Piled Embankments With Geosynthetic Reinforcement: Numerical Analysis of Scale Model Tests

Piled Embankments with Geosynthetic
Reinforcement
Numerical Analysis of Scale Model Tests
Author: T.J.M. den Boogert

Student number: 1323407
Date: 21 February 2011

Place: Delft
Piled Embankments with Geosynthetic
Reinforcement
Numerical Analysis of Scale Model Tests
Student
Author: T.J.M. den Boogert (1323407)
Address: Loosduinsekade 2B
2571 BM Den Haag
Tel: 06-48833384
Email: [email protected]
Deltares
Address: Stieltjesweg 2
2600 MH Delft
Tel: 088-3357200
Document: MSc Thesis
Status: Final TU Delft Civil Engineering and Geoscience
Address: Stevinweg 1
Date: 21 February 2011 2628 CN Delft
Place: Delft Tel: 015-2785440
21 February 2011
Examination Committee
Prof. ir. A.F. van Tol
Section Geo-Engineering, Civil Engineering and Geoscience, TU Delft
Tel: 015 2782092
Ing. H.J. Everts

Section Geo-Engineering, Civil Engineering and Geoscience, TU Delft
Tel: 015 2785478
Ir. S.J.M. van Eekelen

Section Geo-engineering, Infra Engineering, Deltares
Tel: 088 3357287
Ing. P.G. van Duijnen

Movares
Tel: 030 2653669
Piled embankment with geosynthetic reinforcement – Master of Science thesis I

21 February 2011
Preface
During the past year I have been working on my graduation project. This graduation project
is part of the Master Civil Engineering at Technical University Delft. Within Civil Engineering
I have chosen Geo-Engineering as my specialization. Geo-Engineering covers all aspects
of the utilisation of earth. This includes design of underground constructions, foundations
and soil improvement techniques.
Piled embankment with geosynthetic reinforcement is a soil improvement technique. This

technique has not been applied a lot in the Netherlands. To understand the load distribution
in the piled embankments scale tests have been performed. This thesis analyses the
numerical behaviour of the scale tests. Numerical analysis has been performed with Plaxis
3D Tunnel.
First of all, I would like to thank my examination committee: prof. ir. A.F. van Tol, ing. H.J.
Everts, ir. S.J.M. van Eekelen and ing. P.G. van Duijnen, for their ideas, discussions,
remarks, mental support and reviews of the chapters.
I also would like to thank Deltares for providing a workspace and facilities to perform this
graduation project. The colleagues at Deltares, especially ir. J.A.M. Teunissen, thank you
for your input.
Also an acknowledgement goes to Plaxis BV for providing the Plaxis 3D Tunnel software.
Last but not least, I would like to thank Richard, my family, my family in-law and friends for
their interest and support.
Enjoy reading,
Theresa den Boogert
Piled embankment with geosynthetic reinforcement – Master of Science thesis III

21 February 2011
Summary
Highways and railways are constructed on embankments. Constructed on soft soil these
embankments cause large settlement. To avoid large residual settlement, soil improvement
techniques are needed. Piled embankment with geosynthetic reinforcement is such a soil
improvement technique. Piles and geosynthetic reinforcement (GR) are applied underneath
the embankment. This technique is highly settlement reductive or even settlement free.
Besides this, the construction time is short and the technique can be applied nearby
sensitive structures.
The load on the embankment is transferred to the piles by arching. Due to the phenomenon
arching, a large part of the load is transferred directly to the piles (load part A). The load
underneath the arch (load part B) is transported via the geosynthetics to the pile. The last
part (load part C) is transported to the soft subsoil between the piles.
In the last few years, the CUR committee 159B has been working on the new Dutch Design
Guideline for the design of piled embankments. To validate the guideline several field tests
have been performed. From the field measurements is concluded that the design method is
very conservative. Improving the design guideline will reduce the construction costs. To
understand the physical behaviour of the piled embankment and to validate the theory,
experimental scale tests have been performed. In these tests, the type of geosynthetics,
number of geosynthetic layers and fill material were varied. During these tests the load
distribution, deformation and strains were measured.
The results of the scale tests are analysed and published (Van Eekelen et al., 2011a;
2011b, 2011c). Some of the conclusions of these experiments are: (1) A smooth
relationship was found between the net load (top load – subsoil support C) and several
measured parameters: load parts A and B, and GR settlement. (2) Consolidation of the
subsoil resulted in both an increase of load part B and also an unexpected increase of load
part A. Thus consolidation resulted in an improvement of the arching. (3) The measured
response to consolidation depends on the friction angle of the fill. (4) Strains in the GR
occur mainly in the tensile strips that lie on top of and between adjacent piles. (5) The line
load on a GR strip between piles has the shape of an inverse triangle, although the load
may be even more concentrated around the pile caps than indicated by the inverse triangle.
This follows from the shape of the deformed GR, measured between two piles. To improve
the understanding of the phenomenon arching and where possible, to confirm the observed
load distributions and displacements, numerical analysis of the scale tests have been
performed and reported in this report.
The design of reinforced piled embankments involve complex soil-structure interaction.

Numerical techniques such as finite element analysis are available for modelling this
complex soil-structure interaction. In literature, these numerical techniques are validated by
comparing the numerical outcomes with results from analytical calculations, measured
insitu data or scale tests. The numerical analyses are performed with different geometries.
From these results can be concluded that the phenomenon arching is a three dimensional
problem and therefore a three dimensional model will give more accurate results. It is found
that the Hardening Soil model is the most suitable material model, since it describes the
behaviour of the soil the most accurate. Another important conclusion is that it is necessary
to use updated mesh in the numerical analysis, because the influence of change in the
geometry of the model is of importance for the tensile forces in the geosynthetic
reinforcement.
Piled embankment with geosynthetic reinforcement – Master of Science thesis V

Summary
The performed scale test set-up consist of: four piles, a soaked and watertight foam
cushion between the piles, a sand layer of 1.5 to 2 cm on top of the foam cushion and piles,
one stiff steel frame, to which the GR is attached, an embankment fill of 42 cm granular
material and a top load applied by a water cushion. A tap on the foam cushion allows
drainage of the foam cushion during the test. The scale test is controlled by applying a top
load and drainage of the foam cushion.
To determine the strength and stiffness parameters of the granular material, triaxial tests
have been performed. The triaxial test were carried out on three 300 mm x 600 mm
samples (diameter x height) with a confining pressure applied by vacuum pressure. The
radial and axial displacements were measured at 1/3 and 2/3 of the sample height. The
stiffness and strength parameters are determined from the results according BS1377
(1990). The friction angle of the granular material is 49 degrees, the cohesion is zero and
the dilatancy angle 11 degrees.
The numerical analysis of the scale test is performed with Plaxis 3D Tunnel version 2.4.
The geometry of the Plaxis model is one quarter of the geometry of the scale test. This
reduces the amount of elements and therefore the calculation time. In the scale test circular
piles are applied. Circular geometry cannot be modelled in Plaxis, therefore the circular pile
is mathematically converted to a square pile. The sand and granular material are modelled
with Mohr Coulomb (MC) model and Hardening Soil (HS) model. The scale test is driven by
applying top load and by drainage of the foam cushion. To model the drainage of the foam
cushion, the measured water pressure is assigned to the subsoil clusters in Plaxis by a
water pressure head.
During the execution of the scale tests, there is a loss of load due to friction. The friction is
between 10% to 20% of the soil weight and top load and therefore an important part of the
scale test. Normally an interface is applied in Plaxis to model this friction, however in Plaxis
3D Tunnel it is not possible to apply an interface at the back side of the model (slice parallel
to the front plane). A possibility to simulate the friction is by applying an extra soil segment
with lower strength properties. This option gives good results with the MC model, except in
the phase where there is no subsoil support. Also the calculation time increases
significantly when the extra soil body is applied in combination with the HS model.
Therefore it is not realistic to apply an extra soil segment as friction. To decrease the load
in the system, the amount of load loss is subtracted from the top load.
The smooth relationship between the net load and load part A and B and GR displacement
from the scale test are also found in the Plaxis calculations. The drainage of the foam
cushion in Plaxis results in an increase of load part A and B, thus results in improvement of
arching. The calculated tensile forces in the GR are concentrated in ‘tensile strips’ that lie
on top and between adjacent piles. The largest displacement of the GR is found at the
middle of four piles. The results of the Plaxis calculations are compared to the scale test
results. During the first part of the test, the load distribution (load part A and B) shows
similar results as the measured load distribution. During the second part of the test the load
part A is overestimated and load part B is underestimated. The displacement of the
geosynthetic reinforcement calculated with Plaxis is underestimated compared to the scale
test results. In general the results of the HS model are better than the MC model.
By varying a number of parameters in the model, possible causes for the underestimated
displacement are investigated. This research concludes that the stiffness of the foam
cushion and the water pressure in the foam cushion does not have influence on the GR
displacement. The vertical effective stresses are concentrated on and directly next to the
pile and are relatively small between the piles. The internal friction angle does have a large
influence on the geosynthetic reinforcement displacement, because when the internal
friction angle is decreased, the arch decreases and the settlements increase. However, the
measured geosynthetic reinforcement displacement from the scale test is still not found in
the numerical results calculated in Plaxis.
VI Piled embankment with geosynthetic reinforcement – Master of Science thesis

21 February 2011
Table of Contents
EXAMINATION COMMITTEE ....................................................................................................... I
PREFACE .................................................................................................................................... III
SUMMARY .................................................................................................................................... V
TABLE OF CONTENTS.............................................................................................................. VII
LIST OF FIGURES....................................................................................................................... XI
LIST OF TABLES ...................................................................................................................... XV
LIST OF SYMBOLS ................................................................................................................. XVII
1 INTRODUCTION .................................................................................................................... 1
1.1 BACKGROUND INFORMATION ........................................................................................... 1
1.2 PROBLEM DEFINITION AND OBJECTIVES ............................................................................ 3
1.3 READER’S GUIDE ............................................................................................................ 3
2 PILED EMBANKMENT WITH GEOSYNTHETIC REINFORCEMENT .................................. 5
2.1 SUBSOIL ......................................................................................................................... 5
2.2 PILES…………………………………………………………………………………………...5
2.3 GEOSYNTHETIC REINFORCEMENT .................................................................................... 7
2.4 EMBANKMENT ................................................................................................................. 8
2.5 LOAD……………………………………………………………………………………………8
2.6 LOAD DISTRIBUTION ........................................................................................................ 9
2.6.1 Arching in the embankment................................................................................... 9
2.6.2 Spreading forces in the slope .............................................................................. 10
3 ANALYTICAL METHODS FOR THE DESIGN OF PILED EMBANKMENT WITH
GEOSYNTHETIC REINFORCEMENT ................................................................................. 11
3.1 GERMAN METHOD EBGEO ........................................................................................... 11
3.1.1 Vertical load distribution ...................................................................................... 11
3.1.2 Tension load in the geosynthetic reinforcement.................................................. 13
3.2 DUTCH METHOD CUR 226 ............................................................................................ 13
3.2.1 Geometry and boundary conditions .................................................................... 14
3.2.2 Vertical load distribution ...................................................................................... 15
3.2.3 Tensile load in geosynthetic reinforcement ......................................................... 15
4 NUMERICAL METHODS FOR THE DESIGN OF PILED EMBANKMENT WITH
GEOSYNTHETIC REINFORCEMENT ................................................................................. 19
4.1 DEFINITIONS ................................................................................................................. 19
4.2 EXPERIENCE NUMERICAL DESIGN ................................................................................... 21
4.2.1 Experience numerical design with Plaxis ............................................................ 21
4.2.2 Experience numerical design with other programs ............................................. 23
4.3 RESULTS OF EXPERIENCE WITH FINITE ELEMENT METHOD USING PLAXIS ......................... 24
4.3.1 Geometry ............................................................................................................. 24
4.3.2 Geosynthetics ...................................................................................................... 25
4.3.3 Constitutive models ............................................................................................. 25
4.3.4 Parameters .......................................................................................................... 27
4.4 CONCLUSION ................................................................................................................ 27
Piled embankment with geosynthetic reinforcement – Master of Science thesis VII

List of Tables
5 SCALE TEST ........................................................................................................................ 29

5.1 GENERAL DESCRIPTION SCALE TEST............................................................................... 29
5.2 GEOMETRY & SET-UP.................................................................................................... 30
5.2.1 Piles ..................................................................................................................... 30
5.2.2 Subsoil ................................................................................................................. 31
5.2.3 Geosynthetic reinforcement................................................................................. 32
5.2.4 Embankment fill ................................................................................................... 33
5.2.5 Load ..................................................................................................................... 33
5.2.6 Scaling effect ....................................................................................................... 34
5.3 MEASUREMENTS ........................................................................................................... 35
5.3.1 Consolidation foam cushion ................................................................................ 35
5.3.2 Load distribution .................................................................................................. 35
5.3.3 Strains.................................................................................................................. 37
5.3.4 Deformation ......................................................................................................... 37
5.3.5 Top load ............................................................................................................... 38
5.3.6 Summary of measurements ................................................................................ 38
5.4 TEST PROCEDURE, DEFINITION OF EQUILIBRIUM POINTS AND NET LOAD ............................ 39
5.5 SCALE TEST K2 ............................................................................................................. 40
5.5.1 Description test K2 .............................................................................................. 40
5.5.2 Specification measurements scale test K2 .......................................................... 42
5.5.3 Test procedure scale test K2 ............................................................................... 42
5.5.4 Specification equilibrium points scale test K2 ..................................................... 45
5.5.5 Checking the measurements scale test K2 ......................................................... 45
5.5.6 Load distribution scale test K2............................................................................. 47
5.5.7 Deformation scale test K2 ................................................................................... 49
5.5.8 Strains.................................................................................................................. 50
6 TRIAXIAL TEST ON GRANULAR FILL MATERIAL ........................................................... 51
6.1 TEST PROCEDURE ......................................................................................................... 51
6.2 CALCULATIONS ............................................................................................................. 54
6.3 EXECUTION TEST .......................................................................................................... 55
6.4 DATA ANALYSES & RESULTS .......................................................................................... 56
6.5 CONCLUSIONS .............................................................................................................. 60
7 NUMERICAL MODEL OF SCALE TEST ............................................................................. 61
7.1 GEOMETRY AND BOUNDARY CONDITIONS ........................................................................ 61
7.2 CONSTITUTIVE RELATIONS AND MATERIAL PARAMETERS .................................................. 62
7.2.1 Piles ..................................................................................................................... 62
7.2.2 Subsoil ................................................................................................................. 63
7.2.3 Geosynthetic reinforcement................................................................................. 64
7.2.4 Embankment fill ................................................................................................... 65
7.2.5 Steel box wall and frame ..................................................................................... 67
7.3 MESH GENERATION ....................................................................................................... 67
7.4 FRICTION ...................................................................................................................... 68
7.4.1 Wall friction .......................................................................................................... 68
7.4.2 Pile friction ........................................................................................................... 68
7.4.3 Geosynthetic reinforcement friction ..................................................................... 68
7.5 CALCULATION PHASES ................................................................................................... 68
7.6 DETERMINATION OF LOADS AND SETTLEMENTS ............................................................... 69
7.7 CONCLUSION ................................................................................................................ 71
VIII Piled embankment with geosynthetic reinforcement – Master of Science thesis

21 February 2011
8 VALIDATION AND ANALYSIS OF NUMERICAL MODEL ................................................. 73

8.1 NUMERICAL RESULTS OF SCALE TEST K2 ....................................................................... 73
8.2 COMPARISON NUMERICAL RESULTS TO SCALE TEST K2 RESULTS ..................................... 76
8.2.1 Load distribution .................................................................................................. 76
8.2.2 Displacements ..................................................................................................... 79
8.3 ANALYSIS OF NUMERICAL MODEL RESULTS ..................................................................... 80
8.3.1 Subsoil ................................................................................................................. 80
8.3.2 Geosynthetic Reinforcement ............................................................................... 87
8.3.3 Granular material ................................................................................................. 92
8.4 CONCLUSIONS .............................................................................................................. 99
9 CONCLUSIONS AND RECOMMENDATIONS.................................................................. 101
9.1 CONCLUSIONS ............................................................................................................ 101
9.2 RECOMMENDATIONS ................................................................................................... 102
REFERENCES .......................................................................................................................... 103
APPENDIX ................................................................................................................................ 105
APPENDIX A CALCULATION STIFNESS FOAM CUSHION

APPENDIX B PROPERTIES OF SAND AND GRANULAR MATERIAL
APPENDIX C SCALE TEST PROCEDURE
APPENDIX D JOURNAL SCALE TEST
APPENDIX E TRIAXIAL TEST
APPENDIX F RELATIVE DENSITY
APPENDIX G MATERIAL MODELS
APPENDIX H NUMERIAL SIMULATIONS OF SOIL-STRUCTURE INTERFACE
BEHAVIOUR
APPENDIX I RESULTS OF PLAXIS MODELS
Piled embankment with geosynthetic reinforcement – Master of Science thesis IX

21 February 2011
List of Figures
Figure 1.1: Bumpy road ........................................................................................................ 1
Figure 1.2: Cross section piled embankment ........................................................................ 2
Figure 1.3: Load distribution in a reinforced piled embankment ............................................ 2
Figure 2.1: Augeo piles ......................................................................................................... 6
Figure 2.2: Timber piles ........................................................................................................ 6
Figure 2.3: Settlement free system ....................................................................................... 6
Figure 2.4: Settlement reducing system ................................................................................ 6
Figure 2.5: Prefab pile caps .................................................................................................. 7
Figure 2.6: Geosynthetic reinforcement and piles underneath the embankment .................. 7
Figure 2.7: Geogrids ............................................................................................................. 8
Figure 2.8: Non-woven and woven geotextiles ..................................................................... 8
Figure 2.9: Interaction between grid and granular material ................................................... 8
Figure 2.10: Arching (Van Eekelen, 2010) ............................................................................ 9
Figure 2.11: Spreading forces in the slope ......................................................................... 10
Figure 3.1: Analytical approach of Zaeske (Zaeske, 2001) ................................................. 11
Figure 3.2: Analytical approach of Zaeske (Zaeske, 2001) ................................................. 12
Figure 3.3: Load distribution (left) and schematization of loading (right) on geosynthetic
reinforcement strip as assumed by EBGEO and CUR (Van Eekelen, 2010) ...................... 13
Figure 3.4: Schematization of the cable equation ............................................................... 13
Figure 3.5: Cross section and top view of piled embankment geometry (CUR 226, 2010) . 14
Figure 3.6: Maximum strain in the geosynthetic reinforcement (CUR 226, 2010) ............... 16
Figure 3.7: Vertical force on the GR between the piles (CUR 226, 2010) ........................... 17
Figure 4.1: Axisymmetrical geometry .................................................................................. 19
Figure 4.2: Plane strain geometry ....................................................................................... 20
Figure 4.3: 3 Dimensional geometry ................................................................................... 20
Figure 4.4: Load transfer mechanisms of piled embankments ............................................ 21
Figure 4.5: Calculated and measured load on pile (Satibi, 2009) ....................................... 24
Figure 4.6: Calculated and measured tensile force in the geosynthetic reinforcement ....... 25
Figure 4.7: Load-settlement curves from analyses with different soil models compared with
measurements of the scale tests performed by Zaeske (2001) (Satibi, 2009) .................... 26
Figure 4.8: Calculated and measured data of the load on a pile (left) and data of the tensile
forces in the geosynthetic reinforcement (right) (Satibi, 2009) ............................................ 26
Figure 4.9: Influence of the friction angle ’ on subsoil surface settlements at the pile centre
(Satibi, 2009)....................................................................................................................... 26
Figure 4.10: Influence of the dilatancy angle  on subsoil surface settlements at the pile
centre (Satibi, 2009)............................................................................................................ 26
Figure 4.11: Influence of the oedometer loading stiffness on subsoil surface settlements at
the pile centre (Satibi, 2009) ............................................................................................... 27
Figure 5.1: Section used from a piled embankment ............................................................ 29
Figure 5.2: Side view test set-up ......................................................................................... 30
Figure 5.3: Top view test set-up .......................................................................................... 30
Figure 5.4: Piles in test set-up ............................................................................................ 31
Figure 5.5: First cushion (left) and second cushion (right) .................................................. 31
Figure 5.6: Design steel frame ............................................................................................ 32
Figure 5.7: Set-up geosynthetic reinforcement ................................................................... 32
Figure 5.8: Itterbeck sand ................................................................................................... 33
Figure 5.9: Granular material .............................................................................................. 33
Figure 5.10: Compaction with steel weight ......................................................................... 33
Figure 5.11: Water cushion ................................................................................................. 34
Piled embankment with geosynthetic reinforcement – Master of Science thesis XI

List of Figures
Figure 5.12: Test set-up ..................................................................................................... 34

Figure 5.13: Test set-up pressure cells for one layer of geosynthetic reinforcement .......... 35
Figure 5.14: Test set-up pressure cells for two layers of geosynthetic reinforcement ........ 36
Figure 5.15: Strain gauges ................................................................................................. 37
Figure 5.16: Bicycle gear cables......................................................................................... 37
Figure 5.17: Liquid leveling system .................................................................................... 38
Figure 5.18: Water pressure cells ....................................................................................... 38
Figure 5.19: Test set-up deformation .................................................................................. 38
Figure 5.20: x-y scanner ..................................................................................................... 38
Figure 5.21: Tensile test results on GR of K2 (5 test samples) in cross direction (left) and
longitudinal direction (right)................................................................................................. 41
Figure 5.22: Isochronous curves of K2 ............................................................................... 42
Figure 5.23: Locations of the steel strain cables (yellow) and the settlement transducers
(blue) of test K2 .................................................................................................................. 42
Figure 5.24: Test procedure (top load, average settlement due to drainage and subsoil
support) with selected equilibrium points (dotted lines) ...................................................... 44
Figure 5.25: Measured load parts A (on piles 1 and 2, on top of reinforcement), A+B (below
reinforcement) and C (determined from force transducers C1 to C4, corrected for weight
changes of the water foam cushion) with selected equilibrium points (dotted lines) ........... 44
Figure 5.26: Measured load parts A and A+B on both piles ............................................... 45
Figure 5.27: Comparison total weight and total measured weight ...................................... 46
Figure 5.28: Comparison corrected pressure foam cushion and total force on force
transducers (C1+C2+C3+C4) ............................................................................................. 47
Figure 5.29: Measured load distribution for scale test K2 ................................................... 48
Figure 5.30: Vertical displacements geosynthetic reinforcement in time for scale test K2.. 49
Figure 5.31: Vertical displacements of geosynthetic reinforcement for scale test K2 ......... 49
Figure 5.32: Left: strain (gear) cables  in K2, the arched areas are parts of the GR that act
as tensile strips. Right: measured strains in K2. ................................................................. 50
Figure 6.1: Sample prepared in steel mould ....................................................................... 52
Figure 6.2: Measuring rings with horizontal and vertical displacement sensors ................. 52
Figure 6.3: Large scale triaxial apparatus ........................................................................... 53
Figure 6.4: Control and data acquisition unit ...................................................................... 53
Figure 6.5: Location displacement sensors on top ring (left) and on bottom ring (right) ..... 55
Figure 6.6: Sample shape at failure .................................................................................... 56
Figure 6.7: Deviatoric stress strain curve ........................................................................... 57
Figure 6.8: Volume change curve ....................................................................................... 57
Figure 6.9: Stress path curve.............................................................................................. 59
Figure 6.10: Mohr circle of effective stress ......................................................................... 59
Figure 7.1: Top view scale test set-up ................................................................................ 61
Figure 7.2: 3D geometry Plaxis .......................................................................................... 61
Figure 7.3: Cross section of Plaxis model and dimensions ................................................ 62
Figure 7.4: Water pressure development in foam cushion ................................................. 64
Figure 7.5: Pile head after execution scale test .................................................................. 65
Figure 7.6: Creating a 3D model and finite element mesh (Plaxis) ..................................... 67
Figure 7.7: 15-node wedge element with: nodes (left) and stress points (right).................. 68
Figure 7.8: Area for calculation of the load part A and part A+B ......................................... 70
Figure 8.1: Cross section of model and top view ................................................................ 73
Figure 8.2: Effective principal stresses for Mohr Coulomb model ....................................... 74
Figure 8.3: Effective principal stresses for Hardening Soil model ....................................... 74
Figure 8.4: Vertical displacement of vacuum phase for MC model ..................................... 74
Figure 8.5: Vertical displacement of vacuum phase for Hardening Soil model ................... 74
Figure 8.6: Tensile forces in GR of vacuum phase for Mohr Coulomb model in x-direction
(top) and in y-direction (bottom).......................................................................................... 75
Figure 8.7: Tensile forces in GR of vacuum phase for Hardening Soil model in x-direction
(top) and in y-direction (bottom).......................................................................................... 75
Figure 8.8: Total displacement of GR in vacuum phase for Mohr Coulomb model ............. 75
Figure 8.9: Total displacement of GR in vacuum phase for Hardening Soil model ............. 75
XII Piled embankment with geosynthetic reinforcement – Master of Science thesis

21 February 2011
Figure 8.10: Total displacement A-A of GR for Mohr Coulomb model ................................ 76
Figure 8.11: Total displacement A-A of GR for Hardening Soil model ................................ 76
Figure 8.12: Load part A+B ................................................................................................. 77
Figure 8.13: Load part C ..................................................................................................... 77
Figure 8.14: Load part A ..................................................................................................... 77
Figure 8.15: Load part B ..................................................................................................... 77
Figure 8.16: Load part A in percentage ............................................................................... 78
Figure 8.17: Load part B in percentage ............................................................................... 78
Figure 8.18: Displacement z1 (Centre 4 piles) .................................................................... 79
Figure 8.19: Displacement z2 (Centre 2 piles) .................................................................... 80
Figure 8.20: Displacement z3 (Close to pile) ...................................................................... 80
Figure 8.21: Load part A+B, stiffness foam cushion ........................................................... 81
Figure 8.22: Load part C, stiffness foam cushion ................................................................ 81
Figure 8.23: Load part A, stiffness foam cushion ................................................................ 81
Figure 8.24: Load part B, stiffness foam cushion ................................................................ 81
Figure 8.25: Displacement z1 (centre 4 piles), stiffness foam cushion ............................... 82
Figure 8.26: Displacement z2 (centre 2 piles), stiffness foam cushion ............................... 82
Figure 8.27: Displacement z3 (close to pile), stiffness foam cushion .................................. 82
Figure 8.28: Water pressure in foam cushion in the second drainage phase with a load of
100kPa, location is z2 of MC model (left) and of HS model (right). ..................................... 84
Figure 8.29: Load part A+B, water pressure ....................................................................... 85
Figure 8.30: Load part C, water pressure ........................................................................... 85
Figure 8.31: Load part A, water pressure ............................................................................ 85
Figure 8.32: Load part B, water pressure ............................................................................ 85
Figure 8.33: Displacement z1 (centre 4 piles), water pressure ........................................... 86
Figure 8.34: Displacement z2 (centre 2 piles), water pressure ........................................... 86
Figure 8.35: Displacement z3 (close to pile), water pressure ............................................. 86
Figure 8.36: Geometry of Plaxis model with only load B ..................................................... 87
Figure 8.37: Load part B, GR .............................................................................................. 88
Figure 8.38: Displacement z1 (centre 4 piles), GR ............................................................. 89
Figure 8.39: Displacement z2 (centre 2 piles), GR ............................................................. 89
Figure 8.40: Displacement z3 (close to pile), GR ................................................................ 89
Figure 8.41: Load part A+B, stiffness GR ........................................................................... 90
Figure 8.42: Load part C, stiffness GR................................................................................ 90
Figure 8.43: Load part A, stiffness GR ................................................................................ 90
Figure 8.44: Load part B, stiffness GR ................................................................................ 90
Figure 8.45: Displacement z1 (centre 4 piles), stiffness GR ............................................... 91
Figure 8.46: Displacement z2 (centre 2 piles), stiffness GR ............................................... 91
Figure 8.47: Displacement z3 (close to pile), stiffness GR .................................................. 91
Figure 8.48: Location of cross section C-C and locations A-J in cross section and top view
of the model ........................................................................................................................ 92
Figure 8.49: Effective vertical stress distribution for locations A to D 1) ............................... 93
Figure 8.50: Effective vertical stress distribution for locations E to H 1) ............................... 93
Figure 8.51: Effective vertical stress distribution for locations I to J 1) ................................. 93
Figure 8.52: Effective vertical tensile stresses in the stress points of the elements ............ 94
Figure 8.53: Vertical effective stresses in cross section E-E ............................................... 94
Figure 8.54: Results of triaxial tests on samples (CUR-166, 2005)..................................... 96
Figure 8.55: Load part A+B, decrease phi .......................................................................... 97
Figure 8.56: Load part C, decrease phi ............................................................................... 97
Figure 8.57: Load part A, decrease phi ............................................................................... 97
Figure 8.58: Load part B, decrease phi ............................................................................... 97
Figure 8.59: Displacement z1 (centre 4 piles), decrease phi .............................................. 98
Figure 8.60: Displacement z2 (centre 2 piles), decrease phi .............................................. 98
Figure 8.61: Displacement z3 (close to pile), decrease phi ................................................. 98
Figure 8.62: Comparison of measured load distribution for tests T2 (sand) and T3
(granular) (Van Eekelen, et al., 2011a) ............................................................................... 99
Piled embankment with geosynthetic reinforcement – Master of Science thesis XIII

List of Figures
XIV Piled embankment with geosynthetic reinforcement – Master of Science thesis

21 February 2011
List of Tables
Table 2.1: Distributed load for a 600 kN truck ....................................................................... 9
Table 3.1: Boundary conditions piled embankment ............................................................ 14
Table 4.1: Summary of experience with Plaxis ................................................................... 28
Table 5.1: Geometry piles ................................................................................................... 30
Table 5.2: Geosynthetic reinforcement ............................................................................... 33
Table 5.3: Geometry embankment...................................................................................... 33
Table 5.4: Scaling effect ..................................................................................................... 34
Table 5.5: Dimensions scale test and reality ....................................................................... 35
Table 5.6: Summary of measurements and equipment ...................................................... 39
Table 5.7: Average stiffness dependent on strain, for short-term behaviour of GR ............ 41
Table 5.8: Test procedure with selected equilibrium points ................................................ 43
Table 5.9: Specification equilibrium points. ......................................................................... 45
Table 5.10: Load distribution at equilibrium points for scale test K2 ................................... 48
Table 5.11: Deformations and strains at equilibrium points for K2 ...................................... 50
Table 6.1: Characteristics and results of the samples in the triaxial tests ........................... 58
Table 6.2: General characteristics of the triaxial tests......................................................... 60
Table 6.3: Summary of strength and stiffness properties of the granular material .............. 60
Table 7.1: Pile properties .................................................................................................... 63
Table 7.2: Subsoil properties .............................................................................................. 63
Table 7.3: Geosynthetic reinforcement properties .............................................................. 65
Table 7.4: Sand and granular material properties for the Mohr Coulomb model ................. 66
Table 7.5: Sand and granular material properties for the Hardening Soil model ................. 66
Table 7.6: Steel box wall and frame properties ................................................................... 67
Table 7.7: Plaxis calculation phases ................................................................................... 69
Piled embankment with geosynthetic reinforcement – Master of Science thesis XV

21 February 2011
List of Symbols
Symbol Dimension Description
A kN/pile Measured load part transferred to the pile by arching

