.09076 m4 (see Apendix 1)

Relative difference Ibeam(*) + Ihollow core element / Ibeam: +1.29%, which is very small.

(*) beam= hanging beam + affected hollow core element.

$translationBooks

PART 3 – EXAMPLES AND COMPLETE CASE STUDIES

Example of a beam grillage

calculation according to different
methods - Part 2
Example of a beam grillage calculation according to different
methods
Comparison of the results - multi-criteria analysis

Editor : Didier GUTH - Arcadis - June 14th, 2020

4) Results comparison
4.1 Support reactions

We calculate the "SLS" reactions, i.e., we cumulate G+Q (span 1 or 1+2). The unit cases are given in Appendix 1.

Comparison logic:

Column " %1 " = max of (Grill1, Grill2, EF1, EF2) / min (Grill1, Grill2, EF1, EF2) – 1
Column " %2 " =Guyon-Massonnet / Average (Grill1, Grill2, EF1, EF2) – 1

It should be noted that the edge of the deck has been deliberately loaded to amplify some phenomena.

The differences are small between the reactions of the models in bar grillage and bar grillage + shells,
On the other hand, between the Guyon-Massonnet approach and the grillage approach, the differences are
relatively big, up to 33%, on the first beams. The studies carried out at CEREMA (center for studies and
expertise on risks, environment, mobility and development) led to the same observations -ref [11].
There are two main reasons for these differences:
partly due to the fact that the effect of the 5 kN/m load at the edge of the corbel, and to a lesser
extent the weight of the corbel itself, is not correctly apprehended by the Guyon-Massonnet
calculation, which does not take into account the real effect of the load offset,

the most important fact is because the distribution coefficients are quite reliable for the distribution of
moments (ref [11]), but are quite erroneous to estimate the part of the loads close to the supports that
actually passes through a beam.

It is useful to refer to SETRA’s PRAD73 file which specifies the correction to be made:

Let X be the longitudinal abscissa of the load P, measured from the support axis and e the spacing between beams:

1. if x=0 the transverse distribution is done by assuming the hollow core element articulated on the beams
2. If x≥4e the cross-sectional distribution obeys the GUYON-MASSONNET hypotheses.
3. If x<0<4e the cross-sectional distribution is as follows:

the fraction 1-x4eP is distributed by assuming the hollow core element articulated on the beams
the fraction x4eP is distributed following the GUYON-MASSONNET model.

It should be noted that despite these differences, the automatic calculation software considered that distributed loads
of type A(l) (uniformly distributed load UDL nowadays) could be legitimately distributed with the coefficients of the
Guyon-Massonnet method.

Identifying the support nodes:

101 = beam 1 on abutment

111 = beam 1 central support

201 = beam 2 on abutment

211 = beam 2 on central support

Etc…

SLS Max

Support GM Grill1 Grill2 EF1 EF2 %1 %2

101 247.5 366.5 373.8 361.3 365.8 3% -33%

111 793.5 1023.0 1028.3 1019.1 1015.3 1% -22%

201 236.0 199.3 197.4 223.6 215.0 13% 13%

211 759.2 763.4 758.9 804.2 786.4 6% -2%

301 224.5 238.4 238.7 224.7 223.2 7% -3%

311 725.1 751.5 748.6 743.5 743.5 1% -3%

401 213.4 196.3 195.0 196.2 197.3 1% 9%

411 692.6 652.6 650.7 645.5 651.3 1% 7%

501 203.7 185.3 183.2 181.3 183.5 2% 11%

511 664.4 611.1 608.4 598.7 607.1 2% 10%

601 196.0 177.0 175.2 173.8 176.1 2% 12%

611 642.0 587.7 585.2 575.7 584.7 2% 10%

701 190.3 173.1 171.7 169.7 172.0 2% 11%

711 625.9 575.7 573.8 564.2 572.9 2% 9%

801 186.6 170.7 170.0 168.3 170.1 1% 10%

811 615.8 569.9 569.0 561.4 568.8 2% 9%

901 184.6 173.7 173.4 170.0 170.3 2% 7%

911 610.9 576.5 576.1 569.6 572.4 1% 6%

1001 183.9 165.3 167.8 171.0 166.0 3% 10%

1011 610.0 564.2 568.3 592.6 583.5 5% 6%

1101 184.1 210.6 212.3 215.4 215.7 2% -14%

1111 612.1 691.0 693.1 690.9 681.8 2% -11%

 max: max:

∑ 9602 9623 9619 9621 9622 13% 33%

G+Q support reactions

few differences between grill models

more significant differences between Guyon-Massonnet and grills, but they are reasonable, except for the
edge beam (33%).

4.2 Moments and shear

The moments and shear by Guyon Massonnet's method are calculated by assigning to these forces the same
distribution coefficient as before.

The Grid1 and EF2 models allow for the recovery of forces directly in the bars, while the Grid2 and EF1 models require
recalculation using a data conversion tool.

4.2.1 Guyon-Massonnet method

The forces in each beam calculated with the Guyon-Massonnet approach are reminded below (they are not weighted in
this table).

Guyon-Massonnet coefficients -> to divide by 11

Beam 1 2 3 4 5 6 7 8 9 10 11

Y= -4.5 -3.6 -2.7 -1.8 -0.9 0 0.9 1.8 2.7 3.6 4.5

Distributed
1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000
superstructure

Left cross beam 3.287 2.523 1.860 1.317 0.893 0.577 0.348 0.186 0.072 -0.013 -0.081

Right cross beam -0.081 -0.013 0.072 0.186 0.348 0.577 0.893 1.317 1.860 2.523 3.287

->∑ cross beam 3.206 2.510 1.932 1.503 1.241 1.154 1.241 1.503 1.932 2.510 3.206

Q span 2.407 2.089 1.751 1.407 1.087 0.812 0.586 0.406 0.262 0.144 0.042

Support 1.347 1.412 1.459 1.445 1.322 1.136 0.933 0.738 0.560 0.401 0.257

Beam 1 2 3 4 5 6 7 8 9 10 11

Moment on support

-1198. -1198. -1198. -1198. -1198. -1198. -1198. -1198. -1198. -1198. -1198.
Self-weight -1198.0
0 0 0 0 0 0 0 0 0 0 0

Superstructure -2320.0 -210.9 -210.9 -210.9 -210.9 -210.9 -210.9 -210.9 -210.9 -210.9 -210.9 -210.9

Cross beam -391.0 -114.0 -89.2 -68.7 -53.4 -44.1 -41.0 -44.1 -53.4 -68.7 -89.2 -114.0

Span -2109.0 -461.5 -400.5 -335.7 -269.8 -208.4 -155.7 -112.4 -77.8 -50.2 -27.6 -8.1

-18389. -1984. -1898. -1813. -1732. -1661. -1605. -1565. -1540. -1527. -1525. -1530.
M=
0 4 6 3 1 4 6 4 2 8 7 9

Moment Span 1

Self-weight 674.0 674.0 674.0 674.0 674.0 674.0 674.0 674.0 674.0 674.0 674.0 674.0

