You are on page 1of 15

Modified Models for Predicting Dynamic Properties of

Granular Soil Under Anisotropic Consolidation


Meysam Bayat 1 and Abbas Ghalandarzadeh 2

Abstract: The shear modulus and damping ratio are important parameters for the design of structures subjected to dynamic loading and can
be obtained by in situ and laboratory measurements. Previous research has lacked quantitative study of the effects of anisotropic consoli-
Downloaded from ascelibrary.org by University of Hong Kong on 02/03/23. Copyright ASCE. For personal use only; all rights reserved.

dation, especially in extension mode, on the dynamic properties of granular soil. The objective of the current study was to evaluate the
dynamic properties of sand-gravel mixtures for practical applications. To this end, resonant column, cyclic triaxial, and S-wave velocity
measurements under anisotropic confining conditions were conducted. The influence of mean effective consolidation stress, consolidation
stress ratio under constant and variable mean effective stresses, and gravel content on the maximum shear modulus Gmax , G=Gmax -γ and D-γ
curves are discussed. Simple formulations are presented to predict Gmax and the reference strain (γ r ) of sand-gravel mixtures using parameters
that are easy to obtain from test data. A modified empirical model is proposed based on the test results to estimate the shear modulus
degradation and damping ratio. The modified model is validated using experimental data from previous studies. The results indicate that
the proposed empirical model is capable of evaluating the shear modulus and damping ratio of granular soil. DOI: 10.1061/(ASCE)
GM.1943-5622.0001607. © 2019 American Society of Civil Engineers.
Author keywords: Shear modulus; Damping ratio; Consolidation stress ratio; Cyclic triaxial; Bender element; Resonant column.

Introduction smooth and aligned. One solution is the use of local strain instru-
mentation on triaxial specimens.
The inputs required for geotechnical seismic analysis include the Most previous studies on the dynamic characteristics of granular
maximum shear modulus (Gmax ), normalized shear modulus soil have focused on the isotropic stress condition; however, natural
(G=Gmax ), and damping ratio (D) for small to large shear strains soil under the K 0 condition or slopes are invariably subjected to
(γ). Atkinson and Sallfors (1991) categorized the strain levels into anisotropic stress. In the current study, an advanced cyclic triaxial
three groups: the very small strain level, where the normalized apparatus incorporating bender elements, local displacement meas-
shear modulus is constant in the elastic range; the small and urement, and a resonant column apparatus have been used to mea-
medium strain levels, where the normalized shear modulus varies sure G and D under isotropic and anisotropic conditions for a broad
nonlinearly with the shear strain; and the large strain level, where range of strain amplitudes.
the soil is close to failure and the normalized shear modulus is rel-
atively small. Advanced laboratory testing techniques such as
bender element (BE), resonant column (RC), and cyclic triaxial Previous Investigations on Anisotropic Consolidation
(CT) testing have been developed to study the dynamic properties Casagrande and Carrillo (1944) introduced two independent sources
of soil. Accurate determination of soil stiffness over a wide range of of anisotropy (inherent and induced) that influence the mechanical
strains is difficult to achieve in a conventional triaxial apparatus. In behavior of soil. Inherent anisotropy is attributed to sedimentation
a conventional triaxial test, the determination of the axial stiffness conditions and grain characteristics such as particle shape and
of a triaxial specimen is based on external measurements of dis- roughness that are independent of strain and loading history. In-
placement, which include a number of extraneous movements, duced anisotropy is induced after sedimentation, during the process
particularly in the small to medium strain range. Specimen dis- of nonelastic deformation. Induced stress anisotropy is defined as a
placements measured outside the triaxial cell introduce significant physical characteristic due exclusively to the strain associated with
errors in the computed strains. There are two main sources of er- applied stress that can occur under complex stress conditions. The
rors: the compliance of the apparatus because the tie bars and load consolidation stress ratio can be used to describe induced anisotropy
cell will be compressed during loading, and bedding errors because or anisotropic consolidation as
contacts between the specimen ends and the apparatus may not be
σcl0
K¼ 0 ð1Þ
σca
1
Assistant Professor, Dept. of Civil Engineering, Islamic Azad Univ.,
Najafabad Branch, Najafabad 8514143131, Iran. ORCID: https://orcid where σcl0 and σca
0 = lateral and axial effective consolidation stress,
.org/0000-0001-5525-5199. Email: bayat.m@pci.iaun.ac.ir respectively.
2
Associate Professor, School of Civil Engineering, College of Engineer- Stress conditions on the failure surface under different combi-
ing, Univ. of Tehran, Tehran 1417466191, Iran (corresponding author).
nations of static and cyclic shear stress can be simulated in the
ORCID: https://orcid.org/0000-0001-9939-7842. Email: aghaland@ut.ac.ir
Note. This manuscript was submitted on January 24, 2019; approved on laboratory by means of element testing such as cyclic triaxial tests.
August 22, 2019; published online on December 30, 2019. Discussion per- Researchers have thus far studied the effects of anisotropic consoli-
iod open until May 30, 2020; separate discussions must be submitted for dation and principal stress rotation on the mechanical behavior
individual papers. This paper is part of the International Journal of Geo- of granular material using a triaxial apparatus or hollow cylinder
mechanics, © ASCE, ISSN 1532-3641. apparatus (Doanh et al. 2012; Kumruzzaman and Yin 2010;

