Parameter Plaxis PDF

§
MODELLING OF A DEEP EXCAVATION IN SOFT CLAY

A COMPARISON OF DIFFERENT CALCULATION METHODS TO IN-SITU
MEASUREMENTS
Master of Science Thesis in the Master’s Programme Infrastructure and Environmental Engineering
EMIL JOHANSSON
EMIL SANDEMAN
Department of Civil and Environmental Engineering

Division of GeoEngineering
CHALMERS UNIVERSITY OF TECHNOLOGY
Gothenburg, Sweden, 2014
Master’s Thesis 2014:22
MASTER’S THESIS 2014:22
MODELLING OF A DEEP EXCAVATION IN SOFT CLAY

A COMPARISON OF DIFFERENT CALCULATION METHODS TO IN-SITU
MEASUREMENTS
Master of Science Thesis in the Master’s Programme Infrastructure and

Environmental Engineering
EMIL JOHANSSON
EMIL SANDEMAN

CHALMERS UNIVERSITY OF TECHNOLOGY
Gothenburg, Sweden, 2014
Modelling of a deep excavation in soft clay
A comparison of different calculation methods to in-situ measurement
EMIL JOHANSSON
EMIL SANDEMAN
© EMIL JOHANSSON, EMIL SANDEMAN, 2014.
Examensarbete / Institutionen för bygg- och miljöteknik,

Chalmers tekniska högskola 2014:22

Chalmers University of Technology
SE-412 96 Göteborg
Sweden
Telephone: + 46 (0)31-772 1000
Sweco Civil AB
Rosenlundsgatan 4
Box 1094
SE-405 23 Göteborg
Sweden
Telephone +46 (0)31-627 500
Cover:
Deformed mesh │u│ (scaled up 50.0 times)
Obtained from the Hardening Soil calculation in PLAXIS
Chalmers Reproservice / Institutionen för bygg- och miljöteknik

Göteborg, Sweden 2014
Modelling of a deep excavation in soft clay
A comparison of different calculation methods to in-situ measurements
EMIL JOHANSSON
EMIL SANDEMAN
Chalmers University of Technology
SUMMARY
The largest infrastructure project ever undertaken in Gothenburg, the Västlänken tunnel project, brings
new geotechnical challenges when a 6 km long train tunnel is to be built underneath the city. The
several deep excavations supported by retaining walls needed in the soft Gothenburg clay raises the
question: Which retaining wall design method is the most accurate?
The aim of this thesis is to compare the forces and deformations measured at a multi-anchored sheet
pile wall, installed during the construction of Götatunneln with similar conditions, to the results of
different calculation methods. The calculation methods are hand calculations, the one-dimensional
finite element software Novapoint GS Supported Excavation and the Mohr-Coulomb, Hardening Soil
and Hardening Soil with small strain stiffness constitutive models in the two-dimensional finite
element software PLAXIS 2D.
The different models were created based on previously conducted investigations of the soil, evaluation
of CRS and triaxial tests and blueprints of the sheet pile wall and its anchors. A parametric study was
also performed to see how the different models reacted when varying certain parameters and assessing
the importance of investigating these parameters.
The results show that the PLAXIS 2D models give the best results, with the Hardening Soil model
being the most accurate. Another advantage with the Hardening Soil model is that it provides the
possibility of validating the model by comparing the stress paths obtained from triaxial tests to
simulated tests in PLAXIS SoilTest.
The parametric study showed that if no triaxial tests are available, the Mohr-Coulomb model can give
fairly accurate results as well, using empirical correlations to evaluate the stiffness. It also showed that
basing a Hardening Soil with small strain stiffness model on empirical correlations for small strain
stiffness should not be done, since there are different methods of correlating these parameters which
give different results.
Keywords: Sheet pile wall, Multi-anchored, Deep excavation, Finite element, Novapoint GS
Supported Excavation, PLAXIS 2D, PLAXIS SoilTest, Mohr-Coulomb, Hardening Soil, Hardening
Soil with small strain stiffness, Geotechnics
TABLE OF CONTENTS
1 Introduction ..................................................................................................................................... 1
1.1 Aim .......................................................................................................................................... 1
1.2 Method .................................................................................................................................... 1
2 Literature study................................................................................................................................ 3
2.1 Similar studies ......................................................................................................................... 3
2.2 Earth pressure coefficient ........................................................................................................ 4
2.3 Evaluation of lab tests ............................................................................................................. 5
2.3.1 CRS test ........................................................................................................................... 5
2.3.2 Triaxial test ...................................................................................................................... 6
2.4 Hand Calculations ................................................................................................................... 9
2.5 Novapoint GS Supported Excavation .................................................................................... 11
2.6 PLAXIS ................................................................................................................................. 13
2.6.1 Mohr-Coulomb .............................................................................................................. 14
2.6.2 Hardening Soil and Hardening Soil with small strain stiffness ..................................... 15
3 Models, calculations and results .................................................................................................... 19
3.1 Site information ..................................................................................................................... 19
3.1.1 Soil ................................................................................................................................ 21
3.1.2 PLAXIS SoilTest ........................................................................................................... 22
3.1.3 Pore water pressure ....................................................................................................... 24
3.1.4 Sheet pile wall and anchors ........................................................................................... 24
3.2 Hand calculations .................................................................................................................. 25
3.2.1 Input .............................................................................................................................. 25
3.2.2 Calculation..................................................................................................................... 26
3.2.3 Results ........................................................................................................................... 26
3.3.1 Input .............................................................................................................................. 26
3.3.2 Calculation..................................................................................................................... 27
3.3.3 Results ........................................................................................................................... 27
3.4 PLAXIS – General model setup ............................................................................................ 28
3.5 PLAXIS – Mohr-Coulomb .................................................................................................... 28
3.5.1 Input .............................................................................................................................. 28
3.5.2 Results ........................................................................................................................... 29
3.6 PLAXIS – Hardening soil ..................................................................................................... 30
CHALMERS, Civil and Environmental Engineering, Master’s thesis 2014:22

3.6.1 Input .............................................................................................................................. 30
3.6.2 Results ........................................................................................................................... 30
3.7 PLAXIS – Hardening soil with small strain .......................................................................... 32
3.7.1 Input .............................................................................................................................. 32
3.7.2 Results ........................................................................................................................... 32
4 Analysis of results ......................................................................................................................... 34
4.1 Sheet pile wall deformations ................................................................................................. 34
4.1.1 Discussion ..................................................................................................................... 35
4.2 Anchor forces ........................................................................................................................ 35
4.2.1 Discussion ..................................................................................................................... 36
4.3 Bending moment and shear force .......................................................................................... 36
4.3.1 Discussion ..................................................................................................................... 38
4.4 Earth pressures ...................................................................................................................... 39
4.4.1 Discussion ..................................................................................................................... 40
4.5 Subsidence behind the sheet pile wall ................................................................................... 40
4.5.1 Discussion ..................................................................................................................... 40
5 Parametric studies.......................................................................................................................... 42
5.1 E from empirical evaluation in PLAXIS MC ........................................................................ 42
5.2 φ’cv and c’ in PLAXIS HS ..................................................................................................... 43
5.3 Interface in PLAXIS HS........................................................................................................ 46
5.4 Initial shear modulus G0 in HSs ............................................................................................. 47
5.5 γ0.7 in HSs............................................................................................................................... 48
6 General discussion......................................................................................................................... 50
6.1 Triaxial tests .......................................................................................................................... 50
6.2 Hand calculations .................................................................................................................. 50
6.4 PLAXIS SoilTest and evaluated parameters ......................................................................... 51
6.5 Mohr-Coulomb and Hardening Soil model ........................................................................... 51
6.6 Hardening Soil with small strain stiffness model .................................................................. 52
6.7 Parametric studies.................................................................................................................. 52
6.8 Further investigations ............................................................................................................ 52
7 Conclusions ................................................................................................................................... 53

APPENDIX 1:1-1:9 Soil parameters
APPENDIX 2:1-2:5 PLAXIS SoilTest
APPENDIX 3:1-3:3 Blueprints
APPENDIX 4:1-4:6 Hand calculations
APPENDIX 5:1-5:4 Parametric studies

PREFACE
This report is the result of a Master Thesis project conducted at the department of GeoEngineering at
Chalmers University of Technology during spring 2014.
We would like to thank Anders Kullingsjö for providing us with raw data as well evaluated parameters
obtained during his work on his doctoral thesis. We also want to thank our supervisor at Chalmers,
Minna Karstunen, for all the hints and useful information provided when we got lost in the confusing
jungle of geotechnical FE analyses.
Above all, we would like to thank Jonas Thelander at SWECO for his help with initiation of the work
as well as always being encouraging and supportive throughout the entire process of writing this
thesis. Also thanks for putting up with us babbling in your room.
Gothenburg, May, 2014
Emil Johansson
Emil Sandeman

LIST OF NOTATIONS
ROMAN LETTERS
cinter Shear strength within the interface in PLAXIS [kPa]
csoil Shear strength of the soil [kPa]
cu Undrained shear strength [kPa]
c’ Apparent cohesion [kPa]
d Distance from shaft bottom to a point where below, the passive [m]
earth pressure is equal to or larger than the active earth pressure
m Power for stress-level dependency of stiffness [-]
pa Active earth pressure (GSS) [kPa]
pp Passive earth pressure (GSS) [kPa]
pp Isotropic preconsolidation pressure (PLAXIS) [kPa]
pref Reference pressure (PLAXIS) [kPa]
pv Vertical earth pressure (GSS) [kPa]
p’ Mean effective stress [kPa]
q Deviator stress [kPa]
qd Design load next to shaft for hand calculations [kPa]
r Roughness of the SPW (GSS) [-]
u Pore pressure [kPa]
uda Design pore pressure [kPa]
vf Limit earth strain (GSS) [-]
w Weight of SPW [kN/m2]
D Point where beneath that point the total passive pressure is larger [-]
than or equal to the total active pressure
E Young’s modulus [kPa]
Eu Undrained Young’s modulus [kPa]
E50 Undrained secant stiffness [kPa]
Eur Undrained unload-reload stiffness [kPa]

E’ Effective initial stiffness [kPa]
E’50 Effective secant stiffness [kPa]
E’oed Effective oedometer stiffness [kPa]
E’ur Effective unload-reload stiffness [kPa]
Effective secant stiffness at reference pressure (σ’3) [kPa]
Effective oedometer stiffness at reference pressure (σ’1) [kPa]
Effective unload-reload stiffness at reference pressure (σ’3) [kPa]
EA Normal stiffness [kN/m]
EI Flexural rigidity [kNm2/m]
G Undrained shear modulus [kPa]
G0 Undrained initial shear modulus [kPa]
G0ref Undrained initial shear modulus at reference pressure (σ’3) [kPa]
G’0 Effective initial shear modulus [kPa]
G’0ref Effective initial shear modulus at reference pressure (σ’3) [kPa]
H Height from shaft bottom to top of the sheet pile wall [m]
I Moment of inertia [m4]
IP Plasticity index [-]
K0 Coefficient of lateral earth pressure at rest [-]
K0,nc Coefficient of lateral earth pressure at rest [-]

for normally consolidated soil
KA Coefficient of lateral active earth pressure [-]
KP Coefficient of lateral passive earth pressure [-]
Ky Spring stiffness (GSS) [kPa]
Kyi Initial spring stiffness (GSS) [kPa]
̅y Slope of straight line between initial earth pressure and limit earth [kPa]
pressure (GSS)
M Slope of line in a p’-q plot [-]
M0 Compression modulus, elastic part [kPa]
ML Compression modulus, plastic part [kPa]

