You are on page 1of 12

Numerical Modeling of Cyclic Triaxial Experiments for

Granular Soil
Zhangbo Nie1; Shichun Chi2; and Song Gong3
Downloaded from ascelibrary.org by Indian Inst Of Technology Library, Kharagpur on 04/05/17. Copyright ASCE. For personal use only; all rights reserved.

Abstract: In this study, cyclic triaxial tests under isotropic and anisotropic stress consolidation conditions were performed on Nuozhadu granu-
lar soil, and a constitutive model based on the Bouc–Wen model and endochronic theory was applied to investigate the dynamic stress–strain rela-
tionship and pore-water pressure of granular soils under cyclic triaxial stress states. In the isotropic stress consolidation condition, dynamic stress
degradation occurred as a result of increased pore-water pressure and the reduction of effective stress. This paper presents the general validity of
the model and its ability to describe several features of granular soil, including soil cyclic mobility, plastic strain accumulation, and stiffness and
strength changes resulting from cyclic loading. Then, the equation of the model and key model parameters related to soil behavior, such as hyster-
esis and contractive and dilative behaviors, are further discussed. The proposed model was used to simulate the dynamic stress–strain relationship,
effective stress path, and increased pore-water pressure under undrained conditions. The numerical results were found to have good agreement
with the experimental curves. DOI: 10.1061/(ASCE)GM.1943-5622.0000832. © 2016 American Society of Civil Engineers.
Author keywords: Constitutive law; Cyclic mobility; Plastic strain accumulation; Bouc–Wen model; Dynamic triaxial test.

Introduction initial anisotropic sustained deviatoric stress and cyclic deviatoric


stress ratio. Chiaro et al. (2012) focused on the effects of static shear
Soils are often subjected to transient and cyclic loading such as that stress on the large deformation behavior of loose saturated sand dur-
induced by earthquakes or road traffic (Youd and Idriss 2001; ing undrained cyclic loading. Liu and Xu (2013) investigated a se-
Yasuda et al. 2012). The response of granular soil subjected to ries of strain-controlled cyclic triaxial tests with different initial
cyclic loading is governed by the strong nonlinearity of stress–strain conditions and found that liquefaction occurs not only in loose satu-
behavior. Particularly, under undrained conditions and large cyclic rated soils but also in dense soils or unsaturated soils. Kong et al.
loadings, the pore-water pressure may increase, reducing the effec- (2016) and Liu et al. (2016) performed cyclic triaxial tests on
tive stresses and, consequently, reducing the shearing resistance of Zipingpu gravel and investigated the stress–dilatancy relationship
saturated granular soil. To understand more deeply the response of of gravelly soils in triaxial stress states. El Shamy et al. (2013)
soil, it is necessary to account for its changing properties in the investigated the impact of shear-induced anisotropy for the evalua-
course of cyclic deformation and its inelastic behavior that results in tion of thermal conductivity of granular soils subjected to external
progressive densification or associated pore pressure changes. loading conditions. Manahiloh et al. (2016) researched the principle
Many experiments have investigated the dynamic characteriza- of effective stress for unsaturated granular soils with a combination
tion of granular soil. Mohamad and Dobry (1986) discussed the of microstructural analysis and image-processing techniques. Sze
influence of stress reversals during cyclic triaxial tests and proposed and Yang (2014) studied the impact of specimen preparation on the
an interpretation for the liquefaction resistance of anisotropically cyclic loading behavior of saturated sand.
consolidated sand in cyclic triaxial tests. Hyde et al. (2006) devel- Significant progress has been made in the last decade in the sim-
oped an empirical model relating the number of cycles to failure ulation of the fundamental relationships of stress, strain, and pore-
under conditions of both liquefaction and cyclic mobility to the water pressure of granular soils under cyclic loading conditions.
Most of constitutive models are based on yield and bounding surfa-
1
ces or try to approach the problem with a critical-state framework
Ph.D. Candidate, State Key Laboratory of Coastal and Offshore [e.g., Papadimitriou et al. (2001); Papadimitriou and Bouckovalas
Engineering, Dalian Univ. of Technology, Dalian, Liaoning 116024, China;
(2002); Zhang and Wang (2012); Li and Ming (2000); Dafalias and
Professor, Institute of Earthquake Engineering, Faculty of Infrastructure
Engineering, Dalian Univ. of Technology, Dalian, Liaoning 116024, China.
Manzari (2004)]. Others have been developed on the basis of gener-
E-mail: niezhangbo_0809@163.com alized plasticity theories [e.g., Pastor et al. (1990); Liu et al.
2
Professor, State Key Laboratory of Coastal and Offshore Engineering, (2014)]. Camusso and Barla (2009) presented a numerical model
Dalian Univ. of Technology, Dalian, Liaoning 116024, China; Professor, for the loose and cemented components of the subsoil using particle
Institute of Earthquake Engineering, Faculty of Infrastructure Engineering, methods. Bao and Sture (2011) presented a kinematic-cyclic consti-
Dalian Univ. of Technology, Dalian, Liaoning 116024, China (correspond- tutive model based on fuzzy-set concepts and incremental plasticity
ing author). E-mail: schchi@dlut.edu.cn theory for modeling the cyclic mobility of saturated granular soil.
3
Former Graduate Student, China Gezhouba Group International Compared with other plasticity theories, endochronic theory does
Engineering Co., LTD, Tower F, Ocean International Center, 208 Ciyunsi not include the conception of a yield surface or the specification of
Beili, Chaoyang District, Beijing 100025, China. E-mail: gongsong211@
loading or unloading rules and takes the effects of strain history into
hotmail.com
Note. This manuscript was submitted on April 11, 2016; approved on account by using intrinsic time. López-Querol and Blázquez (2006,
September 13, 2016; published online on November 18, 2016. Discussion 2007), Blázquez and López-Querol (2006), and López-Querol et al.
period open until April 18, 2017; separate discussions must be submitted (2008) developed a liquefaction constitutive model based on endo-
for individual papers. This paper is part of the International Journal of chronic theory applied to the densification of sandy soil. Bouc–Wen
Geomechanics, © ASCE, ISSN 1532-3641. hysteresis models were used in the seismic analysis of soil by

© ASCE 04016147-1 Int. J. Geomech.