© ASCE 04019197-1 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019197


Sivathayalan 2011; Sivathayalan and Vaid 2002). Ishihara (1996) dynamic characteristics of gravelly soil using a large-scale cyclic
stated that the improved resistance of soil specimens subjected to triaxial apparatus. Their results show that GC is a key parameter
anisotropic consolidation is due to the increase in mean effective affecting the nonlinear behavior of soil at large strains. The results
consolidation stress, which is due to an increase in initial static also indicate that G=Gmax ratios of the specimens that contained
shear stress. Youd and Idriss (2001) stated that the cyclic resistance 60% or 80% gravel increased at shear strains larger than 0.1%;
of dilative soil increases as the initial static shear stress increases however, this trend was not observed in the behavior of specimens
and that of contractive soil decreases as the initial static shear containing 20% or 40% gravel. Lo Presti et al. (1997) studied the
stress decreases. In other words, initial static shear stress can in- effects of loading frequency and number of loading cycles (N) on
crease or decrease the cyclic strength depending on the level of dynamic properties of dry sand. The results indicate that the damp-
initial density and initial static shear (Vaid and Chern 1983; Yang ing ratio of dry sand is almost insensitive to loading frequency and
and Sze 2011). affected by the number of loading cycles. Investigations by Hardin
Zhou et al. (2017) studied the effects of void ratio, effective and Drnevich (1972) and Seed et al. (1984, 1986) have established
confining stress, and initial stress ratio on the dynamic properties that the number of loading cycles have minor influence on shear
Downloaded from ascelibrary.org by University of Hong Kong on 02/03/23. Copyright ASCE. For personal use only; all rights reserved.