MSd Design bending moment [kNm]
Ncb Bearing capacity factor (hand calculations) [-]
P Anchor force [kN]
PA Total active force on the sheet pile wall (hand calculations) [kN]
PH Horizontal anchor force [kN]
Rf Scaling factor (GSS) [-]
Rinter Interface value (PLAXIS) [-]
GREEK LETTERS
α Inclination of anchors [°]
γ Soil weight [kN/m3]
γ0.7 Shear strain level where G = 0.722G0 [-]
γm Soil weight [kN/m3]
γs Shear strain [-]
γSda Design partial factor (hand calculations) [-]
ε1 Axial strain [-]
κ Parameter depending on the roughness of the SPW (GSS) [-]
υ Poisson’s ratio [-]
υu Undrained Poisson’s ratio [-]
υur Poisson’s ratio for unloading-reloading [-]
υ’ Effective Poisson’s ratio [-]
υ’ur Effective Poisson’s ratio for unloading-reloading [-]
φ Friction angle of soil [°]
φd Design friction angle of soil [°]
φinter Friction angle of interface (PLAXIS) [°]
φsoil Friction angle of soil (PLAXIS) [°]
φ’cv Critical state friction angle [°]
σ1 Major principal stress [kPa]
σ3 Minor principal stress [kPa]

σa Active horizontal earth pressure [kPa]
σf Limit earth pressure (GSS) [kPa]
σi Load intensity on SPW (hand calculations) [kN/m]
σpnetto Net earth pressure (hand calculations) [kPa/m]
σv Vertical stress [kPa]
σv,as Vertical earth pressure above shaft [kPa]
σyi Initial earth pressure (GSS) [kPa]
σ’ Effective stress [kPa]
σ’1 Effective major principal stress [kPa]
σ’3 Effective minor principal stress [kPa]
σ’c Preconsolidation pressure [kPa]
σ’r Confining pressure in triaxial test [kPa]
σ’v Effective vertical stress [kPa]
ABBREVIATIONS
CRS Constant Rate of Strain
FE Finite element
GSS Novapoint GS Supported Excavation
HS Hardening Soil
HSs Hardening Soil with small strain stiffness
MC Mohr-Coulomb
OCR Overconsolidation Ratio
PWP Pore Water Pressure
SPW Sheet Pile Wall
TSA Total Stress Automatic

1 INTRODUCTION
The Västlänken tunnel project is the largest infrastructure project undertaken in Gothenburg. The
objective is to construct a 6 km train tunnel underneath the city to increase accessibility for commuters
from outside the city. In order to construct this tunnel, several deep excavations supported by multi-
anchored sheet pile walls (SPW) need to be performed in the soft Gothenburg clay. This raises the
question of which sheet pile wall analysis method give the most accurate results. A comparison of
different methods and deformation measurements performed on a real sheet pile wall is therefore of
interest.
A similar project to Västlänken is Götatunneln, which is a road tunnel constructed in the vicinity
completed in 2006. Part of this project was a multi-anchored SPW temporarily installed to support a
ten meter deep excavation in soft clay. Forces in anchors and deformations in the SPW and in the area
behind it were measured during the excavation and construction of the tunnel. This sheet pile wall
would therefore be suited as a reference object for a comparison of different calculation methods.
1.1 AIM
The aim is to compare the forces and deformations measured at a SPW to the results from a hand
calculation method, the one-dimensional finite element (FE) software Novapoint GS Supported
Excavation (GSS) and three constitutive models in the two-dimensional FE software PLAXIS 2D
(PLAXIS). This is done by modelling a multi-anchored SPW and comparing the results from these
methods to in-situ measurements to assess which analysis method is the most efficient to use in the
design process.
1.2 METHOD
At first, a desk study is performed to obtain the knowledge necessary when designing a SPW,
evaluating soil parameters and using the software GSS and PLAXIS.
Soil parameters are either obtained from field and lab tests evaluated by Kullingsjö (2007, p. 136) or
evaluated from raw data from triaxial and CRS tests conducted at Chalmers University of Technology
during the construction of Götatunneln 2000-2006. No additional tests were performed during the
work with this Master’s project.
One FE model will be created in GSS and one in PLAXIS. Three different constitutive models will be
used in PLAXIS; a Mohr-Coulomb (MC) model, a Hardening Soil (HS) model and a Hardening Soil
with small strain stiffness (HSs) model. The models are based on a temporary SPW used during the
construction of Götatunneln, Gothenburg.
The input parameters used for the different models, the calculation conditions and the results from all
the calculations are presented in section 3. An analysis of the results, where the output from the
different models is compared to measurements performed as site, as well as to each other, is conducted
to determine which method is the most accurate. The following results are compared and analysed in
section 4:
 Horizontal deformations in the SPW

 Anchor forces
 Bending moments and shear forces
 Earth pressures
 Subsidence behind the wall
CHALMERS, Civil and Environmental Engineering, Master’s thesis 2014:22 1

The comparisons are presented with a short interpretation and discussion. The evaluation of the
different calculation methods will focus solely on the accuracy of the methods and the results from
these comparisons are then used to assess which software and model is preferable to use in the design
process.
Parametric studies are also performed for the PLAXIS models to determine how varying certain
parameters affects the results, and if those parameters should be investigated in more detail. In section
6 there is a discussion about the more general issues and discoveries encountered during the work with
this project are discussed.
2 CHALMERS, Civil and Environmental Engineering, Master’s thesis 2014:22

2 LITERATURE STUDY
In this section, studies that are similar to the one performed in this thesis are presented as well as a
description of earth pressures and how the different lab tests are performed and evaluated. A
description of how the different constitutive models function and what input is required is also found
in this section.
2.1 SIMILAR STUDIES

Rankine published a method for analysing earth retaining walls in 1857 in which it is assumed that the
lateral earth pressure is fully mobilized and is increased linearly with depth in a homogenous soil. This
is a good first assumption but is not the case in reality because it requires quite large movements
before full mobilization occurs (Powrie & Simpson, 2001, p. 2).
Since FE analyses have become more and more common due to the rapid development of computer
power in recent years, several studies have been conducted and different constitutive models have
been developed in order to capture more realistic soil behaviour.
An example of such a study is Effects of deep excavations in soft clay on the immediate surrounding
by Kullingsjö (2007) where different methods of how to analyse ground deformation when performing
deep excavations in soft clay are presented together with how the lateral earth pressures can be
estimated. Back calculations are then performed in order to predict and estimate ground deformations
adjacent to a multi anchored SPW.
These back calculations are performed using FE analyses with three different constitutive models,
namely a linear elastic Mohr-Coulomb model, a total stress based e-ADP model and an effective stress
based MIT-S1 model. The results from these different calculations are then compared with in-situ
measurements to determine their accuracy, advantages and disadvantages (Kullingsjö, 2007).
The study concludes that there are no additional benefits gained when using the studied FE methods
since the results were inaccurate. It also concludes that the collaboration between the contractor and
the geotechnical consultancy is important in order to ensure a reliable construction that behaves in an
acceptable manner (Kullingsjö, 2007). The study performed in this thesis is similar to Kullingsjö’s,
using the same raw data and in situ measurements but comparing different constitutive models.
Another similar study was performed by H. F. Schweiger (2009) where calculations with the same
constitutive models in PLAXIS as used in this thesis were performed. Schweiger suggests that the MC
model is not well suited for this kind of analyses and that a more advanced constitutive model is
necessary to obtain reliable results. However, the study focused on comparing the constitutive models
to each other and did not compare the results from these models to real measurements.
In Schweiger and Breymann (2005) the HS model has been compared to in-situ measurements
performed during deep excavations in soft Salzburg clay. Five individual excavations has been
analysed in PLAXIS and compared to the measured deformations and the paper concludes that the
agreement between the HS model and reality is overall good and that this constitutive model gives
quite accurate results when modeling such problems.
Furthermore, in Kempfert and Gebresellasie (2006, p. 169) it is stated that Schweiger (2000) has
compared 14 individual analyses of a 16.8 m deep excavation. The results differ quite a lot between
the analyses and the reason for this is, except for differences in modeling in general, that there is a
difference in identifying parameters, particularly the stiffnesses. Kempfert and Gebresellasie (2006, p.
170) therefore suggest that a standardization of FE analysis of excavations is necessary.

Simpson and Powrie (2001) present, compare and discuss different design approaches for calculating
earth pressures when analysing retaining walls. It highlights some different approaches used in the
industry for calculating such problems and what they believe is the most accurate design approach as
well as a motivation to why they believe so. They also discuss Eurocode 7 and describe some
problems that can occur when following this design approach if the user is not aware of them.
2.2 EARTH PRESSURE COEFFICIENT

When analysing retaining structures, the change in earth pressure due to movement is taken into
account using different earth pressure coefficients. In the initial state, when there is no lateral strain in
the soil mass, the soil is at rest and the corresponding coefficient is the earth pressure coefficient at
rest, K0.
If the structure moves, the horizontal stresses on the side it moves away from decreases and this is
called the active side. The side it moves towards, where the horizontal stresses increase, is called the
passive side, see Figure 1. The new horizontal stresses on the active and passive side are calculated
using Ka and Kp respectively.
Figure 1 – Illustration of active and passive side for an excavation supported by a retaining wall.
K0,nc, which is the lateral earth pressure coefficient at rest for a normally consolidated soil, can be
calculated using Jaky’s formula (Jaky, 1944), see Equation (1).
(1)
The Ka and Kp can be calculated according to Rankine’s earth pressure theory using Equation (2) and
(3).
(2)
(3)

The amount of wall movement required to mobilize full active and passive pressure differs, see Figure
2. It only requires small movement in the case of expansion while it takes larger lateral strains to fully
mobilize in compression.
Figure 2 - The relationship between lateral strain and earth pressure coefficients (Craig & Knappett, 2012, p. 417).
If the soil is overconsolidated, it behaves somewhat differently from a normally consolidated soil. The
earth pressure at rest is higher in an overconsolidated soil since a stress increase and consolidation has
occurred in the past. This increase of K0 can be taken into account by using a correlation to the
overconsolidation ratio (OCR), see Equation (4), suggested in Eurocode 7.
√ √ (4)
2.3 EVALUATION OF LAB TESTS

The lab tests evaluated during this project are CRS tests and triaxial tests and the evaluation process is
described below. All other evaluated data from lab and field tests used in this thesis are obtained from
Kullingsjö (2007).
2.3.1 CRS TEST

The CRS (Constant Rate of Strain) test is a test where a soil specimen is compressed with a constant
deformation rate. The deformation rate is usually 0.0025 mm/min which gives approximately 18 %
deformation in 24 hours in Gothenburg clay.
The vertical stress, σv, the vertical deformation, ε1 and the pore water pressure (PWP), u, are measured
during the test. By plotting the vertical effective stress, σ’v, against ε1, the preconsolidation pressure,
σ’c, and the compression moduli, M0 and ML, can be evaluated using the method provided in
Kompressionsegenskaper – Geotekniska laboratorieanvisningar, del 10 (Sällfors & Andreasson,
1985), see Figure 3.

