Rep HS Model

THE HARDENING SOIL MODEL -
A PRACTICAL GUIDEBOOK
Z Soil.PC 100701 report
revised 2.01.2020
by
Rafal F. Obrzud
Andrzej Truty
Edition 2020
Zace Services Ltd, Software engineering

P.O.Box 2, CH-1015 Lausanne
Switzerland
(T) +41 21 802 46 05
(F) +41 21 802 46 06
http://www.zsoil.com,
hotline: hotline@zsoil.com
since 1985
Contents
Table of Contents 1
List of symbols 9
FAQ’s 1
1 Introduction 3
1.1 Why do we need small strain extension in the HS model? . . . . . . . . . . 4
1.2 Application fields of constitutive models . . . . . . . . . . . . . . . . . . . . 8
2 Short introduction to the Hardening Soil models 13

2.1 HS-standard model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.1.1 Shear mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.1.1.1 Shear yield mechanism . . . . . . . . . . . . . . . . . . . 15
2.1.2 Stress dependent stiffness . . . . . . . . . . . . . . . . . . . . . . . 17
2.1.2.1 Stress dependent stiffness based on the mean effective stress 20
2.1.2.2 Parameter migration between minor and mean stress formu-
lations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.1.2.2.1 An integrated tool for automatic parameter migration 27
2.1.3 Shear hardening law . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.1.4 Plastic flow rule and dilatancy . . . . . . . . . . . . . . . . . . . . . 29
2.1.5 Volumetric mechanism . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.1.6 Additional strength criterion . . . . . . . . . . . . . . . . . . . . . . 34
2.1.7 Initial state variables . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.1.8 Troubleshooting for Initial State analysis . . . . . . . . . . . . . . . 39
2.2 HS-small and HS-Brick models . . . . . . . . . . . . . . . . . . . . . . . . 40
2.2.1 Non-linear elasticity for small strains . . . . . . . . . . . . . . . . . 40
2.2.2 Brick extension to the Hardening Soil model to desribe soil behavior
in the range of small strains . . . . . . . . . . . . . . . . . . . . . . 44
2.2.3 Modifications of the plastic part for the HS-small and HS-Brick models 46
2.3 Model parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3 Parameter determination 55
3.1 Experimental testing requirements for direct parameter identification . . . . 56
3.1.1 Direct parameter identification for the Hardening-Soil Standard . . . 56
3.1.2 Direct parameter identification for the HS-small . . . . . . . . . . . 63
3.1.3 Parameter identification sequence . . . . . . . . . . . . . . . . . . . 64
3.1.4 Model parameters for ”undrained” simulations . . . . . . . . . . . . 65
3.2 Alternative parameter estimation for granular materials . . . . . . . . . . . . 67
3.2.1 Initial stiffness modulus and small strain threshold . . . . . . . . . . 67
3.2.2 Secant and unloading-reloading moduli . . . . . . . . . . . . . . . . 79
3.2.3 Oedometric modulus . . . . . . . . . . . . . . . . . . . . . . . . . . 84
3.2.4 Unloading-reloading Poisson’s ratio . . . . . . . . . . . . . . . . . . 86
3.2.5 Stiffness exponent . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.2.6 Friction angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
3.2.7 Dilatancy angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
3.2.8 Coefficient of earth pressure ”at rest” . . . . . . . . . . . . . . . . . 95
3.2.9 Void ratio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
3.2.10 Overconsolidation ratio . . . . . . . . . . . . . . . . . . . . . . . . 98
3.3 Alternative parameter estimation for cohesive materials . . . . . . . . . . . . 100
3.3.1 Initial stiffness modulus and small strain threshold . . . . . . . . . . 100
3.3.2 Strength parameters . . . . . . . . . . . . . . . . . . . . . . . . . 110
3.3.3 Failure ratio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
3.3.4 Dilatancy angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
3.3.5 Stiffness moduli . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
3.3.6 Oedometric modulus . . . . . . . . . . . . . . . . . . . . . . . . . . 123
3.3.7 Stiffness exponent . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
3.3.8 Overconsolidation ratio . . . . . . . . . . . . . . . . . . . . . . . . 127
3.3.10 Void ratio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
3.4 Automated assistance in parameter determination . . . . . . . . . . . . . . 136
4 Benchmarks 139
4.1 Triaxial drained compression test on dense Hostun sand . . . . . . . . . . . 139
4.2 Isotropic compression of dense Hostun sand . . . . . . . . . . . . . . . . . . 143
4.3 Oedometric compression test . . . . . . . . . . . . . . . . . . . . . . . . . 144
4.4 Oedometric compression test - K0NC -path test . . . . . . . . . . . . . . . . 145
5 Case studies 147

5.1 Excavation in Berlin Sand . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
5.2 Twin tunnels excavation in London Clay . . . . . . . . . . . . . . . . . . . 155
5.3 Spread footing on overconsolidated Sand . . . . . . . . . . . . . . . . . . . 166
A Determination of undrained shear strength 183

A.1 Non-uniqueness of undrained shear strength . . . . . . . . . . . . . . . . . . 183
A.2 Determination of su from field tests . . . . . . . . . . . . . . . . . . . . . . 185
B Estimation of compression index 187
C Estimation of shear wave velocity 191
References 209
List of Symbols
Stress and Strain Notation Cc slope of the normal compression

ε strain line in log10 scale (=2.3λ)
εv volumetric strain Ck coefficient of curvature (=d230 /(d10 ·
= (ε1 + ε2 + ε3 ) d60 )
γs shear strain CN overburden correction factor for
σ stress SPT N60 -value
τ shear stress Cr slope of unload-reload consolida-
tion line in log10 scale
p total mean stress
= 13 (σ 1 + σ 2 + σ 3 ) Cu coefficient of uniformity (=d60 /d10 )
p0 mean effective stress D scaling parameter (by default =1.0
for HS-Std, =0.25 for HS-SmallStrain)
q deviatoric stress
= √12 [(σ 1 − σ 2 )2 + Dr relative density
E Young’s modulus
+ (σ 2 + σ 3 )2 + (σ 3 − σ 1 )2 ]1/2
e void ratio
ED dilatometer modulus (= 34.7(p1 −
Roman Symbols p0 ))
su undrained shear strength emax maximal void ratio
E0 maximal soil stiffness ft limit tensile strength
e0 initial void ratio G tangent shear modulus
Eoed tangent oedometric modulus Gur unload-reload shear modulus
E50 secant modulus corresponding to Gs secant shear modulus
50% of qf H parameter which defines the rate
Eur unloading-reloading stiffness of the volumetric plastic strain
G0 (or Gmax ) maximal small-strain shear ID dilatometer material index (= (p1 −
modulus p0 )/(p0 − u0 ))
K0 coefficient of in situ earth pres- IP plasticity index (= wL − wP )
sure ”at rest” (K0 > K0NC for KD dilatometer horizontal stress in-
OCR > 1) dex (= (p0 − u0 )/σ 0v0 )
K0NC coefficient of earth pressure ”at M parameter of HS model which de-
rest” of normally-consolidated soil fines the shape of the cap surface
K0SR stress reversal K0 coefficient defin- m stiffness exponent for minor stress
ing stress point position at inter- formulation
section between hardening mech- M∗ (or Mc∗ ) slope of critical state line
anisms (= 6 sin φ0c /(3 − sin φ0c ))
q POP (=σ 0v0 + σ 0c ) preoverburden pres- Me∗ slope of critical state line (= 6 sin φ0c /(3+
sure sin φ0c ))
Bq pore pressure parameter for CPTU MDMT constrained modulus derived from
c cohesion intercept the Marchetti’s dilatometer
c∗ intercept for M ∗ slope in q − p0 MD one-dimensional drained constrained
plane (=6c cos φ/(3 − sin φ)) modulus
mp stiffness exponent for p0 -formulation φ0cs critical state friction angle
p0 corrected first DMT reading φ0m mobilized friction angle
p1 corrected second DMT reading φ0tc effective friction angle determined
pa atmospheric pressure (average sea- from triaxial compression test
level pressure is 101.325 kPa) ψ dilatation angle
pc effective preconsolidation pressure ψm mobilized dilatation angle
in terms of mean stress ρ soil density
pco initial effective preconsolidation pres- σ ref reference stress
sure
σ 0c effective vertical preconsolidation
qa asymptotic deviatoric stress stress
qc cone resistance σL minimal limit minor stress
qf deviatoric stress at failure
Qt normalized cone resistance for CPT
qt corrected cone resistance Abbreviations
Rf failure ratio (= qf /qa ) CKo UC Ko consolidated undrained com-
u pore pressure pression
Vs shear wave velocity CKo UE Ko consolidated undrained exten-
wL liquid limit sion
wn water content CAP Cap model with Drucker-Prager
wP plastic limit failure criterion
z depth CIDC consolidated isotropic drained com-
pression
OCR overconsolidation ratio (= σ 0c /σ 0vo )
CIUC consolidated isotropic undrained
PI plasticity index
compression
CIUE consolidated isotropic undrained
Greek Symbols extension
γP S plastic strain hardening parame- CPTU cone penetration test with pore
ter for deviatoric mechanism pressure measurements (electric piezo-
γ SAT saturated unit weight cone)
γD dry unit weight CSL critical state line
γs shear strain DMT Marchetti dilatometer test
γw water unit weight DSS direct simple shear
γ 0.7 value of small strain for which Gs /G0 FVT field vane test
reduces to 0.722 HS-Brick Hardening Soil-Brick model
κ slope of unload-reload consolida- HS-small Hardening Soil-small model
tion line in ln scale HS-standard Hardening Soil-standard model
Λ plastic volumetric strain ratio (= MC Mohr-Coulomb model
1 − κ/λ)
MCC Modified Cam clay model
λ slope of primary consolidation line
NCL normal consolidation line
in ln scale
OED oedometric test
ν Poisson’s coefficient
PMT pressuremeter test
ν ur unloading/reloading Poisson’s co-
efficient SBPT self-boring pressuremeter test
φ friction angle SCPT static penetration test with seis-
φ0c effective friction angle from com- mic sensor
pression test SLS serviceability limit state analysis
φ0e effective friction angle from ex- SPT standard penetration test
tension test TC triaxial compression
UCS unified classification system
ULS ultimate limit state analysis
Sign convention: Throughout this report, the

sign convention is the standard convention of
soil mechanics, i.e. compression is assigned as
positive.
FAQs
1. When should the HS model be applied and what are its advantages?
2. Which formulation to describe the stress dependent stiffness should be chosen?
3. How to migrate stiffness moduli between two different formulations for stress dependent
stiffness?
4. How to set model parameters for an ”undrained” simulation?
5. How to troubleshoot convergence problem at the initial state analysis?

ref ref
6. What is a typical ratio Eur /E50 ?
ref ref
7. What is a typical ratio Eoed /E50 ?
8. What are typical parameter ranges?
9. Why do three different ratios K0 , K0SR and K0NC have to be defined in order to run a
simulation with the HS model?
10. What should be specified in σ ref

oed cell?
11. What should be specified in σ ref cell?
12. What is the difference between preconsolidation defined with OCR and q POP ?
13. When should I activate the small strain extension?
14. How to identify model parameters?
15. What is the suggested parameter identification sequence?
16. How to use the Virtual Lab v2020 ?

2 Z Soil.PC 100701 report (revised 2.01.2020)
Chapter 1
Introduction
The use of the finite element analysis has become widespread and popular in geotechnical
practice as a mean of controlling and optimizing engineering tasks. However, the quality
of any prediction depends on the adequate model adopted in the study. In general, a more
realistic prediction of ground movements requires using the models which account for pre-
failure behavior of soil. Such behavior, mathematically modeled with non-linear elasticity,
is characterized by a strong variation of soil stiffness which depends on the magnitude of
strain levels occurring during construction stages. Pre-failure stiffness plays a crucial role in
modeling typical geotechnical problems such as deep excavations supported by diaphragm
walls or tunnel excavations in densely built-up urban areas.
The present study completes the ZSoilr report elaborated by Truty (2008) on the Hardening
Soil models. It includes also the new Hardening Soil-Brick (HS-Brick) version (available
in ZSoilr since 2020), after Cudny and Truty (2020), which eliminates important drawbacks
(see discussion by Niemunis and Cudny (2018)) of the small strain overlay by Benz (2007)
used in the original Hardening Soil-small (HS-small) model.
The objectives of the present report can be summarized as follows:
• to highlight the need of using advanced constitutive models in daily engineering practice;
• to recall the main features of the Hardening Soil model and to facilitate under-
standing its mathematical background;
• to provide to practicing engineers who foresee using the Hardening Soil model with a helpful
guideline on specifying an appropriate testing program or making use of already acquired
experimental results in order to identify or estimate model parameters;
• to show importance of using the Hardening Soil model in typical geotechnical analyses such
as shallow footing, retaining wall excavation and tunnel excavation in an urban area;
• to compare predictions obtained with the HS-standard, HS-small and HS-Brick models
CHAPTER 1. INTRODUCTION
1.1 Why do we need small strain extension in the HS model?
It is commonly known that soil behavior is not as simple as its prediction with a simply-
formulated linear constitutive models which are commonly used in numerical analyses. Com-
plex soil behavior which stems from the nature of the multi-phase material, exhibits both
elastic and plastic non-linearities and, deformations include irreversible plastic strains. De-
pending on the history of loading, soil may contract or dilate, its stiffness may depend on the
magnitude of stress levels, soil deformations are time-dependent, etc. In fact, soil behavior for
a given stress level is considered to be truly elastic in the range of small strains as schemat-
ically presented in Figure 1.1. In this strain range, soil may exhibit a nonlinear stress-strain
relationship. However, its stiffness is almost fully recoverable in unloading conditions. In the
aftermath of pre-failure non-linearities of soil behavior, one may observe a strong variation of
stiffness starting from very small shear strains, which cannot be reproduced by models such
as linear-elastic Mohr-Coulomb model (see Figure 1.2).
Figure 1.1: Typical representation of stiffness variation in function of the shear strain amplitudes;
comparison with the ranges for typical geotechnical problems and different tests (based
on Atkinson and Sallfors, 1991, and updated be the author); SCPT - seismic cone
penetration test; CPTU - piezocone penetration test; DMT - Marchetti’s dilatometer
test; PMT - Pressuremeter test.

1.1. WHY DO WE NEED SMALL STRAIN EXTENSION IN THE HS MODEL?
(a) Shear characteristics q − ε1
(b) Normalized secant shear stiffness characteristics Gs /Go − ε1
Figure 1.2: Comparison of different model responses for drained triaxial compression condition using
equivalent parameters and OCR =1.2
Z Soil.PC 100701 report (revised 2.01.2020) 5

Engineers who are looking for reliable predictions of the engineering system response should
be aware that by applying linear-elastic, perfectly plastic models in the finite element analysis,
soil ground movements may be underestimated, which may influence the magnitude of efforts
which are computed in supporting structural elements. The models which account for high
stiffness at very small strains concentrate the development of high amplitudes of strain around
the close neighborhood of the source of deformations similarly to what is observed in reality.
This can be the case of braced excavations (e.g. Figure 1.4) or tunnel excavations (e.g. Fig-
ure 1.3) where the varying stiffness increases soil deformations at the unloading boundaries,
appropriately reducing them away from the unloaded zone (Addenbrooke et al., 1997). Fur-
thermore, it is often observed in numerical analyses that not differentiating between loading
and unloading stiffness moduli in the Mohr-Coulomb model may result in an unrealistic lifting
of the diaphragm wall, associated with unloading of the bottom of the excavation (see e.g.
Figure 1.4(c)).
The Hardening Soil model in its three variants HS-standard, HS-small and HS-Brick can be a
remedy for modeling problems which have been listed above, as they account for most of soil
behavior features (see Section 2). Despite the mathematical complexity of the HS model, its
parameters have explicit physical meaning and can be determined with conventional soil tests.
There is no difference in parameters estimation between HS-small and HS-Brick
models.
Figure 1.3: Typical model response to the excavation problem using the standard Mohr-Coulomb on
the left (unrealistic dominant heaving of the tunnel’s bottom) and the Hardening-Soil
model on the right (a realistic trough above the tunnel).

1.1. WHY DO WE NEED SMALL STRAIN EXTENSION IN THE HS MODEL?
(a)
(b)
(c)
Figure 1.4: An example of deep excavation in Berlin Sand (after Truty, 2008). Comparison of model
predictions: (a) Hardening-Soil vs Standard Mohr-Coulomb model, (b) wall deflections,
(c) surface settlements.

Figure 1.5: Recommendations for the model choice for different soils and two type analyzes: Ser-
viceability Limit State (SLS) and Ultimate Limit State (ULS). Meaning of rating: (A)
recommended, (B) recommended except situations when small strain reversals may oc-
cur and transient analyses, (C) can be used, (D) can be used but not recommended in
terms of quality of results, empty grid cell means not applicable
1.2 Application fields of constitutive models
The finite element code ZSoilr includes soil models from simple linear elastic, perfectly plastic
(e.g. Mohr-Coulomb), elasto-plastic cap models (e.g. Cap, Modified Cam Clay) to advanced
nonlinear-elasto-plastic cap model HS-small (ZACE, 2010). Table 1.1 summarizes each class
of models in terms of basic model attributes. The table includes the main model features,
failure criteria, hardening laws, and a comparison of required and corresponding soil param-
eters. It can be noticed that different models require a specification of different material
properties. However, most of them are common to all presented models.
The choice of a constitutive model depends on many factors but, in general, it is

related to the type of analysis that the user intends to perform (e.g. ultimate limit
state analysis (ULS) or serviceability limit state analysis (SLS)), expected precision of
predictions and available knowledge of the soil. In general, SLS analysis requires an applica-
tion of advanced constitutive models which predict the stress-strain relation more accurately
than simple linear-elastic, perfectly plastic models. A perceived general applicability of con-
stitutive models is schematically proposed in Figure 1.5.
It should be emphasized that the classical HS-small model was used in the past
and still can be used to carry out SLS analyses. However, this must be done with
caution. In certain cases due to small strain reversals the model strain history can
be reset causing significant stiffness overshooting. This may lead to severe under-
estimation of deformations. The HS-Brick model behavior is superior with respect
to the HS-small one, moreover it is theoretically much simpler and absolutely free
of the aforementioned deficiencies.
First approximation
Typically, the Mohr-Coulomb model (MC) is used for testing of the FE mesh discretiza-

1.2. APPLICATION FIELDS OF CONSTITUTIVE MODELS
tion and should be considered as a first quick approximation in the preliminary analyses. In
general, MC model can be applied for the estimation of the ultimate limit state (e.g. stability
analyses) or modeling of less influential, massive soil bed layers. The model is often used in
the cases where the number of soil tests and the parameter database are limited.
The use of MC is not recommended for clays and soft soils because the model overestimates
soil stiffness of normally- and lightly consolidated soils1 (there is no preconsolidation pressure
threshold beyond which important plastic straining occurs)and loading and unloading stiffness
are not distinguished.
Soft soils
In many cases, modeling of soft and near normally-consolidated clay type soils can be per-
formed with the family of volumetric cap models, i.e. Cap model and the Modified Cam
Clay model, under the assumption that the deformation of the considered soil layer are
dominated by the volumetric plastic strains. The Modified Cam clay is however not
recommended if soil exhibits a distinct non-associated (dilatant) behavior. This
shortcoming comes from the fact that the direction of strain increment is associated with
that of stress increments and the dilatancy cannot be modeled. In addition, natural soils,
especially soft clays, may exhibit viscous behavior which can be distinctly observed during
secondary consolidation. In the ZSoilr , creep behavior (including swelling) can be modeled
by means of constitutive models which exhibit pure linear elastic behavior for stress paths
that penetrate the interior of the yield surface (e.g. the Cap model).
All type of soils
Most soil types can be modeled using the family of HS models as their formulation incor-
porates two hardening mechanisms. The shear mechanism deals with the plastic straining
which is dominated by shearing what can be observed in granular soils and in overconsolidated
cohesive soils. Having formulated the volumetric hardening mechanism which is governed by
the compressive plastic strains, HS models are also suitable for modeling soft soils. It was
demonstrated on many examples that the HS models, especially the HS-small, give realistic
deformations for diaphragm walls and ground movements behind the wall in modeling ex-
cavation problems, e.g. Finno and Calvello (2005); Kempfert (2006); Benz (2007); Truty
(2008) and Section 5.1.
Since HS models are developed in the isotropic framework for both elastic behavior and hard-
ening mechanisms (uniform expansion of the yielding surfaces in all directions), modeling
of heavily overconsolidated soils which exhibit strong anisotropy should be treated
carefully.
As regards the HS-standard model, it does not include the formulation which deals with the
stiffness in the range of small strains, and therefore the stiffness parameters should be chosen
according to dominant strain levels in the modeled task. The HS-standard model is not able
to reproduce hysteretic elastic behavior nor cyclic mobility (gradual softening due to cyclic
loading).
Although the HS-small model reproduces the hysteretic elastic behavior in gen-
eral, its high sensitivity to presence of small strain reversals precludes its usage
1
It is generally assumed that a normally consolidated soil has OCR = 1, lightly overconsolidated OCR
between 1 and 3, whereas heavily overconsolidated OCR = 6 − 8 (Bowles, 1997).

for cyclic loadings. On the other hand the HS-Brick can be applied to a certain
extent for cycling loading as long as the cyclic mobility is not crucial for a given ap-
plication and as long as dynamically-induced liquefaction effects are not considered.
General limitations
Note that none of the models mentioned above is able to reproduce debonding (destructura-
tion) effects which can be observed as softening in the sensitive soils. It should also be noted
that the cap hardening parameter (preconsolidation pressure) is not coupled with the degree
of saturation, and therefore modeling of collapsible behavior of partially saturated soils is not
possible with the implemented models.

Table 1.1: Comparison of selected soil models implemented in ZSoilr .
Model Hardening Soil Cap Mohr-Coulomb

Type of model nonlinear elasto-plastic, shear and compression elasto-plastic, compression strain hardening elastic-perfectly plastic
strain hardening
Basic features • hyperbolic stress-strain relation • non-linear stress-strain relation for normally • linear stress-strain relation
and lightly overconsolidated material
• stress dependent stiffness according to power • constant stiffness (possibility of introducing lin- • constant stiffness (possibility of introducing lin-
law ear stress dependent stiffness through superele- ear stress dependent stiffness through superele-
ments) ments)
• non-linear dilatancy according to Rowe’s law • linear soil dilatancy • linear soil dilatancy (+ cut-off for excessive
(+ cut-off for excessive plastic dilatancy) plastic dilatancy)
• distinction between primary loading and un- • distinction between primary loading and un-
loading loading
• evolution of preconsolidation pressure • evolution of preconsolidation pressure
• plastic straining in primary compression • plastic straining in primary compression
• plastic straining in primary deviatoric loading
Z Soil.PC 100701 report (revised 2.01.2020)

• small strain stiffness (HS-small/HS-Brick)
• hysteretic, nonlinear elastic stress-strain rela-
tionship (small strains only)
Failure criterion Mohr-Coulomb (hexagon in π-plane) Drucker-Prager (circular in π-plane) Mohr-Coulomb (hexagon in π-plane)
Cap yield surface ellipsoidal, defined by van Eekelen criterion in π- ellipsoidal, circular in π-plane none
plane
Hardening deviatoric-shear and isotropic-compaction isotropic-compaction none
Flow rule non-associated for shear hardening non-associated for shear hardening non-associated
associated for isotropic hardening associated for isotropic hardening
Corresponding soil pa-
rameters
Small strain stiffness E0 and γ 0.7 (HS-small/HS-Brick only) none none
Elastic characteristics υ ur and Eur υ and E υ and E
E50 and m
Failure criterion φ and c φ and c φ and c
Dilatancy ψ and emax ψ ψ and emax
Cap surface parameters ref (can be determined from λ)
Eoed λ none
OCR and K0NC OCR
1.2. APPLICATION FIELDS OF CONSTITUTIVE MODELS
11

Chapter 2
Short introduction to the Hardening

Soil models
The Hardening Soil model (HS-standard) was designed by Schanz (1998); Schanz et al.
(1999) in order to reproduce basic macroscopic phenomena exhibited by soils such as:
• densification, i.e. a decrease of voids volume in soil due to plastic deformations, e.g.
Figure 2.13;
• stress dependent stiffness, i.e. observed phenomena of increasing stiffness moduli with
increasing stress level (mean stress), e.g. Figure. 2.4;
• soil stress history, i.e. accounting for preconsolidation effects;
• plastic yielding, i.e. development of irreversible strains with reaching a yield criterion,
e.g. Figure 2.2;
• dilatation, i.e. an occurrence of negative volumetric strains during shearing, e.g. Figure
2.13.
Contrary to other models such us the Cap model or the Modified Cam Clay (let alone the
Mohr-Coulmb model), the magnitude of soil deformations can be modeled more accurately by
incorporating three input stiffness parameters corresponding to the triaxial loading stiffness
(E50 ), the triaxial unloading-reloading stiffness (Eur ), and the oedometer loading modulus
(Eoed ).
An enhanced version of the HS-standard, the HS-small strain model (HS-small) was formu-
lated by Benz (2007) in order to handle a commonly observed phenomena of:
• strong stiffness variation with increasing shear strain amplitudes in the domain of small
strains (Figure 1.1);
• hysteretic, nonlinear elastic stress-strain relationship which is applicable in the range of

small strains (Figure 2.22).
These features mean that the HS-small should be able to produce more accurate and reliable
approximation of displacements which can be useful for dynamic applications or in modeling
CHAPTER 2. SHORT INTRODUCTION TO THE HARDENING SOIL MODELS
Figure 2.1: Schematic presentation of the Hardening-Soil model framework vis-à-vis the degradation
of shear stiffness with increasing shear strains.
unloading-conditioned problems, e.g. deep excavations. This model was used in the practice
for several years however recent findings by Niemunis and Cudny (2018) have shown a serious
drawback that may show up in an uncontrolled manner in that model. It is related to the
spurious reset of the strain history in the model due to tiny strain reversals. Such situations
may appear in all kind of transient analyses (consolidation, transient dynamics, creep etc..).
To remedy the HS-small model deficiences a new HS-Brick version was recently proposed by
Cudny and Truty (2020).
Although all HS models can be considered as advanced soil models which are able to faithfully
approximate complex soil behavior, they include some limitations related to specific behavior
observed for certain soils. These models are not able to reproduce softening effects associated
with soil dilatancy and soil destructuration (debonding of cemented particles) which can be
observed, for instance, in sensitive soils. As opposed to the HS-small and HS-Brick model,
the HS-standard does not account for large amplitudes of soil stiffness related to transition
from very small strain to engineering strain levels (ε ≈ 10−3 − 10−2 ). Therefore, the user
should adapt the stiffness characteristics to the strain levels which are expected to take place
in conditions of the analyzed problem. Moreover, the HS-standard model is not capable to
reproduce hysteretic soil behavior observed during cycling loading.
As an enhanced version of the HS-standard model, the HS-Brick accounts for small strain
stiffness and therefore, it can be used to some extent to model hysteretic soil behavior under
cyclic loading conditions with the exception of gradual softening which is experimentally
observed with an increasing number of loading cycles. The HS-small should not be used in
these situations at all.

2.1. HS-STANDARD MODEL
2.1 HS-standard model
2.1.1 Shear mechanism
The shear mechanism is introduced in order to handle the soil hardening which is induced
by the plastic shear strains. Domination of plastic shear strains can be typically observed for
granular materials such as sands, and heavily consolidated cohesive soils.
2.1.1.1 Shear yield mechanism
The hardening yield function for shear mechanism f1 , is described using the concept of
hyperbolic approximation of the relation between the vertical strain ε1 and deviatoric stress
q for a standard triaxial drained compression test (Figure 2.2). The yield condition is thus
expressed as follows:
Figure 2.2: Hyperbolic stress-strain relationship and the definition of different moduli in the triaxial
drained test condition.
qa q q
f1 = −2 − γP S for q < qf (2.1)
E50 qa − q Eur
where γ P S is the plastic strain hardening parameter, qa is the asymptotic deviatoric stress
which is defined by the ultimate deviatoric stress qf and the failure ratio 1 Rf is defines as:
qf
qa = (2.2)
Rf
1
A suitable value of the failure ratio is set by default Rf = 0.9. For most soils, the value of Rf falls
between 0.75 and 1. See also Section 3.3.3.

It means that for larger values of the hardening parameter γ PS , the hyperbolic relation is re-
strained by the ultimate deviatoric stress qf described by the Mohr-Coulomb criterion (Figure
2.2 and Figure 2.3):
2 sin(φ)
qf = (σ 3 + c cotφ) (2.3)
1 − sin(φ)
which is defined by the friction angle φ and the cohesion c.
Figure 2.3: Cap surface of the volumetric hardening mechanism, yield loci for the different values of
the hardening parameter γ PS and the Mohr-Coulomb criterion limiting the larger values
of γ PS .

2.1.2 Stress dependent stiffness
The secant modulus E50 which corresponds to 50% of the value of qf is defined to be minor
stress dependent using the frequently adopted power law:
m
σ ∗3 + c cotφ

ref
E50 = E50 (2.4)
σ ref + c cotφ
where σ ∗3 = max (σ 3 , σ L ), i.e. stiffness degrades with decreasing σ 3 up to the limit minor
stress σ L which can by assumed be default σ L = 10kPa; and σ ref is the minor stress at which
ref
E50 has been identified. In the triaxial compression test, σ ref corresponds to the confining
stress σ 3 (cf. Figure 3.4).
Since ZSoil 2020 another σ 3 based stress dependency is introduced omitting the term c cotφ
in the original form. The new formula eliminates the drawback which comes out when a
total stress undrained analysis is carried out assuming φ = 0 (and c = Cu ). In the standard
form the stiffness stress dependency becomes inactive and current stiffness modulus is always
equal to its reference value. The newly added formula takes the following form:
m
σ ∗3

ref
E50 = E50 (2.5)
σ ref
Note that E50 largely controls the magnitude of the plastic strains which are related to
the shear yield mechanism. In natural soil, the exponent m varies between 0.3 and 1.0.
Janbu (1963) reported values of 0.5 for Norwegian sands and silts, whereas Kempfert (2006)
provided values between 0.38 and 0.84 for soft lacustrine clays (see also Section 3.3.7). The
user may set the material stiffness to be independent on the stress level by setting
m = 0 (i.e. constant stiffness like in the standard Mohr-Coulomb model).
By analogy with E50 , the modulus Eur which defines the slope of the unloading-reloading
curve is also defined as minor stress dependent:
m
σ ∗3 + c cotφ

ref
Eur = Eur (2.6)
σ ref + c cotφ
or with the newly introduced stress dependency type without the term c cotφ (since ZSoil
2020) as
∗ m
ref σ3
Eur = Eur (2.7)
σ ref
ref ref
Note that the same σ ref applies to the stiffness moduli E50 , Eur and E0ref .

An example of stress dependency is graphically presented in Figure 2.4.

(a) (b)
(c)
σ ∗3 + c cotφ
Figure 2.4: Example of stress dependency (based on ) at initial state defined by Eq.
σ ref + c cotφ
(2.6) for different values of parameter (a) m, (b) φ, and (c) c.

2.1.2.1 Stress dependent stiffness based on the mean effective stress

In the case of dynamic analyzes or when modeling of excavation problems, a rotation of
principal stresses may occur resulting in spurious oscillations of stiffness moduli which, in
the classical formulation, depend merely on the minor stress σ 3 . For example, in the case
of excavation of a circular tunnel, one may observe the rotation of principal stresses at the
tunnel sides, and the vertical stress which initially defines the soil stiffness σ 3 = σ 0v becomes
σ 3 = σ 0h which may lead to a stiffness underestimation. This is because σ 0h decreases almost
to zero dropping the unloading-reloading stiffness to its minimal value which is limited by σ L .
Moreover, at the bottom of an excavated tunnel σ 3 which is equal to the vertical stress may
also drop to zero resulting in slightly overestimated swelling of the tunnel bottom.
In order to remedy this problem, the user can use the stress dependency formulation which
depend on the mean effective stress p0 which can be written in a general form as:
∗ mp
p
E=E ref
with p∗ = max(p0 , σ L ) (2.8)
σ ref
where:
σ ref - reference stress
E ref - reference modulus corresponding to the reference stress σ ref
mp - stiffness exponent for p0 -formulation which is equal to mp = m if c = 0,
otherwise mp 6= m
p0 - mean effective stress (σ 01 + σ 02 + σ 03 )/3
σL - the limiting stress (in order to avoid zero stiffness when p0 tends to 0)
ref ref
A unique σ ref applies to the stiffness moduli E50 , Eur and E0ref .
In the current version (ZSoil v2020 ), the evolution of the preconsolidation parameter (refer
to Eq. 2.25) remains the same for all formulations of stiffness dependency and it does not
ref
include the component c cot φ. Note that parameter transformation does not apply to Eoed
because this input parameter is not a model parameter but it serves as the reference value
for calibrating parameters M and H.
Table 2.1: Comparison of different types of stiffness stress dependency.
Minor effective stress σ 03 Mean effective stress p0

Drawback: Spurious oscillations of stiffness Stiffness oscillations independent of the principal
moduli in the case of dynamic analyzes and ex- stresses rotation
cavation problems in materials with K0 > 1
Drawback: Underestimation of unloading- Unloading-reloading stiffness depends on p0
reloading stiffness in excavation problems which is larger than σ 3 in excavation problems
Direct identification of the stiffness moduli E50 Drawback: E50 and Eur are not directly iden-
and Eur for the constant σ 3 during triaxial com- tifiable from triaxial test as p0 varies during tri-
pression test axial compression test. A simple transformation
method is provided in Figure 2.7.
Drawback: in case of σ 3 + c cotφ stress depen- Stiffness is stress dependent independently on
dency type stiffness is not stress dependent in a assumed φ value
simplified undrained analysis assuming φ = 0

In ZSoil v2020 , p-stress dependency can be activated in Elastic dialog window, as illustrated
below.
2.1.2.2 Parameter migration between minor and mean stress for-

mulations
Note that material stiffness described with σ 3 + c · cot φ, σ 3 and p0 formulations can be
ref ref
different in triaxial conditions in the case when the same values of parameters, Eur , E50 ,m
and Rf , are assumed to describe the HS model, as illustrated in Figure 2.5. It is so because
the material stiffness is obtained with respect to different scaling variables.
˙
The small strain stiffness described by σ 3 + ccotφ-formulation and p0 -formulation can also be
different across the soil initial stiffness profile assuming the same σ ref , if the same reference
moduli E0σ,ref (σ ref ) = E0p,ref (σ ref ) and the same value of stiffness exponent m = mp are taken.
In the other words, assuming the same departing stiffness profile for both formulations one
should be aware that:
• the reference moduli are different E0σ3 ,ref (σ ref ) 6= E0p,ref (σ ref ) if K0 6= 1
• the stiffness exponents m and mp are different if c > 0
Table 2.2 shows in detail equivalency of parameters in terms of compatibility of material

stiffness for different initial stress conditions and cohesion value.