Acircular m2 Area of circular pile
Area of the fill material Afill  1.1 m2
2 2
Afill m
Afoam m2 Area of the foam cushion
Ai m 2
Zone of influence per pile, calculated with Ai  sx  sy
2
Aint mm Initial area of cross section
2
ALx ;Ly m Zone of influence
Ap 2
m Pile cap area
As m2 Area of the soil body
2
Asquare m Area of square pile
a m Diameter of a circular pile cap
aeq m Equivalent diameter of a rectangular pile cap, calculated with aeq  4  Ap / 
B kN/pile Measured load part transferred to the pile by the GR

beq m Equivalent width of the pile cap
C kN/pile Measured load supported by subsoil
c kPa Cohesion
3
c increment kN/m Increase cohesion with depth
2
dA0 m Area of the upper side of the element
2
dAs m Area of a side of the element
2
dAu m Area of the lower side of the element
2
dV m Volume of the element
E - Efficacy
2
E '50
ref
kN/m Secant stiffness for primary loading in standard triaxial test
3
E increment kN/m Increase of stiffness with depth
2
E' ref
oed kN/m Tangent stiffness for primary loading in oedometer test
2
E ref kN/m Young’s modulus
Es kN/m2 Linear or non-linear modulus of elasticity for subsoil
2
E ref
ur kN/m Unloading / reloading stiffness
F kN Vertical force on the reinforcement strip between two piles
Fframe kN Force due to the weight of frame
Fs kN Subsoil reaction
g kN/m2 Permanent load on the embankment
H m Embankment height
HT kN/m Constant horizontal component of the tensile force in the cable
h m Height layer
hfoam m Height foam cushion
hg m Height of the embankment for full arching, H
Piled embankment with geosynthetic reinforcement – Master of Science thesis XVII

List of Symbols

hsand m Height sand layer
J  EAg kN/m Stiffness of geosynthetic reinforcement
K 0nc - K0-value for normal consolidation
Kkrit - Critical ratio of the principal stresses
3
k foam kN/m Stiffness foam cushion
3
ks kN/m Modulus of subgrade reaction
kx , ky , kz m/day Permeability
Lw m Clear width between the pile caps
Lx,y m Distance between pile caps (sx,y - Bers)
m - Stress dependent stiffness according to a power law
n - Stress concentration ratio
P kN Total force carried by the pile caps
Pmeasured kPa Measured pressure within foam cushion
Vertical load (permanent +variable) on the embankment, calculated by
p kN/m2 p  g q
2
pb kN/m Average vertical pressure above geosynthetic
2
pGR kN/m Pressure against the geosynthetic reinforcement in Plaxis
2
pPlaxis kN/m Water pressure within the foam cushion in Plaxis
p ref kN/m2 Reference stress for stiffness
2
pscale test kN/m Measured water pressure within the foam cushion
q kN/m2 Variable load (traffic load) on the embankment
q0 kN/m2 Uniform surcharge on the embankment
q( x ) kN/m2 The loading on the cable
R kN/pile Friction between embankment fill and box walls
Rf - Failure ratio
r mm Radius of the sample
rint mm Initial radius of the sample
s m Centre-to-centre distance of the piles in diagonal direction
sx ; y Centre to centre distance of the piles in longitudinal direction (x) or cross
m
direction (y)
Tensile force in the geosynthetic reinforcement caused by the vertical load due
Tv kN/m
to arching
t m Thickness of the subsoil
Wn kN/pile Net load
Ws kN/pile Top load on the embankment
w m Vertical deformation of the cable
w m Average settlement over the area
x m Horizontal distance
y ref m Reference level
z m Distance between the pile cap and geosynthetic reinforcement
Average reduction height foam cushion, due to the drainage of the cushion
zaverage m
(determined from the measured amount of litres drained from the cushion)
zPlaxis m Input value for Plaxis: pressure head
r mm Change in radius of the sample from the start of compression
m rad Average angle of part arch of element
 kN/m 3
Volumetric weight
 sat kN/m 3
Saturated volumetric weight
 unsat kN/m3 Unsaturated volumetric weight
XVIII Piled embankment with geosynthetic reinforcement – Master of Science thesis

21 February 2011

 water kN/m 3
Volume weight of water (9.8 kN/m3)
 - Average strain in the geosynthetic reinforcement
1 - Axial strain
3 - Radial strain
v - Volumetric strain
 - Poisson’s ratio
 ur - Poisson’s ratio for elastic unloading/reloading
 - Stress reduction ratio
c kN/m
2
Vertical stress on the pile caps
 n,f kPa Normal stress at failure
s kN/m
2
Vertical stress on the soil
 tension kN/m2 Tensile strength
 v ;g 2 Vertical stress on the level of geosynthetic reinforcement between the pile
kN/m
caps
 vstat
;p kN/m2 Total vertical load on the level of geosynthetic reinforcement
 v ;t kN/m
2
Total vertical load on the level of geosynthetic reinforcement
 kN/m2 Soil stress perpendicular to sides of the element
z kN/m2 Effective vertical soil stress
f kPa Shear stress at failure
 ° Internal friction angle
 ° Dilatancy angle
Piled embankment with geosynthetic reinforcement – Master of Science thesis XIX

21 February 2011
Chapter 1
Introduction
To introduce the subject of this master thesis the background information, problem
definition and objectives are described. The chapter is completed with the composition of
this report.
1.1 Background information

To distribute the load of a highway or railway track to the ground, an embankment is
constructed as a foundation. The construction of such an embankment will cause an
increase of the stress in the subsoil. Due to the increase of stress creep and consolidation
occur in the subsoil. Creep follows when stress is applied on soft soil, causing the soil
particles to press together. The volume of the soil will decrease resulting in settlements.
Consolidation occurs when the soft soil is saturated. By the increase of stress water is
squeezed out of the subsoil causing settlements.
In the western part of the Netherlands, the

ground consists mainly of soft soils, like
clay and peat. When large differential
settlements after construction occur, this
can cause damage to constructions which
result in high maintenance costs. See
Figure 1.1 for an example of large
differential settlements. When a highway
or railway track has to be constructed and
the subsoil consists of soft soil, the soil
has to be improved. Several techniques
have been developed and applied for soft
soil improvement. One of the techniques
Figure 1.1: Bumpy road
is a reinforced-piled embankment.
A piled embankment consists of a field of piles and a geosynthetic reinforcement (GR) in

the embankment on top of the piles. The embankment (fill) is usually constructed of
granular material. See Figure 1.2 for a cross section of a piled embankment. A piled
embankment with reinforcement can be constructed rather fast and has a small settlement
after construction or is even settlement-free. This results in lower maintenance costs.
Another advantage is that a piled embankment prevents horizontal movement of the
subsoil. The consequence is that the embankment can be build next to sensitive
constructions, like cables, pipelines, sensitive foundations or existing roads.
Piled embankment with geosynthetic reinforcement – Master of Science thesis 1

Chapter 1 Introduction
asfalt = asphalt
fundering = foundation
geogrid = geogrid
palen = piles
puingranulaat = crushed granular
rubble material (hereafter called
‘granular material’)
samendrukbare lagen = soft subsoil
sloot = ditch
zand = sand
Figure 1.2: Cross section piled embankment
The static and dynamic loads on a reinforced piled embankment are distributed to the
ground divided in three parts based on arching. The first part of the load (load part A) is
transferred directly to the piles by arching. Part of the load underneath the arch, is
transported via the GR to the piles (load part B). The rest of the load underneath the arch is
transported to the soft subsoil between the piles (load part C). See Figure 1.3 for a picture
with the load distribution.
Figure 1.3: Load distribution in a reinforced piled embankment
In the last few years, the CUR committee 159B has been working on the new Dutch Design
Guideline for the design of piled embankments. This design guideline is based on the
German EBGEO with adjustments to Dutch circumstances. To validate the guideline
several field tests have been performed. From the field measurements is concluded that
the design methods are very conservative. Improving the design guideline will reduce the
construction costs. To understand the physical behaviour of the piled embankment and to
validate the theory, experimental scale tests have been performed. Deltares is analysing
the results of the scale test and the results will be published in 2011.
2 Piled embankment with geosynthetic reinforcement – Master of Science thesis

21 February 2011
1.2 Problem definition and objectives

Several experimental scale tests have been performed. During these tests, different types
of geosynthetics and in a different amount of layers have been used. The three parts of the
load distribution were measured with pressure cells. Next to that, the vertical displacements
of the GR were measured with water pressure cells. The results of the scale tests were
analysed and published in Experiments on piled embankments: the influence of subsoil, fill,
and reinforcement (Van Eekelen, et al., 2011a).
Some of the most notable conclusions of Van Eekelen et al. (2011a) were that:
 A smooth relationship was found between the net load (top load – subsoil support
C) and several measured parameters: load parts A and B, and GR settlement;
 Consolidation of the subsoil resulted in both an increase of load part B and also an
increase of load part A. Thus consolidation resulted in an improvement of the
arching;
 The measured response to consolidation depends on the friction angle of the fill;
 Strains in the GR occur mainly in the tensile strips that lie on top of and between
adjacent piles;
 The line load on a GR strip between piles has the shape of an inverse triangle,
although the load may be even more concentrated around the pile caps than this
indicates. This follows from the shape of the deformed GR, measured at two points
between two piles (See also Van Eekelen, et al., 2011b; 2011c)
The purpose of this research is to perform numerical analysis of the scale model tests to
better understand and where possible, to confirm the observed load distributions and
displacements.
1.3 Reader’s guide

In this report a numerical analysis of the geosynthetic-reinforced piled embankments is
reported.
First the different parts of geosynthetic-reinforced piled embankment are described in

chapter 2. This includes a more detailed description of the load distribution in the
embankment. In chapter 3 an analytical method for designing the piled embankment is
described. The analytical method consists of the determination of the load distribution and
tensile forces in the GR due to vertical load. The piled embankment can also be analysed
numerically. The experience with numerical analysis of piled embankments described in the
literature is analysed in chapter 4. Before a numerical analysis of the scale test can be
performed, the geometry, test set-up and results of the scale tests are described in more
detail in chapter 5. The parameters of the granular material used in the embankment were
unknown and therefore analysed by performing triaxial tests on the granular material. The
results of the triaxial tests are described in chapter 6. Chapter 7 contains the numerical
model of the scale test. The results of the numerical model are analysed and compared to
the results of the scale test in chapter 8. In the last chapter the conclusions and
recommendations are given.

21 February 2011
Chapter 2
Piled embankment with geosynthetic

reinforcement
A piled embankment consists of various parts. The subsoil where the embankment is built
on, the piles and the embankment that is reinforced with a geosynthetic reinforcement. The
load on the embankment and how the load is distributed in the embankment is important for
the design of the piled embankment. In this chapter, these parts are explained.
2.1 Subsoil
A piled embankment normally is applied in soft subsoil. Due to the reinforcement and the
piles of the piled embankment, the soft subsoil is almost not loaded by the embankment
and dynamic load. Therefore, the settlements are small.
Due to small settlements, there are small strains in the subsoil. The subsoil will react, but
this reaction will be small. Therefore a linear relationship between the subsoil reaction F
and the settlement w can be used. The subsoil can be modelled as a spring with a spring
constant. This constant is the modulus of subgrade reaction ks.
Fs  ks  ( As  w ) (2.1)
With:
Es
ks  (2.2)
t
Where:
Fs kN Subsoil reaction
3
As m2 Area of the soil body
w m Average settlement over the area
Es kN/m2 Linear or non-linear modulus of elasticity for subsoil
t m Thickness of the subsoil
2.2 Piles
Several pile types are applied for piled embankments. CUR 226 (2010) indicates several
pile types. The most commonly used type of piles is the soil displacement type (Figure 2.1
and Figure 2.2):
 Driven piles of prefab concrete or timber;
 In-situ poured piles with reinforcement (in the Netherlands, AuGeo and HSP have
been applied several times in a piled embankment).
There are two methods to design a piled embankment: a settlement free construction or a
settlement reducing construction (CUR 226, 2010).

Chapter 2 Piled embankment with geosynthetic reinforcement
If a settlement free system is applied, the settlements are almost none (< 3 cm), because
the piles are designed to carry the total load. The effect of a setting free system is that the
soft soil between the piles can settle whereby a gap arises between the soft soil and the
geosynthetics. For the design of the geosynthetics, load part C (subsoil) can be assumed
zero, dependent on the local circumstances. The piles underneath are end-bearing piles.
The vertical bearing capacity of the end-bearing piles is determined using the common
Dutch design guidelines, as described in NEN 6743 (1991).
Figure 2.1: Augeo piles Figure 2.2: Timber piles
A settlement reducing system is a system whereby both the piles and the subsoil carry the
load. Some settlement of the piled embankment is accepted. The pile head will settle as
much as the subsoil between the piles. The piles underneath are end bearing piles or
friction piles. This second design approach results in less construction costs. For the
bearing capacity of the friction piles in a settlement reducing system, CUR 226 presents a
so-called ‘Interaction Model’.
Figure 2.3: Settlement free system Figure 2.4: Settlement reducing system
The centre-to-centre distance of the piles in Dutch piled embankments lies between 0.8
and 3.0 meter. This means that for a long trace of a piled embankment a large amount of
piles is needed. To decrease the construction time and thereby to increase the profit, the
installation speed of the piles is important.

21 February 2011
The forces on the piles are relative small, therefore thin piles can often be installed. Thin
piles have a disadvantage; they are sensitive for horizontal loads. Therefore, piles can be
sensitive for settlements on the sides of the embankment and for future constructions
nearby. Usually, piles are designed to resist horizontal loads from future elevations of the
embankment sides or even for future constructions. During construction of driven soil
displacement piles the vibrations and sound nuisance are often of minor importance,
because the soil resistance is minimal due to the soft soil and the slenderness of the piles.
After piling and before placing geosynthetics,

pile caps are placed on top of the piles.
These pile caps provide a smaller span of
the geosynthetics and a larger contact
surface between the geosynthetics and the
pile. Depending on the type of piles used for
the embankment, the pile cap is prefab or in
situ fabricated. The corners of the pile cap
should be rounded off to prevent damage of
the geosynthetics. The piles caps are placed
above the water table for visibility. When
necessary the water table is lowered by
drainage. See Figure 2.5 for a picture of
prefab pile caps. (CUR 226, 2010) Figure 2.5: Prefab pile caps
2.3 Geosynthetic reinforcement

A piled embankment is reinforced with one or more layers of geosynthetic reinforcement in
the base of the embankment. The function of the geosynthetic reinforcement and several
types of geosynthetics are described below.
The function of the geosynthetic

reinforcement is to secure the arching in
the embankment, and to transport the
load between the piles to the piles. In
Figure 2.6 this is illustrated. The load can
be split up in vertical load and horizontal
load. The load distribution in the
geosynthetics is explained in paragraph
2.6.
The strength of geosynthetic

reinforcement can be uniaxial or biaxial.
A uniaxial material is a material of which
the strength in longitudinal direction is
Figure 2.6: Geosynthetic reinforcement and (much) stronger than in cross direction.
piles underneath the embankment (Dalen, 2008) A biaxial material is a material of which
the strength in both the directions is (nearly) equal. The application of uniaxial or biaxial is
depending on the needed strength. Above a certain strength (around 100 kN/m) it is more
economical to apply two layers of uniaxial material perpendicular and on top of each other.
(CUR 226, 2010).
Different types of geosynthetic reinforcement that are available are geogrids, woven or non-
woven geotextiles, see Figure 2.7 and Figure 2.8. The type of material and the fabrication
process influences the properties of these geosynthetics.
There are two phenomena that have to be considered. One of them is slack. Before the
geosynthetic reinforcement can transfer tensile forces, an initial deformation is needed.
This initial deformation can be applied by installing the geosynthetic reinforcement in the

deformed position. The geosynthetics can be less stiff when applied in the deformed
position, because the tensile forces are smaller. To install the initial deformation is not
easy; on the other hand, to install the geosynthetic reinforcement completely horizontally is
impossible. In the construction phase, the initial deformation at the top of the embankment
can be corrected by levelling the embankment.
Figure 2.7: Geogrids Figure 2.8: Non-woven and woven geotextiles
Another phenomenon is interaction between grid

and granular material. This is applicable when a
geogrid is used for reinforcement in combination
with granular material. As seen in Figure 2.7 the
geogrids have a very open structure. In these
openings granular material can get stuck. Therefore
the geogrid gives support to the granular material,
see Figure 2.9. This improves the interaction Figure 2.9: Interaction between grid
and granular material
between the embankment and the geogrid.
When geosynthetics are applied creep has to be taken into account. Due to creep of the
GR, the tensile strain increases in time with a constant load. The geosynthetic
reinforcement influences mainly the lifetime of a piled embankment.
2.4 Embankment
The embankment fill has to have a high degree of internal friction. In the Netherlands, the
embankment fill usually consists of a granular material. In a material with a relative high
degree of internal friction, arching will develop better. The development of arching can also
be improved by increasing the height of embankment. Next to the height and degree of
internal friction, the drained behaviour of the fill material is important. This is important to
prevent water over pressure in the embankment.
2.5 Load
The load on the geosynthetic reinforcement and piles is divided in static load and dynamic
load. Static load is the weight of the embankment and the weight of the railway or road
construction. Dynamic loading occurs due to the now and then passing (heavy) vehicles
and trains.
The Dutch Guideline CUR 226 (2010) gives rules how to translate the dynamic load into a
uniformly distributed load. The axle loads of a standard truck are spread according
Boussinesq over the total height of the embankment. The asphalt top layer has extra
spreading capacity, therefore this may be taken into account with a virtual extra height. The
influence of all three axles of the standard truck are summed. Table 2.1 presents a
summary of the larger table that is part of the Dutch Guideline. The stress (max;ave) is the

21 February 2011
average stress on the maximal loaded pile grid (sx*sy), with sx,y (m) the centre-to-centre
distance between the piles. (Van Eekelen et al., 2010b)
Height embankment Maximal average stress (max;ave) [kN/m2] for pile grid (sx*sy)
H [m] 1.5 x 1.5 m2 2.0 x 2.0 m² 2.5 x 2.5 m²
1.0 61.3 51.3 44.8
2.0 33.7 30.0 27.8
3.0 21.1 19.8 19.0
Table 2.1: Distributed load for a 600 kN truck
The influence of dynamic load has been researched by Heitz (2006). He stated that the
arching is reduced when the dynamic load is too large. CUR 226 adopted his arch
reduction model.
2.6 Load distribution

The load distribution differs for each part of the embankment. The load distribution in the
middle part of the embankment is based on arching and in the slope of the embankment
there are spreading forces. The load distributions are explained below.
2.6.1 Arching in the embankment

Arching is a phenomenon whereby the load is transferred laterally to stiffer elements. The
soil between the piles is going to subside, however the soil above the piles cannot settle
and will give resistance to the soil body that is subsiding. This causes shear forces and
creates an arch.
When a very weak geosynthetic is used, the embankment fill will settle between the piles,
causing shear forces and arching development. The shear forces reduce the forces on the
geosynthetic reinforcement, but the force on the pile caps increases. On the other hand
when a very stiff geosynthetic is applied the tensile force in the geosynthetic reinforcement
will be very high and the displacement very small. Therefore arching cannot develop fully.
The load above the arch is load part A and transported directly to the piles. The size of the
load is determent by the steepness of the arch, which is depending on the angle of internal
friction of the embankment material.
The load underneath the arch (load part

B) produces a vertical load on the
geosynthetic reinforcement. The vertical
load on the geosynthetic causes very
small deflections in the geosynthetic A
which generates tensile forces in the
geosynthetic. The tensile forces are
transferred to the piles and to the firm B
layers.
Underneath the geosynthetic

reinforcement the load is carried by the
subsoil, load part C. Subsoil support
relieves the force in the geosynthetic C
reinforcement. Figure 2.10: Arching (Van Eekelen, 2010)
In CUR 226 (2010) different mechanisms are described that causes an arch increase or
decrease. For example dynamic load or subsoil subsidence. Below the different
mechanisms are explained.
The dynamic load can compact the embankment fill. This will increase the internal friction
angle causing a firmer arch. A condition for compaction is that the embankment fill is

supported by the subsoil and the subsoil is dynamical stiff. By compaction the load
distribution will develop better, whereby load part A increases and load part B+C will
decrease.
The stiffness of the arch can be increased by subsoil subsidence. Subsoil subsidence can
be caused by changes in groundwater level, load from for example a working platform
(below the geosynthetic reinforcement) or by (minor) cyclic loading. Heitz (2006) performed
tests with cyclic loading on peat in a large oedometer test. The test resulted in plastic
deformation of peat. By subsoil subsidence, load part C decreases, whereby load A+B
increase.
When the subsoil subside creep or relaxation can occur in the geosynthetic reinforcement.
Creep or relaxation of the geosynthetic reinforcement will decrease load B and increase
load parts A and C.
Granular material used for the embankment fill often contains cement. This cement causes
connections between the grains, whereby the cohesion increases. An increase of cohesion,
results in a better arch development and an increase of load part A. Because A increases,
load parts B and C decreases. There is a risk of brittle failure of the arch.
The connections between the grains can be broken down or the grains can be rearranged
by a large single or cyclic load. This mechanism is arch failure, causing load part A to
decrease and load part B+C to increase. After arch failure, the arch can increase again by
the previous described mechanisms.
2.6.2 Spreading forces in the slope

The slope of the embankment has a tendency to slide laterally creating spreading forces.
This can be seen in Figure 2.11. The spreading forces cause an horizontal load, which
causes an extra tensile force in the geosynthetic reinforcement. The length of the
reinforcement beyond the outer row of the piles should be long enough to ensure the
transfer of the tensile load. This thesis focuses only on the tensile forces in the
reinforcement that are due to vertical load. The lateral load falls beyond the scope of this
thesis.
Figure 2.11: Spreading forces in the slope

21 February 2011
Chapter 3
Analytical methods for the design of piled

embankment with geosynthetic reinforcement
There are several guidelines available to design the geosynthetic reinforcement (GR) in a
piled embankment. The British Standard BS8006 was published in 1995. In 2009, a second
(draft) version was published. Then the German Emphehlungen für Bewehrungen aus
Geokunststoffen (EBGEO) followed in 2010 and the Dutch guideline CUR 226, also in
2010. This thesis only discusses the Dutch guideline. The Dutch guideline is based on the
EBGEO. In the first paragraph the theory of the EBGEO calculation is discussed and in the
second paragraph the calculation method of the CUR 226. This thesis only discusses the
tensile forces in the GR due to the vertical load. The tensile forces due to the lateral load
fall beyond the scope of this thesis.
3.1 German method EBGEO

The EBGEO is based on the analytical solution of the stress distribution by Zaeske (2001).
Zaeske’s study consists of three parts: experiments simulating a piled embankment, a
numerical study and an analytical analysis. During the experiments a number of
parameters were varied, such as pile stiffness, embankment height, load and two geogrid
types. Zaeske’s analytical model has been validated with the test results. The vertical load
distribution of his model is explained in paragraph 3.1.1 and the determination of the tensile
force in the geosynthetic reinforcement in paragraph 3.1.2.
3.1.1 Vertical load distribution

The stress distribution model describes the vertical equilibrium of a soil element of a soil
arch in a three-dimensional situation. Figure 3.1 and Figure 3.2 shows the approach. The
arch consists of several layers, the upper arch has a radius of half the diagonal pile spacing
and the lowest arch is almost flat and has a radius of infinity.
Figure 3.1: Analytical approach of Zaeske (Zaeske, 2001)

Chapter 3 Analytical methods for the design of piled embankment with geosynthetic reinforcement
Figure 3.2: Analytical approach of Zaeske (Zaeske, 2001)
Each element should be in vertical equilibrium. The components in the analysis are:
 Pressure acting on top of arch, Fv ,t   'z  d 'z   dA0
 Bearing reaction below arch, Fv ,b   'z  dAu
 Self-weight of element, Fv ,g    dV
  
Shear friction on the four sides of the element, Fv ,s  4     dAs  sin  m 

 2 
The components of the element vertical equilibrium is given in a differential equation:
Fv  Fv ,t  Fv ,b  Fv ,g  Fv ,s  0
   (3.1)
Fv   'z  dAu   'z  d 'z   dAo  4 '  dAs  sin  m     dV  0
 2 
Where:
dAu m2 Area of the lower side of the element
2
dA0 m Area of the upper side of the element
2
dAs m Area of a side of the element
2
dV m Volume of the element
d m
m rad Average angle of part arch of element, m   
2
 ' kN/m
2
Soil stress perpendicular to sides of the element
 'z kN/m2 Effective vertical soil stress
 kN/m 3
Volumetric weight of the fill material
The solution of this differential equation gives the residual stress between the piles and is
given in equation (3.4) in paragraph 3.2.2.

21 February 2011
3.1.2 Tension load in the geosynthetic reinforcement

The vertical load on the reinforcement has to be determined to calculate the tension load in
the geosynthetic reinforcement. It is assumed that the vertical load is transported to a
reinforcement strip between the piles (Figure 3.3, left). The resulting load on the strip is a
triangular vertical load (Figure 3.3, right).
Fk
Q
Ax
w r
L0
Figure 3.3: Load distribution (left) and schematization of loading (right) on geosynthetic
reinforcement strip as assumed by EBGEO and CUR (Van Eekelen, 2010)
q(x)
V T
q(x)
x
H
θ
y(x) H + dH (
w
T+dT V+dV
L0
Figure 3.4: Schematization of the cable equation
The strip without subsoil support could be modelled as a cable equation, see Figure 3.4.
With subsoil support, however, the differential equation is more complicated and given in
equation (3.2).
d 2w q( x ) ks  w ( x )
2
  (3.2)
dx HT HT
Where:
w m Vertical deformation of the cable
x m Horizontal distance
HT kN/m Constant horizontal component of the tensile force in the cable
2
q( x ) kN/m The loading on the cable
3
The tensile force in the reinforcement is depending on the strain and the stiffness of the
reinforcement. To calculate the strain, Zaeske constructed a graph. This graph is based on
the solution of the cable equation. How the graph is used is explained in paragraph 3.2.3.
3.2 Dutch method CUR 226

This paragraph gives the design equations for a piled embankment. The equations follow
from the theory explained in the previous paragraph.

3.2.1 Geometry and boundary conditions

The geometry of the piled embankment is sketched in Figure 3.5. The parameters of the
sketch are explained below the figure.
Figure 3.5: Cross section and top view of piled embankment geometry (CUR 226, 2010)
H m Embankment height
sx m Centre to centre distance of the piles in longitudinal direction
sy m Centre to centre distance of the piles in cross direction
Ai m 2
Zone of influence per pile, calculated with Ai  sx  sy
s m Centre-to-centre distance of the piles in diagonal direction
Ap 2
m Pile cap area
a m Diameter of a circular pile cap
aeq m Equivalent diameter of a rectangular pile cap, calculated with aeq  4  Ap / 
z m Distance between the pile cap and geosynthetic reinforcement
In the embankment of the scale test arching will develop. For piled embankments there are
a number of boundary conditions as described below in Table 3.1.
Description Parameter Criteria

Embankment height H H /(s  a)  0.66
Boundary condition for arch reduction due to heavy q /  v ; t  0 .5
dynamic load
Relation between pile cap and pile distance a/s a / s  0.15
Maximum distance between first layer of
z z  0.15
reinforcement and pile cap
Ratio between the centre-to-centre distance of the s x / sy 2  s x / sy  3 2
3
piles in x and y directions
Embankment fill consists of granular material with a    35
internal friction angle
Minimum strength of geosynthetic reinforcement Tr ;d Tr ;d  30
Ratio between the strength of the geosynthetic Tr ;d 0.1  Tr ; x ;d / Tr ;y ;d  10
reinforcement in x and y direction
Relation of stiffness between pile and subsoil ks ; paal / ks ks ; paal / ks  10
Table 3.1: Boundary conditions piled embankment

21 February 2011
3.2.2 Vertical load distribution

The total vertical stress on the level of the geosynthetic reinforcement is:
v ;t    H  g  q (3.3)
The vertical stress on the geosynthetic reinforcement can be calculated with the solution of
the differential equation (3.1) described by Zaeske based on arching. The solution is
presented in equation (3.4).
  hg2  2 


 p
   
 
 stat
 1        H  1  hg2  2
 
 hg   1    1  hg  2
2  (3.4)
v ;g
 H    4  
  
With:
1
1    s  a 
2
(3.5)
8
s 2  2  a  s  a2
2  (3.6)
2  s2
  '
K krit  tan 2  45   (3.7)
 2
a   K krit  1
 (3.8)
2  s
Where:
 v ;t kN/m
2
Total vertical load on the level of geosynthetic reinforcement
p kN/m
2
Vertical load (permanent +variable) on the embankment, calculated by p  g  q
2
q kN/m Variable load (traffic load) on the embankment
2
g kN/m Permanent load on the embankment
 v ;g kN/m
2
Vertical stress on the level of geosynthetic reinforcement between the pile caps
 kN/m3 Volumetric weight of the embankment fill
Kkrit - Critical ratio of the principal stresses
hg m Height of the embankment for full arching, H
The vertical stress on the pile caps based on arching is calculated with:
 vstat 
; p     H  p    v ;g 
Ai
Ap

  v ;g (3.9)
Where:
 vstat
;p kN/m2 Total vertical load on the level of geosynthetic reinforcement
2
Ai m Zone of influence per pile
Ap m2 Pile cap area
3.2.3 Tensile load in geosynthetic reinforcement

The tensile force is calculated with equation (3.10). To use the equation the stiffness of the
reinforcement and maximal strain have to be determined. The stiffness of the geosynthetic
reinforcement is determined from the isochrone curves, that describe the strain/stress
relationship of the geosynthetic reinforcement. These isochrone curves should be provided
by the manufacturer. The maximal strain can be derived from the design graph given in
Figure 3.6,. The graph is a solution of differential equations, solved and presented
graphically by Zaeske (2001).

Tv    EAg (3.10)
Where:
Tensile force in the geosynthetic reinforcement caused by the vertical load due
Tv kN/m
to arching
 - Average strain in the geosynthetic reinforcement
J  EAg kN/m Stiffness of geosynthetic reinforcement
k s  L2w
EAg
F beq
EAg
Figure 3.6: Maximum strain in the geosynthetic reinforcement (CUR 226, 2010)
To use the design graph there are two dimensionless parameters that have to be
calculated. These parameters are:
F
beq
Located on the x-axis (3.11)
EAg
k s  L2w
Located in the curve (3.12)
EAg
Where:
F kN Vertical force on the reinforcement strip between two piles
beq m Equivalent width of the pile cap
EAg kN/m Stiffness geosynthetic reinforcement
3
ks kN/m Modulus of subgrade reaction, see paragraph 2.1
Lw m Clear width between the pile caps

21 February 2011
The vertical load F on the reinforcement due to arching is determined by using equation
(3.13). In the equations, the zone of influence is used. The principle of the zone of influence
is shown in Figure 3.7.
1 a2 s   
Fx  ALx   v ;g     s x  sy    arctan  y   
2 2 180  v ;g
  s x  
(3.13)
1 2 s   
Fy  ALy   v ;g     s x  s y  
a
 arctan  x    
2 2  s  180  v ;g
  y  
Where:
Fx ; y kN Vertical force on a geosynthetic strip between two piles
2
ALx ;Ly m Zone of influence, defined in Figure 3.7
 v ;g kN/m
2
Vertical load on a geosynthetic strip between two piles
sx m Centre to centre distance of the piles in longitudinal direction
sy m Centre to centre distance of the piles in cross direction
Figure 3.7: Vertical force on the GR between the piles (CUR 226, 2010)
The equivalent width of the pile cap is calculated with:
1
beq  a  (3.14)
2
Where:
beq m Equivalent width of the circular pile cap (or the width of the square pile cap)
a m Diamater of the circular pile cap
The clear width between the pile caps is calculated with:
Lwx  s x  beq
(3.15)
Lwy  sy  beq

21 February 2011
Chapter 4
Numerical methods for the design of piled

embankment with geosynthetic reinforcement
In practise, numerical analysis is used to calculate the deformation of the soil and the
moments in the piles. The deformation and pile loads are important for the deformation
requirements and the dimensions of the pile.
There are several programs that can perform a numerical analysis, for example FLAC and
Plaxis. To compare the result of the performance of a piled embankment, some definitions
are described in paragraph 4.1. The experience with numerical analysis in the literature is
discussed in paragraph 4.2. For this thesis the experience with Plaxis is applicable,
therefore the results of the experiences with Plaxis found in the literature is described in
more detail in paragraph 4.3.
4.1 Definitions
In order to evaluate the performance of a piled embankment construction, some definitions
have been used.
Axisymmetrical geometry
In an axisymmetrical geometry one pile is modelled with the subsoil and embankment fill.
The existing geometry is schematized to a cylinder with the centre of the pile as central
axis. The axisymmetrical model is illustrated in Figure 4.1.
Figure 4.1: Axisymmetrical geometry
Plane strain geometry

With a plane strain geometry the pile is transformed to a continuous wall with an equivalent
thickness (Figure 4.2). The 2 dimensional cross section B-B is transported perpendicular of
the cross section to infinity. Due to symmetry, only half of the geometry is used in the
calculations.