Superstructure 1305.0 118.6 118.6 118.6 118.6 118.6 118.6 118.6 118.6 118.6 118.6 118.6

Cross beam 220.0 64.1 50.2 38.6 30.1 24.8 23.1 24.8 30.1 38.6 50.2 64.1

Span 1615.0 353.4 306.7 257.1 206.6 159.6 119.2 86.0 59.6 38.5 21.1 6.2

M= 10774.0 1210.1 1149.5 1088.4 1029.3 977.0 934.9 903.5 882.3 869.7 864.0 862.9

Shear C0

Self-weight 144.0 144.0 144.0 144.0 144.0 144.0 144.0 144.0 144.0 144.0 144.0 144.0

Superstructure 278.0 25.3 25.3 25.3 25.3 25.3 25.3 25.3 25.3 25.3 25.3 25.3

Cross beam 47.0 13.7 10.7 8.3 6.4 5.3 4.9 5.3 6.4 8.3 10.7 13.7

Span 295.0 64.6 56.0 47.0 37.7 29.2 21.8 15.7 10.9 7.0 3.9 1.1

V= 2251.0 247.5 236.0 224.5 213.4 203.7 196.0 190.3 186.6 184.6 183.9 184.1

Shear P1, left

Self-weight -240.0 -240.0 -240.0 -240.0 -240.0 -240.0 -240.0 -240.0 -240.0 -240.0 -240.0 -240.0

Superstructure -464.0 -42.2 -42.2 -42.2 -42.2 -42.2 -42.2 -42.2 -42.2 -42.2 -42.2 -42.2

Cross beam -78.0 -22.7 -17.8 -13.7 -10.7 -8.8 -8.2 -8.8 -10.7 -13.7 -17.8 -22.7
 Span -422 -92.3 -80.1 -67.2 -54.0 -41.7 -31.2 -22.5 -15.6 -10.1 -5.5 -1.6

V= -3682.0 -397.3 -380.1 -363.1 -346.8 -332.7 -321.5 -313.5 -308.4 -305.9 -305.5 -306.5

4.2.2 Model Grill1

ULS transverse shear force:

ULS moment:

4.2.3 Model Grill2

ULS transverse shear force:

Transverse shear force hollow core element and rib, ULS:

If we load the cross beams, we end up with the same amount but the distribution is different.

When using meshes, do not forget to sum the vertical shear in the two bars of the ladder beam to obtain the shear in
the compound section.

N and M forces, ULS

Hence the evaluated torsor:

N= -1604.0 kN M=622.0 kN.M

Table 0.25 x 0.84 m2 (the participating width is calculated)

Hanging beam 0.40 x 1.00 m2

lower fiber upper fiber

σ(N) -4.01 MPa -4.01 MPa

σ(M) -9.33 MPa 9.33 MPa

σ(N+M) -13.44 MPa 5.32 MPa

it is deduced, for the rib:

Vupper= 0.285m and σupper, table= 9.99 MPa

Vlower= 0.715m

Constraints integration:

Calculated moment= 1645 kN.m

Torsor:

N= 2169.0 kN M=-1494.0 kN.M

Table 0.25 x 0.38 m2 (the participating width is calculated)

Hanging beam 0.40 x 1.00 m2

lower fiber upper fiber

σ(N) 5.42 MPa 5.42 MPa

σ(M) 22.41 MPa -22.41 MPa

σ(N+M) 27.83 MPa -16.99 MPa

it is deduced, for the rib:

Vupper= 0.379m and σupper, table= -28.19 MPa

Vlower= 0.621m

Constraints integration:

Calculated moment= -2872 kN.m

4.2.4 Model EF1

Span M ULS=601 and N ULS=-1783 kN

Support M ULS=-1177 and N ULS=3035 kN
Torsor:

N= -1783.0 kN M=601.0kN.M

Table 0.25 x 1.05 m2 (the participating width is calculated)

Hanging beam 0.40 x 1.00 m2

lower fiber upper fiber

σ(N) -4.46 MPa -4.46 MPa

σ(M) -9.02 MPa 9.02MPa

σ(N+M) -13.47 MPa 4.56 MPa

it is deduced, for the rib:

Vupper= 0.253m and σupper, table= 9.07 MPa

Vlower= 0.747m

Constraints integration:

Calculated moment= 1740 kN.m

(Ed: paradoxically, the width of the hollow core element is 1.05 while in theory we can only have 0.90 m…)

Torsor:

N= 3035.0 kN M=-1177.0 kN.M

Table 0.25 x 0.84 m2 (the participating width is calculated)

Hanging beam 0.40 x 1.00 m2

lower fiber upper fiber

σ(N) 7.59 MPa 7.59 MPa

σ(M) 17.66 MPa -17.66 MPa

σ(N+M) 25.24 MPa -10.07 MPa

it is deduced, for the rib:

Vupper= 0.715m and σupper, table= -18.90 MPa

Vlower= 0.285m

Constraints integration:

Calculated moment= -3112 kN.m

V=405 and 672 kN respectively.

4.2.5 Model EF2

We understand the interest of this modeling, since the efforts can be obtained by direct reading:

Mspan=1604 kN.m, Mmin=-2932 kN.m

 V=404 and 671 kN respectively

4.2.6 Summary

We note that the differences in moments are "relatively" small, both with the Guyon-Massonnet approach and the
other approaches.

For the shear as before, and logically, for the support reactions the differences are much more significant between the
Guyon-Massonnet method and the other models (from 15 to 20%) - for the edge beam and for a distributed load.

4.3 Load close to a support

We will study the impact of a load placed close to a support:

Global support reactions

It is immediately noticeable that the Guyon-Massonnet distribution coefficients remain close to a value of 1.40 on the
first 5 beams, i.e. the Guyon-Massonnet approach will assume that this load is distributed almost uniformly in these 5
beams (as a reminder, the implementation of the method assumes sinusoidal loads distributed over the entire length
of the beam, which is obviously not the case here).

Results (reminder, only load):

We note a difference of 80% on this reaction alone! Node 401 is the support of the 4th beam on the abutment of the
loaded span:
Table of support reactions

However, if we apply the method of PRAD Folder 73.

3. If x<0<4e the cross-sectional distribution is as follows:

the fraction 1-x4eP is distributed by assuming the hollow core element articulated on the beams
the fraction x4eP is distributed following the GUYON-MASSONNET model.

The load has been applied on the beam to simplify the calculations, so we will have:

R=(1 1.25/4/0.90)*200+1.25/4/0.90*1.445/11*200=140kN, value completely in accordance with the reactions of the

grillage calculation models.

If we are looking for relatively specific efforts on support devices, we cannot do without the PRAD 73 approach.

This complicates the calculations that we could try to automate, but remains essential.

4.4 Corbel load

This short chapter aims to show the differences that one could have between a Guyon-Massonnet, FE and manual
calculation.