500
400
Deviatoric stress q [kPa]
300
200
100
3 + c cot( )
p (without parameter migration)
0
0.00 0.02 0.04 0.06 0.08 0.10 0.12
Axial strain 1 [-]
˙ and p0 assuming the same parameter values
Figure 2.5: Triaxial test curves obtained for σ 3 + ccotφ
ref ref
for Eur , E50 , m and Rf (without parameter migration)
Table 2.2: Equivalency of model parameters in terms of compatibility of material stiffness for dif-
ferent initial stress conditions and cohesion value
c=0 c=0 c 6= 0 c 6= 0
K0 = 1 K0 6= 1 K0 = 1 K0 6= 1
σ 3 +c·cot φ → σ 3 all parameters m = mσ3 m > mσ 3 m > mσ 3
equivalent E0ref = E0σ3 ,ref E0ref 6= E0σ3 ,ref E0ref > E0σ3 ,ref
ref σ 3 ,ref ref σ 3 ,ref ref σ 3 ,ref
Eur = Eur Eur < Eur Eur < Eur
ref
E50 = E503 ,ref
σ ref
E50 < E503 ,ref
σ ref
E50 < E503 ,ref
σ
Rf = Rfσ3 Rf = Rfσ3 Rf < Rfσ3

σ 3 + c · cot φ → p m = mp m = mp m > mp m > mp
E0ref = E0p,ref E0ref > E0p,ref E0ref 6= E0p,ref E0ref > E0p,ref
ref p,ref ref p,ref ref p,ref ref p,ref
Eur > Eur Eur > Eur Eur > Eur Eur > Eur
ref p,ref ref p,ref ref p,ref ref p,ref
E50 > E50 E50 > E50 E50 > E50 E50 > E50
σ3 σ3 σ3 σ3
Rf < Rf Rf < Rf Rf < Rf Rf < Rf
σ3 → p mσ3 = mp mσ 3 = mp mσ3 < mp mσ3 < mp
E0ref = E0σ3 ,ref E0ref > E0σ3 ,ref E0ref 6= E0σ3 ,ref E0ref > E0σ3 ,ref
ref σ 3 ,ref ref σ 3 ,ref ref σ 3 ,ref ref σ 3 ,ref
Eur > Eur Eur > Eur Eur > Eur Eur > Eur
ref σ 3 ,ref ref σ 3 ,ref ref σ 3 ,ref ref σ 3 ,ref
E50 > E50 E50 > E50 E50 > E50 E50 > E50
σ3 σ3 σ3 σ3
Rf < Rf Rf < Rf Rf < Rf Rf < Rf

In order to preserve compatibility of material behavior in terms of its stiffness between dif-
ferent power law formulations, a migration of parameters should be considered. A parameter
migration can be carried out according to a general procedure as follows:
ref
1. Identify or define the parameters of the base power law, i.e. E0ref , Eur ref
, E50 , m and
Rf ; notice that parameters identified from experimental triaxial curves correspond to
σ 3 + c · cot φ formulation
ref
2. Select a stiffness modulus (Eur or E0ref ) for which the stiffness exponent m will be adjusted
for a selected target power law
3. Make migration of m and the selected modulus by applying either continuous profile re-
gression analysis (e.g. for E0ref ) or a regression analysis based on discrete values of Eur
(e.g. for three values obtained from 3 triaxial tests); both approaches are described below
ref
4. Migrate E50 and the failure ratio Rf by running optimization of three triaxial curves, i.e.
a curve-fitting of numerical triaxial drained compression curves produced with base power
ref
law parameters and and sought values of E50 and Rf describing the target power law
Figure 2.6: Migration of stiffness exponent and unloading-reloading modulus based on unloading-
reloading cycles from triaxial curves
Regression analysis based on discrete stress points

In order to obtain a similar response on unloading-reloading cycle for a given stress level for
p,ref
σ 3 +c·cot φ and p formulations, one may adjust Eur to the same value of σ ref by proceeding
the following procedure:
ref
1. Determine Eur (from triaxial test if available), m for given σ ref , φ0 and, c0 assuming σ 3 -
formulation
e.g. for σ ref = 100kPa, the following parameters were obtained from triaxial compression tests: Eur =
64000kPa, m = 0.75, φ0 = 31o and c = 5kPa

2. Specify a discrete number of different initial stress conditions described by the effective
vertical stress σ 0v0 and K0 (in order to determine σ 3 )
e.g. σ 0v0 = {50, 100, 200}kPa and K0 = 1.0 (in the case of an isotropic triaxial test), so σ 3 =
min(σ 0h0 , σ 0v0 )
3. Compute three values of the average mean stress which correspond to the half the unloading-
reloading cycle, see Figure 2.6 (q1 and q2 can be taken from experimental triaxial curves
or arbitrarily assumed)
e.g. the user can assume that the unloading begins at q1 = 0.65 · qf and end at q2 = 0.65 · qf with qf
denoting the failure stress which can be computed for the known φ, c and σ 3
4. Perform a linear regression analysis between logarithmically transformed two factors: the
reference stiffness moduli obtained for three different initial stress conditions and the nor-
malizing term of the target power law (pavg /σ ref in the case of the mean stress formulation),
the linear regression allows the power law exponent and the stiffness modulus of the target
power law to be found (refer to Figure 2.6)
Continuous profile regression analysis

Parameter migration between the σ 3 -formulation and p0 -formulation can also be carried out by
performing a simple regression analysis using a spreadsheet (refer to Figure 2.7). This
method requires plotting an evolution of the stiffness modulus computed using σ 3 -formulation
against the normalized mean effective stress. Two coefficients which define the power-type
trendline y = axb give the values of E p,ref and mp , respectively.
The above procedure optimizes the stiffness exponent m for a continuously populated
ref
stress interval and can be used to migrate E0ref and Eur .

(a) Input variables and computed data in a spreadsheet
(b) Determination of p0 -dependency parameters using

regression analysis
Figure 2.7: Migration between the σ 3 -formulation and p0 -formulation for the initial stiffness param-
eter E0ref using a spreadsheet.

500
400
Deviatoric stress q [kPa]
300
200
3 + c cot( )
100
p (without parameter migration)
p (with parameter migration)
0
0.00 0.02 0.04 0.06 0.08 0.10 0.12
Axial strain 1 [-]
Figure 2.8: Curve-fitting of numerical triaxial drained compression curves produced with base power
law parameters and and sought values of E50 ref and R for the target power law
f
Migration of E50 and Rf - curve-fitting approach

Ensuring compatibility of base and target power laws at larger strains requires migrating the
ref
secant stiffness moduli E50 and the failure ratio Rf which shapes the triaxial curve beyond
0.5qf .
ref
Migration of E50 and Rf consists of running an optimization of three triaxial curves - a
curve-fitting of numerical triaxial drained compression curves which are produced with base
ref
power law parameters and and sought values of E50 and Rf for the target power law, as
illustrated in Figure 2.8. The optimization should be constrained in terms of Rf value in
order to avoid the over-fitting of the failure axial strain. Therefore, the maximum available
change of Rf is fixed to 0.02. Moreover, the unloading-reloading branch can be ignored
ref
during curve-fitting since Eur has been previously identified.

2.1.2.2.1 An integrated tool for automatic parameter migration

Virtual Lab v2020 offers and an integrated toolbox for moduli migration (Figure 2.9). The
tool includes the migration approaches which are presented above.
The algorithm uses the regression analysis based on discrete stress points approach to migrate
the stiffness exponent m and Eur and E0ref and the curve-fitting approach for E50 and Rf .
Moreover, a continuous profile of E0ref is plotted in for a stress interval selected for parameter
migration. Note that those two values completed with the reference stress value are taken
as three discrete stress points for the regression analysis.
Figure 2.9: A toolbox in Virtual Lab v2020 for the parameter migration between different stress
dependent stiffness formulations

2.1.3 Shear hardening law
The shear hardening yield function f1 can be decomposed into part which is a function of
stress - two first components, whereas the last component is a function of plastic strains
γ P S = εp1 − εp2 − εp3 . Assuming that in the contractancy domain, the volumetric plastic strain
εpv = εp1 + εp2 + εp3 is observed to be very small εpv ≈ 0, it can thus be written:
γP S ∼ p
= 2ε1 (2.9)
Hence, for the primary loading in drained triaxial conditions, ε1 is evaluated using the yield
condition f1 (Eq.(2.1)) and decomposition of the elastic and the plastic strains:

q 1 qa q 2q qa q
ε1 = εe1 + εp1 = + − = (2.10)
Eur 2 E50 qa − q Eur 2E50 qa − q
For the drained triaxial conditions and the confining stress remaining constant (i.e. σ 2 =
σ 3 = const.), the modulus Eur remains constant and the elastic strains can be computed
from:
q q
εe1 = and εe2 = εe3 = ν ur (2.11)
Eur Eur
where ν ur denotes unloading/reloading Poisson’s ratio.
The hyperbolic relation between the axial strain and the deviatoric stress presented in Equation
2.10 can be rearranged into:
ε1
q= (2.12)
1 ε1 Rf
+
2E50 qf
which can also be rewritten in the following form:
ε1
q= (2.13)
a + bε1
These equations are graphically presented in Figure 2.10.

Note that for an anisotropically consolidated clay, the initial state deviatoric stress (after
consolidation but before compression) which corresponds to the state of zero strains is:
1 − K0
q0 = σ 3 (2.14)
K0
so the deviatoric stress after consolidation becomes:
qm = q − q0 (2.15)
and Eq.(2.12) can be rewritten as:

ε1
qm = (2.16)
1 ε1 Rf
+
2E50 qm,f
with qm,f denoting qm at failure.

(a) (b)
Figure 2.10: Graphical representation of Eq.(2.12) and identification of failure ratio Rf (a) hyper-
bolic curve plotted with laboratory data points (b) typical triaxial drained compression
results presented in the hyperbolic form (laboratory data from Kempfert, 2006).
2.1.4 Plastic flow rule and dilatancy

The plastic flow rule is derived from the following plastic potential:
σ1 − σ3 σ1 + σ3
g1 = − sin ψ m (2.17)
2 2
where the mobilized dilatancy angle ψ m is calculated in the HS-standard model according to:
sin ψ m = 0 if φm < φcs (cut-off in contractancy domain) (2.18a)

sin φm − sin φcs
sin ψ m = if φm ≥ φcs (Rowe’s dilatancy) (2.18b)
1 − sin φm sin φcs
The mobilized friction angle2 , φm , is computed from:
σ1 − σ3
sin φm = (2.19)
σ 1 + σ 3 − 2c cotφ
and the critical state friction angle which is a material property and is independent of the
stress conditions, is defined by the friction angle φ and the ultimate dilatancy angle ψ as:
sin φ − sin ψ
sin φcs = (2.20)
1 − sin φ sin ψ
The volumetric plastic strain generated by the shear plastic mechanism is expressed as follows
(dλ1 is the plastic multiplier for shear mechanism):
∂g1 ∂g1
dεpv = dλ1 + dλ1 = −dλ1 sin ψ m (2.21)
∂σ 1 ∂σ 3
2
The mobilized friction angle φm describes the stress ratio τ /σ (at the critical state φm = φcs ).

It is worth noting that dilatancy may occur for larger values of the mobilized friction angle
φm > φcs , whereas for smaller stress ratios (φm < φcs ), the material contracts and the
mobilized dilatancy angle is controlled by the cut-off criterion as presented in Figure 2.11.
Figure 2.11: Rowe’s dilatancy law and the cut-off criterion in the contractant domain for the HS-
standard model.
The dilatancy angle ψ can be derived from the maximum slope d of the ε1 − εv curve (Figure
2.13):
dεv dεp 2 sin ψ
d= ≈ vp = − (2.22)
dε1 dε1 1 − sin ψ
In order to cancel dilatancy which is produced by Rowe’s law at the critical state, an additional
cut-off criterion is introduced to respect the maximal defined void ratio emax (Figure 2.13):
if e ≥ emax sin ψ m = 0 (cut-off) (2.23)
otherwise Eq.(2.18b) is used to calculate sin ψ m .

(a) Non-associated flow for deviatoric mechanism (b) Associated flow for volumetric mechanism
Figure 2.12: Plastic flow rules in the HS model (a) graphical explanation of mobilized dilatancy ψ m
which increases from 0 up to the input dilatancy angle ψ once M-C line is reached,
(b) contractancy increases with compressive p0 stress from zero to maximum value at
M-C failure only when cap is mobilized.
eps-1 [-]
eps-v [-]
1 + emax
εv − εv0 = − ln
d
1 + e0
Dilatancy cut-off
emax reached
1
dilatancy
Figure 2.13: Strain curve for a standard triaxial drained compression test with the dilatancy cut-off.

2.1.5 Volumetric mechanism
The volumetric plastic mechanism is introduced to account for a threshold point beyond
(preconsolidation pressure) which important plastic straining occur characterizing a normally-
consolidated state of soil. Since the shear mechanism generates no volumetric plastic strain
in the contractant domain, the model without volumetric mechanism could significantly over-
estimate soil stiffness in virgin compression conditions particularly for normally- and lightly
overconsolidated cohesive soils. Such a problem can be observed when using, for instance,
the Mohr-Coulomb model.
The second yield mechanism is proposed in the form of the cap surface similarly to other
hardening models available in ZSoilr , e.g. Modified Cam Clay or Cap. The yield function
which is graphically presented in Figure 2.14 and 2.3, is thus defined as:
q2
f2 = + p02 − p2c (2.24)
M 2 r2 (θ)
where r(θ) obeys van Eekelen’s formula in order to assure a smooth and convex yield surface
(cf. also the formulation of the Modified Cam Clay model); M is the model parameter which
defines the shape of the cap surface and is related to K0NC , and pc denotes the preconsolidation
pressure which defines an intersection of the cap surface with the hydrostatic axis p0 .
Figure 2.14: 3D representation of strength anisotropy in the HS model with the Mohr-Coulomb
failure surface and the cap surface which obeys van Eekelen’s formula.
Evolution of the hardening parameter pc is described by the hardening law (since ZSoil 2020
term c cot φ is removed):
m
pc
dpc = H dεpv (2.25)
σ ref
where H is the parameter which controls the rate of volumetric plastic strains and is related
to the tangent oedometric modulus Eoed at given reference oedometric (vertical) stress level
(see Figure 3.7(a)).

The rate of the volumetric plastic strain is then computed:
dεpv = dλ2 2 p0 (2.26)
Note that the parameters M and H can easily be calculated using internal ZSoilr calculator
by providing values of K0NC and tangent oedometric modulus Eoed ref
corresponding to the
reference oedometric vertical stress σ ref
oed ; both must be captured from the primary
loading curve (Normal Consolidation Line NCL); note that σ refoed is used to compute
∗ ∗
the initial stress point defined by p and q (see Figure 3.7).
The automatic optimization of M and H must fulfill two conditions:
• K0NC produced by the model in the oedometric conditions is the same as K0NC specified by
the user.
ref
• Eoed generated by the model in the oedometric conditions is the same Eoed specified by
the user.
The internal optimization procedure runs a strain driven oedometer test with the verti-
cal strain amplitude ∆εv = 10−5 and the tangent oedometric modulus is computed as
Eoed = δσ v /δεv ∼
= ∆σ/∆εv .
The plastic potential in the volumetric mechanism is derived from the yield criterion ne-
glecting r(θ) term (Truty, 2008).

2.1.6 Additional strength criterion

Sometimes, it is necessary to control excessive tensile stresses which are built up during the
analysis, particularly when using materials with high values of cohesion. The tensile strength
condition is thus described with the Rankine’s criterion:
f3 = −σ 3 − ft = 0 (2.27)
where ft is the user-defined tensile strength (default value ft = 0) and σ 3 denotes the
minimal principal stress.
The plastic potential is associated with the cut-off condition.

2.1.7 Initial state variables

Setting the initial stress state is necessary for calculating the initial values of hardening
parameters γ PS and pc0 . This calculation can be performed by a numerical procedure at the
beginning of FE analysis based on the initial effective stress conditions σ 00 = σ 0 (σ 0x0 , σ 0y0 , σ 0z0 )
and its distance from the maximal stress point σ ’SR which is supposed to be experienced
by the soil (see Figure 2.17). In order to calculate this distance, the user has to define the
following variables:
1. Stress history variable which can be set in two ways:

• through the overconsolidation ratio OCR = σ vc /σ 0v0 where σ vc is the vertical pre-
consolidation stress (see Figure 2.15). By applying this option, a constant OCR profile
is obtained.
• through the maximal preoverburden pressure offset q POP = σ vc − σ 0v0 (see Figure
2.15). This option is useful to describe the deposits with varying OCR over the depth
(typically superficial layers of natural soil subject to mechanical overconsolidation or
dessication).
2. Historical coefficient of earth pressure at rest K0SR which corresponds to

the maximal stress point σ ’SR (Figure 2.17), and its value can be assumed as:
• K0NC consolidation: K0SR = K0NC (automatically copied from K0NC cell), applicable
to most study cases if soil was subject to K0NC natural consolidation (oedometric
conditions), or for simulating a triaxial compression or extension triaxial on a K0NC -
consolidated sample.
• Isotropic consolidation: K0SR = 1 (automatically assigned) when simulating isotropi-
cally consolidated compression or extension triaxial tests.
• Anisotropic consolidation: K0SR (user-defined) in situations when historical consoli-
dation was different than K0NC -consolidation specified for Eoed
ref
.
3. Current in situ stress configuration σ 00 = σ 00 (σ 0x0 , σ 0y0 , σ 0z0 )

σ 0y0 = ρ ·g · y
σ 0x0 = σ 0y0 · K0x
σ 0z0 = σ 0y0 · K0x

Figure 2.15: Definition of the initial preconsolidation state by means of a constant OCR and the
resulting vertical preconsolidation stress σ vc .
Figure 2.16: Definition the initial preconsolidation state by means of preoverburden pressure q POP
and the resulting variable OCR profile (typically observed for superficial soil layers).
In such a case, a variable K0 can also be considered by applying, for instance, Eq.
(3.89).

At the beginning of a FE analysis, ZSoilr sets the stress reversal point (SR) with:
σ ’SR
y = σ y · OCR or σ ’SR
y = σ 0y0 + q POP (2.28a)
and
σ ’SR ’SR SR
x = σ y K0 and σ ’SR
z = σ ’SR
y K0
SR
(2.28b)
Then ZSoilr uses the calculated σ ’SR stress state to compute initial values of the hardening
parameter γ PS
0 from the condition f1 = 0, and pc0 from f2 = 0.
Figure 2.17: Initial stress state setup. Note that for normally-consolidated soil σ 0SR coincides with
σ 00 , and therefore the Initial K0 state which is required to be set by the user is equal to
K0NC specified in the Non-linear material menu. For overconsolidated soil, the initial
state coefficient K0 is typically larger than K0NC (cf. Section 3.3.9).
Note that σ 00 in ZSoilr is computed with the Initial State driver based on gravity-induced
vertical stress and on the user-specified K0 (Initial Ko State menu):
Using the automatic K0 evaluation option, the initial K0 profile is computed as a function
of the preconsolidation state and the effective friction angle (Mayne and Kulhawy, 1982):
0
K0 = K0NC OCRsin φ (2.29)

with:
K0NC = 1 − sin φ0 (2.30)

and the upper bound value for K0 is limited to the passive lateral earth pressure coefficient:
1 + sin φ
Kp = (2.31)
1 − sin φ
It is worth noting that this option is meaningful for cohesive soils.

The initial K0 state can also be set via the Preprocessor by applying effective stresses σ 0x0 ,
σ 0y0 ,σ 0z0 using Initial stresses option, Fig.2.18. This option can be useful in the case when
imposing a variable K0 is needed (see an example in Section 5.3).
Figure 2.18: Imposing Initial stresses option in PrePro.
NB. In whole ZSoilr interface, compressive stresses are negative. In the case of
Deformation+Flow analysis, the effective stresses have to be introduced when using
Initial stresses option.

2.1.8 Troubleshooting for Initial State analysis

It may happen that the FE analysis does not converge using the Initial State driver. In such
a case the following double-check sequence is suggested to be performed:
1. Control whether Initial K0 State is specified for each material which is defined by the
material formulation Hardening-Soil.
2. Control whether K0 defined in Initial K0 State is larger than K0SR defined in Non linear
menu of the HS-small strain stiffness model.
3. Try to start Initial State analysis from a very small Initial load factor (e.g. 0.2) applying
a very small Increment (e.g. 0.1 to 0.4, see below).
Another efficient way to converge and accelerate the initial state analysis is to define the
initial stress state via the Preprocessor by means of the Initial stresses option (see Fig.2.19).
In the Initial stresses dialog window, you can put default values for the simplified stress
definition which is applied on the whole model domain (i.e. all the soil layers) or bounding
box of selected set of elements. This option will help to find the initial guess but the final
solution for each soil layer will be computed based on the user-imposed K0 defined locally in
the material definition, Fig. 2.1.7. Setting the initial guess for the initial stresses is also very
important when mixture of small and large elements or distorted elements are present in the
mesh.
Figure 2.19: Defining initial stresses using simplified definition.

2.2 HS-small and HS-Brick models
The basic HS-standard model can be extended to the HS-small model to account for the S-
shaped stiffness reduction as presented in Figure 1.1. In such a case stress paths penetrating
the para-elastic domain can be traced using a non-linear elastic stress-strain relationship.
As the HS-small is implemented as an extension of the HS-standard model, the list of model
parameters remains the same as for the HS-standard (cf. Table 2.3) and it is only extended
with two parameters defining small strain behavior, i.e. the maximal shear modulus G0 and
a characteristic shear strain level γ 0.7 at which the secant shear modulus Gs reduces to
70% of its initial value G0 (alternatively the E0 modulus can be specified). The effect of
small strain stiffness is taken into account in the HS model once the small strain extension
is activated (classical HS-small version or the new Brick one)(see below).
The study cases presented in Section 5.1,5.2 and 5.3 demonstrate sensitivity of numerical
simulations to small strain extension. For example, Figures 5.25(a) demonstrate that it is
better to run a simulation with underestimated small-strain parameters than not to account
for the small-strain stiffness at all.
2.2.1 Non-linear elasticity for small strains

In order to describe the nonlinear S-shaped stiffness reduction, the commonly known in soil
dynamics, hyperbolic Hardin-Drnevich relation is adopted. This relation relates the current
secant shear modulus Gs with an equivalent monotonic shear strain level γ hist , and it takes
the following forms:
Gs 1
for primary loading: = γ
G0 1 + a hist (2.32a)
γ 0.7
Gs 1
for unloading/reloading: = γ
G0 1 + a hist (2.32b)
2γ 0.7
with a = 0.385 modifying the original Hardin-Drnevich formula. Note that for γ hist = γ 0.7 ,
the ratio Gs /G0 is equal to 0.722 which means 72.2% reduction in the case of more accurate
considerations (see Figure 2.20).
The equivalent monotonic shear strain is computed from:
3
γ hist = εq (2.33)
2
2.2. HS-SMALL AND HS-BRICK MODELS
Figure 2.20: Reduction of the secant shear modulus Gs using Eq.(2.32a) and interpretation of the
parameter γ 0.7 .
with εq denoting the second deviatoric strain invariant, and in triaxial test conditions γ hist
can be expressed as:
γ hist = ε1 − ε3 (2.34)
The corresponding tangent shear modulus G can be expressed as:

2
G γ 0.7
= (2.35)
G0 γ 0.7 + aγ hist
The modified Hardin-Drnevich formula is only valid if γ hist ≤ γ c , with γ c being the cut-off
shear strain at which:
Eur
G = Gur where Gur = (2.36)
1 + ν ur
The stiffness cut-off allows applying the Hardin-Drnevich formula in the elastic domain (see
Figure 2.21), whereas further stiffness reduction is governed by the hardening mechanism.
The cut-off shear strain can be computed from:
r !
γ 0.7 G0
γc = −1 (2.37)
a Gur
In Eq. (2.32b), the term 2γ 0.7 replaces γ 0.7 appearing in Eq.(2.32a) for virgin loading in order
to fulfill Masing’s rule which describes the hysteretic behavior in loading/unloading conditions
(see Figure). The rule assumes that (i) initial tangent shear modulus in unloading is equal
to the initial tangent shear modulus during initial loading, and (ii) size of the unloading and
reloading curves is twice of the initial loading curve.
Although the Hardin-Drnevich law seems to be quite simple its implementation is not straight-
forward as an efficient tool is needed to detect stress reversal points and the corresponding
equivalent shear strain γ hist . Such an implementation was elaborated by Benz (2007) and
published in form of a Fortran code in his PhD thesis. However in the recent discussion by

Figure 2.21: Reduction of the tangent shear modulus G in the HS-small model based on Hardin-
Drnevich formula (Eq.(2.35)).
Figure 2.22: Hysteretic soil behavior using Masing’s rule.
Niemunis and Cudny (2018) some serious drawbacks of this implementation were detected.
In order to shortly describe the problem let us consider a simple triaxial CD test with a
complex loading program in which a small stress unloading occurs. The test setup and the
vertical loading equivalent to the σ 1 is shown in Figure 2.23. The HS-small vs the new HS-
Brick model responses are shown in Figure 2.24. It is clearly visible that the small unloading
cycle erases strain history in the HS-small model and a significant stiffness overshooting is
generated. This way stress-strain loop becomes open. Such effects may lead to significant
deformation underestimation especially in transient analyses (consolidation, dynamics).

Figure 2.23: Element test on overconsolidated material (q P OP = 150 kPa) with a complex loading
program
Figure 2.24: σ 1 − ε1 curves for HS-small and HS-Brick small strain extensions

2.2.2 Brick extension to the Hardening Soil model to desribe soil

behavior in the range of small strains
The complete theory, equations and all implementation details of the new HS-Brick model
are given in the paper by Cudny and Truty (2020). Hence only some basic ideas are presented
here. The main goal of the development was to replace the HS-small model tracing strain
history algorithm after Benz but to keep the set of HS-small model parameters. The general
concept of nested yield surfaces by Mróz et al. (1981) and its specialized version after Simpson
et al. (1979); Simpson (1992), called as Brick model, formulated in the strain space, were
used to trace stress paths penetrating the interior of current yield surfaces (shear (in the limit
the M-C envelope) and cap yield surfaces). This concept is represented in analogy to a man
pulling some finite amount of bricks (see Figure 2.25)) on strings. At the initial state strings
of different lengths are slack and then when the men is monotonically moving (going in a given
direction) strings one by one become progressively taut. The man movement represents strain
while strings lengths represent radii of circular yield surfaces in the strain space. Each time
the next brick starts to be pulled by a man the reference shear stiffness modulus is degraded
in a stepwise manner. This way the S-shaped curve is approximated by a piecewise constant
segments (see Figure 2.26). In case when some strings due to the change of the loading
direction become loose the high stiffness modulus is regained. In the implemented version of
the Brick model, the pulling process is analyzed in general six dimensional strain space. The
relative strain distances between the man and b−th brick are measured using shear strain
invariant γ b of the strain distance. A general bricks strain history update procedure is given
in Win.2-1. In the implementation 10 bricks are used to approximate stiffness degradation
curve (Nb = 10). Some symbols used in this window are explained in Figure 2.26.
Window 2-1: Updating bricks strain history

Z Soil.PC
• For given:
b
F bricks strain history: εn (b = 1..Nb )
F total accumulated strain: εn

b
F string lengths: s (b = 1..Nb )
b
F stiffness degradation factors: ∆ω
G
ref
F reference small strain shear modulus: G0
F strain increment: ∆εn+1
• Update: εn+1 , εbn+1

• Compute current reference shear tangent stiffness modulus: Gref t
• Initialize: ω G = 0.0, εn+1 = εn + ∆εn+1
• Loop over each brick b
b b
F Compute strain distance: ε̃ = εn+1 − εn
b b 1 T b
F Compute strain distance deviator: ẽ = ε̃ − 1 ε̃ 1
3
b b
F Compute: γ n+1 ẽ
b b
F if γ n+1 ≤ s
string of b-th brick is slack

F else
String of b-th brick is taut
b b b
b
Compute relative distance: d = γ n+1 − s /γ n+1
b
Update: ω G = ω G + ∆ω
G
b b b
Update brick strain history: εn+1 = εn + d εn+1 − εbn+1
• Compute reference tangent shear stiffness modulus: Gref ref
t = (1 − ω G ) G0
Window 2-1

Figure 2.25: Brick model analogy
Figure 2.26: Stepwise approximation of shear stiffness reduction curve
As it has been shown in the paper by Cudny and Truty (2020) HS-Brick behavior is superior
with respect to the classical HS-small model. All technical details and algorithmic aspects
are disscussed in this paper.

2.2.3 Modifications of the plastic part for the HS-small and HS-
Brick models
The HS-small and HS-Brick models also require some modifications in the plastic part of the
HS-standard code. These modifications concern the plastic flow rule and dilatancy in the
domain of contractancy.
Introducing the cut-off for the contractancy domain (as it is in the HS-standard model, cf.
Eq.(2.18a)) could yield too small volumetric strains. Therefore, allowing a certain amount
of contractancy for the mobilized friction angle φm before it reaches the critical state (φm <
φcs ). Introducing the scaling parameter D into Eq.(2.18b) match Rowe’s dilatancy in the
contractancy domain to the formula proposed by Li and Dafalias (2000), see Figure 2.27.
The Rowe’s dilatancy law for the HS-small and HS-Brick models is reformulated as follows:
sin φm − sin φcs
sin ψ m = D (2.38a)
1 − sin φm sin φcs
where:
D = 0.25 if sin ψ m < sin φcs (2.38b)
D = 1.00 if sin ψ m ≥ sin φcs (2.38c)
Parameter D is automatically updated to the value 0.25, once the small strain extension is
activated.
Another modification concerns the hardening laws for parameters γ PS and pc . The modifica-
tion is executed by introducing hi function which is required for an appropriate approximation
of γ − G curve in the case when a stress path starts directly from one or two yield surfaces.
Evolution of the hardening parameters is defined as follows:

PS ∂g1 ∂g1 ∂g1
dγ = dλ1 hi − − = dλ1 hi for shear mechanism (2.39)
∂σ 1 ∂σ 2 ∂σ 3
and m
pc
dpc = dλ2 2Hhi p for volumetric mechanism (2.40)
σ ref
with the function hi being defined as:
Eur
1+
hi = Gm 2E50 (2.41)

Figure 2.27: Scaled Rowe’s dilatancy vs the formula proposed by Li and Dafalias (2000).
where the stiffness multiplier Gm is calculated as:

Gt
Gm = min( ) (2.42)
Gur

Table 2.3: List of parameters defining the HS-standard and HS-small/HS-Brick models.
Parameter Unit HS- HS- Function

standard small/HS-
Brick
Stress depen- 3 3 defines stiffness moduli stress dependency
dency type function which is based on σ 3 + c cotφ, pure
σ 3 or p
Small strain – 3 defines whether classical HS-small or HS-Brick
formulation (recommended) version is to be used
E0ref [kPa] – 3 defines the initial tangent slope of ε1 −q curve
at the reference minor principal stress σ ref3
γ 0.7 [–] – 3 defines a characteristic shear strain level γ s at
which the ratio Gs /G0 = 0.722
ref
Eur [kPa] 3 3 defines unloading/reloading stiffness at engi-
neering strains (ε ≈ 10−3 ) at the reference
minor principal stress σ ref 3
ref
E50 [kPa] 3 3 defines the secant stiffness at 50% of the ulti-
mate deviatoric stress qf at the reference mi-
nor principal stress σ ref
3
σ ref [kPa] 3 3 reference stress used to scale stiffness moduli
E0ref , Eur
ref ref
, E50 to current values with respect
to a current minor principal stress σ 3 (or a
current mean stress p0 if this formulation is
selected)
m [–] 3 3 defines stress dependent stiffness through
Eq.(2.6)
ν ur [–] 3 3 defines the ratio ε3 /ε1 in an unloading-
reloading cycle (elastic deformations)
Rf [–] 3 3 used to compute the hardening parameter γ PS
with the use of the asymptotic deviatoric stress
qa defining the hyperbolic function f2 (default
Rf = 0.9)
c0 [kPa] 3 3 defines the intercept of the Mohr-Coulomb line
at null stress condition
φ0 [o ] 3 3 defines the slope of the Mohr-Coulomb yield
criterion
ψ [o ] 3 3 defines the maximal slope of ε1 − εv curve for
dilatancy
emax [–] 3 3 defines the cut-off limit corresponding to the
maximal void ratio observed in material at the
ultimate state
ft [kPa] 3 3 defines the maximal tensile strength for mate-
rial
D [–] 3 3 controls Rowe’s dilatancy law in the contrac-
tancy domain (default D = 0 for HS-Std,
D = 0.25 for HS-SmallStrain)
M [–] 3 3 defines the shape of the elliptical cap yield sur-
face
H [kPa] 3 3 defines the rate of the plastic volumetric strain
and the preconsolidation pressure
OCR or q POP [–] or [kPa] 3 3 sets the initial position of stress with respect
to the cap surface and it is used to compute
the hardening parameter γ PS and pc0
K0SR [–] 3 3 sets a historical position of the stress point
σ ’SR (K0SR = σ ’SR ’SR
h /σ v ) with respect to the
initial stress configuration for an overconsoli-
dated material and it is used to compute the
hardening parameter γ PS and and pc0
Figure 2.28: Dialog window for the Elastic group of parameters which define the HS model in-
cluding the small strain extension (in the top window possible choices of stiffness
stress dependency (barotropy) is shown while in the bottom one possible small strain
extensions.