Chapter 4 Numerical methods for the design of piled embankment with geosynthetic reinforcement
Figure 4.2: Plane strain geometry
3 Dimensional geometry
With a 3D geometry the actual situations can be modelled as shown in Figure 4.3.
Figure 4.3: 3 Dimensional geometry
Efficacy
The efficacy E of the pile support is defined as the proportion of embankment weight
carried by the pile caps. Efficacy increases as the effect of arching increases. Efficacy is
defined as (Hewlett et al., 1988):
P
E (4.1)
s H
2
Where:
E - Efficacy
P kN Total force carried by the pile caps
 kN/m 3
Volumetric weight of the embankment
H m Height of embankment
s m Pile centreline spacing
Stress Reduction Ratio

The term Stress Reduction Ratio is used by Kempton et al. (1998) to describe the degree
of soil arching. It is the ratio of the average vertical stress remaining to be carried by the
subsoil and/or reinforcement after arching has occurred. A stress reduction ratio of 1.0
implies no arching, the stress reduction ratio reduces to zero when the effects of arching
increases. The term soil arching is the same as the term stress reduction ratio. The stress
reduction ratio is defined by equation (4.2) and the description of the symbols is depicted in
Figure 4.4.
pb
 (4.2)
 H  q0

21 February 2011
Where:
 - Stress reduction ratio
 kN/m 3
Volumetric weight of the embankment
q0 kN/m2 Uniform surcharge on the embankment
2
pb kN/m Average vertical pressure above geosynthetic
Figure 4.4: Load transfer mechanisms of piled embankments
Stress Concentration Ratio

The Stress concentration ratio is the ratio between the stress on the pile caps and the
stress on the subsoil/reinforcement. This value also increases as the effect of arching
increases. The stress concentration ratio is defined by:
c
n (4.3)
s
Where:
n - Stress concentration ratio
c kN/m 2
Vertical stress on the pile caps
s kN/m 2
Vertical stress on the soil
4.2 Experience numerical design

For the numerical analysis of piled embankments, experiences with several numerical
programs have been reported in the literature: FELSTA, FLAC, Plaxis and ABAQUS. These
programs use numerical techniques to solve partial differential equations and integral
equations. This section describes the experiences with these programs.
4.2.1 Experience numerical design with Plaxis

Plaxis is based on the finite element method (FEM) (Plaxis). It is designed to perform
geotechnical analysis of deformation and stability of soil structures. The experience
described in the literature is listed in this section.
A project where Plaxis was used to model the interaction between soils and various
structural components is City West Link Road Section 3, New South Wales, Australia.
Timber piles were used in a transition construction between the bridge and the
embankment. Next to that, settlement plates were installed to monitor the performance of
the timber-piled embankments. The with plane strain Plaxis calculated and measured

forces in the structures and deformation of the subsoil were compared and correspond well
(Hsi, 2001).
In the CUR 2002-7 two projects are analyzed in Plaxis 2D, Bus lane ‘Monickendam’ and
Second Severn Crossing. The projects are modelled with axisymmetrical and plane strain
geometries. Because these geometries does not resemble the reality, the results of the
calculation or the parameters of the calculation are adjusted. To adjust the results of the
calculation is not advised, because a large amount of engineering judgement is needed. It
is advised to apply factors on the input parameters. The validation of both the projects is
limited.
Gangakhedkar (2004) used both 2D axial symmetric and 2D plane strain finite element
analysis to study effects of certain factors like pile modulus, stiffness of the geosynthetic
reinforcement, height of the embankment and the effect of the soil layer directly below the
geosynthetic. The soft soil model in Plaxis is used to represent the weak foundation soil.
The embankment fill was modelled with the Mohr Coulomb model. A fully bonded interface
between soil-pile and soil-geosynthetics was assumed, because the influence of the
interfaces was estimated to be small due to the small deformations. The geosynthetics
endures creep which results in an increase in the strains. To a certain extent, this will cause
a reduction in the tensile strength. Updated mesh is used to consider the influence of
change in the geometry of the model caused by large deformation of the soft soils.
Gangakhedkar carried out numerical analysis with Plaxis and analytical analysis for five
case histories. The maximum settlements at the pile cap decrease when the tensile
stiffness of the geosynthetics is increase or the height of the embankment is decreased.
The increase in the pile modulus causes an increase in the difference of the stiffness
between the pile and soil. This causes more differential settlements and eventually
increases the tensile strength in the geosynthetics. These effects influence the stress
concentration ratio. The stress concentration ratio increases by increasing the modulus of
the pile or the tensile stiffness of the geosynthetic layer.
Slaats (2008) modelled the scale test performed by Zaeske (2001) in Plaxis 2D and Plaxis
3D Foundation. The soil is modelled with the Hardening Soil model. The results of Plaxis
2D and Plaxis 3D are almost equal for the vertical force on the pile caps. The results of
Plaxis are compared to the result of the scale test performed by Zaeske. The vertical force
on the pile caps calculated with Plaxis is overestimated, and thus the load on the
reinforcement is underestimated. The settlements on top of the embankment were
underestimated with Plaxis. Next to that, a piled embankment is modelled in Plaxis 2D to
perform a parameter study. The parameter study was not performed in Plaxis 3D
Foundation, because it had difficulties to generate a mesh of a soil profile including a slope.
Satibi (2009) investigated the influence of geometrical idealizations, the influence of the
stiffness of geosynthetic reinforcements, the influence of capping ratio (area pile cap / area
influence zone) and the influence of constitutive models on numerical analysis of soil
arching in a piled embankment. For the analysis Plaxis is used. The piled embankment
tests performed by Zaeske (2001) are used as a reference case study. The Hardening Soil
model is applied for modelling the material constitutive behaviour of the embankment fill
and the peat underneath. The results described in the report by Satibi (2009) can be found
in paragraph 4.3.
In the Dutch design guideline, CUR 226 (2010), there is a description of how pile moments
and deformations could be calculated with a finite element program. Case studies of low
and high embankments have been analysed. The case study for low embankments is a
piled embankment in Houten and the case study for high embankments are the large scale
tests done by Farag (2008). The large scale tests were performed to investigate the lateral
spreading forces in the embankment. For the piled embankment in Houten the
embankment fill and subsoil were modelled with the Hardening Soil model. The piles are
modelled as beams. The embankment fill of the large scale test is also modelled with de

21 February 2011
Hardening Soil model. The soft subsoil is modelled linear elastic, because the subsoil is a
foam cushion. The piles are modelled as volume elements with non porous linear elastic
behaviour.
4.2.2 Experience numerical design with other programs

The experiences with other numerical programs are described in this paragraph.
FELSTA
The first experience with numerical analysis has been reported by Jones et al. (1990). They
analysed the influence of several parameters on the actual tensile forces in the
geosynthetic reinforcement with FELSTA (Finite Element Structural Analysis). In the
analysis they used the hyperbolic stress-strain relationships described by Duncan & Chang
(1970). They show that support of the soft soil can decrease the tensile force in the
geosynthetic considerably.
ABAQUS
Liu et al. (2007) performed numerical analysis of a case history with ABAQUS. ABAQUS is
finite element software. The case history is a geogrid-Reinforced and Pile Supported
Highway Embankment in a northern suburb of Shanghai, China. The pressures acting on
the piles and the soil surfaces between piles, pore-water pressures, settlements and lateral
displacements are measured in the field. The field measurements are compared with the
results from three-dimensional fully coupled finite element back analyses. The embankment
fill was modelled using a linear elastic-perfectly plastic model with Mohr-Coulomb failure
criterion. The pile was modelled as isotropic linear elastic material and the subsoil with
Modified Cam Clay. The measured pressures on the pile and the computed values differ
7%. In the journal is concluded that field loading tests on the foundation clay show that
bearing capacity of the soft soil supported with piles can be improved by at least a factor of
3. The rates of consolidation are slightly over predicted by the 3D analysis, however the
computed values agree reasonably well with the field measurements.
FLAC
Another program that is used to perform numerical analysis is FLAC (Fast Lagrangian
Analysis of Continua). FLAC is based on finite difference method (Itasca). Because no
matrices are formed, large 3D calculations can be made without excessive memory
requirements.
Russell & Pierpoint (1997) analysed two case histories numerically to compare several
design models. The two case studies were: the A13 piled embankment and the Second
Severn Crossing trial embankment. The numerical analysis was performed with the three-
dimensional finite difference program FLAC3D. In 3D numerical analysis, the arching
mechanism was assumed to be a dome, resting on four pile caps. The embankment fill was
modelled with the linear elastic Mohr-Coulomb model. The reinforcement consisted of one-
dimensional linear cable elements. Two methods of constructing the embankment were
analysed: construction of the embankment in one step or construction stepwise. It is shown
that higher maximum settlements, stress on the pile cap and geosynthetic tensions were
found when the embankment is constructed in one step. The construction of the
embankment in one layer resembles reality most. The results of the numerical analysis also
show that large deformations occurred at the base of the embankments although the
differential settlements at the surface were negligible. Russell & Pierpoint noted that the
reinforcement tensile forces were concentrated in the reinforcement elements spanning
directly between the pile caps. The highest tensions were found at the edge of the pile
caps.
Kempton et al. (1998) compare the load transfer and settlements behaviour between 2D
plane strain and 3D analyses of a quarter cell embankment. The numerical analyses is
performed using FLAC. The results show that although the tendency is the same, the
magnitudes are considerably different. The stress reduction ratio with respect to H/s are

much lower in 3D analyses compared to 2D. Higher maximum and differential settlements
are observed in 3D analyses.
A numerical study has been performed with FLAC to investigate the influence of the height
of the fill, the tensile stiffness of geosynthetic and the elastic modulus of pile material (Han
et al., 2002). The carried out axial symmetric calculations on one pile and its ‘effective’
equivalent circle. For the soil and the fill, the nonlinear hyperbolic elastic model described
by Duncan and Chang (1970) is applied. The piles and geosynthetic are both considered
linear elastic materials. The interfaces between piles and soil and geosynthetics and soil
are considered less important, because of the small deformations. They are, also for
simplicity reasons, assumed fully bonded. The study indicated that the stress reduction
ratio decreases with an increase in the height of embankment fill, an increase in the elastic
modulus of the pile material, and a decrease in the tensile stiffness of geosynthetic. The
results also suggest that the stress concentration ratio and maximum tension in
geosynthetic increase with increasing the height of embankment fill, increasing the stiffness
of geosynthetic, and increasing the elastic modulus of the pile material. It is also shown that
the maximum tensile force occur near the edge of a pile.
4.3 Results of experience with Finite Element Method using Plaxis

The results of experiences with Plaxis are described in this paragraph regarding the
geometry, geosynthetics, constitutive models and parameters. The experience is based on
the reports of Satibi (2009) and the CUR 226 (2010). Satibi used the scale tests performed
by Zaeske (2001) as a reference case. Satibi applied the Hardening Soil model for
modelling the behaviour of the embankment fill and the soft subsoil. This model is used,
because it can simulate both stiff and soft soil behaviour. It concludes stress-dependent
stiffness and yielding formulations for simulating volumetric compression and shear
hardening. Only when the influence of the constitutive model is researched different models
are applied.
4.3.1 Geometry
Numerical analysis of soil arching in piled embankments can be done with different
geometries(Satibi, 2009): axisymmetrical geometry, plane strain geometry and 3D
geometry. The geometries are explained in paragraph 4.1. Axisymmetrical analysis will
produce an ‘umbrella’ shaped arch resting on a single central pile cap (Kempton, et al.,
1998a). Plane strain analysis will produce half-tube type arching. The analysis of a 3D
geometry is the best representation of reality. However, it requires much more effort and
computer power.
For the analyses of the geometries,

Satibi used the scale tests without
geosynthetic reinforcement performed
by Zaeske (2001) as reference case.
An axisymmetrical geometry gives the
same results as the 3D geometry for
the load on the pile, embankment
settlements and the height of the arch
in the piled embankment. The results of
the load on the pile corresponds with
the measurements of the case study.
This can be seen in Figure 4.5. On the
other hand, the plane strain geometry
does not give the same results as the
case study. The plane strain geometry
leads to soil arching instability through Figure 4.5: Calculated and measured load on
punching failure. Punching failure is pile (Satibi, 2009)
that the pile is penetrating into the
embankment instead of supporting it. With plane strain geometry the actual squared pile

21 February 2011
transforms into a thin wall. Soil arching stability can be recognised by plasticity conditions
and shear strain at the soil stress points in the embankment.
As described in CUR 226 (2010) the piles can be modelled as springs, beams or volume
elements. The preference is to model the piles as beam elements, because the moments
and shear forces can be observed directly from the FE program. The results of a 2D model
are per meter and have to be translated to a pile depth, because in a 2D model the pile is
modelled as a wall. To translate the results there are a number of factors that have to be
taken into account: subsidence behaviour of the pile (wall), the behaviour of the soil
between the piles, the stiffness of the pile and the calculated deformations. These factors
are described in paragraph 5.2.2 of the CUR 226 (2010).
4.3.2 Geosynthetics
Satibi (Satibi, 2009) also investigated the influence of applying geosynthetic reinforcement.
The use of geosynthetic reinforcements reduces the embankment settlements and
increases the load on the pile. Also the soil arching stability is improved, whereby the
occurrence of punching failure is reduced. When an additional layer of geosynthetics is
applied, it should be placed in the lower
half of the arching height. In the lower half
of the arch there is a larger differential
settlement, which results in larger tensile
forces in the geosynthetics. In the finite
element calculations it is necessary to
update the geometry of the mesh during
incremental loading. When the mesh is
updated the tensile strains in the
geosynthetic elements are captured and
the geosynthetic is no longer horizontal.
The calculated tensile forces correspond Figure 4.6: Calculated and measured tensile
with the measured tensile forces in the force in the geosynthetic reinforcement
reference case as shown in Figure 4.6. (Satibi, 2009)
The CUR 226 (2010) recommends to validate the finite element calculations of the tension
force in the geosynthetic reinforcement with the analytically calculated tension force. Farag
(2008) found a factor of 3 - 3.5 between the by Farag measured and numerically calculated
maximum strain in a 3D FE program. CUR 226 showed that this factor can be reduced to
0.85 - 0.9 by modelling an air gap under the geosynthetic reinforcement and between the
pile caps. Modelling this air gap has no significant effect on the pile moments. The vertical
deformations are twice as large when an air gap is applied.
4.3.3 Constitutive models

Satibi (2009) considers three constitutive models for modelling the embankment fill: Mohr
Coulomb model (MC), Hardening Soil model (HS) and Hardening Soil Small model (HS-
Small). The constitutive models for the piles and subsoil are not changed. Figure 4.7 shows
the calculated curves and measured curve of applied uniformly distributed external load
against average embankment surface settlement from different soil models. The load
settlement behaviour of the HS and HS-Small model are closer to the measured data then
the MC model.

Figure 4.7: Load-settlement curves from analyses with different soil models compared
with measurements of the scale tests performed by Zaeske (2001) (Satibi, 2009)
Calculated and measured loads on the pile are similar, as shown on the left side of Figure
4.8. This does not apply to the tensile force in the geosynthetic reinforcement. As shown on
the right side of Figure 4.8, the result of the constitutive models are not similar to the
measured data. Satibi recommends to use an advanced soil constitutive model such as the
HS model, because it allows a more realistic choice of soil parameters, gives a better
approximation of embankment settlement compared to reality and a clear identification of
soil arching stability.
Figure 4.8: Calculated and measured data of the load on a pile (left) and data of the
tensile forces in the geosynthetic reinforcement (right) (Satibi, 2009)
Figure 4.9: Influence of the friction angle ’ Figure 4.10: Influence of the dilatancy angle
on subsoil surface settlements at the pile  on subsoil surface settlements at the pile
centre (Satibi, 2009) centre (Satibi, 2009)

21 February 2011
4.3.4 Parameters
Using the HS model involves several soil parameters. To understand the influence of these
parameters, a sensitivity analysis is performed by Satibi (2009). Figure 4.9 shows that the
internal friction angle has a significant influence on the settlements of the subsoil surface. A
higher internal friction angle will lead to less settlements in the embankment. The influence
of the dilatancy angle is less significant, as shown in Figure 4.10.
The HS model uses three stiffness

parameters, which are the secant
ref
deviatoric loading stiffness E50 ,
ref
oedometer loading stiffness Eoed and
ref
unloading-reloading stiffness Eur . Satibi
concludes that among these stiffness
parameters, the oedometer loading
stiffness is the only stiffness that shows a
noticeable influence on Satibi’s calculation
results. It influences the differential
settlements at the subsoil surface. In Figure 4.11: Influence of the oedometer
Figure 4.11 the differential settlements at loading stiffness on subsoil surface
the subsoil surface reduce as the settlements at the pile centre (Satibi, 2009)
oedometer loading stiffness increases.
The influence of the oedometer loading stiffness is particularly noticeable for values lower
than 30 MPa. Satibi also researched the influence of the power law parameter m and the
Poisson ratio. The power law parameter regulates the stress level dependent stiffness of
the soil, however does not influence the embankment settlements. The variation of the
unloading-reloading Poisson’s ratio shows no influence on the calculated settlements and
the load on pile. This is not the case for the Poisson’s ratio of the Mohr Coulomb model,
where the influence of Poisson’s ratio is high.
4.4 Conclusion
For this report the experience with Plaxis is most applicable. The experiences with Plaxis
are summarized in Table 4.1 and discussed in this paragraph.
Numerical analysis can be performed with plane strain geometry, axial symmetric geometry
or 3D geometry. Arching is a 3D phenomenon and therefore it is obvious to use the 3D
geometry. However, a piled embankment can also be modelled in an axial symmetrical
geometry. This gives the same results as a 3D geometry. According to Satibi (2009) a
plane strain geometry cannot be used to model soil arching, because soil arching instability
occurs by punching failure. Slaats (2008) found that the results of 2D plane strain and 3D
geometry do correspond. Experience done by Kempton et al. (1998) showed that the
results of 2D plane strain and 3D geometry have the tendency to be the same, however the
stress reduction ratio is much lower in 3D analysis compared to 2D analysis.
In numerical analysis of piled embankments, different soil models have been applied. There
is a difference in the load transfer mechanism as different soil models use different
assumptions for stiffness and yielding behaviour of the soil. For example, the Mohr-
Coulomb model contains one stiffness for the entire soil. This produces different stress
fields due to arching compared to advanced soil models. The advanced soil models
consider stress level dependency. Satibi (2009) concluded that applying the Hardening Soil
model for the embankment fill gives the most realistic average surface settlements and
arching effect. The model that gives the most realistic results for the soft subsoil is not
discussed in the literature. Mostly the Hardening Soil model is used, but this model does
not consider creep effect. Creep can have significant effect on the pile moments, therefore
the Soft Soil Creep model should be applied.

The influence of soil parameters on the settlements is consistent with the influence on the
load transfer. Higher soil strength parameters and soil stiffness leads to lower values for the
maximum and differential settlement at the embankment surface.
In the literature, it is demonstrated that the performance of the embankment can be

influenced by adjusting different parameters. Higher maximum settlements, higher stress
on the pile cap and larger geosynthetic tensions are reached when the embankment is
constructed in one-step compared to construction it stepwise. The tensile forces in the
reinforcement are concentrated in the reinforcement elements spanning directly between
the pile caps. The highest tensile force can be found on the edge of the pile caps. It is
indicated that the stress reduction ratio decreases by increasing in the height of the
embankment, increasing the elastic modulus of the pile material, and decreasing the tensile
stiffness of the geosynthetic reinforcement. Besides this, the stress concentration ratio
increases by increasing the height of the embankment, the stiffness of the geosynthetics
and the elastic modulus of the pile material.
Compared to analytical
Embankment fill model
or measured results
Subsoil model
Pile model
Geometry
Literature
Interface
Program
Details
Hsi (2001) Plaxis 2D-PS - - - - Measured -
data
CUR 2002-7 Plaxis 2D-AS HS SS LE - Analytical Updated
(2002) 2D-PS mesh
Gangakhedkar Plaxis 2D-AS MC SS - Soil-pile Analytical Updated
(2004) 2D-PS Soil-GR data mesh
Slaats (2008) Plaxis 2D-PS HS HS LE Soil-pile Scale test Updated
3D-FD Soil-GR Zaeske mesh
Satibi (2009) Plaxis 2D-AS HS HS LE Soil-pile Scale test Updated
2D-PS MC Zaeske mesh
3D-FD HSS
CUR 226 Plaxis 3D-Tun HS
LE LE Soil-pile Scale test Updated
(2010) Farag mesh
CUR 226 Plaxis 3D-Tun HS HS - Soil-pile Project Updated
(2010) Houten mesh
Table 4.1: Summary of experience with Plaxis
Legend
Program Models Interface
2D-AS: 2D axial symmetric LE: Linear elastic Soil-pile: interface between soil and pile
2D-PS: 2D Plane strain MC: Mohr Coulomb Soil-GR: interface between soil and geosynthetics
3D-FD: 3D Foundation HS: Hardening Soil model
3D-Tun: 3D Tunnel SS: Soft Soil model
HSS: Hardening Soil Small model

21 February 2011
Chapter 5
Scale test
To understand the behaviour of arching in piled embankments a scale test has been
developed by Deltares. This test is based on the tests performed by Zaeske (2001) and
Heitz (2006).
Twelve tests were performed with different settings. These differences lie in the type of
geosynthetics, the amount of layers of geosynthetics and the type of embankment fill. This
chapter describes the global geometry, set up and measuring instruments of the scale test.
5.1 General description scale test

A symmetrical part of a piled embankment is modelled in the scale test. In Figure 5.1 a
piled embankment with the section for the test set-up is schematized. The section is taken
in the centre of the embankment where there is no lateral load. Lateral load occurs at the
edges of the embankment and influences the load distribution. The section of the
embankment is modelled in a metal box where lateral displacement cannot occur. In Figure
5.2 and Figure 5.3 a top and side view of the scale test are given.
Figure 5.1: Section used from a piled embankment
The metal box consists of three parts which can be separated: bottom box, top box and
cover. Four piles are situated in the bottom box. Between the piles, a foam cushion is
placed on a steel plate. Geosynthetic reinforcement is attached to a steel frame and
situated on top of the foam cushion. Depending on the type of test, one layer or two layers
of geosynthetic reinforcement are applied. Two layers of geosynthetic reinforcement are
separated by a layer of granular material. On top of the geosynthetic reinforcement, an
embankment is constructed of granular material or sand. The geosynthetic reinforcement
and the embankment are situated in the top box. On top of the embankment, the water
cushion and a steel plate are applied. The rest of the space is filled up with distance

Chapter 5 Scale test
keepers and the box is closed off by a cover. A number of steel screw threads and bolts are
applied to enclose the box.
0.225
Frame for geosynthetic
reinforcement
0.10
Pile 3 Pile 1
Top load water cushion
0.45
Embankment fill
0.42
Pile
0.10
Pile 2 Pile 4
0.225
Foam cushion Tap
0.10 0.10 0.10

0.10
0.225 0.45 0.225 0.225 0.45 0.225
Figure 5.2: Side view test set-up Figure 5.3: Top view test set-up
5.2 Geometry & Set-up

A piled embankment consists of a number of elements: piles, subsoil, geosynthetic
reinforcement, embankment and traffic load. These elements have to be scaled and
modelled to make the test more practical. The following paragraphs give a description of
each of these elements.
5.2.1 Piles
In the Netherlands the centre-to-centre distance of the piles in a piled embankment is
between 0.9 and 2.5 meter. The piles have a diameter of 15 to 32 cm with a larger pile cap
on top. The centre-to-centre distance of the piles in the scale test is scaled to 0.55 m. This
brings the scale to approximately 1:3. The diameter of the piles is scaled to 10 cm. The
geometry of the piles is summarized in Table 5.1. With the diameter of the piles and the
centre-to-centre distance the inside size of the steel box becomes 1.1 meter.
Parameter Symbol Dimension

Centre-to-centre distance piles sx, sy 0.55 m
Diagonal centre-to-centre distance piles s 0.78 m
Diameter piles d 0.10 m
Area piles Ap 0.0079 m2
Equivalent width square piles aeq 0.089 m
Diagonal distance between piles s-d 0.68 m
Table 5.1: Geometry piles
Material Density [kg/m3] Specific weight [kN/m3] E modulus [kN/m2]

Hard-PVC 1390 13.64 2943000
The four piles are placed in the bottom box of the test set-up and go through the bottom
plate. The piles are constructed of solid hard PVC with a height of 18.6 cm. Two piles are
lower to compensate the thickness of the pressure cells on top of the piles. Figure 5.4
shows the piles.

21 February 2011
Figure 5.4: Piles in test set-up
5.2.2 Subsoil
Normally a piled embankment is applied in soft soils. Using these soils in a test set-up has
a number of disadvantages. A load on soft soil can cause plastic deformation. When
plasticity has occurred, the soil properties are changed and the soft soil cannot be used
again. Furthermore, soft soil has a long consolidation time, which has effect on the
execution time of the scale test. Therefore, the subsoil is schematized with a watertight
foam cushion that does not have these disadvantages. An advantage is that the foam
cushion can be reused for each following experiment and the consolidation time of the
foam cushion is shorter. In addition, the consolidation process can be controlled. This
means that the consolidation process can be started and stopped on demand. For
example, a load can be applied on the embankment without causing consolidation directly.
Another advantage of the foam cushion is that the subsoil support can be reduced to zero.
A disadvantage of a foam cushion is that it has different characteristics compared to real
soil.
The foam cushion lies in the bottom part of the metal box and is attached with a tap to a
burette to control the consolidation process. The consolidation process starts when the tap
is opened and the water in the cushion can flow to the burette. The last step of the scale
test is to reduce the subsoil support to zero by applying vacuum on the burette. To protect
the top of the foam cushion against sharp material a geosynthetic filter is applied. The
friction between the cushion, the piles and the box walls is reduced by applying Vaseline or
an oily substance in combination with plastic sheets.
The first foam cushion, made by Deltares, was painted with latex to make it watertight. The
latex was fixed directly on the foam. During the painting, the cushion was resting on two
supports causing a hardly visible deflection. This resulted in an inclined pile gap, which
needed to be compensated by adapting the pile. Due to leakage the foam cushion was
replaced by a waterbed company manufactured foam cushion. Compared to the first
cushion the second cushion is wrapped in a plastic sheet and not fixed to the foam. The
pile that was adapted to an inclined position is corrected to a straight position. Figure 5.5
shows pictures of the foam cushions.
Figure 5.5: First cushion (left) and second cushion (right)

The foam and the wrap of the cushion have a certain stiffness which is determined by a
water-test. Water is applied to simulate a uniform load. The advantage of water is that there
are no shear forces, so arching will not occur.
During several load steps, water is pressed out of the foam cushion. For each load step an
equilibrium is reached when the water stopped draining from the foam cushion. The
amount of drained water is equal to the average volume change. The change in load steps
and volume determines the stiffness of the foam cushion. In Appendix A the calculation of
the stiffness is elaborated. The average value of the stiffness of the first foam cushion is 48
kN/m3. The water test was not performed on the second foam cushion. The foam of the first
and second cushion are the same. Therefore the average stiffness of the second foam
cushion is estimated to be in the same range as the average stiffness of the first foam
cushion.
5.2.3 Geosynthetic reinforcement

It is impossible to scale the geosynthetic reinforcement, simply because a scaled version is
not available. To get enough settlement and strain in the geosynthetic the pile diameter is
relatively small. The geosynthetic reinforcement has one condition: the grid spacing has to
be small enough for several grid ribs to overlap the pile. Otherwise there are not enough
grid ribs to distribute the tension to the piles with breaking of the geosynthetics as a result.
For the tests, the geosynthetic reinforcement is attached to a steel frame. In the tests, one
layer or two layers of geosynthetic reinforcement are applied. The steel frame and the
stiffness of the geosynthetic reinforcement are described below.
The steel frame is designed to prevent horizontal displacement of the geosynthetic

reinforcement. The frame can move freely in vertical direction with the modelled subsoil.
The design of the steel frame is illustrated in Figure 5.6. The geosynthetic reinforcement is
folded around the tube and pressed between the strips with nuts and bolts.
Figure 5.6: Design steel frame Figure 5.7: Set-up geosynthetic

reinforcement
The geosynthetic reinforcement is sensitive for the sharp edges of the piles. Therefore a
sand layer of 2 cm is applied on the foam cushion and piles to protect the geosynthetic
reinforcement, see Figure 5.7. When two layers of geosynthetic reinforcement are applied,
a layer of 5 or 10 cm of embankment fill divides the geosynthetic reinforcement layers.
The stiffness of the geosynthetic reinforcement is determined according DIN EN ISO 10

319. In Table 5.2 the average tensile stiffness J2% is given for the geosynthetic material
used during the 12 scale tests. The tensile stiffness is determined at the first 2% strain and
the average of five tensile stiffness tests.