A point load of 100 kN/m has been placed on the edge of the deck.

Guyon-Massonet-coefficient μ:
The calculation requires a Fourier series development and the taking into account of at least 3 harmonics. We end up
with: -23.3 kN.m/m:

Manual calculation: M=-100 kN/m*0.45 m= -45 kN.m/m

FE calculation

M≈30 kN.m/m

Conclusion:

There are significant differences between the 3 approaches. The manual approach remains safe, especially since it
does not take into account the beneficial "equalizer" effect of a cross beam, which exists in many cases.
$translationBooks

PART 3 – EXAMPLES AND COMPLETE CASE STUDIES

Example of a beam grillage

calculation according to different
methods - Part 3
Example of a beam grillage calculation according to different
methods
Comparison of the results - multi-criteria analysis

Editor : Didier GUTH - Arcadis - June 14th, 2020

5) Automated reinforcement calculation

5.1 (Automated) bending reinforcement

Section C8. of the wiki discusses an example of a ribbed floor - we repeat the same approach below.

Calculation on T-sections (the moments are taken from the previous §).

Calculation on isolated ribs (values that would be recovered from an automated calculation).

Reinforcement of the hollow core elements (we loaded at the right of the beams to avoid cumulating global
and local effects):
SUMMARY:

Again, as in the body of the Wiki (section C8), using the automatic reinforcement features on a ladder beam model,
separately on the upper element and on the lower element does not lead to a reinforcement that conforms to the
reinforced concrete design assumptions. It is therefore necessary to recalculate a global torsor to determine the
reinforcement (without forgetting the splice reinforcement).

Caution: the ladder beam model can lead to underestimating the shear on the last section if the hollow core element is
loaded. (see 4.3.2)

5.2 Reinforcement under local load (20 kN/m² over 1.25 x 0.90 m²

(Under this one isolated case)

FE approach global + local effects:

Myy, hollow core element span, smoothed=1.34 /0.625 m=2.14 kN.m/m (stretched lower fiber).
Myy, rib support: 0.47/0.625=0.75 kN.m/m (lower fiber also stretched).

Axis cut (unsmoothed):

ST1 approach, crossbar, global effects:

We recover the My moments in the concrete strips of the hollow core element:

Slightly enveloped approach: Mon ribs≈3.38 kN.m/2.5m=1.35 kN.m/m (stretched lower fiber).

Do not forget to take into account the stresses of fixed ended plates, as the stresses from the bar model only reflect
the overall effects:
Results of the calculation in a fixed ended plate - Calculation with 0.90 m spans, to remain homogeneous with the
global FE model calculation

Above Mplate on ribs=-1.29 kN.m/m, Mplate in span=0.64 kN.m/m

Total support≈1.35-1.29 ≈+0.06 kN.m/m (under this load case, in the middle axis)

(Nevertheless, it can be seen that the moment decreases very quickly at the loading end -> it would be possible to
calculate the reinforcements with the perfect fixed end moment which in this case represents few reinforcements
M=-1.29 kN.m/m -> As≈1.29/0.85/0.22/30=0.24 cm²/m).

Total hollow core element span = 1.35 global +0.64 local = 1.99 kN.m/m.

Guyon-Massonnet approach :

Myy=1.20 kN/m/m

In this case too, we have to add the forces of fixed ended plates, so we would have 1.20+0.64=1.84 kN.m/m.

Conclusion: the 3 approaches lead to values that do not show significant differences. The Guyon-Massonnet method
will be extremely fast, if an automated software is available.

For a local bending calculation, the bar model should be refined (the strips here are 2.50m wide for a load of 1.25 m
long).

6) Specific points
6.1 Skew

We will study the incidence of the skew in two configurations, 70 degrees which is assumed to be the limit for not
taking into account the skew, and 50 degrees. We start from the second FE model because it is very easy to modify.

The bar grillage requires a specific joint layout of the transverse strips since these, in spite of the skew, must remain
perpendicular to the ribs, as can be seen in this example:
© EGIS

6.1.1 Skew model with 70 degrees

6.1.2 Skew model with 50 degrees

6.1.3 Comparison

Moments and transverse shear in the most stressed beam:

1.35* [G+(Span 1, Span 2, Span 1+2)

"%2" is the maximum relative difference between Guyon-Massonnet/FE efforts (%=GM/max(FE)-1)

Supporting Reactions:

Reminder of the support nodes numbering for reactions on the next page.

And for reactions, G+Q (kN) - denomination "FEi", i is the skew in degrees:
6.2 Presence of cross girders

Considering that the efforts without cross girders are as follows:

6.2.1 Cross girders only on supports

The cross girders contribute to the redistribution of transverse shear forces on supports:

6.2.2 Intermediate cross girders

There too...

The cross girders make it possible to redistribute the forces between the beams more efficiently.

6.2.3 Guyon-Massonnet approach

With this method, there is no choice but to "spread" the stiffness of the cross girder - these values will be reduced to 1
linear meter, thus redivided by 8,333 m.

Impact on parameters α and θ:

Without cross girders With cross girders

α=0.4599 α=0.1244

θ=0.6422 θ=0.3203

With the cross girders, the distribution curves become almost rectilinear, we are getting closer to a calculation
according to the Courbon method.
This means that offset beams always take up more loads than central beams (see Courbon's formula).

We don’t redo the whole calculation process, but as a comparison, the coefficients in both cases:

Without cross girders

With cross girders

Graphic for cross beams (in orange, with cross girders):

 Graph for q load in span:

Graph for the 4th loaded beam ("Support" case):

6.3 The crossfall

In order to follow "reality", we could be tempted to model the structures with the crossfall. It should be noted that this
is of little interest and that, instead of stresses, vertical axis moments (Mz) in particular will appear and the question
arises as to whether or not to use these stresses in the dimensioning. Illustration...

6.3.1 Model Grill1

Model without crossfall

Grillage n°1 (plan), moment Mz under ULS envelope (-> Mz=0 everywhere)

Applying a crossfall

Grillage n°1, with a roof profile at 2.5 (mechanical characteristics unchanged for the beams).

Vertical axis moment on ribs

Associated transverse shear

Likewise, in the hollow core element

For horizontal axis moments in the edge beam:

and the vertical axis transverse shear:

Insignificant differences compared to the plan calculation.

For support reactions: we compared SLS reactions (G+Qtr 1 or G+(Tr1and2)) and the maximum difference is less than
4%).

It does not seem wise to model the crossfall as long as it remains reasonable to avoid having to manage "parasitic"
efforts.

6.3.2 Model Grill2 (ladder beam)

Finally, it can be noted that by the structure of the model grillage in ladder beam (with 3D frame), with or without
crossfall makes vertical axis moments appear due to the nature of the model.

Crossfall cross section

3D view

Moment Mz (vertical axis) in the longitudinal hollow core element bars

(An automatic calculation would lead to a reinforcement cross-section of ≈63.2/0.9/0.85/43.5=1.9 cm², but would not
correspond to any mechanical continuity logic of the hollow core element).