Int. J. Geomech., -1--1


Gerolymos and Gazetas (2005), Gerolymos et al. (2007), and Cylindrical specimens with a diameter of 61.8 mm and a height
Drosos et al. (2012) because they are capable of reproducing the of 120 mm were prepared in the cyclic triaxial apparatus. The speci-
complex nonlinear characteristics of cyclic behavior, including men was mounted on the base pedestal of the triaxial cell with its
cyclic mobility and liquefaction. Erlicher and Bursi (2008) provided top cap connected to a fixed point. Each specimen was brought to
a thermodynamic analysis of Bouc–Wen models with both strength initial saturation through CO2 flushing and then deaired water flush-
and stiffness degradation based on the relationship between the ing, which was then followed by backpressure saturation with an
flow rules of Bouc–Wen models and endochronic plasticity theory. effective pressure of 20 kPa. The quality of saturation was checked
This paper presents a theoretical framework for predicting the to guarantee B > 0.95 (B = parameter of Skempton). Then, the
dynamic characteristics of granular soil, verified by experiments on specimens were consolidated to the initial confining stress p0. After
the cyclic behavior of granular soil. Interpretation and comparison that, dynamic loads with various cyclic stress ratio (CSR = s d/2p0)
Downloaded from ascelibrary.org by Indian Inst Of Technology Library, Kharagpur on 04/05/17. Copyright ASCE. For personal use only; all rights reserved.

focusing on the cyclic deformation pattern, pore-water pressure values were applied to the specimens in undrained conditions,
buildup, and cyclic stress path shown in the cyclic triaxial experi- where s d is the dynamic axial stress.
mental results are provided. Then, a constitutive law, based on the
Bouc–Wen dynamic shear stress–strain relationship and endo- Analysis of Test Results
chronic densification model, that can be used to predict different
The behavior of saturated granular soil is illustrated in Figs. 2 and 3,
dynamic failure mechanisms for granular soil is presented. The con-
which show different stress–strain relationships related to the initial
tractive and dilative behavior of the soil were analyzed separately.
consolidation ratio. Fig. 2 shows results for the specimen with initial
Comparisons between tests results and numerical predictions of the
consolidation ratio Kc = 1.0, relative density Dr = 0.8, confining
model, both for cyclic mobility and plastic strain accumulation, are
stress p0 = 200 kPa, cyclic stress ratio CSR = 0.35 subject to sym-
shown for different samples with the same initial relative densities
metric cyclic loading. As shown in Fig. 2(a), in the case of reversal
and different initial effective consolidation stresses.
loading, it is apparent that hysteresis loops occurred in the course of
the load cycle, with strength and stiffness degradation and accumu-
Cyclic Triaxial Test Results and Analysis lated strain amplitude developing rapidly at the low load level. As
cyclic loading progressed, there was a progressive increase in
excess pore-water pressure Du and corresponding reduction in
Sample Materials and Testing Procedure mean effective stress p, as shown in Figs. 2(b and c). With each sub-
The cyclic triaxial device used in this study was designed and manu- sequent cycle of applied deviatoric load q, the effective stress path
factured by GDS Corporation (League City, TX). The device was curve crossed the phase transition line (PTL), which means the sam-
able to vary the deviatoric stress and confining pressure simultane- ple alternated between being incrementally dilative (p increasing)
ously within a range of frequencies between 0.01 and 10 Hz. In this and incrementally contractive (p decreasing) in its response. The
apparatus, the vertical stress was applied by a servo loading system, dilatancy-induced volume increase led to an immediate reduction in
and the confining pressure and back pressure were applied through pore-water pressure and associated increase in effective confine-
a standard GDS 200-cc/3-Mpa digital pressure controller. ment. The axial strains remained relatively small until the excess
The granular soil used in this research for the preparation of pore pressure ratio ru approached 100%; after that, strain amplitude
remolded samples was obtained from the Nuozhadu rockfill dam grew with each additional cycle of loading. As ru approached
project in China, which was set between the rockfill and core wall 100%, the peak pore pressure dropped and returned to peak immedi-
of the dam to avoid large uneven displacements. The dry density of ately as a result of the rise of structure pore pressures, which
the sample was r d = 1.80 g/cm3, the relative density was Dr = 0.8. occurred because the inner structure changed dramatically. The
because of the limitation of the triaxial apparatus, the prototype ma- excess pore pressure ratio increased to a maximum value of 100%,
terial was scaled down; the particle size distribution curves are which corresponds to the sample temporarily having zero effective
shown in Fig. 1. The coefficient of the particle size distribution was stress (i.e., p = 0 and q = 0) and a double-amplitude axial strain of
Cu = 6.38, and the curvature coefficient was Cc = 1.05. 3% in a triaxial test. This state of ru = 100% has been referred to as
initial liquefaction.
Fig. 3 shows the response of a specimen at an effective confining
100 stress of p0 = 200 kPa and consolidation ratio of Kc = 1.5 subjected
to a cyclic stress ratio of CSR = 0.85. As cyclic loading progressed,
there was a progressive increase in excess pore-water pressure and a
Percent Finer by Weight (%)

80
corresponding reduction in mean effective stress, as in the isotropi-
cally consolidated sample. However, the excess pore pressure ratio
60 ru could not increases to a maximum value of 100% because of the
existence of initial deviatoric stress. The existence of initial static
stress restricts the development of pore pressure. The resulting
40 stress–strain loops shifted to the right, exhibiting the cyclic ratchet-
ing shown in Fig. 3(a), which is an example of plastic strain
accumulation.
20
Prototype
Test Description of the Theoretical Model
0 −1 0 1 2
10 10 10 10 Granular soil tends to densification during cyclic loading. The pore-
Paricle Diameter (mm)
water pressure increases and the effective stress reduces because
Fig. 1. Particle size distribution for Nuozhadu granular soil the saturated sand remains totally undrained or does not drain fast
enough. In isotropic consolidation, cyclic mobility behavior occurs

© ASCE 04016147-2 Int. J. Geomech.