Figure 3 - Evaluation of σ’c, M0 and ML from a CRS-test according to Sällfors method.
2.3.2 TRIAXIAL TEST

The triaxial test is performed on an undisturbed piston sample of clay which is inserted into a cell
containing liquid (usually water or paraffin oil). The oil is used to control the cell pressure and thus
control the horizontal effective stress σ’3. To allow for drainage, a filter paper is attached to the
sample. The clay sample is then enclosed within a rubber membrane to prevent the liquid from
penetrating the sample1.
A triaxial test is performed in two stages. During the first stage the sample is subjected to a pressure
corresponding to approximately 85 % of the preconsolidation pressure for both vertical and horizontal
stresses. The vertical preconsolidation pressure can be obtained from a CRS test, see section 2.3.1 and
the horizontal preconsolidation pressure can be calculated by multiplying the vertical pressure with
K0,nc.
Since the friction angle is evaluated from this test it is not possible to calculate K 0,nc in advance.
Instead K0 = 0.6 can be assumed2. It is important to avoid exceeding the preconsolidation pressure
since the structure of the clay can be affected which would make the evaluation of the effective
parameters corresponding to the peak strength impossible (SGF, 2012, p. 42).
During shearing, a vertical strain is applied to the specimen, resulting in a vertical effective stress σ’1,
see Figure 4, which is increased until shear failure occurs. This increase is known as the deviator
stress, q, and is calculated as the difference between axial and radial stress, see Equation (5).
(5)
1
Hedborg, Peter; Research Engineer at the Division of Geo Engineering at Chalmers, interview during study
visit 2014-02-06.
2
Ibid.

Figure 4 - The axial and radial stresses acting on a triaxial test sample.
The drainage condition in the sample is controlled by a gauge and is either drained, partially drained or
undrained. The gauge is open in a drained test which allows water to either enter or leave the sample
meaning that the PWP remains constant if the test is run at a sufficiently low rate. The PWP is
regulated with a back pressure that often is set equal to the in-situ PWP (SGF, 2012, p. 9). If the gauge
is closed, the water conditions are instead undrained and allow an increase in the PWP.
The test can be performed as either a compression test or an extension test. The axial pressure is
higher than the radial pressure in the compression test (σ’1 > σ’3), while the opposite (σ’1 < σ’3) is the
case for the extension test. Usually the consolidation stages are identical so that the tests only differ
during the shearing stage.
If the PWP is measured throughout the test this information can be used when evaluating the critical
state friction angle. Critical state is reached when the undrained peak strength is passed (Larsson,
2008, p. 49), see Figure 5. The height of the undrained peak strength depends on the deformation rate
in the test, the slower the rate the lower the peak. The critical state can be explained as the state where
the real long term behaviour of the soil is captured.
Figure 5 - Shows the different peak shear strengths depending on deformation rate in an undrained triaxial test
(Larsson, 2008, p. 49).

The parameters φ’cv and cu can be evaluated by the following procedures, illustrated in Figure 6:
a) Examine at what strain the PWP has stabilized in a ε1-σ’ plot

b) Examine the deviator stress at the axial strain level obtained.
c) Create a p’-q plot and draw a tangent from the origin to the failure envelope at the deviator
stress level obtained.
The parameter p’ is the mean effective stress in a triaxial test and is calculated using Equation (6).
(6)
The slope of the line produced in the p’-q plot, M, in Figure 6c, is then used to calculate the critical
state effective friction angle φ’cv for a compression test with Equation (7). The undrained shear
strength cu is obtained from the ε1-q, see Figure 6b.
( ) (7)
Figure 6 – Procedure for evaluating φ’cv and cu.

2.4 HAND CALCULATIONS
Simple hand calculations are performed in order to give an estimation of the earth pressures acting on
the SPW, the anchor forces and the bending moment in the SPW. For the hand calculations the method
provided in Sponthandboken (Ryner, et al., 1996) is used. This provides a rough estimation of the
anchor forces and design bending moment in the SPW which can be compared to measurements.
The parameter used in this section is:
 Undrained shear strength cu
The first step is to calculate the active horizontal earth pressure, σa, acting on the SPW above the shaft
bottom, see Equation (8) for friction material and Equation (9) for cohesion material (Ryner, et al.,
1996). These equations are based on Rankine theory but utilize the design values of the soil
parameters.
( ) (8)
(9)
In these equations γSda is a design partial factor, σv is the vertical soil pressure and u is the pore
pressure. The index d signifies a design value.
For this project, where a comparison with actual measured data is performed, the design values are no
different than the evaluated values, since all partial factors are set to 1. This gives the Equations (10) -
(11).
( ) (10)
(11)
Below the shaft bottom, a net earth pressure for clay, σpnetto, is calculated according to Equation (12)
(Ryner, et al., 1996, p. 50).
(12)
In this equation γSd,Ncb is a design partial factor, γ is the soil weight, H is the height of the SPW above
shaft bottom, qd is the load applied behind the SPW and σv,as is the vertical soil pressure above the
shaft bottom.
Ncb is a bearing capacity factor that depends on the geometry of the shaft. For an anchored SPW with
satisfactory vertical support Ncb = 5.7 is selected (Ryner, et al., 1996, p. 44). The distance d, which is
from the shaft bottom to the point D where beneath that point the total passive pressure, AP, is larger
than or equal to the total active pressure, AA, can then be found, see Figure 7.

Figure 7 - Earth pressure distribution for different D (Ryner, et al., 1996, p. 54).
The total force, PA, acting on the SPW is calculated down to the point D, see Equation (13).
∫ (13)
The load intensity, σi, is then calculated using Equation (14):
(14)
This load intensity is then distributed between the wales so that the load between two wales is
distributed evenly between them, see Figure 8. The force obtained, PH, is a horizontal force and, if the
anchors are inclined, it needs to be transformed before comparing to the measured results. The
transformation to a force P with angle α is done by Equation (15).
(15)
Figure 8 - Load intensity and how it is distributed between wales.

The design moment in the SPW usually occurs by the top wale, and is calculated as described in
Equation (16). In between the wales there is usually a lower bending moment, but beneath the bottom
wale there is another critical area. The bending moment in that span is calculated as described in
Equation (17) according to Sponthandboken (Ryner, et al., 1996, p. 55). The design bending moment
is given by the equation that gives the largest absolute value.
( ) (16)
(17)
2.5 NOVAPOINT GS SUPPORTED EXCAVATION

Novapoint GS Supported Excavation (GSS) is a one dimensional FE software included in Novapoint
GeoSuite Toolbox. In this software, the SPW is modelled as an elastic beam element divided into
vertical linear elastic beam elements (Vianova GeoSuite AB, 2010). The surrounding soil is
represented by continuous non-linear springs with varying stiffness between the different nodes, see
Figure 9. This way of modelling the SPW does not consider vertical deformations and load changes.
Figure 9 - Elements in the SPW (Vianova GeoSuite AB, 2010).
GSS provides a few different models but the model used in this project is the Total Stress Automatic,
TSA, soil model. TSA performs a total stress analysis, which is appropriate for clays with low
permeability, and the parameters used are:

 Shear modulus G
The parameters used to model the SPW are:
 Roughness r
 Young’s modulus E
 Moment of inertia I

The roughness, r, defines the amount of friction against the SPW relative to the normal stress (Vianova
GeoSuite AB, 2010).The direction in which the friction acts is also defined on both sides of the SPW
as either upwards or downwards. For a multi-anchored SPW the friction direction is upwards both
behind and in front of the SPW, see Figure 10, and for a SPW installed to bedrock, r = 0 can be
considered a conservative value (Vianova GeoSuite AB, 2010).
Figure 10 - Friction directions for an anchored SPW (Vianova GeoSuite AB, 2010).
The active and passive limit earth pressures are based on Rankine theory and calculated using
Equations (18) - (19).
(18)
(19)
In these equations, pv is the vertical earth pressure and κ is a parameter which depends on the
roughness of the SPW. For a SPW with r = 0, a passive earth pressure and an upwards friction
direction gives the parameter κ = 1, while an active earth pressure and an upwards friction direction
gives the parameter κ = 2 (Vianova GeoSuite AB, 2010).
The initial spring stiffness on both sides of the wall is K yi = 4*G (Vianova GeoSuite AB, 2010) and
the characteristics of the springs are then generated for each node. The spring stiffness is assumed to
be load dependent and is reduced when the earth pressures approach their respective limits, see Figure
11. G can be evaluated from Eu, obtained from triaxial tests, using Equation (20).
(20)
In Figure 11, an earth pressure-displacement relationship for a soil spring is shown. This curve shows
how the soil stiffness is calculated for each load step and it is generated by adopting a hyperbola
between the points of the initial earth pressure, σyi, and the limit earth pressures, (vf, σf), as seen in
Equation (21).
(21)
The term Rf in Equation (22) is a scaling factor which is calculated using Equation (22).
̅
(22)

The secant which intersects the hyperbola at the initial pressure and the limit pressure is called ̅ . The
inclination of the curve represents the spring stiffness, K y, and for the passive case it is calculated for
different passive earth pressures as seen in Equation (23).
( )
( ) (23)
Figure 11 - Earth pressure - displacement relationship for a soil spring (Vianova GeoSuite AB, 2010).
2.6 PLAXIS
The two-dimensional FE software PLAXIS is used to model the SPW with three different constitutive
models, Mohr-Coulomb (MC), Hardening Soil (HS) and the Hardening Soil with small strain stiffness
(HSs). The MC model have a different stress-strain relationship compared with HS and HSs. MC uses
an elastic-perfectly plastic relationship HS and HSs use a strain hardening elasto-plastic model.
There is an input in PLAXIS, which is used in all three models, called interface, Rinter. It is used to
model the interaction between the soil and a structure, which is done by reducing the shear strength of
the soil, see Equations (24) - (25), in a region close to the structure, see Figure 12. It also adds an extra
node to the element on the interface to allow different displacements in the soil and the structure and
thus allowing them to separate from each other. Using an interface value which is evaluated for the
soil-structure interaction improves the accuracy of the model since in reality there is not perfect
adhesion between the two and there might be some disturbance of the soil during installation which
would lower the soil strength close to the SPW. The interfaces can also be used to make a structure
impermeable in PLAXIS (Brinkgreve, et al., 2012c, pp. 96 - 102).
(24)
(25)

Figure 12 - Interface region close to a structure, from PLAXIS.
2.6.1 MOHR-COULOMB
The MC model is a linear elastic perfectly plastic model, which means that when the material is
yielding, a perfectly plastic behaviour is assumed.
The parameters used in the MC model are:

 Undrained Young’s modulus Eu
 Coefficient of lateral earth pressure K0
 Undrained Poisson’s ratio υu
The user must be aware that the MC model can overestimate the undrained shear strength of a soil if
the drainage type Undrained (A) is selected for a material in PLAXIS, see Figure 13. The input
required for this drainage type are the effective parameters and when PLAXIS calculates the undrained
shear strength, cu, it is increased until the stress path reaches the failure envelope, represented by the
inclined line in Figure 13. This gives a cu which is higher than in reality and thus overestimates the
undrained shear strength. However, this overestimation does not occur if the undrained parameters and
the drainage type Undrained (B) are used since cu is an input parameter when using this drainage type.
The stress paths for these models do not correspond well with real stress paths. This is a general
shortcoming of the MC model and is due to the elastic-perfectly plastic assumption made in the model.
Figure 13 - Stress paths for materials using MC model, with Undrained (A) and (B) respectively, compared to a real
stress path.