Moment Mz (vertical axis) in longitudinal hanging beam

An automatic calculation would lead to a reinforcement cross-section of ≈27.6/0.9/0.35/43.5=2.0 cm² on the vertical
side of the rib.

Moment Mz (vertical axis) on transverse straps

An automatic calculation would lead to a reinforcement section of ≈279.4/0.9/2.45/43.5=2.91 cm², on the vertical
face, but would not correspond to any logic of mechanical continuity of the hollow core element).

6.4 Small discussion about torsion

Charles Massonnet in [4] wrote: "It is acceptable to neglect the effect of torsion in the case of open-section steel girder
bridges without slabs. But it can never be neglected in monolithic girder bridges or in prestressed bridges (...)".

6.4.1 Torsional inertia calculation

The first question for the practitioner is how to calculate the torsional inertia. The PRAD 73 and VIPP67 documents,
references [13], [14], [15] give rules to be applied. In particular, remember that the inertia of the hollow core element
must be divided by 2 (be3/3 → be3/6) to remain compatible with the plate theory.

In addition, the PRP 75 guide recommended reducing this inertia by 10%.

Finally, it should be noted that this calculation of torsional inertias assumes that compressed concrete is a perfectly
elastic material; it may be prudent to reduce them by 10% to take into account the fact that concrete is not, in reality,
perfectly elastic.

6.4.2 What to do with all the calculated efforts?

Law of equilibrium

For the hollow-core element, a reinforcement can be calculated from a reinforced concrete sizing method
(Capra, Wood and Armer, ...) based on bending and torsional moments (see WIKI - part 1 – E,
For the rib, the situation is a little more complex...

The PRP 75 guide should be read in detail to understand the method to be applied. The method consists in taking a
quarter of the sum of the torsional moments (/lm), in algebraic value, arriving at the same node.

We apply it to the model Grill1

Torsional moment in the structure - longitudinal beams

Torsional moment in the structure – crossbeams

We perform the calculation according to the PRP 75 method for nodes 301 and 401.
Effort Mx

From the above sign convention, we apply the method of the PRP 75 guide and calculate Mxy (kN.m/m).

Efforts at the end of bar 301

Efforts at the end of bar 401

In the hollow core element, this Mxy effort would have to be taken into account via one of the known
methods: Capra, Wood and Armer, ...
In the beam, this moment would have to be taken into account to justify the straight section and the
reinforcement... Let us note that the VIPP and PRAD guides do not address this subject of torsion very much.
(to be confirmed)

Effect of a ULS torsion of 1.35*11.20 kN.m/m*0.90 m=13.6 kN.m (BAEL calculation).

For massive sections such as those in this example, the torsion is absolutely not dimensional.

Model EF2:

Torsional moments in the hollow core element

Torsional moments in the ribs

The forces are almost identical to those of the bar grillage. Again, the logic of the PRP 75 guide should be applied.

Guyon-Massonnet method

With a Guyon-Massonnet type calculation, we arrive to 8.2 kN.m/m and 8.8 kN.m/m, we remain in the rough estimate,
but... with the paradox that the Guyon-Massonnet method will give zero torsional moments on supports (respectively
maximal at mid-span), whereas this is precisely where they are maximal in a grid (respectively zero, NB: for this load
case).

7) Conclusion
This example showed (if the reader was not convinced) that the way of modeling will lead to different results.

We propose below some guidelines to guide the choice of modeling. The author's assessments.

[0] the need to automate calculations for a certain number of similar structures may influence the choice of modeling
and lead to the application of the Guyon-Massonnet method while the calculation of a single bridge may make this
method less profitable.

[1] subject to having automated the calculation of the coefficients of the Guyon-Massonnet method.

[2] only one wire beam to be modeled.

[3a] the calculation, per load case, of the part taken up by each beam can be tedious - a preliminary calculation of
influence surfaces may be necessary.

[3b] easier if the software knows how to make convoys run on a surface - a preliminary calculation of influence
surfaces may be necessary.
[4] may require significant work on spreadsheet unless the software (e.g. ST1) knows how to incorporate user
variables, variables assigned to each beam and each load case.

[4a] requires a global model in parallel or a good spreadsheet to manage concomitances.

[4b] on some software, it is complicated to have concomitant reactions for envelopes.

[5] relative precision and tedious to compute in the absence of prior programming of the decompositions into series
and coefficient ν and τ. Nevertheless, the rough estimate is quickly given in this case by the GM method.

[6] requires the calculation of additional fixed end slab forces because these models only provide the overall
transverse bending.

[7] the cumulation of general and local transverse bending gives an advantage, but does not allow one to easily
distinguish the part of each one. It seems permissible to keep a somewhat enveloping approach.

[8] See CEREMA's CHAMOA guide.

[9] the geometrical setting of the transverse strips (perpendicular to the ribs) requires a small geometrical study
beforehand.

[10] their stiffness is spread out and the recovery of the efforts in the cross girders seems complex from the.

[11] joint layout of the crossbars to be planned + study under local load to be planned - same logic as for the
transversal hollow core element.

There are several things to think about before embarking on the model:

What degree of accuracy is expected?

Will it be necessary to justify all sections or just check a few sections?
Is it necessary to change the calculation method if we check a design from the 1960s, for example?

It should be noted that the different models will lead to results with differences of up to a few tens of % on the
calculated efforts. It is interesting to know that when performing a counter-calculation ... it is impossible to obtain a
perfect convergence of the results.

On the other hand, it is important to understand the reasons for the discrepancies. This document hopes to contribute
to that understanding.

Finally, it should be noted that none of the models will be able to automate 100% of the justifications and
reinforcement.

Anything else?

8) Appendix - Guyon-Massonnet calculation data

Pi, bar model - Rib features:

Potential abscissa of charges:

We calculate the distribution coefficients using a software that implements the Guyon-Massonnet method.
 Fictitious isostatic calculation span: 22.867 m
Bending and torsional inertias of the main beam (calculated according to the method recommended in
documents [13] to [15]).

Main ribs: Hollow core element:

Detail for the rib: the characteristics are the same as for the FE software, nothing to report.

Sx=0.6250 m², Iy=0.09076, Iz=(0.90^3*0.25+0.40^3*1.00)/12=0.0205 m4

Ix= ?

K(hollow core element) ? 0.9/0.25=3.6 -> kcalc=0.2722 to be divided by 2 for the hollow core element k=0.1361,

K(rib)=k(2*1/0.4)=k(5.0)=0.292

Ix=0.1361*0.90*0.25^3+0.292*1.00*0.4^3=0.02060191 m4 = 2,060,191 cm4, not much different from the bar

software: 2,238,759 cm4 (previous page).

Flexion and torsion inertia of the hollow core element:

NB: b=25 m to force a calculation with 1/6 be3 (does not pose a problem, since the method spreads out the inertias
and reduces them to 1 lm of hollow core element).