of rockfill materials. Their results show that Gmax is affected by modulus for granular soils. As the number of loading cycles in-
confining pressure, void ratio, and initial stress ratio. However, creases, the amplitude of shear strain also increases, resulting in the
they studied the dynamic properties of rockfill materials only in degradation of shear modulus. Thus, shear modulus is a function of
compression mode. The experimental results showed that Gmax de- the number of loading cycles as well as the shear strain amplitude.
creased as the void ratio increased and increased as the confining But for a given level of strain amplitude, the number of loading
pressure or initial static deviatoric stress increased. As mentioned, cycles has very little influence on shear modulus. The effect of
consolidation stress ratio is a key parameter to defining the induced the number of cycles on dynamic properties was investigated by
anisotropic condition, but is not a comprehensive parameter for Darendeli (2001) for the loading cycles of N ¼ 1, 10, and 1,000.
explaining the effect of anisotropy on cyclic triaxial testing. The results show that the number of loading cycles, N, has no sig-
Stress conditions of soil indicate that Gmax in soil is mainly nificant influence measured on shear modulus and damping ratio. A
anisotropic. For example, previous studies indicated that the results comprehensive laboratory investigation using a highly sensitive
of bender element tests on granular soil under true triaxial condi- cyclic triaxial device has been carried out by Kokusho (1980) to
tions are mainly influenced by the principal stresses in the direction evaluate the effects of relative density on the shear modulus and
of wave propagation and particle motion and are nearly indepen- damping ratio for saturated Toyoura sand. The results indicate that
dent of the out-of-plane stress component (Bellotti et al. 1996; the manner of decreasing shear modulus with shear strain is almost
Roesler 1979; Wang and Mok 2008). Sun et al. (2013) carried out identical irrespective of the relative density. Kokusho (1980) indi-
a series of resonant column tests to study the effect of consolidation cated that the relation between the damping ratio and shear strain is
ratio on the dynamic shear modulus of undisturbed soil. They also established irrespective of the relative density of the specimens;
reported an increase in Gmax as well as G=Gmax as a result of an however, the values of the damping ratio slightly increased as the
increase in initial shear stress. Payan et al. (2016a) studied the void ratio became larger. Wichtmann and Triantafyllidis (2013)
effect of stress anisotropy on the Gmax of sand using the results also confirmed that for a constant value of the strain amplitude,
of bender element tests. The results showed that the Gmax of sand the normalized shear modulus and damping ratio are independent
under anisotropic compression stress states was greater than under of relative density.
isotropic stress states at a given mean effective stress. Despite these
research efforts, the dynamic characteristics of granular soil under
anisotropic stress conditions, especially in extension mode, are Empirical Model for Estimating G max
not fully understood and are currently the subject of interesting
research. Previously published experimental data show that there are a num-
ber of parameters affecting the Gmax of soil. These parameters are
related to the soil fabric and structure (void ratio and grading size
Previous Investigations on Dynamic Properties of distribution) and to its geological history (stress state and strain
Granular Soils history) (Cai et al. 2015; Carlton and Pestana 2016; Hardin and
According to previous research, the most important parameters that Black 1966; Payan et al. 2017; Senetakis et al. 2012; Senetakis
affect the dynamic properties of soil include the external parameters and Payan 2018). Based the experimental data, Hardin and Black
of loading properties (stress–strain path, stress–strain amplitude, (1966) stated that Gmax depends primarily on the void ratio of the
stress–strain rate, and stress–strain duration) and material proper- soil, e, and the mean effective confining pressure, σm0
ties (soil type, grain size distribution curve, shape of soil grains and Gmax ¼ AfðeÞðσm0 Þn ð2Þ
density) (Bayat and Ghalandarzadeh 2018; Seed and Idriss 1982).
Among the effective parameters, the most important ones affecting where A = material constant reflecting the effect of soil type, grain
the shear modulus and damping ratio of granular soil are amplitude size distribution, grain shape, and soil structure; n = mean confin-
of shear strain, effective confining pressure, and density. The less ing stress exponent reflecting the effect of stress state; and fðeÞ is a
important parameters include frequency of loading and degree of void ratio function reflecting the effect of soil density, which will
saturation (Hardin and Drnevich 1972). Tanaka et al. (1987) studied increase with a decrease in e. Menq (2003) summarized the values
the dynamic properties of reconstituted gravelly soil. Their results of A presented in the literature for reconstituted sandy and gravelly
showed that effective confining pressure has an important effect on soils. The results indicate that the values of A of both sandy and
the nonlinear dynamic parameters of gravelly soil as well as on the gravelly soils vary widely.
nonlinear dynamic parameters of sandy soil, such that gravelly soil In Eq. (2), the mean effective stress under anisotropic condition
behaves more linearly with an increase in isotropic confining stress. with lateral stress equal in all directions is calculated as
The results indicate that the soil containing a lower gravel content
0
(GC ¼ 25%) behaved slightly more linearly than soil containing a σca þ 2σcl0
σm0 ¼ ð3Þ
higher gravel content (GC ¼ 50%). Lin et al. (2000) studied the 3