Int. J. Geomech., -1--1


150 1
Kc=1.0
p0=200kPa
100 CSR=0.35 0.8

Dynamic Stress σd (kPa)


50
0.6

ru
0
0.4
Downloaded from ascelibrary.org by Indian Inst Of Technology Library, Kharagpur on 04/05/17. Copyright ASCE. For personal use only; all rights reserved.

−50

0.2
−100

−150 0
−3 −2 −1 0 1 2 0 10 20 30 40
Axial Strain ε1 (%) Number of Cycles
(a) (b)

200
Failure Line
150
PTL
100

50
q (kPa)

−50

−100

−150
0 50 100 150 200 250
p (kPa)
(c)

Fig. 2. Undrained cyclic triaxial test for granular soil (Kc = 1.0, Dr = 0.8, p0 = 200 kPa, CSR = 0.35): (a) stress–strain hysteresis loop; (b) excess pore-
water pressure; (c) effective stress path

600 1
Kc=1.5
p =200kPa
500 0
CSR=0.85
0.8
Deviator Stress q (kPa)

400

300 0.6
ru

200

100 0.4

0
0.2
−100

−200 0
0 2 4 6 8 0 10 20 30 40
Axial Strain ε1 (%) Number of Cycles
(a) (b)

Fig. 3. Undrained cyclic dynamic triaxial test for granular soil (Kc = 1.5; Dr = 0.8; p0 = 200 kPa; CSR = 0.85): (a) stress–strain hysteresis loop;
(b) excess pore-water pressure

© ASCE 04016147-3 Int. J. Geomech.

Int. J. Geomech., -1--1


until liquefaction takes place. On the other hand, plastic strain accu- From Eqs. (1) and (2), the parameter z can be calculated
mulation and the development of large deformation are observed as through a straightforward step-by-step numerical integration,
the soil withstands the superimposition of initial static and cyclic which is readily implemented within the framework of codes
shear stress. In this section, modeling for these phenomena and the available in MATLAB.
prediction of the rise in the pore-water pressure is introduced. Taking into account the elastoplastic decomposition of the shear
strain increment, it can be written as
Shear Stress–Strain Constitutive Relationship dg ¼ dg e þ dg p (6)

The Bouc–Wen model is a versatile action–reaction relationship pro- where the elastic strain increment d g e and plastic strain increment
d g p are computed as
Downloaded from ascelibrary.org by Indian Inst Of Technology Library, Kharagpur on 04/05/17. Copyright ASCE. For personal use only; all rights reserved.

posed by Bouc (1967) and modified by Baber and Wen (1981) that is
capable of reproducing a variety of stress–strain relationships, includ-
dt
ing monotonic and cyclic loading. It was applied in this study to model dg e ¼ (7)
the cyclic shear stress–strain response of granular soil. The constitu- G
tive relationship for the granular soil reaction can be written as
d g p ¼ ð1  aÞ½ðb þ gÞsinðz  d g Þjzjn d g (8)
t ¼ aG g þ ð1  aÞt y z (1)

where t and g = shear stress and shear strain, respectively; G = tan-


gent shear modulus; t y and g r = value of shear stress and strain at Densification Law
initiation of yielding in the soil, respectively; and a = nonnegative
parameter that controls the postyielding shear stiffness. The hyste- López-Querol and Blázquez (2006, 2007), Blázquez and López-
retic parameter z is a dimensionless inelastic parameter expressed in Querol (2006), and López-Querol et al. (2008) developed a densifi-
the following differential form: cation and dilatancy model of sand associated with the application
  of a certain level of cyclic strain. The basic concept underlying the
dz 1 dg dz theory is based on the notion of intrinsic time which, is a material
¼ g z (2)
dt g r dt dt property and depends on the deformation history (Valanis 1971).
The densification is produced mainly by continuous interparticle
where z = intrinsic time and measures the rearrangement of soil slips that result in a rearrangement of grain configurations. Thus,
grains and is expressed as the densification increment, d« p, can be expressed as being propor-
tional to the intrinsic time z calculated in Eq. (3), which is positive
 
b in compression, as follows:
d z ¼ 1 þ sinðz  d g Þ jzjn1 jd g j (3)
g
dɛp ¼ Fð z ; Dr Þd z (9)
In Eq. (3), parameters b, g, and n are dimensionless quantities
Densification-hardening function F is the function of initial rela-
that control the shape of the soil stress–strain hysteretic loops.
tive density Dr and rearrangement of intrinsic time z . Densification
Parameter n governs the sharpness of the transition from the linear
increments per cycle decrease with an increase in the number of
to the nonlinear range during initial virgin loading. Decreasing n
cycles and intrinsic time z .
leads to smoother transitions, with plastic behavior occurring at
lower loading levels. 1
F¼ (10)
Considering stiffness degradation effects induced by cyclic load- 1 þ av z
ing, Matasovic and Vucetic (1993) took the normalized residual
excess pore-water pressure as the degradation parameter, whereas The parameter av can be termed a strain-hardening constant.
Baber and Wen (1981) introduced the positive increasing energy-
related function into the Bouc–Wen model. In this paper, the tangent 4
10
elastic shear modulus, G, which changes with the loss of the initial Test data
stiffness Gmax of the soil as a result of dynamic loading, is given by Fitting curve
 m
Gmax 1 p
G¼ ¼ kpa (4)
h 1 þ h 0 Ed pa
log(Gmax/pa)