2.6.2 HARDENING SOIL AND HARDENING SOIL WITH SMALL STRAIN STIFFNESS
HS and HSs are two similar models, both more advanced than the MC model. The main difference
between MC and these models is that HS and HSs accounts for the stress dependency of soil stiffness,
i.e. changes the moduli depending on the stress state in the soil, see Figure 14.
Figure 14 - Stiffness evaluated from triaxial test, which shows how stiffness varies with strain after Brinkgreve et al.
(2012b, p. 83).
Since tests on sensitive soft soils are typically performed under undrained conditions, the evaluated
stiffnesses are in undrained state and need to be recalculated to effective state. For E u and Eur this is
done using Equation (26) (Muir Wood, 1990, p. 46) with an assumed undrained Poisson’s ratio, υu, of
0.5 and an effective Poisson’s ratio, υ’, of 0.2.
(26)
E50 cannot theoretically be recalculated to its effective counterpart, E’ 50, since E50 is not linear elastic,
which this relationship requires.
The yield surfaces in HS are shown in Figure 15. The yield cap, or the volumetric yield surface, is
formed as an ellipse where the length of the ellipse on the p-axis, pp, is calculated based on OCR
(Brinkgreve, et al., 2012b, pp. 73-74). Where the ellipse intersects the q-axis depends on K0,nc. The
inclination of the deviatoric yield surface is found by calculating the ratio q/p’ when σ’1 = σ’c and σ’3
= σ’c * K0,nc. Depending on the calculated stress state, and dependent on which zone that stress state is
in, different moduli apply, see Figure 15.

Figure 15 - The yield surfaces and the different stiffnesses in a HS model shown in a p’-q plot.
The difference between HS and HSs is that HSs also provides the possibility of modelling the soil
stiffness during very small strains. This stiffness is significantly higher than what can be measured
during conventional soil testing, see Figure 16. To measure such small strains would require seismic
measurement of shear waves (Larsson & Mulabdic', 1991, p. 24).
Figure 16 - Typical stiffness-strain behaviour of a soil with strain ranges for different structures and laboratory tests
(Sällfors & Atkins, 1991).
The parameters used in HS are:
 Apparent cohesion c’
 Critical state friction angle φ’cv
 Effective secant stiffness (pref = σ’3)
 Effective oedometer stiffness (pref = σ’1)
 Effective unload-reload stiffness (pref = σ’3)
 Power for stress-level dependency of stiffness m
 Effective Poisson’s ratio for unloading-reloading υ’ur
 Reference pressure pref
 Coefficient of lateral earth pressure K0

The additional parameters for HSs are:
 Initial shear modulus at reference pressure (σ’3)

 Shear strain level where G = 0.722G0 γ0.7
In the both the HS and the HSs models the stiffness varies with stress according to Equations (27)-(30)
(Brinkgreve, et al., 2012b, p. 64).
( ) (27)
( ) (28)
( ) (29)
In these equations, σ’1 is the major principal stress, σ’3 is the minor principal stress and pref is the
reference pressure. E’50 and E’ur are evaluated as described in section 2.3.2. E’oed can be evaluated
from oedometer tests but can also be evaluated through the relationship seen in Equation (30) which is
the evaluation method used in this project. Both ML and σ’c used in this equation are obtained from
CRS tests.
( ) (30)
The reference stiffness, E’iref, is calculated for all evaluated stiffnesses for the same reference pressure
to simplify the comparison of them and to use them as input in PLAXIS. For E’50 and E’ur this is done
by setting σ’3 in Equations (28) - (29) to the maximum radial stress applied during the triaxial test.
With m = 1 for soft soils (Brinkgreve, et al., 2012b, p. 63), the different reference stiffness can be
calculated by solving the Equations (27) - (30) for E’iref.
The HSs model also implements the use of the initial shear modulus, G0. Since there are no tests
performed which provide such information, G0 is evaluated using the correlation with undrained shear
strength, cu, and plasticity index, IP, presented by Larsson and Mulabdic´ (1991, p. 116) seen in
Equation (31).
( ) (31)
This undrained shear modulus is then transformed to an effective modulus using Equation (32).
(32)

The shear modulus, G’0, varies with pressure in the same way as the stiffnesses does, see Equation
(33), and thus a G’0ref value needs to be calculated as in Equation (34), and as mentioned above, m = 1.
( ) (33)
(34)
The parameter γ0.7 is the shear strain level at which the shear modulus is 72.2 % of the initial shear
modulus, see Figure 17. It can be calculated using Equation (35) (Benz, 2007, p. 23), based on a
correlation with IP.
(35)
Figure 17 – Shear modulus reduction curve for two different γ0.7.

3 MODELS, CALCULATIONS AND RESULTS
A description of the general parameters, such as the evaluated soil parameters, pore pressures, the
geometry of the excavation and the SPW and anchor characteristics can be found in this section. The
input and calculation conditions for the different calculation methods and material models used in this
thesis are also described, as well as the results obtained from the different calculations.
3.1 SITE INFORMATION

The SPW was installed during the construction of the southernmost tunnel face of Götatunneln in
central Gothenburg; see the area marked by an ellipse in Figure 18.
Figure 18 – Location of the excavation shaft (VISS, 2014).
The studied part of the SPW is located between two buildings, see Figure 19. The surrounding
buildings are located approximately 20 meters from the studied cross-section and it is therefore
assumed that they do not affect the SPW. The excavation might affect the surrounding buildings by
causing settlements in the area, but such impacts on the surrounding buildings are not considered in
this study.

Figure 19 – Plan of the excavation shaft with the studied SPW marked as a section (Kullingsjö, 2007, p. 107).
The geometry used in the calculations is based on Kullingsjö (2007, p. 108). The shaft is
approximately 11 meters deep, the SPW is installed down to and attached to the bedrock and has wales
and anchors at three different levels, as can be seen in Figure 20.
The installation procedure for the SPW and anchors is illustrated in Figure 20 and is as follows
(Kullingsjö, 2007, p. 109):
a) Installation of the SPW.

b) Excavation to a depth of 2 meters on both sides of the SPW.
c) Excavation to a depth of 4 meters in front of the SPW.
d) Installation of wales and anchors at a depth of 3.5 meters.
e) Excavation to a depth of 6.5 meters in front of the SPW and 8.5 meters just beside the SPW.
f) Installation of wales and anchors at a depth of 7.5 meters.
g) Excavation to a depth of 9 meters in front of the SPW and 11.5 meters just beside the SPW.
h) Installation of wales and anchors at a depth of 10.5 meters.
i) Excavation to a depth of 11.7 meters in front of the SPW.
This information is used to model the calculation phases for the different software.
Figure 20 – Schematic figure of the geometry of the excavation shaft (Kullingsjö, 2007, p. 108), modified by authors.

3.1.1 SOIL
The soil consists of 3 meters of fill on top of 19 meters of slightly overconsolidated clay. Beneath that
there is approximately 1.5 meters of sand and then there is bedrock, see Figure 21. The
overconsolidation of the clay is likely due to the fact that the site was used for storing iron arriving at
the harbour previously located in the area (Kullingsjö, 2007, p. 135).
Figure 21 – Soil strata and evaluated soil parameters for the analysed area (Kullingsjö, 2007, p. 136)
A compilation of the in situ conditions and the OCR can be found in Appendix 1:1. The input value
for the undrained shear strength used in the hand calculations, GSS and MC model is evaluated from a
compilation of several different tests, which can be seen together with the selected distribution in
Appendix 1:2
Raw data from CRS and triaxial tests obtained from Kullingsjö are evaluated to determine some input
parameters needed for the different models. These tests were performed by Kullingsjö and Hedborg
during the work with Kullingsjö’s doctoral thesis. The reference pressure used when evaluating all
moduli in this project is pref = 50 kPa. The soil parameters that are not evaluated from CRS or triaxial
tests are obtained from Kullingsjö (2007, p. 136), see Figure 21.
The parameter φ’cv which is used for the clay layers in all models is evaluated from triaxial tests. The
evaluated values and the selected distribution can be seen in Appendix 1:3. The range of φ’cv is from
33° to 35° which is similar to the values evaluated by Kullingsjö (2007, p. 239).

The undrained Young’s modulus, Eu, is also evaluated from triaxial tests, and the evaluated values, as
well as the selected distribution, can be found in Appendix 1:4. The chosen distribution is selected by
performing SoilTest as described in section 3.1.2.
A similar procedure is used to obtain the effective secant stiffness E’50. A first estimation of the secant
stiffness is evaluated for the undrained case which is then recalculated to an effective parameter using
Equation (26). In reality there is no such relationship between effective and undrained secant stiffness,
but since it is only used as a first estimation before calibrating with SoilTest, it is considered
acceptable. The evaluated and the selected E’50 used in the HS and HSs models can be seen in
Appendix 1:5.
The effective oedometer stiffness E’oed is evaluated using a relationship with ML and σ’c, evaluated
from CRS tests, as described in section 2.6.2. There is however a limit to how E’oed can vary in
relation to other parameters in PLAXIS and since the evaluated parameters are not within this limit
they cannot be used as input. The evaluated and selected values can be seen in Appendix 1:6.
Since only one triaxial test with unloading and reloading was performed, this is the only level where
there is evaluated effective unload-reload stiffness. Since this also is a parameter that is calibrated
using SoilTest it is possible to evaluate a distribution for the entire soil which can be seen, together
with the value evaluated from triaxial tests, in Appendix 1:7.
Both G0 and γ07 have been evaluated from correlations as described in section 2.6.2. The evaluated
values and selected distribution can be seen in Appendix 1:8-1:9.
The division of the clay into different layers will vary between the different calculation methods and
constitutive models due to the variation in input parameters between them. The soil parameters
selected for the fill layer are obtained from TRV Geo (Trafikverket, 2011, pp. 35-41) since no tests
were performed in that layer.
3.1.2 PLAXIS SOILTEST

The SoilTest function in PLAXIS can be used to validate the evaluation of field and lab tests. After
creating a material in the model, SoilTest can be used to simulate tests in this material, based on the
input parameters and constitutive model, which then can be compared to real tests (Brinkgreve, et al.,
2012a, p. 10). The simulated tests produce plots, such as the p’-q and the ε1-q seen in Figure 22, and
the parameters in the created material can be varied until adequate fits between these simulated plots
and the plots from triaxial tests are found.
The parameters that are varied are the different stiffnesses while φ’cv is kept constant, to get a failure
envelope which corresponds well with the triaxial tests. Since SPW are designed to keep deformations
small, the focus is on fitting the results for the small deformation intervals rather than the larger ones if
a fit cannot be found for the entire plot.

Figure 22 - Comparison between SoilTest and a triaxial test at 9 meter depth.
A comparison between the results from tests simulated using SoilTest and results from the triaxial
tests for different depths can be found in Appendix 2:1-2:5.