Figure 2.29: Dialog window for the Nonlinear group of parameters which define the HS model
including the initial state setup.

2.3. MODEL PARAMETERS
2.3 Model parameters
Although the HS model is mathematically complex, its parameters have the physical meaning
and they can be derived from the standard laboratory test, i.e. the triaxial compression and
oedometer tests. A complete list of parameters that the user needs to specify before running
application is provided in Table 2.3. The details related to the identification of specific pa-
rameters are provided in the subsequent Chapter 3.
The following abbreviations apply to Table 2.4:
• CICD - triaxial test: consolidated isotropically compression drained
• CICU - triaxial test: consolidated isotropically compression undrained
• OED - oedometer test
• CPT - cone penetration test
• CPTU - piezocone cone penetration test
• DMT - Marchetti’s dilatometer test
• SCPTU - piezocone cone penetration test with seismic sensor
• DMT - Marchetti’s dilatometer test with seismic sensor
• SPT - standard penetration test

Table 2.4: List of parameters which should be provided by the user (advanced parameters in gray).
Model Direct estimation Alternative

parameter Unit test test or solution
Small stiffness (HS-SmallStrain only)
E0ref [kPa] SCPT, DMT or bore-hole, unloading-reloading branch
cross-hole or other geophys- of CICD; geotechnical evi-
ical method dence; sands: CPT
γ 0.7 [–] CICD with local gauges geotechnical evidence
Elastic constants
ref
E50 (σ ref ) [kPa] min. 1 CICD at σ ref
3 sands: CPT
ref
Eur (σ ref ) [kPa] min. 1 CICD at σ ref 3
geotechnical evidence
σ ref [kPa] 1 CICD
ν ur [–] min. 1 CICD with geotechnical evidence
unloading-reloding curve
m [–] 3 CICD at different σ 3 geotechnical evidence
Shear mechanism
c [kPa] 3 CICD or CICU at different
σ3
φ [o ] 3 CICD or CICU at different geotechnical evidence; sand:
σ3 CPT, DMT, PMT, SPT
ψ [o ] min. 1 CICD geotechnical evidence
Rf [–] min. 1 CICD default Rf = 0.9, geotech-
nical evidence
emax [–] min. 1 CICD on a dense geotechnical evidence
or preconsolidated soil spec-
imen
ft [kPa] isotropic extension default ft = 0
D [–] min. 1 CICD default D=0 for HS-
Standard and D=0.25
HS-SmallStrain
Volumetric (cap) mechanism
ref
Eoed (σ ref
oed ) [kPa] min.1 OED clays: CPT, DMT
ref
σ oed [kPa] idem idem
Initial state variables (soil history)
OCR or q POP [–/kPa] min. 1 OED clay: CPT, CPTU, DMT
K0SR [–] K0 -consolidation ”Jaky’s formula”

2.3. MODEL PARAMETERS
The table below presents typical ranges of HS-model parameters in soils. It also indicates rel-
vant sections where the intersted user may find more information about parameter estimation
in case of lack laboratory data.
Table 2.5: Typical values and ranges for parameters of the HS model.
Model
parameter Unit Coarse soils Fine soils
Small stiffness (HS-SmallStrain only)
E0ref [kPa] Sec.3.2.1 Sec.3.3.1
γ 0.7 [–] 7 · 10−5 < γ 0.7 < 4 · 10−5 γ 0.7 > 9 · 10−5
cf. Sec.3.2.1 cf. Sec.3.3.1
Elastic constants
ref
E50 (σ ref ) [kPa] ref
Eur /3, cf.3.2.2 Eurref
/3, cf. Sec.3.3.5
ref
Eur (σ ref ) [kPa] Sec.3.2.2 Sec.3.3.5
σ ref [kPa] typically taken as 100 kPa, cf. also in
Virtual Lab report
ν ur [–] 0.15 < ν ur < 0.25, cf. Sec.3.2.4
m [–] 0.5, cf. Sec.3.2.5 0.5 < m < 1.0, cf.
Sec.3.3.7
Shear mechanism
c0 [kPa] 0÷5 ≥0
φ0 [o ] o
25 < φ < 50 o
18o < φ < 42o
cf. Sec.3.2.6 cf. Sec.3.3.2
ψ [o ] Sec.3.2.7 cf. Sec.3.3.4
Rf [–] 0.75 < Rf < 1 with average Rf = 0.9, cf. Sec.3.3.3
emax [–] Sec.3.2.9 Sec.3.3.10
ft [kPa] default ft = 0
D [–] default D=0 for HS-Standard and D=0.25 HS-SmallStrain
Volumetric (cap) mechanism
ref
Eoed (σ ref
oed ) [kPa] ≈ E50ref
(σ ref ), cf. Sec.3.2.3 and 3.3.6
σ ref
oed [kPa] ≈ σ ref /K0NC , cf. Sec.3.2.3
K0NC [–] good-working equation: K0NC = 1 − sin φ0
Initial state variables (soil history)
OCR or q POP [–/kPa] OCR ≥ 1 or q POP ≥ 0
K0SR [–] = K0NC for natural soils, cf. Sec. 2.1.7
0
K0 [–] good-working equation: K0 = K0NC OCRsin φ , cf. Sec.3.3.9


Chapter 3
Parameter determination
As most of the constitutive models for soils, the HS-standard model has been designed
based on behavior of soil specimen which is observed during laboratory tests with the use of
standard devices such as triaxial cell and oedometer. Therefore, still responding to certain
test requirements such as drained compression, model parameters can be derived directly
from the experimental curves. Direct parameter identification is presented in Section 3.1.
Sometimes, the test requirements cannot be fulfilled (e.g. performing drained compression
test on low permeable clay specimen may prove to be too time consuming). Then, the model
can still be calibrated using, for instance, the measurements derived from the undrained tri-
axial compression test or the model parameters can be estimated based on results obtained
through in situ tests or approximated using parameter correlations observed in geotechnical
practice. Such an indirect parameter determination is presented in Section 3.2 for sand type
materials, and in Section 3.3 for cohesive soils.
Additional parameter which describes the small stain stiffness in HS-small and HS-Brick
models can easily be determined using the measurements derived from one of the in situ
probes equipped with a seismic sensor which allows measuring the velocity of shear waves.
Owing to time and economical constraints of laboratory testing, and the effect of speci-
men disturbances during soil sampling, the use of laboratory devices to determine G0 seems
less reasonable. Nevertheless, an approximate value of G0 can be derived from unloading-
reloading branch derived from the triaxial compression test.
The following sections provide a comprehensive guideline on parameter identification which

may help the user to effectively apply the advanced constitutive models. In this context,
the guideline may be helpful in specifying an appropriate testing program or making use of
already acquired experimental results which need a specific treatment in order to estimate
model parameters.
CHAPTER 3. PARAMETER DETERMINATION
3.1 Experimental testing requirements for direct parameter identi-

fication
3.1.1 Direct parameter identification for the Hardening-Soil Stan-

dard
Direct parameter identification for the HS-standard model requires the use of two commonly
used laboratory devices:
• triaxial cell with consolidated isotropically drained compression test (CICD); three pro-
grammed compression tests at different confining pressures σ 3 should provide:
F stress paths in p0 − q plane which are used to determine strength parameters φ (= φ0 )

and c (= c0 ), according to Figure 3.1 or Mohr’s circles;
Note that φ and c can also be derived from the undrained compression test1 (CU)
considering that the failure stress envelopes derived from drained and undrained tests
are essentially similar, see an example in Figure 3.2;
F relationships ε1 − q which is used to determine unloading-reloding modulus Eur 2,3 , as
shown in Figure 3.4; secant modulus E50 and failure ratio Rf according to Figure 3.3;
and stiffness stress dependency parameter m according to Figure 3.5;
F relationships ε1 − εv which is used to determine the dilatancy angle ψ and the maximal
void ratio emax , as shown in Figure 3.6.
• oedometer; the test should provide pre- and post-yield evolution of the void ratio (or
specimen height) with respect to changes of vertical effective stress, σ 0v − e, which is used
to estimate:
F the preconsolidation pressure σ 0c for cohesive deposits, which is then used to determine
the overconsolidation ratio OCR defined as:
σ 0c
OCR = 0 (3.1)
σ v0
F the tangent oedometric stiffness Eoed for corresponding σ ref

oed which have to be cap-
4
tured from the primary loading curve (postyield branch), see Figure 3.7.
The preconsolidation pressure σ 0c understood as a threshold point beyond which the important
plastic straining occur, is difficult to establish unambiguously. Among a number of methods
proposed in literature for determining σ 0c , the following ones are commonly used owing to
their simplicity:
In the case of the undrained test, the maximum principal stress ratio (σ 01 /σ 03 )max and the maximum
1
deviatoric stress (σ 1 − σ 3 )max can be considered as the failure criteria.

2
Note that Eur corresponds to Young’s modulus E which is specified by the user in Mohr-Coulomb or
Cap model.
3
Note that E0 > Eur > E50 or G0 > Gur > G50 .
4
This condition implies that at given oedometric pressure, both shear and volumetric mechanisms are
active following K0NC consolidation line, as shown in Figure 3.7(b).

3.1. EXPERIMENTAL TESTING REQUIREMENTS FOR DIRECT PARAMETER
IDENTIFICATION
• empirical, graphical Casagrande’s method (Casagrande, 1936), see Figure 3.9(a),
• simple graphical method proposed by Pacheco-Silva (1970), see Figure 3.9(b),
• and as a last resort, σ 0c can be taken as the vertical stress which corresponds to the
intersection point of the reloading and the virgin compression lines, cf. Figure 3.9(a).
It should be noticed that the in situ preconsolidation pressure may vary from that derived from
laboratory tests considering specimen disturbances due to sampling, transporting or specimen
trimming, etc. Leroueil et al. (1983a) demonstrated that the in situ preconsolidation pressure
is observed as:
σ 0c,in situ = α · σ 0c,lab (3.2)
where α = 1.1 for normally consolidated clays (OCR < 1.2), α = 1.0 for lightly consoli-
dated clays (1.2 < OCR < 2.5), and α = 0.9 for overconsolidated clays (2.5 < OCR < 4.5).
Eoed and σ ref

oed are the input variables which are used to calculate parameters M and H with
the aid of the internal ZSoilr calculator, see material interface for nonlinear characteristics
of the HS model.
Clearly, the tangent oedometric modulus Eoed can also be determined from:
2.3(1 + e0 )
Eoed = σv (3.3)
Cc
Note that the above definition is correct when the stress point is on the primary loading
curve.
In the case of incompleteness of experimental results, the input model parameters can be es-
timated using approximative parameter correlations which are provided in Section 3.2 and 3.3.
Sometimes, the compression index Cc can also be expressed through the isotropic compression
index λ which is the slope of the virgin compression line plotted in ln p0 − e axes. Since
log10 x = 0.43 ln x, one can derive:
Cc = 2.3λ (3.4)
A number of correlations for estimating Cc are provided in Appendix B.

3M ∗

3 − sin φ
then φ = arcsin c = c∗ (3.5)
6 + M∗ 6 cos φ
Figure 3.1: Determination of the residual Mohr-Coulomb envelope and strength parameters φ and
c from typical stress paths derived from the triaxial drained compression tests driven at
three different confining pressures σ 3 .
Figure 3.2: Compatibility of strength envelopes derived from drained and undrained triaxial tests
(from Kempfert, 2006).

IDENTIFICATION
Figure 3.3: Determination of the slope a and b for identification of the secant modulus E50 and
failure ratio Rf from typical triaxial drained compression results ε1 − q. Best precision
of the interpreted parameters is obtained by plotting the trendline for two closest data
points adjacent to 0.5qf .
Figure 3.4: Determination of E moduli (input model parameters) from a typical curve derived from
the triaxial drained compression tests.

Identification algorithm:
(i) (i)
1. Find three values of E50 corresponding to σ 3 respectively.
2. Find a trend line y = ax + b by assigning variables
(i)
y as ln E
50 (i)and
σ + c cotφ
x as ln
σ ref + c cotφ
and assuming σ ref (typically equal to 100kPa)
3. Then the determined slope of the trend line a is the parameter m.
Figure 3.5: Determination of the stiffness stress dependency parameter m from three curves derived
from the triaxial drained compression tests.

IDENTIFICATION
εv

d |d|
ψ = − arcsin = arcsin
2−d 2 + |d|
ε1
Figure 3.6: Determination of the dilatancy angle ψ from εv − ε1 curve obtained in the triaxial
drained compression test (note that d is negative in the classical notation used in soil
mechanics).
1 + 2K0NC ref
∗
p = σ oed and q ∗ = (1 − K0NC )σ ref
oed
3
(a) (b)
Figure 3.7: Assumptions to the automatic determination of parameters M and H: at given σ ref oed
which is located at post-yield plastic curve, both shear an volumetric mechanism are
active.

Figure 3.8: Determination of the compression index Cc from typical results derived from oedometer
test for estimating the tangent modulus Eoed .
(a) (b)
Figure 3.9: Estimation of preconsolidation pressure σ 0c (a) Casagrande’s method, (b) Pacheco Silva’s
method.

IDENTIFICATION
3.1.2 Direct parameter identification for the HS-small

Direct parameter identification for the HS-small and HS-Brick models requires the mea-
surements derived from geophysical tests or one of the advanced in situ probes equipped
with a seismic sensor which allows measuring the shear wave velocity Vs in the subsoil.
Two commonly known devices, i.e. the seismic piezocone (SCPT or SCPTU) or seismic
Marchetti’s dilatometer (SDMT), can be used to determine small strain stiffness G0 (or
Gmax ) from the following expression:
G0 = ρVs2 (3.6)
where ρ is a density of soil. Typical ranges for Vs in different types of soils, as well as a variety
of methods that can be used to estimate Vs from in situ tests (SPT, DMT, CPT, PMT) are
given in Appendix C.
Note that in natural conditions G0 is stress dependent and, in the HS-small/HS-Brick model,
this parameter is defined by analogy to other stiffness moduli using one of the selected
expressions:
∗ m
ref σ 3 + c cotφ
G0 = G0 (3.7)
σ ref + c cotφ
∗ m
ref σ3
G0 = G0 (3.8)
σ ref
∗ m
ref p
G0 = G0 (3.9)
σ ref
Having determined G0 , the parameter E0 which is defined by the user in the material dialog,
can be calculated from:
E0 = 2(1 + ν ur )G0 (3.10)
assuming that Poisson’s coefficient ν ur is a constant in the model.
Soil stiffness at very small strains can also be approximated based on the initial part of the
ε1 − q curve or the unloading-reloading branch derived from the triaxial compression test,
as demonstrated on Figure 3.4. However, an exact determination of the initial soil stiffness
Ei may prove to be difficult, especially in soft soils. Therefore, one should realize that the
initial slope Ei derived from triaxial test can be more than once lower than soil stiffness E0
observed in natural conditions.
Identification of the parameter γ 0.7 at which the secant shear modulus Gref s is reduced to
ref
0.722G0 , requires the use of advanced laboratory devices in order to determine the S-shape
curve at very small strain levels. In practice, it may prove to be time-consuming and expensive
and therefore, it is suggested to estimate γ 0.7 by means of typically observed experimental
curves. In the case of granular materials γ 0.7 mostly depends on the mean effective stress
p0 (see Figure 3.19 ) but also on overconsolidation. In cohesive materials, γ 0.7 may mostly
depend on the plasticity index IP (PI) (see Figures 3.38 or 3.40), however the stress level
(p0 ) and the overconsolidation (OCR) may also increase the value of γ 0.7 . Having assumed
all other model parameters, it is also recommended to run a one-element simulation of the
triaxial compression test in order to examine the shape of log(ε1 ) − G (or E) curve derived
from the computed ε1 − q results.

3.1.3 Parameter identification sequence

The following sequence should be followed during parameter identification for the HS model:
1. Identify φ0 and c0 (e.g. Figure 3.1)
2. Identify ψ (e.g. Figure 3.6)

(i) (i)
3. Identify Rf and E50 for different confining pressures σ 3 (e.g. Figure 3.3)
(i) (i)
4. Identify Eur for different confining pressures σ 3 (the value of ν ur can be assigned between
0.1 and 0.2)
(i) (i)
5. Identify m based on the identified values of E50 or/and Eur (e.g. Figure 3.5)
6. Assign the reference stress σ ref (it can be the confining pressure in the triaxial test that
best corresponds to in situ stress conditions)
7. Evaluate E0 and γ 0.7

ref
8. Calculate E0ref , Eur ref
, E50 in terms of σ ref (note that by applying the stress stiffness depen-
dency law such as Eq.(2.4) the stiffness moduli also depend on φ, c and m)
9. Double-check the following relationships:
• E0ref > Eur

ref ref
> E50
ref ref
• Eur /E50 >2
ref
• 3.6 < E0ref /E50 ref
< 30 (typically 6 < E0ref /E50 < 14)
10. Evaluate K0NC and Eoed

ref
for the corresponding vertical reference stress σ ref
oed in order to
compute M and H (note that M and H should be recomputed if one of the following
parameters has been changed: Eur , ν ur , m, E50 , φ, ψ, c, Rf )
11. Evaluate the profile of soil preconsolidation in order to specify a constant OCR or a variable
OCR profile by means of q POP
12. Set K0SR = K0NC
13. Evaluate the in situ stress state in order to specify K0 (for example accounting for precon-
solidation state Eq.(3.89))
An example of parameter identification using a spreadsheet is given in Table 5.8.

IDENTIFICATION
3.1.4 Model parameters for ”undrained” simulations

In ZSoilr , the undrained behavior can be obtained in the two-phase analysis (Deformation+flow)
by running on of the following drivers:
• Consolidation driver (undrained or partially-drained conditions can be obtained depend-

ing on the action time and soil permeability 5 )
• Driven Load (Undrained) (perfectly undrained conditions)
Since the constitutive models are formulated in effective stresses, it is recommended to use
the effective stress parameters:
0
• Effective stiffness parameters E00 , Eur 0
, E50 , ν 0ur permeability)
• Effective strength parameters φ0 , c0
The main advantages of working with the effective parameters are as follows:
• deviatoric stress at failure (that corresponding to the undrained shear strength qfu = 2Su )
depends on preconsolidation history (cf. Figure 3.45)
• undrained shear strength is stress dependent
Another approach which can be considered for simulating the undrained behavior is that
relying on ”undrained” strength parameters, i.e. φ = φu = 0, c = Su , ν ur = 0.4999.
Note, however, that for this approach the undrained shear strength is constant and stress
independent.
In such approach, the following parameters should be considered:
• Effective strength parameters φ = φu = 0, c = Su (ψ = 0 for normally- and lightly cohesive

soils in order to avoid excessive gain in material resistance after reaching the failure stress
point)
0
• Effective stiffness parameters E00 , Eur ud
, E50 , ν ur = 0.4999
• High OCR, e.g. 1000 to skip the cap mechanism in computations (no plastic volumetric
deformations)
The following sequence should be respected when setting the ”undrained” parameters:
1. Insert the effective parameters E00 , Eur

0
in Elastic dialog window and the ”undrained”
Poisson’s coefficient ν ur = 0.4999.
5
In order to obtain perfectly undrained conditions for consolidation driver use very short time steps. For
real problems this may require two-phase stabilization to be active.

In the Nonlinear dialog window, disable

Automatic evaluation of H and M parameters in order to avoid unfeasible calcula-
tion of M and H parameters for the null friction angle.
2. Set ”undrained” parameters φu = 0o , c = Su , and ψ = 0 for normally- and lightly cohesive

soils in order to avoid excessive gain in material resistance after reaching the failure stress
point (cf. Truty and Obrzud (2015)).
3. Set high OCR, e.g.1000, in order to skip the cap mechanism during analysis
u
4. Set input E50 equal to undrained one E50 . The ”undrained” secant modulus can be com-
puted from:
0
u 3E50
E50 = (3.11)
2(1 + ν)
where ν should correspond to the ratio ε3 /ε1 obtained for plastic straining and the efective
0
secant modulus E50 , i.e. ν ≈ 0.3.

3.2. ALTERNATIVE PARAMETER ESTIMATION FOR GRANULAR MATERIALS
3.2 Alternative parameter estimation for granular materials
3.2.1 Initial stiffness modulus and small strain threshold

The present section provides a number of approaches for estimating the initial soil stiffness
and the small strain threshold γ 0.7 . Some correlations allows to directly approximate the
input parameter E0 , the others provide solutions for estimating the initial shear modulus G0 .
Then the input parameter E0 can be obtained through:
E0 = 2(1 + ν ur )G0 (3.12)
assuming a constant value of the unloading-reloading Poisson’s coefficient ν ur .
Geotechnical evidence. Experimental data shows that the initial stiffness of soils may
depend on the stress level, soil porosity and overconsolidation. These factors can be taken
into account using a modified equation proposed by Hardin and Black (1969):
0 m
k p
G0 = A · f (e) · OCR , in [MPa] (3.13)
pref
where G0 is the maximum small-strain shear modulus in MPa, p0 is the mean effective stress
in kPa, pref is the reference stress equal to the atmospheric pressure pref = 100 kPa, OCR
is the overconsolidation ratio and A, f (e), k, m are the correlated functions and parameters
which are given in Table 3.1 and 3.2 for different types of soils. It is observed that the
empirical exponent k varies from 0 for sands and 0.5 for high plasticity clays.
Biarez and Hicher (1994) proposed a simple relationship for all soils with wL < 50%:
0.5
p0

140
E0 = , in [MPa] (3.14)
e pref

Table 3.1: Parameters for estimation of G0 in different types of granular soils using Eq.(3.13).
Soil tested D50 Cu A f (e) k m Reference

[mm] [-] [-] [-] [-] [-]
Kenya carbonate sand 0.13 1.86 101-129 e−0.8 0 0.45-0.52
Fioravante (2000)
Quiou carbonate sand 0.75 4.40 71 e−1.3 0 0.62
Lo Presti et al. (1993)
(2.17 − e)2
Ottawa sand No.20-30 0.72 1.20 69 0 0.50
1+e Hardin and Richart Jr (1963)
(2.17 − e)2
SLB sand (subround) 0.62 1.11 82-130 0 0.44-0.53
1+e Hoque and Tatsuoka (2004)
(2.17 − e)2
Toyoura sand (subangu- 0.16 1.46 71-87 0 0.41-0.51
lar) 1+e Hoque and Tatsuoka (2004)
(2.17 − e)2
Toyoura sand (subangu- 0.19 1.56 84-104 0 0.50-0.57
lar) 1+e Chaudary et al. (2004)
Toyoura sand (subangu- 0.22 1.35 72 e−1.3 0 0.45
lar) Lo Presti et al. (1993)
(2.17 − e)2
Ticino sand (subangu- 0.50 1.33 61-64 0 0.44-0.53
lar) 1+e Hoque and Tatsuoka (2004)
(2.27 − e)2
Ticino sand (subangu- 0.54 1.50 71 0 0.43
lar) 1+e Lo Presti et al. (1993)
Ticino sand (subangu- 0.55 1.66 79-90 e−0.8 0 0.43-0.48
lar) Fioravante (2000)
(2.17 − e)2
Ham River sand (suban- 0.27 1.67 72-81 0 0.5-0.52
gular) 1+e Kuwano and Jardine (2002)
(2.17 − e)2
Silica sand (subangular) 0.20 1.10 80 0 0.5
1+e Kallioglou et al. (2003)
(2.17 − e)2
Hostun sand (angular) 0.31 1.94 80 0 0.47
1+e Hoque and Tatsuoka (2000)
(2.17 − e)2
Silica sand (angular) 0.32 2.80 48 0 0.50
1+e Kallioglou et al. (2003)
(1.46 − e)2
Silica sand 0.55 1.80 275 0 0.42
1+e Wichtmann and Triantafyllidis
(2004)
(2.17 − e)2
Hime gravel (subround) 1.73 1.33 53-94 0 0.45-0.51
1+e Chaudary et al. (2004)
(2.17 − e)2
Chiba gravel (subround) 7.90 10 76 0 0.50
1+e Modoni et al. (1999)
Table 3.2: Parameters for estimation of G0 in granular soils using Eq.(3.13).
Soil tested emin emax A f (e) m Reference

[-] [-] [-] [-] [-]
(2.17 − e)2
Clean uniform sands 0.5 1.1 57 0.4 Iwasaki and Tatsuoka (1977)
1+e
with Cu < 1.8
1
All soils with wL < 0.4 1.8 58∗ 0.5 Biarez and Hicher (1994)
e
50%
(2.97 − e)2
Undisturbed crushed 0.6 1.5 33 0.5 Hardin and Black (1969)
1+e
sands
∗ obtained from Eq. 3.14 assuming ν = 0.2

Figure 3.10: Graphical representation of empirical relations presented in Table 3.6 and 3.18 (after
Benz, 2007).

Geotechnical evidence. In case of lack of test data at very small strain levels, E0 can be
evaluated from an empirical relation proposed by Alpan (1970). This relation which is pre-
sented in Figure 3.11, relates so-called ”static” modulus Es to the ”dynamic” modulus Ed .
For the sake of HS-SmallStrain model, Es can be considered as Eur obtained at engineering
strain levels (ε ≈ 10−3 ), whereas E0 can be considered as ≈ Ed .
100
Ed
Ro
Es ck
s
co h
es i
10 ve
so i
ls
gran
u l ar
soil
s
Es [kPa]
1
1000 10000 100000 1000000
Figure 3.11: Approximative relation between ”static” soil stiffness (here Es ≈ Eur ) and ”dynamic”
modulus Ed corresponding to E0 proposed by Alpan (1970).
Geotechnical evidence. It can be observed in laboratory test that secant stiffness reduces
with mobilization of the shear strength. Mayne (2007) provides a selection of secant modulus
curves, represented by the ratio Gs /G0 or Es /E0 . The collected results were derived from
monotonic laboratory shear tests performed on an sorted mix of clayey ans sandy materials,
and they are presented in Figure 3.13. Such experimental results can be approximated with
a hyperbolic model by Fahey and Carter (1993) (see Figure 3.12):
g
Gs q
=1−F (3.15)
G0 qmax
where f and g are soil-specific model parameters (typically 0.8 < f < 1.0).
Experimental results reported in Lo Presti et al. (1998) show that g increases (E0 /E50
decreases) with overconsolidation, especially for quartz and calcerous sands. It has also been
recognized that E0 /E50 increases with soil cementation and structurization. Therefore, higher
values of E0 /E50 can be expected for sensitive clays.
Experimental results and Eq. (3.15) have been used in this report to develop a method which
provides reasonable first-guess values of E0 based on a known value of E50 . Considering
that q/qmax corresponding to E50 is equal to 0.5, and assuming that F = 1, the hyperbolic
equation can be used to find lower and upper limits of E50 /E0 by adjusting g parameter. The
adjustment of g with respect to overconsolidation or the consolidation stress ratio Kc (which
for natural soils increases with overconsolidation) was carried out using the experimental
results for Toyura sand presented in Figure 3.14. A summary of adjusted g values and
corresponding E0 /E50 ratios with respect to overconsolidation is provided in Table 3.3.

Figure 3.12: Reduction of G0 /G50 by the hyperbolic model - Equation (3.15) (assuming F = 1).
q/qmax = 0.5
Figure 3.13: Observed secant stiffness modulus reduction curves from static torsional and triaxial
shear data on clays and sands (from Mayne, 2007) and the superposed hyperbolic
curves obtained for different g values. The adjusted g values provides reasonable first-
guess lower and upper limits for estimating E0 from a determined E50 value (see Table
3.3).

q/qmax = 0.5
Figure 3.14: Normalized secant modulus of Toyura vs mobilized shear strength (Lo Presti et al.,
1998, from) and adjusting of g parameter with respect to stress consolidation ratio.
Table 3.3: Typical values of E0 /E50 ratio for granular soils with respect to the preconsolidation
state (this table has been developed based on literature review and Fahey’s stiffness
reduction model).
OCR [–] E0 /E50 [–] Corresponding g [–]

Degree of preconsolidation ”Min” ”Max” For ”Min” For ”Max”
Unknown 4.1 18.0 0.40 0.083
Normally consolidated 1 ÷ 1.2 6.3 18.0 0.25 0.083
Lightly overconsolidated 1.2 ÷ 2.5 4.6 12.5 0.35 0.12
Overconsolidated 2.5 ÷ 4.5 3.9 8.1 0.43 0.19
Heavily overconsolidated 4.5 ÷ 10 3.6 5.5 0.47 0.29
The higher values of E0 /E50 ratio are suggested for aged, cemented and structured sands.

Figure 3.15: Cone resistance vs. maximal shear modulus G0 for sands (after Robertson and Cam-
panella, 1983).
CPT. Initial small strain stiffness for sands can be approximated from cone resistance mea-
surements qc derived from CPT. Robertson and Campanella (1983) related the maximal shear
modulus G0 with qc for different effective vertical stresses σ 0v0 , as presented in Figure 3.15.
CPT. Based on calibration chamber and field measurements Rix and Stokoe (1992) proposed
the correlation for uncemented quartz sands (cf. Figure 3.16). The wide range of G0 /qc at
low values of normalized cone resistance is explained by variations in soil compressibility; more
compressible sands may give lower values of Qt and hence higher values of G0 /qc (Lunne
et al., 1997).
!−0.75
G0 qc
= 1634 p 0
qc avg σ v0
(3.16)
Average
Range = Average ±
2
with G0 , qc and σ 0v0 in kPa.
DMT. In general, G0 in sands can be estimated based on dilatometer modulus ED using the
correlations obtained based on calibration chamber tests by Baldi et al. (1986) and on field
tests by Belotti et al. (1986):
G0
= 2.72 ± 0.59 (3.17a)
ED
G0
= 2.20 ± 0.7 (3.17b)
ED
The relations are graphically presented in Figure 3.17.

Figure 3.16: Comparison of empirical correlations for estimating G0 (Gmax ) in sands (Eq.(3.16))
and clays (Eq.(3.47)) from CPT data (correlations interpreted for typical qt ranges).
Figure 3.17: Comparison of empirical correlations for estimating G0 (Gmax ) in sands from DMT
data (Eq.(3.17a) and Eq.(3.17b).

SPT. G0 can be estimated for sands from the correlation proposed by Ohta and Goto (1976,
as referred in Kramer (1996)), cf. Figure 3.18:
0.5
p0

0.3333
G0 = 438N1,60 pa (3.18)
pa
with G0 and p0 -mean effective stress in kPan and pa = 100kPa, N1,60 - ”overburden-
corrected” N60 -value (cf. Table 3.10).
Another correlation for sand based on SPT date was proposed by Imai and Tonouchi (1982,
as refered in Kramer (1996)), cf. Figure 3.18:
0.68
G0 = 15600N60 (3.19)
(G0 in kPa.)
Figure 3.18: Comparison of empirical correlations for estimating G0 (Gmax ) in sands from SPT
data, Eq.(3.18) plotted for K0 = 0.4 and Eq.(3.19).

Geotechnical evidence. It has been proved experimentally that the strain threshold γ 0.7
does not depend on soil density in the case of non-cohesive granular soils (cf. 3.19). On the
other hand, γ 0.7 can be affected by the magnitude of the confining pressure σ 00 which corre-
sponds to the mean effective stress p0 for in situ conditions (cf. Darendeli and Stokoe (2001);
Wichtmann and Triantafyllidis (2004)). Hence, the parameter γ 0.7 can be approximated from
a diagram presented in Figure 3.19. Note that the model formulation does not account for
stress dependency of γ 0.7 . If needed this parameter can be incorporated into boundary value
problems through definition average mean effective stress for defined sub-layers.
Figure 3.19: Influence of relative density ID and the confining pressure p0 on strain threshold γ 0.7
for sands (from Wichtmann and Triantafyllidis, 2004).
Experimental evidence for sands reported in Darendeli and Stokoe (2001), Figure 3.20, allows
to write the following approximating relationship:
0 0.35
ref p
γ 0.7 = γ 0.7
pa
with (3.20)
−4
γ ref
0.7 (pa ) = 1.26 · 10 - reference strain threshold at pa
pa = 1atm ≈ 100kPa - atmospheric pressure
An estimation of γ 0.7 for granular soils can also be carried out using a linear interpolation
which is obtained through interpretation of the results presented in Figure 3.19:
p0
γ 0.7 = 8.75 · 10−5 + γ ref
0.7 for p0 ≤< 400kPa
pref
with (3.21)
= 1.0 · 10−4

γ ref
0.7 p
ref
- reference strain threshold at p ref
pref = 100kPa - reference pressure
where pref = 100kPa.