21 February 2011
Sequence Geosynthetic Stiffness J2% Distance between GR

reinforcement [kN/m] [cm]
1 N1 2 laid grids biaxial 2090 5
2 N2 1 laid grid biaxial 1045 -
3 T1 1 woven geotextile 670 -
4 K1 1 woven grid biaxial 2399/2904 -
5 T2 1 woven geotextile 970/1810 -
6 K2 2 woven grids uniaxial 2269 0
7 T3 1 woven geotextile 970/1810 -
8 S1 2 extruded grids triaxial 757 10
9 T4 1 knitted grid biaxial 1715/1742 -
10 K3 2 woven grids uniaxial 2269 5
11 S2 2 extruded grids triaxial 757 10
12 N3 2 laid grids biaxial 2090 5
Table 5.2: Geosynthetic reinforcement
5.2.4 Embankment fill

An embankment is built up from sand or granular material, see Figure 5.8 and Figure 5.9.
The embankment is scaled to 42 cm layer of sand or granular material. The geometry is
summarized in Table 5.3. Most of the tests were performed with granular material. The
embankment is constructed in several layers of 10 cm and every layer is compacted with a
steel weight. See Figure 5.10. The friction between granular material or sand and the box
walls is reduced with a rubber sheet and Vaseline or an oily substance.
Parameter Symbol Dimension

Minimal height of embankment according to CUR226 0.66(s-d) 0.44 m
Height of embankment in tests 0.42 m
Table 5.3: Geometry embankment
Figure 5.8: Itterbeck sand
Figure 5.9: Granular material Figure 5.10: Compaction with steel weight
During the scale tests Itterbeck sand with a fraction of 125-250 µm is used. The fraction of
the granular material is based on research from centrifuge model tests. Centrifuge model
tests have proven that the grains have to be 20-40 times smaller then the pile diameter to
cause a deformation pattern comparable to reality. The fraction of the granular mater is 0 -
16,5 mm. The sieve curves are presented in Appendix B.
5.2.5 Load
Load induced by traffic is dynamic. Studies done by Heitz (2006) show that the load
distributions for dynamic loading and static loading are different. Dynamic loading can
cause improvement of arching, as well as degradation of the arches. This is mainly
depending on the degree of reinforcement. However, to simplify the scale test the traffic

load is modelled as a static load. The static load is applied in the scale test by a water
cushion that results in an equally distributed top load.
A waterbed manufacturer has fabricated the water

cushion, see Figure 5.11. A sheet on the embankment
and a wooden plate on top of the cushion protect the
water cushion during the tests. The wooden plate fits
the metal box exactly to prevent that the water cushion
moves between the wooden plate and the metal box.
On top of the wooden plate a steel plate is applied.
During the scale test, the load on the embankment is
applied by water pressure in the water cushion. The
water pressure is controlled with a burette. In Figure
5.12 the location of the burette for the water cushion is Figure 5.11: Water cushion
presented.
Burette for pressure

water cushion
Burette for discharge

Metal box foam cushion
Figure 5.12: Test set-up
5.2.6 Scaling effect

The scaling has influence on different parameters of the test. For example, the length is 1/3
times smaller in the scale test then in the reality. This means that the area is (1/3)2=1/9
smaller. The stress that is measured in the test is the same as in reality. In Table 5.4 for a
number of parameters the scaling is given. In Table 5.5 the real dimensions of the
embankment are calculated based on the dimensions of the scale test. The scaling has no
effect on the calculations with the design guidelines.
Parameter Dimension Scale 1:x

Length L m 1:x
Area A=LxL m2 1:x2
Stress  =F/A kN/m2 1:x
Force F= xA kN 1:x2
Tensile stiffness geosynthetic reinforcement F/L kN/m 1:x
Deflection/distance between the piles m/m 1:1
Table 5.4: Scaling effect

21 February 2011
Scale test Prototype

Steel box 1.1 x 1.1 m2 3.3 x 3.3 m2
Diameter piles 0.1 m 0.3 m
Centre-to-centre distance piles 0.55 m 1.65 m
Embankment height 0.42 m 1.32 m
Top load 100 kPa 100 kPa
Weight fill  H 20 x 0.42 = 8.4 kN/m2 20 x 1.32 = 26.4 kN/m2
Traffic load 100 – 26.4 + 8.4 =83.2 kN/m2
Table 5.5: Dimensions scale test and reality
5.3 Measurements
In order to get insight in the arching mechanism different parameters are measured during
the scale test. In this paragraph the different measurements are described.
5.3.1 Consolidation foam cushion

By consolidation, the water in the foam cushion is pressed out and collected in the burette.
The amount of water is determined by weighing the burette. This is measured constantly. In
Figure 5.12 the location of the burette for the discharge of the foam cushion is given in the
test set-up.
5.3.2 Load distribution

As described in the previous chapter the load distribution has three parts:
 Load part A is directly transferred to the piles through arching;
 Load part B is transferred through the geosynthetic reinforcement to the piles;
 Load part C is transferred to the soft subsoil.
How the load parts are measured is described below.
Load parts A and B

Load parts A and B are measured with pressure cells. The pressure cells are placed on two
diagonal piles, to check the system. The pressure cells should measure the same value on
the diagonal piles. The location in the set-up differs for one layer or two layers of
geosynthetics.
For tests with one layer of geosynthetics the pressure cells are placed on the pile and on
the geosynthetics. In Figure 5.13 a picture is presented of the test set-up for one layer of
geosynthetics with the pressure cells. The pressure cell on top of the GR measures load
part A. The pressure cell on top of the pile (below the GR) measures load parts A and B.
Load part B is calculated by subtracting load part A of parts A and B.
Load B transferred through the GR: B  ( A  B )  A
A+B C
P
gives C
C1 C2
Figure 5.13: Test set-up pressure cells for one layer of geosynthetic reinforcement

For the tests with two layers of geosynthetics the location of the pressure cells were
variable. In Figure 5.14 a possible set-up is given. In general the pressure cells are placed
on the pile, on the first layer of GR, under the second layer of GR and on top of the second
GR.
Load part B transferred through the top GR: Btl  ( A  B)tl  Atl
Load part B transferred through the bottom GR: Bbl  ( A  B)bl  Abl
Load part B transferred through both GR’s: Btotal  Bbl  Btl
Load part B transferred through whole layer: Bwl  ( A  B)bl  Atl
Atl
Abl
Atl+Btl
Abl+Bbl C
P
gives C
C1 C2
Figure 5.14: Test set-up pressure cells for two layers of geosynthetic reinforcement
Load part C
Load part C is measured twice. The measurement are equal for one layer or two layers of
geosynthetic reinforcement and should correspond.
The first way of measuring load C is by four force transducers below the cushion. In Figure
5.13 and Figure 5.14 the force transducers are indicated with C1 and C2. The total load C is
calculated by adding up the four force transducers.
Load C resting on the soft soil: C  C1  C2  C3  C4
The second way of measuring load C is based on the pressure in the cushion below. The
pressure is measured with a water pressure meter outside the water cushion. In the figures
above the pressure meter is indicated with P. The load C is calculated by multiplying the
pressure with the area of the water cushion. The measured pressure should be corrected
with the stiffness of the foam cushion and the location of the water pressure meter.
Load C resting on the soft soil: C  As  Pmeasured  zaverage  k foam   water   (5.1)
Where:
C kN Measured load supported by subsoil
Pmeasured kPa Measured pressure within foam cushion
3
k foam kN/m Stiffness foam cushion as determined in Appendix A
 water kN/m 3
Volume weight of water (9.8 kN/m3)
Average reduction height foam cushion, due to the drainage of the cushion
zaverage m
(determined from the measured amount of litres drained from the cushion)
As m2 Area
In this thesis all measured values of A, B and C are presented in kN/pile.

21 February 2011
5.3.3 Strains
Strains can be measured with strain gauges. However strain gauges have a number of
disadvantages. To attach the strain gauges to the GR glue is used, see Figure 5.15.
Unfortunately on most geosynthetic reinforcements this is not possible. For example the
glue doesn’t stick on the material, the geosynthetic is woven and the surface is unsuitable
or the geometry of the geosynthetic is to narrow. When it is possible to glue the strain
gauges on the GR it is difficult to interpret the results. For instance, the influence of the glue
on the measurements can be rather large. Besides the difficulty of attaching the strain
gauges, the strain gauges are also sensitive for sharp edges and fail before the strains
become interesting. When it was possible to attach the strain gauges to the geosynthetics,
the strain gauges were installed.
Figure 5.15: Strain gauges Figure 5.16: Bicycle gear cables
Because the strain gauges cannot be installed on every geosynthetic reinforcement, a

different method has been applied. To measure the strains, bicycle gear cables were
installed. A bicycle gear cable is a rubber tube with an inner steel cable. The ends of the
steel cable and the rubber tubes are fixed on the geosynthetic reinforcement with tie-wraps,
see Figure 5.16. Displacement transducers measure the change of length l in Figure 5.16.
It is difficult to measure the exact displacement of the geosynthetic reinforcement, because
it is uncertain whether the steel cable follows the geosynthetic reinforcement exactly. This
can lead to an underestimation of the strains. There was no experience available with
bicycle gear for measuring strains. Therefore, the bicycle gear is tested first in a tensile
test. During the tensile test, the strains are measured with the bicycle gear and with image
processing. The results of the measuring methods corresponded very well. During the test
series the measurement system with the bicycle gear cables has been optimized further.
Two improvements are: the cable was attached to the GR over its full length and the steel
cable was pre-stressed better before the start of the test.
5.3.4 Deformation
The vertical deformation of the geosynthetic reinforcement is measured with a liquid
levelling system. This system consists of a water pressure cell in a tube filled with water
leading to a box with a constant water level. When vertical settlements occur the water
pressure will change. This is measured and translated to a deformation. The system is
presented in Figure 5.17. The water pressure cell is attached to the geosynthetics with tie-
wraps, see Figure 5.18. The location of the measured settlement is one cm from the tip of
the cell.

water
pressure
cell
Figure 5.17: Liquid leveling system Figure 5.18: Water pressure cells
The deformation on top of the embankment is measured with a x-y scanner. This is done
before the scale test and after. So the total displacement on top of the embankment can be
calculated. The x-y scanner is showed in Figure 5.20.
Box with water for

Metal box water pressure cells
Figure 5.19: Test set-up deformation Figure 5.20: x-y scanner
5.3.5 Top load

The top load is applied with a water cushion and is measured and controlled with a water
pressure meter outside the cushion.
5.3.6 Summary of measurements

Table 5.6 summarizes the measurements and equipment used during the tests. The
location of the measurement equipment differs per test. For every test the exact location of
the equipment is sketched in the test set-up.

21 February 2011
Measurement Equipment Unit

Consolidation foam cushion Weight burette Kg
Load part A Pressure cells kPa
Load parts A + B Pressure cells kPa
Water pressure meter in foam
Load distribution kPa
cushion
Load part C
Four force transducers below
kN
foam cushion
Strain gauges, when possible
Strains in GR
Bicycle gear cables %
Liquid levelling system with cmwk (cm
GR level
Deformation water pressure meters water column)
Ground level XY scanner M
Water pressure meter in
Top load kPa
water cushion
Table 5.6: Summary of measurements and equipment
5.4 Test procedure, definition of equilibrium points and net load

Before the test can be started it has to be set-up with the elements described in paragraph
5.2 and 5.3. First, the foam cushion is placed between the piles in the bottom box. The
foam cushion is filled with water. Then the sand layer and steel frame with the GR are
applied on the foam cushion. Before the fill material is applied, the settlement transducers
and strain gauges are set to zero. Then the water cushion is placed on top of the
embankment and the metal box is closed with the cover.
The actual test is started. First the foam cushion is drained with about six litres. This is a
consolidation step as a result of the weight of the embankment. The top load is applied in
four steps of 25 kPa until a top load of 100 kPa is reached. The consolidation takes place
by draining water from the water cushion in steps of six litres until the pressure in the foam
cushion has been reduced to a low value close to zero. The load steps and consolidation
steps alternate each other.
In some tests, vacuum pressure is applied to the foam cushion below the GR. This is done
after the second drainage step with a top load of 100 kPa. By applying vacuum, the subsoil
support becomes zero. After this the measurements are stopped and the pressure on the
top water cushion is removed. The metal box can be dismantled. In Appendix C the test
procedure is described in detail and illustrated with pictures.
After every drainage or top load step, equilibrium has to be reached by allowing the system
to stabilise. An equilibrium point means that no changes in displacement or load distribution
occur any more. It represents a certain unique combination of top load, average settlement
of the foam cushion and load distribution. Normally the equilibration point is reached in two
hours.
The results of the scale test are presented as a function of time or net load Wn. Net load is
defined by equation (5.2). The net load increases in every phase (top load or drainage). In
(Van Eekelen, et al., 2011a) is concluded that the net load on the fill and the GR
determines its behaviour. Because the measurements show a smooth relationship between
the net load and several measured parameters: load parts A and B, and GR displacement.
Therefore the results are presented as function of the net load.

Wn  Ws  C  R (5.2)
With:
R  0.25 Afil   H  Ws   ( A  B  C ) (5.3)
Where:
Wn kN/pile Net load
R kN/pile Friction between embankment fill and box walls
Area of the fill material Afill  1.1 m2

2 2
Afill m
 kN/m3 Volumetric weight of embankment fill
A kN/pile Measured load part transferred to the pile by arching
B kN/pile Measured load part transferred to the pile by the GR
C kN/pile Measured load supported by subsoil
5.5 Scale test K2

The results of scale test K2 is described in this paragraph. The content of this paragraph is
based on the Factual report piled embankment laboratory tests (Van Eekelen et al., 2010a)
and on the Experiments on piled embankments: the influence of subsoil, fill and
reinforcement (Van Eekelen, et al., 2011a).
5.5.1 Description test K2
Embankment scale test K2

The embankment was constructed with crushed granular rubble (granular material). The
embankment was 42 cm high, measured from reinforcement till ground level. The total
weight of the embankment was 855.8 kg. The total volume of the embankment is
0.42*1.10*1.10 m3 - ½ volume frame = 0,504 m3, resulting in a volume weight of 16.65
kN/m3.
Geosynthetic reinforcement scale test K2

The embankment was reinforced with two layers of uniaxial geosynthetic reinforcement, the
layers are placed direct on top of each other and perpendicular to each other. The two
layers of geosynthetic reinforcement are fixed together onto one steel frame. The
specifications of this material are given in Figure 5.21 and Table 5.7.

21 February 2011
140 140
120 120
100 100
Force (kN/m)
Force (kN/m)
80 80
60 60
40 40
20 20
0 0
0 2 4 6 8 10 12 14 16 18 0 1 2 3 4 5 6
strain strain
Figure 5.21: Tensile test results on GR of K2 (5 test samples) in cross direction (left) and
longitudinal direction (right)
Two layers
Longitudinal
Cross direction perpendicular on
direction
each other
strain stiffness strain stiffness strain stiffness
% kN/m % kN/m % kN/m
0-2 136 0-2 2133 0-2 2269
0-3 113 0-3 2142 0-3 2255
0-4 134 0-5 2382 0-4/5 2516
0-15,8 192 0-5,2 2410
Table 5.7: Average stiffness dependent on strain, for short-term behaviour of GR
The GR stiffness is dependent on the strain and the duration of loading, as well as other
factors. The Jsht given in Table 5.7, and is determined with short term tests (sht), in
accordance with CEN ISO 10319. These ISO tests are much faster than the piled
embankment model tests. This means that the GR in the model tests is loaded longer and
will behave differently (less stiff) from the behaviour suggested by this Jsht. The values of
Table 5.7 for Jsht are determined from tests on samples of the roles from which the samples
for the piled experiments tests are also taken. The stiffness behaviour corresponds with the
stiffness given by the line "constant loading rate" (= short term) curve in the isochronous
stress-strain curves of the applied geosynthetics (Figure 5.22). From Figure 5.22 follows
that the ratio (A1) of the stiffness between this constant loading rate curve and the 1-day
curve is: 1.45 for 2% strain, and 1.40 for 3% strain. The 1-day-GR stiffness at 2% strain is
calculated in equation (5.4). The 1-day-GR stiffness is significant lower than the short term
stiffness. The influence of the GR stiffness in the Plaxis simulations is analysed in
paragraph 8.3.2.
J1%;sht 2269
J2%;1day    1565 (5.4)
A1 1.45

Figure 5.22: Isochronous curves of K2
5.5.2 Specification measurements scale test K2

The locations of the measurement equipment (steel strain cables and settlement
transducers) are shown in Figure 5.23.
2 1
3 6 4
5
3 2 1
Figure 5.23: Locations of the steel strain cables (yellow) and the settlement transducers
(blue) of test K2
5.5.3 Test procedure scale test K2

The sequence of test K2 as it was carried out is given in Table 5.8 and in more detail in
Appendix D. From this table the equilibrium points are selected. The top load, average
settlements due to drainage and subsoil support during the scale test K2 are plotted in
Figure 5.24. The measured load parts A, A+B and C during the scale test are shown in
Figure 5.25. In both the graphs the equilibrium points are shown with dotted lines.

21 February 2011
Date and time sec Hours

17-11-09 14:04 0 0
17-11-09 14:05 - 18:11 62 - 14856 0.02-4.13 Embankment 42 cm, XY scan, top load 0 kPa.
17-11-09 17:38 12894 3.58 Put settlement transducers to zero.
Equilibrium point 1 (after construction of
17-11-09 17:38 12894 3.58 embankment)
17-11-09 17:41 13074 3.63 Drainage 1 (around 6 liters)
18-11-09 9:40 70848 19.61 Placing top cushion and close box
Equilibrium point 2 (after first drainage, before
18-11-09 9:45 70908 19.70 applying top load)
18-11-09 10:02 71900 19.97 Increase top pressure to 25 kPa
Equilibrium point 3 (p = 25 kPa, just before
18-11-09 11:55 78688 21.86 drainage step 2)
Equilibrium point 4 (p = 25 kPa, in between
18-11-09 13:48 85470 23.74 drainage steps 2 and 3)
Reference-container of vertical displacement
18-11-09 13:55 85898 23.86 transducers slipped
18-11-09 15:40 92206 25.61 increasing top load to 50 kPa)
18-11-09 15.42 92330 25.65 Increase pressure to 50 kPa
18-11-09 17:41 99456 27.63 drainage step 4)
19-11-09 8:26 152550 42.38 drainage step 5)
19-11-09 10:29 159920 44.42 Increase pressure to 75 kPa
19-11-09 12:05 165670 46.02 drainage step 6)
19-11-09 14:11 173230 48.12 drainage step 7)
19-11-09 15:57 179620 49.89 Increase pressure to 100 kPa
19-11-09 17:21 184630 51.29 drainage step 8)
20-11-09 9:15 241900 67.19 drainage step 9)
20-11-09 10:15 245470 68.19 Measuring software crashed, reboot software
20-11-09 12:18 252860 70.24 vacuum suction foam cushion)
20-11-09 12:24 253230 70.34 Drainage 10 (vacuum suction foam cushion)
Equilibrium point 15, not taken into account (p =
20-11-09 12:41 254260 70.63 100 kPa, foam cushion at 0 kPa)
20-11-09 12:46 254570 70.71 Removing top load
20-11-09 15:00 Dismantling test setup
Table 5.8: Test procedure with selected equilibrium points

45
100
40
average settlements cushion below (mm
80
35
60 top load 30
subsoil support (kPa)

top load (kPa)
40
25
20 average settlement cushion below 20
0 15
-20 10
sub soil
-40 5
-60 0
-80 -5
0 10 20 30 40 50 60 70
time (hours)
Figure 5.24: Test procedure (top load, average settlement due to drainage and subsoil
support) with selected equilibrium points (dotted lines)
A1 A2 A1+B1 A2+B2 C equilibrium points

35
30
force per pile (kN)
25
20
15
10
0
0 10 20 30 40 50 60 70
time (hours)
Figure 5.25: Measured load parts A (on piles 1 and 2, on top of reinforcement), A+B
(below reinforcement) and C (determined from force transducers C1 to C4, corrected for
weight changes of the water foam cushion) with selected equilibrium points (dotted
lines)

21 February 2011
5.5.4 Specification equilibrium points scale test K2

The equilibrium points, selected in the previous paragraph, are specified in Table 5.9.
Time t Top load p Top load p Pressure q in Average settlement foam

[hours] [kPa] [kN/pile] foam cushion cushion, due to drainage
(load part C) [mm]
[kN/pile]
1 3.6 0.0 0.0 2.3 0.0
2 19.7 1.4 0.4 0.8 5.1
3 21.9 25.0 7.6 4.4 5.1
4 23.7 25.0 7.6 2.7 10.3
5 25.6 25.0 7.6 1.5 15.3
6 27.6 50.0 15.1 5.4 15.2
7 42.4 50.0 15.1 3.6 20.4
8 44.4 50.0 15.1 1.8 25.5
9 46.0 75.0 22.7 5.6 25.5
10 48.1 75.1 22.7 3.5 30.6
11 49.9 75.0 22.7 1.9 35.5
12 51.3 100.0 30.3 5.5 35.5
13 67.2 99.8 30.2 3.6 40.7
14 70.2 100.0 30.2 1.3 46.1
Table 5.9: Specification equilibrium points.
5.5.5 Checking the measurements scale test K2

Several measurements were carried out twice, so that it is possible to check these
measurements. The following measurements are checked:
 The total load on two piles;
 The total load in the system and the measured load;
 The pressure in the foam cushion multiplied by the area of the foam cushion and
the measured load part C (C1+C2+C3+C4) (kN) in an equilibrium point (tap is
closed).
Total load on two piles

The measured load parts A and A+B are plotted in Figure 5.26 as a function of time for
both pile 1 and 2. In the factual report (Van Eekelen, et al., 2010a) is concluded: Although
the difference between A1 and A2 increase during the first 30 hours, the measurements on
pile 1 and 2 agree rather well.
22 A1 A2
A1+B1 A2+B2
force per pile (kN)
16
10
-2
0 10 20 30 40 50 60 70
time (hours)
Figure 5.26: Measured load parts A and A+B on both piles

Total load in the system

To check the total load in the system, the following is compared in Figure 5.27:
 The soil weight plus the top load (H+p)*area embankment;
 the total measured weight A+B+C = 2(A1;bl+B1;bl) + 2(A2;bl+B2;bl) + C1+C2+C3+C4
The figure shows that the soil weight plus the top load is increasingly higher than the
measured load A+B+C. The difference is probably friction between embankment and box,
and friction between the cushion and box and piles. The friction is between the 10% and
20% of the total load.
140
pressure top cushion
+ soil weight
total load (kN)
100 A+B+C
60
20
friction
-20
0 10 20 30 40 50 60 70
time (hours)
Figure 5.27: Comparison total weight and total measured weight
Pressure in foam cushion

Two measurements of the subsoil support are compared in Figure 5.28.
 The measured and corrected pressure in the foam cushion. The corrected pressure
is calculated with equation (5.1) in paragraph 5.3.2;
 The sum of the total force on the cushion (C1+C2+C3+C4) / area cushion.
Each top load increase gives a sudden increase in the subsoil support, and thus an
increase of water pressure in the foam cushion. Each drainage of the foam cushion results
in a reduction of the water pressure in the foam cushion, followed by a phase of slight
pressure increase during rest (closed tap).
The figure shows that the measurements agree very well. Small differences develop during
both the top load increase and drainage phases. However, the agreement remains
relatively good.

21 February 2011
25 pressure cushion below

sum of force transducers
difference corrected pressure and sum
corrected pressure cushion below
subsoil support (kPa) 20
15
10
-5
0 10 20 30 40 50 60 70
time (hours)
Figure 5.28: Comparison corrected pressure foam cushion and total force on force
transducers (C1+C2+C3+C4)
5.5.6 Load distribution scale test K2

Load parts A and B are presented in kN per pile and as % of the total load (A+B+C) in
Figure 5.29. The figures shows a smooth relationship between the net load (top load minus
subsoil support C) and load parts A and B. The figure shows two types of loading: top load
and drainage of subsoil (consolidation). The loads are summarized in Table 5.10.
During the top load phase, the top load is increased. This results in an increase of load
parts A and B in kN per pile. However, load part A decreases when considered as a
percentage of the total load. Figure 5.29 even shows that it has a tendency to return to the
same value of approximately 50% each time the top load is increased. The percentage of
load part B also decreases as a result of a top load increase as seen in Figure 5.29. But
this decrease is limited.
During each drainage step, load parts A and B increase. Thus the arch shows an
unexpected improvement as a result of subsoil drainage. This improvement is higher for
lower top loads.
Each drainage step results in an lower part A percentage (A%) than at the end of the
previous drainage step. As the top load increases, the A% therefore decreases for drainage
conditions. It is possible that these two extremes will eventually become nearly identical for
higher top loads. Load part B in percentage (B%) increases during the scale test.
The scale test K2 starts with one drainage step and the top load remains zero. Arching is
found immediately, as shown by the relatively high initial A% value of 54% seen in Figure
5.29.

18 90 16 80
Drainage
Drainage
Drainage
Top load
Drainage
Top load
Top load
Top load
Drainage
Drainage
Drainage
Vacuum
Drainage
Drainage
Drainage
Vacuum
Top load
Top load
Top load
Top load
16 80 14 70
14 70
12 60
Load part A [kN/pile]
Load part B [kN/pile]

12 60
Load part A [%]
Load part B [%]

10 50
10 50
8 40
8 40
6 30
6 30
4 20
4 20
2 A [%] 10 2 B [%] 10
A [kN/pile] B [kN/pile]
0 0 0 0
0 5 10 15 20 25 0 5 10 15 20 25
Net load [kN/pile] Net load [kN/pile]
Figure 5.29: Measured load distribution for scale test K2
Nr. Time t Total Average Average Average A B C*

[hours] measure A B C* [% of [% of [% of
d load [kN/pile] [kN/pile] [kN/pile] A+B+C] A+B+C] A+B+C]
(A+B+C)*
[kN/pile]
1 3,6 2,4 0,02 0,08 2,31 1% 3% 96%
2 19,7 2,5 1,32 0,32 0,83 54% 13% 34%
3 21,9 8,9 3,63 0,86 4,42 41% 10% 50%
4 23,7 8,8 4,93 1,21 2,67 56% 14% 30%
5 25,6 8,7 5,69 1,57 1,49 65% 18% 17%
6 27,6 15,4 7,65 2,43 5,36 50% 16% 35%
7 42,4 14,7 8,34 2,84 3,51 57% 19% 24%
8 44,4 14,5 9,07 3,72 1,75 62% 26% 12%
9 46,0 21,1 10,56 5,00 5,51 50% 24% 26%
10 48,1 20,6 11,30 5,85 3,45 55% 28% 17%
11 49,9 20,2 11,65 6,70 1,84 58% 33% 9%
12 51,3 26,4 12,66 8,41 5,36 48% 32% 20%
13 67,2 25,5 12,93 9,04 3,53 51% 35% 14%
14 70,2 25,0 13,47 10,39 1,16 54% 42% 5%
15 70,6 24,8 13,92 10,80 0,04 56% 44% 0%
Table 5.10: Load distribution at equilibrium points for scale test K2
* C: calculated from corrected measurements foam cushion.
** The first step often does not have a real equilibrium, because no drainage phase has not
been carried out, and the embankment is ‘floating’ a bit on the saturated foam cushion.

21 February 2011
5.5.7 Deformation scale test K2

The location of the settlement transducers is described in paragraph 5.5.2. The measured
displacements z1, z2 and z3 are shown in time (Figure 5.30) and against the net load
(Figure 5.31). The results are summarized in Table 5.11. Figure 5.31 also shows a smooth
relationship between the net load (top load minus subsoil support C) and the GR
settlements. Van Eekelen et al. (2011b, 2011c) show that it is possible to determine the
load distribution from the measured shape of the deformed GR. The conclude that the load
on the GR is concentrated more or less around the piles. The prove that the load on the GR
can be adequately modelled as inverse triangular line loads on the GR strips between
adjacent piles.
70
pile 3 pile 1
60
z3
50 z2 z1
settlement (mm) .
40
pile 2 pile 4
30
20 z1
z2
10
z3
0
-10
0 10 20 30 40 50 60 70
time (hours)
Figure 5.30: Vertical displacements geosynthetic reinforcement in time for scale test K2
70
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
60
Displacement z1, z2 and z3 [mm]
50
40
30
20
z1
10 z2
z3
0
0 5 10 15 20 25
Net load [kN/pile]
Figure 5.31: Vertical displacements of geosynthetic reinforcement for scale test K2

Nr. Time t Displacement z1 Displacement z2 Displacement z3

[hours] (centre of 4 (centre of 2 (close to 1 pile)
piles) piles) [mm] [mm]
[mm]
1 3,6 0,00 0,00 0,00
2 19,7 3,57 -0,03 0,93
3 21,9 8,96 5,34 6,50
4 23,7 12,43 8,66 9,81
5 25,6 16,71 12,48 13,32
6 27,6 22,16 17,39 18,05
7 42,4 25,35 20,09 20,70
8 44,4 28,97 23,13 23,48
9 46,0 33,97 27,57 27,76
10 48,1 37,31 30,50 30,46
11 49,9 40,94 33,16 32,64
12 51,3 45,71 37,26 36,39
13 67,2 48,80 39,60 38,58
14 70,2 51,28 41,05 40,16
15 70,6 54,07 41,72 40,83
Table 5.11: Deformations and strains at equilibrium points for K2
5.5.8 Strains
Figure 5.32 shows the strains measured with gear cables (paragraph 5.3.3) in test K2. The
figure shows that the strains localized in the tensile strips between the pile caps, and that
the maximum strains were measured on top of the piles.
18
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
16
14
eps1
12 eps2
eps3
10 eps4
Strain [%]
eps5
8 eps6
6
6
5 4 4
2
1
0
3 2 0 5 10 15 20 25
Net Load [kN/pile]
Figure 5.32: Left: strain (gear) cables  in K2, the arched areas are parts of the GR that
act as tensile strips. Right: measured strains in K2.

21 February 2011
Chapter 6
Triaxial test on granular fill material

Most model tests reported in this thesis, were carried out with a fill crushed granular rubble,
hereafter called ‘granular material’. The properties of this granular material were not known.
Therefore, triaxial tests were carried out on the material, and the strength and stiffness
parameters were calculated with the results of these triaxial tests. The diameter of the
samples for the triaxial tests is determined by the grain size of the material. The British
Standard (BS1377, 1990) states that the diameter of the largest particle shall not be greater
than one-fifth of the sample diameter. The material tested in the triaxial test is granular with
a maximum grain size of 16.5 mm according to a sieve test in Appendix B. Following the
British Standard the sample diameter should be at least 82.5 mm. However looking closer
to the material there are grains that are larger. For example, the diameter of a grain is 16.5
mm but the length is 30 mm. This means that the minimal diameter should be 150 mm. The
triaxial apparatus that is used at the Technical University Delft for larger samples, tests
samples with a dimension of 300 mm x 600 mm (diameter x height). The test procedure of
this triaxial apparatus is slightly different from a small triaxial test.
6.1 Test procedure

The sample is prepared on a footplate with help of a plastic membrane, rubber o-rings and
a steel mould. The plastic membrane is constructed from plastic sheet. On the sheet a
template is drawn and welded together. A plastic membrane is not as stretchable as a
rubber membrane and will not deform easily. This is compensated by increasing the width
of the membrane. So, the diameter of the membrane is in the centre larger than at the top
and the bottom of the sample. The membrane is attached to the footplate by stretching the
membrane around the footplate. By greasing the side of the plate and stretching o-rings
over the membrane an airtight seal is ensured. The o-rings are kept in place by two groves
in the footplate.