3.1.3 PORE WATER PRESSURE
The PWP is based on measurements performed during the construction of the SPW, which are
presented by Kullingsjö (2007). These measurements show that the groundwater surface is at
approximately 2 meters below the ground surface and that the hydraulic head measured at different
depths vary between approximately 2-2.5 meters beneath the ground surface, see Figure 21.
For simplified use in the models it is assumed that the groundwater surface is 2 meters below the
ground surface and that the PWP is hydrostatic.
3.1.4 SHEET PILE WALL AND ANCHORS

The SPW is an anchored Z-section AZ36 SPW, see Appendix 3:1-3:3. The characteristics of the SPW
are presented in Table 1. The SPW is installed down to and attached to the bedrock, at 23.5 m depth,
and the base of the SPW was sealed for groundwater flow to prevent hydraulic uplift (Kullingsjö,
2007, p. 106).
Table 1 - Characteristics of AZ36 SPW (ArcelorMittal, n.d.).
Parameter Value Unit
Weight, w 1.90 [kN/m2]

Young’s modulus, E [kPa]
Moment of inertia, I [m4/m]
Normal stiffness, EA [kN/m]
Flexural rigidity, EI [kNm2/m]
Poisson’s ratio, υ 0.3 [-]
The anchors used are cable anchors which are installed with an angle of 45° and anchored in bedrock.
The anchors are installed in three rows, symbolized by the arrows in Figure 23, at depths 3.5, 7.5 and
10.5 meters (Kullingsjö, 2007, pp. 110-116). The input parameters, c-c distance and the prestress
applied to the different anchors can be seen in Table 2.
Table 2 - Characteristics of anchors, see Appendix 3:1-3:3.
Parameter Row 1 Row 2 Row 3 Unit
Length 28.3 22.6 18.4 [m]

Number of strands 12 12 12 [-]
Young’s modulus, E [kPa]
Normal stiffness, EA [kN/m]
c-c distance 5.9 2.2 2.1 [m]
Prestress 170 523 548 [kN/m]
The deformation figure and anchor forces obtained from measurements performed on the SPW during
the final excavation stage (Kullingsjö, 2007) are shown in Figure 23.

Figure 23 – Measured deformations and anchor forces in the SPW during the final excavation stage.

The hand calculations are performed in Excel using the method provided in Sponthandboken (Ryner,
et al., 1996), see section 2.4. It is not possible to obtain a deformation figure with values on horizontal
deformations when using this method. Instead, the calculated anchor forces and the design bending
moment are used when comparing with measured results and other models.
3.2.1 INPUT
The soil layer division and the parameters assigned to each layer are presented in Table 3. The
groundwater table is set two meters below the surface and the PWP is assumed to increase
hydrostatically.
Table 3 - Soil layer division and parameters used for hand calculations.
Undrained
Soil Increase in undrained Friction
shear
Layer Depth [m] weight, γ shear strength with angle,
strength, cu
[kN/m3] depth, cu,inc [kPa/m] φ[°]
[kPa]
Fill 0-3 18 0 0 30
Clay 3-23.5 16 29 1 0

The parameters used for the fill layers are taken from Kullingsjö’s doctoral thesis (2007, p. 290) and
these are the design parameters stated in a contract between the Swedish Road Administration and the
contractor for this section of the SPW.
3.2.2 CALCULATION
Excel is used to calculate the earth pressure distribution acting on the SPW. The results from this
calculation are then used to calculate the resulting anchor forces and design bending moment. The
calculations and the results can be seen in Appendix 4:1-4:6.
3.2.3 RESULTS
The resulting anchor forces are presented in Table 4. The design bending moment acting on the SPW
is calculated to 154 kNm/m.
Table 4 - Forces obtained from hand calculations.
Anchor force,
Anchor row
P [kN/m]
Row 1 417
Row 2 343
Row 3 294

Since GSS is performing one-dimensional FE calculations, it is only possible to obtain the horizontal
deformations along the SPW. The vertical deformations and how the excavation is affecting the
surrounding area are not possible to obtain from the software.
3.3.1 INPUT
The model created in GSS, seen in Figure 24, has the soil layer division and parameters presented in
Table 5.
Table 5 - Soil layer division and parameters used for calculations in GSS.
Coefficient Undrained
Undrained
Soil of lateral initial
Depth shear
Layer weight, γ earth shear
[m] strength,
[kN/m3] pressure at modulus,
cu [kPa]
rest, K0 [-] G0 [MPa]
Clay 1 3-10 15-16.5 29-36 0.56-0.52 4.3-5.7

Clay 2 10-18 16.5-18 36-44 0.52-0.48 5.7-7
Clay 3 18-23.5 18-19.8 44-49.5 0.48-0.47 7-8
The constitutive model used for the clay is TSA, see section 2.5, since the clay is considered
impermeable enough to prevent pore pressures from dissipating. The reason that K0 varies is due to
OCR-correction, see section 2.2.
The fill layer is modelled using effective parameters in the model Effective Stress Simplified (Vianova
GeoSuite AB, 2010) since it is a drained layer. The parameters used for the fill layer are the same as
for the hand calculations except for c’, which is set to 1 kPa since it has to be a positive number in the

software, and E’, which is assumed to be the same as the Young’s modulus found in TRV Geo
(Trafikverket, 2011, p. 41). The value selected is for a firmly packed layer since the area has
previously been used as storage for iron coming in from the harbour (Kullingsjö, 2007, p. 135).
The properties of the SPW and the anchors used are described in section 3.1.4.
3.3.2 CALCULATION
The calculation phases are based on the installation sequence described in section 3.1. It is however
not possible to model local excavations within the shaft in the software. Instead, when there is an
excavation just beside the wall, the excavations in the software are modelled to the depth of the local
excavation.
Since the software only allows modelling an excavation in front of the SPW, the excavation to 2
meters depth on both sides of the wall could not be modelled. This was considered by setting the
ground surface 2 meters beneath the original surface, giving a fill layer 1 meter thick instead of 3
meters. The effects from not being able to include this excavation are assumed to be small.
3.3.3 RESULTS
The results from the GSS calculation can be seen in Figure 24, which shows the geometry of the open
shaft and the displacement curve.
Figure 24 – Geometry of final excavation stage and the corresponding displacement curve.
The maximum displacement is approximately 8 mm and occurs 1.5-2 m below shaft bottom. The
anchor forces are presented in Table 6.
Table 6 – Anchor forces from GSS calculation.
Anchor row Anchor force [kN] Anchor force per meter [kN/m]
Row 1 994 168
Row 2 1156 525
Row 3 1151 548

3.4 PLAXIS – GENERAL MODEL SETUP
The calculation setup is identical for all the PLAXIS models, except for constitutive models and
drainage conditions. The calculations are performed using plane strain conditions with 15 node
elements and medium mesh coarseness, see Figure 25. The mesh is gradually refined until the
difference in output between the finer mesh and the previous mesh is negligible.
The calculation type is plastic and the loading input is staged construction. The water table is reduced
to the bottom of the shaft and the excavated clusters are set to dry in each step. The phases are
modelled based on the information in section 3.1.
Figure 25 - The mesh during the final excavation stage in the MC model, with a global medium coarseness.
3.5 PLAXIS – MOHR-COULOMB

3.5.1 INPUT
The soil layer division and the input parameters used in PLAXIS are presented in Table 7 below. The
drainage conditions are set to Undrained (B) to avoid overestimation of the shear strength in the soil,
see section 2.6.1. The interface is set to 0.67 in the fill layer and to 0.5 for all the clay layers
(Karstunen, 2013b, p. 43).

Table 7 – Soil layer division and parameters used for the MC model in PLAXIS.
Coefficient
Critical
Undrained of lateral
Soil Initial state
Depth shear earth
Layer weight, γ stiffness, friction OCR [-]
[m] strength, pressure
[kN/m3] Eu [MPa] angle,
cu [kPa] at rest, K0
φ’cv [°]
[-]
Fill 0-3 18 0 50 30 - 0.47

Clay 1 3-6 16 29-32 12.4 35 1.8 0.57
Clay 2 6-8 16 32-34 7 35 1.65 0.55
Clay 3 8-11.5 16 34-37.5 20 35 1.55 0.53
Clay 4 11.5-15 17 37.5-41 16 33 1.32 0.52
Clay 5 15-23.5 18 41-49.5 22 35 1.29 0.48
3.5.2 RESULTS
Figure 26 shows the deformation figure with 27 mm as the maximum horizontal displacement. The
total principal strains span from -4.1*10-3 to 6.5*10-3, which proves that the assumption made in
section 3.1.2 that small strains occur is valid and thus the calibration of the soil parameters in SoilTest
is performed for the right strain levels.
Figure 26 – Total horizontal displacements in the SPW calculated in the MC model, scaled up 100 times.

The shear force and the bending moment in the SPW is also obtained from PLAXIS, see Table 8, as
well as the anchor forces after the final excavation stage, which can be found in Table 9.
Table 8 – Shear force and bending moment in the SPW calculated in the MC model.
Type Maximum Minimum Unit

Shear force 260 -212 [kN/m]
Bending moment 237 -198 [kNm/m]
Table 9 – Anchor forces calculated in the MC model.
Anchor row Anchor force per meter [kN/m]

Row 1 152
Row 2 498
Row 3 569
3.6 PLAXIS – HARDENING SOIL

3.6.1 INPUT
The soil layer division and the input parameters used in PLAXIS are presented in Table 10 below. The
drainage conditions are set to Undrained (A) since effective parameters are required for the soil
stiffness stress dependency, described in section 2.6.2. The input parameters are obtained using
SoilTest, see section 3.1.2.
Table 10 - Soil layer division and parameters used for the HS model in PLAXIS.
Coefficient
Effective Effective Unload- Critical
Soil of lateral
secant oedometer reload state
Depth weight, OCR earth
Layer stiffness, stiffness, stiffness, friction
[m] γ [-] pressure
E’50 E’oed E’ur angle,
[kN/m3] at rest, K0
[MPa] [MPa] [MPa] φ’cv [°]
[-]
Clay 1 3-6 16 8 5.8 16 35 1.8 0.57

Clay 2 6-11.5 16 8 4.7 16 35 1.6 0.54
Clay 3 11.5-15 17 10 5 20 33 1.32 0.52
Clay 4 15-23.5 18 5 1.65 12 35 1.29 0.48
The interface is set to 0.67 in the fill layers and to 0.5 for all the clay layers (Karstunen, 2013b, p. 43).
The fill layer is modelled the same as in the MC model since there are no tests to provide additional
information for the HS model.
3.6.2 RESULTS

Figure 27 – Total horizontal displacement in the SPW calculated in the HS model, scaled up 50 times.
Table 11 – Shear force and bending moment in the SPW calculated in the HS model.

Table 12 – Anchor forces calculated in the HS model.

Row 1 159
Row 2 511
Row 3 580

3.7 PLAXIS – HARDENING SOIL WITH SMALL STRAIN
3.7.1 INPUT
The input for the HSs model is the same as for the HS model, except for the additional parameters
presented in Table 13. The interface is set to 0.67 in the fill layers and to 0.5 for all the clay layers
(Karstunen, 2013b, p. 43).
Table 13 - Soil layer division and additional parameters used for the HSs model in PLAXIS.
Shear strain level

Depth Effective initial shear
Layer where G’/G’0 ≈ 0.7,
[m] modulus, G’0 [MPa]
γ0.7 [-]
Clay 1 3-6 20
Clay 2 6-11.5 20
Clay 3 11.5-15 23
Clay 4 15-23.5 28
3.7.2 RESULTS
Figure 28 – Total horizontal displacement in the SPW calculated in the HSs model, scaled up 100 times.