A fit of the above correlation to the experimental curves is presented in Figure 3.21.

Eq.(3.20)
Figure 3.20: Comparison of predictions for γ 0.7 at different mean stresses p0 using Eq.(3.20) for the
data reported in Darendeli and Stokoe (2001).
Eq.(3.21)
Figure 3.21: Comparison of predictions for γ 0.7 at different mean stresses p0 using Eq.(3.21) with
experimental data for sands reported in Wichtmann and Triantafyllidis (2004).

In the case of granular soils, it is also observed that the strain threshold may be affected
by OCR and for sands with high content of fines additionally on IP . In order to account
these effects, the correlation presented in Eq.(3.54) can be used (this generalized formula
by Darendeli (2001) was developed based on a database containing four groups of soils, i.e.
”clean” sands, sands with high content of fines, silts and clays.

3.2.2 Secant and unloading-reloading moduli

In general, the relation between the stiffness moduli is as follows:
E50 < Es < Eur (3.22)
where Es denotes ”static” modulus or secant modulus taken from the initial part of the ε1 −q
experimental curve at ε1 = 0.1%.
In the case when E50 or Eur cannot be directly determined from experimental curves, it may
be relevant for many practical cases to set:
ref
Eur
ref
= 2 to 6 (an average can be assumed equal to 3) (3.23)
E50
Higher ratios can be assumed for loose sands (3 to 6), whereas lower ones for dense sands
E ref
(2 to 4). However, note that the following condition should be satisfied: urref
> 2.
E50
E ref
However, note that the following condition should be satisfied: ur
ref
> 2.
E50
In absence of laboratory results, the stiffness moduli can be approximated based on typically
observed order of magnitudes of ”static” modulus Es which are given in Table 3.4. Assuming
that soil behavior follows the stress-strain relation described by Equation 2.10 and assuming
ε1 = 0.1%, the ”static” modulus can be represented with:
q 1
Es = = (3.24)
0.001 1 0.001 · Rf
+
2E50 qf (φ, c)
The above equation can be represented graphically in Figure 3.22, and can be used to estimate
E50 from the known value of Es .
Table 3.4: Typical values for the ”static” modulus Es [MPa] (compiled from Kezdi, 1974; Prat
et al., 1995, and extended by the authors).
Soil Density
Very loose Loose Medium Dense Very dense
Soil Type Min Max Min Max Min Max Min Max Min Max
Gravels/Sand well-graded 10 30 30 80 80 160 160 320 320 480
Sand, uniform 5 10 10 10 30 30 50 50 80 120
Sand/Gravel silty 3 7 7 12 12 20 20 30 30 40
Sand/Gravel clayey 2 6 6 10 10 15 15 21 21 29

Figure 3.22: Estimating the ratio between the static modulus Es and secant modulus E50 .
For sands, the secant modulus E50 can be estimated based on the known porosity, Figure
3.23.
Figure 3.23: Normalized secant modulus E50 vs. porosity n0 for different sands (after Schanz and
Vermeer, 1998).

CPT. Secant modulus E50 for sands can be approximated from cone resistance measurements
qc derived from CPT. Robertson and Campanella (1983) related E50 with qc for different
effective vertical stresses σ 0v0 , as presented in Figure 3.24.
Figure 3.24: Cone resistance qc vs. secant E50 modulus for sands (after Robertson and Campanella,
1983).
DMT. The ”static” modulus corresponding to Eur can be evaluated for silty sands and sand
from the vertical drained constrained modulus MDMT which is derived from three intermediate
dilatometer parameters, i.e. the material index ID , the horizontal stress index KD , and
the dilatometer modulus ED . Note that MDMT may correspond to Eoed only for normally-
consolidated soil. The unloading-reloading modulus Eur can be evaluated assuming that:
Eur = Es = aEoed = aMDMT (3.25)

where a ≈ 0.9 as a = (1 + ν ur )(1 − 2ν ur )/(1 − ν ur ) and the dilatometer vertical drained
constrained modulus MDMT :
MDMT = RM ED (3.26)
with ED = 34.7(p1 − p0 ) and RM depending on soil type behavior, i.e.
• silty sand (1.8 < ID < 3): RM = RM,0 + (2.5 − RM,0 ) log KD and
RM,0 = 0.14 + 0.15(ID − 0.6)
• sand (ID ≥ 3): RM = 0.5 + 2 log KD
• if KD > 10: RM = 0.32 + 2.18 log KD
• if RM < 0.85: set RM = 0.85

SPT. Table 3.5 gives some empirical equations which can be used to evaluate the ”static”
modulus based on SPT N -value. A graphical interpretation of correlation performance and
a comparison with typical values for the ”static” modulus Es is presented in Figure 3.25.
Table 3.5: Empirical correlations relating the ”static” modulus Es with SPT N -value.
Soil Type Empirical formula Remarks Reference

Es in [kPa]
Gravels/Sands 6000N55 can be considered as a good
well-graded approximation for Eur in well- Bowles (1997)
graded sands and gravels
Gravelly sand 1200(N60 + 6) + 4000 seems to give too low values
for in for Eur in gravelly sand Begemann
(1974)
600(N55 + 6) seems to gives too
Bowles (1997)
+2000 if N > 15 low values for Eur in gravelly
sands
Sand (3.5 to 5.0) × 104 log(N60 ) perceived as a good approx-
imation of lower and upper Tromienkov
bound for Eur in uniform sands (1974)
Sand NC 2750N55 can be considered as a good
approximation for Eur in loose Bowles (1997)
√ to medium NC sands
7000 N55 can be considered as a lower-
bound approximation for Eur Bowles (1997)
in NC uniform sands
780N60 + 20000
Sand NC or sand can be considered as a good authors’ equa-
1 − ν2
& gravel approximation for Eur in uni- tion from plot
form NC sands of D’Appolonia
et al. (1970)
1100N60 + 40000
Sand OC can be considered as a good authors’ equa-
1 − ν2
approximation for Eur in uni- tion from plot
form OC sands of D’Appolonia
et al. (1970)
Sand saturated 500(N60 + 15) gives too low values for Eur
Webb (1969)
Fine sands and 100(44N60 )0.75 ± 5000 can be considered as the
silty sands lower-bound approximation Schultze and
for Eur in silty sands Menzenbach
(1961)
Clayey sand 333(N60 + 15) can be considered as the
lower-bound approximation Webb (1969)
for Eur in clayey sands
N60 corresponds to the energy ratio Er = 60. Since the energy × blow count
should be a constant for any soil, the following equation can be applied
Er1 × N1 = Er2 × N2 (Bowles, 1997). For example, N55 = N60 × 60/55.

Figure 3.25: Performance of empirical correlations from Table 3.5 compared with typical values for
the ”static” modulus Es from Table 3.4.

3.2.3 Oedometric modulus

In case of lack of oedometric test data for granular material the oedometric modulus can
approximately be taken as:
ref ∼ ref
Eoed = E50 (3.27)
In such a case, the oedometric vertical reference stress σ ref
oed should be matched to the reference
minor stress σ ref since the latter typically corresponds to the confining (horizontal) pressure
σ ref = σ 3 = σ 0h :
σ ref NC
oed = σ ref /K0 (3.28)
On the other hand, when defining σ ref
oed = σ ref in the model, the following relationship should
be taken: m
ref ∼ ref
Eoed = E50 K0NC (3.29)
Figure 3.26: Comparison of reference stiffness moduli for sands from oedometer and triaxial tests
(after Schanz and Vermeer, 1998).
Figure 3.27: Normalized stiffness modulus of various sands derived from oedometer tests (after
Schanz and Vermeer, 1998).

DMT. The constrained tangent modulus MD (corresponding to Eoed in oedometer test) can
be interpreted from three intermediate DMT parameters, i.e. the material index ID , the
horizontal stress index KD , and the dilatometer modulus ED , by applying the correlation
presented in Eq.(3.26).
ref
Important note. In the case of the HS model, Eoed can be taken as equal to MDMT if the
latter has been derived from DMT but only for normally-consolidated soil. In such a case
σ oed can be taken as σ 0v0 which corresponds to the testing depth for which MDMT has been
evaluated.

3.2.4 Unloading-reloading Poisson’s ratio

Experimental measurements from local strain gauges show that the initial values of Pois-
son’s ratio in terms of small mobilized stress levels (q/qmax ) varies between 0.1 and 0.2
for clays, sands and rocks (Figure 3.28). Therefore, the characteristic value for the elastic
unloading/reloading Poisson’s ratio of ν ur = 0.2 can be adopted for most soils.
Figure 3.28: Poisson’s ratio ν vs. mobilized stress level derived from local strain measurements on
sand, clay and soft rock (after Mayne et al., 2009).

3.2.5 Stiffness exponent

Geotechnical evidence. In natural soil, the exponent m varies between 0.3 and 1.0. Janbu
(1963) reported values of 0.5 for Norwegian sands ans silts. Typical values for m obtained in
clean sands and gravels are provided in Table 3.6.
Figure 3.29: Typical values for m obtained for sands from triaxial test vs. initial porosity n0 (from
Schanz and Vermeer, 1998).
Figure 3.30: Typical values for m obtained for sands from oedometric test vs. initial porosity n0
(after Schanz and Vermeer, 1998).

Table 3.6: Typical values for m observed in clean sands and gravels for the shear modulus G0 (from
Benz, 2007).
Soil tested m [-] Reference

Kenya carbonate sand 0.45-0.52 Fioravante (2000)
Quiou carbonate sand 0.62 Lo Presti et al. (1993)
Ottawa sand No.20-30 0.50 Hardin and Richart Jr (1963)
SLB sand (subround) 0.44-0.53 Hoque and Tatsuoka (2004)
Toyoura sand (subangular) 0.41-0.51 Hoque and Tatsuoka (2004)
Toyoura sand (subangular) 0.50-0.57 Chaudary et al. (2004)
Toyoura sand (subangular) 0.45 Lo Presti et al. (1993)
Ticino sand (subangular) 0.44-0.53 Hoque and Tatsuoka (2004)
Ticino sand (subangular) 0.43 Lo Presti and Jamiolkowski (1998)
Ticino sand (subangular) 0.43-0.48 Fioravante (2000)
H.River sand (subangular) 0.5-0.52 Kuwano and Jardine (2002)
Silica sand (subangular) 0.5 Kallioglou et al. (2003)
Hostun sand (angular) 0.47 Hoque and Tatsuoka (2000)
Silica sand (angular) 0.50 Kallioglou et al. (2003)
Silica sand 0.42 Wichtmann and Triantafyllidis (2004)
Hime gravel (subround) 0.45-0.51 Chaudary et al. (2004)
Chiba gravel (subround) 0.50 Modoni et al. (1999)
Table 3.7: Suggested ranges of stiffness exponent m observed for oedometric modulus Eoed (von
Soos, 1991).
Soil type m [-]

Gravel: poorly-graded (uniform) 0.4 ÷ 0.6
Gravel: sandy, well-graded 0.5 ÷ 0.7
Gravel: silty or clayey, well-graded, not crushed 0.5 ÷ 0.7
Gravel-sand-clay mixture, crushed 0.7 ÷ 0.9
Sand: fine, uniform 0.6 ÷ 0.75
Sand: coarse, uniform 0.55 ÷ 0.7
Sand: well-graded and gravelly sand 0.55 ÷ 0.7
Sand: with fines, not crushed 0.65 ÷ 0.8
Sand: with fines, crushed 0.75 ÷ 0.9

3.2.6 Friction angle

In coarse soils, the value of friction angle is mostly influenced by soil density, shape of particles
and soil gradation. Friction angle can be approximated from Table 3.8 or estimated based on
in situ test results from SPT, CPT or DMT.
Brinch Hansen and Lundgren (1958) proposed to estimate the friction angle with the following
empirical correlation:
φ = 36o + φ1 + φ2 + φ3 + φ4 (3.30)
using the following corrections to account for different soil features:
Soil type Relative density

o
• gravel φ1 = +2 • very loose φ2 = −6o
• gravel+sand φ1 = +1o • loose φ2 = −3o
• sand φ1 = 0o • medium φ2 =0
• dense φ2 = +3o
• very dense φ2 = +6o
Soil gradation Particles shape
o
• well-graded φ3 = +3 • angular φ2 = +1o
• medium φ3 = 0o • subangular φ2 = 0o
• poorly-graded φ3 = −3o • subrounded φ2 = −3o
• rounded φ2 = −5o
Table 3.8: Empirical values for φ and Dr of granular soils based on SPT at about 6 m depth and
normally consolidated (after Bowles, 1997, and modified by the authors).
Description Very loose Loose Medium Dense Very Dense

Relative density Dr 0 0.15 0.35 0.65 0.85
SPT N70 :
fine 1-2 3-6 7-15 16-30
medium 2-3 4-7 8-20 21-40 > 40
coarse 3-6 5-9 10-25 26-45 > 45
SPT φ
fine 26-28 28-30 30-34 33-38
medium 27-28 30-32 32-36 36-42 < 50
coarse 28-30 30-34 33-40 40-50
SPT. The friction angle for granular soils with a small content of fine grains can be determined
using the chart suggested by Peck et al. (1974). This chart also correlates the SPT number
with the bearing factors Nγ and Nq which are standardly used for dimensioning of foundations.

Loose
Soil density
loose
Very
Compact Dense Very dense
Number of blows per 30 cm

Bearing factors Nγ , Nq
Friction angle [o ]
Figure 3.31: Determination of the friction angle φ0 and bearing factors for granular soils based on
the SPT number (from Peck et al., 1974).
Table 3.9: Estimation of the friction angle φ0 from the SPT number.
Standard Penetration Resistance Friction angle, φ [o ]

Soil type N60 , blows/0.3m Peck et al. (1974) Meyerhof (1956)
Very loose sand <4 < 29 < 30
Loose sand 4-10 29-30 30-35
Medium sand 10-30 30-36 35-40
Dense sand 30-50 36-41 40-45
Very dense sand > 50 > 41 > 45

Table 3.10: Empirical correlations relating SPT N -value with the effective friction angle for sands.
Empirical formula Reference Remarks

φ0peak = 15.4N1,60 + 20o tends to overestimate φ0 in
p
Hatanaka and Uchida (1996) very dense sands σ 0v0 <
100kPa
√
φ0peak = 15.4N60 + 20o (±3o ) simplification: σ 0v0 = pa
Hatanaka and Uchida (1996) with-
out overburden correction
φ0 = 20N1,60 + 15o tends to overestimate φ0 in

p
Teixeira (1996) very dense sands σ 0v0 <
100kPa
 0.34
N60
φ0 = tan−1  tends to overestimate φ0 in
 

σ 0v0 Schmertmann (1975) very dense sands σ 0v0 <
 
12.2 + 20.3
pa 100kPa
0.34
0 −1 N60
φ = tan simplification: σ 0v0 = pa
32.5 Schmertmann (1975) without
overburden correction
s
N55
φ0 = 25o + 28 after Townsend et al. (2003) tends to overestimate φ0 in
σ 0v0
very dense sands σ 0v0 <
100kPa
√
φ0 = 15o + 18N70 Japan Road Association
Shioi and Fukui (1982) standards
φ0 = 27o + 0.36N70 Japanese National Railway

Shioi and Fukui (1982) standards
N1,60 = N60 CN (”overburden-corrected” N60 -value)

0.5
with the overburden correction factor CN = (pa /σ 0v0 ) (Liao and Whitman, 1986)
use CN = 1.7 if CN > 1.7
pa - atmospheric pressure (average sea-level pressure is 101.325 kPa)

Figure 3.32: Cone resistance vs. peak friction angle φ0 for sands (after Robertson and Campanella,
1983).
CPT. The most widely accepted relationship which relates the cone resistance qt with φ0 for
granular materials is the expression proposed by Robertson and Campanella (1983) (Figure
3.32):
0 qt
φ = arctan 0.10 + 0.38 log (3.31)
σ 0v0
DMT. Two direct empirical correlations suggested in Totani et al. (1999) can be used to
estimate lower and upper bounds of the range of the friction angle:
φ0max = 31 + KD /(0.236 + 0.066KD ) (3.32)
φ0min = 28 + 14.6 log KD − 2.1(log KD )2 (3.33)

with KD denoting horizontal stress index which is calculated based on the first dilatometer
reading p0 , i.e. KD = (p0 − u0 )/σ 0v0 .
Geotechnical evidence. Typical values of the friction angle for granular soils are provided
in Tables 3.11, 3.12, 3.13 and 3.14.

Table 3.11: Representative values of φ observed in sands (after Schmertmann, 1978).
Relative Friction angle φ [o ]

density Fine Grained Medium Grained Coarse Grained
Dr [%] Uniform Well-graded Uniform Well-graded Uniform Well-graded
40 34 36 36 38 38 41
60 36 38 38 41 41 43
80 39 41 41 43 43 44
100 42 43 43 44 44 46
Table 3.12: Representative values of φ observed in cohesionless soils (after Carter and Bentley,
1991).
φ [o ]
Soil type Loose Dense
Uniform sand, round grains 27 34
Well-graded sand, angular grains 33 45
Sandy gravels 35 50
Silty sand 27-33 30-34
Inorganic silt 27-30 30-35
Table 3.13: Representative values of φ observed in compacted sands and gravels (after Carter and
Bentley, 1991).
Soil type UCS class φ [o ]

Well-graded sand-gravel mixtures GW > 38
Poorly-graded sand gravel mixtures GP > 37
Silty gravels, poorly graded sand-gravel-clay GM > 34
Clayey gravels, poorly graded sand-gravel-clay GC > 31
Well-graded clean sand, gravelly sands SW 38
Poorly-graded clean sands SP 37
Table 3.14: Representative relationships between relative density Dr and friction angle φ for granular
soils.
State of compaction Relative density Dr [%] φ [o ]

Very loose 0-15 < 25
Loose 15-35 25-30
Medium 35-65 30-37
Dense 65-85 37-43
Very dense 85-100 > 42

3.2.7 Dilatancy angle

It is typically observed in laboratory tests that for a very dense sand the value of the dilatancy
angle ψ is about 1/3 of the peak friction angle φ0 . In the case of loose sands, the dilatancy
angle reduces to a few degrees, where as normally consolidated clays may exhibit no dilatancy
at all. For example, Bolton (1986) reports ψ max = 14.7o for a dense sand which corresponded
to the peak friction angle φ0max = 44.8o derived from a drained, plane strain compression test.
Bolton (1986) proposed that for well-compacted granular soils, the maximal dilatancy angle
can be estimated from:
ψ = 3.75IR under triaxial conditions (3.34a)
ψ = 6.25IR under plain strain conditions (3.34b)

with the relative dilatancy index IR can be estimated for well-compacted granular soils from:
IR = 5Dr − 1 (0 < IR < 4) (3.35)

emax − e
with Dr denoting the relative density index = .
emax − emin
Another reasonable approximation of ψ can be obtained using the following equation:
ψ = φ − 30o (3.36)
for the values of friction angle larger than 30o .

Notice that the above approximation gives ψ ≈ φ0 /3 for a dense gravel, whereas small values
of ψ will be obtained for the the friction angles that correspond to loose sands (typically
φ ≈ 30o to 32o ).

3.2.8 Coefficient of earth pressure ”at rest”

Various relationships can be found in literature for estimation of the coefficient of earth
pressure ”at rest” for normally consolidated soils. They commonly relate the value of K0NC
to the effective friction angle φ0 , and the most popular are6 :
K0NC = 1 − sin φ0 (3.37a)

√ √
K0NC = ( 2 − sin φ0 )/( 2 + sin φ0 ) (Simpson, 1992) (3.37b)
These equations are illustrated in Fig.3.33.
Figure 3.33: Typical relationships between K0NC and φ0 observed for soils.
In the case of sands, the notion of preconsolidation pressure is not as meaningful as for co-
hesive soils, and therefore OCR = 1 (i.e. K0SR = K0NC ) can be assumed when calculating
parameters H and M .
In the case of running a simulation of isotropic consolidation (the case of isotropically

consolidated triaxial compression tests, i.e. CIU or CID), the coefficient should be assumed
as K0SR = 1.
6
Note that Eq.(3.37a) is often erroneously called ”Jaky’s equation” as it is a simplified form of his original
expression K0NC = (1 − sin φ0 )/(1 + sin φ0 )(1 + 2/3 sin φ0 ) (Jaky, 1947) which gives essentially the same
results as Eq.(3.87).

3.2.9 Void ratio

Void ratio for a saturated soil can be calculated from:
e = wn Gs (3.38)
where wn is the water content, Gs is the specific gravity of soil solids and S is the saturation
ratio.
In the case of partially unsaturated soil, the void ratio can be obtained from:
Gs γ w (1 + wn ) Gs γ w
γ= or e = − 1 or e = wn Gs /S (3.39)
1+e γd
where γ d is the dry unit weight.

Hence, the maximum void ratio emax can be calculated from:
Gs γ w
emax = −1 (3.40)
γ d,min
The maximum void ratio emax can also be estimated according to approximate relationship
presented in Figure 3.34 between the void ratio the coefficient of uniformity for different
granular soils.
Typical values of void ratios and dry unit weights observed in granular soils are provided in
Table 3.15.
Table 3.15: Typical values of void ratio and dry unit weights observed in granular soils (after Das,
2008).
Voids ratio Dry unit weight

e [–] γ [kN/m3 ]
Soil type Max Min Min Max
Gravel 0.6 0.3 16 20
Coarse sand 0.75 0.35 15 19
Fine sand 0.85 0.4 14 19
Standard Ottawa sand 0.8 0.5 14 17
Gravelly sand 0.7 0.2 15 22
Silty sand 1 0.4 13 19
Silty sand and gravel 0.85 0.15 14 23

Figure 3.34: Generalized charts for estimating emax , and emin from gradational and particle shape
characteristics (from UFC, 2004; Das, 2008).

3.2.10 Overconsolidation ratio

CPT. Estimation of OCR from CPT data in sands can be carried out using a relationship
developed based on statistical multiple regression analysis of 26 separate sands worldwide with
flexible-walled calibration chambers (Kulhawy and Mayne, 1990; Lunne et al., 1997; Mayne
et al., 2001; Mayne, 2007). Primarily clean siliceous (quartz and feldspar) sand samples where
subject to normally-consolidated and overconsolidated states 1 ≤ OCR ≤ 15. Sands in the
test were usually dry or saturated without back pressures. The obtained empirical equation
is the following:
1
 
 0.22  0

qt sin φ − 0.27
0.192
 pa 
OCR =  (3.41)
 
0 0.31 
σ v0
(1 − sin φ0 )
 
pa
with qt denoting cone tip resistance, σ 0v0 effective overburden stress, and atmospheric pressure
is pa = 100 kPa.


CPT. Estimation of K0 from CPT data in sands can be carried out using a relationship
developed based on statistical multiple regression analysis of 26 separate sands worldwide with
flexible-walled calibration chambers (Kulhawy and Mayne, 1990; Lunne et al., 1997; Mayne
et al., 2001; Mayne, 2007). Primarily clean siliceous (quartz and feldspar) sand samples where
subject to normally-consolidated and overconsolidated states 1 ≤ OCR ≤ 15. Sands in the
test were usually dry or saturated without back pressures. The obtained empirical equation
is the following: 0.22 0.31
qt pa
K0 = 0.192 OCR0.27 (3.42)
pa σ 0v0
with qt denoting cone tip resistance, OCR overconsolidation ratio, and atmospheric pressure
is pa = 100 kPa.
DMT-CPT. In 1980’s, researchers reported that a unique correlation for K0 from KD cannot
be established for sands as data points showed that such correlation in sand also depends
on φ or Dr . Initially developed Marchetti’s chart K0 − qc − KD (Marchetti, 1985) was
updated by Baldi et al. (1986) by incorporating all subsequent calibration chamber research
and converted into a simple algebraic equations:
K0 = 0.376 + 0.095KD − 0.0017qc /σ 0v0 for ”freshly” deposited sand (3.43a)
K0 = 0.376 + 0.095KD − 0.0046qc /σ v0 for ”seasoned” sand) (3.43b)

Marchetti et al. (2001) recommend to use Eq.(3.43a) in ”freshly deposited” sand, whereas
for ”seasoned” sands Eq.(3.43b).

3.3 Alternative parameter estimation for cohesive materials
3.3.1 Initial stiffness modulus and small strain threshold

The present section provides a number of approaches for estimating the initial soil stiffness
and the small strain threshold γ 0.7 . Some correlations allows to directly approximate the
input parameter E0 , the others provide solutions for estimating the initial shear modulus G0 .
Then the input parameter E0 can be obtained through:
E0 = 2(1 + ν ur )G0 (3.44)
assuming a constant value of the unloading-reloading Poisson’s coefficient ν ur .
Geotechnical evidence. Experimental data shows that the initial stiffness of soils may
depend on the stress level, soil porosity and overconsolidation. These factors can be taken
into account using a modified equation proposed by Hardin and Black (1969):
0 m
k p
G0 = A · f (e) · OCR , in [MPa] (3.45)
pref
where G0 is the maximum small-strain shear modulus in MPa, p0 is the mean effective stress
in kPa, pref is the reference stress equal to the atmospheric pressure pref = 100 kPa, OCR
is the overconsolidation ratio and A, f (e), k, m are the correlated functions and parameters
which are given in Table 3.16 and 3.18 for different types of soils. It is observed that the
empirical exponent k varies from 0 for sands and 0.5 for high plasticity clays. It means that
k may increase with soil plasticity and its value can be taken from Table 3.17.
Biarez and Hicher (1994) proposed a simple relationship for all soils with wL < 50%:
0.5
p0

140
E0 = , in [MPa] (3.46)
e pref

3.3. ALTERNATIVE PARAMETER ESTIMATION FOR COHESIVE MATERIALS
Table 3.16: Parameters for estimation of G0 in clays using Eq.(3.45).
Soil tested IP A f (e) k mp Reference

[%] [-] [-] [-] [-]
Quaternary Italian - 60 e−1.30 N/A 0.50 Jamiolkowski et al. (1995)
clays (see Fig.3.35)
Avezzano clay 10-30 74 e−1.27 N/A 0.46 Lo Presti and Jamiolkowski (1998)
(Holocene-
Pleistocene)
Fucino clay 45-75 64 e−1.52 N/A 0.40 Lo Presti and Jamiolkowski (1998)
(Holocene-
Pleistocene)
Garigliano clay 10-40 44 e−1.11 N/A 0.58 Lo Presti and Jamiolkowski (1998)
(Holocene)
Panigaglia clay 44 44 e−1.30 N/A 0.50 Lo Presti and Jamiolkowski (1998)
(Holocene)
Montaldo di Castro 15-34 50 e−1.33 N/A 0.40 Lo Presti and Jamiolkowski (1998)
clay (Pleistocene)
Reconstituted 27 44 1 N/A 0.85 Rampello et al. (1997)
Valericca clay
(Pleistocene)
Pisa clay (Pleis- 23-46 50 e−1.43 N/A 0.44 Lo Presti and Jamiolkowski (1998)
tocene)
London clay (recon- 41 13 1 0.25∗ 0.76 Viggiani and Atkinson (1995)
stituted)
Speswhite kaolin 24 40 1 0.2∗ 0.65 Viggiani and Atkinson (1995)
clay (reconstituted)
(2.97 − e)2
Kaolin clay 35 45 N/A 0.50 Marcuson and Wahls (1972)
1+e
2
(4.40 − e)
Bentonite clay 60 4.5 N/A 0.50 Marcuson and Wahls (1972)
1+e
∗
overconsolidation ratio OCR based on the mean stress (OCR = p0c /p0 )
Table 3.17: Overconsolidation ratio exponent k used in Eq.(3.45).
Plasticity Index Exponent

IP [%] k [-]
0 0.00
20 0.18
40 0.30
60 0.41
80 0.48
≥ 100 0.50
source: Kramer (1996, after Hardin&Drnevich 1972)

Figure 3.35: Typical void ratio-G0 dependency using an empirical equation by Jamiolkowski et al.
(1995) from Table 3.16.
Table 3.18: Parameters for estimation of G0 in cohesive soils using Eq.(3.45). The relationships
are illustrated in Figure 3.10.
Soil tested emin emax A f (e) mp Reference

[-] [-] [-] [-] [-]
1
All soils with wL < 0.4 1.8 58∗ 0.5 Biarez and Hicher (1994)
e
50%
(2.97 − e)2
Undisturbed clayey 0.6 1.5 33 0.5 Hardin and Black (1969)
1+e
soils
(2.97 − e)2
Undisturbed cohesive 0.6 1.5 16 0.5 Kim and Novak (1981)
1+e
soils
(7.32 − e)2
Loess 1.4 4.0 1.4 0.6 Kokusho et al. (1982)
1+e
∗ obtained from Eq. 3.14 assuming ν = 0.2

Geotechnical evidence. In case of lack of test data at very small strains, E0 for cohesive
soils can be evaluated from an empirical relation proposed by Alpan (1970), see Figure 3.11.
The chart relates so-called ”static” modulus Es to the ”dynamic” modulus Ed . For the sake
of HS-SmallStrain model, Es can be considered as Eur obtained at engineering strain levels
(ε ≈ 10−3 ), whereas E0 can be considered as ≈ Ed .
Geotechnical evidence. It can be observed in laboratory test that secant stiffness reduces
with mobilization of the shear strength. Mayne (2007) provides a selection of secant modulus
curves, represented by the ratio Gs /G0 or Es /E0 . The collected results were derived from
monotonic laboratory shear tests performed on an sorted mix of clayey ans sandy materials,
and they are presented in Figure 3.13.
Geotechnical evidence and authors’ experience show that E0 /E50 ratio varies in natural
clays from 4.6 to 30 depending on soil aging and particle bonding. The higher values of
E0 /E50 ratio are suggested for aged, cemented and structured clay, whereas the lower ones
for insensitive, unstructured and remoulded clays.
CPT. A relationship between G0 and corrected tip resistance qt for clays has been proposed
by Mayne and Rix (1993). The correlation also depends upon the inplace void ratio e0 (cf.
Figure 3.36):
G0 = 49.4qt0.695 e−1.13
0 in MPa (3.47)
with qt in [MPa].
The effective vertical stress σ 0v0 can used to recalculate the estimated modulus E0 = 2G0 (1+
ν ur ) to the reference one E0ref using the stiffness dependency power law assuming that minor
stress σ 3 equal to min (σ 0v0 ; σ 0v0 · K0 ).
CPT. Based on a database for ten Norwegian marine soft clay sites, Long and Shane (2010)
proposed an expression obtained by modifying the original expression by Simonini and Cola
(2000). The relationship, apart of qt , also accounts for pore pressure measurements (cf.
Figure 3.36). The modification was related to replacing ∆u/qc ratio with pore pressure
parameter Bq (= (u2 − u0 )/(qt − σ v0 )) and tunning empirical coefficients.
G0 = 4.39qt1.225 (1 + Bq )2.53 ±50% (3.48)
with G0 , qt in kPa and Bq is a dimensionless pore pressure parameter.

DMT. G0 can be estimated based on DMT data using a correlation reported by Hryciw
(1990). The correlation was originally proposed for sands, silts and clays, however it seems
to underestimate G0 in sands (cf. Figure 3.37):
530 γ DMT /γ W − 1 0.25 0.5

G0 = 0.25 K0 (pa σ 0v0 ) (3.49)
0
(σ v0 /pa ) 2.7 − γ DMT /γ W
with G0 , pa , σ 0v0 in same units; γ DMT /γ W dilatomter-based unit weight ratio obtained using
Marchetti’s chart for soil type and unit weight estimation.

Figure 3.36: Comparison of empirical correlations for estimating G0 (Gmax ) in clays from CPT
data; plot for Eq. (3.47) obtained for e0 = 0.8.
Figure 3.37: Estimating G0 from DMT data using Eq.(3.49) proposed by Hryciw (1990) (plot
obtained for K0 = 0.5).