where pa = atmospheric pressure; k, m, and h 0 = dimensionless pa- 3


10
rameters; and h is defined as a linearly increasing function of the
energy Ed dissipated by the hysteretic model. In the Bouc–Wen
model, the dissipated energy can be calculated using the following
relationship (Sasani and Popov 2001):
ð
1
Ed ¼ ð1  aÞG g r zd g  ð1  aÞGð g r zÞ2 (5) 2
2 10 0 1
10 10
log(p/pa)
On the right-hand side of Eq. (5), the first term is the total work
done by the hysteretic stress, and the second term is the correspond- Fig. 4. Gmax/pap/pa relationship
ing elastic stored energy.

© ASCE 04016147-4 Int. J. Geomech.

Int. J. Geomech., -1--1


1 Dilatancy Flow Law

When the stress path and the phase transformation line intersect, the
0.8 behavior of the soil has been transformed from contraction to dila-
tion. In this situation, the stress–dilatancy relationship, which

0.6 Table 1. Model Parameters for Granular Soil


G/Gmax

Description Parameter Value


0.4
Bouc–Wen hysteresis model parameters a 0
Downloaded from ascelibrary.org by Indian Inst Of Technology Library, Kharagpur on 04/05/17. Copyright ASCE. For personal use only; all rights reserved.

b 0.5
0.2 g 0.5
n 1
Elastic shear modulus parameters k 476.09
0 −5 m 0.605
−4 −3 −2
10 10 10 10 h0 0.05
γ Shear strength parameters c 30.02
(a) w 41.98
Dilatancy angle w pt 32.62
0.2
Densification parameter av 8.00
Dilatancy parameter ap 1.33

0.15
60

0.1
ξ

40

Dynamic Stress σd (kPa) 20


0.05

0 −5 −4 −3 −2
10 10 10 10 −20
γ
(b)
−40
Fig. 5. Summary comparison of simulations versus data from cyclic Test
triaxial tests for Nuozhadu granular soil: (a) G/Gmax  g ; (b) j  g −60
−1 −0.5 0 0.5 1
Axial Strain ε1 (%)
(a)

60

40
Dynamic Stress σd (kPa)

0
20

0
−0.1
Mpt−q/p

−20

−40
−0.2 Model
σ =100kPa
3 −60
σ3=200kPa −1 −0.5 0 0.5 1
Axial Strain ε1 (%)
σ3=400kPa (b)
−0.3
−0.8 −0.6 −0.4 −0.2 0 0.2
Dilatancy rate d Fig. 7. Hysteresis loop for Nuozhadu granular soil (Kc = 1.0; Dr = 0.8;
p0 = 100 kPa; CSR = 0.27): (a) experimental measurements; (b) numeri-
Fig. 6. Stress dilatancy under monotonic loading cal model results

© ASCE 04016147-5 Int. J. Geomech.

Int. J. Geomech., -1--1


relates the plastic volumetric strain increment d« p and the plastic where Md is the tangent rebound modulus of the soil corresponding
shear strain increment d g p calculated by Eq. (8), is defined as to the initial (s v0) and current (s v) vertical effective stresses and
  can be determined experimentally for each type of sand; and Kd and
dɛp q
¼ ð1 þ a p Þ Mpt  (11) n* are model coefficients.
jd g p j p Byrne et al. (2004) proposed the following relationship for
undrained conditions:
where Mpt = slope of phase transformation line. The parameter ap
depends on the type of sand, the initial relative density, and the Ke
du ¼ e dɛ
p
(13)
cyclic stress ratio and must be estimated from test data. 1 þ K Kf =np

where parameter np = porosity of the soil skeleton; and Ke and Kf =


Downloaded from ascelibrary.org by Indian Inst Of Technology Library, Kharagpur on 04/05/17. Copyright ASCE. For personal use only; all rights reserved.

Pore Pressure Response bulk stiffness of the soil skeleton and the pore fluid, respectively,
defined as
Cetin and Bilge (2012) investigated semi-empirical excess pore-
water pressure models, including stress-based models, strain-based 2ð1 þ  Þ
Ke ¼ G (14)
models, energy-based models, and models based on plasticity theory. 3ð1  2 Þ
Martin et al. (1975) derived the relationship between the tenden-
cies of plastic volumetric straining, d« p, and the increment of the p20
excess pore-water pressure du as Kf ¼ (15)
ð1  Sr0 Þpa
ðs v Þ0:5 p
du ¼ Md dɛp ¼ Kd dɛ (12) where  = Poisson’s ratio of the soil; and Sr0 = initial saturation of
ðs v0 Þn the soil.

1
60
Test
0.8 40

20
0.6
q (kPa)
ru

0
0.4
−20

0.2
−40
Test
0 −60
0 20 40 60 80 0 20 40 60 80 100 120
Number of Cycles p (kPa)
(a) (a)

1 60
Model

0.8 40

20
0.6
q (kPa)
u

0
r

0.4
−20

0.2
−40
Model
0 −60
0 20 40 60 80 0 20 40 60 80 100 120
Number of Cycles p (kPa)
(b) (b)

Fig. 8. Excess pore-water pressure for Nuozhadu granular soil (Kc = Fig. 9. Effective stress path for Nuozhadu granular soil (Kc = 1.0; Dr =
1.0; Dr = 0.8; p0 = 100 kPa; CSR = 0.27): (a) experimental measure- 0.8; p0 = 100 kPa; CSR = 0.27): (a) experimental measurements; (b) nu-
ments; (b) numerical model results merical model results

© ASCE 04016147-6 Int. J. Geomech.