Table 14 – Shear force and bending moment in the SPW calculated in the HSs model.

Table 15 – Anchor forces calculated in the HSs model.

Row 1 162
Row 2 515
Row 3 548

4 ANALYSIS OF RESULTS
The results from the different calculations described in section 3 are compared to each other and to in-
situ measurements in this section. The different comparisons are performed to assess the accuracy of
the different methods and to illustrate their differences. Interpretations of the different results are also
included here.
4.1 SHEET PILE WALL DEFORMATIONS

When comparing the resulting deformations from the different models it is clear that they all give
results of the same magnitude, varying between approximately 7-46 mm, see Figure 29. The model
that gives the most different results compared to the measured deformations is the GSS model. The
maximum deformation differs approximately 80 % from the measured one and the model also fails to
capture the shape of the real deformation curve.
Figure 29 - Comparison of calculated deformations in the SPW from different models to measured deformations.

The results from the PLAXIS models correspond fairly well to the size of the deformation,
71-121 % of measured maximum deformation. Above the middle anchor however, the measured
deformations suggest that there are almost no deformations in the SPW.
4.1.1 DISCUSSION
The reason for the SPW deforming above the middle anchor in the PLAXIS models might be that the
idealized characteristics of the SPW and anchors used in the model might differ from reality, since
these are based on specifications from the manufacturer. When comparing the deformation curves
below the shaft bottom, a better compliance is obtained. Another reason might be that problems
occurred during installation of the measuring instruments or that these were damaged during the
construction period.
The HS model gives both the most accurate maximum deformation and a deformation curve which
matches the measured one most accurately. Another advantage is that it is the only model that
overestimates the deformations at all depths, which indicates that the results are just on the safe side.
4.2 ANCHOR FORCES

A comparison of the anchor forces show that they generally correspond very well to the measured
values and to each other, see Figure 30. There is also not much difference between the prestress and
the stress during the final excavation stage, when the shaft is completely open, for any of the models.
Figure 30 - Comparison of anchor forces calculated with different models and measured anchor forces.

When comparing the results from hand calculations it is clear that they do not correspond very well,
see Figure 31. This is because the prestress is not considered in this method. The result instead shows
what anchor forces would be necessary to reach force equilibrium between the earth pressures and the
anchor forces.
Figure 31 - Comparison of hand calculated anchor forces to measured forces and prestress.
4.2.1 DISCUSSION
The measured values are lower than the calculated ones at 7.5 m and 10.5 m depth. This might be due
to time effects in the clay or relaxation in the anchors (Kullingsjö, 2007, p. 126), which is not possible
to consider in the models used.
4.3 BENDING MOMENT AND SHEAR FORCE

The bending moments calculated in the different models are very similar down to 10.5 m depth, where
the lowest anchor row is, see Figure 32. Below that, the difference is large and the shapes of the
different distributions are varying. This trend of the bending moments being similar above the anchors
and differing below is also something observed by Schweiger (2009, p. 7) when comparing the same
constitutive models used in PLAXIS.
There are however some differences to Schweiger’s comparison, one of which being the fact that the
MC model gives the lowest bending moments in that comparison while in this comparison the HSs
model gives the lowest bending moments of the PLAXIS models. The largest difference though is the
distribution of the bending moments. The PLAXIS calculations performed in this study shows
different shapes for the different models, while they are uniform in Schweiger’s comparison (2009, p.
7).

Figure 32 - Comparison of calculated bending moments in the SPW from different models.
The maximum bending moment is rather similar for all PLAXIS models, while it is a bit lower in the
GSS model. The hand calculations do not give similar results at all, with a design bending moment of
approximately 154 kNm/m.
The shear forces calculated in the different models are, much like the bending moments, very similar
down to where the lowest anchor row is, see Figure 33, due to the high prestresses. Below that, the
models still produce quite similar distribution shapes with the magnitude varying somewhat. However,
all calculations give similar values for the maximum shear force.

Figure 33 - Comparison of calculated shear forces in the SPW from different models.
The results from the HS model indicate that there is no shear force at all from 16-20 m depth. This
explains why the bending moment calculated in the HS model is virtually constant in this span.
4.3.1 DISCUSSION
The reason for the HSs model giving lower bending moments than the MC model in this comparison
and not in the one performed by Schweiger is probably due to the results of the HSs model being very
sensitive to the choice of parameters G0 and γ0.7, see sections 5.4 and 5.5 respectively.
The differences in bending moment distribution between Schweiger’s comparison and this one are
probably due to the SPW analysed by Schweiger being installed in a much stiffer soil than the
Gothenburg clay modelled in this project. Another reason might be the difference in length of the
SPW in the two studies.
The reason for the design bending moments obtained in the hand calculations being different is most
likely due to not being able to consider the effect of the prestressed anchors.

4.4 EARTH PRESSURES
The earth pressure distribution comparison also shows that the results from the PLAXIS models are
more similar to each other than to other models, see Figure 34. The pressure distributions behind the
SPW are very similar to each other down to 9 m depth, except for the hand calculations. Below that,
the results from the PLAXIS models are fairly similar to each other, with the MC model giving
slightly lower earth pressures than the other two, and the GSS model and hand calculations giving
higher pressures.
Figure 34 - Comparison of earth pressures acting on the SPW calculated using different models.
In front of the SPW the GSS model and the hand calculations give very similar results while the
PLAXIS models generally give lower pressures. At 15 m depth, the earth pressures from the HS and
HSs models decrease instantly. Below that the results from the PLAXIS models are quite similar, with
HSs giving slightly higher earth pressures than the other two.

4.4.1 DISCUSSION
The hand calculation method assumes full mobilization of the earth pressures which does not apply in
this case since the SPW is anchored in bedrock, which prevents the movement required for full
mobilization. In the GSS model, the earth pressures are very similar to the hand calculations in front of
the SPW, and are thus nearly fully mobilized. This might be an explanation to why the resulting
deformations in GSS are smaller than the other FE models. The instant decrease in earth pressure for
the HS and HSs models at 15 m depth is probably due to there being a soil layer boundary there over
which the friction angle changes.
4.5 SUBSIDENCE BEHIND THE SHEET PILE WALL

The measured subsidence that eventually occur behind the SPW after the final excavation stage can be
seen in Figure 35, together with the resulting subsidence calculated using the different PLAXIS
models. Neither GSS nor the hand calculations are sophisticated enough to provide such information
and are thus not included.
Figure 35 - Comparison of subsidence behind the SPW calculated from the PLAXIS models and measured results.
The comparison shows that all three constitutive models underestimate the maximum measured
subsidence. None of the models capture the shape of the deformation curve for the first 20 meters
behind the wall, but after that the HS model seems to give quite realistic results. All three models also
show that heave occurs just behind the SPW, which is not something that has been measured. The
prediction of heave instead of subsidence was also something that was observed by Schweiger (2009)
when comparing the same models.
4.5.1 DISCUSSION
The reasons for heave occurring in the models and not in reality might be due to a number of things.
For example, the soil might have been disturbed during installation, causing subsidence just around the
SPW, which is not considered in the models. Another explanation might be that the models show that
the SPW bends inwards at the top which could result in a heave.

Furthermore, the calculations were performed with undrained conditions, which is most likely not the
real case. There might be partial drainage occurring in the area, which could cause consolidation
settlements in the soil. The parameter Eoed might also affect the subsidence behind the SPW. Since it is
much higher in the model than what was evaluated from CRS-tests, it might lead to an
underestimation of the settlements.

5 PARAMETRIC STUDIES
The models used for the parametric study are identical to the original models, except for certain
parameters being changed in order to determine the impact of these specific parameters. The
parameters are varied as described below and the results are then compared to the results from the
original models as well as to the measured results. Interpretations and discussion of these comparisons
are also included in this section.
5.1 E FROM EMPIRICAL EVALUATION IN PLAXIS MC

Triaxial tests are the basis for the evaluation of Eu, which is used in the original MC model. However,
the relatively high cost of triaxial tests combined with the limited knowledge about how to evaluate
them makes the use of these tests uncommon3.
If no triaxial tests are available, another procedure where the undrained Young’s modulus is evaluated
empirically as E50 can be used, see Equation (36) (Trafikverket, 2011, p. 41). Two scenarios are
calculated, one with the Young’s modulus set to E50 for the entire soil, case A, and one where the
Young’s modulus is set to 3*E50 in front of the SPW4 to account for unload-reload stiffness, case B,
see Figure 36.
(36)
Figure 36 – Two different calculation scenarios with E evaluated from an empirical correlation to c u.
The undrained shear strength distribution used in this parametric study differs from the one used in the
original MC model. This is due to not considering the undrained shear strengths evaluated from
triaxial tests, which generally give higher undrained shear strength values than conventional soil tests.
The new shear strength distribution and the empirically evaluated parameters together with the
parameters evaluated from lab tests are shown in Appendix 5:1-5:2.
A comparison between the results from a MC model with E50 evaluated using this correlation and one
with Eu evaluated from triaxial tests might provide information about the necessity of triaxial tests and
is therefore of interest. The resulting deformations in the SPW from the different calculations are
shown in Figure 37.
3
Thelander, Jonas; geotechnical engineer at Sweco, interview 2014-03-31.
4
Johansson, Pia; geotechnical engineer at Sweco, interview 2014-03-31.

Figure 37 - Comparison of resulting deformations in SPW from calculations with varying E in the MC model and
measured deformations.
This comparison shows that when using the empirical values, dividing the soil into two zones (case B),
gives more accurate results than using the same distribution of undrained Young’s modulus for the
entire soil (case A). This indicates that using empirically evaluated stiffness parameters in the MC
model might be a good option if no triaxial tests are performed.
5.2 φ’CV AND C’ IN PLAXIS HS

When evaluating drained strength parameters according to Swedish practice, φ’cv is usually assumed to
be 30°. This would when evaluating triaxial tests give a line which does not pass through the origin
but instead intersect the q-axis; half the value at which it does so is called apparent cohesion, c’.

According to TRV Geo (Trafikverket, 2011, p. 44) c’ can be estimated using either Equation (37),
using a correlation to the undrained shear strength, or Equation (38), using a correlation to the
preconsolidation pressure. Since TRV Geo (Trafikverket, 2011) is widely used as a reference in
Sweden, it is of interest to compare the methods presented there to the method used in the original HS
model.
(37)
(38)
The results from this comparison show that the calculations where φ’cv is lowered to 30º give smaller
deformations and that c’ had an insignificant impact on the result. Therefore, another calculation is
performed where φ’cv is lowered to 25º and c’ = 1 kPa, also shown in Figure 38. This further lowered
the deformations obtained. A comparison between these calculations and the original model, with
φ’cv = 35º and c’ = 1 kPa, see Figure 38.
Figure 38 – Comparison of different input values for φ’ and c’.