Reduced to 2.50 m for the grillage calculation (1/10 of the above value).

Sx=0.625 m², Iy=0.003255 m4 , Iz=2.50^3*0.25/12=0.3256 m4, Ix=0.00651 m4

9) Model ST1
Plan model features:

Main ribs: Hollow core element:

We reduce the hollow core element to b=2.50 m for the grillage calculation (1/10 of the above value).

Sx=0.625 m², Iy=0.003255 m4 , Iz=2.50^3*0.25/12=0.3256 m4 et Ix=0.00651 m4

Characteristics for the ladder beam: ribs and longitudinal hollow core element

Rib Superstructure hollow core element

Loading
NB: do not forget not to activate the self weight of the straps, otherwise it will be counted twice.

10) Appendix - Bars + Shells model

10.1 3D Model

Hanging beams features

Hollow core element features

10.2 2D Model

11) Appendix - Results tables

Support reactions

Beam 1: nodes 101, 111, 121

Beam 2: nodes 201, 211, 221, etc…

Little difference, but we can see that the models with shells overload the edge beams (c effect).

Superstructures:

Large difference between the Guyon Massonnet approach and all the approaches due to offset loads in the corbel, in
particular.
Span loading -span 1:

Span loading – span 1+2

12) Appendix - ST1 files
Text files can be downloaded from the website.
$translationBooks

PART 3 – EXAMPLES AND COMPLETE CASE STUDIES

Practical application of smoothing

Practical application of smoothing
Example of the Br wheel in Handbook 61, title II
Finite Elements Working Group

Editor: D. Guth, with the support of V. Bruno and F. Dubois.

Introduction
1.1 Purpose

The purpose of this document is, on the basis of an extremely simple example, to show the differences that can be
obtained according to the modeling which is chosen, depending on a certain number of parameters such as: mesh
size, modeling of the load, extent of smoothing, and on several software.

It is used as an example in §D.5.2.1 of the Wiki (https://wiki-gtef.frama.wiki/accueil-gtef:partie-2:chapitre-d:d5 ).

It does not claim to exhaustively represent the scenarios that the modeler may encounter, however it is easily
replicable by everyone on their favorite software and can be used as a benchmark.

It aims at illustrating the common problems that engineers encounter in their daily work: how to model loads (point or
distributed?), over which length to smooth (2h, 4h, other?), where to read the values (at the nodes, at the mesh
centers?), what shape for the elements? (triangular? quadrilateral?), is the smoothing (or "averaging") performed by
the software correct? etc.

If you want to share your results coming from other software than Robot (R), Sofistik (S) or Pythagoras (P), the right
address is [email protected] .

1.2 Study configuration

We are going to calculate the forces resulting from the impact of a Br wheel (Handbook 61 title II), centered on a one-
way joist slab bridge.

The calculations were carried out by SETRA with a Poisson's ratio of 0.15 for the concrete.

The guide does not specify whether the guide values have been smoothed or not.

→ We calculate the effect of a Br wheel, with a diffusion on E=10 cm.

The impact of 60cm (transverse) x 30 cm (direction of traffic) is, therefore, to be modeled as diffused on an impact of
80 cm x 50 cm².

The impact force is 10 tons, which we take equal to 100 kN, i.e. a pressure of 250 kPa.

The modeled slab has dimensions of b=40 m (b≈∞) and a=6 m:

The size of the meshes is 25x25 cm², where the thickness h of the slab is modeled as squared.

In addition, a calculation with 100x100 cm² meshes is also performed to evaluate the impact of increasing the mesh
size on the results.

1.3 Studied cases

We study, with 3 software (R, S and P), in whole or in a part, the following configurations:

25x25cm² mesh / distributed load / point load - quadrilateral elements (potentially triangular);
100x100cm² mesh / distributed load / point load - quadrilateral elements.

Synthesis and conclusion

All graphic outputs are provided in the Appendix.

We compare the calculated values to the BT1 values that serve as a reference point.

Moments in kN.m/m

Mxx Difference/BT1 Myy Difference/BT1

BT1 SETRA 23.5 28.7 Appendix 7

Distributed load - mesh 25x25

Model R1, peak values 24.2 2.7% 29.0 1.0% Appendix 1

Model R2, peak values 23.2 -1.3% 28.5 -0.6% Appendix 3

Model R1, 25x25, manual smoothing for 2

hours 23.3 -0.7% 28.0 -2.4% Appendix 1

Model R1, 25x25, automated smoothing

23.4 -0.3% 28.1 -2.3% Appendix 2
for 2 hours

Model R2, 25x25, automated smoothing

22.1 -6.2% 27.5 -4.5% Appendix 3
for 2 hours

Software S, quadrilateral-25x25,
23.6 0.4% 28.6 -0.3% Appendix 5
smoothing over 2 hours

Software S, triangular-25x25 21.0 -11.9% 26.9 -6.7% Appendix 5

Software P, quadrilateral-25x25 21.6 -9.0% 27.4 -4.7% Appendix 6

Model R1, 25x25, manual smoothing for 4

21.9 -7.4% 26.7 -7.5% Appendix 1
hours

Model R1, 25x25, automated smoothing

22.0 -7.1% 26.7 -7.4% Appendix 2
for 4 hours

Model R2, 25x25, automated smoothing

20.9 -12.4% 26.2 -9.8% Appendix 3
for 4 hours

Software S, quadrilateral-25x25,
22.8 -3.3% 27.8 -3.2% Appendix 5
smoothing for 4 hours

Distributed load – mesh 100x100

Model R1, 100x100, automated

21.9 -7.3% 27.9 -2.7% Appendix 2
smoothing for 2 hours

Model R1, 100x100, automated

20.5 -14.5% 26.6 -8.0% Appendix 2
smoothing for 4 hours

Software P, 100x100 13.5 -74.3% 19.7 -45.8% Appendix 6

 Point load – mesh 25x25

Model R1 25x25 37.1 36.7% 41.8 31.3% Appendix 2

Model R1 25x25 28.1 16.3% 34.9 17.6% Appendix 2

Model S 45.7 48.6% 52.6 45.4% Appendix 5

Model P 28.1 16.4% 34.8 17.6% Appendix 6

Conclusion:

From these calculations, we can already extract the following conclusions:

It is important to understand what the software does in terms of display and smoothing. Of course, we can
always size with the worst case scenario, but even if we could justify it, this approach would not be
economical. You should not hesitate to insist to the editor to ensure that the software does not remain a black
box for the modeler;
Avoid point forces, they are too unfavorable and, moreover, have no physical meaning. If describing impact
surfaces for rolling loads is complex, for example, the single force must at least be transformed into several
point forces;
Given the differences observed, it is useless to provide results with 8 decimal places, 1 is enough;
This example illustrates why, during a verification, an external or outside control of the studies, it is necessary
to specify in the hypothesis/modeling note the mesh size, the type of element and the chosen smoothing
logic. It is also possible to set the tolerance which will be accepted to consider the results as "correct". It is
indeed unlikely that two software will provide the same results; the software must be able to converge with a
maximum difference of 5% (given it is agreed on at the beginning of the study and in the case of a linear
calculation).
As indicated in the Wiki, the smoothing logic remains subject to the engineer's judgment, however the
thickness of the slab and the load impact area must guide the choice.