Geotechnical evidence. Experimental measurements reveal that in the case of fine plastic
soils, the reference strain threshold γ 0.7 at which Gs /G0 = 0.722 may be affected by many
factors such as soil plasticity, stress history, confining pressure, number of cyclic loadings
and others. A well known experimental database reported in Vucetic and Dobry (1991)
(Figure 3.38) illustrates the relationship between γ 0.7 and plasticity index IP . Based on this
experimental data, γ 0.7 can be approximated by the following empirical correlation:
−6
γ 0.7 = γ ref
0.7 + 5 · 10 IP for IP < 15
1.15 log(IP )−5.1 (3.50)
γ 0.7 = 10 for IP ≥ 15
−4
with the reference strain threshold γ ref
0.7 (IP = 0) = 1 · 10 and plasticity index IP in %.
A fit of the above correlation to experimental data is illustrated in Figure 3.38. Since the
IP -dependent chart was compiled from the original data which showed a considerable scatter,
a 50% error can be assumed in estimations giving max and min ranges.
Results for IP < 100 have been experimentally proved in many research, whereas extrapola-
tion for soils which exhibit IP > 100 should be treated carefully.
Figure 3.38: Comparison of predictions for γ 0.7 from the plasticity index IP (PI) using Eq.(3.50)
with experimental data reported for cohesive soils (after Vucetic and Dobry, 1991).
Recently, Vardanega and Bolton (2011) reported a database of 20 clays and silts (OCR=1-
17) for which a hyperbolic fit to the normalized reduction curve data has been proposed as
follows:
G 1
= 0.74 (3.51)
G0

γ
1+
γ 0.5
with the reference threshold parameter γ 0.5 which corresponds to Gs /G0 = 0.5. They also
confirmed that strain threshold for cohesive soils depends on the plasticity index IP (Figure
3.39), liquid limit wL (Figure 3.41) and plastic limit wP . Based on these curves, they proposed
a number of correlations for prediction of the reference strain threshold γ 0.5 . Linear regression

analyzes were characterized by reasonable R2 values although an error band of ±50% was
observed. For the purpose of this report, the original coefficients obtained through regression
analyzes for γ 0.5 where recalculated to the model parameter γ 0.7 :
0.5975IP [−]
γ 0.7 = ±50%, R2 = 0.75, n = 61, for IP = 10 − 150% (3.52a)
1000
0.3442wL [−]
γ 0.7 = ±50%, R2 = 0.75, n = 61, for wL = 25 − 240% (3.52b)
1000
0.7517wP [−]
γ 0.7 = ±50%, R2 = 0.57, n = 61, for wP = 12 − 90% (3.52c)
1000
Note that the above correlations account for no history stress nor confining pressure effects.
Figure 3.39: Comparison of predictions using Eq.(3.51) and (3.52a) with curves from Vucetic and
Dobry (1991).

Figure 3.40: IP -dependent predictions using Eq.(3.51) and (3.52a) with possible 50% error.
Figure 3.41: wL -dependent predictions using Eq.(3.51) and (3.52b) with possible 50% error.

Note that the diagrams proposed in Vucetic and Dobry (1991) and Vardanega and Bolton
(2011) are independent on stress history. In order to account an observed increase of γ 0.7
with the increasing OCR, Stokoe et al. (2004) proposed the following formula to predict γ 0.7
for plastic soils:
−6 0.3
γ 0.7 = γ ref
0.7 + 5 · 10 IP OCR (3.53)
−4
with the reference strain threshold γ ref
0.7 (IP = 0, OCR = 1) = 1 · 10 .
Darendeli (2001) proposed a correlation for the reference strain threshold which additionally
accounts for the effect of confining pressure:
γ 0.7 = (a1 + a2 IP · OCRa3 ) σ a04 (3.54)
Since the original correlation was developed for the reference strain threshold γ 0.5 , for the
purpose of the HS model, the empirical coefficients a1 , a2 , a3 and a4 have been adjusted in
order to predict γ 0.7 :
γ 0.7 - reference strain threshold in [%]
IP - plasticity index in [%]
p00
σ0 = ref normalized confining pressure
p
(2K0 + 1)σ 0v0
p00 = kPa in situ mean effective stress
3
pref ∼= 100kPa (atmospheric pressure)
a1 = 1.25e − 2
a2 = 3.7e − 4
a3 = 0.3
a4 = 0.35
Note that in the current formulation of the HS model no stress dependency for γ 0.7 is
considered. If needed, this parameter can be incorporated into boundary value problems
through defining average mean effective stresses for defined sub-layers.

-
6
Figure 3.42: A graphical interpretation of Eq.(3.54): an example of estimation of shear strain thresh-
old γ 0.7 for IP = 40%, OCR = 4 and mean effective stress p00 = 400kPa.

3.3.2 Strength parameters

It is commonly known that the strength of clays in terms of effective stresses is mostly fric-
tional and the effective cohesion c0 ≈ 0. Small values of cohesion which are observed during
testing may appear in partially saturated clays where the meniscus effects (suction) draw soil
particles together resulting in inter-particles stresses. Larger magnitudes of cohesion can be
often observed in cemented soils due to bonding effects.
Geotechnical evidence. The values of the effective friction angle φ0 observed for fine soils
fall in a wide range from 18o to 42o . Some representative values of φ0 for compacted clays
are provided in Table 3.19 after Carter and Bentley (1991).
Table 3.19: Representative values of φ0 observed in compacted clays (after Carter and Bentley,
1991).
Soil type UCS class φ [o ]

Silty clays, sand-silt mix SM 34
Clayey sands, sandy-clay mix SC 31
Silts and clayey silts ML 32
Clays of low plasticity CL 28
Clayey silts MH 25
Clays of high plasticity CH 19
In fine soils, soil consistency affects the magnitude of the void ratio. Therefore, it seems to
be relevant to relate the friction angle with the soil consistency, as the value of friction angle
depends on the void ratio. Typical ranges of shear strength parameters for some common
soils types given by Senneset et al. (1989) have been extended and are presented in 3.20.
Table 3.20: Typical ranges of shear strength parameters for some common soils types (extended
after Senneset et al., 1989).
Soil type φ [o ] c’ [o ]
Clay
Silt
CPTU. The estimation of effective stress parameters from the total stress analysis of undrained
penetration is difficult. The solution needs to account for excess pore water pressure for which
the distribution around the cone is highly complex and difficult to model analytically. Inter-
pretation methods can be thus viewed as rather approximative.
The effective friction angle φ0 can be estimated using the solution which is based on the
bearing capacity theory (Sandven et al., 1988)7 :
Nq − 1
qt − σ vo = Nm (σ 0vo + a) with Nm = (3.55)
1 + Nu Bq
7
The approach proposed by researchers from the Norwegian University of Science and Technology (NTNU)
is referred to NTNU method.

where a0 denotes the attraction (a0 = c0 cot φ0 ), β is the angle of plastification, Nq and Nu
are the bearing capacity factors (Nq = Nq (φ0 , β) and Nu ∼ = Nu (φ)).
Mayne (2005, 2007) proposed a simplified expression which is applicable to the ranges of
20◦ ≤ φ0 ≤ 45◦ and 0.1 ≤ Bq ≤ 1.0 (see Figure 3.43). By setting for the above method
the effective cohesion intercept c0 = 0 and plastification angle β = 0, the values of φ0 were
evaluated line-by-line and the following approximate expression was obtained:
φ0 ≈ 29.5◦ Bq0.121 (0.256 + 0.336Bq + log Qt ) (3.56)
Resistance number Nm
Friction angle φ0 [o ]
Figure 3.43: Friction angle for sands, silts and clays based on approximation of NTNU original
method (from Mayne, 2005).
Strength parameters vs. undrained shear strength. Best-quality predictions of the

strength parameters φ0 and c0 for cohesive soils can be essentially derived from laboratory
tests. In case of lack of laboratory data, these model parameters can be approximately cal-
ibrated based on the undrained shear strength su . The magnitude of su can be determined
from a variety of in situ tests such as field vane tests (FVT), pressuremeter tests, cone pen-
etration tests (CPT or CPTU), etc. Some interpretation formula for determining su from
commonly used field tests are provided in Appendix A.
Considering that the undrained shear strength in the undrained triaxial conditions is defined
as:
1 1
su = (σ 1 − σ 3 )f = qf (3.57)
2 2
the model parameters φ and c can be adjusted so that they satisfy the normalized condition:
sin situ 1/2qfsim
u ∼
= ’ sim (3.58)
p’0 in situ p0
where sinu
situ
and p’0 in situ denote field test results of the undrained shear strength and the
effective mean stress respectively, whereas qfsim is the failure deviatoric stress obtained through

a numerical simulation of the undrained compression triaxial test at given initial effective mean
stress p’0 sim . Note that the above relation should be considered as approximative since su is
not a unique soil parameter as, it depends, among others, on the type of test, which involves
particular strain paths related to dominant shear modes appearing during testing (cf. Wroth,
1984; Jamiolkowski et al., 1985).
Conceptually, normalization of data in terms of initial stress conditions removes the effect of
depth. Although the mechanisms of particular field test are influenced by both σ 0h0 and σ 0v0 ,
for practical reasons, the normalization can be carried out in terms of σ 0v0 since there is often
little information about σ 0h0 . Therefore, calibration of the strength parameters can be carried
out to satisfy the following relation:
sin
u
situ 1/2qfsim
≈ (3.59)
σ ’v0in situ σ ’1,0sim
where σ ’1,0sim denotes the axial effective stress at the beginning of simulation.
The calibration procedure can be summarized as follows:
1. Assess field values of su for, at least, two different depths (different σ 0v0 ) and plot the data
on σ 0v0 − su chart.
2. Run two simulations of the undrained triaxial compression test (preferably aniotropically-
consolidated with the specified K0SR 6= 1.0) for different σ ’1,0sim and corresponding OCRs
with an initial guess of parameters φ and c, and an assumed failure ratio Rf (note that
an explicit ultimate deviatoric stress can be obtained for the dilatancy angle ψ = 0, or
non-zero ψ with the assumed dilatancy cut-off).
3. Plot numerical results of 1/2qf on σ 0v0 − su chart (as in Figure 3.44) and check the degree
of fit for numerical and in situ trend lines.
4. Return to step 2 if the degree of fit is not satisfactory and modify parameters φ and c. Note
that each modification of φ and c requires updating K0NC and evaluating of parameters M
and H before the next calculation run.
In the case of overconsolidated material, if the initial mean effective stress p00 lies before the
mean effective stress p∗ which value corresponds to the intersection between deviatoric and
isotropic mechanisms (see Figure 3.45), qf can be estimated directly from the Mohr-Coulomb
criterion (cf. Figure 3.1):
qf = α (p00 + c cot φ) (3.60)
where α is related to the friction angle φ which depends on the dominating shear mode which
is appropriate to a given in situ test:

sim (p∗ denotes the

Figure 3.44: Undrained shear strength su against normalizing effective stress σ ’1,0
intersection between deviatoric and isotropic mechanisms, see Figure 3.45).

6 sin φ
α= for triaxial compression conditions (3.61a)
3 − sin φ
6 sin φ
α= for triaxial extension conditions (3.61b)
3 + sin φ
√
α = 3 sin φ for plane strain conditions (3.61c)
Figure 3.45: Effective stress paths derived from simulations of undrained compression test in
normally- and overconsolidated soil.

3.3.3 Failure ratio

For most soils, the value of Rf falls between 0.75 and 1 and an average value of the failure
ratio can be taken as Rf = 0.9. Kempfert (2006) reported some Rf values derived from
triaxial compression tests for three lacustrine soft soils in southern Germany:
• CICD test: Rf = 0.73 − 0.88, with the average value of 0.82
• CICU test: Rf = 0.70 − 0.99, with the average value of 0.89

3.3.4 Dilatancy angle

While for granular soils the maximal dilatancy angle can be related to the relative density
(Dr ), an estimation of ψ for clays relies on geotechnical experience. For instance, it can be
observed in laboratory tests that normally consolidated clays may exhibit no dilatancy at all.
Therefore, it is proposed that for cohesive soils, it can be assumed that dilatancy depends on
the preconsolidation state and ψ can be approximately taken as:
• ψ = 0o for normally- and ligthly-overconslidated soil
• ψ = φ0 /6 for overconsolidated soil
• ψ = φ0 /3 for heavily overconsolidated soil
Another reasonable approximation of ψ can be obtained using the following equation:
ψ = φ − 30o (3.62)
for the values of friction angle larger than 30o .

Notice that the above approximation gives ψ ≈ φ0 /3 for a dense gravel, whereas small
values of ψ will be obtained for the the friction angles that correspond to compacted silts or
clays(typically φ ≈ 30o to 35o ).

3.3.5 Stiffness moduli

In general, the relation between the stiffness moduli is as follows:
E50 < Es < Eur (3.63)
where Es denotes ”static” modulus or secant modulus taken from the initial part of the ε1 −q
experimental curve at ε1 = 0.1%.
In the case when one of the stiffness moduli cannot be directly determined, it may be relevant
for many practical cases to set:
ref
Eur
ref
= 3 to 6 (an average can be assumed equal to 4) (3.64)
E50
ref
Eur
However, note that the following condition should be satisfied: ref
> 2.
E50
The secant modulus E50 can be approximated based on the known value of the ”static”
modulus Es and according to the approach described in Section 3.2.2, Figure 3.22. A rough
approximation of the order of magnitudes for the ”static” modulus Es is given in Table 3.21.
Table 3.21: Typical values for the ”static” modulus Es [MPa] (compiled from Kezdi, 1974; Prat
et al., 1995).
Soil Consistency
Very Soft Soft Medium Stiff Very Stiff Hard
Soil Type Min Max Min Max Min Max Min Max Min Max Min Max
Silts
slight plasticity 2.5 4 5 8 10 15 15 20 20 40 40 80
low plasticity 1.5 3 3 6 6 10 10 15 15 30 30 60
Clays
low to medium plast. 0.5 3 2 5 5 8 8 12 12 30 30 70
high plasticity 0.35 2 1.5 4 4 7 7 12 12 20 20 32
Silt organic 0.5 5
Clay organic 0.5 4
Assuming that during unloading/reloading soil behaves elastically, the modulus Eur can be
related with the constrained unloading/reloading oedometric modulus Eoed,ur through:
(1 + ν ur )(1 − 2ν ur )
Eur = Eoed,ur (3.65)
1 − ν ur
Note, however, that Eur is not a unique value for a given soil in the oedometric test because
Eur depends on the previous maximal stress level σ 0c attained before the unloading and the
corresponding void ratio ec , as shown in Figure 3.46. Therefore, assuming an infinitesimal
change of the compression stress, i.e. ∆σ 0 → 0, the unloading/reloading oedometric modulus
Eoed,ur should be approximated by similarity with Eq.(3.3) as:
2.3(1 + e0 ) 0
Eoed,ur = σc (3.66)
Cs

where Cs is the swelling index (Figure 3.46).

Since the Eoed,ur was approximated for the stress point belonging to the primary loading line
(K0NC -line), such a determined reference unloading/reloading modulus Eurref
corresponds thus
ref
to the reference stress σ which can be estimated from:
σ ref = K0NC σ 0c (3.67)
Figure 3.46: Idealized plot of one-dimensional oedometric compression test.
DMT. The ”static” modulus corresponding to Eur can be evaluated for cohesive soils the ver-
tical drained constrained modulus MDMT which is derived from three intermediate dilatometer
parameters, i.e. the material index ID , the horizontal stress index KD , and the dilatometer
ref
modulus ED . Note that MDMT may correspond to Eoed only for normally-consolidated soil.
The unloading-reloading modulus Eur can be evaluated assuming that:
Eur = Es = aEoed = aMDMT (3.68)

where a ≈ 0.9 as a = (1 + ν ur )(1 − 2ν ur )/(1 − ν ur ) and the dilatometer vertical drained
constrained modulus MDMT :
MDMT = RM ED (3.69)
with ED = 34.7(p1 − p0 ) and RM depending on soil type behavior, i.e.
• clayey silt to sandy silt (0.6 < ID < 1.8): RM = RM,0 + (2.5 − RM,0 ) log KD and
RM,0 = 0.14 + 0.15(ID − 0.6)
• clay to silty clays (ID ≤ 0.6): RM = 0.14 + 2.36 log KD
• if KD > 10: RM = 0.32 + 2.18 log KD
• if RM < 0.85: set RM = 0.85
Undrained vs drained moduli - theoretical relationship. In case of lack of drained

compression test data, the stiffness moduli can be calibrated based on the results derived from
the undrained triaxial compression test (e.g. CIUC or CAUC). Since water filling skeleton

pores has no shear stiffness, the shear modulus is not affected by the drainage condition so
one can write:
Eu E
= Gu = G = (3.70)
2(1 + ν u ) 2(1 + ν)
where ν u is the Poisson’s coefficient in undrained conditions.
Considering that the undrained conditions imply ε1 = ε3 , and therefore ν u = 0.5, the above
equation can be rewritten as:
Eu 3
= (3.71)
E 2(1 + ν)
and for the drained Poison’s coefficient ranging for most soils between 0.12 and 0.4:
Eu
≈ 1.07 to 1.34 (3.72)
E
An assumption of ν u = 0.5 for undrained conditions can also be expressed with the condition
of no volume change (∆εv = 0). Since the undrained bulk modulus Ku tends to infinity in
such conditions , ν u → 0.5:
∆σ Eu
Ku = = (3.73)
∆εv 3(1 − 2ν u )
The undrained ”static” modulus Esu can be estimated based on a value of undrained shear
strength su using an empirical correlation:
E u = Kc su (3.74)
with an empirical correlation coefficient which depends on the plasticity index PI and OCR,
and can be estimated from Figure 3.47.
Figure 3.47: Evaluating the undrained modulus from E u from su : chart for estimating the correla-
tion coefficient Kc in Eq. (3.74) (from Duncan and Buchignani, 1976).

Undrained vs drained moduli - curve fitting. The ”drained” model moduli can also
be calibrated by means of curve-fitting. It is recommended because in the ”undrained” test
the effective stress σ 03 (which corresponds to the reference stress) does not remain constant
during compression due to a development of excess pore water pressure. The calibration of
0 0
”drained” stiffness moduli (Eur , E50 ) from ”undrained” test requires fitting laboratory data,
i.e. curve ε1 − q, with the results obtained through an axis-symmetric, one-element simula-
tion of the undrained compression. The flowchart of the parameter calibration is presented
in detail in Figure 3.48 and it is described below.
Considering that strength parameters φ0 and c0 can be directly derived from undrained test
data, they should be kept unmodified during curve-fitting. In order to avoid excessive gain
in material resistance after reaching the failure stress point, the dilatancy angle can be set
ψ = 0o during ”undrained” simulations (cf. Truty and Obrzud (2015)). The soil unit weight
should be set to γ = 0 in order to cancel body force loading. As regards the fluid weight the
option X Skip gravity term should be chosen. A non-zero value for the initial void ratio e0
should be set.
In order to represent the undrained behavior a finite value of the fluid bulk modulus should
be set, e.g. the bulk modulus of water is 2.2e6 kPa. The ratio between fluid and soil bulk
moduli should be of order Kf /K = 105 ÷ 106 . It corresponds to the penalty formulation
which may fail to converge if the penalty factor is too large. In such a case computation will
be terminated and some null pivots will be reported in the ∗ .log file. In order to remedy such
a problem, the value of the fluid bulk modulus should be decreased. Note that the two-phase
stabilization should not be activated in the single-element test as no pressure oscillation is
observed (pore pressure is constant over the element). No initial pressure BC need to be
introduced.
The simulation should be carried out using Axisymmetry analysis type,
Deformation+Flow problem type and Consolidation driver should be set for when running
compression of the element.

Figure 3.48: Flowchart for calibration of stiffness moduli E0 , Eur , E50 based on ε1 − q curve derived
from the undrained compression triaxial test (CU) and a single-element test.

Geotechnical evidence. Kempfert (2006) have provided typical results for the ratios be-
tween stiffness moduli. These ratios are presented below in Tables 3.22 and 3.23.
Table 3.22: Relationship between triaxial stiffness moduli and oedometric moduli for three lacustrine
clays in Germany, from Kempfert (2006).
Ei /Eoed E50 /Eoed Eur /Eoed,ur Eoed,ur /Eoed

Soil 1 2.08 1.03 2.33-2.52 2.60
Soil 2 1.63 0.77 1.29-2.09 3.63
Soil 3 2.82 1.45 1.32-2.51 6.65
Average 2.17 1.08 4.29
Ei was derived from the initial slope of the triaxial curve ε1 − q
Eoed,ur denotes unloading/reloading oedometer modulus
Table 3.23: Relationship between stiffness moduli derived from drained and undrained triaxial tests
for three lacustrine clays in Germany, from Kempfert (2006).
Drainage conditions Ei /E50 Eur /Ei Eur /E50

Soil 1 drained 2.02 3.20 5.93
undrained 1.48
Soil 2 drained 2.17 3.10 6.72
undrained 1.84
Soil 3 drained 1.94 6.55 12.66
undrained 3.02
Average drained 2.04 4.28 8.43
undrained 2.11
Ei was derived from the initial slope of the triaxial curve ε1 − q

3.3.6 Oedometric modulus

The one-dimensional constrained tangent modulus Eoed (which, in literature, is often assigned
as MD when determined from in situ tests) is obtained for steady state measurements based
on the oedometer test through the expression:
δσ 0v
Eoed = MD = (3.75)
δεv
which can also be expressed as:
2.3(1 + e)σ 0v (1 + e)σ 0v
Eoed = = (3.76)
Cc λ
where Cc is the compression index(Cc = 2.3λ). A number of empirical correlations which
can be used to evaluate Cc are given in Appendix B.
In case of lack of relevant data the oedometric modulus can be approximately taken as:
ref ∼ ref
Eoed = E50 (3.77)
In such a case, the oedometric vertical reference stress σ ref
oed should be matched to the reference
minor stress σ ref since the latter typically corresponds to the confining (horizontal) pressure
σ ref = σ 3 = σ 0h :
σ ref NC
oed = σ ref /K0 (3.78)
ref ref
As an example, Kempfert (2006) reports E50 /Eoed ratio for three lacustrine clays in Germany
which varies from 0.77 to 1.45 with the average 1.08 (see Table 3.22) .
ref
Important note. In the case of the HS model, Eoed can be taken as equal to MD if the
latter has been derived from CPT, CPTU or DMT but only for normally-consolidated
soil. In such a case σ oed can be taken as σ 0v0 which corresponds to the testing depth for
which MD has been evaluated.
CPT. The constrained modulus for clays can be interpreted from the CPT or CPTU test
using the measured cone resistance qc and an empirical coefficient αm . Lunne et al. (1997)
quote the values of αm for different types of soils proposed by Sanglerat (1972).
CPTU. The constrained modulus can be interpreted from the CPTU using the net cone
resistance qt − σ v0 (qt denotes the corrected cone resistance):
MD = αn (qt − σ v0 ) (3.79)
where αn is observed for most clays between 5 and 15 while for normally consolidated clays,
it is between 4 to 8 (Sandven et al., 1988; Senneset et al., 1989). A more general correlation
was suggested by Kulhawy and Mayne (1990) (cf. Figure 3.49):
MD = 8.25(qt − σ v0 ) (3.80)
As discussed be Lunne et al. (1997), the estimation of ”drained” parameter MD from an
undrained penetration test using general empirical correlations may suffer from errors as large
as ±100%. An individual site-specific calibration is thus recommended for αn . They also
concluded that it is difficult to correlate ”drained” parameters without accounting for the
pore pressure measurements as the cone resistance is measured in total stress.
DMT. The constrained modulus MD can be interpreted from three intermediate dilatometer
parameters, i.e. the material index ID , the horizontal stress index KD , and the dilatometer
modulus ED , by applying the correlation presented in Eq.(3.69).

Table 3.24: Estimation of constrained modulus MD for clays (after Lunne et al. (1997)).
MD = αm · qc
qc < 0.7 MPa 3 < αm < 8 Clay of low plasticity (CL)
0.7 < qc < 2.0 MPa 2 < αm < 5
qc > 2.0 MPa 1 < αm < 2.5
qc < 2.0 MPa 1 < αm < 3 Silts of low plasticity (ML)
qc > 2.0 MPa 3 < αm < 6
qc < 2.0 MPa 2 < αm < 6 Highly plastic silts and clays (MH, CH)
qc < 1.2 MPa 2 < αm < 8 Organic silts (OL)
qc < 0.7 MPa Peat and organic clay (Pt ,OH)
50 < w < 100 1.5 < αm < 4 (w-water content [%])
100 < w < 200 1 < αm < 1.5
w > 200 0.4 < αm < 1
Figure 3.49: General MD correlation for CPTU data proposed by Kulhawy and Mayne (1990) (from
Lunne et al., 1997).

3.3.7 Stiffness exponent

Geotechnical evidence. The formulation of HS models assumes the same exponent m for
four different stiffness moduli, i.e. E0 , E50 , Eur and Eoed . Kempfert (2006) demonstrated
that in reality, the following relation may appear m0 < m50 < mur ; they also provide some
typical values derived from drained tests (oedometer loading and triaxial tests) for three la-
custrine soft soils:
moed mavg
oed m0 mavg
0 m50 mavg
50 mur mavg
ur
(no. of (no. of (no. of (no. of
tests) tests) tests) tests)
Soil 1 0.73- 0.75 (2) 0.3- 0.34 (3) 0.39- 0.45 (3) 0.74 0.74 (1)
0.76 0.42 0.51
Soil 2 0.58- 0.64 (2) 0.52- 0.68 (4) 0.66- 0.72 (4) 0.61- 0.64 (2)
0.69 0.79 0.84 0.67
Soil 3 0.58 0.58 (1) 0.42- 0.51 (3) 0.38- 0.48 (3) 0.79- 0.84 (3)
0.56 0.54 0.89
Kempfert (2006) also highlighted that the exponent m for undrained tests can be generally
higher than for drained tests.
Viggiani and Atkinson (1995) reports the exponent numbers m for different clays at very
small strains as a function of the plasticity index IP (see Figure 3.50(a)) whereas Hicher
(1996) presents them as a function of the liquid limit wL (see Figure 3.50(b)).
(a) (b)
Figure 3.50: Power law exponent m related to (a) plasticity index IP (Viggiani and Atkinson, 1995),
and (b) liquid limit wL (Hicher, 1996).

Table 3.25: Typical values for mp * observed in clays for the shear modulus G0 (from Benz, 2007).
Soil tested IP [%] mp [-] Reference

Avezzano clay (Holocene-Pleistocene) 10-30 0.46 Lo Presti and Jamiolkowski (1998)
Fucino clay (Holocene-Pleistocene) 45-75 0.40 Lo Presti and Jamiolkowski (1998)
Garigliano clay (Holocene) 10-40 0.58 Lo Presti and Jamiolkowski (1998)
Panigaglia clay (Holocene) 44 0.50 Lo Presti and Jamiolkowski (1998)
Montaldo di Castro clay (Pleistocene) 15-34 0.40 Lo Presti and Jamiolkowski (1998)
Recon. Valericca clay (Pleistocene) 27 0.85 Rampello et al. (1997)
Pisa clay (Pleistocene) 23-46 0.44 Lo Presti and Jamiolkowski (1998)
London clay (reconstituted) 41 0.76 Viggiani and Atkinson (1995)
Speswhite kaolin clay (reconstituted) 24 0.65 Viggiani and Atkinson (1995)
Kaolin clay 35 0.50 Marcuson and Wahls (1972)
Bentonite clay 60 0.50 Marcuson and Wahls (1972)
0 mp
p
The number mp was obtained for the relation G0 ∝
σ ref
Table 3.26: Suggested ranges of stiffness exponent m observed for oedometric modulus Eoed (von
Soos, 1991).
Soil type m [-]

Silt: low plasticity 0.6 ÷ 0.8
Silt: medium and high plasticity 0.7 ÷ 0.9
Clay: low plasticity 0.9 ÷ 1.0
Clay: medium plasticity 0.95 ÷ 1.0
Clay: high plasticity 1.0
Silt or clay: organic 0.85 ÷ 1.0
Peat 1.0
Mud 0.9 ÷ 1.0

3.3.8 Overconsolidation ratio

In the case of a soil which is located below the ground water table, a qualitative estimation
of the overconsolidation ratio can be done based on the Atterberg limits and the natural
moisture content. Assuming that the soil is saturated, it can be expected that smaller void
ratios have less water space and wn would be smaller Bowles (1997). From this observation,
the following may be deduced:
if wn is close to wL soil is normally consolidated

if wn is between wP and wL soil is lightly consolidated
if wn is close to wL soil is lightly- to heavily overconsolidated
if wn is larger than wL soil is on verge of being a viscous liquid
In the latter case (wn > wL ), stability of soil in in situ conditions may be ensured by
overburden pressure and interparticle bonds, unless visual inspection indicates a liquid mass.
Geotechnical evidence. Mayne (1988) provides empirical upper and lower limits derived
from laboratory tests:
• undrained shear strength su determined under isotropically consolidated-undrained triaxial

conditions (CIUC):
1.43 1.43
su su
1.82 0 ≤ OCR ≤ 4 0 (3.81)
σ v0 σ v0
• su determined under isotropically consolidated-undrained triaxial conditions (CAUC):

1.25 1.25
su su
3.70 0 ≤ OCR ≤ 5.26 0 (3.82)
σ v0 σ v0
CPTU. One of the best working approaches relates the overconsolidation ratio OCR to the
net cone resistance qt − σ v0 :
qt − σ v0
OCR = kσt (3.83)
σ 0v0
where kσt is an empirical coefficient which falls in the interval from 0.1 to 0.5 for non-
fissured clays (Larsson and Mulabdić, 1991; Hight and Leroueil, 2003). The higher values
are suggested for cemented, aged and heavily consolidated soils (between 0.9 and 2.2). For
good-quality interpretation, this coefficient needs to be calibrated for specific site conditions
based on the benchmark values derived from oedometer test. However, the first-order ap-
proximates of OCR can be obtained using the values of kσt from multiple regression analyzes
which are based on historical syntheses from many characterization sites (see Table 3.27).
Mayne (2006b) suggests assuming kσt = 0.30 for first-order estimates.
Another approach combines measurements of cone resistance qt and pore pressure u2 mea-
sured behind the cone:
qt − u2 qe
OCR = kσe 0
= kσe 0 (3.84)
σ v0 σ v0
with kσe being obtained through site-specific correlations. By analogy to the previous ap-
proach, the first-order approximates of OCR can be obtained using the values of kσe through

Table 3.27: Comparison of the empirical coefficients obtained from multiple regression analyzes for
non-fissured clays.
Results of regression analysis

Ref. Geographical Number Number
region of sites/points kσt R2 of sites/points kσe R2
[1] Sweden 9/110 0.292 - 9/110 0.50 -
[2] Canada 31/153 0.294 0.90 31/153 0.546 0.96
[3] Worldwide 123/1121 0.305 0.84 84/811 0.50 0.75
[1] Larsson and Mulabdić (1991); [2] Demers and Leroueil (2002);
[3] Chen and Mayne (1994, 1996)
regression analyzes (see Table 3.27). Mayne (2006b, 2007) suggested assuming kσe = 0.60
for the first-order estimates. This approach is often used as a comparative to the previous one
and local correlations are strongly recommended. The formula is also viewed as less reliable
in soft, lightly overconsolidated clays the qt results accompanied by large values of u2 yield
in a small number for qt − u2 (Houlsby and Hitchman, 1988; Lunne et al., 1997).

DMT. Based on dilatometer measurements, estimation of OCR for clays can be carried out
with the formula proposed by Marchetti which relates the horizontal stress index8 KD to
OCR from oedometer tests with the following correlation:
OCR = (0.5KD )1.56 (3.85)
The application of this correlation is restricted to materials with ID < 1.2, free of cementation
which have experienced simple one-dimensional stress histories (Totani et al., 2001).
An improved relationship which takes into account a large range of soil plasticity in the
exponent was proposed by Lacasse and Lunne (1988):
1.35÷1.67
OCR = 0.225KD (3.86)
where the exponent varies from 1.35 for plastic clays, up to 1.67 for low plasticity materials.
Figure 3.51: Various correlations KD − OCR for cohesive soils from various geographical areas
(from Totani et al., 2001).
8
Horizontal stress index which is calculated based on the first dilatometer reading p0 , i.e. KD = (p0 −
u0 )/σ 0v0 .


The coefficient of earth pressure ”at rest” for normally-consolidated clays can be estimated
through Eq.(3.37a) or the similar expression suggested in Brooker & Ireland (1965):
K0NC = 0.95 − sin φ0 (3.87)
For cohesive soils, K0NC can also be related through empirical correlations with soil plasticity:
K0NC = 0.19 + 0.233 log IP (Alpan, 1967) (3.88a)

and similar
K0NC = 0.44 + 0.0042IP (Holtz & Kovacs, 1981) (3.88b)
where IP is the plasticity index in %.
It is commonly known that in cohesive soils the preconsolidation plays an important role and
K0 typically increases with the overconsolidation ratio OCR. Estimations of the initial stress
state for overconsolidated soil take a general form:
K0 = K0NC OCRm (3.89)
where m is a coefficient which for estimation of K0NC for most practical purposes can be
taken as:
m = 0.5 suggested by Meyerhof (1976) (3.90a)
m = sin φ0 suggested in Mayne & Kulhawy (1982) (3.90b)
However, the upper bound value for K0 should be limited by the passive lateral earth pressure
coefficient:
1 + sin φ
Kp = (3.91)
1 − sin φ
The equations are presented graphically in Fig.3.52 (K0NC was calculated using Eq. (3.37a)).
SBPT. Approximation of K0 from the self-boring pressuremeter test requires determination

of the horizontal effective stress σ 0h0 since the vertical effective stress σ 0vo can be estimated
based on depth, unit weight and groundwater information. In the case of SBPT, the in situ
total horizontal stress can be directly estimated from the ”lift-off” pressure (Jamiolkowski
et al., 1985; Clough et al., 1990; Amar et al., 1991). The ”lift-off” pressure corresponds to
the internal cavity pressure ψ 0 when the membrane starts to deform the wall of a borehole,
therefore ψ 0 ∼
= σ 0h0 + u0 (Figure 3.53). The ”lift-off” is typically estimated based on the
averaging procedure including the measurements of three feeler arms spaced at 120◦ around
the instrument (Dalton and Hawkins, 1982; Mair and Wood, 1987). In general, the lateral
stress measurements can be considered as fairly accurate in clays, particularly in soft deposits
(Jamiolkowski et al., 1985).