Int. J. Geomech., -1--1


In this paper, Eqs. (13)–(15) are used for calculating pore pres- The elastic shear modulus parameters, k and m in Eq. (4), can be
sure response. determined by plotting the tangential modulus at the beginning of
triaxial shearing and the mean effective stress in a double logarith-
mic diagram, as illustrated in Fig. 4.
Numerical Simulation of the Dynamic Triaxial Test: Then, the yield shear strain g r = t y/Gmax can be calculated. The
Calibration of Model Parameters Bouc–Wen model parameters a, b, g, n and the stiffness degrada-
tion parameter h 0 are then introduced to manipulate the shape of
The constitutive law presented earlier is a system of equations capa- the hysteresis loops in cyclic triaxial tests, the values of which
ble of reproducing the cyclic soil behavior and was calibrated with essentially influence the energy dissipation behavior of granular
soil. Both of them should be calibrated by best fitting the hysteresis
available experimental data for granular soil. Initial relative density
Downloaded from ascelibrary.org by Indian Inst Of Technology Library, Kharagpur on 04/05/17. Copyright ASCE. For personal use only; all rights reserved.

loops presented in cyclic loading experiments and matching the


Dr and porosity np were selected in accordance with the actual pro-
experimental shear modulus and damping curves simultaneously.
ject values. For Nuozhadu granular soil, initial relative density Dr =
Fig. 5 presents summary comparisons of model simulations of data
0.8, porosity np = 0.26, Poisson’s ratio  = 0.33, and initial satura-
for the shear modulus ratio G/Gmax g and damping ratio j  g
tion of soil Sr0 = 0.95.
from cyclic triaxial tests under various confining stresses.
The shear strength of granular soil t y was attained in static triax-
The slope of the phase transformation line Mpt was obtained as
ial tests as
6 sin w pt
t y ¼ c þ s tan w (16) Mpt ¼ (17)
3  sin w pt
where c and w = cohesion strength and internal friction angle of soil,
respectively. Both of them can be calibrated directly from the failure where w pt = dilatancy angle, which can be obtained from static tri-
envelopes according the Mohr-Coulomb failure criterion. axial tests. It depends on the type of sand and on the initial relative

300
1

200
Dynamic Stress σd (kPa)

0.8
100
0.6
0
ru

−100 0.4

−200 0.2
Test Test
−300
−2 −1 0 1 0
Axial Strain ε1 (%) 0 10 20 30
Number of Cycles
(a)
(a)

300 1

200
0.8
Dynamic Stress σd (kPa)

100
0.6
0
ru

0.4
−100

0.2
−200
Model Model
−300 0
−2 −1 0 1 2 0 10 20 30
Axial Strain ε1 (%) Number of Cycles
(b) (b)

Fig. 10. Hysteresis loop for Nuozhadu granular soil (Kc = 1.0; Dr = Fig. 11. Excess pore-water pressure for Nuozhadu granular soil (Kc =
0.8; p0 = 400 kPa; CSR = 0.37): (a) experimental measurements; (b) nu- 1.0; Dr = 0.8; p0 = 400 kPa; CSR = 0.37): (a) experimental measure-
merical model results ments; (b) numerical model results

© ASCE 04016147-7 Int. J. Geomech.

Int. J. Geomech., -1--1


density and is independent of the initial effective pressure of the granular soil under widely variable initial conditions, including vari-
soil. The densification parameter av can be generalized for sands ous consolidation conditions (Kc = 1.0, 2.5) and confining pressures
with different initial relative density values. Based on the data (p0 = 100, 400 kPa), were implemented.
reported in the literature, Cuellar et al. (1977) proposed the follow- A comparison of experimental and numerical results for the
ing relationship: stress–strain hysteresis loops is shown in Fig. 7 for granular soil
with p0 = 100 kPa, Kc = 1.0, Dr = 0.8, and CSR = 0.27. The cyclic
av ¼ 33:33D2r þ 61:66Dr  20 (18) dynamic stress–strain curves show a cycle-by-cycle degradation
in strength and stiffness. Stress amplitude attenuated and strain
The dilatancy parameter ap is calibrated according to the rela- amplitude increased cumulatively with increasing numbers of
tionship between the stress ratio and the dilatancy rate, d, as shown vibrations. As shown in Fig. 8, under undrained conditions, the
Downloaded from ascelibrary.org by Indian Inst Of Technology Library, Kharagpur on 04/05/17. Copyright ASCE. For personal use only; all rights reserved.

in Fig. 6. pore pressure ratio increased rapidly until induced liquefaction


Table 1 presents the model parameters, which were calibrated at ru = 100%. Fig. 9 presents the measured effective stress paths
on the basis of the monotonic and cyclic triaxial tests discussed (q-p) and their simulations, where the stress point accelerates
previously. toward the failure line after passing through the PTL. Similarly,
Figs. 10–12 show a comparison of the experimental and numeri-
cal results of cyclic loading with p0 = 400 kPa, Kc = 1.0, Dr =
Comparison of Model Simulations 0.8, and CSR = 0.37. It can be observed that the proposed model
was able to reproduce both contractive and dilative trends in
To evaluate the performance of the proposed densification–dilatancy behavior.
model, the following simulations of cyclic triaxial tests for Nuozhadu Figs. 13–18 show a comparison of the experimental and nu-
merical results for Nuozhadu granular soil under cyclic loading