© ASCE 04019197-2 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019197


In the current study, the void ratio function in Eq. (2) is that Regularity is deemed to be an effective parameter to incorporate the
proposed by Hardin and Black (1966) as follows: effect of particle shape on Dmin .
ð2.17 − eÞ2
fðeÞ ¼ ð4Þ
1þe Test Apparatus, Materials, and Testing Procedure

Cyclic Triaxial, Resonant Column, and Bender Element


G=G max Relationships Tests
A number of hyperbolic models have been proposed to describe the In the current study, the dynamic properties of soil at the medium to
dynamic behavior of soil (Amir-Faryar et al. 2017; Darendeli 2001; large shear strain range (shear strain ranging from 0.004% to 1%)
Menq 2003; Oztoprak and Bolton 2013; Zhang et al. 2005; Zhou were extracted from the results of cyclic triaxial testing as shown in
et al. 2017). Darendeli (2001) and Zhang et al. (2005) proposed a Fig. 1. The triaxial device used in this study was able to exert both
modified hyperbolic model that uses only two parameters: compression and extension loads while the specimen was consoli-
Downloaded from ascelibrary.org by University of Hong Kong on 02/03/23. Copyright ASCE. For personal use only; all rights reserved.

dated under anisotropic conditions. The procedure defined by the


1
G=Gmax ¼  α ð5Þ ASTM D3999 (ASTM 2011b) was used to identify the dynamic
γ
1þ γr
characteristics of the soil. Constant cyclic shear stress amplitude
loading started at a small amplitude and was gradually increased
where α = curvature parameter, which controls the curvature of the during triaxial testing. Each stage of loading consisted of 40 load-
G=Gmax curve; and γ r = reference shear strain value at which ing cycles. Calculations were performed to compute the value of
G=Gmax ¼ 0.5. The concept of reference strain (a shear strain shear modulus and damping ratio on each of the hysteresis loops
amplitude at which G=Gmax ¼ 0.5) is very useful for comparison at Cycles 1–5, 10, 20, and 40 as recommended in ASTM D3999
of nonlinearity parameters. Darendeli (2001) and Menq (2003) pre- (ASTM 2011b). Drainage was not allowed during each stage of
sented a power relation between γ r and σm0 . Some researchers are loading, and the specimen drainage valves were opened after each
of the opinion that grain size distribution curve parameters such as stage of cyclic loading to reestablish the effective consolidation
the Cu of granular material affect their dynamic parameters. For stress or dissipate the excess pore water pressure before moving
example, Menq (2003) presented γ r and α for sandy and gravelly on to the next stage. Measuring the volume change of the soil spec-
soil as a function of Cu and σm0 . imens during loading stage indicated that the void ratio was reduced
slightly during cyclic loading. In the current study, the average in-
crease of relative density, Dr , was about 12%, with a range from
Damping Relationships about 4% for low confining pressures to 19% for high confining
pressures. The hysteresis loop at the 10th cycle of loading in each
Researchers such as Hardin and Drnevich (1972); and Zhang et al. stage was used to compute the shear modulus and damping ratio
(2005) have assumed that D is a function of (γ=γ r ). On the other because it best represents typical seismic loading (Bayat and
hand, some researchers have expressed D as a function of G=Gmax . Ghalandarzadeh 2019; Zhang et al. 2005). The slope of a secant
The advantage of this approach is that the better defined G=Gmax is line that connected the extreme points on the hysteresis loop
used to infer D, which is more difficult to accurately measure. A (stress–strain curve) was defined as the elasticity modulus, E. Es-
common approach to model D is to relate the damping ratio to timations of the shear modulus, G, and corresponding cyclic shear
G=Gmax (Zhang et al. 2005). Payan et al. (2016b) investigated in- strains, γ, were obtained using Poisson’s ratio, ν, which was
fluence of particle shape on the small strain damping ratio (Dmin ) of taken as 0.5 for the saturated undrained condition. The damping
dry sand using torsional resonant column tests. The specimens are ratio can also be calculated as
consolidated under isotropic confining pressures ranging from 50
WD 1 Aloop
to 800 kPa. The results show that Dmin is a function of grading D¼ ¼ × ð6Þ
characteristics (D50 and Cu ), mean effective stress, and regularity. 4πW S 2π Gγ 2