Figure 3.52: Typical relationships between K0 and OCR observed for clays based on the correlation
proposed by Mayne and Kulhawy (1982).
Figure 3.53: Example of the total horizontal stress estimation from the lift-off pressure in soft clay
at Panigaglia site (after Jamiolkowski et al., 1985).

Figure 3.54: Approximation of K0 based on the normalized pore pressure difference PPSV from
dual sensor cone (after Sully and Campanella, 1991).
CPTU. At present, no reliable method exists for interpretation K0 from CPT data. Rough
evaluations related directly to CPTU measurements can be made using various approximative
methods.
Observing that the pore pressure distribution around the cone is a function of σ 0ho , Sully and
Campanella (1991) proposed to approximate K0 based on a linear regression analysis using
the normalized difference between pore pressure measured at the cone tip u1 and behind the
tip at the sleeve shoulder u2 :
K0 = 0.11 · PPSV + 0.5 (3.92)
where PPSV= (u1 − u2 )/σ 0v0 and the empirical coefficient aK was obtained equal to 0.11,
see Figure 3.54. Note that the regression analysis reveals a considerable scatter and this
identification approach should be used carefully.
Masood and Mitchell (1993) proposed the estimation of K0 based on measurements at the
friction sleeve fs . In this method, K0 is a function of the normalized sleeve friction fs /σ 0v0
and the overconsolidation ratio OCR, as presented in Figure 3.55. Thus the approach requires
prior evaluation of OCR and reliable measurements of fs .
The most common technique for estimating K0 employs an empirical formula which is based
on the normalized cone resistance:

qt − σ vo
K0 = kK (3.93)
σ 0vo

Figure 3.55: Proposed relationship between K0 , fs and OCR Masood and Mitchell (after 1993).
where kK is an empirical coefficient. Using the regression analysis, Kulhawy and Mayne
(1990) obtained the value of kK = 0.1 for several K0 values estimated from the self-boring
pressuremeter test (SBPT), see Figure 3.56.
Figure 3.56: General K0 correlation for CPTU data proposed by Kulhawy and Mayne (1990)
(adapted from Lunne et al., 1997).
DMT. The original correlation for K0 based on the dilatometer data, relative to uncemented
clays is (Marchetti, 1980):
0.47
KD
K0 = − 0.6 (3.94)
1.5
In highly cemented clays, the above equation may significantly overestimate K0 , since part
of KD is due to the cementation.

DMT. K0 can also be interpreted from dilatometer test data. Since the original Marchetti
relationship tends to overestimate K0 , its estimation can be carried out through the correlation
suggested in Lacasse and Lunne (1988):
0.44÷0.64
K0 = 0.34KD (3.95)
where the lower exponent value is associated with highly plastic clays, whereas higher values
are suggested for for low plasticity materials.

3.3.10 Void ratio

Typical values of voids ratio and dry unit weights observed for cohesive soils are provided in
Table 3.28 and 3.29.
Table 3.28: Typical values of void ratios and dry unit weights observed in cohesive soils (from
Hough, 1969).
Voids ratio
e [–]
Soil type Min Max
Silty or sandy clay 0.25 1.8
Gap-graded silty clay w. gravel or larger 0.2 1
Well-graded gravel/sand/silt/clay 0.13 0.7
Clay (30 to 50% of 2microns size) 0.5 2.4
Colloidal clay (over 50% of 2microns size) 0.6 12
Organic silt 0.55 3
Uniform, inorganic silt 0.4 1.1
Organic clay (30 to 50% of 2microns size) 0.7 4.4
Table 3.29: Typical values of void ratios and unit weights observed in granular soils (from Terzaghi
et al., 1996).
Voids ratio Dry unit weights Wet weights

Soil type e [–] [kN/m3 ] [kN/m3 ]
Glacial till, very mixed grained 0.25 20.8 22.7
Soft glacial clay 1.2 12 17.4
Stiff glacial clay 0.6 16.7 20.3
Soft slightly organic clay 1.9 9.1 15.5
Soft very organic clay 3 6.7 14.4
Soft bentonite 5.2 4.2 12.5

3.4 Automated assistance in parameter determination
Virtual Lab v2020 is a highly-interactive module which provides users with:
• assistance in selecting a relevant constitutive law with regards to the general behavior of
the real material
• first-guess parameter estimation based on field test records
• automated parameter selection (first-guess values of model parameters for soil for any
incomplete or complete specimen data)
• user-engaged parameter selection (interactive parameter selection which involves browsing

different parameter correlations including field tests data)
• ranges of parameter values which can be considered in parametric studies
• automated parameter identification from laboratory experimental data
• possibility of running numerical simulations of elementary laboratory tests in order to visu-

alize the constitutive model response for the defined model parameters
• possibility of comparing numerical simulations of elementary laboratory tests with curves

obtained in the laboratory
A separated report provides Help to Virtual Lab.

The toolbox is initialized by clicking on Virtual Lab which is visible once one of the following
continuum models has been chosen as the material definition (Figure 3.57):
• Mohr-Coulomb
• Hardening-Soil Small Strain
• Cam-Clay
• Cap

3.4. AUTOMATED ASSISTANCE IN PARAMETER DETERMINATION
Figure 3.57: Initializing Virtual Lab from the Materials window


Chapter 4
Benchmarks
4.1 Triaxial drained compression test on dense Hostun sand
Files:
HS-std-dh-sand-100kPa.inp,
HS-small-dh-sand-100kPa.inp,
HS-brick-dh-sand-100kPa.inp
The following section presents a validation of all HS model versions ie. HS-standard, HS-
small, HS-Brick on a triaxial drained compression test for Hostun sand. Material properties
are taken from PhD thesis by Benz (2007) and are given in the following table:
Parameter Unit Value Parameter Unit Value

ref
Eur [kPa] 90000 ψ [o ] 16.0
ref
E50 [kPa] 30000 ft [kPa] 0.0
σ ref [kPa] 100 D [–] 0.0/0.251
m [–] 0.55 M [–] 1.5656/1.5597
ν ur [–] 0.25 H [kPa] 51649/52339
Rf [–] 0.9 OCR [–] 1.0
c [kPa] 0.0 E0ref [kPa] 270000
φ [o ] 42.0 γ 0.7 [–] 0.0002
Parameters M and H were estimated automatically by the code assuming K0NC = 0.4 and
Eoed = 30000 kPa at reference stress equal to 100 kPa. It must be emphasized that M and
H values will not be equal to the ones given by Benz because of the different form of the
hardening law for the preconsolidation pressure pc . All results obtained with Z Soil match
very well results published by Benz.
CHAPTER 4. BENCHMARKS
σ1
(a) (ε1 ) (Z Soil) (b) G(γ) (Z Soil)
σ3
σ1
(c) (ε1 ) (zoom) (Z Soil) (d) εv (ε1 ) (Z Soil)
σ3
(e) Solution by Benz Benz (2007)
Figure 4.1: Results for the confining pressure σ 3 = 100 kPa

4.1. TRIAXIAL DRAINED COMPRESSION TEST ON DENSE HOSTUN SAND
σ1
σ3
σ1
σ3

σ1
σ3
σ1
σ3

4.2. ISOTROPIC COMPRESSION OF DENSE HOSTUN SAND
4.2 Isotropic compression of dense Hostun sand
File: HS-isotropicCompr.inp
This benchmark is solved analytically for the HS-standard model with classical stress depen-
dency function using (σ 3 + c cotφ) term. The decomposed total strain increments for the
elastic and plastic part in isotropic compression conditions (p = pc ) are presented in the fol-
lowing expression (NB. the increment of total volumetric strain is measured from the initial
configuration of equilibrium p0 = pc0 to the current one):
σ ref pc 1−m pco 1−m

−
p 1 − m σ ref σ ref
∆εv = (4.1)
H
" 1−m 1−m #

σ ref + c cotφ

3(1 − 2ν ur ) pc + c cotφ pco + c cotφ
∆εev = ref
−
Eur 1−m σ ref + c cotφ σ ref + c cotφ
(4.2)
∆εv = ∆εev + ∆εpv (4.3)
Verification was carried out on an single axisymmetric finite element which is subject to an
external uniformly distributed load varying from q = 50 kN/m2 to q = 250 kN/m2. The
initial effective stresses are σ o = {−50, −50, 0, −50}T kPa. Material data for the dense
Hostun sand (see section (4.1)) is used in the simulation. Numerical and analytical solutions
are compared in in the following figure.

4.3 Oedometric compression test
Files: HS-oedometer.inp,HS-oedometer-5steps.inp, HS-oedometer-2steps.inp, HS-

oedometer-1.inp
This single-element axisymmetric benchmark demonstrates that using the HS-standard model
an assumed Eoed = 30000 kPa at the reference stress σ ref NC
oed kPa and K0 =0.4 are correctly
reproduced. This benchmark uses material data for dense Hostun sand presented in Section
4.1.
In the first 3 tests an element is loaded vertically by an uniformly distributed load which varies
from q = 75 kN/m2 to q = 275 kN/m2 and is applied in 10, 5 and 2 steps respectively.
Stress strain diagram and the resulting K0NC coefficient are shown in the next two figures.
In the last test, a stress driven program is run using a single step in which vertical traction
varies from 99.99 kPa to 100.01 kPa. The initial effective stress state is defined as σ o =
{−99.99 ∗ 0.4, −99.99, 0, −99.99 ∗ 0.4}T kPa.
The result of this test yields the tangent oedometric modulus at σ ref = 100 kPa which is
∆σ y 0.02
computed using central finite difference scheme as follows Eoed = = =
∆εy 6.6367e − 7
30135 kPa, which is close to the assumed value. A small error (below 0.5%) results from the
assumed tolerance in M and H estimation procedure.
σ y − εy plot for different number of load increments
Estimated K0NC

4.4. OEDOMETRIC COMPRESSION TEST - K0NC -PATH TEST
4.4 Oedometric compression test - K0NC -path test
Files: KoNC-HS-std.inp,HS-oedometer-5steps.inp, HS-oedometer-2steps.inp, HS-

oedometer-1.inp
This single-element benchmark shows that HS-standard model is able to correctly reproduce
K0NC -path for the oedometric test for different values of the friction angle. The results are
compared with the data obtained with other constitutive models.
The oedometric test on the normally-consolidated soil (OCR = 1) is modeled in axisymme-
(a) (b)
Figure 4.4: K0NC -path test for the HS-standard model and different values of the friction angle
φ = 20o , 30o , 40o : a) stress paths in p0 − q plane, b) K0NC with increasing loading.
try, with a single element subject to an external, uniformly distributed load which varied from
q = 50 kN/m2 to q = 1000 kN/m2 . The initial stress state corresponds to the K0NC = 1−sin φ
value. Results which are presented in Figure 4.4 show that HS-standard model correctly re-
produces K0NC stress paths which obey the empirical expression K0NC = 1 − sin φ for three
different values of the friction angle. It is worth noting that during whole loading history the
K0NC is constant.
The same model is used to compare the results from the oedometric test using different
constitutive models: the Modified Cam clay (MCC), the standard Mohr-Coulomb (M-C) and
the Cap model (CAP). Simulations for different models were carried out for the same value
of the friction angle φ = 30o (and equivalent M = 1.2 for the MCC model). Hence, the
value of cohesion was assumed as c = 0 kPa. The value of Poisson’s ratio was assumed
ν ur = 0.2 for the HS model and ν = 0.3 for other models. Figure 4.5 shows that starting
from the initial ”zero” stress setup and gradually increasing the vertical stress, only the HS
model is able to reproduce expected σ 0h /σ 0v path. It is so because the parameter M which
defines the shape of the cap surface in the HS-model, is optimized so that the tangent to
the cap surface at the stress reversal point is perpendicular to K0NC stress path.

ν
1−ν
Figure 4.5: K0NC -path test using different models for φ = 30o (K0NC = 0.5): Hardening Soil (HS),
Modified Cam Clay (MCC), standard Mohr-Coulomb (M-C) and Cap model (CAP).

Chapter 5
Case studies
5.1 Excavation in Berlin Sand
File: HS-std-Exc-Berlin-Sand-2phase.inp
File: HS-small-Exc-Berlin-Sand-2phase.inp
File: MC-Exc-Berlin-Sand-2phase.inp
This example demonstrates the importance of modeling excavation problems using Hardening
Soil model. The study case presents an analysis of main differences between HS-standard,
HS-small, HS-Brick and Mohr-Coulomb models based on numerical simulation of deep exca-
vation in Berlin Sand.
An engineering draft of the problem and the sequence of both excavation and construction
steps, are given in Figure 5.1. Material data for calibration of sand was taken from Benz
(2007) and Schweiger (2002). The data with standard MC model was generated assuming
that stiffness of sand varies according to the power law:
√
E = 20000 y kPa for y ≤ 20m
√
E = 60000 y kPa for y > 20m
where y is the depth expressed in [m]. The same strength parameters apply to both the MC
model and HS models.
CHAPTER 5. CASE STUDIES
Figure 5.1: An engineering draft and the sequence of both excavation and construction stages for
the deep excavation in Berlin sand.
Sequence of stages:
1. Generating an initial stress state for an assumed K0in situ in sand layers
2. Installation of the diaphragm wall
3. Lowering the ground water level in the excavated zone up to the elevation -17.90m
4. Excavation step 1 (up to -4.80m)
5. Introducing the first row of anchors (distance 2.30 m) and applying the prestress P0 = 768
kN
7. Introducing of the second row of anchors (distance 1.35 m) and and applying the prestress
P0 = 945 kN
9. Introducing of the third row of anchors (distance 1.35 m) and and applying the prestress
P0 = 980 kN

5.1. EXCAVATION IN BERLIN SAND
Figure 5.2: Excavation in Berlin Sand: FE mesh.
A finite element model of the problem is shown in the figure below. The mesh represents:
• deposits consisting of two sand layers which are described with two different groups of
stiffness characteristics
• three rows of prestressed anchors
• diaphragm wall
• contact interfaces between sand and the wall
• zone of artificial contact elements which are used to model a hydraulic barrier (preserving
continuity of displacement field and discontinuity of pore pressure)
• external displacement boundary conditions (BC) (box type)
• pressure BC which are applied via fluid head and set up along the right hand side boundary,
as well as along the left hand boundary up to the level of impermeable barrier (pressure
fluid head BC is applied with the aid of seepage elements)

Table 5.1: Excavation in Berlin Sand: material properties for soils
Material Model Data group Properties Unit Value

1 Sand (-20m↑) HS-small Elastic Eur [kN/m2 ] 180000
σ ref [kN/m2 ] 100.0
ν ur – 0.2
m – 0.50
σL – 10.0
E0ref [kN/m2 ] 405000
γ 0.7 – 0.0002
Density γD [kN/m3 ] 16
γF [kN/m3 ] 10
e0 – 0.66
Non-linear φ [◦ ] 35
ψ [◦ ] 5
c [kN/m2 ] 1
E50ref [kN/m2 ] 45000
Rf – 0.9
D – 0.25/0.0(HS-std)
ft – 0.0
M – 1.85
H [kN/m2 ] 129305
K0NC – 0.426
pmin
c0 [kN/m2 ] 10.0
Initial K0 state K0x0 /K 0 – 0.43
0z
2 Sand (-20m↓) HS-small Elastic Eur [kN/m2 ] 300000
σ ref [kN/m2 ] 100
ν ur – 0.2
m – 0.55
σL – 10.0
E0ref [kN/m2 ] 675000
γ 0.7 – 0.0002
Density γD [kN/m3 ] 16
γF [kN/m3 ] 10
eo – 0.66
Non-linear φ [◦ ] 38
ψ [◦ ] 6
c [kN/m2 ] 1
E50ref [kN/m2 ] 75000
Rf – 0.9
D – 0.25/0.0(HS-std)
ft – 0.0
M – 2.955
H [kN/m2 ] 128964
K0NC – 0.38
pmin
c0 [kN/m2 ] 10.0
Initial K0 state Kox0 /K 0 – 0.38
oz

Table 5.2: Excavation in Berlin Sand: material properties for the diaphragm wall, anchors and
interfaces
Material Model Data group Properties Unit Value

3 Wall Beams Elastic E [kN/m2 ] 30000000
ν – 0.15
Density Unit weight [kN/m3 ] 24
Geometry Interval [m] 1.0
A [m2 ] 0.8
Iz [m4 ] 0.0426667
4 Anchors Truss Elastic E [kN/m2 ] 210000000
Density Unit weight [kN/m3 ] 0.0
A [m2 ] 0.0015
A [m2 ] 0.0015
A [m2 ] 0.0015
7 Interface Contact Non-linear φ [◦ ] 28
ψ [◦ ] 0
c [kN/m2 ] 0.0

The chart below presents four unloading functions which are defined and associated with
the excavated elements in order to gradually unload each excavated region. Note that the
same unloading functions must be applied to interface elements adjacent to the excavated
continuum. All existence functions and unloading functions which are applied for excavated
zones of sand are shown in the chart below.
30.0m
EXF1 2 4 6 8 t UF1 2 4 6 8 t
2 4 6 8 t UF2 2 4 6 8 t
EXF2
2 4 6 8 t UF3 2 4 6 8 t
EXF3
EXF4 2 4 6 8 t UF4 2 4 6 8 t
Figure 5.3: Excavation in Berlin Sand: unloading functions

(a) (b)
Figure 5.4: Excavation in Berlin Sand: Bending moments and wall deflections at the last stage of
excavation
Remarks:
1. The largest bending moments are generated by HS-standard model due to excessive plastic
soil deformation caused by lack of small strain stiffness. The shape of the M diagrams is
similar for all models.
2. The most significant overshoot is observed in the bottom part of the wall. In the basic
MC model elastic stiffness remains unchanged and insensitive to the current stress state
while HS-standard and HS-small/HS-Brick models exhibit strong stress dependency (cf.
Eq.(2.6)).
3. Predictions of wall deflection by the HS-small/HS-Brick models match in situ measure-

ments. A certain discrepancy is observed at the top part of the wall. This effect can be
explained by the fact that creep neither cracking of concrete is not included.

(a)
(b)
Figure 5.5: Excavation in Berlin Sand: Soil deformations at last stage of excavation
(a) vertical heaving of subsoil, (b) settlements of the ground behind the wall (y = 0 m)
Remarks:
1. The HS-Standard and MC models with variable stiffness generate similar heaving.
2. Vertical heaving generated by the HS-small and HS-Brick models is significantly reduced
with respect to results which are generated by HS-Standard and MC models; the HS-Brick
model yields larger heaving than the HS-small (because of spurious strain history reset)
3. The MC model results in an unrealistic lifting of the retaining wall associated with unloading
of the bottom of an excavation. Settlements behind the wall are realistically generated
with HS-standard and HS-small/HS-Brick models.

5.2. TWIN TUNNELS EXCAVATION IN LONDON CLAY
5.2 Twin tunnels excavation in London Clay
File: HS-Brick-Exc-London-Clay-2phase.inp
This example demonstrates the importance of modeling tunnel construction problems using
advanced constitutive models such as Hardening Soil models which enable one to consider pre-
failure non-linear stiffness. The study highlights the differences in predictions of subsurface
displacements during tunnel excavations in the stiff, heavily overconsolidated London Clay
modeled using:
• Non-linear elastic, perfectly plastic models: HS-Brick and HS-small.
This study reanalyzes the excavation model of the twin Jubilee Line Extension Project tun-
nels beneath St James’s Park (London, UK) which has been reported in the original paper
by Addenbrooke et al. (1997). Predictions of displacements obtained using Hardening Soil
models are additionally compared with the results obtained by Addenbrooke et al. (1997) for
the isotropic non-linear elastic model J4 and field data.
The problem statement, i.e. subsurface stratigraphy and tunnels orientation is presented in
Figure 5.6. The following paragraphs present the analysis details, excavation/construction
stages and the material data assumed in the analyzes.

Analysis details
• analysis type: Plane Strain: Deformation + Flow
• driver type: Consolidation
• mesh: Figure 5.7
• constitutive models:
F Sand: HS-Brick
F Thames Gravel: HS-Brick
F London Clay: HS-Brick
F Woolwich and Reading Bed Clay: M-C
Excavation/construction stages
1. Generating the initial state in subsoil for the assumed K0insitu across the FE mesh presented
in Figure 5.7
2. Adding seepage elements around the westbound tunnel which permits water outflow and
excavation of the westbound tunnel with gradual unloading - 100% unloading within period
of 8 hours
3. Installation of the westbound tunnel lining at 63% of unloading (this corresponds to the
assumed volume loss level of 3.3 % achieved with the HS-Brick model) and removing
seepage elements; parameters for tunnel lining are given in Table 5.7
4. Consolidation during 8.5 months
5. Adding seepage elements around the eastbound tunnel which permits water outflow and
excavation of the eastbound tunnel during 8 hours with gradual unloading within period
of 8 hours
6. Installation of the westbound tunnel lining at 62.8% of unloading and removing seepage
elements; this unloading level corresponds to the assumed volume loss of 2.9% achieved
with the HS-Brick model
7. Consolidation

Figure 5.6: Soil stratigraphy and diagonally oriented tunnels at St James’s Park, London, UK.
Figure 5.7: Finite element mesh

Material data
• Unit weights - see Table 5.3
• Stiffness parameters
F for HS models - E0 , Eur , E50 and γ 0.7 calibrated using laboratory ε1 − q data points
for the isotropically consolidated undrained extension triaxial test (CIUE) at p00 = 750
kPa, as shown in Figure 5.9. The constant m was assumed for London Clay equal to
0.7 which is little lower than the one (0.75) reported in Viggiani and Atkinson (1995).
The stiffness parameters are given in Table 5.4. The E0 value was calibrated to fit as
best as possible the ε1 − q curve for the aforementioned triaxial undrained extension test
and resulting degradation curve.
• Strength and plastic potential parameters - typical values for London Clay (see Table 5.3)
have been adapted from the original paper (Addenbrooke et al., 1997) for all considered
models. The reported very high dilatancy angles (unrealistic in practice) were significantly
reduced for all soil layers.
• Initial state parameters (see Table 5.6)
F the value of the overconsolidation ratio OCR for London Clay was assumed equal to 5.
F although the estimates of K0in situ for the London Clay are typically reported of around
1.5, the value K0 = 1.0 has been adopted in the analysis. It was observed that the
isotropic Hardening Soil models may give incorrect predictions for K0 >> 1.0. The
comparative results produced by model J4 in Addenbrooke et al. (1997) were obtained
for K0 = 1.5.
• Permeability - sand and gravel were modelled as highly permeable materials, whereas clayey
soils were attributed with an anisotropic permeability decreasing with depth, as shown in
Figure 5.8. The fluid bulk modulus was assumed equal to β f = 2.2 GPa.
• Characteristics for the tunnel lining which were adopted after the original paper are sum-
marized in Table 5.7.

Figure 5.8: Permeability profile assumed in the analysis.

Comments:
• The HS-Brick model matches quite well the laboratory data points (in red) both in the range of
very small strains (up to 0.01%) and in the range of small strains (between 0.01 and 0.4%).
(a) Very small strains
(b) Small strains
Figure 5.9: Stress-strain curves: HS-Brick vs experiment (isotropically consolidated undrained ex-
tension test (p00 = 750 kPa)).

Figure 5.10: Variation of the undrained secant stiffness-strain curve ε1 − Esud : HS-Brick model
response vs laboratory data points obtained in the isotropically (CIEU) consolidated
undrained extension tests (p00 = 750 kPa).

Comments:
R
uy (x)dx
• Volume loss is understood here as a fraction ∗ 100% computed at ground surface
π R2
(a) westbound tunnel (volume loss 3.3% achieved at unloading level of 63 %)
(b) eastbound tunnel (volume loss 2.9% achieved at unloading level of

62.8 %)
Figure 5.11: Evolution of the volume loss with the unloading level for the westbound and the
eastbound tunnels

Comments:
• The HS-Brick and HS-small yield a similar shape of surface settlements with the maximum value
of about 15 mm; this result is better than the one obtained using the J4 model which is also
isotropic; this result clearly shows an anisotropic behavior of London clay
Figure 5.12: Surface settlement profiles after excavation of the westbound tunnel: comparison of
different models.
Comments:
• The HS-Brick model yields a better prediction than the HS-small model and the J4 one. Problem
of the stiffness overshooting in the HS-small model is here well visible.
Figure 5.13: Surface settlement profiles after excavation of eastbound tunnel: comparison of differ-
ent models.

(a) (b)
Figure 5.14: Excavation of the westbound tunnel (a) vertical displacements profile at the tunnel
axis (b) vertical displacements profile 4m (left) from the axis.

Table 5.3: Unit weight, permeability, yield surface and plastic potential parameters for Mohr-
Coulomb and HS-models.
Sand Gravel London Clay Woolwich and

Reading Clay
Strength c0 = 1 kP c0 = 0 kPa c0 = 5.0 kPa c0 = 200 kPa
parameters φ0 = 30o φ0 = 35o φ0 = 25o φ0 = 27o
Dilatancy angle ψ = 0o ψ = 5o ψ = 5o ψ = 5o
Bulk unit weight γ dry = 18 γ sat = 20 γ sat = 20 γ sat = 20
kN/m3 γ sat = 20
Permeability coeffi- 10−5 10−4 see Fig. 5.8 see Fig. 5.8
cient m/s
Table 5.4: Stiffness parameters for London Clay, gravel and sands for HS models at the reference
stress σ ref = 100 kPa.
Layer E0 [MPa] γ 0.7 [-] Eur [MPa] E50 [MPa] m [-] ν [-]
London Clay 160 0.0001 18 4 0.7 0.25
Gravel 300 0.0002 100 33 0.5 0.20
Sands 150 0.0002 50 16.7 0.5 0.25
Table 5.5: Stiffness parameters for Woolwich and Reading Bed Clay
Layer E [MPa] ν [-]

Woolwich and Reading clay 800 0.25
Table 5.6: Initial state parameters assumed in this study.
K0NC [-] K0SR [-] K0 [-] OCR [-]

London Clay 0.577 0.577 1.0 5
Sand 0.5 0.5 0.5 1.1
Gravel 0.5 0.5 0.5 1.1
Woolwich and Read- 1.0 -
ing Clay
Table 5.7: Tunnel lining characteristics.
Young’s mod- Poisson’s ra- Cross sectional Momentum of Lining-soil interface fric-
ulus E tio υ area A inertia Iz tion angle φi
28 GPa 0.15 0.168 m2 /m 3.95136·10−4 20o
m4 /m

5.3 Spread footing on overconsolidated Sand
File: HS-small-Footing-Texas-Sand-2phase.inp
File: HS-Brick-Footing-Texas-Sand-2phase.inp
File: MC-Footing-Texas-Sand-2phase EurVar.inp
A verification of the HS-small/HS-Brick models for a square footing load test is demonstrated
in this section. This load test examines the model in terms of its sensitivity to loading and
unloading modes applied to overconsolidated sand. This simple boundary value problem also
illustrates:
• Combined parameter identification from triaxial tests and dilatometer tests
• Interpretation of laboratory and in situ measurements
• Imposing soil stress history through the q POP (obtaining a variable OCR profile)
• Imposing a variable K0 profile
• Sensitivity of numerical prediction to small-strain stiffness parameters
• Sensitivity of numerical prediction to initial stress setup
The study case demonstrates main differences between HS-standard, HS-small, HS-Brick and
standard Mohr-Coulomb (MC) for the analyzes of footing.
Problem statement.
A number of load tests on square footing were performed at Texas A&M University’s National
Geotechnical Site (Briaud and Gibbens, 1997). In this example, the measurements derived
from a load test of 3x3m ”North” footing are compared to numerical predictions. The test
setup and soil stratigraphy is presented in Figure 5.15. The subsoil mainly consists of sandy
clay to silty sand layers and it has been confirmed by the interpreted DMT data (see Figure
5.21(a)). Vertical displacements were measured at the corners of the square footing.

5.3. SPREAD FOOTING ON OVERCONSOLIDATED SAND
Figure 5.15: A draft of the test setup and soil stratigraphy at A&M University’s National Geotech-
nical Site in Texas.
Parameter identification from a triaxial test. The parameters for the HS model have
been determined from available triaxial compression tests for two sampling depths 0.6m and
3.0m, and three confining pressures 34.5, 138 and 345 kPa. The results from several reso-
nant column tests with the confining pressure 100kPa have been used to evaluate the order of
magnitude of small strain characteristics. The reader can analyze parameter identification in
the spreadsheet presented in Table 5.8. Note that the determination of φ0 and c0 was carried
out only for 34.5kPa and 138kPa tests since the ultimate state (failure deviatoric stress qf )
was not achieved at 345kPa.
A comparison of numerical results and experimental data derived from triaxial compression
tests is presented in Figure 5.17. Note that the preconsolidated state of soil specimens was
taken into account in the triaxial test simulations by applying the initial value of the minimal
preconsolidation pressure pc0 = 250kPa. The value was evaluated assuming that:
σ 0v0 = 14.9kN/m3 × 3.0m = 44.7kPa
and applying an estimated q POP (see Figure 5.20):
σ ’SR
y = σ 0vc = σ 0v0 + q POP = 44.7kPa + 350kP a = 394.7kP a cf. Eq.(2.28a)
0
σ ’SR SR
x = σ v0 · K0 = 394.7kPa ×0.41 = 161.8kPa

pc0 = (2σ ’SR ’SR

x + σ y )/3 ≈ 250kPa
Note that prescribing the initial preconsolidation state may have an influence on the results
of the numerical simulation of the triaxial compression test. In the presented example, the
volumetric mechanism is not activated during the triaxial compression test for the confining
pressures 34.5kPa and 138kPa because the specimens are preconsolidated, i.e. the initial
stress state p00 is largely inferior with respect to p0c0 . On the other hand, the simulation for
345kPa is affected by both shear and cap mechanisms as the specimen is normally consoli-
dated. In the other words, one may expect less stiffness at 50% of qf , where the volumetric
straining also occurs. Note that in the case of the full scale simulation of the footing load
test, the initial preconsolidation state will be prescribed through the preoverburden pressure
q POP which imposes the preconsolidation pressure with respect to the initial effective vertical
stress σ 0v0 (see Figure 5.20).
Since the results derived from resonant columns present a considerable scatter (even for the
same confining pressure 100kPa, see Figure 5.17 in the right hand bottom corner), they
were used to evaluate the first guess of small-strain stiffness parameters. It is commonly
recognized that the small-strain stiffness derived from in situ seismic probes is typically larger
compared to that measured with laboratory devices. This is typically attributed to specimen
disturbances during soil sampling.
The results derived from numerical simulations presented in Figure 5.17 illustrate the sensitiv-
ity analysis of the HS-small/HS-Brick model to two parameters E0 and γ 0.7 which define the
model behavior for very small amplitudes of shear strain. It can be noticed that a consider-
able scatter of the results derived from resonant columns does not allow performing a precise
parameter calibration, especially for parameter γ 0.7 . Considering probable disturbances due
to sampling, the simulations of triaxial tests were conducted for lower small-strain stiffness
parameters, whereas higher values of parameters were assumed to simulate in situ load test.
(a) Identification of slope m (b) Identification of slope M ∗ and in-

tercept c∗
Figure 5.16: Interpretation of triaxial compression test data from A&M test site (sampling depths:
0.6m and 3.0m, refer to Table 5.8).

Table 5.8: Parameter identification spreadsheet for the triaxial test - spread footing benchmark.
Sample 0.6m 3.0m Ref.

Confining pressure σ 3 [kPa] 34.5 138 345 34.5 138 345 Fig.5.17
Ultimate dev. stress qf [kPa] 112.3 417.7 994.4# 114.9 417.9 994.9#
Vertical stress σ 1 [kPa] 146.8 555.7 1339.4 149.4 555.9 1339.9
Mean stress at failure [kPa] 71.9 277.2 676.5 72.8 277.3 676.6
# qf not achieved, identification of φ and c for 34.5kPa and 138kPa
Identification of φ and c Mean

Slope M ∗ [-] 1.488 1.482 Fig.
5.16(b)
Intercept in p0 − q plane c∗ [kPa] 5.262 7.06 Fig.
5.16(b)
Friction angle φ [deg] 36.6 36.5 36.5 cf. Fig.3.1
Cohesion c [kPa] 2.6 3.5 3.1 cf. Fig.3.1
Identification of Rf and E50 (σ 3 ) Mean

Slope a [-] 8.48e- 2.12e- 8.82e- 8.40e- 2.11e- 8.75e- Fig.5.18
03 03 04 03 03 04
Intercept b [-/kPa] 5.01e- 3.68e- 1.57e- 4.90e- 3.68e- 1.65e- Fig.5.18
05 05 05 05 05 05
Failure ratio Rf (= b · qf ) [-] 0.95 0.89 0.88 0.97 0.88 0.87
Secant modulus E50 (= 1/2b) [kPa] 9988 13605 31786 10210 13576 30358
ref and m
Identification of E50 Mean
ln E50 9.21 9.52 10.37 9.23 9.52 10.32
Reference stress σ 3 [kPa] 100 100 100
a = (σ 3 + c cot φ)/(σ ref + c cot φ) 0.367 1.367 3.366 0.375 1.363 3.339
ln a -1.001 0.313 1.214 -0.981 0.309 1.206
Stiffness exponent m [-] 0.521 0.494 0.51 Fig.5.16(a)
Ref. secant modulus E50 ref [kPa] 16013 15821 15917
Identification of ψ Mean
Slope d 4.88e- 3.92e- 9.86e- 4.82e- 3.70e- 1.11e-
02 02 03 02 02 02
Dilatancy angle ψ [deg] 1.37 1.10 1.35 1.04 1.27

Figure 5.17: Triaxial compression test data from A&M test site (sampling depths: 0.6m and 3.0m)
and numerical results for HS-small and HS-standard. Parameters for HS-Std model
were derived from parameter identification presented in Table 5.8; the results for dif-
ferent small-strain parameters are compared to the results from a couple of resonant
column tests (right hand bottom corner: γ s = ε1 − ε3 and G50 = ∆q/2∆γ s ). The
other parameters used for numerical simulations are summarized in Table 5.9.