Chapter 6 Triaxial test
A steel mould is placed

around the footplate and the
membrane, see Figure 6.1.
On the footplate a perforated Top plate with
PVC plate and a geotextile rubber o-rings
are placed. This is done to
prevent fine material from
being sucked out of the
sample into the footplate and Steel mould
vacuum system. The sample
is constructed in 10 layers.
After each layer the sample is Vacuum tube
compacted with a steel
weight. When the sample has
reached a height of 600 mm, Footplate
a geotextile, a PVC plate, a
top plate with grease and o- Vacuum
rings are installed on the measuring tube
sample. Before the steel Figure 6.1: Sample prepared in steel mould
mould is removed vacuum is
applied on the sample via outlets at the top and bottom plate. The vacuum prevents the
sample from collapsing. A high vacuum pressure can not be reached, because the granular
material damages the membrane resulting in small holes. Via the small holes, air is sucked
into the sample. A second membrane is placed over the first membrane to reach a higher
vacuum pressure. The second membrane is also sealed by grease and o-rings. The height
of the sample is measured after applying the second membrane.
The next step is to install the measurement equipment. The measurement equipment is
attached to two rings. Plastic blocks are glued to the membrane at 1/3 and 2/3 of the
sample height. The rings are installed, centred and positioned on the plastic blocks by
studs and adjustable springs. On each ring three LVDT’s (Linear Variable Differential
Transformers) are placed to measure the radial displacements. The LVDT’s are pushed by
vacuum pressure against small glass plates that are glued to the sample. Between the
rings there are three axial LVDT’s installed. The rings and LVDT’s are shown in Figure 6.2.
The sample is now ready to be rolled into the triaxial rig and positioned exactly under the
hydraulic actuator by which the sample will be loaded.
Radial displacement
sensor
Axial displacement
sensor
Adjustable studs
and springs for
centering of rings
Figure 6.2: Measuring rings with horizontal and vertical displacement sensors

21 February 2011
Shaft of hydraulic
actuator
Crosshead of frame
Load cell
300x600 mm
sample
Frame column
Roll on/off apron
Figure 6.3: Large scale triaxial apparatus
The large scale triaxial apparatus consists of a loading frame, a hydraulic actuator, a load
cell and a controller for application and measurement of displacement controlled static axial
loading. Figure 6.3 shows the components. The load on the sample is provided by the
hydraulic actuator. The reaction of the force applied by the actuator is absorbed by the
loading frame. The load cell and the internal displacement transducer measure the applied
load and vertical displacement. A MTS controller controls and responds on the signals of
the load cell and displacement transducer.
The internal vacuum can be controlled with a vacuum regulator on a multi-programmer.

The level of vacuum is measured and recorded with analogue gauges and continuous
electronic measurement. The vacuum pressure is equal to the cell pressure or confining
pressure. To determine the friction angle and cohesion the triaxial test should be carried
out at least two times with different vacuum pressures. The tests were executed with
vacuum pressures of 0.16 bar, 0.55 bar and 0.9 bar. This is equivalent to 16 kPa, 55 kPa
and 90 kPa.
Analogue vacuum gauges

Vacuum regulator
MTS controller
Multi-programmer
Figure 6.4: Control and data acquisition unit

The measurements of the load cell, actuator, two vacuum pressure meters, six radial and
three axial displacement sensors are registered. The multi-programmer provides the
memory and the required conversion of digital to analogue data or vice versa. Figure 6.4
shows the computer, multi-programmer and MTS controller.
The load cell and hydraulic actuator is placed almost on top of the sample. Via the MTS
controller an engine is started and the hydraulic actuator is applying a force to cause a
settlement. The computer registers the measurements.
6.2 Calculations
The stress dependency of the failure behaviour of a granular material is established by
performing a triaxial test with at least two confining stress levels. The stress dependency of
the failure behaviour is described by the Mohr-Coulomb failure criterion.
 f  c   n,f  tan( ) (6.1)
Where:
f kPa Shear stress at failure
c kPa Cohesion
 n,f kPa Normal stress at failure
In the top and bottom part of the sample shear stresses develop (Van Niekerk, 2002). The
shear stresses are induced by friction between the sample and the rigid top or bottom
plates. In contradiction to the top and bottom part, in the central part there are no shear
stresses but a uniform stress distribution. Therefore the parameters are calculated for the
central part of the sample. The equations used to calculate the parameters are based on
the British Standard BS 1377 : Part 8 (1990). The triaxial test performed is not the normal
triaxial test with cell pressure as described in the British Standard, but with vacuum
pressure. This results in differences in equations. All the equations that are applied for the
calculation of the parameters are described in Appendix E. This chapter gives only a brief
summary of the equations that are different from the equations used for a normal triaxial
test.
The volumetric strain should be calculated with the change in volume. Only the volume
change is not measured directly in the large triaxial test. The volumetric strain is therefore
calculated with the axial and radial strain in equation (6.2). The axial and radial strains are
determined for the middle part of the sample
 v   1  2   3 (6.2)
Where:
1 - Axial strain
In the normal triaxial test, the calculation of the area of the cross section is based on the
assumption that the sample deforms as a right cylinder. In this triaxial test the area can be
calculated more accurately with equation (6.3). The equation is based on the radius and
measured radial displacement. This is done continuously, for each measured point.
As    r 2 (6.3)

21 February 2011
With:
r  rint  r (6.4)
Where:
2
As mm Area of cross section
2
6.3 Execution test

Three samples are tested in the triaxial test. The samples are prepared as explained in
paragraph 6.1. During the triaxial test the displacement velocity is controlled at 2 mm per
minute. The measurements are registered every second. The raw data can be found in
Appendix E.
The location of the LVDT’s on the rings are presented in Figure 6.5. The figure shows
which radial displacement sensors are situated above each other. The numbers of the
displacement sensors correspond with the numbers of the data. The vertical centre-to-
centre distance between the rings is 200 mm.
10 Axial 10 Axial
7 Radial 5 Radial 11 Radial 12 Radial
9 Axial 8 Axial 9 Axial 8 Axial

6 Radial 13 Radial
Figure 6.5: Location displacement sensors on top ring (left) and on bottom ring (right)
The vacuum pressure of the test with 16 kPa vacuum pressure was not constant. The
vacuum pressure started at 20 kPa and decreased during the test to 11.2 kPa. The
decrease of vacuum pressure is caused by leakage of the sample membrane. The
confining pressure of 16 kPa is based on the average vacuum pressure during failure. In
the calculations the measured vacuum pressure was used. The vacuum pressure of the
other two tests was constant. This can be seen in the stress distribution graphs of Appendix
E.
During the third test, with a vacuum pressure of 90 kPa, there were some problems with the
engine of the hydraulic actuator. As a result of overheating, the engine had a power shut
down, whereby the force on the sample decreased to zero. When the engine was restarted,
the force increased again. Due to this problem the results of the triaxial test at 90 kPa
vacuum pressure show unloading and reloading.
The sample that was used for the test with 16kPa of vacuum pressure is built up from the
material used in the sample used for a vacuum pressure of 55kPa. If new material was
used the water content would have been lower. The water content has influence on the
strength parameters of the sample. To prevent this, the material was reused. A negative
result is that the material has a different grain distribution, because some grains have been
crushed in the triaxial test executed before.

6.4 Data analyses & Results

The parameters are calculated from the raw data
and presented in general graphs in Figure 6.7 to
Figure 6.10. Detailed graphs can be found in
Appendix E. The data analyses and results are
explained in this paragraph.
The samples were shaped spherical at failure and

no shear zones were visually noticeable on the
samples. See Figure 6.6.
The displacement of the LVDT’s is analysed

before the parameters are calculated. The
displacement of the LVDT’s of the three tests are
plotted against time. These graphs can be found
in Appendix E. From the graphs can be concluded
that a number of LVDT’s have a divergence in
their results. These LVDT’s are excluded from the
calculations of the parameters. There can be
different causes for the divergence. For example
the glass plates on the sample rotates, because of
moving grains. Or the LVDT moves from the glass Figure 6.6: Sample shape at failure
plate during the test.
In Figure 6.7 the stress strain curve is plotted. The unloading and reloading effect, due to
the machine problems, of sample 90kPa can be observed. The deviatoric stress of sample
16kPa decreases after reaching failure. This is caused by the decrease of vacuum
pressure.
Figure 6.8 shows the volume change of the samples. The volume change of samples
55kPa and 90kPa are in the same range. For sample 16kPa the initial volume decrease is
followed by volume increase very quickly in comparison to the other two samples. The fast
transition between volume decrease and increase can be caused by the low confining
pressure.

21 February 2011
Figure 6.7: Deviatoric stress strain curve
Figure 6.8: Volume change curve

The stress path curve is presented in Figure 6.9. The stress paths of sample 55kPa and
90kPa are connected at equal strains. The stress path points are connected with a linear
line at the same strain. Because the cohesion is zero, the linear line starts in the origin.
From the inclination of these lines the angle of internal friction can be calculated for the
corresponding axial strain.
The angle of internal friction and cohesion can also be determined from the Mohr circles.
This can be seen in Figure 6.10. The principal stress and confining stress at failure are
drawn for the three tests. The angle of the tangent to the circles is the angle of internal
friction. For failure this is 47 degrees. The cohesion is zero.
The grading of the applied granular material is between 0 – 16.5 mm as determined in the
sieve test as shown in Appendix B. The grading of the samples is not tested. When the
samples in the plastic membrane were examined the conclusion was that the samples were
not constructed uniformly. Locally there were some groupings of fine grains or coarse
grains. This was consistent for all three samples. The largest difference between the
densities of the samples is 13 kg/m3. This is 0.8% of 1682 kg/m3. The moisture content of
the samples lay between 6% and 7%.
Characteristics and results of the three triaxial tests are presented in Table 6.1. More
detailed graphs of the strains, stresses, stiffness parameters and strength parameters can
be found in Appendix.
Sample Sample Sample

Characteristics
16kPa 55kPa 90kPa
Height mm 602 601 598
Diameter mm 300 300 300
3
Density kg/m 1682 1681 1669
Moisture content % 6.30 6.64 6.98
Confining pressure kPa 16 55 90
Deviatoric stress at failure kPa 134 343 546
Axial strain at 50% of failure - 0.00184 0.00453 0.00502
Young’s modulus at 0% of failure MPa 75.25 189.91 315.94
Young’s modulus at 50% of failure MPa 36.41 37.86 54.38
Poisson ratio at 0% of failure - 0.08 0.10 0.09
Poisson ratio at 50% of failure - 0.41 0.33 0.29
Cohesion kPa 0 0 0
Internal friction angle at failure ° 47 47 47
Table 6.1: Characteristics and results of the samples in the triaxial tests

21 February 2011
1
s  1   3 
2
1
t   1   3 
2
Figure 6.9: Stress path curve
Figure 6.10: Mohr circle of effective stress

6.5 Conclusions
Triaxial tests were carried out on three 300 mm x 600 mm samples (diameter * height). The
volume weights of the samples were respectively 1682; 1681 and 1669 kg/m3 which lies in
the same range as the volume weights of the model tests (average 1677 kg/m3). The
triaxial tests were executed with vacuum pressures of respectively 16 kPa, 55 kPa and 90
kPa.
There are a number of factors that influence the parameters of the samples. These are:
grading, degree of compaction and moisture content. In order to properly compare the
three triaxial tests, these factors need to be in the same range for each performed triaxial
test.
The differences in density and moisture content are not large and will not influence the
results of the triaxial tests considerably. However, differences in grading will have influence.
The offsets caused by the displacement of the sensors will effect the outcome of the
parameters.
The general characteristics of the triaxial tests are presented in Table 6.2. The results of
the triaxial tests, calculated following BS1377, 1990, are summarized in Table 6.3.
Characteristics Values
Type of soil - Granular material
Type of sample - Disturbed sample
Displacement controlled mm/min 2
Table 6.2: General characteristics of the triaxial tests
Strain [%] E  3 55 [MPa] E  3 90 [MPa]  [-]  [ ]  [ ]

1 23 35 0.39 42 -
2 14 21.5 0.45 45 -
3 10.5 16.5 - 48 5
5 7 11 - 49 11
7 5 8 - 48 15
Table 6.3: Summary of strength and stiffness properties of the granular material

21 February 2011
Chapter 7
Numerical model of scale test

The tests are simulated with Plaxis 3D Tunnel version 2.4. This chapter describes how the
scale test is build up in Plaxis.
7.1 Geometry and boundary conditions

The scale test is modelled in the finite element program. The geometry of the scale test is
symmetric, therefore only one quarter of the scale model, one pile with surrounding soil, is
modelled. The modelled quarter of the test is hatched with blue in Figure 7.1. The geometry
of this quarter has one layer of geosynthetics and is shown in Figure 7.2. The size of one
quarter of the scale model is w x l x h = 0.55 x 0.55 x 0.64 meter.
Figure 7.1: Top view scale test set-up Figure 7.2: 3D geometry Plaxis
In the scale test set-up circular piles are applied. Plaxis 3D Tunnel cannot model circular
piles and therefore square piles are applied in Plaxis. The area of the circular pile should be
the same as the area of the square pile. With the following equation the width of the square
pile is determined at 0.088 meter.
1
Acircular  Asquare   a 2  aeq
2
4
(7.1)
1
aeq   0.12  0.088m
4

Chapter 7 Numerical model of scale test
Where:
Acircular m2 Area of circular pile
2
Asquare m Area of square pile
a m Diameter of circular pile
a eq m With of square pile
The height of the subsoil and the pile are 0.2 meter. On top of the subsoil and pile a 0.02
meter layer of sand is modelled. The geosynthetics and frame are situated on top of the
sand layer. The rest of the model is filled up with 0.42 meter granular material. In the scale
test, the geosynthetic reinforcement (GR) is attached to a stiff, steel frame. Because the
frame causes disturbance in the scale model, the frame is modelled. It is modelled as a
rectangular frame. The height of the frame is 0.025 meter. This is the same as the height of
the frame in the scale test. The frame width is 0.05 meter. The width is smaller than the
width in reality, because in reality the frame is not solid. The frame can move vertically.
Figure 7.3: Cross section of Plaxis model and dimensions
During the execution of the scale tests there is a loss of load due to friction between the
steel wall and granular material. The friction is between 10% and 20% of the weight and
applied load during the scale test. The calculation of the amount of friction is described in
paragraph 5.4. The friction is an important part of the model. Normally an interface is
applied to model the friction, which should be applied at the left and back side of the model.
However in Plaxis 3D Tunnel an interface cannot be applied at the back side. Paragraph
7.4.1 describes an alternative approach to apply friction in the model.
7.2 Constitutive relations and material parameters

The constitutive relations and material parameters of the materials applied in the model are
described in this paragraph.
7.2.1 Piles
In the scale test the PVC piles support the GR and granular material. The piles does not
deform and therefore the piles are modelled in the model as a stiff column. PVC is a non-
porous material, whereby the permeability is zero. The piles are constructed in such a way
that there is no lateral deformation due to a vertical load in the scale test. Therefore the
Poisson’s ratio is zero. The unsaturated volumetric weight and Young’s modulus are
determined with ‘Tabellen voor bouw- en waterbouwkunde’ (2000).

21 February 2011
Properties Piles
General properties
Model - Linear elastic
Material type - Non-porous
 unsat Unsaturated volumetric weight kN/m3 13.64
 sat Saturated volumetric weight kN/m3 NA
k x , ky , kz Permeability m/day NA
Stiffness properties
E ref Young’s modulus kN/m2 2.9E6
 Poisson’s ratio - 0.0
Table 7.1: Pile properties
7.2.2 Subsoil
The subsoil in the scale tests is modelled with a watertight, soaked foam cushion in which
the water pressure is controlled.
The behaviour and stiffness of the foam cushion is analysed by performing a water test.
The water test is described in Appendix A. The foam cushion behaves linear elastic and
therefore the constitutive relation for the subsoil is based on Hook’s law. The Young’s
modulus is determined by multiplying the stiffness of the foam cushion with the height of
the cushion.
Properties Subsoil
General properties
Model - Linear elastic
Type - Undrained
 unsat Unsaturated volumetric weight kN/m3 10.15
 sat Saturated volumetric weight kN/m3 10.15
k x , ky , kz Permeability m/day NA
E 'ref Young’s modulus kN/m2 10
' Poisson’s ratio - 0.2
Table 7.2: Subsoil properties
The scale test is controlled by draining the foam cushion and therefore decreasing the
water pressure in the foam cushion. To simulate the drainage of the scale test, the water
pressure is prescribed in the model by applying a phreatic level to the clusters of the
subsoil. The cluster phreatic level represents a pressure head. Therefore the measured
water pressure cannot be used as input value in Plaxis. The water pressure has to be
converted into a pressure head. This is done by dividing the measured water pressure with
the unit weight of water. Next to that the reaction of the weight of the frame and sand layer
on the water pressure is excluded from the measured water pressure. This is done by
resetting the water pressure to zero after applying the sand layer and frame in the scale
model set-up. Therefore the measured water pressure, is the pressure against the GR
during the scale test. In the Plaxis model the weight of the frame and sand layer have to be
included in the water pressure, by adding up the weight to the measured water pressure.
How the measured water pressure is converted is shown in Figure 7.4 and described in
equation. To calculate the input value of the pressure head, the unit weight of the frame

and sand layer are added to the measured water pressure. Next to that the hydrostatic
pressure is added. This value is divided by the unit weight of water. The conversion of the
measured water pressure is described in equation (7.2).
During the construction of the embankment, a water overpressure develops in the foam
cushion. The water pressure is higher than the weight of the granular material. In the scale
test this overpressure has no effect, because the water is enclosed in a rubber sheet. In
Plaxis this water pressure will push the granular water upwards. To avoid this, the water
pressure in the phases were the embankment is constructed, is lowered to a water
pressure that is not higher than the embankment weight.
Fframe
pscale test    sand  hsand
Afoam
Fframe
pscale test    sand  hsand   water  hfoam
Afoam
Figure 7.4: Water pressure development in foam cushion
Fframe
pscale test    sand  hsand   water  hfoam
Afoam (7.2)
zPlaxis 
 water
Where:
zPlaxis m Input value for Plaxis: pressure head
2
pscale test kN/m Measured water pressure within the foam cushion
2
Afoam m Area of the foam cushion
 kN/m 3
Volumetric weight
hsand m Height sand layer
hfoam m Height foam cushion
7.2.3 Geosynthetic reinforcement

The GR is assumed to deform linearly with the load. The axial stiffness of the GR is
determined from five tensile tests. The tensile tests are performed according to DIN EN ISO
10319. From the tensile tests the average tensile stiffness is determined during the first 2%
strain. The tensile stiffness depends on the material of the GR and the number of layers.
For the considered scale tests the axial stiffness is summarized in Table 7.3. The GR is
attached to a steel frame. The properties of the steel frame can be found in paragraph
7.2.5.

21 February 2011
Properties 1 layer GR
Type Elastic
EA Axial stiffness at   2% per layer GR kN/m 2269
Table 7.3: Geosynthetic reinforcement properties
7.2.4 Embankment fill

During the scale tests, a thin Itterbeck sand layer and a 0.42 m thick layer of granular
material is used. Where possible the parameters of the materials are determined with
triaxial test results. Otherwise the parameters are based on relationships. The parameters
are determined for both the Mohr Coulomb model and the Hardening Soil model. The Mohr
Coulomb model is used for initial calculations with Plaxis, because Mohr Coulomb needs
less calculation time. However, the Hardening Soil model will give more accurate results.
The parameters for Mohr Coulomb model are summarized in Table 7.4 and for Hardening
Soil model in Table 7.5. The source of the parameters is described in Appendix G.
Sand and granular material have a high permeability and are therefore modelled as drained
materials. The volumetric weight of the granular material is determined in the scale test.
This value is used in the model. In the scale test there is nearly no water in the
embankment fill, therefore the saturated volumetric weight and the permeability are not
necessary for calculations in Plaxis. Sand and granular material are cohesionless. To avoid
numerical complications with Plaxis the input value of cohesion is one kPa.
The sand layer is split up in two parts, a part

above the pile and a part direct on the subsoil.
The parameters are different for both parts.
The sand on top of the pile is expected to
behave very stiff, because the sand on the
pile will be clamped between the GR and the
pile. Therefore, the sand on the pile will be
compressed more and will have higher
stiffness and strength properties. The higher
stiffness of the sand above the pile will
increase the development of arching. The
sand on the subsoil will follow the settlements
of the subsoil and geosynthetics. The
Figure 7.5: Pile head after execution
parameters of the sand on the subsoil and on
scale test
the pile are determined from two triaxial tests
performed on the Itterbeck sand. The volumetric weight of the sand is not determined in the
scale test, therefore the volumetric weight of the applicable triaxial test is used in the
model.
Sand above Sand next Granular

Soil properties
pile to pile material
General properties
Type - Drained Drained Drained
 unsat Unsaturated volumetric weight kN/m3 20.06 18.67 16.65
 sat Saturated volumetric weight kN/m3 NA NA NA
k x , ky , kz Permeability m/day NA NA NA
Strength properties
c 'ref Cohesion kN/m2 1 1 1
' Internal friction angle  40.93 32.52 47


Soil properties
' Dilatancy angle  10.93 2.52 11
E 'ref Young’s modulus kN/m2 51470 19660 58870
' Poisson’s ratio - 0.30 0.26 0.31
Advanced properties
E 'increment
Increase of stiffness with depth kN/m3 0 0 0
y ref Reference level for E increment m 0 0 0

3
c 'increment Increase cohesion with depth kN/m 0 0 0
y ref Reference level for c increment m 0 0 0
 tension Tensile strength kN/m 0 2
0 0
Table 7.4: Sand and granular material properties for the Mohr Coulomb model

Properties
General properties
Type - Drained Drained Drained
 unsat Unsaturated volumetric weight kN/m3 20.06 18.67 16.65
 sat Saturated volumetric weight kN/m3 NA NA NA
k x , ky , kz Permeability m/day NA NA NA
Strength properties
c 'ref Cohesion kN/m2 1 1 1
' Internal friction angle  40.93 32.52 47
' Dilatancy angle  10.93 2.52 11
Secant stiffness for primary
E '50 kN/m2
ref
51470 19660 58870
loading in standard triaxial test
Tangent stiffness for primary
E 'oed kN/m2
ref
51470 19660 58870
loading in oedometer test
Stress dependent stiffness
m - 0.5 0.5 0.74
according to a power law
E 'urref Unloading / reloading stiffness kN/m2 154410 58980 176610
 'ur Poisson’s ratio for elastic
- 0.2 0.2 0.2
unloading/reloading
p ref Reference stress for stiffness kN/m2 1 100 100
K 0nc K0-value for normal
- 0.345 0.462 0.337
consolidation
c 'increment Increase cohesion with depth kN/m3 0 0 0
y ref Reference level for c increment m 0 0 0
Rf Failure ratio - 0.9 0.9 0.9
 tension Tensile strength kN/m 0 2
0 0
Table 7.5: Sand and granular material properties for the Hardening Soil model

21 February 2011
7.2.5 Steel box wall and frame

Steel is modelled as a linear elastic material. Steel is non porous, whereby the permeability
is zero. The unsaturated volumetric weight of steel is 77 kN/m3 and the Young’s modulus is
2.1E8 kN/m2. The weight of the box wall is unknown, therefore the properties described
above are used for the steel box. The weight of the frame is known (37.3 kg) and
conversed to a unsaturated volumetric weight. The box wall and frame are designed not to
deform laterally, therefore the Poisson’s ratio is zero in the model.
Properties Box wall Frame

Model - Linear elastic Linear elastic
Type - Non-porous Non-porous
 unsat Unsaturated volumetric weight kN/m3 77 70.45
 sat Saturated volumetric weight kN/m3 NA NA
k x , ky , kz Permeability m/day NA NA
 Poisson’s ratio - 0 0
E ref Young’s modulus kN/m2 2.1E8 2.1E8
Table 7.6: Steel box wall and frame properties
7.3 Mesh generation

The geometry has to be divided in a mesh of elements before finite element calculations
can be performed. When the geometry and material properties are fully defined in the
cross-section of the input program, a 2D mesh can be generated. The refinement of the 2D
mesh should full fill the requirements before the 3D mesh is generated.
Before the 3D mesh can be generated, the cross-section planes (z-planes) have to be
defined in z-direction. The previously generated 2D mesh is repeated at each z-plane. The
3D mesh is created by connecting the corners of the 2D triangular elements to the
corresponding points of the elements in the next z-plane. This results in a 3D mesh of
wedge elements. Figure 7.6 shows how the 3D model is created.
Figure 7.6: Creating a 3D model and finite element mesh (Plaxis)
The three dimensional wedge elements consist of triangular and quadrilateral elements.
The triangular elements in the vertical plane have six nodes and the quadrilateral elements
in the horizontal plane have eight nodes. Two three dimensional 15-node wedge elements
are shown in Figure 7.7. The 15 node wedge element consist of six stress points.

Figure 7.7: 15-node wedge element with: nodes (left) and stress points (right)
The Plaxis simulations of the scale tests are performed with a coarse, medium and fine
mesh. A finer mesh has more elements. This causes more time-consuming calculations
and a larger amount of data. An advantage of a finer mesh is that the results are more
accurate. Especially stress peaks in the area of the pile top will be reduced. The amount of
data resulting from calculations with a fine mesh appeared to be unmanageable. Therefore
most of the calculations were carried out with a 3D mesh determined from an 2D mesh, for
which the generation was set to coarse with local refinements at the top of the pile.
7.4 Friction
During the execution of the scale tests there is a loss of load due to friction. This friction
develops between the granular fill and structures in the test set-up. For example the friction
develops between the granular material and the box wall. In the next paragraphs the friction
between different parts are described.
7.4.1 Wall friction

Wall friction is caused by friction between the steel wall and granular material. This
measured friction lies between the 10% and 20% of the soil weight and top load. Therefore,
the friction is an important part of the model. Normally an interface is applied to model the
friction. This interface should be applied at the left and back side of the model. In Plaxis 3D
Tunnel, an interface cannot be applied at the back side, therefore the friction is modelled by
an extra segment of soil with lower strength parameters. To validate this assumption the
difference between the results of a model with an interface on one side and a model with
an extra segment of soil on one side is analysed in the Appendix H.
7.4.2 Pile friction

Pile friction is the friction between the pile and foam cushion. This friction is assumed to be
small and its influence on the arching within the fill is limited. As described in the previous
paragraph it is not possible to model an interface at the back and front side of the pile.
Therefore the friction between the pile and foam cushion is neglected.
7.4.3 Geosynthetic reinforcement friction

There is friction between the soil and the GR when the GR is slipping between the soil. In
the scale test is assumed that the slipping of the GR is very limited, because of the
interaction between the large grains of granular material and large holes in the applied
geogrids. Therefore it is not necessary to model an interface between the soil and
geosynthetics.
7.5 Calculation phases

The calculation phases of the model are based on the scale test procedure described in
paragraph 5.4. The load and water pressure differ for each phase. The details of the
calculation phases are described in Table 7.7.

21 February 2011
0 Initial phase
Normally the water pressure and initial stresses are generated during the initial
phase. The initial stresses are calculated for the model defined in the input
program. The soil body clusters in the z-plane cannot be adapted. Therefore
the initial stresses are calculated for a model were the pile begins in the front
plane and ends in the rear plane. This is not a representation of reality.
Therefore the water pressure and the ∑Mweight are set to zero. When
∑Mweight is set to zero, the soil weight is not activated.
1 Applying subsoil
In this phase the subsoil and pile are activated. When applied, the steel wall is
also activated.
2 Applying soil weight
The soil weight is activated in this phase by setting ∑Mweight to one.
3 Applying sand and frame
The sand layer and frame with GR are activated. In the same step, the water
pressure is applied in the subsoil clusters. Because the water pressure in the
subsoil is prescribed, the subsoil has to behave as drained material. This is
achieved by ignoring the undrained behaviour.
4-7 Applying granular material
The granular material is activated in four phases, because the granular material
is also installed and compacted in layers during the scale test set-up. The water
pressure in the subsoil clusters is increased in each phase with the load of the
granular material applied in that phase.
8 Applying first drainage step of subsoil
The water pressure in the subsoil is decreased to the measured water
pressure. The same load is applied as in the scale test. The displacements are
reset to zero at the beginning of this phase, as performed in the test. This
phase is calculated with an updated mesh.
9 Applying load 25 kPa
The load of 25 kPa is applied during this phase and the water pressure is
increased to the water pressure found during the scale tests.
10 - 11 Applying drainage of subsoil
The subsoil is drained in two steps as performed during the scale test.
12 - 20 Applying load and drainage of subsoil
The phases 9 until 11 are repeated for a load of 50 kPa, 75 kPa and 100 kPa.
21 Applying vacuum
The subsoil and water pressure are deactivated in this phase. Because of this
the subsoil support becomes zero.
Table 7.7: Plaxis calculation phases
7.6 Determination of loads and settlements

During the scale tests, the loads on, and the settlements of the geosynthetics are
measured. These values can be compared with the values calculated with Plaxis. The
following section describes how the loads and settlements are derived from the results of
Plaxis.
During the scale tests, loads A and A+B are measured with a load cell. In Plaxis the loads
cannot be measured directly. The average stress is determined of the stress points in the
area of the load cell. Then the loads are calculated by multiplying the average vertical
stress with the area of the load cell. The area of load parts A and A+B are hatched in
Figure 7.8. Load part B is calculated by subtracting load A from load A+B, as seen in
equation (7.3).

Load part A
Load part A+B
Figure 7.8: Area for calculation of the load part A and part A+B
B  (A  B)  A (7.3)
Where:
A kN/pile Load part transferred to the pile by arching
B kN/pile Load part transferred to the pile by the GR
A+B kN/pile Total load on the pile
Load C is the pressure against the foam cushion. This pressure is not generated directly by
Plaxis. It can be calculated by subtracting the weight of the frame and sand from the water
pressure as described in equation (7.4).
Fframe
pGR  pPlaxis   water  hfoam    sand  hsand (7.4)
Afoam
Where:
2
pPlaxis kN/m Water pressure within the foam cushion in Plaxis
2
pGR kN/m Pressure against the geosynthetic reinforcement in Plaxis
2
Afoam m Area of the foam cushion
 kN/m 3
Volumetric weight
h m Height layer
The settlements of the GR are given directly in the output data of Plaxis. The location of the
settlements depends on the location of the measurement equipment during the scale test.

21 February 2011
7.7 Conclusion
The scale test is modelled in Plaxis 3D Tunnel. To reduce the calculation time, only one
quarter of the scale test is modelled. This is possible, because the scale test is symmetric.
The pile in the scale test is circular, this cannot be modelled in Plaxis. Therefore the area of
the circular pile is converted to a square. The model is build up of one pile, subsoil, sand
layer on the subsoil, GR with steel frame and granular material. The load distribution in the
scale test is driven by the water pressure in the foam cushion. The measured water
pressure is converted to a pressure head. This pressure head is assigned to the subsoil
clusters in the Plaxis model. The material parameters of sand and granular material are
determined for MC model and HS model. To model the equilibrium points of the scale test,
the phases of the Plaxis model are based on the equilibrium points. The water pressure in
the foam cushion and the top load on the embankment are dependent on the measured
values of the scale test and are used as input value in the Plaxis phases. The load
distribution cannot be determined directly from the Plaxis results, therefore the load
distribution is calculated by multiplying the area of the pile with the average stress on the
pile. During the execution of the scale tests there is a loss of load due to friction. The
friction is between 10% to 20% of the soil weight and top load, therefore an important part
of the model. In Plaxis 3D Tunnel it is not possible to apply an interface at the back side of
the model (parallel to the front plane). Another option is to apply an extra soil segment with
lower strength parameters. This option give good results with MC model, but when applied
with HS model the calculation time increases significant. Therefore it is not realistic to apply
an extra soil segment as friction. To decrease the load in the system, the amount of load
loss is subtracted from the top load.

21 February 2011
Chapter 8
Validation and analysis of numerical model

In the previous chapter two constitutive models are described for modelling the fill in the
scale test. The Plaxis simulations with the Mohr Coulomb (MC) model and the Hardening
Soil (HS) model are validated and analyzed in this chapter. The friction in the scale test is
neglected in this chapter. That means that the friction is not included in the simulations, and
that the top load applied in the simulations, is equal to the measured top load minus the
measured friction.
8.1 Numerical results of scale test K2

In this paragraph the Plaxis simulation of scale test K2 is discussed. The numerical
analysis of the scale test is performed with the MC model and the HS model. The geometry
and parameters are described in the previous chapter. For each phase, as described in the
previous chapter, the deformed mesh, vertical deformation, effective vertical stress, tensile
force in the geosynthetic reinforcement (GR) and the deformation of the GR are presented
in Appendix I. The vacuum phase has the largest deformation and will be presented in this
chapter. The results are presented for cross section B-B or for the points z1, z2 and z3 on
the GR. The location of cross section B-B and the locations of z1, z2 and z3 are shown in
Figure 8.1.
Figure 8.1: Cross section of model and top view
Figure 8.2 to Figure 8.5 give calculated results shown in cross section model (B-B). The
calculated principle stresses in the vacuum phase for both models are shown in Figure 8.2
and Figure 8.3. From the figures soil arching can be observed. The principal stresses of the
HS model are larger and are more concentrated on the pile.