The explanation for smaller deformations when decreasing φ’cv might be due to a number of things.
One explanation can be found in Figure 39 and Figure 40, where the stress paths for two calculations
at depth 11.57 m with different φ’cv are plotted together with schematic yield and failure lines. It can
be seen that the yield surfaces differ for different φ’cv and that the stress paths thus are in different
areas. A large part of the stress path for the original model is within the cap hardening zone, where the
stiffness is based on Eoed. Since Eoed is generally lower than the other moduli, larger deformations are
to be expected.
Figure 39 - Stress path for a point in the soil, 11.57 m deep behind the SPW, for a calculations with the original model.
Figure 40 - Stress path for a point in the soil, 11.5 m deep behind the SPW, for a calculations with φ’cv = 25°

Another reason might be that when lowering φ’cv, K0,nc is increased which changes the initial
conditions for the soil in the model. This leads to the stress state in the soil increasing which in turn
leads to higher moduli, due to them being stress dependent. This change of initial conditions is also
what causes the volumetric yield cap to start at different p’.
The stress paths both yield seemingly independent of both the volumetric yield cap and the deviatoric
yield line. This is probably due to not being able to consider three-dimensional effects properly using
these plots. This phenomenon and the variations in deformations are probably due to how the HS
model functions.
5.3 INTERFACE IN PLAXIS HS

The interface value is not measured but instead taken from a table of suggested values (Karstunen,
2013b, p. 43) and it would therefore be interesting to see how the calculation results vary when
changing this value. Calculations for interface values between 0.1 and 1 are thus carried out in the HS
model. The resulting deformations in the SPW for the different calculations can be seen in Figure 41.
Figure 41 – Comparison of different values for the interface in the PLAXIS HS model.

This comparison shows that the input values for the interface has low impact on the resulting
deformations; it only differs 5 mm between the calculations performed with interface value 0.3 and
0.7. When using the minimum and maximum values, 0.1 and 1, in the calculations the results differ
more, approximately 20 mm, but such values are should be used with care. This suggests that when
selecting interface values within the range 0.3-0.7, they have little impact on problems where small
deformations occur and focus should not be on evaluating this parameter further for such cases.
This might be due to the fact that the SPW is anchored in bedrock, which prevents it from moving in
the bottom. If this would not have been the case and the SPW would be able to fail through rotation,
the interface value might have a greater influence.
5.4 INITIAL SHEAR MODULUS G0 IN HSS

Since G0 is based on empirical correlations, see section 2.6.2, it is of interest to see how much
changing this parameter affects the results of the calculations. Two additional calculations are
performed where G0 is varied with 20 %, see Appendix 5:3. The resulting deformations in the SPW
from these two calculations are then compared to the results of the original HSs model and the
measured deformations.
The resulting deformations in the SPW from the different calculations are shown in Figure 42.
Figure 42 - Comparison of resulting deformations in SPW from calculations with varying G0 and measured
deformations.

These results show that increasing G0 results in smaller deformations while increasing this parameter
results in larger deformations. Since there are no obvious advantages to using the HSs model for this
case compared to other models and since there is an uncertainty when selecting the parameter G0,
which affects the result, the HSs model should not be used without properly measuring G0.
5.5 γ0.7 IN HSS

Since there are different empirical correlations for γ0.7 (Benz, 2007) it would be interesting to see how
the values that these correlations give differ and how this difference affects the result in the HSs
model. To calculate γ0.7 in this parametric study Equation (39), which is based on a correlation to ,
c’, φ’cv and K0, is used instead of Equation (35).
(39)
The values evaluated using this correlation, γ0.7,b, can be seen together with the ones described in
section 2.6.2, γ0.7,a, and their respective selected distributions in Appendix 5:4. The deformation in the
SPW obtained from the calculations with varying γ0.7 is presented in Figure 43.
Figure 43 - Comparison of resulting deformations in SPW from calculations with varying γ 0.7 and measured
deformations.

This comparison that there is quite the difference between the calculation with the original γ 0.7,a and
the calculation with γ0.7,b, described above, which are both based on empirical correlations. This
indicates the same thing as the results in section 5.4, i.e. that there is no advantage to using the HSs
model compared to other constitutive models if the input parameters G0 and γ0.7 are not accurately
measured.

6 GENERAL DISCUSSION
A general discussion about issues and discoveries encountered during the work with this master’s
project is found in this section. Some general sources of error are also discussed here, as well as some
suggestions for further investigations. The specific discussions and analyses of the comparisons of the
results and the parametric studies can be found in section 4 and 5.
6.1 TRIAXIAL TESTS

Something that was discovered during the work with this project is that there is a surprising lack of
Swedish guidelines and standards on how to evaluate triaxial tests for clay soils. Most guides tend to
focus on evaluation of s-t plots rather than p’-q plots, and it is usually recommended to evaluate φ’ to
30° together with an apparent cohesion.
There also seems to be some differences between Swedish and international practice when it comes to
notation. For example, in Swedish literature, compression and extension triaxial tests are commonly
referred to as active and passive triaxial tests, probably due to the use of triaxial tests for evaluating
shear strength in active and passive zones during slope stability analysis.
The Swedish standard practice of evaluating φ’ to 30° together with the apparent cohesion in clays
might be due to the geotechnical problems in Sweden consisting largely of infrastructure projects
where embankments are common. Embankments increase the stresses in the soil and for such high
stress intervals this Swedish method of evaluating effective shear strength parameters gives a good
representation of the soil behaviour.
If, however, the case is a deep excavation with a SPW, such as in this thesis, these high stress
increases do not occur. For such cases, evaluating the critical state friction angle gives a better
representation of the soil since it includes these small stress intervals.
When creating soil models in GSS and PLAXIS, information about different soil stiffness parameters
is necessary. These can be obtained from a triaxial test or from empirical correlations to undrained
shear strength. Even though the triaxial test is more expensive than conventional soil tests, the
information obtained can be used to either improve the accuracy of the created model or to create a
more sophisticated model which would provide more accurate results. This could help avoid over-
dimensioning e.g. a SPW and thus save money in the construction stage, which would justify use of
the triaxial tests.

Hand calculations are often used as a first step in a SPW design to determine the installation depth, a
first assumption of the properties of the SPW and the number of anchors needed. These first
assumptions are then refined using more sophisticated calculation method.
However, since the SPW was already designed in this project, a back calculation was performed to
check the accuracy of the different constitutive models used. The hand calculations gave very poor
results when doing this, since there is no way of considering the prestressing of the anchors. Therefore
no conclusions about the accuracy of the hand calculations can be drawn.

When comparing the deformation figure obtained in GSS it is somewhat similar in shape to the
measured one but it is inaccurate regarding the magnitude of the deformations. The anchor forces are
close to the measured ones which is probably due to the high prestress levels. The same applies to the
shear forces obtained in GSS; it is also probably due to the prestress. When comparing bending
moment obtained GSS with the ones obtained from the other models it differs significantly, and this
could be an explanation to the large difference in the deformations. The fact that modeling excavations
behind the SPW as well as modeling local excavations within the shaft is not possible, might affect the
accuracy of the calculations.
It is difficult to assess what makes the bending moment differ since the user manual for GSS seems
incomplete, with poor descriptions of the input parameters needed for the different calculation models,
in comparison to the quite detailed PLAXIS manuals. There were some instances where the only
solution to determine which parameter to use was to ask the support staff at Novapoint. An upgrade of
the user manual would make this software easier to understand and use in the future.
6.4 PLAXIS SOILTEST AND EVALUATED PARAMETERS

The stiffness parameters were changed when simulating tests in SoilTest in order to find an accurate
match between the stress paths of the simulated tests and the real ones, and the results were still not
ideal. However, it is most important to get a good match in the early stages of the test, before shear
failure, since this is how the soil behaves when small deformations are occurring. The stress paths
correspond fairly well to each other for the early stages of the tests but tend to differ further on.
When examining the ε1-q plot, the simulated test and the real test also have a good compliance in the
early stages of the test, when the strains are still small. During shear failure and after, the curves differ
more from each other. This is due to the fact that the models used do not consider strain softening,
which occurs in the triaxial tests on soft natural soils.
In PLAXIS there is a limit to how can vary with regards to and K0. Therefore, PLAXIS
cannot handle which are as low as the ones evaluated from the CRS-tests, and they thus need to
be increased. This is probably due to the fact that the models were not developed for Scandinavian
soils. These parameters were adjusted using the suggestions from PLAXIS when trying to achieve the
best possible stress path fit in SoilTest. It was not considered a problem to change this parameter since
it is not very important when considering the small horizontal deformations that occur during deep
excavations. However, such changes might produce unrealistic results when analysing subsidence
behind the SPW.
The standard value used in PLAXIS for the reference pressure, pref, is 100 kPa. The depth at which
such an in-situ horizontal effective pressure is reached is larger than the depth of the SPW studied in
this project. Using pref = 50 kPa for the models in this project is considered more realistic.
6.5 MOHR-COULOMB AND HARDENING SOIL MODEL

The MC model might be easily perceived as a simpler model to use than other models since it needs
fewer input parameters. This might be the case if a model is to be created without using triaxial tests,
since there are fewer parameters to base on empirical correlations, and if done properly it produces
fairly accurate results, see section 5.1.
The empirical correlations used in this thesis to estimate E50 differ in magnitude if the soil has high or
low plasticity. This means that evaluation of stiffness in soils with plasticity close to the limits of high

or low plasticity give different values depending on which side of the limit it is on, making the
evaluation quite arbitrary.
Another disadvantage of evaluating stiffness parameters using empirical correlations over triaxial tests
is that the soil’s behaviour in the model cannot be validated using SoilTest, since there are no triaxial
tests to compare with. If triaxial tests are performed however, there are no benefits to using the MC
model rather than the more sophisticated HS model.
6.6 HARDENING SOIL WITH SMALL STRAIN STIFFNESS MODEL

Regarding the HSs model it is very difficult to see any advantages of using it rather than the other
models, since there were no in-situ measurements of G0 and γ0.7 performed during this project. Using a
model with an increased level of sophistication only improves the result if the input data is evaluated
properly.
6.7 PARAMETRIC STUDIES

The parametric studies were only performed for the PLAXIS models since the results from these
models were the most accurate. Another reason for this was to increase the knowledge of how
PLAXIS functions in general, but also about how the different constitutive models function.
6.8 FURTHER INVESTIGATIONS

It is difficult to say how general the results and conclusions drawn from this study are, since only one
area with a SPW has been studied. It would have been interesting to study another SPW with similar
measurements and see how the models behaved in comparison to that, and if the conclusions are valid
for other cases as well.
Comparing hand calculations according to Sponthandboken to the other methods analysed is difficult
since it is supposed to be used in the design process and not for back calculations. It is therefore
difficult to draw any conclusions about the accuracy of this method when used to design a SPW. A
comparison between different design methods could be performed to better assess the quality of this
calculation method.

7 CONCLUSIONS
The preferred method to use in the design process is the Hardening Soil constitutive model in
PLAXIS. This model produces accurate results for horizontal deformations in the sheet pile wall and
anchor forces when compared to in-situ measurements performed on a real sheet pile wall.
It is also possible to calibrate the input parameters for the Hardening Soil model by comparing
simulated triaxial tests in SoilTest with results from real triaxial tests, and thus check that the soil
behaviour is realistic. When calibrating the parameters like this it is important to validate that the
strain interval for which they were calibrated is not exceeded in the calculations, since this would give
an incorrect representation of the soil.
The accuracy of the Hardening Soil with small strain stiffness model is difficult to assess since no in-
situ measurements of the small strain stiffness were available. There are some uncertainties when
using empirical correlations for the small strain stiffness parameters, and therefore this model should
not be used without accurately measuring these parameters.
If no triaxial tests are available, the Mohr-Coulomb constitutive model in PLAXIS is the preferred
alternative. This is due to the model producing fairly accurate results when using empirical
correlations for evaluating stiffness parameters instead of evaluating these from triaxial tests.
However, validation of the model by comparing the behaviour of the soil in the model to true soil
behaviour is not possible. There are no advantages to using a more sophisticated model when basing
the stiffnesses on empirical correlations since it only increases the level of uncertainty in the model.