Figure 5.18: Interpretation of triaxial test data in ε1 − ε1 /q plane: samples 0.6m and 3.0m.
Note that slope a is used to calculate Rf whereas the intercept b is used to compute
E50 as shown in Figure 3.3 (sampling depths: 0.6m and 3.0m, refer to Table 5.8).

Parameter estimation from the dilatometer test.

Records derived from two dilatometer tests (DMT-1 and DMT-2) were used for profiling silty
sand. The field tests, i.e. CPT, pressuremeter and Marchetti’s dilatometer, showed that
the silty sand site is overconsolidated due to removal of an overburden surcharge and soil
aging. Estimation of two initial state variables OCR and K0 was carried out based on the
interpreted results of the horizontal stress index KD (see Figure 5.19(a)). It can be noticed
that the OCR decreases with increasing depth which is characteristic to superficial layers of
subsoil which may be directly subject to mechanical unloading such as erosion, excavations,
changes in ground water level, or due to other phenomena such as dessication or formation
of particle bonds.
In addition, the profile of the effective friction angle φ0 which has been obtained applying
empirical correlations for DMT data, was compared to φ0 derived from the triaxial test (see
Figure 5.19(b)). The following correlations were used for parameter profiling:
• OCR: Eq.(3.86) with the exponent value for low plastic materials equal to 1.67,
• K0 : Eq.(3.95) with the exponent value for low plastic materials equal to 0.64,
• Eur : Eq.(3.25),
• φ0 : upper bound Eq.(3.32), lower bound Eq.(3.33), and the mean value as the average of
both.
While the profile φ0 derived from DMT was used to verify the values obtained with triaxial
tests, OCR and K0 profiles assumed in the model were directly interpreted from in situ data.
As regards OCR, the variable profile which is illustrated in Figure 5.20(a) was obtained based
on the effective vertical stress by applying preoverburden pressure q POP , see Figure 5.20(b):
σ ’SR
y = σ 0vc = σ 0v0 + q POP cf. Eq.(2.28a)
and
OCR = σ 0vc /σ 0v0
where:
σ 0vc - preconsolidation pressure
σ 0v0 - effective vertical stress
Clearly, the stress history (soil overconsolidation) could be also obtained by applying and
removing a surcharge before the installation of the footing. However, such manipulation
leaves behind a strain history. Typically, a strain history in natural deposits may be erased
relatively fast due to stress relaxation and soil aging effects such as cementing of soil particles.
Therefore, most boundary value problems should be started from zero initial strains which is
a default setting in ZSoilr . Applying the stress history through q POP option allows the user
to account for the overconsolidation effect (variable OCR typically observed at superficial
layers of subsoil) with zero strains at the beginning of the analysis.
As regards K0 , the profile assumed in the model was obtained by fitting DMT interpretation,
as shown in Figure 5.22.

(a) (b)
Figure 5.19: Interpretation of dilatometer test data: a) profile of horizontal stress index KD , b)
DMT-based profiling of the effective friction angle compared to φ0 derived from the
triaxial test

(a) (b)
Figure 5.20: Initial stress state profiling: a) OCR profile derived from DMT data and OCR profiles
by applying different q POP (in the reference simulation q POP = 350kPa has been
considered), b) the vertical effective stress and the preconsolidation stress profiles
assumed in the analysis.

(a) (b)
Figure 5.21: Dilatometer test data: a) profile of material index ID , b) profile of dilatometer modulus
ED (both profiles have been used to determine Eur profile based on an empirical
solution given in Eq.3.25).

(a) (b)
Figure 5.22: Profiles interpreted from DMT data and assumed profiles for in FE model: a) coefficient
K0 , b) stiffness modulus interpreted from DMT data, stiffness moduli assumed for
HS-small/HS-Brick (E50 , Eur and E0 ) and the Young modulus E assumed for the
Mohr-Coulomb model.

Material data. Material data which was assumed for numerical simulation of footing test
loading are summarized in Table 5.9.
• Physical properties - taken from the original report (Briaud and Gibbens, 1997)
• Small strain properties - evaluated from the resonant column tests
• Deformation characteristics - E50 derived from triaxial compression test data, and Eur
double-checked with DMT data
• Strength characteristics - derived from triaxial compression test data
• Initial state variables - evaluated from DMT data
Table 5.9: Model parameters used in simulations of the spread footing at Texas site.
Hardening-Soil Mohr-Coulomb
Parameter symbol Unit Sand (triaxial Sand (in situ Sand (in situ
test) simulation) simulation)
Physical properties
γD [kN/m3 ] - 14.9 14.9
e0 [-] 0.75 0.75 0.75
γ0 [kN/m3 ] - 9.2 9.2
emax [-] 0.91 0.91 0.91
Small strain properties

E0 [kPa] 260000 380000 -
γ 0.7 [kN/m3 ] 0.0002 0.00045 -
Deformation characteristics
Eur [kPa] 70000 70000 imposed pro-
file for E:
Fig.5.22(b)
E50 [kPa] 16000 16000 -
σ ref [kPa] 100 100 -
ν [-] 0.2 0.25 0.3
m [-] 0.51 0.51 -
Strength characteristics
φ [o ] 36.5 36.5 36.5
c [kPa] 3 3 3
ψ [o ] 1.3 1.3 1.3
Rf [-] 0.92 0.92 -
Oedometric test characteristics

K0NC [-] 0.41 0.41 -
Eoed [kPa] 20000 16000 -
σ ref
oed [kPa] 100 247 -
Initial state variables

q POP [kPa] - 350 -
pc0 [kPa] 250 - -
K0 (in situ) [-] 1 variable variable
K0SR [-] 1 0.41 -

Numerical simulation of the footing problem. A 3D model of the footing problem which
is shown in Figure 5.23 was considered in the analysis. The 3D mesh can represent a quarter
of the real setup thanks to the two symmetry planes. The mesh includes:
• simplified stratigraphy of the subsoil deposits consisting of one sand layer; mesh dimension
19.8m×19.8m×17m
• quarter of 3x3m footing which is embedded in the soil at 0.76m and its thickness is equal
to 1.22m
• interface between footing faces and soil
• external displacement boundary conditions (BC) (box type)
• pressure BCs which are applied via fluid head and set up along the external boundaries;
ground water level is set to 4.9m below soil surface
• nodal force representing a hydraulic jack (due to two symmetry planes the applied nodal
is equal to 0.25 × F
In addition, Figure 5.24 illustrates how to impose a variable profile of K0 by means of the
Initial Stresses option. Note that an existence function is attributed to the superele-
ments describing the initial stresses. It means that the effect of imposed initial stress applies
only to the first analysis step, i.e. during the generation of the initial state. This intervention
is needed to avoid imposing soil’s initial stress onto the material which replaces excavated soil.
Analysis details
• analysis type: 3D deformation+flow
• driver type: steady state
• constitutive models:
F Sand: four variants: HS-standard, HS-small, HS-Brick, Standard Mohr-Coulomb

F Footing: elastic material E = 20GPa, ν = 0.2
The following stages are considered in the model:
1. Generating an initial stress state for an assumed K0in situ in sand layer and initial pore
pressure BC
2. Installation of the square footing with simultaneous replacement of soil’s material in the
embedded part of footing
3. Gradual application of test load at the square footing

Figure 5.23: 3D mesh representing a quarter of 3x3m footing.
Figure 5.24: Imposing a variable K0 profile by means of the Initial Stresses option.

Analysis of results. The following paragraph presents the results derived from numerical
simulations.
Figure 5.25(a) shows that the HS-small/HS-Brick satisfactorily reproduce the overall load
settlement curve. The charts also show that failing to account for small strain stiffness by
using the HS-standard model may lead to overestimation of settlements in overconsolidated
sand.
Figure 5.25(b) demonstrates the incapability of the standard Mohr-Coulomb model to realis-
tically reproduce the evolution of settlements under mixed loading conditions. In this analysis,
an imposed profile of Young modulus E corresponding to Eur has been chosen to describe
stiffness of the M-C model (see Figure 5.22(b)).
(a) HS-small/HS-Brick vs HS-standard (b) HS-small/HS-Brick vs Mohr-Coulomb
Figure 5.25: Experimental and computed load-settlement results for 3x3m footing test on sand at
A&M site: comparison of models.

(a) HS-small/HS-Brick: sensitivity to E0 (b) HS-small/HS-Brick: sensitivity to γ 0.7
A&M site: sensitivity to small-strain stiffness parameters.
Figure 5.26(a) demonstrates sensitivity of HS-small/HS-Brick to E0 . It can be noticed that

underestimating small-strain stiffness may lead to an overestimation of settlements, affecting
especially the initial part of the load-settlement curve.
Figure 5.26(b) illustrates the sensitivity of HS-small/HS-Brick to the small-strain threshold
parameter γ 0.7 . It can be noticed that the lower value of this parameter allows an earlier
degradation of small-strain stiffness.

(a) HS-small/HS-Brick: sensitivity to the initial (b) HS-Brick: sensitivity to q POP

stress setup
A&M site: sensitivity to the initial stress setup.
Figure 5.27(a) illustrates the sensitivity of numerical predictions to the initial setup of the
coefficient of earth pressure ”at rest”. In the reference simulation, the profile of K0 for over-
consolidated soil has been evaluated by means of the Marchetti’s dilatometer test (DMT),
whereas in the second simulation a constant profile of K0 equal to K0NC = 1 − sin(φ) has
been assumed. This analysis reveals slightly larger settlements for K0NC = 0.41 which is at-
tributed to lower initial soil stiffness. Note that soil stiffness in the HS model depends on the
minor effective stress so in the case where K0 = K0NC the initial stiffness is defined by σ 0h0 ,
whereas for the reference simulation, the initial soil stiffness is defined by σ 0v0 up to around
2.2m as K0 > 1. The user should be aware that soil stiffness in the HS model evolves during
simulation with amplitudes of stress level.
Finally, Figure 5.27(b) reveals the sensitivity of a numerical simulation to the initial precon-
solidation setup. The chart presents numerical predictions for three q POP values: 350kPa
(reference simulation), 300kPa and 400kPa. The sensitivity of the OCR profile to the speci-
fied values of q POP is presented in Figure 5.20(a).

Appendix A
Determination of undrained shear

strength
A.1 Non-uniqueness of undrained shear strength
It has been widely recognized that the in situ behavior of soils may be significantly different
from that of laboratory samples. This can be mainly attributed to the quality of the intact
specimens which may depend on drilling and sampling methods and sample geometry (DeG-
root and Sandven, 2004). The disturbance of samples may increase during their insertion into
a sampling tube, transportation, relaxation of stresses, drying, temperature changes, trimming
and, finally, their installation in the testing cells, etc. (Hight et al., 1992). Different sampling
devices such as piston samplers, thin walled tubes or downhole block samplers can provide
specimens for which different magnitudes of preconsolidation pressure or undrained shear
strength su are measured (e.g. Hight et al., 1992; Tanaka and Tanaka, 1999). Experience
shows that sample disturbance may lead to underestimation of the apparent preconsolidation
pressure or undrained shear strength (Karlsrud, 1999; Fioravante, 2004).
Inconsistencies in su values measured by laboratory and field tests may also stem from the
non-uniqueness of this property. The undrained shear strength su is not a unique soil pa-
rameter (Wroth, 1984; Jamiolkowski et al., 1985), as it depends on the type of test, which
involves particular strain paths (cf. Figure 1.1).
The differences in interpreted results also stems from the time-dependent behavior of soils
(e.g. Vaid and Campanella, 1977; Leroueil, 1988; Sheahan et al., 1996; Penumadu et al.,
1998). The undrained shear strength increases linearly with the logarithm of the shear strain
(Bjerrum, 1972; Nakase and Kamei, 1986). For instance, the testing speed for SBPT or
CPTU can be one or more orders of magnitude greater than that used in the triaxial com-
pression test TC (ε̇ = 0.01%/min), see Figure 1.2. Laboratory tests with the use of a model
pressuremeter in clays have revealed an increase of su of about 10% for every tenfold increase
of strain rate (Prapaharan et al., 1989). While the overstimation of su derived from pres-
suremeter test due to the strain rate effect can be reasonably small, in the order of 10-20%,
the differences for CPTU can be much larger. The extrapolated results of Bjerrum (1972)
(Figure 1.2) can indicate that neglecting the strain effect in the analysis of penetration may
lead to the considerable overestimation of su of about 40% with respect to the value obtained
for the conventional triaxial compression. Since the undrained shear strength is a function of
the stress history, the similar effects can be observed for the derived values of preconsolida-
tion pressure σ 0p . The study carried out by Leroueil et al. (1983b) revealed the increase of
APPENDIX A. DETERMINATION OF UNDRAINED SHEAR STRENGTH
(a) (b)
Figure 1.1: Undrained shear strength in normally consolidated soil (a) as a function of shear modes
for various tests: triaxial undrained compression tests (CIUC and CK0 UC), plain strain
compression test (PSC), direct simple shear test (DSS) and field vane test (FVT) (after
Wroth, 1984), (b) profiles for field and laboratory tests on Onsøy clay (from Lacasse
et al., 1981).
.
about 10-14% for σ 0p per log cycle of volumetric strain rate ε˙v in the constant rate of strain
oedometer test (CRS).
1.5 TEST
CPTU
SBPT
TC
STRAIN
C
RATE ed
lat
PS
1.4 o
ap
8)
Vaid & Campanella tr

ex
99
(1977) CIUC
(1
1.3
al.
et
UC
cu / cu,0.01
du
Prapaharan et al.(1989) K
ma
o
1.2 PMT )C
72
nu
19
Pe
(
m
1.1 ru
B jer
1
Nakase & Kamei (1986) CKoUC
0.9 M10
M15
M30
0.8 −3 −2 −1 0 1 2 3
10 10 10 10 10 10 10
Strain Rate, ε [%/min]
Figure 1.2: The strain rate effect on undrained shear strength su for different shear modes and
a schematic comparison of test strain rates for triaxial compression (TC), self-boring
pressuremeter (SBPT) and piezocone (CPTU).

A.2. DETERMINATION OF SU FROM FIELD TESTS
A.2 Determination of su from field tests
FVT. The interpretation of su from the standardly used field vane test (see Figure 1.3) can
be carried out with the conventional formula:
6M M
su = 3
= 0.2728 3 (1.1)
7πD D
in which M is the maximum recorded torque, and D is the diameter of vane.
Since the values of su obtained with the above equation can be too conservative, Chandler
(1988) suggested increasing the factor 0.2728 to 0.2897.
Figure 1.3: Standard dimensions of the most commonly used field vane test (from Chandler, 1988).
DMT. The values of su can be correlated with the Marchetti’s dilatometer data through the
original formula suggested by Marchetti (1980):
su = 0.22σ 0v0 (0.5KD )1.25 (1.2)
where KD is the horizontal stress index which is calculated based on the first dilatometer
reading p0 (KD = (p0 − u0 )/σ 0v0 ).

APPENDIX A. DETERMINATION OF UNDRAINED SHEAR STRENGTH

Appendix B
Estimation of compression index
Table B.1: Some correlation equations for estimating consolidation parameters (after Holtz et al.,
1986; Bowles, 1997; Kempfert, 2006).
Compression index Applicability Reference

Cc = 0.009(wL − 10) Normally consolidated clays Terzaghi and Peck
of low to medium sensitivity (1967)
and wL < 100(±30% er-
ror)
Cc = 0.141Gs (γ SAT /γ D ) All clays Rendon-Herrero
(1983)
Cc = 0.141G1.2
s [(1 + e0 )/Gs ]
2.38
All clays (94 data points) Rendon-Herrero
(1983)
Cc = 1.15(e0 − 0.35) All clays, uniformly packed Nishida (1956)
Cc = 1.15(e − 0.91) All clays, loosely packed Nishida (1956)
e > 0.9
Cc = 0.009wn + 0.005wL All clays Koppula (1986)
Cc = 0.00234wL Gs All inorganic clays Nagaraj and Murthy
(1985)
Cc = 0.37(e0 + 0.003wL + 0.0004wn − 0.34) 678 data points Azzouz et al. (1976)
Cc = 0.00917(wL − 13) 56 data points Mayne (1980)
Cc = −0.156 + 0.411e + 0.00058wL 72 data points Al-Khafaji and An-
dersland (1992)
Cc = 0.156e + 0.0107 Remoulded all clays Azzouz et al. (1976)
Cc = 0.208e + 0.0083 Chicago clays Azzouz et al. (1976)
Cc = 0.007(wL − 10) Remoulded clays Skempton (1944)
Cc = 0.007(wL − 7) Remoulded clays Azzouz et al. (1976)
Cc = 0.005Gs IP Remoulded clays Wroth and Wood
(1978)
Cc = 0.5(IP /100)Gs Remoulded clays Wroth and Wood
(1978)
Cc = 0.046 + 0.0104IP Best for IP < 50% Nakase (1988)
Cc = 1.15e Deformable but incom- Nishida (1956)
pressible soil
Cc = 0.0115wn Organic soils - meadow Azzouz et al. (1976)
mats, peats, and organic
silt and clay
Cc = 0.75(e0 − 0.50) Soils with low plasticity Azzouz et al. (1976)
Cc = 0.30(e0 − 0.27) Inorganic silty sand-silty Hough (1969)
clay
Cc = PI/74 Data from different soils Kulhawy and Mayne
(1990)
APPENDIX B. ESTIMATION OF COMPRESSION INDEX
Cc = 0.0093wn Chicago clays and Alberta Koppula (1981)

Province in Canada (109
data points)
Cc = 17.66/100000wn2 + 0.00593wn − 0.135 Chicago clays Azzouz et al. (1976)
Cc = 0.015(wn − 8) Soils in Taipei Moh et al. (1989)
Cc = 0.54(e − 0.23) Soils in Taipei Moh et al. (1989)
Cc = 0.0046(wL − 9) Brazilian clays Azzouz et al. (1976)
Cc = 1.21 + 1.005(e − 1.87) Motley clays from Sao Azzouz et al. (1976)
Paulo
Cc = 0.0037(wL + 25.5) Cincinnati and Northern Dayal (2006)
Kentucky
Cc = 0.0135wn − 0.1169 Cincinnati and Northern Dayal (2006)
Kentucky
Cc = 0.0042IP + 0.165 Cincinnati and Northern Dayal (2006)
Kentucky
Cc = 0.46e − 0.049Gs + 0.0023 Cincinnati and Northern Dayal (2006)
Kentucky
Cc = 0.4965e − 0.0014wn − 0.123 Cincinnati and Northern Dayal (2006)
Kentucky
Cc = −0.247e + 0.004wL + 0.01wn + 0.021 Cincinnati and Northern Dayal (2006)
Kentucky
Cc = 0.0018/(1 − 0.0109n) Remoulded clays in Japan Park and Koumoto
(Ariake) (2004)
Cc = 0.00269/(1 − 0.0115n) Undisturbed clays in Japan Park and Koumoto
(Ariake) (2004)
Cc = 0.02 + 0.014IP North Atlantic clays Nacci et al. (1975)
Cc = 0.02 + 0.014IP North Atlantic clays Nacci et al. (1975)
Cc = 0.011(wL − 6.36) East coast of Korea Yoon et al. (2004)
Cc = 0.01(wn + 2.83) East coast of Korea Yoon et al. (2004)
Cc = 0.39(e − 0.13) East coast of Korea Yoon et al. (2004)
Cc = −0.16γ D + 2.4 East coast of Korea Yoon et al. (2004)
Cc = 0.0098wn + 0.194e − 0.0025IP − 0.256 East coast of Korea Yoon et al. (2004)
Cc = 0.012(wL + 16.4) South coast of Korea Yoon et al. (2004)
Cc = 0.013(wn − 3.85) South coast of Korea Yoon et al. (2004)
Cc = 0.54(e − 0.37) South coast of Korea Yoon et al. (2004)
Cc = −0.0003wn + 0.538e + 0.002wL − 0.3 South coast of Korea Yoon et al. (2004)
Cc = 0.165 + 0.01IP South coast of Korea Yoon et al. (2004)
Cc = 0.01(wL − 10.9) West coast of Korea Yoon et al. (2004)
Cc = 0.011(wn − 11.22) West coast of Korea Yoon et al. (2004)
Cc = 0.37(e − 0.28) West coast of Korea Yoon et al. (2004)
Cc = −0.066γ D + 1.15 West coast of Korea Yoon et al. (2004)
Cc = 0.0038wn + 0.12e + 0.0065wL − 0.248 West coast of Korea Yoon et al. (2004)

Table B.2: Some correlation equations for estimating recompression index (after Holtz et al., 1986;
Bowles, 1997; Kempfert, 2006).
Recompression index Regions of applicability Reference

Cs = 0.000463wL Gs
Nagaraj and Murthy
(1985)
Cs = 0.00194(PI − 4.6) Best for PI< 50%
Nakase (1988)
Cs = PI/370 Data from different soils
Kulhawy and Mayne
(1990)
Cr /Cc is observed between 0.05 and 0.5, with typical values between 0.1-0.25 and the average
0.2 (cf. Mayne, 1980, 2007; Kempfert, 2006); lower values are observed for cemented soils
PI - plastic index in [%], wL - liquid limit in [%], wn natural water content in [%] .

APPENDIX B. ESTIMATION OF COMPRESSION INDEX

Appendix C
Estimation of shear wave velocity
Table C.1: Typical values of shear wave velocity and density for different geomaterials (after
Lavergne, 1986).
Shear wave velocity Dry unit weight

Vs [m/s] γ [g/cm3 ]
Soil type Min Max Min Max
Screes, organic topsoil 100 300 1.7 2.4
Dry sands 100 500 1.5 1.7
Wet sands 400 1200 1.9 2.1
Clays 200 800 2.0 2.4
Marls 750 1500 2.1 2.6
Sandstones 1200 2800 2.1 2.4
Lime stones 2000 3300 2.4 2.7
Chalk 1100 1300 1.8 2.3
Salt 2500 3100 2.1 2.3
Anhydrite 2200 3100 2.9 3.0
Dolomite 1900 3600 2.5 2.9
Granite 2500 3300 2.5 2.7
Basalte 2800 3400 2.7 3.1
Carbon 1000 1400 1.3 1.8
Ice 1700 1900 0.9 0.9
APPENDIX C. ESTIMATION OF SHEAR WAVE VELOCITY
Table C.2: Typical values of shear wave velocity for different geomaterials (after FOWG, 2003).
Shear wave velocity

Vs [m/s]
Soil type Min Max
Top soil layers (3 to 6m), lightly compact., desagregated, unsat. 110 480
Ballast (gravelly or sandy), unsaturated 220 450
Ballast, saturated 400 600
Ballast cemented 1000 1500
Silt from the lake bottom, not completely saturated 290 540
Silt from the lake bottom, saturated 390 530
Silt at banks, unsaturated 120 400
Moraine 500 1150
Loess 150 300
Marl and mollase sandstone, soft, desagregated 520 1050
Marl , not desagregated 1000 1900
Sandstone, hard 1100 2200
Molasse at plateau 600 2500
Schist 1100 3100
Limestone 1800 3700
Gneiss 1900 3500
Granite 2500 3900
Table C.3: Typical values of shear wave velocity for different geomaterials (after Lindeburg, 2001).
Shear wave velocity

Vs [m/s]
Soil type Min Max
Hard rocks >1400
Firm to hard rocks 700 1400
Gravelly soils and soft to firm rocks 375 700
Stiff clays and sandy soils 200 375
Soft soils 100 200
Very soft soils 50 100
Table C.4: Typical values of shear wave velocity for different geomaterials (after NAVFAC, 1986).
Shear wave velocity

Vs [m/s]
Soil type Min Max
Hard rock 1500
Rock 760 1500
Very dense soil and soft rock (N60 > 50, su > 100kPa) 360 760
Stiff soil (15 > N60 < 50, 50 < su < 100kPa) 180 360
Soft soil (N60 < 15, su < 50kPa) 180
Any soil with PI> 20%,w> 40%, su < 25kPa site eveluation

SPT. Most published correlations for SPT are based on uncorrected N-values. These corre-
lations are given in:
• for all soil types: Table C.5 (insight prrovided in 3.1),
• for sands: Table C.6 (Figure 3.2),
• for silts: Table C.7 (Figure 3.3),
• for clays: Table C.8 (Figure 3.4).
Table C.5: SPT-Based correlations for all type of soils.
Reference Country Soil Type Equation Correlation

coeffi-
cient
Ohba & Toriumi (1970)1 Japan Alluvium Vs 0.31
= 84N60 –
Fujiwara (1972)2 Japan – Vs 0.337
= 92.1N60 –
Ohsaki & Iwasaki (1973)1 Japan – Vs 0.39
= 81.3N60 0.886
0.333
Japan – Vs = 76N60 –
Imai and Yoshimura
(1975)
Imai (1977)2 Japan Quaternary and Pleis- 0.337
Vs = 91N60 –
tocene Alluvium
0.348
Japan Quaternary and Pleis- Vs = 85.34N60 0.719
Ohta and Goto (1978) tocene Alluvium
Imai & Tonouchi (1982)1 Japan – 0.314
Vs = 97N60 0.868
0.5
– – Vs = 56N60 –
Seed et al. (1983)
Shanghai Soft Holocene Deposits Vs = 116.1(N60 + 0.7
Jinan (1987) 0.3185)0.202
0.36
Greece – Vs = 107.6N60 –
Athanasopoulos (1995)
Sisman (1995)2 – – Vs 0.51
= 32.8N60 –
Iyisan (1996)2 – – Vs 0.516
= 51.5N60 –
Kiku et al. (2001)2 Turkey – Vs 0.292
= 68.3N60 –
0.308
Turkey Quaternary Alluvium Vs = 90N60 0.73
Hasancebi and Ulusay and Detritus
(2007)
1
Referenced by Sykora (1987)
2
Referenced by Hasancebi and Ulusay (2007)

Table C.6: SPT-Based correlations for sands.

coeffi-
cient
0.54
– – Vs = 31.7N60 –
Shibata (1970)
0.36
Japan – Vs = 87.2N60 –
Ohta et al. (1972)
Vs = 80.6N60 –
tocene Alluvium
Ohta & Goto (1978b)1 Japan Quaternary and Pleis- 0.333
Vs = 88.4N60 0.719
tocene Alluvium
Imai & Tonouchi (1982)1 Japan Quaternary and Pleis- 0.314
Vs = 87.8N60 0.69
tocene Alluvium
Sykora & Stokoe (1983)1 – – 0.29
Vs = 100.5N60 0.84
0.3
Japan – Vs = 125N60 –
Okamoto et al. (1989)
0.49
Taiwan – Vs = 57.4N60 0.62
Lee (1990)
Taiwan – Vs = 157.13 + 4.74N60 0.691
Lee (1992)
0.178
Greece Alluvium Vs = 145N60 0.70
Pitilakis et al. (1999)
0.319
Turkey Quaternary Alluvium Vs = 90.82N60 0.65
(2007)
0.33
Western – Vs = 73N60 –
Dikmen (2009) central
Anatolia,
Turkey
1
2
Table C.7: SPT-Based correlations for silts.

coeffi-
cient
0.32
Taiwan – Vs = 105.64N60 0.73
Lee (1990)
Taiwan – Vs = 103.99(N60 + 1)0.334 0.798
Lee (1992)
0.178
0.36
Anatolia,
Turkey

Table C.8: SPT-Based correlations for clays.

coeffi-
cient
Vs = 80.2N60 –
tocene Alluvium
Ohta & Goto (1978b)1 Japan Quaternary and Pleis- 0.333
Vs = 86.9N60 0.719
tocene Alluvium
Imai & Tonouchi (1982)1 Japan Quaternary and Pleis- 0.274
Vs = 107N60 0.721
tocene Alluvium
0.31
Taiwan – Vs = 114.43N60 0.62
Lee (1990)
Taiwan – Vs = 138.36(N60 + 1)0.242 0.695
Lee (1992)
0.445
Greece Vs = 76.55N60 –
Athanasopoulos (1995)
0.271
0.269
Turkey Quaternary Alluvium Vs = 97.89N60 0.75
(2007)
0.48
Anatolia,
Turkey
1
2
Figure 3.1: Comparison of ”all soil type”-correlations for estimating Vs from SPT data (correla-
tions are given in Table C.5).

Figure 3.2: Comparison of -correlations for estimating Vs in sands from SPT data (correlations are
given in Table C.6).
Figure 3.3: Comparison of -correlations for estimating Vs in silts from SPT data (correlations are

Figure 3.4: Comparison of -correlations for estimating Vs in clays from SPT data (correlations are
CPT. A number of correlations can be used to estimate shear wave velocity based on CPT
data. Correlations were developed for a generic soil types: Table C.9, for sands: Table C.10,
for clays: Table C.11. Graphical comparisons of these correlations are presented in Figure for
sands
Table C.9: CPT-based correlations for all soil types.
Reference Soil Type Vs [m/s] Remarks

sands, silts, (10.1 log qt − 11.4)1.67 (fs /qt · 100)0.3 qt , fs in kPa
Hegazy and Mayne (1995) clays, mixed
soil types
saturated 118.8 log fs + 18.5 fs in kPa
Mayne (2006a) clays, silts,
and sands
Table C.10: CPT-based correlations for sands.

Giola Taura Sand w/Gravel, 277qt0.13 σ 00.27
v0 qt , σ 0v0 in MPa
Baldi et al. (1989) Po River sands
– 13.18qt0.192 σ 00.179
v0 qt , σ 0v0 in kPa
Hegazy and Mayne (1995)
– 12.02qt0.319 fs−0.0466 qt , fs in kPa

Table C.11: CPT-based correlations for clays.

soft to firm to stiff in- 9.44qt0.435 e0−0.532 qt in kPa
Mayne and Rix (1993) tact clays to fissured
clays, Piedmont silts
– 1.75qt0.627 qt in kPa, n = 481, R =
Mayne and Rix (1993) 0.86
– 14.13qt0.359 e0−0.473 qt in kPa
– 3.18qt0.549 fs0.025 qt , fs in kPa
Figure 3.5: Comparison of empirical correlations for estimating Vs from CPT data: qt and fs ,
plotted for different values of friction ratio Rf = fs /qt × 100 (correlations are given in
Table C.9 and C.10).

Figure 3.6: Comparison of empirical correlations for estimating Vs in sands from CPT data (corre-
lations are given in Table C.10).
Figure 3.7: Comparison of empirical correlations for estimating Vs in sands from CPT data (corre-
lations are given in Table C.11).