Chapter 8 Validation and analysis of numerical model
Figure 8.2: Effective principal stresses for Figure 8.3: Effective principal stresses for
Mohr Coulomb model Hardening Soil model
The calculated vertical displacements are shown in Figure 8.4 and Figure 8.5. The
displacement pattern is similar for the models. For the vacuum phase, the Mohr Coulomb
model gives more displacement then the Hardening Soil model.
Figure 8.4: Vertical displacement of Figure 8.5: Vertical displacement of

vacuum phase for MC model vacuum phase for Hardening Soil model
The tensile forces in the GR for the MC and HS model are shown in Figure 8.6 and Figure
8.7. The values of the tensile forces for both models are similar. The tensile forces are
concentrated in the ‘tensile strips’. This corresponds with the measured strains (Van
Eekelen, et al., 2011a), as shown in paragraph 5.5.8. The tensile strips lie on top of and
between adjacent piles. The maximum tensile forces can be found at the edge of the piles.
The exact location of the peak values cannot be determined, because the mesh is to
coarse and there are too much irregularities in the tensile forces.

21 February 2011
y y
x x
y y
x x
Figure 8.6: Tensile forces in GR of Figure 8.7: Tensile forces in GR of

vacuum phase for Mohr Coulomb vacuum phase for Hardening Soil
model in x-direction (top) and in y- model in x-direction (top) and in y-
direction (bottom) direction (bottom)
The total displacement of the GR for the MC and HS model is given in Figure 8.8 and
Figure 8.9. For cross section B-B, the total displacement is shown in Figure 8.10 and
Figure 8.11. From the figures can be concluded that in the vacuum phase the MC model
gives more GR displacement than the HS model. The maximum displacement is
concentrated at the edge of the model.
Figure 8.8: Total displacement of GR in Figure 8.9: Total displacement of GR in

vacuum phase for Mohr Coulomb model vacuum phase for Hardening Soil model

Figure 8.10: Total displacement A-A of GR Figure 8.11: Total displacement A-A of GR
for Mohr Coulomb model for Hardening Soil model
8.2 Comparison numerical results to scale test K2 results

The load distribution and displacements calculated with Plaxis are compared to the
measured scale test results in this paragraph.
8.2.1 Load distribution

The load distribution for the Plaxis models and the scale test are plotted in Figure 8.12, to
Figure 8.15. The load distribution is split up in the total load on the pile (A+B), the load
transferred through arching (A), the load transferred through the GR and the load
transported to the subsoil (C). The composition of the figures is described in paragraph
5.5.6. The determination of the loads is described in paragraph 7.6.
The load transferred to the pile (load part A+B) in the Plaxis calculations are in agreement
with the measured results as shown in Figure 8.12. This is logical, because this A+B is a
function of the total load in the system (A+B+C) and the subsoil support C, which is both
prescribed in the calculation.
Load part C is prescribed in the calculations. The load in Plaxis should be the same as the
measured load in the scale test. From Figure 8.13 is concluded that load part C are indeed
the same in the calculations as in the measured scale test.
The load that is transferred to the piles through arching is load A. The development of load
part A through the load and drainage phases is shown in Figure 8.14. During the first part
of the tests, until the net load is ca. 11 kN/pile, the agreement between the measured and
calculated load parts is quite well. Then the calculated load part A diverges from the
measured results and is overestimated compared to the measured results of load part A.
This is applicable for the MC model and HS model, although the results of the HS model
are a bit better. The overestimation of load part A can indicate that the calculated
development of arching in the granular material is too high and therefore the properties of
the granular material are too high.
Load part B from the calculations is compared to the scale test results in Figure 8.15. This
figure shows that the load transferred through the GR (load part B) is underestimated by
MC model. The HS model approaches the measured load part B a little bit better, except
for the vacuum phase. In this phase load part B is overestimated.
The smooth relationship between the net load and load parts A and B found with the
measured results is in agreement with the calculated results (Figure 8.14 and Figure 8.15).
These figures also show that both A and B increase as a result of both subsoil drainage
(consolidation) and top load increase.

Load part A [kN/pile] Load parts A+B [kN/pile]
0
2
4
6
8
10
12
14
16
18
20
22
0
5
10
15
20
25
30
Drainage Drainage
0
0
Top load Top load
21 February 2011
Drainage Drainage
5
5
Top load Top load
Drainage Drainage
Figure 8.14: Load part A

10
10
Top load Top load
Figure 8.12: Load part A+B

Drainage Drainage
15
15
Net load [kN/pile]

Net Load [kN/pile]
Top load Top load
20
20
Drainage Drainage
Vacuum Vacuum
25
25
Load part C [kN/pile]
Plaxis HS
Plaxis MC
0
2
4
6
8
10
12
14
16
0
1
2
3
4
5
6
7
Scale test K2
Drainage Drainage
0
0
Top load Top load
Drainage Drainage
5
5
Top load Top load
Drainage Drainage
Figure 8.15: Load part B

Figure 8.13: Load part C
10
10
Top load Top load
Piled embankment with geosynthetic reinforcement – Master of Science thesis

Drainage Drainage
15
15
Net Load [kN/pile]

Net Load [kN/pile]
Top load Top load
20
20
Drainage Drainage
Vacuum Vacuum
25
25
77
Scale test K2 [%]

Plaxis MC [%]
Plaxis HS [%]
90
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
80 60
70
Load part A [% of total load]
Load part B [% of total load]

50
60
40
50
40 30
30
20
20
10
10
0 0
0 5 10 15 20 25 0 5 10 15 20 25
Net load [kN/pile] Net Load [kN/pile]
Figure 8.16: Load part A in percentage Figure 8.17: Load part B in percentage
Load parts A (A%) and B (B%) are presented in percentage of the total load (A+B+C) in
Figure 8.16 and Figure 8.17. The figure shows two types of loading: top load and drainage
of subsoil.
During the top load phase, the top load is increased. Load part A% decreases during the
top load step as shown in Figure 8.16. Load part A% measured in the scale test has a
tendency to return to the same value of approximately 50% each time the top load is
increased. This is not found in the Plaxis results of load part A%. The percentage of load
part B also decreases as a result of a top load increase as seen in Figure 8.17. But this
decrease is limited. This is in agreement with the measured load B%.
The measured part A% of the scale test results has an lower value after each drainage step
than at the end of the previous drainage step. This also is not confirmed by the calculated
load part A% in Figure 8.16.
The scale test K2 starts with one drainage step with a zero top load. Arching is found
immediately, as shown by the relatively high initial measured A% value of 54%. This high
value is also found in Plaxis, however the high values are 44% for MC model and 47% for
HS model.

21 February 2011
8.2.2 Displacements
The displacements of the Plaxis models at z1, z2 and z3 are compared to the measured
displacements. The location of the displacements are shown in Figure 8.1. The
displacement results are presented in Figure 8.18, Figure 8.19 and Figure 8.20. The
composition of the figures is described in paragraph 0.
Comparison of the Plaxis displacements to the scale tests displacements, shows clearly
that the Plaxis displacements are underestimated significantly. Farag (2008) also found
much lower settlements in his Plaxis calculations. In the CUR 226 (2010) this is solved by
modelling a gap underneath the GR in the Plaxis calculations. The underestimation of the
displacements can be caused by different parts of the model. This will be analysed in
section 8.3.
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
60
50
Displacement z1 [mm]
40
Scale test K2
30
Plaxis MC
Plaxis HS
20
10
0
0 5 10 15 20 25
Net Load [kN/pile]
Figure 8.18: Displacement z1 (Centre 4

piles)

Scale test K2
Plaxis MC
Plaxis HS
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Top load
Top load
Top load
Top load
Vacuum
Vacuum
60 60
50 50
40 40
30 30
20 20
10 10
0 0
0 5 10 15 20 25 0 5 10 15 20 25
Net Load [kN/pile] Net Load [kN/pile]
Figure 8.19: Displacement z2 (Centre 2 Figure 8.20: Displacement z3 (Close to

piles) pile)
8.3 Analysis of numerical model results

From the comparison of the results was concluded that the displacements of the Plaxis
calculations do not agree with the displacements of the scale test. The calculated GR
settlements are namely much too small. This inconsistency can be caused by: subsoil, GR
and granular material. These causes are analysed consecutively in this paragraph. For this
analyses, several Plaxis calculations are carried out. These calculations are all variations
on one basic calculation, using the Hardening Soil model to model the fill behaviour. The
results of the basic calculation with the HS model are given by a red sphere .
8.3.1 Subsoil
In the scale test, a foam cushion models the behaviour of the subsoil. The calculated
settlements of the GR are influenced by two parts of the foam cushion: the stiffness of the
(foam in the) foam cushion and the water pressure in the foam cushion.
Stiffness of the subsoil

The stiffness of the foam cushion is very low, therefore the influence of the stiffness on the
displacement of the GR is expected to be very small. To investigate the influence of the
stiffness on the displacement the Young’s Modulus of 10 kN/m2 is changed to 1 kN/m2. The
results of decreasing the Young’s Modulus can be seen in Figure 8.21 to Figure 8.27. From
the figures it is concluded that the influence of the stiffness is very small, because the
resulting load parts A and B and the displacements do not deviate from the basic HS
model. Therefore the influence of the stiffness is too small to give an explanation for the
small calculated GR settlements.

21 February 2011
Scale test K2
Plaxis HS with foam cushion E=10
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
Vacuum
30
6
25
5
Load parts A+B [kN/pile]

20
4
15 3
10 2
5 1
0 0
0 5 10 15 20 25 0 5 10 15 20 25
Figure 8.21: Load part A+B, stiffness foam Figure 8.22: Load part C, stiffness foam
cushion cushion
22
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
16 Vacuum
Vacuum
20
18 14
16 12
14
10
12
10 8
8 6
6
4
4
2
2
0 0
0 5 10 15 20 25 0 5 10 15 20 25
Figure 8.23: Load part A, stiffness foam Figure 8.24: Load part B, stiffness foam
cushion cushion

Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
60
50
40
Scale test K2
30
20
10
0
0 5 10 15 20 25
Net Load [kN/pile]
Figure 8.25: Displacement z1 (centre 4

piles), stiffness foam cushion
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Top load
Top load
Top load
Top load
Vacuum
Vacuum
60 60
50 50
40 40
30 30
20 20
10 10
0 0
0 5 10 15 20 25 0 5 10 15 20 25
Figure 8.26: Displacement z2 (centre 2 Figure 8.27: Displacement z3 (close to

piles), stiffness foam cushion pile), stiffness foam cushion

21 February 2011
Water pressure in the subsoil

There is a relatively high water pressure in the foam cushion, therefore it is expected that
the water pressure has a large influence on the displacements. To investigate this, several
analysis were performed:
1. Investigation of the development of the water pressure in the foam cushion;
2. Calculations with three alternative Plaxis models:
a. Model without foam cushion;
b. Model without water pressure;
c. Model without water pressure and foam cushion.
The water pressure development in the foam cushion of the MC model and HS model are
shown in Figure 8.28. It should be noted that Plaxis calculates the water pressures in the
stress points of the elements. These stress points do not lie at the same locations as the
element nodes. However, in the plot, the water pressure is extrapolated to the element
nodes. Therefore it seems that there is no water pressure at the top of the foam cushion
(underneath the sand layer). Then the water pressure seems to increase rapidly until the
prescribed value is reached. This is the value in the stress point, that determines the water
pressure at the top of the foam cushion in the calculations. From there the water pressure
increases hydrostatically. The MC model shows a peak in top of the foam cushion and the
HS model does not show this peak value.
Figure 8.29 to Figure 8.35 compare the basis HS calculation with the variations a, b and c.
A general conclusion from these figures is that by excluding the water pressure from the
models, the results of the drainage phases are the same as the result of the top load
phase.
When the foam cushion is excluded from the model (a), the total load on the pile and the
water pressure in the foam cushion do not change in comparison to the basic HS model.
This is seen in Figure 8.29 and Figure 8.30. The development of load part A during the
phases is shown in Figure 8.31 and for load part B in Figure 8.32. Load part A and B
deviated after applying a top load of 75kPa, they both decrease. This has effect on the
calculated displacements, the displacement are a bit larger. The displacements of variation
are still underestimated when compared to the scale test results.
Comparing the results of the model (b) and model (c) shows no differences. This means
that the influence of the stiffness of the foam cushion is not visible in the results. The water
pressure controls the stress development in the embankment fill. In Figure 8.29 is shown
that when the water pressure is excluded from the model, the load on the pile (load parts
A+B) increases and is constant for each top load. This is as expected, as C is changed to
zero (Figure 8.30). The calculated A+B of the variations with no water pressure approaches
load part A+B of the HS model for the situations with a small water pressure (end of
drainage phases).The development of load part A during the phases is shown in Figure
8.31 and for load part B in Figure 8.32. Load A does not change much, on the other hand
load B increases significantly. This is because load C is zero and the total weight of the
granular material and the top load are transported to the pile.
Removing the water pressure (models b and c) change the calculated settlements
considerably. During the first phases the calculated displacements are even larger than the
displacements measured in the scale test as seen in Figure 8.33, Figure 8.34 and Figure
8.35. However, during the top load of 50 kPa (around 10 kN/pile) the calculated
displacement becomes smaller than measured in the scale test. The calculated
development of the settlements during the test differ considerably from the measured
settlements.

From the variations a, b and c can be concluded that the influence of the foam cushion on
the load distribution and displacements is very small. When only the water pressure is
modelled (model a), the calculated displacements agree a bit better with the measurements
than the those calculated with the basic HS model. By excluding the water pressure from
the model, the displacements are still under estimated compared to the measured
displacements.
0,44 m
0,2 m
Figure 8.28: Water pressure in foam cushion in the second drainage phase with a load of
100kPa, location is z2 of MC model (left) and of HS model (right).
It should be noted that the water pressure is calculated in the stress points, while the
results are given in the element nodes. This explains why the water pressure on top of
the foam cushion seems to be reduced to zero. This not the case in the calculations.

21 February 2011
Scale test K2
Basic model HS
a. No foam cushion
b. No water pressuer
c. No foam cushion, no water pressure
7
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
Vacuum
30
6
25

20
4
15
3
10 2
5 1
0 0
0 5 10 15 20 25 0 5 10 15 20 25
Figure 8.29: Load part A+B, water pressure Figure 8.30: Load part C, water pressure
22
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
16
Vacuum
20
18 14
16 12
14
10
12
10 8
8 6
6
4
4
2
2
0 0
0 5 10 15 20 25 0 5 10 15 20 25
Figure 8.31: Load part A, water pressure Figure 8.32: Load part B, water pressure

Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
60
50
40
Scale test K2
Basic model HS
30
a. No foam cushion
b. No water pressuer
20 c. No foam cushion, no water pressure
10
0
0 5 10 15 20 25
Net Load [kN/pile]

piles), water pressure
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Top load
Top load
Top load
Top load
Vacuum
Vacuum
60 60
50 50
40 40
30 30
20 20
10 10
0 0
0 5 10 15 20 25 0 5 10 15 20 25

piles), water pressure pile), water pressure

21 February 2011
8.3.2 Geosynthetic Reinforcement

An incorrect stiffness of the GR can be a cause of the difference in settlements. During the
scale test, both the GR settlements and the total vertical load on the GR (load part B) is
measured. To check the behaviour of the GR, a Plaxis model is calculated with
approximately only the GR and load B. The foam cushion and most of the fill are removed.
To simplify the model, load B is simulated as an equally distributed load. The equally
distributed load cannot be applied directly on the GR, therefore a 5 cm layer of granular
material is applied and then the equally distributed load. The geometry is shown in Figure
8.36. The results of the Plaxis model are presented in the Figure 8.37 to Figure 8.40.
Figure 8.36: Geometry of Plaxis model with only load B
From Figure 8.37 is concluded that the load on the GR in the Plaxis model corresponds
with load B measured in the scale test. The displacements of z1, z2 and z3 are shown in
Figure 8.38, Figure 8.39 and Figure 8.40. The differences between calculated and
measured displacements are smaller. The calculated displacements z1 and z2 are larger
than the measured displacements. However, the (last phase) calculated z3 is smaller than
the measured z3. This is caused by the simplified equally distributed load. In the final part
of the test, the calculated and measured average of z1, z2 and z3 agree quite well.
Therefore, this model shows that the GR in the model simulates the GR in the model quite
well, and that the stiffness of the GR in the Plaxis calculation does not give too small GR
settlements. Thus, the model for the GR is not one of the causes of the too small calculated
GR settlements, and the applied stiffness gives good results.
However, it is still useful to consider the influence of the GR stiffness in the calculations.
The influence of the GR stiffness is analysed by calculating a model with a GR stiffness of
1565 kN/m. The results are presented in Figure 8.41 to Figure 8.47. From the figures is
concluded that the influence of the stiffness on the results is very small.

Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
16
14
12
10
8
Scale test K2
6 Plaxis equal distributed load B
0
0 5 10 15 20 25
Net Load [kN/pile]
Figure 8.37: Load part B, GR

21 February 2011
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
60
50
40
30 Scale test K2
Plaxis equal distributed load B
20
10
0
0 5 10 15 20 25
Net Load [kN/pile]

piles), GR
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Top load
Top load
Top load
Top load
Vacuum
Vacuum
60 60
50 50
40 40
30 30
20 20
10 10
0 0
0 5 10 15 20 25 0 5 10 15 20 25

piles), GR pile), GR

Scale test K2
Plaxis HS EA=2269
Plaxis HS EA=1565
7
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
Vacuum
30
6
25

20
4
15
3
10 2
5 1
0 0
0 5 10 15 20 25 0 5 10 15 20 25
Figure 8.41: Load part A+B, stiffness GR Figure 8.42: Load part C, stiffness GR
22
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
16
Vacuum
20
18 14
16 12
14
10
12
10 8
8 6
6
4
4
2
2
0 0
0 5 10 15 20 25 0 5 10 15 20 25
Figure 8.43: Load part A, stiffness GR Figure 8.44: Load part B, stiffness GR

21 February 2011
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
60
50
40
Scale test K2
30
Plaxis HS EA=2269
Plaxis HS EA=1565
20
10
0
0 5 10 15 20 25
Net Load [kN/pile]

piles), stiffness GR
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Top load
Top load
Top load
Top load
Vacuum
Vacuum
60 60
50 50
40 40
30 30
20 20
10 10
0 0
0 5 10 15 20 25 0 5 10 15 20 25

piles), stiffness GR pile), stiffness GR

8.3.3 Granular material

The behaviour of the constitutive model of the granular material can also cause the
displacement differences. In this chapter, the vertical stress distribution in the granular
material are discussed, and after that, the parameters of the granular material are
discussed.
Effective vertical stresses in granular material

The vertical stress in the granular material will increase linearly with depth if no arching
occurs. This is different when arching occurs. Zaeske (2001) for example, assumes that the
vertical stresses in the granular material have a distribution as shown in Figure 3.1. His
model is adopted in both EBGEO and CUR226. Between the piles, the vertical stress
should reduce below the arch. Closer to the piles, the maximum vertical stress should be
found at a lower position. For locations A-J on cross section C-C the vertical stresses
calculated by Plaxis are plotted in Figure 8.49 to Figure 8.51. It should be noted that the
figures are not plotted on the same scale. In all figures, the vertical load at ground level
equals 100 kPa. The figures clearly show that the maximum vertical stresses at locations C
and G are found at a lower position than the maxima at locations B and H. This is in
agreement with Zaeske’s model in Figure 3.1.
At location A (Figure 8.49), the maximum is found (more or less) around ground level. This
indicates that the fill is relatively thin, just enough for the arch to develop fully (full arching)
or even less than that (partial arching). This is in accordance with the design of the test
series: purpose was to analyse tests with a minimum height of the fill, so that the arch could
possibly not develop fully.
The largest vertical stresses are found at the positions on top of the piles (locations D, E
and F). At these positions, the vertical stress increase with depth, also in accordance with
Zaeske’s model in Figure 3.1. The maximum vertical stress is found in the centre of the pile
(location E). The difference between locations C and G with the locations on the piles show
that the vertical stress concentrates strongly in the area on and just around the piles. This
concentration is stronger than suggested by Zaeske.
The large difference between locations C and D, and also between locations F and G
shows that it would have been better to use a finer mesh just around the pile. This is not
done because of the limited possibilities of the available computers.
Figure 8.48: Location of cross section C-C and locations A-J in cross section and top
view of the model

21 February 2011
0,44 m 0,44 m 0,44 m

0,44 m
0,02 m 0,02 m 0,02 m

0,02 m
0,2 m 0,2 m 0,2 m

0,2 m
Location A Location B Location C Location D

Figure 8.49: Effective vertical stress distribution for locations A to D 1)
0,44 m 0,44 m 0,44 m 0,44 m
0,02 m 0,02 m 0,02 m 0,02 m
0,2 m 0,2 m 0,2 m 0,2 m
Location E Location F Location G Location H

Figure 8.50: Effective vertical stress distribution for locations E to H 1)
0,44 m 0,44 m
0,02 m 0,02 m
0,2 m 0,2 m
Location I Location J
Figure 8.51: Effective vertical stress distribution for locations I to J 1)
1) Figures are not on the same scale and at ground level the vertical stress is 100kPa

Tensile stresses
In location C (Figure 8.49) and location G (Figure 8.50), located near the pile there seem to
occur tensile stresses in the granular material. There should not be tensile stresses in the
granular material. To examine the correctness of Figure 8.49 to Figure 8.51, the tension
points in the elements are plotted in Figure 8.52. From this figure can be concluded that the
tensile forces occur in the steel frame and pile, but not in the granular material. The tensile
forces in Figure 8.49 to Figure 8.51 probably are caused by extrapolation of the stresses in
the integration points to the elements nodes.
0.6
0.5
0.4
y-axis [m]
0.3
0.2
0.1
y
0 -0.5
-0.5 -0.4
-0.4 -0.3
-0.3 -0.2 -0.1 0
-0.1
-0.2
z x
x-axis [m]
z-axis [m]
Figure 8.52: Effective vertical tensile stresses in the stress points of the elements
The distribution of the effective vertical stresses on the GR determines the displacement of
the GR. Therefore the location of the high vertical effective stresses is analysed. Figure
8.53 confirms the observation in Figure 8.49 to Figure 8.51 that the vertical effective
stresses are concentrated strongly directly on top of and next to the pile. The vertical
effective stresses on the GR are relatively small compared to the stresses directly next to
the pile. This is not in accordance with EBGEO (2010) and CUR226 (2010), that assumes
that the maximum vertical stress on the GR is found in the centre of a tensile strip, just
between two adjacent piles (Figure 3.3). The transition zone around the pile tip could be
analysed more accurately with a finer mesh.
E E
Figure 8.53: Vertical effective stresses in cross section E-E

21 February 2011
Strength parameters of the granular material

Arching develops due to shear resistance. In the constitutive models, the shear strength is
determined by cohesion and internal friction angle. This chapter analyses whether a too
high values for the granular strength parameters can be one of the causes of the too low
calculated GR settlements.
In the granular material there is no cohesion. Therefore, changing the cohesion is not
considered. By reducing the internal friction angle, arching will develop less and the
settlements will become larger. The internal friction angle was determined from the results
of three tests:
1. triaxial tests;
2. a tilting square box;
3. the spontaneous slope of the model fill.
The triaxial test (1) is described in Chapter 6. The square box (2) was turned until the
material starts falling out of the box. Tilting a square box will give a relatively small internal
friction angle, because the density is low compared to the model- and triaxial tests. When
the top box of the scale test set-up was removed after completion of the scale test, the
granular material fell down until a natural slope developed (3). The angle of this slope is
measured. The values of the internal friction angle for all these three tests lie between 42
and 49 degrees. Therefore changing the internal friction angle would be unrealistic.
However, these measurements all gave the peak values for the internal friction angle.
Figure 8.54 shows that, for ungoing deformation, the strength of the granular material
reduces after reaching the peak strength. This is also observed in the triaxial test with the
greatest confining stress, see Figure 6.7,the shear strength reduces with about 10%. This
occurs in the shear planes that occur within the fill during the test. It is possible that the
influence of this softening has an important role in the test results. Therefore, the
calculations are repeated with a smaller internal friction angle of 41 degrees. The results
are presented in Figure 8.55 to Figure 8.61.

Figure 8.54: Results of triaxial tests on samples (CUR-166, 2005)
The results show that a smaller internal friction angle in the calculations indeed gives larger
deformations, Figure 8.59, Figure 8.60 and Figure 8.61. On the other hand load part A is a
bit smaller than the measured load A (Figure 8.57) and load part B (Figure 8.58) is a bit
larger. The load transported through the arch seems to collapse in the second drainage
phase with a top load of 75kPa. Thus part of the too small calculated displacement is
caused by the application of the peak internal friction angle.

21 February 2011
Scale test K2
Plaxis HS phi=47
Plaxis HS phi=41
7
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
Vacuum
30
6
25

20
4
15
3
10 2
5 1
0 0
0 5 10 15 20 25 0 5 10 15 20 25
Figure 8.55: Load part A+B, decrease phi Figure 8.56: Load part C, decrease phi
22
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
16
Vacuum
20
18 14
16 12
14
10
12
10 8
8 6
6
4
4
2
2
0 0
0 5 10 15 20 25 0 5 10 15 20 25
Figure 8.57: Load part A, decrease phi Figure 8.58: Load part B, decrease phi

Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
60
50
40
Scale test K2
30
Plaxis HS phi=47
Plaxis HS phi=41
20
10
0
0 5 10 15 20 25
Net Load [kN/pile]

piles), decrease phi
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Top load
Top load
Top load
Top load
Vacuum
Vacuum
60 60
50 50
40 40
30 30
20 20
10 10
0 0
0 5 10 15 20 25 0 5 10 15 20 25

piles), decrease phi pile), decrease phi

21 February 2011
It is concluded that a smaller friction angle results in less arching (A) increase during
drainage of the subsoil. This is in agreement with the results found by Van Eekelen et al.
(2011a). They compare two tests (tests T2 and T3) that are the same, except that the fill of
T2 is sand (phi =39 degrees), and the fill of T3 is the same as the fill of test K2. The tests
were nearly the same as K2, except that small difference in the sequence of subsoil
support and top load increase appeared. The results are plotted in Figure 8.62. The
differences between the measured tendency of the granular and sand test show much
agreement with the differences found in the calculated tendencies in Figure 8.57 and
Figure 8.58. This is especially true for the behaviour during the consolidation phases. It is
therefore concluded that obviously, the influence of the friction angle on the tendencies in
the arching behaviour is modelled quite well with the hardening soil model in Plaxis.
16 100% 12 100%
consoli-
consoli-
top load step
top load step
top load step
top load step
top load step
top load step

consolidation.
consolidation.
consolidation.
consolidation.
consolidation.
consolidation.
dation
dation
14 90% 90%
10
80% 80%
load part B (kN/pile) .

12
load part A (kN/pile) .
70% 70%
8
load part A (%)
load part B (%)

10 60%
60%
8 50% 6 50%
40% 40%
6
4
30% 30%
4
sand T2 (kN/pile) 20% 20%
granular T3 (kN/pile) 2
2 sand T2 (%) 10%
10%
granular T3 (%)
0 0% 0 0%
-5 0 5 10 15 20 -5 0 5 10 15 20
net load Wn (kN/pile) net load Wn (kN/pile)
Figure 8.62: Comparison of measured load distribution for tests T2 (sand) and T3
(granular) (Van Eekelen, et al., 2011a)
8.4 Conclusions
The Plaxis results are analysed and compared to the measured results of the scale test.
First the general conclusions are given, followed by the conclusions of the load distributions
and GR displacement. An important conclusion in this chapter is the deviated GR
displacement. The cause is investigated and the conclusions are described in the last
paragraph
General
 The load distribution and displacements calculated by the Hardening Soil model
give a better approximation of the scale test results than the Mohr Coulomb model.
 The Plaxis calculations give a smooth relationship between net load and load parts
A and B, and the GR settlements. This is in accordance with the measurements.
Load distribution
 During the phases of the test, arching is clearly observed in the granular material;
 During the first part of the test the load distribution of MC model and HS model
agree quite well with the measured load distribution;
 During the second part of the test the load transferred through arching is
overestimated and the load transferred through the GR is underestimated by Plaxis;
 The water pressure is in agreement with the water pressure measured in the scale
test.
 Plaxis finds an increasing arching during drainage of the subsoil (consolidation).
This is not in agreement with EBGEO (2010) and CUR 226 (2010), but in
agreement with the measurements.

 The vertical stress concentrates strongly on top of and just around the piles. This
confirms the conclusions of Van Eekelen et al. (2011a; 2011b, 2011c), that the
vertical load distribution on the tensile strips is different from the assumptions of
Zaeske (2001) as adopted in EBGEO (2010) and CUR226 (2010).
 The development of the vertical stress in depth shows that the arch has less height
close to the piles than in the centre between two piles.
Geosynthetic Reinforcement
 The displacements of the GR calculated with Plaxis are underestimated compared
to the scale test results. The possible causes of this underestimation are described
below.
 The tensile forces are concentrated in ‘tensile strips’ between the piles of the GR,
the exact location of the greatest tensile forces cannot be determined with this
coarse mesh;
 The largest displacement of the GR is found at the middle of four piles;
 Plaxis finds for a smaller friction angle less arching increase during consolidation.
This is in accordance with the measurements.
Deviated displacement
The Plaxis calculations approach the measured load distribution, but show far too low GR
displacement in comparison to the measurements. This is caused by a combination of
causes. This has been investigated and it is concluded that:
 The stiffness of the foam cushion does not have influence on the GR
displacements;
 By excluding the water pressure and foam cushion the calculated settlements
increase, but the measured settlements are not found.
 By excluding the foam cushion from the model the displacements improve a bit.
 The behaviour of the GR is not the cause of the too low GR displacements. When
only load B is applied on the GR the displacements of the GR are overestimated by
Plaxis in comparison with the measured scale test results.
 The vertical effective stresses are concentrated on and directly next to the pile. The
effective vertical stress on the GR between the piles is relatively small.
 Part of the too small calculated displacement is caused by the application of the
peak internal friction angle of the granular material.