REFERENCES
ArcelorMittal, n.d. Profile: AZ 36. [Online]
Available at:
https://www.arcelormittal.com/projects/europe/foundationsolutions/EN/sheet_piling/AZ_sections/AZ3
6.htm
[Accessed 02 04 2014].
Benz, T., 2007. Small-Strain Stiffness of Soils and its Numerical Consequences. Stuttgart: Institut für
Geotechnik der Universität Stuttgart.
Breymann, H. & Schweiger, H. F., 2005. Chapter 123. FE-analysis of five deep excavations in
lacustrine clay and comparison with in-situ measurements. Amsterdam, Taylor & Francis.
Brinkgreve, R., Engin, E. & Swolfs, W., 2012a. PLAXIS 2D 2012l - General Information. Delft,
Netherlands: Plaxis bv.
Brinkgreve, R., Engin, E. & Swolfs, W., 2012b. PLAXIS 2D 2012 Materials Manual. Delft,
Netherlands: Plaxis bv.
Brinkgreve, R., Engin, E. & Swolfs, W., 2012c. PLAXIS 2D 2012 Reference Manual. Delft: Plaxis bv.
Craig, R. F. & Knappett, J. A., 2012. Craig's Soil Mechanics. Eight Edition ed. Abingdon: Spon Press.
Jaky, J., 1944. The coefficient of earth pressure at rest. Journal of the Society of Hungarian Architects
and Engineers, 78(22), pp. 355-358.
Karstunen, M., 2013a. Lecture 7: Soil modelling in the course BMT041 Infrastructure Geo-
engineering. Gothenburg: Chalmers University of Technology.
Karstunen, M., 2013b. Lecture 8: Earth retaining structures in the course BMT041 Infrastructure
Geo-engineering. Gothenburg: Chalmers University of Technology.
Kempfert, H.-G. & Gebreselassie, B., 2006. Excavations and Foundations in Soft Soils. Berlin:
Springer-Verlag.
Kullingsjö, A., 2007. Effects of deep excavations in soft clay on the immediate surroundings. Diss.
Göteborg: Chalmers University of Technology.
Larsson, R., 2008. Jords egenskaper, Linköping: Swedish geotechnical institute.
Larsson, R. & Mulabdic', M., 1991. Shear moduli in Scandinavian clays. Linköping: Swedish
Geotechnical Institute.
Muir Wood, D., 1990. Soil behaviour and critical state mechanics. Cambridge: Cambridge University
Press.
Powrie, W. & Simpson, B., 2001. Embedded retaining walls: theory, practise and understanding.
Istanbul, 15th International Conference on Soil Mechanics and Geotechnical Engineering.
Ryner, A., Fredriksson, A. & Stille, H., 1996. Sponthandboken. Stockholm: Byggforskningsrådet.
Schweiger, H., 2000. Ergebnisse des Berechnungsbeispieles Nr. 3 “3 - fach verankerte Baugrube":
Gegenüberstellung der eingesandten Berechnungsergebnisse. i: Verformungsprognose für tiefe
Baugruben: AK 1.6 “Numerik in der Geotechnik“. u.o.:u.n., pp. 7-67.

Schweiger, H. F., 2009. Influence of constitutive model and EC7 design approach in FE analysis of
deep excavations. Proceeding of ISSMGE International Seminar on Deep Excavations and Retaining
Structures, Budapest, pp. 99-114.
SGF, 2012. Triaxialförsök - En vägledning, Linköping: SGF - Svenska Geotekniska Föreningen.
Sällfors, G. & Andreasson, L., 1985. Kompressionsegenskaper - Geotekniska laboratorieanvisningar,
del 10. Stockholm: Statens råd för byggnadsforskning.
Sällfors, G. & Atkins, J., 1991. Experimental determination of soil properties. Proceedings of the 10th
European Conference on Soil Mechanics and Foundation Engineering, Volume 3, pp. 915-954.
Trafikverket, 2011. TRV Geo, Trafikverkets tekniska krav för geokonstruktioner. u.o.:Trafikverket.
Vianova GeoSuite AB, 2010. GS Supported Excavation Help, Solna: Vianova GeoSuite AB.
VISS, 2014. Vattenkartan. [Online]
Available at: http://www.viss.lansstyrelsen.se/MapPage.aspx
[Accessed 12 03 2014].

Appendix 1:1
Compilation of soil parameters
Appendix 1:2
Appendix 1:3
Appendix 1:4
Appendix 1:5
Appendix 1:6
Appendix 1:7
Appendix 1:8
Appendix 1:9
Appendix 2:1
Comparison between PLAXIS SoilTest and a real triaxial test at 5 meter depth.
Appendix 2:2
Appendix 2:3
Appendix 2:4
Appendix 2:5
Appendix 3:1
- Blueprints from Götatunneln
Appendix 3:2
Appendix 3:3
Appendix 4:1
- Hand calculations
Earth pressure calculation

Input
Soil Undrained Increase in undrained
Maximum Friction angle,
Layer weight, γ shear shear strength with depth,
depth, zmax [m] ϕ [°]
[kN/m^3] strength, cu cu,inc [kPa/m]
Fill 1 1 18 0 0 30
Fill 2 3 18 0 0 30
Clay 23 16.5 29 1 0
Equations
Active pressure
Friction material
Cohesion material
Passive pressure
Cohesion material
Net pressure
Cohesion material
Appendix 4:2
Active side Passive side

vertical Active vertical Passive Net earth
Depth, pressure, pressure, pressure, pressure, pressure,
z [m] σv,a [kPa] σa,a [kPa] σv,p [kPa] σp,p [kPa] σpnetto [kPa]
0 0 0 0 0 0
1 20 4 0 0 4
1 20 4 0 0 4
2 38 8 0 0 8
3 56 19 0 0 19
3 56 0 0 0 0
4 73 13 0 0 13
5 89 27 0 0 27
6 106 42 0 0 42
7 122 56 0 0 56
8 139 71 0 0 71
9 155 85 0 0 85
10 172 100 0 0 100
11 188 114 0 0 114
12 205 129 0 0 129
12 205 129 0 76 -12
13 221 143 17 95 -18
14 238 158 33 113 -24
15 254 172 50 132 -29
16 271 187 66 150 -35
17 287 201 83 169 -41
18 304 216 99 187 -46
19 320 230 116 206 -52
20 337 245 132 224 -58
21 353 259 149 243 -63
22 370 274 165 261 -69
23 386 288 166 264 -75
23.5 403 304 167 266 -78
Appendix 4:3
Appendix 4:4
Total force acting on the SPW

P1 2 kN
P2 8 kN
P3 38 kN
P4 578 kN
Sum P 626 kN
Load Intensity
σi [kPa]
where :H = 12 m, d = 0 m
σi = 58 kN/m
Appendix 4:5
Appendix 4:6
Load interval
1 4.25 m
2 3.5 m
3 3m
Horisontal forces, PH
1 247 kN/m
2 203 kN/m
3 174 kN/m
Anchor force, P
1 417 kN/m
2 343 kN/m
3 294 kN/m
Bending momen between anchors, Msd

87 kNm/m
59 kNm/m
44 kNm/m
Bending moment under anchor 3

16 kNm/m
Bracket moment anchor 1

154 kNm/m e=1
Appendix 5:1
- Parametric studies
Appendix 5:2
Appendix 5:3
Appendix 5:4

Parameter Plaxis PDF

Uploaded by

Copyright:

Available Formats

Parameter Plaxis PDF

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

Parameter Plaxis PDF

Uploaded by

Copyright:

Available Formats

§

MODELLING OF A DEEP EXCAVATION IN SOFT CLAY

Department of Civil and Environmental Engineering

MODELLING OF A DEEP EXCAVATION IN SOFT CLAY

Master of Science Thesis in the Master’s Programme Infrastructure and

Department of Civil and Environmental Engineering

© EMIL JOHANSSON, EMIL SANDEMAN, 2014.

Examensarbete / Institutionen för bygg- och miljöteknik,

Department of Civil and Environmental Engineering

Chalmers Reproservice / Institutionen för bygg- och miljöteknik

CHALMERS, Civil and Environmental Engineering, Master’s thesis 2014:22

CHALMERS, Civil and Environmental Engineering, Master’s thesis 2014:22

APPENDIX 2:1-2:5 PLAXIS SoilTest

APPENDIX 3:1-3:3 Blueprints

APPENDIX 4:1-4:6 Hand calculations

APPENDIX 5:1-5:4 Parametric studies

CHALMERS, Civil and Environmental Engineering, Master’s thesis 2014:22

Gothenburg, May, 2014

CHALMERS, Civil and Environmental Engineering, Master’s thesis 2014:22

cinter Shear strength within the interface in PLAXIS [kPa]

csoil Shear strength of the soil [kPa]

cu Undrained shear strength [kPa]

c’ Apparent cohesion [kPa]

m Power for stress-level dependency of stiffness [-]

pa Active earth pressure (GSS) [kPa]

pp Passive earth pressure (GSS) [kPa]

pp Isotropic preconsolidation pressure (PLAXIS) [kPa]

pref Reference pressure (PLAXIS) [kPa]

pv Vertical earth pressure (GSS) [kPa]

p’ Mean effective stress [kPa]

q Deviator stress [kPa]

qd Design load next to shaft for hand calculations [kPa]

r Roughness of the SPW (GSS) [-]

u Pore pressure [kPa]

uda Design pore pressure [kPa]

vf Limit earth strain (GSS) [-]

w Weight of SPW [kN/m2]

E Young’s modulus [kPa]

Eu Undrained Young’s modulus [kPa]

E50 Undrained secant stiffness [kPa]

Eur Undrained unload-reload stiffness [kPa]

CHALMERS, Civil and Environmental Engineering, Master’s thesis 2014:22

E’50 Effective secant stiffness [kPa]

E’oed Effective oedometer stiffness [kPa]

E’ur Effective unload-reload stiffness [kPa]

Effective secant stiffness at reference pressure (σ’3) [kPa]

Effective oedometer stiffness at reference pressure (σ’1) [kPa]

Effective unload-reload stiffness at reference pressure (σ’3) [kPa]

EA Normal stiffness [kN/m]

EI Flexural rigidity [kNm2/m]

G Undrained shear modulus [kPa]

G0 Undrained initial shear modulus [kPa]

G0ref Undrained initial shear modulus at reference pressure (σ’3) [kPa]

G’0 Effective initial shear modulus [kPa]

G’0ref Effective initial shear modulus at reference pressure (σ’3) [kPa]

I Moment of inertia [m4]

IP Plasticity index [-]

K0 Coefficient of lateral earth pressure at rest [-]

K0,nc Coefficient of lateral earth pressure at rest [-]