References
References
Addenbrooke, T., Potts, D., and Puzrin, A. (1997). The influence of pre-failure soil stiffness on the numerical
analysis of the tunnel construction. Géotechnique, 47(3):693–712. 6, 155, 158
Al-Khafaji, A. and Andersland, O. (1992). Equations for compression index approximation. J Geotech Eng
ASCE, 118(1):148–153. 187
Alpan, I. (1970). The geotechnical properties of soils. Earth-Science Reviews, 6:5–49. 70, 103
Amar, S., Clarke, B., Gambin, M., and Orr, T., editors (1991). The application of pressuremeter test results
to foundation design in Europe. Part 1: Predrilled pressuremeters/self-boring pressuremeters. Balkema,
Rotterdam. 130
Athanasopoulos, G. A. (1995). Empirical correlations vs-nspt for soils of greece: a comparative study of
reliability. In Cakmak, A., editor, 7th Int Conf on Soil Dynamics and Earthquake Engineering, pages
19–36, Chania, Crete. 193, 195
Atkinson, J. and Sallfors, G. (1991). Experimental determination of soil properties. In Proc. 10th ECSMFE,
volume 3, pages 915–956, Florence, Italy. 4
Azzouz, A., Krizek, R., and Corotis, R. (1976). Regression analysis of soil compressibility. Soils and Founda-
tions, 16(2):19–29. 187, 188
Baldi, G., Bellotti, R., Ghionna, V., Jamiolkowski, M., and Lo Presti, D. (1989). Modulus of sands from
cpts and dmts. In Proc. 12th International Conference on Soil Mechanics and Foundation Engineering,
volume 1, pages 165–170, Rio de Janeiro, Brazil. Balkema, Rotterdam. 197
Baldi, G., Bellotti, R., Ghionna, V., Jamiolkowski, M., Marchetti, S., and Pasqualini, E. (1986). Flat
dilatometer tests in calibration chambers. In In Situ 86, ASCE Spec. Conf. on ”Use of In Situ Tests in
Geotechn Engineering”, volume GSP 6, pages 431–436. ASCE. 73, 99
Begemann, H. (1974). General report for central and western europe. In European Symp. on Penetration
Testing, Stockholm, Sweden. 82
Belotti, R., Ghionna, V., Jamiolkowski, M., Lancellotta, R., and Manfredini, G. (1986). Deformation charac-
teristics of cohesionless soisl in in situ tests. In ASCE, editor, In Situ 86, Geotechnical Special Publications,
pages 47–73, New York. ASCE. 73
Benz, T. (2007). Small-strain stiffness of soils and its numerical consequences. Phd, Universitat Sttutgart.
3, 9, 13, 41, 69, 88, 126, 139, 140, 141, 142, 147
Biarez, J. and Hicher, P. (1994). Elementary mechanics of soil behaviour. Saturated remoulded soils. Balkema,
Rotterdam. 67, 68, 100, 102
Bjerrum, L. (1972). Embankments on soft ground. In Proc. Am.Soc. Civ Engs Specialty Conf. on Perfomence
of Earth and Earth supported Structures., volume 2, pages 1–54, Perdue University. 183
Bolton, M. (1986). The strength and dilatancy of sands. Géotechnique, 36(1):65–78. 94
Bowles, J. (1997). Foundation analysis and design. McGraw-Hill. 9, 82, 89, 91, 127, 187, 189
References
Briaud, J.-L. and Gibbens, R. (1997). Large scale load test and data base of spread footings on sand.
Technical Report Publication No. FHWA RD-97-068, Texas A&M University. 166, 177
Brinch Hansen, J. and Lundgren, H., editors (1958). Geoteknik. Teknisk Forlag, Kopenhagen. 89
Carter, M. and Bentley, S. (1991). Correlations of soil properties. Penetech Press Publishers, London. 93,
110
Casagrande, A. (1936). The determination of the pre-consolidation load and its practical significance. In
Casagrande, A., editor, In Proc. 1st Int Soil Mech and Found Engng Conf, 22-26 June 1936, volume 3,
pages 60–64, Cambridge, Mass. Graduate School of Engineering, Harvard University, Cambridge, Mass.
57
Chandler, R. (1988). The in-situ measurements of the undrained shear strength of clays using the field tests.
Technical report, ASTM. 185
Chaudary, S., Kuwano, J., and Hayano, Y. (2004). Measurement of quasi-elastic stiffness parameters of
dense toyoura sand in hollow cylinder apparatus and triaxial apparatus with bender elements. Geotechnical
Testing Journal, 27(1):1–13. 68, 88
Chen, B. and Mayne, P. (1994). Profiling the OCR of clays by piezocone tests. Technical Report Rep. No.
CEEGEO-94-1, Georgia Institute of Technology. 128
Chen, B. and Mayne, P. (1996). Statistical relationships between piezocone measurements and stress history
of clays. Can Geotech J, 33:488–498. 128
Clough, G., Briauld, J., and Hughes, J. (1990). The developement of prssuremeter testing. In Pressuremeters,
3rd Int Symp on Pressuremeters, pages 25–45, Oxford. Thomas Telford, London. 130
Cudny, M. and Truty, A. (2020). Refinement of the hardening soil model within the small strain range. Acta
Geotechnica, accepted for publication (https://doi.org/10.1007/s11440-020-00945-5). 3, 14, 44, 45
Dalton, J. and Hawkins, P. (1982). Fields of stress - some measurements of the in-situ stress in a meadow
in the Cambridgeshire countryside. Ground Engineering, 15(4):15–23. 130
D’Appolonia, D., D’Appolonia, E., and Brissette, R. (1970). Closure: Settlement of spread footing on sand.
J Soils Mechanics and Foundation Division ASCE, 96(2):754–762. 82
Darendeli, M. (2001). Development of a New Family of Normalized Modulus Reduction and Material Damping
Curves. PhD thesis, University of Texas. 78, 108
Darendeli, M. and Stokoe, K. (2001). Development of a new family of normalized modulus reduction and
material damping curves. Technical report, University of Texas. 76, 77
Das, B. (2008). Advanced Soil Mechanics. Taylor & Francis, London & New York. 96, 97
Dayal, I. (2006). Consolidation analyses of greater cincinnati soils. MSc. Technical report, University of
Cincinnati. 188
DeGroot, D. and Sandven, R. (2004). General report: Laboratory and field comparisons. In de Fonseca, V.
and Mayne, editors, Proc. ISC-2 on Geotechnical and Geophysical Site Characterization, pages 1775–1789,
Porto, Portugal. Millpress, Rotterdam. 183
Demers, D. and Leroueil, S. (2002). Evaluation of preconsolidation pressure and the overconsolidation ratio
from piezocone tests of clay deposits in quebec. Can Geotech J, 39:174–192. 128
Dikmen, U. (2009). Statistical correlations of shear wave velocity and penetration resistance for soils. Journal
of Geophysics and Engineering, 6:61–72. 194, 195
Duncan, J. and Buchignani, A. (1976). An engineering manual for settlement studies. Technical report,
University of California. 119

References
Fahey, M. and Carter, J. (1993). A finite element study of the pressuremeter in sand using a nonlinear elastic
plastic model. Canadian Geotechnical Journal, 30(2):348–362. 70
Finno, R. and Calvello, M. (2005). Supported excavations: the observational method and inverse modeling.
J Geotech Geoenv Engng ASCE, 131(7):826–836. 9
Fioravante, V. (2000). Anisotropy of small-strain stiffness of ticino and kenya sands from seismic wave
propagation measured in traixial testing. Soils and Foundations, 40(4):129–142. 68, 88
Fioravante, V. (2004). General report: Enhanced characterization by combined in situ testing. In de Fonseca,
V. and Mayne, editors, Proc. ISC-2 on Geotechnical and Geophysical Site Characterization, pages 1585–
1596, Porto, Portugal. Millpress, Rotterdam. 183
FOWG (2003). Documentation de base pour la vérification des ouvrages d’accumulation aux séismes. Tech-
nical report, Swiss Federal Office for Water and Geology. 192
Hardin, B. and Black, W. (1969). Closure to vibration modulus of normally consolidated clays. J Soils
Mechanics and Foundation Division ASCE, 95(6):1531–1537. 67, 68, 100, 102
Hardin, B. and Richart Jr, F. (1963). Elastic wave velocities in granular soils. J Soils Mechanics and
Foundation Division ASCE, 89(1):33–65. 68, 88
Hasancebi, N. and Ulusay, R. (2007). Empirical correlations between shear wave velocity and penetration
resistance for ground shaking assessments. Bulletin of Engineering Geology and theEnvironment, 66:203–
213. 193, 194, 195
Hatanaka, M. and Uchida, A. (1996). Empirical correlation between penetration resistance and phi of sandy
soils. Soils and Foundations, 36(4):1–9. 91
Hegazy, Y. and Mayne, P. (1995). Statistical correlations between vs and cone penetration data for different
soil types. In Society, S. G., editor, Cone Penetration Testing (CPT’95), volume 2, pages 173–8. 197, 198
Hicher, P. (1996). Elastic properties of soils. J Geotech Eng ASCE, 122(8):641–648. 125
Hight, D., Böese, R., Butcher, A., Clayton, C., and Smith, P. (1992). Disturbance of the Bothkennar clay
prior to laboratory testing. Géotechnique, 42(2):199–217. 183
Hight, D. and Leroueil, S. (2003). Characterisation of soils for engineering purposes. In Tan, T., Phoon, K.,
Hight, D., and Leroueil, S., editors, Characterisation and engineering properties of natural soils, volume 1,
pages 254–360, Singapore. Swets and Zeitlinger. 127
Holtz, R., Jamiolkowski, M., and Lancellotta, R. (1986). Lessons from oedometer tests on high quality
samples. J Geotech Eng ASCE, 112(8):768–776. 187, 189
Hoque, E. and Tatsuoka, F. (2000). Kinematic elasticity of granula material. In GeoEng 2000, Melbourne.
68, 88
Hoque, E. and Tatsuoka, F. (2004). Effects of stress ratio on small-strain stiffness during triaxial shearing.
Géotechnique, 54(7):429–439. 68, 88
Hough, B. (1969). Basic soil engineering. Ronald Press Company, New York. 135, 187
Houlsby, G. and Hitchman, R. (1988). Calibration chamber tests of a cone penetrometer in sand.
Géotechnique, 38(1):39–44. 128
Hryciw, R. (1990). Small-strain-shear modulus of soil by dilatometer. J Geotech Eng ASCE, 116(11):1700–16.
103, 104
Imai, T. and Tonouchi, K. (1982). Correlation on n-value with s-wave velocity and shear modulus. 57-72. 75
Imai, T. and Yoshimura, Y. (1975). The relation of mechanical properties of soils to p and s-wave velocities
for ground in japan. technical note. Technical report, OYO Corporation. 193

References
Iwasaki, T. and Tatsuoka, F. (1977). Effects of grains size and grading on dynamic shear moduli of sands.
Soils and Foundations, 17(3):19–35. 68
Jamiolkowski, M., Ladd, C., Germaine, J., and Lancellotta, R. (1985). New developments in field and
laboratory testing of soils. In Proc. 11th ICSMFE, volume 1, pages 57–153, San Francisco. Balkema. 112,
130, 131, 183
Jamiolkowski, M., Lo Presti, D., and Pallara, O. (1995). Role of in-situ testing in geotechnical earthquake
engineering. In Proc. 3rd Int Conf on Recent Advances in Geotechnical Earthquake Engineering and Soil
Dynamics, volume 2, pages 1523–1546, Missouri. 101, 102
Janbu, N. (1963). Soil compressibility as detrmined by oedometer and triaxial tests. In Proc. ECSMFE,
volume 1, pages 19–25, Wiesbaden. 17, 87
Jinan, Z. (1987). Correlation between seismic wave velocity and the number of blow of spt and depth.
Selected Papers from the Chinese Journal of Geotechncial Engineering (ASCE), pages 92–100. 193
Kallioglou, T., Papadopoulou, A., and Pitilakis, K. (2003). Shear modulus and damping of natural sands.
In Di Benedetto, H., Doanh, T., Geoffroy, H., and Sauzeat, C., editors, Deformation characteristics of
geomatrials, pages 401–407, Lyon. Balkema. 68, 88
Karlsrud, K. (1999). General aspects of transportation infrastructure. In Barends, e. a., editor, 12th European
Conf on Soil Mech and Geotech Engng, volume 1, pages 17–30, Amsterdam. Balkema. 183
Kempfert, H. (2006). Excavations and foundations in soft soils. Springer, Heidelberg. 9, 17, 29, 58, 115,
122, 123, 125, 187, 189
Kezdi, A. (1974). Handbook of Soil Mechanics. Elsevier, Amsterdam. 79, 117
Kim, T. and Novak, M. (1981). Dynamic properties of some cohesive soils of ontario. Can Geotech J,
18(3):371–389. 102
Kokusho, T., Yoshida, Y., and Yasuyuki, E. (1982). Dynamic properties of soft clay for wide strain range.
Soils and Foundations, 22(4):1–18. 102
Koppula, S. (1981). Stastical estimation of compression index. Geotechnical Testing Journal, 4(2):68–73.
188
Koppula, S. (1986). Discussion: Consolidation parameters derived from index tests. Géotechnique, 36(2):291–
292. 187
Kramer, S. (1996). Geotechnical Earthquake Engineering. Civil Engineering and Engineering Mechanics.
Prentice Hall, New Jersey. 75, 101
Kulhawy, F. and Mayne, P. (1990). Manual on estimating soil properties for foundation design. Technical
report, Electric Power Research Institute. 98, 99, 123, 124, 133, 187, 189
Kuwano, R. and Jardine, R. (2002). On the application of cross-anisotropic elasticity to granular materials
at very small strains. Géotechnique, 52(10):727–749. 68, 88
Lacasse, S., Jamiolkowski, M., Lancellotta, R., and Lunne, T. (1981). In situ characteristics of two Norwegian
clays. In Proc. 10th Int Conf Soil Mech and Found Engng, volume 2, Stockholm. 184
Lacasse, S. and Lunne, T. (1988). Calibration of dilatometer correlations. In Penetration Testing 88, ISOPT-
1, volume 1, pages 539–548, Orlando, Florida. 129, 134
Larsson, R. and Mulabdić, M. (1991). Piezocone tests in clay. Technical Report Report No.42, Swedish
Geotechnical Institute. 127, 128
Lavergne, M. (1986). Méthodes sismiques. Technip, Paris. 191
Lee, S. H. (1990). Regression models of shear wave velocities. J Chin Inst Eng, 13:519–32. 194, 195

References
Lee, S.-H. (1992). Analysis of the multicollinearity of regression equations of shear wave velocities. Soils and
Foundations, 32(1):205–214. 194, 195
Leroueil, S. (1988). Tenth Canadian Geotechnical Colloquium: Recent developments in consolidation of

natural clays. Can Geotech J, 25:85–107. 183
Leroueil, S., Tavenas, F., and Le Bihan, J.-P. (1983a). Propriétés caractéristiques des argiles de l’est du
Canada. Can Geotech J, 20(4):681–705. 57
Leroueil, S., Tavenas, F., Samson, L., and Morin, P. (1983b). Preconsolidation pressure of champlain clays.
part II. laboratory determination. Can Geotech J, 20(4):803–816. 183
Li, X. and Dafalias, Y. (2000). Dilatancyfor cohesionless soils. Géotechnique, 50(4):449–460. 46, 47
Liao, S. and Whitman, R. (1986). Overburden correction factors for. spt in sand. J Geotech Eng ASCE,
112(3):373–377. 91
Lindeburg, M. (2001). Civil Engineering Reference Manual for the PE Exam. Professional Publications,
Belmont, California. 192
Lo Presti, D. and Jamiolkowski, M. (1998). Discussion: Estimate of elastic shear modulus in holocene soil
deposits. Soils and Foundations, 38:263–265. 88, 101, 126
Lo Presti, D., Pallara, O., Fioravante, V., and Jamiolkowski, M. (1998). Assessement of quasi-linear models
for sands. In Jardine, R., Davies, M., Hight, D., Smith, A., and Stallebras, S., editors, Pre-failure
Deformation of Geomaterials, pages 363–372, London. Thomas Telford. 70, 72
Lo Presti, D., Pallara, O., Lancellotta, R., Armandi, M., and Maniscalco, R. (1993). Monotonic and cyclic
loadin behaviour of two sands at small strains. Geotechnical Testing Journal, 16(4):409–424. 68, 88
Long, M. and Shane, D. (2010). Characterization of norwegian marine clays with combined shear wave
velocity and piezocone cone penetration test (cptu) data. Can Geotech J, 47(7):709–718. 103
Lunne, T., Robertson, P., and Powell, J. (1997). Cone penetration testing in geotechnical practice. Blackie
Academic & Professional, London. 73, 98, 99, 123, 124, 128, 133
Mair, R. and Wood, D. (1987). Pressuremeter testing. Methods and interpretation. Butterworths, London.
130
Marchetti, S. (1980). In situ tests by flat dilatometer. J Geotech Eng ASCE, 106(3):299–321. 133, 185
Marchetti, S. (1985). On the field determination of k0 in sand. discussion session no. 2a. In XI ICSMFE,
volume 5, pages 2667–2673, San Francisco. 99
Marchetti, S., Monaco, P., Totani, G., and Calabrese, M. (2001). The flat dilatometer test (dmt) in soil
investigations. In TC16, I. C., editor, IN SITU 2001, Int Conf On In situ Measurement of Soil Properties,
Bali. ISSMGE. 99
Marcuson, W. and Wahls, H. (1972). Time effects on dynamic shear modulus of clays. J Soils Mechanics
and Foundation Division ASCE, 98(12):1359–1373. 101, 126
Masood, T. and Mitchell, J. (1993). Estimation of in situ lateral stresses in soils by cone-penetration test. J
Geotech Eng ASCE, 119(10):1624–1639. 132, 133
Mayne, P. (1980). Cam-clay predictions of undrained strength. J Geotech Engng ASCE, 106(GT11):1219–
1242. 187, 189
Mayne, P. (1988). Determining OCR in clays from laboratory strength. J Geotech Engng ASCE, 114(1):76–
92. 127
Mayne, P. (2005). Versatile site characterization by seismic piezocone. In Proc. 16th ICSMGE, volume 2,
Osaka. Millpress. 111

References
Mayne, P. (2006a). The 2nd james k. mitchell lecture: Undisturbed sand strength from seismic cone tests.
Geomechanics and Geoengineering, 1(4):239–247. 197
Mayne, P. (2006b). In-situ test calibrations for evaluating soil parameters. In Tan, Phoon, Hight, and
Leroueil, editors, Characterisation and Engineering Properties of Natural Soils, volume 3, pages 1601–
1652, Singapore. Taylor & Francis Group, London. 127, 128
Mayne, P. (2007). Cone penetration testing. a synthesis of highway practice. Technical report, Transportation
Research Board. 70, 71, 98, 99, 103, 111, 128, 189
Mayne, P., Barry, R. C., and DeJong, J. (2001). Manual on subsurface investigations. Technical Report
Publication No. FHWA NHI-01-031, Federal Highway Administration. 98, 99
Mayne, P., Coop, M., Springman, S., Huang, A.-B., and Zornberg, J. (2009). State-of-the-art paper (soa-1):
Geomaterial behavior and testing. In ICSMGE, editor, 17th Int Conf Soil Mech And Geotech Engng,
volume 4, pages 2777–2872, Alexandria, Egypt. Millpress/IOS Press Rotterdam. 86
Mayne, P. and Kulhawy, F. (1982). Ko −OCR relationship in soils. J Geotech Engng ASCE, 108(6):851–872.
37, 131
Mayne, P. and Rix, G. (1993). Gmax-qc relationships for clays. Geotechnical Testing Journal, 16(1):54–60.
103, 198
Meyerhof, G. (1956). Penetration tests and bearing capacity of cohesionless soils. J Soils Mechanics and
Foundation Division ASCE, 82(SM1). 90
Modoni, G., Flora, A., Anh Dan, L., Macuso, C., Koseki, J., Balakrishnaier, K., and Tatsuoka, F. (1999).
A simple experimental procedure for the complete characterization of small strain stiffness of gravels. In
Jamiolkowski, M., Lancellotta, R., and Lo Presti, D., editors, Pre-failure Deformation Characteristics in
Geomaterials, pages 123–130, Torino. Balkema. 68, 88
Moh, Z., Chin, C., Lin, C., and Woo, S. (1989). Engineering correlations for soil deposits in taipei. J Chinese
Institute of Engineers, 12(3):273–283. 188
Mróz, Z., Norris, V. A., and Zienkiewicz, O. C. (1981). An anisotropic, critical state model for soils subject
to cyclic loading. Géotechnique, 31(4):451–469. 44
Nacci, V. A., Wang, A. M., and Demars, K. R. (1975). Engineering behavior of calcareous soils. In In Civil
Engineering in the Oceans III, volume 1, pages 380–400. Am Soc Civ Eng. 188
Nagaraj, T. and Murthy, B. (1985). Prediction of the preconsolidation pressure and recompression index of
soils. Geotechnical Testing Journal, 8(4):199–202. 187, 189
Nakase, A. (1988). Constitutive parameters estimated by plasticity index. J Geotech Engng ASCE,
114(GT7):844–858. 187, 189
Nakase, A. and Kamei, T. (1986). Influence of strain rate on undrained shear characteristics of ko -consolidated
cohesive soils. Soils and Foundations, 26(1):85–95. 183
NAVFAC (1986). Foundations and earth structures: Design Manual DM-7.02. Technical report, Naval
Facilities Engineering Command Publications Transmittal. 192
Niemunis, A. and Cudny, M. (2018). Discussion on ”Dynamic soil-structure interaction: A three-dimensional

numerical approach and its application to the Lotung case study”. Poor performance of the HSS model.
Computers and Geotechnics, 98:243–245. 3, 14, 42
Nishida, Y. (1956). A brief note on compression index of soils. J Soils Mechanics and Foundation Division
ASCE, 82(SM3):1027. 187
Ohta, H. and Goto, N. (1976). Estimation of s-wave velocity in terms of characteristic indices of soil (in
Japanese). Buturi-tansa, 29:251–261. 75

References
Ohta, T., Hara, A., Niwa, M., and Sakano, T. (1972). Elastic shear moduli as estimated from n-value. In
7th Ann. Convention of Japan Society of Soil Mechanics and Foundation Engineering, pages 265–8. 194
Ohta, Y. and Goto, N. (1978). Empirical shear wave velocity equations in terms of characteristic soil indexes.
Earthquake Engineering and Structural Dynamics, 6:167–187. 193
Okamoto, T., Kokusho, T., Yoshida, Y., and Kusuonoki, K. (1989). Comparison of surface versus subsurface
wave source for p-s logging in sand layer. In Engineers, J. S. C., editor, Proc. 44th Ann. Conf., volume 3,
pages 996–7 (in Japaneese). 194
Pacheco-Silva, F. (1970). A new graphical construction for determination of the pre-consolidation stress of
a soil sample. In 4th Brazilian Conf on Soil Mech and Found Engng, volume 1, pages 225–232, Rio de
Janeiro. 57
Park, J. and Koumoto, T. (2004). New compression index equation. J Geotech Geoenv Eng ASCE, 130:223.
188
Peck, R., Hanson, W., and Thornburn, T. (1974). Foundation Engineering Handbook. Wiley, London. 89,
90
Penumadu, D., Skandarajah, A., and Chameau, J. (1998). Strain-rate effects in pressuremeter testing using
a cuboidal shear device: experiments and modeling. Can Geotech J, 35:27–42. 183
Pitilakis, K., Raptakis, D., Lontzetidis, K., Tika-Vassilikou, T., and Jongmans, D. (1999). Geotechnical and
geophysical description of euro-seistests, using field, and laboratory tests and moderate strong ground
motions. Journal of Earthquake Engineering, 3(3):381–409. 194, 195
Prapaharan, S., Chameau, J., and Holtz, R. (1989). Effect of strain rate on undrained strength derived from
pressuremeter tests. Géotechnique, 39(4):615–624. 183
Prat, M., Bisch, E., Millard, A., Mestat, P., and Cabot, G. (1995). La modelisation des ouvrages. Hermes,
Paris. 79, 117
Rampello, S., Viggiani, G., and Amorosi, A. (1997). Small-strain stiffness of reconstituted clay compresses
along constant traiaxial effective stress ratio paths. Géotechnique, 47(3):475–489. 101, 126
Rendon-Herrero, O. (1983). Closure: Universal compression index equation. J Geotech Engng ASCE,
109(GT5):755–761. 187
Rix, G. and Stokoe, K. (1992). Correlation of initial tangent modulus and cone resistance. In Int Symp. on
Calibration Chamber Testing, pages 351–62, Postdam. Elsevier, New York. 73
Robertson, P. and Campanella, R. (1983). Interpretation of cone penetration tests. part I: Sand. Can Geotech
J, 20(4):718–33. 73, 81, 92
Sandven, R., Senneset, K., and Janbu, N. (1988). Interpretation of piezocone tests in cohesive soils. In
De Ruiter, J., editor, ISOPT-1 Penetration Testing, pages 939–953. Balkema, Rotterdam. 110, 123
Sanglerat, G. (1972). The penetrometer and soil exploration. Elsevier, Amsterdam. 123
Schanz, T. (1998). Zur modellierung des mechanischen verhaltens von reibungsmaterialien. Mitt. Inst. fÃ¼r
Geotechnik, Universitat Stuttgart, 1998, 45. 13
Schanz, T. and Vermeer, P. (1998). On the stiffness of sands. In Jardine, R., Davies, M., Hight, D., Smith,
A., and Stallebras, S., editors, Pre-failure deformation behaviour of geomaterials, pages 383–387, London.
Thomas Telford. 80, 84, 87
Schanz, T., Vermeer, P., and Bonier, P., editors (1999). Formulation and verification of the Hardening Soil
model. Beyond 2000 in Computational Geotechnics. Balkema, Rotterdam. 13
Schmertmann, J. (1975). Measurement of insitu shear strength. In In-situ Measurement of Soil Properties,
volume 2. ASCE. 91

References
Schmertmann, J. (1978). Guidlines for cone penetration test, performance and design. Technical report, US
Federal Highway Admininstration. 93
Schultze, E. and Menzenbach, E. (1961). Standard penetration test and compressibility of soils. In 5th
ICSMFE, volume 1, pages 527–532, Pari. 82
Schweiger, H. (2002). Benchmarking in geotechnics. part I: Results of benchmarking. part II reference

solution and parametric study. Technical report, Institute for Slil Mechnaics and Foundation Engineering,
Graz University of Technology. 147
Seed, H., Idriss, I., and Arango, I. (1983). Evauluation of liquefaction potential using field performance data.
Journal Geotechnical Engineering, 109(3):458–482. 193
Senneset, K., Sandven, R., and Janbu, N. (1989). The evaluation of soil paramters from piezocone tests.
Transportation Research Record, 1235:24–37. 110, 123
Sheahan, T., Ladd, C., and Germaine, J. (1996). Rate-dependent undrained shear behavior of saturated clay.
J Geotech Eng ASCE, 122(2):99–108. 183
Shibata, T. (1970). Analysis of liquefaction of saturated sand during cyclic loading. Disaster Prevention Res.
Inst. Bull., 13:567–70. 194
Shioi, Y. and Fukui, J. (1982). Application of n-value to design of foundations in japan. In 2nd ESOPT,
volume 1, pages 159–164. 91
Simonini, P. and Cola, S. (2000). Use of piezocone to predict maximum stiffness of venetian soils. J Geotech
Geoenv Engng ASCE, 126(4):378–382. 103
Simpson, B. (1992). Retaining structures: displacement and design, 32nd Rankine Lecture. Géotechnique,
42(4):541–576. 44
Simpson, B., O’Riordan, N. J., and Croft, D. D. (1979). A computer model for the analysis of ground
movements in London clay. Géotechnique, 29(2):149–175. 44
Skempton, A. (1944). Notes on the compressibility of clays. Quarterly Journal of Geotechnical Society,
London., pages 119–135. 187
Stokoe, K., Derendeli, M., Gilbertl, R., Menq, F., and Choi, W. (2004). Development of a new family of
normalized modulus reduction and material damping curves. In Int Workshop on Uncertainties in Nonlinear
Soil Properties and their Impact on Modeling Dynamic Soil Response, UC Berkley. 108
Sully, J. and Campanella, R. (1991). Effect of lateral stress on CPT penetration pore pressures. J Geotech
Eng ASCE, 117(7):1082–1088. 132
Sykora, D. (1987). Examination of existing shear wave velocity and shear modulus correlations in soils.
miscellaneous paper gl-87-22. Technical report, Department of the Army, Waterways Experiment Station,
Corps of Engineers. 193, 194, 195
Tanaka, H. and Tanaka, M. (1999). Kay factors governing sample quality. In Int Symp on Coastal Engng in
Practice, pages 57–82, Yokohama. 183
Teixeira, A. (1996). Projeto e execuadao de fundacoes. In Seminario de Engenharia de Fundacoes Especiais

e Geotecnica III, volume 1, pages 33–50, Sao Paolo. 91
Terzaghi, K. and Peck, R. (1967). Soil mechanics in engineering practice. Wiley and Sons, New York. 187
Terzaghi, K., Peck, R., and Mesri, G. (1996). Soil Mechanics in Engineering Practice. Wiley, New York. 135
Totani, G., Marchetti, S., Monaco, P., and Calabrese, M. (1999). Impiego della prova dilatometrica (dmt)
nella progettazione geotecnica. In XX National Geotechnical Congress, volume 1, pages 301–308, Parma.
92

Modifications
Totani, G., Marchetti, S., Monaco, P., and Calabrese, M. (2001). Use of the flat dilatometer test (dmt) in
geotechnical design. In Rahardjo, P. and Lunne, T., editors, IN SITU 2001 Int Conf In Situ Measurement
of Soil Properties and Case Histories, Bali, Indonesia. 129
Townsend, F. C., Anderson, J. B., Horta, E., and Sandoval, J. (2003). Development of deep foundations
test site. Technical report, University of Central Florida. 91
Tromienkov, Y. (1974). Penetration testing in western europe. In Europ. Symp. on Penetration Testing,
Stockholm, Sweden. 82
Truty, A. (2008). Hardening soil model with samll strain stiffness. Technical report, Zace Services Ltd. 3, 7,
9, 33
Truty, A. and Obrzud, R. (2015). Improved formulation of the hardening soil model in the contex of modeling
the undrained behavior of cohesive soils. Studia Geotechnica et Mechanica, 37(2):61–68. 66, 120
UFC (2004). Unified facilities criteria: Soils and geology procedures for foundation design of buildings and
other structures (except hydraulic structures). technical manual TM 5-818-1/AFM 88-3. Technical report,
Joint Departments of the Army and Air Force USA. 97
Vaid, Y. and Campanella, R. (1977). Time-dependent behavior of undisturbed clay. J Geotech Eng ASCE,
103(7):693–709. 183
Vardanega, P. J. and Bolton, M. (2011). Practical methods to estimate the non-linear shear stiffness of fine
grained soils. In Int Symp on Deformation Characteristics of Geomaterials, pages 372–379, Seoul, Korea.
105, 108
Viggiani, G. and Atkinson, J. (1995). Stiffness of fine grained soil at very small strain. Géotechnique,
45(2):249–265. 101, 125, 126, 158
von Soos, P. (1991). Grundbataschenbuch. Part 4. Ernst & Sohn, Berlin. 88, 126
Vucetic, M. and Dobry, R. (1991). Effect of soil plasticity on cyclic response. J Geotech Eng ASCE,
117(1):89–107. 105, 106, 108
Webb, D. (1969). Settlement of structures on deep alluvial sandy sediments in duran, south africa. In
Engineers, I. o. C., editor, Conf. on In Situ Behaviour of Soil and Rock, pages 181–188, London. 82
Wichtmann, T. and Triantafyllidis, T. (2004). Influence of a cyclic and dynamic loading history on dynamic
properties of dry sand, part i: cyclic and dynamic torsional prestraining. Soil Dynamics and Earthquake
Eng., 24(2):127–147. 68, 76, 77, 88
Wroth, C. (1984). The interpretation of in situ tests. Géotechnique, 34(4):449–489. 112, 183, 184
Wroth, C. and Wood, B. (1978). The correlation of index properties with some basic engineering properties
of soils. Canadian Geotechnical Journal, 15(2):137–145. 187
Yoon, G., Kim, B., and Jeon, S. (2004). Empirical correlations of compression index for marine clay from
regression analysis. Can Geotech J, 41:1213–1221. 188
ZACE (2010). Zsoil pc2010, user manual, soil, rock and structural mechanics in dry or partially saturated
media. Technical report, ZACE Services Ltd, Software Engineering. 8

Rep HS Model

Uploaded by

Copyright:

Available Formats

Rep HS Model

Uploaded by

Document Information

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

Rep HS Model

Uploaded by

Copyright:

Available Formats

THE HARDENING SOIL MODEL -

Zace Services Ltd, Software engineering

2 Short introduction to the Hardening Soil models 13

5 Case studies 147

A Determination of undrained shear strength 183

B Estimation of compression index 187

C Estimation of shear wave velocity 191

Stress and Strain Notation Cc slope of the normal compression

Sign convention: Throughout this report, the

2. Which formulation to describe the stress dependent stiffness should be chosen?

4. How to set model parameters for an ”undrained” simulation?

5. How to troubleshoot convergence problem at the initial state analysis?

8. What are typical parameter ranges?

10. What should be specified in σ ref

11. What should be specified in σ ref cell?

13. When should I activate the small strain extension?

14. How to identify model parameters?

15. What is the suggested parameter identification sequence?

16. How to use the Virtual Lab v2020 ?

1.1 Why do we need small strain extension in the HS model?

4 Z Soil.PC 100701 report (revised 2.01.2020)

(a) Shear characteristics q − ε1

(b) Normalized secant shear stiffness characteristics Gs /Go − ε1

Z Soil.PC 100701 report (revised 2.01.2020) 5

6 Z Soil.PC 100701 report (revised 2.01.2020)

Z Soil.PC 100701 report (revised 2.01.2020) 7

1.2 Application fields of constitutive models

The choice of a constitutive model depends on many factors but, in general, it is

8 Z Soil.PC 100701 report (revised 2.01.2020)

All type of soils

Z Soil.PC 100701 report (revised 2.01.2020) 9

10 Z Soil.PC 100701 report (revised 2.01.2020)

Model Hardening Soil Cap Mohr-Coulomb

Z Soil.PC 100701 report (revised 2.01.2020)

12 Z Soil.PC 100701 report (revised 2.01.2020)

Short introduction to the Hardening

• soil stress history, i.e. accounting for preconsolidation effects;

• hysteretic, nonlinear elastic stress-strain relationship which is applicable in the range of

14 Z Soil.PC 100701 report (revised 2.01.2020)

2.1 HS-standard model

2.1.1 Shear mechanism

2.1.1.1 Shear yield mechanism

Z Soil.PC 100701 report (revised 2.01.2020) 15

16 Z Soil.PC 100701 report (revised 2.01.2020)

2.1.2 Stress dependent stiffness

Z Soil.PC 100701 report (revised 2.01.2020) 17

An example of stress dependency is graphically presented in Figure 2.4.

18 Z Soil.PC 100701 report (revised 2.01.2020)

Z Soil.PC 100701 report (revised 2.01.2020) 19

2.1.2.1 Stress dependent stiffness based on the mean effective stress

Table 2.1: Comparison of different types of stiffness stress dependency.

Minor effective stress σ 03 Mean effective stress p0

20 Z Soil.PC 100701 report (revised 2.01.2020)

2.1.2.2 Parameter migration between minor and mean stress for-

• the stiffness exponents m and mp are different if c > 0

Table 2.2 shows in detail equivalency of parameters in terms of compatibility of material

Z Soil.PC 100701 report (revised 2.01.2020) 21

Rf = Rfσ3 Rf = Rfσ3 Rf < Rfσ3

22 Z Soil.PC 100701 report (revised 2.01.2020)

Regression analysis based on discrete stress points

string of b-th brick is slack