21 February 2011
Chapter 9
Conclusions and recommendations

In this last chapter the conclusions of the numerical analysis are drawn and
recommendations are made.
9.1 Conclusions
Numerical analysis have been performed to better understand, and where possible, to
confirm the measured load distribution and deformations in the scale model tests.
The scale test is modelled with Plaxis 3D Tunnel. The geometry of the Plaxis model is one
quarter of the geometry of the scale test to reduce calculation time of the numerical
analysis. The geometry of the circular pile is converted to a square pile because circular
piles cannot be modelled in Plaxis 3D Tunnel. The model is build up of one pile, subsoil,
sand layer on the subsoil, geosynthetic reinforcement (GR) with steel frame and granular
material. For the sand and the granular material, the material parameters are determined
for the MC model and the HS model. The initial models are calculated with the MC model,
because the necessary calculation time with this model is shorter. Updated mesh is applied
in the model for the tensile forces in the GR. The scale test is driven by water pressure and
top load increase. To model the water pressure in Plaxis, a pressure head is assigned to
the subsoil clusters.
During the execution of the scale tests there is a loss of load due to friction. The measured
friction is between 10% to 20% of the soil weight and top load, therefore an important part
of the model. In Plaxis 3D Tunnel it is not possible to apply an interface at the back side of
the model (parallel to the front plane). Another option to simulate friction with an extra soil
segment with lower strength parameters is investigated. This option give good results with
the MC model. But the calculation time increases significantly when the extra soil segment
is applied in a model with the HS model. Therefore, it is not realistic to apply an extra soil
segment as friction. In the final model, the amount of load loss (friction) is subtracted from
the top load.
Scale test K2 is calculated with both the MC model and the HS model. From both models is
concluded that arching occurs in the granular material. Both models show that the tensile
forces in the GR are concentrated in ‘tensile strips’ between the piles. The exact location of
the greatest tensile forces cannot be determined with this coarse mesh The largest
displacement of the GR is found at the middle of four piles.
The model is driven by the water pressure in the foam cushion, therefore the water
pressure is an input value. The results show that the water pressure in the Plaxis
calculations corresponds with the measured water pressure in the scale test. In accordance
with the measurements, Plaxis calculations give a smooth relationship between net load
and load parts A and B, and the GR settlements. During the first part of the test, the load
distribution of MC model and HS model agree quite well with the measured load
distribution. During the second part of the test the load transferred through arching is
overestimated and the load transferred through the GR is underestimated. However the HS

Chapter 9 Conclusions and recommendations
model gives better results than MC model. In general Plaxis finds an increasing arch during
drainage of the subsoil (consolidation), this is in agreement with the measurements, but not
in agreement with EBGEO (2010) and CUR 226 (2010). The vertical stress concentrates
strongly on top of and just around the piles. This confirms the conclusions of Van Eekelen
et al. (2011a; 2011b, 2011c) that indicates that the vertical load distribution on the tensile
strips is different from the assumptions of Zaeske (2001) as adopted in EBGEO (2010) and
CUR226 (2010). The development of the vertical stress in depth shows that the arch has
less height close to the piles than in the centre between two piles. The displacement
calculated with Plaxis are underestimated compared to the scale test results. The causes of
this underestimation lie in several factors, as described below.
The Plaxis calculations show far too low GR displacement in comparison to the
measurements. There are several possible causes. To investigate these causes different
models have been calculated. From these models is concluded that the stiffness of the
foam cushion and the water pressure in the foam cushion does not have influence on the
GR displacement. The influence of the GR stiffness on the GR displacement is small. The
vertical effective stresses are concentrated on and directly next to the pile. The effective
vertical stress on the GR between the piles is relatively small. Part of the too small
calculated displacement is caused by the application of the peak internal friction angle of
the granular material. With a smaller internal friction angle less arching increase is found
during drainage of the subsoil. This is in accordance with the measurements.
9.2 Recommendations
The recommendations for the numerical analysis of the scale model tests are described
below.
The triaxial test has been performed with vacuum pressure. During these tests the exact
change of volumetric strain is not measured. The volumetric strain is approximated with the
horizontal and vertical strain. Triaxial tests performed with cell pressure measures the
volume change during the test. Therefore it is recommended to perform triaxial tests with
cell pressure.
The literature stated that the oedometer loading stiffness of the HS model as an influence
on the development of arching in the embankment. Therefore it is recommended to perform
a large oedometer test on the granular material.
To increase the accuracy of the Plaxis results, the mesh should be refined.
The loss of load (due to friction) during the scale test is an important part of the scale test. It
produces disturbance in the load distribution of the granular material. The friction should be
included in the Plaxis model by an interface, therefore it is advised to perform numerical
analysis of the scale model tests with the full 3D version of Plaxis.
The cause of the differences in calculated and measured displacement of the GR has to be
investigated in more detail.

21 February 2011
References
Blok, K.J.B.R. (2000). Tabellen voor bouwkunde en waterbouwkunde. Leiden: Spruyt, van
mantgem & de Does B.V.
BS 1377 : Part 8. (1990). British Standard Methods of tst for Soils for civil engineering purposes,
Part 8. Shear strength tests (effective stress): British Standards Institution, London,
BS 8006. (2010). Code of practice for strengthened/reinforced soils and other fills: British
Standards Institution, London,
CUR-publicatie 166 (vierde druk). (2005). Damwandconstructies: Stichting CUR, Gouda, ISBN
90-3760-063-8.
CUR-publicatie 226. (2010). Ontwerprichtlijn Paalmatrassystemen: Stichting CUR, Gouda, ISBN
978-90-376-0518-1.
CUR-rapport 2002-7. (2002). Gewapende granulaatmatras op palen. Toepassing, ontwerp en
uitvoeringsaspecten: Stichting CUR, Gouda, ISBN 9037602622.
Dalen, J.v. (2008). Soil- and embankment reinforcement on stiff elements. Presentation,
Technical University Delft: Rotterdam Public Works
Duncan, J.M., & Chang, C.Y. (1970). Non linear analysis of stress and strain in soils. Journal of
the Soil Mechanics and Foundations Division, Volume 96(SM5), pp. 1629-1653.
EBGEO (2010). Empfehlung für den Entwurf und die Berechnung von Erdkörper mit
Bewehrungen als Geokunstoffen: Deutsche Gesellschaft für Geotechnik e.V., Essen,
Entwurf 02/2009,
Farag, G.S.F. (2008). Lateral spreading in basal reinforce embankments supported by pile-like
elements. PhD thesis, Universität Kassel, Schrifenreihe Geotechnik, Heft 20.
Gangakhedkar, R. (2004). Geosynthetic reinforced pile supported embankments. Master thesis,
University of Florida, Florida.
Han, J., & Gabr, M.A. (2002). Numerical analysis of geosynthetic-reinforced and pile-supported
earth platforms over soft soil. [Article]. Journal of Geotechnical and Geoenvironmental
Engineering, 128(1), pp. 44-53.
Heitz, C. (2006). Bodengewölbe unter ruhender und nichtruhender Belastung bei
Berücksichtigung von Bewehrungseinlagen aus Geogittern. PhD thesis, Universität
Kassel, Schriftenreihe Geotechnik, Heft 19, Kassel.
Hewlett, W.J., & Randolph, M.F. (1988). Analysis of piled embankments. Ground Engineering,
21(3), pp. 12-18.
Hsi, J.P. (2001). Timber-piled embankments over soft ground. Proceedings of the 15th
International Conference on Soil Mechanics and Geotechnical Engineering, Istanbul,
27-31 August 2001: pp. 2085-2088 (Volume 3).
Itasca. Software FLAC. Retrieved 15 September, 2010, from
www.itascacg.com/flac3d/overview.php
Jones, C.J.F.P., Lawson, C.R., & Ayres, D.J. (1990). Geotextile reinforced piled embankments.
Proceedings of the 4th International Conference on Geotextiles Geomembranes and
Related Products, Den Haag, 28 May - 1 June 1990: pp. 155-160 (Volume 1).
Kempton, G.T., Russell, D., Pierpoint, N.D., & Jones, C.J.P.F. (1998). Two- and three-
dimensional numerical analysis of the performance of geosynthetics carrying
embankment loads over piles. Proceedings of the 6th International Conference on
Geosynthetics, Altanta, Georgia: pp. 767-772
Liu, H.L., Ng, C.W.W., & Fei, K. (2007). Performance of a geogrid-reinforced and pile-supported
highway embankment over soft clay: Case study. [Article]. Journal of Geotechnical and
Geoenvironmental Engineering, 133(12), pp. 1483-1493.
NEN 6743. (1991). Berekeningsmethoden voor funderingen op palen - drukpalen
Plaxis. Plaxis 3D Tunnel Reference Manual.
Plaxis. Plaxis home page. Retrieved 15 September, 2010, from www.plaxis.nl/page/home/
Russell, D., & Pierpoint, N.D. (1997). An assesment of design methods for piled embankments
Ground Engineering, November 1997, pp. 39-44.

References
Satibi, S. (2009). Numerical analysis and design criteria of embankments on floating piles. PhD
thesis, Universität Stuttgart, Stuttgart.
Slaats, H. (2008). Load Transfer Platform, Bending Moments in Slender Piles. Master thesis,
Technical University Delft.
Van Eekelen, S.J.M. (2010). Piled embankments; Dutch Standard. Presentation, Technical
University Delft:
Van Eekelen, S.J.M., Bezuijen, A., Lodder, H.-J., & Van Tol, A.F. (2011a). Experiments on piled
embankments: the influence of subsoil, fill and reinforcement. Geotextiles &
Geomembranes.
Van Eekelen, S.J.M., & Hulst, M.W.J. (2010a). Factual report piled embankment laboratorium
tests, Heusker geosynthetics. Delft: Deltares, report number 1001697-004-GEO-0002.
Van Eekelen, S.J.M., Jansen, H.L., Van Duijnen, P.G., De Kant, M., Van Dalen, J.H., Brugman,
M.H.A., et al. (2010b). The Dutch Design Guideline for Piled Embankments.
Proceedings of 9th International Conference on Geosynthetics, Brazil: pp. 1911-1916
Van Eekelen, S.J.M., Lodder, H.-J., & Bezuijen, A. (2011b). Load distribution on the
geosynthetic reinforcement within a piled embankment. Paper presented at the Soil
mechanics & geotechnical engineering.
Van Eekelen, S.J.M., Lodder, H.-J., & Bezuijen, A. (2011c). Vervormingen van geokunststoffen
in een paalmatras en de daaruit volgende belastingverdeling, Op weg naar
aanscherping van de ontwerprichtlijn deel 1. Geokunst.
Van Niekerk, A.A. (2002). Mechanical Behavior and Perfrmance of Granular Bases and Sub-
bases in Pavements. PhD thesis, Technical University Delft, Delft.
Zaeske, D. (2001). Zur Wirkungsweise von unbewehrten und bewehrten mineralischen
Tragschichten über pfahlartigen Gründungselementen. PhD thesis, Uni Kassel,
Zchriftenreihe Geotechnik, Heft 10, Kassel.

21 February 2011
Appendix

21 February 2011
Table of Contents
A CALCULATION STIFFNESS BEHAVIOUR OF FOAM CUSHION ................................... 115
B PROPERTIES OF SAND AND GRANULAR MATERIAL ................................................. 119
B.1 SAND ...................................................................................................................... 119
B.2 GRANULAR MATERIAL ............................................................................................... 127
C TEST PROCEDURE........................................................................................................... 129
D JOURNAL SCALE TEST K2 ............................................................................................. 137
E TRIAXIAL TEST ................................................................................................................. 141
E.1 CALCULATIONS ........................................................................................................ 141
E.2 RESULTS ................................................................................................................. 144
F RELATIVE DENSITY ......................................................................................................... 155
F.1 LOOSE BULK DENSITY .............................................................................................. 155
F.2 COMPACTED BULK DENSITY ...................................................................................... 155
F.3 RELATIVE DENSITY ................................................................................................... 157
G MATERIAL MODELS ......................................................................................................... 159
G.1 MOHR COULOMB MODEL .......................................................................................... 159
G.2 HARDENING SOIL MODEL .......................................................................................... 160
G.3 PLAXIS RESULTS OF TRIAXIAL TEST ........................................................................... 162
H NUMERICAL SIMULATION OF SOIL-STRUCTURE INTERFACE BEHAVIOUR ........... 165
H.1 GEOMETRY & PARAMETERS OF SOIL-STRUCTURE INTERFACE ..................................... 165
H.2 RESULTS PLAXIS OF SOIL-STRUCTURE INTERFACE BEHAVIOUR ................................... 167
H.3 VALIDATION SOIL-STRUCTURE INTERFACE MODEL ...................................................... 174
H.4 CONCLUSION OF SOIL-STRUCTURE INTERFACE BEHAVIOUR ........................................ 177
I RESULTS OF PLAXIS HARDENING SOIL MODEL ........................................................ 179
I.1 DEFORMED MESH .................................................................................................... 179
I.2 VERTICAL DISPLACEMENT ......................................................................................... 181
I.3 EFFECTIVE VERTICAL STRESSES ............................................................................... 188
I.4 TENSILE FORCES IN GEOSYNTHETIC REINFORCEMENT ................................................ 195
I.5 DEFORMATION OF GEOSYNTHETIC REINFORCEMENT .................................................. 202

21 February 2011
List of Figures
Figure A.1: Water load on foam cushion ........................................................................... 115
Figure A.2: Equilibrium points ........................................................................................... 116
Figure A.3: Equilibrium points ........................................................................................... 117
Figure E.1: Axial displacement - Time curve for Sample 16kPa ....................................... 145
Figure E.2: Radial top displacement - Time curve for Sample 16kPa ............................... 145
Figure E.3: Radial bottom displacement - Time curve for Sample 16kPa ......................... 145
Figure E.10: Stress – Time curve for Sample 16kPa ........................................................ 148
Figure E.13: Deviatoric stress – Axial strain curve for Sample 16kPa .............................. 150
Figure E.14: Volumetric strain – Axial strain curve for Sample 16 kPa ............................. 150
Figure E.19: Young’s modulus – Axial strain curve ........................................................... 153
Figure E.20: Poisson ratio – Axial strain curve ................................................................. 153
Figure E.21: Cohesion and internal friction angle – Axial strain curve .............................. 154
Figure E.22: Dilatancy angle – Axial strain curve .............................................................. 154
Figure F.1: Graph dry density against moisture content ................................................... 157
Figure G.1: Deviatoric stress strain curve ......................................................................... 163
Figure G.2: Volume change curve .................................................................................... 163
Figure H.1: 2D intersection Plaxis and dimensions ........................................................... 165
Figure H.2: Reference model: cross section (left) and 3D Plaxis geometry (right) ............ 166
Figure H.3: Interface model: cross section (left) and 3D Plaxis geometry (right) .............. 167
Figure H.4: Soil segment model: cross section (left) and 3D Plaxis geometry (right)........ 167
Figure H.5: Vertical effective stresses in the granular fill of the second drainage phase,
calculation results of the Interface model .......................................................................... 168
Figure H.6: Vertical effective stresses in the granular fill of the second drainage phase,
calculation results of the Soil segment model ................................................................... 168
Figure H.7: Vertical effective stresses in the granular fill of the vacuum phase, calculation
results of the Interface model ............................................................................................ 168
Figure H.8: Vertical effective stresses in the granular fill of the vacuum phase, calculation
results of the Soil segment model ..................................................................................... 168
Figure H.9: Tensile force N1 within the GR in x direction in the Interface model .............. 169
Figure H.10: Tensile force N1 within the GR in x direction in the Soil segment model...... 169
Figure H.11: Load distribution ........................................................................................... 169
Figure H.12: Load part A+B .............................................................................................. 170
Figure H.14: Load part A ................................................................................................... 170
Figure H.15: Load part B ................................................................................................... 170
Figure H.16: Location of displacement measurements of scale test K2............................ 171
Figure H.17: Displacement z1 (Centre 4 piles) ................................................................. 172

List of Figures

Figure H.19: Displacement z3 (Close to pile) ................................................................... 172
Figure H.20: Friction ......................................................................................................... 174
Figure H.22: Load part C .................................................................................................. 175
Figure H.23: Load part A .................................................................................................. 175
Figure H.24: Load part B .................................................................................................. 175
Figure H.27: Displacement z3 (Close to pile) ................................................................... 176
Figure H.28: Friction ......................................................................................................... 177
Figure I.1: Deformed mesh A-A with HS model of phase Drainage 1 ............................... 179
Figure I.2: Deformed mesh A-A with HS model of phase Drainage 25.2 .......................... 179
Figure I.5: Deformed mesh A-A with HS model of phase Drainage 100.2 ........................ 180
Figure I.6: Deformed mesh A-A with HS model of phase Vacuum ................................... 180
Figure I.7: Vertical displacement A-A with HS model of phase Drainage 1 ...................... 181
Figure I.8: Vertical displacement A-A with HS model of phase Load 25kPa ..................... 181
Figure I.9: Vertical displacement A-A with HS model of phase Drainage 25.1 ................. 182
Figure I.10: Vertical displacement A-A with HS model of phase Drainage 25.2 ............... 182
Figure I.11: Vertical displacement A-A with HS model of phase Load 50kPa ................... 183
Figure I.14: Vertical displacement A-A with HS model of phase Load 75kPa ................... 184
Figure I.17: Vertical displacement A-A with HS model of phase Load 100kPa ................. 186
Figure I.18: Vertical displacement A-A with HS model of phase Drainage 100.1 ............. 186
Figure I.19: Vertical displacement A-A with HS model of phase Drainage 100.2 ............. 187
Figure I.20: Vertical displacement A-A with HS model of phase Vacuum ......................... 187
Figure I.21: Effective vertical stress A-A with HS model of phase Drainage 1 .................. 188
Figure I.22: Effective vertical stress A-A with HS model of phase Load 25kPa ................ 188
Figure I.23: Effective vertical stress A-A with HS model of phase Drainage 25.1 ............. 189
Figure I.31: Effective vertical stress A-A with HS model of phase Load 100kPa .............. 193
Figure I.32: Effective vertical stress A-A with HS model of phase Drainage 100.1 ........... 193
Figure I.33: Effective vertical stress A-A with HS model of phase Drainage 100.2 ........... 194
Figure I.34: Effective vertical stress A-A with HS model of phase Vacuum ...................... 194
Figure I.35: Tensile forces N1 of phase Drainage 1 ......................................................... 195
Figure I.36: Tensile forces N1 of phase Load 25kPa ........................................................ 195
Figure I.37: Tensile forces N1 of phase Drainage 25.1 .................................................... 196
Figure I.40: Tensile forces N1 model of phase Drainage 50.1 .......................................... 197
Figure I.45: Tensile forces N1 of phase Load 100kPa ...................................................... 200
Figure I.46: Tensile forces N1 of phase Drainage 100.1 .................................................. 200

21 February 2011
Figure I.47: Tensile forces N1 of phase Drainage 100.2 ................................................... 201

Figure I.48: Tensile forces N1 of phase Vacuum .............................................................. 201
Figure I.49: Deformation of geosynthetic reinforcement of phase Vacuum ....................... 202

21 February 2011
List of Tables
Table A.1: Equilibrium points in load and volume change during the water test ............... 115
Table A.2: Determination stiffness in load and volume change ........................................ 117
Table B.1: Properties Itterbeck sand 125-250 m ............................................................. 119
Table B.2: Internal friction angle  (o) Itterbeck sand 125-250 m .................................... 119
Table B.3: Sieve Results granular material ....................................................................... 127
Table F.1: Loose bulk density ........................................................................................... 155
Table F.2: Compacted dry density .................................................................................... 156
Table F.3: Relative density ............................................................................................... 157
Table G.1: Mohr Coulomb model properties ..................................................................... 160
Table G.2: Hardening Soil model properties ..................................................................... 161
Table G.3: Density and relative density of the triaxial tests and scale test ........................ 162
Table H.1: Material properties for the extra soil segment in the Soil segment model........ 166

21 February 2011
Appendix A
Calculation stiffness behaviour of foam

cushion
The stiffness behaviour of the foam and wrap of the foam cushion is determined with a
water-test. Water is applied to simulate a uniform load. Arching will not occur in water,
because there are no shear stresses. The water-test is shown in Figure A.1.
Figure A.1: Water load on foam cushion
First a layer of water is placed on the foam cushion. Than the taps of the foam cushion
were opened. Due to the weight of the water, the water in the foam cushion is able to drain
from the cushion freely. When the water stopped flowing out of the cushion, the volume
change of the cushion stopped and equilibrium is reached for a certain load step. This is
done for different load steps. The equilibrium points are presented in Table A.1: .
Equilibrium point Average load Volume change

[-] [kN] [m3]
1 0.81 1.39E-03
2 2.25 7.68E-03
3 3.65 4.10E-02
4 3.97 5.97E-02
5 5.44 8.15E-02
Table A.1: Equilibrium points in load and volume change during the water test

Appendix A Calculation stiffness behaviour of foam cushion
To determine the stiffness the differences in loads and volume have to be considered, the
following equation is used to calculate the stiffness:
dF
dp dF
k  A  (A.1)
z dV dV
A
Where:
3
k kN/m Stiffness (sub grade soil reaction)
dp kN/m2 Change in pressure
z m Height
A m2 Area
dF kN Change in total vertical force on the cushion
3
dV m Volume change due to corresponding load step
The equilibrium points and a trend line are plotted.
Figure A.2: Equilibrium points
From the graph follows the stiffness:

dFvert
k  50 kN / m3
d  vol
Another way of calculating the average stiffness is by calculating the stiffnesses for the
differences between the equilibrium points. The differences between the equilibrium points
and the calculated stiffnesses are given in Table A.2.

21 February 2011
Difference between Load difference Volume change Stiffness

tow equilibrium points deference
[Numbers] [kN] [m3] [kN/m3]
1-2 1.43 6.29E-03 228
1-3 2.83 3.93E-02 72
1-4 3.16 5.83E-02 54
1-5 4.63 8.01E-02 58
2-3 1.40 3.33E-02 42
2-4 1.72 5.20E-02 33
2-5 3.20 7.38E-02 43
3-4 0.32 1.87E-02 17
3-5 1.80 4.06E-02 44
4-5 1.47 2.19E-02 67
Table A.2: Determination stiffness in load and volume change
Figure A.3: Equilibrium points
The values of Table A.2 are comparable, except for the first value. The high first value can
be explained by the influence of the initial situation of the cushion. A small overpressure in
the foam cushion at the beginning of the test will drain water out of the cushion. Or when
there is an under pressure at the beginning of the test the drained water will be less. From
the table can be concluded that there was an under pressure at the beginning of the test.
The first value is therefore neglected.
Figure A.3 shows the calculated stiffnesses and the average stiffness. The average
stiffness, without the first value, is determined at 48 kN/m3.

21 February 2011
Appendix B
Properties of sand and granular material

The properties of sand and granular material are described below.
B.1 Sand
Itterbeck sand has a fraction of 125-250 μm. Deltares uses Itterbeck sand often in
laboratory tests.
The porosity is determined with:

V
n p
Vs
With:
n - Porosity
3
Vp m Pore volume
3
Vs m Total volume
A number of laboratory tests have been performed by Deltares to determine the properties
of the Itterbeck sand. The general properties are described below. The sieve curve is given
on the next page.
Minimal porosity nmin 0.345 -

Maximal porosity nmax 0.465 -
Unit weight  2643 kg/m3
Table B.1: Properties Itterbeck sand 125-250 m
Strain
n=0.453 n=0.408 n=0.368
(%)
2 28.2 35.27 39.78
5 31.24 35.96 40.80
10 32.48 34.59 35.37
At tu 32.52 36.03 40.93
Table B.2: Internal friction angle  (o) Itterbeck sand 125-250 m

Appendix B Properties of sand and granular material
100,000
10,000
Particlesize (mm)
1.400
Sieve size
1,000
1.000
0.710
0.500
0.355
0.250
0.212
0.180
0.150
0.125
0,100
0.106
0.090
0.070
0.063
0.053
Itterbeck sand
0.038
0.016
0,010
100
90
80
70
60
50
40
30
20
10
cummulatief weight passing (%)

21 February 2011
Triaxial test sample 1


21 February 2011


21 February 2011


21 February 2011
B.2 Granular material

The granular material used during the scale tests has a fraction of 0 - 16.5 mm. The
fraction is determined by sieving the material. The sieve results are presented in Table B.3.
In the table the results are compared to the limits according to the Dutch Rules given in
‘Standaard RAW bepalingen 2005’. The internal friction is determined in the next
paragraph.
Summation of Between
weight Lower lower limit
Size sieve Upper limit*
percentage of limit* and upper
particles < d limit?
mm % % %
63 okay
45
31.5 100 okay
22.4 100
16 96 75 100 okay
11.2 79
8 62 50 90 okay
5.6 48
4 39 30 75 okay
2 30 20 60 okay
1 25 13 45 okay
0.500 20 8 35 okay
0.250 13
0.125 7
0.063 4.1 0 9 okay
Table B.3: Sieve Results granular material

100,000
75
31.5
22.4
20
16
11.2
10,000
8.0
5.6
4.0
2.0
Particlesize (mm)
Sieve size
1,000
1.0
0.5
0.250
0.125
0,100
0.063
Granular material
0,010
100
90
80
70
60
50
40
30
20
10
cummulatief weight passing (%)

10

21 February 2011
Appendix C
Test procedure
The steps of the test procedure are described in detail below.
1. Applying grease to reduce friction along

bottom box walls and piles;
2. Placing bottom plate, piles, force

transducers, foam cushion and
pressure cells in the bottom box;
3. Soaking the foam cushion with water;
4. Starting the registration;

Appendix C Test procedure
5. Cover piles and foam cushion with non-

woven geotextile;
6. Applying sand on the non-woven;
7. Attaching geosynthetic reinforcement to

the steel frame;
8. Placing steel frame on the sand layer;

21 February 2011
9. Attaching bicycle gear cables and

settlement transducers to the GR and
installing pressure cells;
10. Measuring weight bottom box;
11. When second GR is applied:

a. Placing 5 or 10 cm fill material;
b. Compact the fill material;
c. Placing pressure cells below top
GR;
d. Placing top GR;
e. Fixing bicycle gear cables and
settlement transducers to top GR
and installing pressure cells;
f. Measuring weight bottom box;
12. Placing rubber sheet in the steel top box;
13. Placing the top box on the bottom box;

14. Applying and compacting the

embankment in layers until 42 cm;
15. Measuring weight bottom and top part

of the box to calculate weight of the
embankment fill;
16. Making xy-scan of the ground surface;
17. Set the settlement transducers and strain gauges to zero and start recording data;

21 February 2011
18. Cover embankment with a sheet and

placing the top water cushion;
19. Cover cushion with wooden plate;
20. Cover wooden plate with steel plate;
21. Placing distance keepers on top of

steel plate;
22. Closing the box with cover and placing

screw threads and bolts;
23. Consolidation as a result of the weight of the embankment by draining the foam
cushion with 6 l;

24. Applying top load in steps of 25 kPa;
25. Around two hours after applying top load the foam cushion is drained with 6 litres;
26. After two hours the loading and consolidation steps are repeated until a top load of
100 kPa;
27. Using vacuum pressure until subsoil support is zero;
28. When equilibrium is reached the recording of data can be stopped;
29. Removing pressure from top water cushion;
30. Opening metal box;
31. Removing top cushion;
32. Making xy-scan of surface;
33. Removing top box;
34. Removing embankment fill;

21 February 2011
35. Removing cables, transducers,

pressure cells and steel frame(s);

21 February 2011
Appendix D
Journal scale test K2

The journal of scale test K2 is presented below.
Journal
Huesker 2
date time action observations, remarks
2009-11-12 Thursday
attach reinforcement Attach two layers of Fortrac 110-25 on one frame, with the
to frame strength directions perpendicular on each other.
Monday
start filling foam
2009-11-16 cushion
Tuesday
2009-11-17 10:00 foam cushion filled
11:13 install strain cables
11:13 start journal
install sand layer and
frame
weight box below in
kg 714,6
zero values strain 1 = 35.79, 2 = 35.62, 3 = 33.45, 4 = 35.85, 5 = 34.90, 6 = 35.29
13:48 cables mm
13:59 zero strain cables
zero height
13:59 transducers (settlement transducers)
14:03 start registration Huesker_110_25
Total pressure cells (TPC's, above piles), force transducers and
14:06 zero transducers pressure foam cushion
weight box including
14:42 top box 863,2
measure points on
14:44 top of geosynthetic 70.1 cm
70.1 cm
70.1 cm
install 42 cm granular in 5 layers with compacting each layer
weight with granular

16:53 layer in kg 1719
17:11 XY scan
open tap foam
17:41 cushion drain 6 liters: from 6.5 to 12.5 l

Appendix D Journal scale test K2
close tap foam

18:11 cushion
18:11 XY-scan is ready
Wednesday
install water cushion
for top load and zero
2009-11-18 9:40 top load
10:02 Start top load: 25 kPa
open tap foam
12:00 cushion drain 6 liters: from 12.6 to 18.6
close tap foam
12:20 cushion drained totally:: 18.6 l
open tap foam
Water tray of the

settlement
transducers has
13:55 moved.
close tap foam
14:20 cushion
Fill up burette and
14:50 empty barrel
15:41 increase top load 50 kPa
open tap foam
17:45 cushion drain 6 liters: from 0.1 to 6.1 l
close tap foam
18:03 cushion
Thursday
open tap foam
2009-11-19 8:40 cushion drain 6 liters: from 6.1 to 12.1 l
close tap foam
9:06 cushion

open tap foam
12:10 cushion drain 6 liters from 12.1 to 18.1 l
close tap foam
12:29 cushion pressure foam cushion = 6.5 kPa, thus another drainage step
14:07 fill up burette
open tap foam
close tap foam
14:39 cushion
empty barrel

open tap foam
17:23 cushion drain 6 liters from 0.1 to 6.1 l
close tap foam
17:42 cushion
Friday
open tap foam
2009-11-20 9:27 cushion 6 l liter draining from 6.11 to 12.1
close tap foam
10:03 cushion
crash measurement
10:15 program start again
12:22 suck vacuum drain 12 liters with vacuum
close tap foam
12:41 cushion

21 February 2011
12.42 remove top load

15:00 dismantle
15:25 stop registration
install XY-scan, but
frame is turned 90 connectors watch in the direction of the stairs, measured with
Monday 15:28 deg. the same outline-start-positions as in the other tests.
2009-11-23 14:58 start registration Huesker_110_25_afbouwen.gef
end values strain

cables vernier gauge (Dutch: schuifmaat) [mm] and from logger [mm]
1 36.57 ; 2.496
2 35.99 ; 0.7523
3 33.97 ; 1.012
4 35.78 ; 0.5298
5 34.92 ; -0.1755
6 35.03 ; 0.071
15:11 stop registration

During lifting top box, the rubber sleeve remains around the
granular, and is not lifted with the top box. This happened for
dismantle: the first time.