Fig. 1. Typical examples of the combined results of the BE, RC, and CT tests (pure sand under confining pressure of 300 kPa).

© ASCE 04019197-3 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019197


where W D = energy dissipated in one cycle of loading; and W S = J2 tanðaÞ J J
a tanðaÞ − × ¼ þ ð9Þ
maximum strain energy stored during the cycle [see Kumar et al. J0 JL a J0 JL
(2017) and Delfosse-Ribay et al. (2004) for more details]. All the
cyclic triaxial tests were performed using a constant cyclic load where h = height of the specimen; ρ = soil density; and J, J 0 , and
with sinusoidal waves at a constant frequency of 1 Hz. In the cur- J L = mass polar moment of inertia of specimen, bottom part, and
rent study, two LVDTs were installed directly on either side of the top part, respectively. According to the relation of the RC appara-
triaxial specimen without puncturing the surrounding rubber mem- tus, it can be concluded that α is one of the main parameters of
brane to measure the relative local axial strain, which was then RC that is related to the geometry of the resonant column apparatus
taken as the average of the two LVDTs. A pair of bender elements that is a function of mass polar moment of inertia of top and
(transmitter and receiver) has been installed at the top and pedestal bottom parts of the resonant column apparatus. Based on the geom-
of the triaxial apparatus to apply shear waves in the axial specimen etry method, the values of J 0 and J L for the RC apparatus are
direction. In order to find the time of travel from the transmitter to 1.1819076 and 0.0688066 kg · m2 , respectively (Bayat and
the receiver, there are three different methods of interpretations Ghalandarzadeh 2019). This device measures the shear modulus
Downloaded from ascelibrary.org by University of Hong Kong on 02/03/23. Copyright ASCE. For personal use only; all rights reserved.