300
Test 4
200

100 3

Axial Strain ε1 (%)


q (kPa)

0
2
−100

−200 1

−300 Test
0 100 200 300 400 500
0
p (kPa) 0 20 40 60
Number of cycles
(a)
(a)
300
4
Model
200

3
Axial Strain ε1 (%)

100
q (kPa)

0 2

−100
1
−200
Model
−300 0
0 100 200 300 400 500 0 20 40 60
p (kPa) Number of Cycles
(b) (b)

Fig. 12. Effective stress path for Nuozhadu granular soil (Kc = 1.0; Fig. 13. Axial accumulation strain for Nuozhadu granular soil (Kc =
Dr = 0.8; p0 = 400 kPa; CSR = 0.37): (a) experimental measurements; 2.5; Dr = 0.8; p0 = 100 kPa; CSR = 0.9); (a) experimental measure-
(b) numerical model results ments; (b) numerical model results

© ASCE 04016147-8 Int. J. Geomech.

Int. J. Geomech., -1--1


with a consolidated stress ratio of Kc = 2.5. The model was found pore pressure ratio could not reach a maximum value of 100%, and
to accurately predict the shape of the dynamic stress–strain loops liquefaction did not happen as a result of the existence of large ini-
and the increase of the accumulation strain with increasing num- tial deviatoric stress.
bers of cycles. It is remarkable that the predictions for cumulated After an analysis of the dynamic behavior of undrained satu-
axial strain were very accurate and reasonably accurate for rated granular soil, equations of the constitutive law were pre-
excess pore-water pressure as the number of cycles increased. sented. The Bouc–Wen model was introduced to describe the
The numerical results were found to have good agreement with cyclic shear stress–strain relationship, taking into consideration
the measurements for both isotropic and anisotropic stress con- the stiffness and strength degradation effects. The plastic shear
solidation conditions. strain increment and plastic modulus were obtained without the
concept of yield surface. Then, the densification and dilative
Downloaded from ascelibrary.org by Indian Inst Of Technology Library, Kharagpur on 04/05/17. Copyright ASCE. For personal use only; all rights reserved.

behaviors of sand were predicted. For undrained conditions, the


Conclusions development of pore pressure occurred with volumetric strain
increment.
In this paper, experimental observations from cyclic triaxial tests To check the performance of the model in reproducing the ex-
were analyzed, and a constitutive model describing the dynamic perimental observations, the measured data and numerical predic-
behavior of sand was presented. tions for various consolidation conditions and confining pressures
Samples of granular soil were isotropically or anisotropically were compared. It was found that most of the experimental results
consolidated and subjected to cyclic loading. For the isotropically obtained from cyclic loading tests could be reproduced satisfacto-
consolidated samples, stress reversal occurred during cyclic load- rily with the proposed model. In particular, cyclic mobility and plas-
ing, and double-amplitude axial strain developed. The samples tic strain accumulation were well captured by the model. The model
showed contractive behavior and axial plastic strain accumulation thus provides helpful tools for earth, rock, dam, and other geotech-
during cyclic loading under anisotropic consolidation. The excess nical engineering applications.

350 1

300
0.8
Deviator Stress q (kPa)

250

200 0.6
ru

150
0.4
100

50
0.2
0
Test Test
−50 0
0 1 2 3 4 5 0 20 40 60
Axial Strain ε1 (%) Number of Cycles
(a) (a)

350 1

300
0.8
Deviator Stress q (kPa)

250

200 0.6
ru

150
0.4
100

50
0.2
0 Model
Model
−50 0
0 1 2 3 4 5 0 20 40 60
Axial Strain ε1 (%) Number of Cycles
(b) (b)

Fig. 14. Stress–strain hysteresis loop for Nuozhadu granular soil (Kc = Fig. 15. Excess pore-water pressure for Nuozhadu granular soil (Kc =
2.5; Dr = 0.8; p0 = 100 kPa; CSR = 0.9: (a) experimental measurements; 2.5; Dr = 0.8; p0 = 100 kPa; CSR = 0.9): (a) experimental measure-
(b) numerical model results ments; (b) numerical model results

© ASCE 04016147-9 Int. J. Geomech.

Int. J. Geomech., -1--1


6 1400

5 1200

Deviator Stress q (kPa)


1000
Axial Strain ε1 (%)
4
800
3
600
2 400
Downloaded from ascelibrary.org by Indian Inst Of Technology Library, Kharagpur on 04/05/17. Copyright ASCE. For personal use only; all rights reserved.

1 200
Test
0
0
0 20 40 60 80 100 Test
Number of Cycles −200
0 1 2 3 4 5 6
(a) Axial Strain ε1 (%)
(a)
6
1400
5
1200
Axial Strain ε1 (%)

Deviator Stress q (kPa)


1000

3 800

2 600

400
1
Model 200
0
0 20 40 60 80 100 0
Number of Cycles
Model
(b) −200
0 1 2 3 4 5 6
Axial Strain ε1 (%)
Fig. 16. Axial accumulation strain for Nuozhadu granular soil (Kc =
(b)
2.5; Dr = 0.8; p0 = 400 kPa; CSR = 0.75): (a) experimental measure-
ments; (b) numerical model results
Fig. 17. Stress–strain hysteresis loop for Nuozhadu granular soil (Kc =
2.5; Dr = 0.8; p0 = 400 kPa; CSR = 0.75): (a) experimental measure-
Acknowledgments ments; (b) numerical model results