21 February 2011
Appendix E
Triaxial test
To determine the strength and stiffness parameters of the granular material triaxial tests
have been performed. First the equations used for the calculations are described. Followed
by the results plotted in graphs.
E.1 Calculations
To determine the parameters of the triaxial tests the following equations have been used.
The equations are based on the British Standard 1377 : Part 8 (1990).
Axial strain
h
1  (E.1)
l int
Where:
1 - Axial strain
h mm Change in length between the rings
l int mm Initial length between the rings
Volumetric strain
 v  1  2   3 (E.2)
With:
r
3  (E.3)
rint
Where:
1 - Axial strain, formula (A.1)
Area of cross section of the specimen normal to its axis

As    r 2 (E.4)
With:
r  rint  r (E.5)

Appendix E Triaxial test
Where:
As mm2 Area of cross section
2
Axial stress or deviatoric stress

P
q  1   3   10 6 (E.6)
As
Where:
q kPa Deviatoric stress
P kN Axial force
As mm2 Area of cross section, formula (E.4)
Principal stress
 1   1   3    3 (E.7)
Where:
1 kPa Principal stress
1   3 kPa Deviatoric stress
3 kPa Confining pressure
Young’s modulus
The Young’s modulus can be calculated for each step of the triaxiaal test. In the report two
values are calculated, E 0 and E 50 . In a graph the Young’s modulus is presented for each
step.
1   3
E (E.8)
1
Where:
E kPa Young’s modulus
E0 kPa Young’s modulus at axial strain by 0% of deviatoric stress at failure
E 50 kPa Young’s modulus at axial strain by 50% of deviatoric stress at failure
1   3 kPa Deviatoric stress at failure
1 kPa Axial strain
Moisture content
m  mdr
w  100% (E.9)
mdr
Where:
w % Moisture content
m kPa Weight of sample
mdr kPa Dry weight of sample

21 February 2011
Stress path parameter

1
s  1   3  (E.10)
2
1
t   1   3  (E.11)
2
Where:
s kPa Stress path parameter
t kPa Stress path parameter
1   3 kPa Deviatoric stress
1 kPa Principal stress
3 kPa Confining pressure
Shear strength parameters

sin    tan   (E.12)
t0
c (E.13)
cos  
Where:
 ° Angle of internal friction
c kPa Cohesion
 ° Help variable determent from the stress path parameter plot
t0 kPa Help variable determent from the stress path parameter plot.
The help variables are determent from the stress path parameter plot. A line can be drawn
through the stress paths at equal axial strain. The parameter t0 is determined where this
line intersects the t axis. The inclination of this line to the horizontal is θ.
Poisson ratio
The Poisson ratio can be calculated when there is a volume decrease. Two values that are
calculated are  0 and  50 . In a graph the poisson ratio is presented for each step.
v
 1  2  (E.14)
1
Where:
1 % Change in axial strain
v % Change in volumetric strain
0 Poisson ratio
0 Poisson ratio at axial strain by 0% of deviatoric stress at failure
 50 Poisson ratio at axial strain by 50% of deviatoric stress at failure

Dilatancy angle
The dilatancy angle can be calculated when there is a volume increase. The results are
shown in a graph below.
 v 2  sin  
 (E.15)
 1 1  sin  
Where:
1 % Change in axial strain
v % Change in volumetric strain
E.2 Results
Different graphs with the results can be found in this paragraph.
Displacement in time
The displacement of the LVDT’s is analysed before the parameters are calculated. The
graphs of the displacement LVDT’s are shown in this paragraph.
From the graphs for triaxial test 16kPa can be concluded that one side of the sample had a
smaller axial and radial settlement than the rest of the sample. The difference is very small,
therefore the average of the settlement is used to calculate the strains.
Three LVDT’s of the triaxial test with 55kPa vacuum pressure have a divergence. These
sensors are axial 8, radial top 6 and radial bottom 12. The sensors are not located near or
above each other. The divergence is about 4 mm, which results in an offset from the trend.
Because of this offset these lines are not used to calculate the parameters.
Triaxial test 90kPa also shows offsets in the results, caused by axial 9 and radial top 5. The
curve of radial top 5 makes a jump when the engine of the hydraulic actuator was restarted.
The data of these two displacement sensors were also not taken into account for the
calculation of the parameters.

21 February 2011
Figure E.1: Axial displacement - Time curve for Sample 16kPa
Figure E.2: Radial top displacement - Time curve for Sample 16kPa
Figure E.3: Radial bottom displacement - Time curve for Sample 16kPa


21 February 2011

Stress distribution in time

To see if there are some abnormalities during the triaxial test, the stresses are plotted
against time.
The results of sample 16kPa are shown in Figure E.10. From this graph can be seen that
the vacuum pressure σ3 is not constant in time. This is caused by little holes in the
membrane causing a decrease in vacuum pressure. The vacuum pressure is decreasing,
when the sample is failing.
The results of sample 55kPa are shown in Figure E.11. From this graph can be concluded
that the vacuum pressure staid constant during the loading.
The vacuum pressure of sample 90kPa remained constant during loading. There were
some problems with the engine of the hydraulic actuator during the test. As a result of
overheating, the engine had a power shut down, whereby the force on the sample
decreased to zero. When the engine was restarted, the force increased again. Due to this
problem Figure E.12 shows unloading and reloading.
Figure E.10: Stress – Time curve for Sample 16kPa

21 February 2011

Deviatoric stress and volumetric strain

The detailed graphs of the deviatoric stress and volumetric strain per Triaxial test are
presented in this paragraph.
Figure E.13: Deviatoric stress – Axial strain curve for Sample 16kPa
Figure E.14: Volumetric strain – Axial strain curve for Sample 16 kPa

21 February 2011


21 February 2011
Stiffness parameters
Young’s modulus and poisson ratio are the stiffness parameters. These parameters are
calculated with equations (E.8) and (E.14) and plotted against axial strain. The plots of the
young’s modulus and Poisson ratio are presented below.
Figure E.19: Young’s modulus – Axial strain curve
Figure E.20: Poisson ratio – Axial strain curve

Strength parameters
Internal friction angle and dilatancy angle are the strength parameters. These parameters
are calculated with the equations (E.13) and (E.15) and plotted against axial strain. These
plots are presented below.
Figure E.21: Cohesion and internal friction angle – Axial strain curve
Figure E.22: Dilatancy angle – Axial strain curve

21 February 2011
Appendix F
Relative density
The loose and compacted bulk density of the granular material is determined to calculate
the relative density.
F.1 Loose bulk density

The loose bulk density of the granular material is determined with the European standard
EN 1097-3.
The loose bulk density is calculated with the following equation.

m  m1
d ;loose  2 (F.1)
V
Where:
d ;loose kg/m
3
Loose bulk density
m1 kg Mass of the empty container
m2 kg Mass of the container and test specimen
3
V m Capacity of the container
The results of three tests are presented below in Table F.1.
Test nr. m1 ρloose

[kg] [kg/m3]
1 10.593 1266
2 10.702 1299
3 10.644 1281 V [m3] 0,00336
Average 1282 m2 [kg] 6.339
Table F.1: Loose bulk density
The average minimum dry density is 1282 kg/m3.
F.2 Compacted bulk density

The maximum dry density of the granular material is determined with the Proctor test
described in the European Standard EN 13286-2.
The compacted bulk density is calculated with the following equation.

m  m1
compacted  2 (F.2)
V
Where:
compacted kg/m
3
Compacted bulk density
m1 kg Mass of the empty container
m2 kg Mass of the container and test specimen

Appendix F Relative density
3
V m Capacity of the container
The compacted dry density is calculated with the following equation.

100 compacted
d ;compacted  (F.3)
100  w
Where:
d ;compacted kg/m
3
Compacted dry density
compacted kg/m
3
w % Water content of the mixture
The results of the proctor tests are presented in Table F.2.

+ wet sample
+ dry sample
Mass basket
Mass basket
Mass empty
ρd;compacted
basket [g]
ρcompacted
Test nr.
[kg/m3]
[kg/m3]
[kg]
[%]
[g]
[g]
m2
w
1 94.5 801.5 775.1 3.9 8.869 1737 1672

2 93.7 750.2 703.1 7.7 9.340 1937 1798
3 94.4 803.1 733.5 10.9 9.821 2141 1930
4 93.9 752.1 679.9 12.3 9.844 2150 1915
5 94.2 1100.9 976.4 14.1 9.884 2167 1899
6 94.0 1178.3 1037.8 14.9 9.822 2141 1864
Max 12.3 1930
V [m3] 0.00236
m1 [kg] 4.769
Table F.2: Compacted dry density
The results of the dry density and moisture content are plotted in Figure F.1. The maximum
dry density is determined from this figure. The maximum dry density is 1930 kg/m3.

21 February 2011
Optimum Moisture Content vs.

Maximum Dry Density
1950
1900
Dry Density [kg/m3]
1850
1800
1750
1700
1650
0 5 10 15 20
Moisture content [% ]
Figure F.1: Graph dry density against moisture content
F.3 Relative density

The relative density can be determined with the loose and compacted dry density. The
relative density is based on the porosity and calculated with the following equation.
n n d  d ;min
Dr  max  100   100 (F.4)
nmax  nmin d ;max  d ;min
Where:
Dr % Relative density
nmax - Maximum porosity
nmin - Minimum porosity
n - Porosity
d ;max kg/m
3
d ;min kg/m
3
Loose bulk density
d kg/m 3
Density
The results of the relative density of the triaxial tests and the scale tests are presented in
Table F.3. The relative density of the triaxial tests lies between 59.7% and 64.4%. The
values are in the same range and dense compacted.
Test ρd Dd
[kg/m3] [%]
Triaxial test 1 1681 61.6
Scale test 1 1695 63.7 ρd;min [kg/m3] 1282
Scale test 2 1699 64.4 ρd;max [kg/m3] 1930
Table F.3: Relative density

21 February 2011
Appendix G
Material models
This appendix describes from what sources and/or from which relationships the soil
properties for the granular fill are determined.
G.1 Mohr Coulomb model

The elastic-plastic Mohr Coulomb model represents a ‘first-order’ approximation of soil
behaviour. This model is used for a first analysis of the problem. For each layer, a constant
average stiffness is estimated. Due to this constant stiffness, the calculation time is
relatively short and a first impression of deformations is obtained. The Mohr Coulomb
model properties, their sources and relationships are presented in Table G.1.

Appendix G Material models
Properties Source
General properties
Material type
Material types are: drained, undrained or non- - -
porous.
The volumetric
 unsat Unsaturated volumetric weight kN/m3 weight used in
triaxial test
 sat Saturated volumetric weight kN/m3 -
k x , ky , kz Permeability m/day -
Strength properties
Cohesion
Cohesion of granular soils and sand is zero. To
c' kN/m2 c 1
avoid numerical instability problems, the cohesion
is set to a small value.
' Internal friction angle  Triaxial test result
Dilatancy angle Triaxial test result
' The dilatancy angle can be determined with triaxial  or     30  0
test results or with the following relationship.
Young’s modulus
E' For loading of soils it is realistic to use the secant kN/m2 Triaxial test result
modulus E50
' Poisson’s ratio - Triaxial test result
Advanced properties
E 'increment
Increase of stiffness with depth kN/m2/m Eincrement  0
y ref Reference level for Eincrement m y ref  0

c 'increment Increase cohesion with depth kN/ m2/m cincrement  0
y ref Reference level for cincrement m y ref  0
 tension Tensile strength kN/m2  tension  0
Table G.1: Mohr Coulomb model properties
G.2 Hardening Soil model

The Hardening Soil model is an isotropic soil model, which takes stress dependent stiffness
into account. This model can be used for simulating stiff soils, but is also suitable for soft
soils. The model distinguishes two main types of hardening, namely shear hardening and
compression hardening. Shear hardening is used to model irreversible strains due to
primary deviatoric loading. Compression hardening is used to model irreversible plastic
strains due to primary compression in oedometer loading and isotropic loading.
The Hardening Soil model properties are presented in Table G.2. The default settings and
sources or relationships of these properties are also presented in this table.

21 February 2011
Properties Relationship
General properties
Material type:
Material types are: drained, undrained, non- - -
porous.
 unsat Unsaturated volumetric weight kN/m3 -
 sat Saturated volumetric weight kN/m 3
-
k x , ky , kz Permeability m/day -
Strength properties
c 'ref Cohesion
Cohesion of granular soils or sand are zero. To cref  1
kN/m2
avoid numerical instability problems, the cohesion
is set to a small value.
' Internal friction angle  Triaxial test result
Triaxial test result
' Dilatancy angle  or     30  0
Secant stiffness for primary loading in standard
E '50 kN/m2
ref
Triaxial test result
triaxial test
Tangent stiffness for primary loading in oedometer  E50ref
E 'oed kN/m2
ref ref
Eoed
test
Stress dependent stiffness according to a power
m - m  0.5
law for sand.
The stress dependent stiffness can be calculated
m

(1)
ln E50 ( 2)
E50 
when triaxial tests have been performed for -
ln  
m
different confining pressure
(1)
3  3( 2 )
Advanced properties
E 'urref Unloading / reloading stiffness kN/m2 Eurref  3E50ref
 'ur Poisson’s ratio for elastic unloading/reloading - ur  0.2
p ref
Reference stress for stiffness kN/m2 p ref  100
K 0nc K0-value for normal consolidation - K 0nc  1  sin 
c 'increment Increase cohesion with depth kN/ m2/m cincrement  0
y ref Reference level for cincrement m y ref  0
Rf Failure ratio - Rf  0 . 9
 tension Tensile strength kN/m 2
 tension  0
Table G.2: Hardening Soil model properties

Appendix G Material models
G.3 Plaxis results of triaxial test

The triaxial tests performed on the granular material are modeled in Plaxis 2D. The model
is axisymmetric. The parameters of the models are described in paragraph 7.2. The density
and relative density of the material is shown in Table G.3. The calculation of the relative
density is described in paragraph F.3. The deviatoric stress and volumetric strain are
plotted against the axial strain in Figure G.1 and Figure G.2.
Test Density Relative density

[kg/m3] [%]
Triaxial test at 55 kPa confining pressure 1681 61.6
Triaxial test at 90 kPa confining pressure 1669 59.7
Scale test K2 1695 63.7
Table G.3: Density and relative density of the triaxial tests and scale test

21 February 2011
600
500
Deviatoric stress [kPa]
400
300
Triaxial test 55kPa

200 HS model 55kPa
MC model 55kPa
Triaxial test 90kPa
100 HS model 90kPa
MC model 90kPa
0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05
Axial strain [-]
Figure G.1: Deviatoric stress strain curve
0.025
0.02
Volumetric strain [-]
0.015
0.01
Triaxial test 55kPa

0.005 HS model 55kPa
MC model 55kPa
Triaxial test 90kPa
0
HS model 90kPa
MC model 90kPa
-0.005
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05
Axial strain [-]
Figure G.2: Volume change curve

21 February 2011
Appendix H
Numerical simulation of soil-structure

interface behaviour
In the model tests, friction occurred between the box wall and the fill material. Within the
Plaxis 3D Foundation, it is not possible to model this friction with interface elements. This
appendix describes and compares options for simulation the friction behaviour, leading to a
choice how to simulate the soil-structure interface behaviour.
H.1 Geometry & parameters of soil-structure interface

To analyse the difference between the results of a model with an interface and a model
with an extra segment of soil on one side the following geometries are used:
1. Reference model: Geometry without friction;
2. Interface model: Geometry with friction on the left side by an interface;
3. Soil segment model: Geometry with friction on the left side by an extra segment of
soil with reduced strength properties;
The geometry of the model without friction is based on the scale test and described in
Chapter 7. The 2D geometry is shown in Figure H.1 and the 3D geometry in Figure H.2.
Figure H.1: 2D intersection Plaxis and dimensions
An interface can only be applied perpendicular to the planes, therefore the interface only is
applied on the left side. The resistance between soil and structure depends on the surface
material of the structure and the type of soil. In the tests, the friction is measured to lie
between 10% and 20%. Therefore, the interface strength parameter Rinter is chosen to be
0.2. The interface strength parameter will have effect on the cohesion, the internal friction
angle and the dilatancy angle. The reduction of these parameters is described in equations
(H.1), (H.2) and (H.3). The 3D geometry of the model with friction on the left side by an
interface is shown in Figure H.3.

Appendix H Numerical simulation of soil-structure interface behaviour
ci  Rinter csoil (H.1)

tan i  Rinter tan soil  tan soil (H.2)
 i  0 for Rinter  1 (H.3)
Where:
Rinter - Interface strength parameter = 0.2 for the calculations with interface elements
ci kN/m2 Cohesion of the interface
2
csoil kN/m Cohesion of the soil
i ° Internal friction angle of the interface
soil ° Internal friction angle of the soil
i Dilatancy angle of the interface
 inter Dilatancy angle of the soil
The friction also can be modelled as an extra segment of soil with lower strength
parameters. With this model the extra segment of soil also can be modelled at the backside
of the model. To get corresponding results the parameters of the extra soil segment need
to be the same as the parameters calculated with the interface. Therefore, the relations in
equations (H.1), (H.2) and (H.3) are used to calculate the lower strength parameters of the
extra soil segment based on a friction of Rinter=0.2. The parameters are shown in Table H.1.
The 3D geometry of the model with an extra segment of soil on the left side can be seen in
Figure H.4.
Interface
Interface
Strength properties granular
sand
material
cref Cohesion kN/m2 1 1
 Internal friction angle  7.27 12.11
 Dilatancy angle  0 0
Table H.1: Material properties for the extra soil segment in the Soil segment model
Figure H.2: Reference model: cross section (left) and 3D Plaxis geometry (right)

21 February 2011
Figure H.3: Interface model: cross section (left) and 3D Plaxis geometry (right)
Figure H.4: Soil segment model: cross section (left) and 3D Plaxis geometry (right)
H.2 Results Plaxis of soil-structure interface behaviour

To validate the models with friction the results of the Plaxis calculations are presented in
this paragraph. First the vertical stresses on the pile and the tensile forces are considered.
Then the load distribution and settlements are presented.
Vertical stresses on the pile

The vertical effective stresses in the soil are presented in cross-sections through the centre
of the pile during the second drainage phase and vacuum phase, with a top load op
100kPa. The Interface model is compared to the Soil segment model, to validate the Soil
segment model. The model with an interface is shown in Figure H.5 and Figure H.7 and the
model with an extra soil segment on the left side in Figure H.6 and Figure H.8.
Comparison of the two figures of the second drainage phase (Figure H.5 and Figure H.6)
show that the Soil segment model gives more disturbance on the left side than the Interface
model and more peak stresses just above the frame.
In the vacuum phase the displacement of the extra soil body next to the frame has to be
prescribed. This has effect on the vertical stress distribution in the granular material. An
arch will arise between the frame and the pile as seen in Figure H.7. Therefore the extra
soil body is not applicable in the vacuum phase.

Figure H.5: Vertical effective stresses in the Figure H.6: Vertical effective stresses in the
granular fill of the second drainage phase, granular fill of the second drainage phase,
calculation results of the Interface model calculation results of the Soil segment
model
Figure H.7: Vertical effective stresses in the Figure H.8: Vertical effective stresses in the
granular fill of the vacuum phase, granular fill of the vacuum phase,
calculation results of the Interface model calculation results of the Soil segment
model
Tensile force in geosynthetic reinforcement

The axial force N1 is the force in x-direction. The axial force N1 in the geosynthetic
reinforcement of the model with an interface is shown in
Figure H.9 and of the model with an extra soil body in
Figure H.10.
The figures are generally the same. Only the axial forces next to the pile of the model with
an extra soil body are a bit higher than the axial forces of the model with an interface. This
is probably caused by the higher stresses as found in Figure A.7.

21 February 2011
x x
Figure H.9: Tensile force N1 within the GR in Figure H.10: Tensile force N1 within the GR
x direction in the Interface model in x direction in the Soil segment model
Load distribution
In Plaxis the load distribution cannot be measured directly. The average vertical stress on
the load cell is determined of the stress points in the area of the load cell. The area where
the average vertical stress is determined is shown in Figure H.11. The loads are calculated
by multiplying the average vertical stress with the area of the load cell.
Load part A
Load parts A+B
Figure H.11: Load distribution
The load distribution for the Plaxis models and the results of the scale test are plotted in
graphs on the next page. As expected, the A and B determined with the Plaxis models with
friction are 10% lower than the results of the Plaxis model without friction. In general, the
load distribution is the same for each Plaxis model. Only during the vacuum phase, the
loads A and B are different. During the vacuum phase load part A decreases significant for
all Plaxis models, although not with the same amount. For the Plaxis model without and
with an interface, load part A+B decreases and for the Plaxis model with an extra soil
segment load part A+B increases.
Load parts A and A+B are overestimated with the Plaxis models compared to the scale
test. This causes an underestimation of load B with the Plaxis models.

Scale test K2
Plaxis without friction
Plaxis with interface R=0.2
Plaxis with friction on one side
40 7
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
35 6
30

25
4
20
3
15
2
10
5 1
0 0
0 5 10 15 20 25 0 5 10 15 20 25
Figure H.12: Load part A+B Figure H.13: Load part A+B
20
25
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
18
16
20
14
15 12
10
10 8
5 4
0 0
0 5 10 15 20 25 0 5 10 15 20 25
Figure H.14: Load part A Figure H.15: Load part B

21 February 2011
Displacement
The settlements of the geosynthetic reinforcement can be determined from the output data
of Plaxis. The calculated settlements of the GR are considered for the locations of the
measurement equipment during the scale test, as indicated in Figure A.11.
Frame
0.225 m
Displacement z3
0.10 m
Displacement z2
pile 3 pile 1
0.45 m
Displacement z1
0.10 m
pile 2 pile 4
0.225 m
0.225 m 0.45 m 0.225 m
0.10 m 0.10 m
Figure H.16: Location of displacement measurements of scale test K2
The friction between 10% and 20% in the Interface model and Soil segment model has no
effect on the settlements z1, z2 and z3 of the GR as shown in Figure H.17, Figure H.18 and
Figure H.19, except for the vacuum phase. During this phase the settlements differ for each
Plaxis model. The largest deformation is found in the model with an extra soil segment for
the friction.

Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
60
50
40
Scale test K2
30 Plaxis without friction
20
10
0
0 5 10 15 20 25
Net Load [kN/pile]
Figure H.17: Displacement z1 (Centre 4

piles)
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Top load
Top load
Top load
Top load
Vacuum
Vacuum
60 60
50 50
40 40
30 30
20 20
10 10
0 0
0 5 10 15 20 25 0 5 10 15 20 25
Figure H.18: Displacement z2 (Centre 4 Figure H.19: Displacement z3 (Close to

piles) pile)

21 February 2011
Friction
The amount of friction is calculated by subtracting the top load and the weight of the soil
from the load parts A+B+C. The amount of friction during the test is shown in Figure H.20.
As expected, the calculated amount of friction increases when an interface or extra soil
segment is applied in the model. The amount of friction during the phases are the same for
each Plaxis model, except for the vacuum phase. For the model with an interface the
friction increases and for the model with an extra soil segment the friction decreases.
The amount of friction of the Plaxis models is too low compared to the friction during the
scale test. This will change when an extra soil segment is applied at the backside of the
Plaxis model. The extra soil segment will bring the amount of friction of the Plaxis models
closer to the friction of the scale test.
During the vacuum phase the results of the calculated friction increases in the Interface
model and decreases in the Soil segment model. When the deformation becomes large,
the soil will yield in the extra soil segment of the Soil segment model. The relation for
yielding soil is described in equation (H.4).
 sin  cos
 (H.4)
 n 1  sin  sin
Where:
 kN/m2 Shear stress
n kN/m
2
Normal stress
A dilatancy angle of zero has the following result.
  0  sin i  tan i  Rinter tan soil  i  12.38 (H.5)

15
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
10
Friction [kN/pile]
Scale test K2
5
0 5 10 15 20 25
Net Load [kN/pile]
Figure H.20: Friction
H.3 Validation soil-structure interface model

To validate if the extra soil body model reduces the load on the pile with 10% to 20% three
models are compared in Figure H.21 to Figure H.28:
 Scale test K2
 Reference model: Geometry without friction;
 Soil segment model: Geometry with friction on the left side by an extra soil segment
of soil with reduced strength properties;
 Total soil segment model: Geometry with friction on the left and back side by an
extra soil segment of soil with reduced strength properties.
In Figure H.21 is shown that the load on the pile (load part A+B) of the model with an extra
soil body on both sides is reduced to the load measured in the scale test. From this can be
concluded that the model with an extra soil body, models the interface behaviour between
the soil and the wall.
In the calculation of the extra soil body an internal friction angle of 12.11 degrees is used.
The CUR 166 (2005) describes the wall friction angle. The wall friction angle for coarse
sand is   13  '  47 3  15.66 . This value is comparable to the value calculated with an
interface strength parameter.

21 February 2011
Scale test K2
Plaxis with friction on both sides
40 7
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
35 6
30

25
4
20
3
15
2
10
5 1
0 0
0 5 10 15 20 25 0 5 10 15 20 25
Figure H.21: Load part A+B Figure H.22: Load part C
20
25
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
18
16
20
14
15 12
10
10 8
5 4
0 0
0 5 10 15 20 25 0 5 10 15 20 25
Figure H.23: Load part A Figure H.24: Load part B

Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
60
50
40
Scale test K2
30 Plaxis without friction
20
10
0
0 5 10 15 20 25
Net Load [kN/pile]
Figure H.25: Displacement z1 (Centre 4

piles)
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Top load
Top load
Top load
Top load
Vacuum
Vacuum
60 60
50 50
40 40
30 30
20 20
10 10
0 0
0 5 10 15 20 25 0 5 10 15 20 25
Figure H.26: Displacement z2 (Centre 4 Figure H.27: Displacement z3 (Close to

piles) pile)

21 February 2011
15
Drainage
Drainage
Drainage
Drainage
Drainage
Top load
Top load
Top load
Top load
Vacuum
10
Friction [kN/pile]
Scale test K2
5
0 5 10 15 20 25
Net Load [kN/pile]
Figure H.28: Friction
H.4 Conclusion of soil-structure interface behaviour

This appendix compared 3 Plaxis models with a different simulation of the soil-structure
interface behaviour and the measurements. For most of the test phases, the calculated
load distribution, displacements and friction are comparable for the Soil segment model and
the Interface model. For the vacuum phase the results are different. The friction applied in
the Plaxis models decreases the load parts A and A+B. This results in a better
approximation of the results of the scale test. However, when the extra soil segment is
applied with the Hardening Soil model the calculation time increases significantly. And in
the vacuum phase there arises an extra arch between the frame and pile, because the
displacement of the soil between the frame and wall is prescribed. Therefore is concluded
that the extra soil segment is not suitable to apply in the Plaxis model. To reduce the load
in the system, the top load is reduced by the amount of friction. The disturbance of the
friction on the load distribution in the embankment is neglected in the model.

21 February 2011
Appendix I
Results of Plaxis Hardening Soil model

The results of the Plaxis Hardening Soil model are presented in this Appendix. The model
does not consider friction.
I.1 Deformed mesh

The minimum deformation can be found at the edge of the geometry and the maximum
deformation at the middle of the geometry. Therefore the deformed mesh is presented for
cross section A-A.
Figure I.1: Deformed mesh A-A with HS Figure I.2: Deformed mesh A-A with HS
model of phase Drainage 1 model of phase Drainage 25.2

Appendix I Results of Plaxis Hardening Soil model
model of phase Drainage 50.2 model of phase Drainage 75.2
model of phase Drainage 100.2 model of phase Vacuum

21 February 2011
I.2 Vertical displacement

For every phase the Vertical displacement is plotted below.
Figure I.7: Vertical displacement A-A with HS model of phase Drainage 1
Figure I.8: Vertical displacement A-A with HS model of phase Load 25kPa

Figure I.9: Vertical displacement A-A with HS model of phase Drainage 25.1

21 February 2011


21 February 2011


21 February 2011
Figure I.20: Vertical displacement A-A with HS model of phase Vacuum

I.3 Effective vertical stresses

The effective vertical stresses in cross section A-A for each phase are presented in this
paragraph.
Figure I.21: Effective vertical stress A-A with HS model of phase Drainage 1
Figure I.22: Effective vertical stress A-A with HS model of phase Load 25kPa

21 February 2011
Figure I.23: Effective vertical stress A-A with HS model of phase Drainage 25.1


21 February 2011


21 February 2011

Figure I.34: Effective vertical stress A-A with HS model of phase Vacuum

21 February 2011
I.4 Tensile forces in geosynthetic reinforcement

The tensile forces N1 are the forces in x direction and the tensile forces N2 in z direction.
The results of the tensile forces N1 and N2 are perpendicular, therefore only the results of
N1 are given.
Figure I.35: Tensile forces N1 of phase Drainage 1
Figure I.36: Tensile forces N1 of phase Load 25kPa

Figure I.37: Tensile forces N1 of phase Drainage 25.1

21 February 2011
Figure I.40: Tensile forces N1 model of phase Drainage 50.1


21 February 2011


21 February 2011
Figure I.48: Tensile forces N1 of phase Vacuum

I.5 Deformation of geosynthetic reinforcement

The maximum deformation of the geosynthetic reinforcement is reached in the vacuum
phase. In this phase there is no subsoil support. The 3D deformed GR is illustrated in
Figure I.49.
Figure I.49: Deformation of geosynthetic reinforcement of phase Vacuum

Piled Embankments With Geosynthetic Reinforcement: Numerical Analysis of Scale Model Tests

Uploaded by

Copyright:

Available Formats

Piled Embankments With Geosynthetic Reinforcement: Numerical Analysis of Scale Model Tests

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Piled Embankments With Geosynthetic Reinforcement: Numerical Analysis of Scale Model Tests

Uploaded by

Copyright:

Available Formats

Piled Embankments with Geosynthetic

Author: T.J.M. den Boogert

Date: 21 February 2011

Ing. H.J. Everts

Ir. S.J.M. van Eekelen

Ing. P.G. van Duijnen

Piled embankment with geosynthetic reinforcement – Master of Science thesis I

Piled embankment with geosynthetic reinforcement is a soil improvement technique. This

Theresa den Boogert

Piled embankment with geosynthetic reinforcement – Master of Science thesis III

The design of reinforced piled embankments involve complex soil-structure interaction.

Piled embankment with geosynthetic reinforcement – Master of Science thesis V

VI Piled embankment with geosynthetic reinforcement – Master of Science thesis

Piled embankment with geosynthetic reinforcement – Master of Science thesis VII

5 SCALE TEST ........................................................................................................................ 29

VIII Piled embankment with geosynthetic reinforcement – Master of Science thesis

8 VALIDATION AND ANALYSIS OF NUMERICAL MODEL ................................................. 73

APPENDIX A CALCULATION STIFNESS FOAM CUSHION

Piled embankment with geosynthetic reinforcement – Master of Science thesis IX

Piled embankment with geosynthetic reinforcement – Master of Science thesis XI

Figure 5.12: Test set-up ..................................................................................................... 34

XII Piled embankment with geosynthetic reinforcement – Master of Science thesis

Piled embankment with geosynthetic reinforcement – Master of Science thesis XIII

XIV Piled embankment with geosynthetic reinforcement – Master of Science thesis

Piled embankment with geosynthetic reinforcement – Master of Science thesis XV

A kN/pile Measured load part transferred to the pile by arching

B kN/pile Measured load part transferred to the pile by the GR

Piled embankment with geosynthetic reinforcement – Master of Science thesis XVII

Symbol Dimension Description

XVIII Piled embankment with geosynthetic reinforcement – Master of Science thesis

Symbol Dimension Description

Piled embankment with geosynthetic reinforcement – Master of Science thesis XIX

1.1 Background information

In the western part of the Netherlands, the

A piled embankment consists of a field of piles and a geosynthetic reinforcement (GR) in

Piled embankment with geosynthetic reinforcement – Master of Science thesis 1

Figure 1.2: Cross section piled embankment

Figure 1.3: Load distribution in a reinforced piled embankment

2 Piled embankment with geosynthetic reinforcement – Master of Science thesis

1.2 Problem definition and objectives

1.3 Reader’s guide

First the different parts of geosynthetic-reinforced piled embankment are described in

Piled embankment with geosynthetic reinforcement – Master of Science thesis 3

Piled embankment with geosynthetic

Piled embankment with geosynthetic reinforcement – Master of Science thesis 5

Figure 2.1: Augeo piles Figure 2.2: Timber piles

6 Piled embankment with geosynthetic reinforcement – Master of Science thesis

After piling and before placing geosynthetics,

2.3 Geosynthetic reinforcement

The function of the geosynthetic

The strength of geosynthetic

Piled embankment with geosynthetic reinforcement – Master of Science thesis 7

Figure 2.7: Geogrids Figure 2.8: Non-woven and woven geotextiles

Another phenomenon is interaction between grid

8 Piled embankment with geosynthetic reinforcement – Master of Science thesis

2.6 Load distribution

2.6.1 Arching in the embankment

The load underneath the arch (load part

Underneath the geosynthetic