Once the force is distributed over an impact area defined at the mean slab layer and the mesh size is in a reasonable
ratio relative to the slab’s thickness and the dimensions of the load impact rectangle, then the peak value for
reinforcement may be selected, where the value smoothed over 2 hours does not lead to too important deviations -
which one can see in the table above.

It should be remembered that rolling loads will not stress the entire cross-sectional area at the same time, because
they move by nature. The choice between a 2h or 4h smoothing must also be made on the basis of a reflection on the
transversal redistributive capacity and/or the incidence of increased constraints by a few %; the consequences are not
necessarily the same between a calculation in SLS characteristics and a calculation of crack opening or fatigue;

If we consider that the SETRA values are correct values (although we have no indication about a possible
smoothing), we realize that the approach consisting in smoothing for 2 hours seems to be the most valid;
The triangular elements give worse values for the same mesh size, which confirms what the Wiki says in Part
2 - A2;

In any case, it is the responsibility of the engineer to get an idea of the impact of the mesh size or of how to model the
loads, through tests on simplified parts of models and evaluating the sensibility of such tests.

In summary:
3) Appendix 1 - Model R1 - manual smoothing
This model has a node under the center of the impact rectangle as opposed to Model 2 which was built to have the
center of a 25x25 mesh under the center of the impact rectangle - See Appendix 3.

This § calculates the smoothed forces, manually, as opposed to Appendix 2. We will see that the differences between
the two approaches are minimal.

Load impact area with mesh size

Moment My (transversal)

Values taken every ≈12.5 cm at the central axis of the one-way joist slab:
 Average value on 2 meshes: 28.04 kN.m/m
Average value on 4 meshes: 26.7 kN.m/m
Moment Mx (longitudinal)

Values taken every ≈12.5 cm at the transverse axis:

Average value on 2 meshes: 23.3 kN.m/m

Average value on 4 meshes: 21.9 kN.m/m
Smoothing detail (from the integral):

We read with 2 meshes = 0.50 cm=2h and with 4 meshes=1 m=4h.

Reminder: effect of a load shift of 0.125 m in X (longitudinal): changes practically nothing, see below.
My, smoothed over 1.00 m -> My=26. kN.m/m

My, smoothed over 0.50 m ->My=28.0 kN.m/m

These values are almost identical to the case without shift.

4) Appendix 2 - Using the smoothing functions of a software - Model R1

Many tests are carried out to judge the sensibility of the mesh size.

A calculation with a point force is also computed.

4.1 Use of automated smoothing - distributed load - mesh size 25x25 cm².

On 0.50 m: My=14.03/0.50=28.06 kN.m/m

With a mesh refinement: My=13.96/0.50m=27.92≈28.06 kN.m/m

For 1 m, My=26.73 kN.m/m

With a mesh refinement: My=26.57 kN.m/m ≈26.73 kN.m/m

Mx, smoothed on 0.50 m=11.71/0.50=23.42 kN.m/m

Mx, smoothed on 1.00 m =21.95 kN.m/m

4.2 Use of automated smoothing - distributed load - mesh size 100x100 cm².

My=13.97/0.5=27.94 kN.m/m

My=26.57 kN.m/m

Mx=10.95/0.5=21.90 kN.m/m

Mx=20.53 kN.m/m
4.3 Smoothing with point force

Whether you smooth on 1.00 m or 50 cm, the smoothed value remains excessive.

My=20.9/0.50 m= 41.80 kN.m/m

My=34.85 kN.m/m, on 1 m.

Mx=18.56/0.5=37.12 kN.m/m

Mx=28.09 kN.m/m
4.3 Reminder: load distribution over several point loads

5) Appendix 3 - Model R2 - Diffused load centered mesh

(With centered mesh under the impact i.e. the center of a mesh is under the center of a load)
Moment My
Moment Mx
6) Appendix 4 - FE calculation data - models R1 and R2
Materials v=0.15

Panel Thickness Materials Mesh Reinforcement NU

Sum of reactions:

Node/Case FX [kN] FY [kN] FZ [kN] MX [kNm] MY [kNm] MZ [kNm]

Case 101 Br

Total sum 0.0 0.0 0.0 0.0 0.0 0.0

Sum of reactions 0.0 0.0 100.00 550.00 -1000.00 0.0

Sum of efforts 0.0 0.0 -100.00 -550.00 1000.00 0.0

Verification 0.0 0.0 -0.0 -0.0 0.0 0.0

Precision 6.43486e-13 4.51850e-24

Case 102 Br displaced

Total sum 0.0 0.0 0.0 0.0 0.0 0.0

Sum of reactions 0.0 0.0 100.00 550.00 -1012.50 0.0

Sum of efforts 0.0 0.0 -100.00 -550.00 1012.50 0.0

Verification 0.0 0.0 -0.0 -0.0 0.0 0.0

Precision 7.68058e-13 4.56828e-24

Case 201 Point

Total sum 0.0 0.0 0.0 0.0 0.0 0.0

Sum of reactions 0.0 0.0 100.00 550.00 -1000.00 0.0

Sum of efforts 0.0 0.0 -100.00 -550.00 1000.00 0.0

Verification 0.0 0.0 -0.0 -0.0 0.0 0.0

Precision 1.03246e-13 4.60955e-24

Case 202 Point 6 forces

Total sum 0.0 0.0 0.0 0.0 0.0 0.0

Sum of reactions 0.0 0.0 100.00 550.00 -1012.50 0.0

Sum of efforts 0.0 0.0 -100.00 -550.00 1012.50 0.0

Verification 0.0 0.0 -0.0 -0.0 0.0 0.0

Precision 4.12474e-13 4.63262e-24

7 Appendix 5 - Sofistik Model
7.1 Distributed load - mesh ≈25 x 25 cm².

(Curve generated from nodal values - method recommended by the editor).

My=28.6 kN.m/m

Displayed on a grid with a mesh size of 25x25 cm:

Smoothing over 1 m would lead to (28.6+27.0)/2=27.8 kN.m/m

The same calculation with triangular elements: My=26.9 kN.m/m

 Mx=23.6 kN.m/m

Values displayed in 25x25 cm² meshes

Smoothing over 1 m would lead to (23.6+21.9)/2=22.8 kN.m/m

Same calculation with triangular elements: Mx=21 kN.m/m

Model view with triangular meshes

7.2 Point force- mesh ≈25 x 25 cm².

My=52.6 kN.m/m
 Mx=45.7 kN.m/m

Data:

Loading (wheel impact):

Impact rectangle 80x50 with 250 Kpa load

Materials (Poisson’s ratio for concrete)

8) Appendix 6 – Pythagore software
Material and load data:

Poisson’s ratio for slab

Distributed load

Point load

Note on the smoothing performed by the software:

The node smoothing performed by the Pythagore software consists of an average, at a given node, of the results
obtained for this node in the four finite elements having this node in common.