The authors are grateful to the anonymous referees for their valuable
d g p ¼ plastic shear strain increment;
comments and suggestions for improving the presentation of this
d« p ¼ plastic volumetric strain increment;
paper. The research described in this paper was financially supported
Ed ¼ dissipated energy;
by the Natural Science Foundation of China (Grants 51179024 and
G ¼ tangent shear modulus;
51379029).
Kc ¼ initial consolidation ratio;
Ke ¼ bulk stiffness of soil skeleton;
Notation Kf ¼ bulk stiffness of pore fluid;
k, m ¼ elastic shear modulus parameters;
The following symbols are used in this paper: Mpt ¼ phase transformation state stress ratio;
B ¼ parameter of Skempton; np ¼ porosity;
b, g, n ¼ parameters controlling the shape of the hysteresis p ¼ mean effective stress;
curve; pa ¼ atmospheric pressure;
Cc ¼ curvature coefficient of particle size distribution; p0 ¼ initial confining stress;
CSR ¼ cyclic stress ratio; q ¼ deviatoric stress;
Cu ¼ coefficient of uniformity of particle size ru ¼ excess pore pressure ratio;
distribution; Sr0 ¼ initial saturation;
c ¼ cohesion strength; z ¼ hysteretic parameter;
Dr ¼ relative density; a ¼ postyielding stiffness ratio;
d ¼ dilatancy rate; ap ¼ dilatancy parameter;
d g e ¼ elastic shear strain increment; av ¼ densification parameter;

© ASCE 04016147-10 Int. J. Geomech.

Int. J. Geomech., -1--1


0.8 Blázquez, R., and López-Querol, S. (2006). “Generalized densification law
for dry sand subjected to dynamic loading.” Soil Dyn. Earthquake Eng.,
0.7 26(9), 888–898.
Bouc, R. (1967). “Forced vibration of mechanical systems with hysteresis.”
0.6 Proc., 4th Conf. on Non-Linear Oscillation, Academia, Prague,
Czechoslovakia.
0.5 Byrne, P. M., Park, S. S., Beaty, M., Sharp, M., Gonzalez, L., and Abdoun,
T. (2004). “Numerical modeling of liquefaction and comparison with
centrifuge tests.” Can. Geotech. J., 41(2), 193–211.
ru

0.4
Camusso, M., and Barla, M. (2009). “Microparameters calibration for loose
0.3 and cemented soil when using particle methods.” Int. J. Geomech., 10
Downloaded from ascelibrary.org by Indian Inst Of Technology Library, Kharagpur on 04/05/17. Copyright ASCE. For personal use only; all rights reserved.

.1061/(ASCE)1532-3641(2009)9:5(217), 217–229.
0.2 Cetin, K. O., and Bilge, H. T. (2012). “Cyclic large strain and induced pore
pressure models for saturated clean sands.” J. Geotech. Geoenviron.
0.1 Eng., 10.1061/(ASCE)GT.1943-5606.0000631, 309–323.
Test
Chiaro, G., Koseki, J., and Sato, T. (2012). “Effects of initial static shear on
0 liquefaction and large deformation properties of loose saturated
0 20 40 60 80 100 Toyoura sand in undrained cyclic torsional shear tests.” Soils Found.,
Number of Cycles
52(3), 498–510.
(a) Cuellar, V., Silver, M. L., Krizek, R. J., and Silver, M. L. (1977).
“Densification and hysteresis of sand under cyclic shear.” J. Geotech.
0.8
Engrg. Div., 103(5), 399–416.
Dafalias, Y. F., and Manzari, M. T. (2004). “Simple plasticity sand model
0.7
accounting for fabric change effects.” J. Eng. Mech., 10.1061
/(ASCE)0733-9399(2004)130:6(622), 622–634.
0.6
Drosos, V. A., Gerolymos, N., and Gazetas, G. (2012). “Constitutive
model for soil amplification of ground shaking: Parameter calibra-
0.5
tion, comparisons, validation.” Soil Dyn. Earthquake Eng., 42,
255–274.
ru