As a result, for the node located exactly in the middle of the plate (and load), these four values are symmetrically
identical, so the smoothed and unsmoothed maximum values coincide, as shown in the following maps.

8.1 Distributed load - mesh size = 25 x 25 cm².

8.1.1 Without smoothing

My = 27.41 kN.m/m

Mx = 21.55 kN.m/m

8.1.2 With smoothing at the nodes

My = 27.41 kN.m /m (no change in peak value)

Mx = 21.55 kN.m/m (no change in peak value due to load symmetry)

8.2 Concentrated load - mesh size = 25 x 25 cm².

8.2.1 Without smoothing

My = 34.83 kN.m/m

Mx = 28.11 kN.m/m

8.2.2 With smoothing

Same peaks as without smoothing

8.3 Distributed load - mesh size = 100 x 100 cm2

8.3.1 Without smoothing

My = 19.69 kN.m/m
Mx = 13.48 kN.m/m

8.3.2 With smoothing

My = 19.69 kN.m/m (no change in peak value due to load symmetry)

Mx = 13.48 kN.m/m (no change in peak value)

8.4 Concentrated load - mesh size = 100 x 100 cm2

8.4.1 Without smoothing

Same peaks as with smoothing (see below).

8.4.2 With smoothing at the nodes

My = 21.71 kN.m/m

Mx = 15.21 kN.m/m

8.5 Distributed load - automatic meshing ≈ 25 x 25 cm2

8.5.1 Without smoothing

My = 27.54 kN.m/m (almost the same result as with the square mesh)

Mx = 22.45 kN.m/m

8.5.2 With smoothing at the nodes

Same peak values as without smoothing.

9) Appendix 7 – SETRA abacus (operating society for transport and

automobile repairs) - Technical Bulletin n°1
Ma (transverse)=2870 t.m.

Mb=2350 t.m
10) Appendix 8 – Pücher abacus
M=Average value read/8/π*100 kN - this method includes a "certain" margin of error given how each person reads the
contours.

Mxx=Mx+νMy et Myy=My+νMx

Numerical application:

We have Mx≈24.3 kN.m/m My≈19.9 kN.m/m

Mxx=24.3+0.15x19.9=27.3 kN.m/m
Myy=19.9+0.15x24.3=23.5 kN.m/m

Transverse moment (according to x)

Longitudinal moment (according to y)

BIBLIOGRAPHY
$translationBooks

BIBLIOGRAPHY

Bibliography

Bibliography
General bibliographic references
1) General Papers

BATOZ, Jean-Louis DHATT, Gouri. Modélisation des structures par éléments finis (volume 1 : solides
élastiques). Hermès, 1990.
BATOZ, Jean-Louis DHATT, Gouri. Modélisation des structures par éléments finis (volume 2 : poutres et
plaques). Hermès, 1990.
BATOZ, Jean-Louis DHATT, Gouri. Modélisation des structures par éléments finis (volume 3 : coques). Hermès,
1990.
BONNET, Marc FRANGI, Attilio. Analyse des solides déformables par la méthode des éléments finis. Éditions
de l'École Polytechnique, 2006, 300 p.
BRULE, Stéphane CUIRA, Fahd. Pratique de l'Interaction Sol Structure sous Séisme. Éditions , 2006, 300 p.
ZIENKEIWICZ, O. C. La méthode des éléments finis. Mc Graw Hill, Paris, 1979.
Collectif. Théorie de la fiabilité - Application à l’évaluation structurale des ouvrages d’art. Rapport d’études.
Sétra, février 2012.
Collectif. NF EN 1990. Eurocode 0 - Bases de calcul des structures. Afnor, mars 2003.

2) Pedagogical works or educational books

CHATEAUNEUF, Alaa. Comprendre les éléments finis. Structures : principes, formulations et exercices
corrigés. Ellipses-Marketing, 2005, 272 p.
CRAVEUR, Jean-Charles. Modélisation par éléments finis. Dunod, 3ème édition, 2008, 328 p.
DEBARD, Yves. RDM 6 : mémorisation des matrices globales, méthodes de calculs. IUT du Mans. Link to the
document.

3) The quality of the EF solution - Global Error Estimator

BABUSKA, I. RHEINBOLDT, W. Error estimates for adaptative finite element computations. SIAM Journal of
Numerical Analysis, 1978, 15(4):736–754.
ZIENKIEWICZ, O.C. ZHU, J.Z. The superconvergent patch recovery and a posteriori error estimates - Part 1 :
the recovery technique. Int. Journal for Num. Met. in Eng., 1992, vol 33, 1331-1364.
ZIENKIEWICZ, O.C. ZHU, J.Z. The superconvergent patch recovery and a posteriori error estimates - Part 2 :
error estimates and adaptivity. Int. Journal for Num. Met. in Eng., 1992, vol 33, 1365-1382.
DELMAS, J. Estimateur d’erreur de ZHU-ZIENKIEWICZ. [R4.10.01] Documentation CODE-ASTER, 2012.
DELMAS, J. Estimateur d’erreur en résidu. [R4.10.02] Documentation CODE-ASTER, 2012.
LADEVEZE, P. PELLE, J. La maîtrise du calcul en mécanique linéaire et non-linéaire. Lavoisier, 2001.

4) The quality of a quantity of interest obtained through FEM (Finite Element Method) - Local Error
Estimator.

BECKER, R. RANNACHER, R. An optimal control approach to shape a posteriori error estimation in finite
element methods. Acta Numerica, 10, 2001, pp. 1-120.
ODEN, J. PRUDHOMME, S. Goal-oriented error estimation and adaptivity for the finite element method.
Computer and Mathematics with Applications, 2001, 41:735–756.
DELMAS, J. Estimateurs d’erreurs en quantité d’intérêt. [R4.10.06] Documentation CODE-ASTER, 2013.

Bibliographic references on dynamics

1) General Papers

CLOUGH, R.W. PENZIEN, J. Dynamics of structures. Computers & Structures, 2003.

2) Pedagogical works or educational books

EUVRARD, D. Résolution numérique des équations aux dérivées partielles de la physique, de la mécanique et
des sciences de l'ingénieur : Différences finies, éléments finis, problèmes en domaines non bornés. Dunod,
1994.

3) Pedagogical works or educational books

BETBEDER-MATIBET, J. Génie parasismique Tome 1, 2 & 3. Hermès, 2003.

CHOPRA, A.K. Dynamics of structures, Theory and Applications to Earthquake Engineering. Prentice Hall,
1995.