0.4
El Shamy, U., De Leon, O., and Wells, R. (2013). “Discrete element method
study on effect of shear-induced anisotropy on thermal conductivity of
0.3
granular soils.” Int. J. Geomech., 10.1061/(ASCE)GM.1943-5622
0.2 .0000165, 57–64.
Erlicher, S., and Bursi, O. S. (2008). “Bouc–Wen-type models with stiffness
0.1 degradation: Thermodynamic analysis and applications.” J. Eng. Mech.,
Model 10.1061/(ASCE)0733-9399(2008)134:10(843), 843–855.
0 Gerolymos, N., and Gazetas, G. (2005). “Constitutive model for l-D cyclic
0 20 40 60 80 100 soil behaviour applied to seismic analysis of layered deposits.” Soils
Number of Cycles Found., 45(3), 147–159.
(b) Gerolymos, N., Vardoulakis, I., and Gazetas, G. (2007). “A thermo-poro-
visco-plastic shear band model for seismic triggering and evolution of
Fig. 18. Excess pore-water pressure for Nuozhadu granular soil (Kc = catastrophic landslides.” Soils Found., 47(1), 11–25.
2.5; Dr = 0.8; p0 = 400 kPa; CSR = 0.75): (a) experimental measure- Hyde, A. F., Higuchi, T., and Yasuhara, K. (2006). “Liquefaction, cyclic
ments; (b) numerical model results mobility, and failure of silt.” J. Geotech. Geoenviron. Eng., 10.1061
/(ASCE)1090-0241(2006)132:6(716), 716–735.
Kong, X., Liu, J., Zou, D., and Liu, H. (2016). “Stress-dilatancy relationship
g ¼ shear strain; of Zipingpu gravel under cyclic loading in triaxial stress states.” Int. J.
g r ¼ yield shear strain; Geomech., 10.1061/(ASCE)GM.1943-5622.0000584, 04016001.
« 1 ¼ axial strain; Li, X. S., and Ming, H. Y. (2000). “Unified modeling of flow liquefac-
z ¼ intrinsic time; tion and cyclic mobility.” Soil Dyn. Earthquake Eng., 19(5),
363–369.
h 0 ¼ stiffness degradation parameter;
Liu, C., and Xu, J. (2013). “Experimental study on the effects of initial con-
 ¼ Poisson’s ratio;
ditions on liquefaction of saturated and unsaturated sand.” Int. J.
j ¼ damping ratio; Geomech., 10.1061/(ASCE)GM.1943-5622.0000350, 04014100.
r d ¼ dry density; Liu, H., Zou, D., and Liu, J. (2014). “Constitutive modeling of dense grav-
s d ¼ dynamic axial stress; elly soils subjected to cyclic loading.” Int. J. Numer. Anal. Methods
t ¼ shear stress; Geomech., 38(14), 1503–1518.
t y ¼ yield shear stress; Liu, J. M., Zou, D. G., Kong, X. J., et al. (2016). “Stress-dilatancy of
w ¼ internal friction angle; and Zipingpu gravel in triaxial compression tests.” Sci. China Tech. Sci.,
w pt ¼ dilatancy angle. 59(2), 214–224.
López-Querol, S., and Blázquez, R. (2006). “Liquefaction and cyclic mobil-
ity model for saturated granular media.” Int. J. Numer. Anal. Methods
References Geomech., 30(5), 413–439.
López-Querol, S., and Blázquez, R. (2007). “Validation of a new endo-
Baber, T. T., and Wen, Y. K. (1981). “Random vibration hysteretic degrad- chronic liquefaction model for granular soil by using centrifuge test
ing systems.” J. Eng. Mech., 107(6), 1069–1087. data.” Soil Dyn. Earthquake Eng., 27(10), 920–937.
Bao, Y., and Sture, S. (2011). “Numerical modeling of cyclic mobility López-Querol, S., Fernández-Merodo, J. A., Mira, P., and Pastor, M.
based on fuzzy-set concepts in plasticity theory.” Comput. Geotech., (2008). “Numerical modelling of dynamic consolidation on granular
38(3), 375–382. soils.” Int. J. Numer. Anal. Methods Geomech., 32(12), 1431–1457.

© ASCE 04016147-11 Int. J. Geomech.

Int. J. Geomech., -1--1


Manahiloh, K. N., Muhunthan, B., and Likos, W. J. (2016). Pastor, M., Zienkiewicz, O. C., and Chan, A. H. C. (1990). “Generalized
“Microstructure-based effective stress formulation for unsaturated plasticity and the modelling of soil behaviour.” Int. J. Numer. Anal.
granular soils.” Int. J. Geomech., 10.1061/(ASCE)GM.1943-5622 Methods Geomech., 14(3), 151–190.
.0000617, D4016006. Sasani, M., and Popov, E. P. (2001). “Seismic energy dissipators for RC
Martin, G. R., Finn, W. D. L., and Seed, H. B. (1975). “Fundamentals of panels: Analytical studies.” J. Eng. Mech, 10.1061/(ASCE)0733
liquefaction under cyclic loading.” J. Geotech. Engrg. Div., 101, -9399(2001)127:8(835), 835–843.
423–438. Sze, H. Y., and Yang, J. (2014). “Failure modes of sand in undrained cyclic
Matasovic, N., and Vucetic, M. (1993). “Cyclic characterization of liquefi- loading: impact of sample preparation.” J. Geotech. Geoenviron. Eng.,
able sands.” J. Geotech. Eng., 10.1061/(ASCE)0733-9410(1993)119: 10.1061/(ASCE)GT.1943-5606.0000971, 152–169.
11(1805), 1805–1822. Valanis, K. C. (1971). “A theory of viscoplasticity without a yield surface.”
MATLAB [Computer software]. MathWorks, Natick, RI. Arch. Mech., 23(4), 517–551.
Mohamad, R., and Dobry, R. (1986). “Undrained monotonic and cyclic tri-
Downloaded from ascelibrary.org by Indian Inst Of Technology Library, Kharagpur on 04/05/17. Copyright ASCE. For personal use only; all rights reserved.

Yasuda, S., Harada, K., Ishikawa, K., and Kanemaru, Y. (2012).


axial strength of sand.” J. Geotech. Eng., 10.1061/(ASCE)0733 “Characteristics of liquefaction in Tokyo Bay area by the 2011 Great
-9410(1986)112:10(941), 941–958. East Japan earthquake.” Soils Found., 52(5), 793–810.
Papadimitriou, A. G., and Bouckovalas, G. D. (2002). “Plasticity model for Youd, T. L., and Idriss, I. M. (2001). “Liquefaction resistance of soils:
sand under small and large cyclic strains: a multiaxial formulation.” Soil Summary report from the 1996 NCEER and 1998 NCEER/NSF workshops
Dyn. Earthquake Eng., 22(3), 191–204. on evaluation of liquefaction resistance of soils.” J. Geotech. Geoenviron.
Papadimitriou, A. G., Bouckovalas, G. D., and Dafalias, Y. F. (2001). Eng., 10.1061/(ASCE)1090-0241(2001)127:4(297), 297–313.
“Plasticity model for sand under small and large cyclic strains.” J. Zhang, J. M., and Wang, G. (2012). “Large post-liquefaction deformation
Geotech. Geoenviron. Eng., 10.1061/(ASCE)1090-0241(2001)127: of sand, part I: physical mechanism, constitutive description and numer-
11(973), 973–983. ical algorithm.” Acta Geotech., 7(2), 69–113.

© ASCE 04016147-12 Int. J. Geomech.

Int. J. Geomech., -1--1

You might also like