Specific bibliographic references in civil engineering

PRAT, M. BISCH, Ph. MESTAT, Ph. MILLARD, A. PIJAUDIER-CABOT, G. La modélisation des ouvrages. Hermès,
1995, 770 p.
PRAT, M. BISCH, Ph. MESTAT, Ph. MILLARD, A. PIJAUDIER-CABOT, G. Calcul des ouvrages généraux de
construction. Hermès, 1995, 767 p.
HASHIN, Z. The elastic moduli of heterogeneous materials. 1962.
de LARRARD, F. Structures granulaires et formulation des bétons (2000). Etudes et Rech. Des LPC, n°OA34
GIMSING, N.J. Cable supported bridges (1983), Wiley & Sons
ACKER, P. EYMARD, R. Fluage du béton : un modèle plus performant et plus simple à introduire dans les
calculs numériques. Annales de l’ITBTP, n°507, série Béton 295, 1992.
ARGIN, M. Stress-strain relationships for concrete and the analysis of structural concrete sections. Solid
Mechanics Division, University of Waterloo, 1971.
GOTO, Y. Cracks formed in concrete around deformed tension bars. ACI J. Proc., n°68 (4e série), 1971, pp.
244-251.
BARON, J. SAUTEREY, R. Les bétons hydrauliques, Presse des Ponts et Chaussées, 2003.
HILL, R. Mathematical theory of Plasticity Cambridge University Press, 1970.
SOMJA, H. Contribution à la modélisation du comportement des ouvrages d’art lors des phases de
construction avec prise en compte des non-linéarités géométriques et matérielles. Thèse de doctorat,
Université de Liège, 2004.
GIRY, C. DUFOUR, F. MAZARS, J. Stress-based Non-local Damage Model. International Journal of Solids and
Structures - INT J SOLIDS STRUCT. 48. 3431-3443. 10.1016/j.ijsolstr.2011.08.012, année 2011.
LORENTZ E. A nonlocal damage model for plain concrete consistent with cohesive fracture. International
Journal of Fracture. 207. 10.1007/s10704-017-0225-z, année 2017.
PEERLINGS, R.H.J. de BORST, R. BREKELMANS, W.A.M. de VREE, J.H.P. Gradient-enhanced damage for quasi-
brittle materials. Int. J. Numer. Methods Engrg, 1996, 39: 3391-3403
PIJAUDIER-CABOT, G BAŽANT, P.Z. Nonlocal Damage Theory. Journal of Engineering Mechanics-asce - J ENG
MECH-ASCE. 113. 10.1061/(ASCE)0733-9399(1987)113:10(1512), année 1987.
Collectif. Practitioner's guide to finite element modelling of reinforced concrete structures. Bulletin n°45, fib,
2008.
VIRLOGEUX, M. Tribune libre. BOA n°14, SETRA, mars 1993. Link to the BOA (sujet bielles-tirants).
KRETZ, Th. & al. Etude des montants des cadres d'entretoisement. Bulletin des Ponts Métalliques n°18, OTUA,
1996.

Reference related to metal fatigue calculations

HOBBACHER, A. F. Recommendations for Fatigue Design of Welded Joints and Components. Springer - 2018.

Bibliographic references on post treatment

1) General Papers

PERCHAT J. Traité de béton armé selon l’Eurocode 2. Editions Le Moniteur, 2017.

2) Reference related to the calculation of reinforcement in reinforced concrete slabs.

Collectif. Algorithme de calcul des densités de ferraillage. Doc. R7.04.05, document de référence de code
aster, May 4th, 2018.

3) Reference related to the calculation of reinforced concrete reinforcements using automated strut-and-
tie method.

MENDOZA-CHAVEZ G. et al.. Discrete sequential optimization for strut-and-tie design. Transactions, SMiRT-25
Charlotte, NC, USA, Division VI, August 4 -9, 2019.

Specific bibliographic references for geotechnical calculations

BATHE K.J. Finite element procedures. Prentice Hall Inc, 1996, 1037 p.
BONNET M. Boundary Integral Equations for Solids and Fluids. Wiley and Sons, Chichester, UK, 1995.
CLOUGH R.W., PENZIEN J. Dynamics of structures. Mc Graw-Hill, 1993, 738 p.
KRAMER S.L. Geotechnical Earthquake Engineering. Prentice Hall Inc, 1996, 653 p.
MESTAT P., PRAT M. Ouvrages en interaction, volume 2 : la maîtrise de la modélisation des ouvrages. Hermès,
1999, 944 p.
LEHMANN, L. Wave propagation in infinite domains. Springer Verlag, 2007, 182 p.
PECKER, A. Dynamique des sols. Presses de l’Ecole Nationale des Ponts et Chaussées, 1984, 284 p.
 POTTS D.M., ZDRAVKOVIC L. Accounting for partial material factors in numerical analysis. Géotechnique, 62,
2012, pp. 1053–1065.
SEMBLAT J.F., PECKER? A. Waves and Vibrations in Soils: Earthquakes, Traffic, Shocks, Construction Works
IUSS Press (Instituto Universitario di Studi Superiori di Pavia), 2009, 493 p.
TSCHUCHNIGG F., SCHWEIGER H.F., SLOAN S.W., LYAMIN A.V., RAISSAKIS I. Comparison of finite-element limit
analysis and strength reduction techniques. Géotechnique, 65, 2015, pp. 249–257.
WOLF, JP. Dynamic Soil-Structure Interaction. Prentice Hall Inc., 1988.
WOLF, JP. Soil-Structure Interaction Analysis in Time Domain. Prentice Hall Inc., 1988.
YOSIDA, N. Seismic ground response analysis. Springer Verlag, 2015, 370 p.
Collectif. Guide « Ponts courants en zone sismique ». Sétra - SNCF - 199x
BETBEDER-MATIBET, J. Génie parasismique – volume 3. Hermès Science Publications, 2003.
Collectif. Règles de construction parasismiques – Dimensionnement des ancrages en zone sismique. FD P06-
029. Afnor. Décembre 2017.

Internet links to useful documentation:

PACOSTE, C. PLOS, M. JOHANSSON, M. Recommendations for finite element analysis for the design of
reinforced concrete slabs. Chambers, Stockholm, Sweden, 2012.

http://publications.lib.chalmers.se/records/fulltext/176734/local_176734.pdf

Website of ECAM: Lignes d'influences

http://clb.perso.univ-pau.fr/rdm/isa2/Codes/HTML/ligne_influence/ligne_influence.html

Website of ENS Cachan: https://eduscol.education.fr/sti/si-ens-paris-saclay/ressources_pedagogiques/dossier-

elements-finis

Website métalétech du CTICM: Lien: https://metaletech.com/

FOLLOW-UP OF ADDS-UP AND
MODIFICATIONS
$translationBooks

FOLLOW-UP OF ADDS-UP AND MODIFICATIONS

Additions and Modifications

Additions and Modifications
2020/07/12: Example E - Calculation of transverse bending of a BP (presumably "beam and plate") box - JP Deveaud.

2020/07/27: Example E, addition of the dimensioned cross-section.