(Kumar and Madhusudhan 2010). In the current study, a number and damping under both isotropic and anisotropic conditions.
of preliminary BE tests were conducted to select the most appro- The RC device includes a compression loading ram that is only
priate method to find the time of travel. A comparison was made capable of loading under isotropic or compression mode.
among the travel times obtained by the different methods adopted To calibrate and verify the accuracy of the CT and RC appara-
for analyzing the results of the BE testing. The results indicate that tuses, several tests were conducted and their results analyzed.
the results of the first time of arrival method have the slightest dif- The load cell and local LVDTs used in triaxial apparatus were
ference with RC results. Based on the results, for finding the travel periodically calibrated. A set of aluminum probes made of a ver-
times, the first time of arrival was employed in all the tests. In this tical aluminum pipe with two horizontal aluminum bars or disks
method, travel time of the signal refers to the time between the start with different resonant frequencies were used to calibrate the
of the transmitted wave and the start of the receiver wave by ignor- RC device.
ing the initial weak signal or near-field effect that affects the shape
of the receiver signal (present if any). The time delay between the
sender and receiver signals and the travel distance (the tip-to-tip Test Program
distance between the two bender elements) are required to calculate In the current study, the consolidation stress ratio, mean effective
the shear wave velocity (V S ). In the current study, the tip-to-tip dis- confining stress, shear strain amplitude, and gravel content were
tance for each position of the BEs was calculated from the position selected as the controlled variables in the design of an experimen-
of the rod. By using the theory of shear wave propagation in an tal program. Based on the results, two general models were devel-
elastic body, the shear modulus calculated from the measurement oped for the degradation of the normalized shear modulus and
of shear wave velocity and the mass density of the soil, ρ, are damping ratio of granular material. The experimental data pre-
given by sented by other researchers were used to validate the proposed
model forms.
Gmax ¼ ρV 2S ð7Þ Table 1 summarizes the tests performed in this study. These in-
clude the resonant column and triaxial tests along with bender
Gu and Yang (2011) indicated that the RC test is the most re- element tests. Table 1 only includes the RC tests under isotropic
liable of the various test methods for evaluation of shear modulus. and compression modes. For example, Fig. 1 shows the shear
In the current study, a free-free RC apparatus in which the actuator modulus and damping ratio of the pure sand specimens at a con-
was mounted on top of the specimen was used to measure the shear fining pressure of 300 kPa under extension (K ¼ 1.5), isotropic
modulus and damping ratio at strain levels ranging from 0.0004% (K ¼ 1), and compression (K ¼ 0.5) modes. As shown in this
to 0.01%. The cuboidal top mass (so-called excitation head) was figure, the results of the tests under isotropic or compression
equipped with two electrodynamic exciters (minishakers), which condition include the results of BE, RC, and CT tests, whereas the
each accelerate a small mass. The minishakers made a sinusoidal results of the test under extension mode only contain BE and
rotational vibration at the top of the specimen. This acceleration CT tests.
and the resulting acceleration of the top mass were measured with
acceleration transducers that were installed in the cuboidal top
mass. The bottom end of the specimen was placed on a base plate Materials
that was free in rotation during the RC test. The weight of the top
Test specimens were prepared by combining subrounded sand
mass caused a small stress anisotropy of about 10 kPa. This may
with gravel to investigate the influence of GC, consolidation stress
be important for low confining pressures but this anisotropy is of
ratio, and mean confining pressure on the dynamic properties of
secondary importance for high confining pressures. To reduce the
soil. The sand used was a clean, uniformly graded fine sand with
error caused by the weight of the top mass, it was considered in
a silt content of less than 1% classified as SP according to the Uni-
the calculation of mean effective stress. A sinusoidal electrical
fied Soil Classification System (USCS) [ASTM D2487 (ASTM
signal was generated by a function generator, amplified, and ap-
2011d)] and the gravel is a uniformly graded soil with maximum
plied to electrodynamic exciters. The frequency of excitation was
and minimum particle sizes of 15 and 4.75 mm, respectively. The
varied until the resonant frequency fR of the system was found.
grain size distribution curves and maximum and minimum void ra-
Based on the boundary condition of the RC apparatus and basic
tios (emax and emin ) based on GC are presented in Figs. 2 and 3,
dynamic relations, the following equations are applicable to
respectively. Fig. 3 shows the significant decrease in the values
determine the shear modulus with the free-free resonant column
of emax and emin with an increase in GC from 0% to 50%. It indi-
device:
cates that gravel particles can float in the sand matrix at GC con-
  tents of 0% to 50% with little contact between them. The values for
2πhf R 2
G¼ ×ρ ð8Þ emax and emin increase with an increase in GC to more than 50%, at
a which point contact between the gravel grains increases.

© ASCE 04019197-4 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019197


Table 1. Summary of the test details
Effective confining Gravel σcl0
K¼ 0
stress (kPa) content (%) σca No. of tests Mean effective consolidation stress
100 0 0.5 14 BE, 10 RC, 14 CT Variable dependent on K
30
50 1
75 1.5
100
300 0 0.5 23 BE, 15 RC, 23 CT Variable dependent on K
30 0.75
50 1
75 1.25
100 1.5
300 0 0.5 5 BE, 3 RC, 5 CT Constant equal to 300 kPa
Downloaded from ascelibrary.org by University of Hong Kong on 02/03/23. Copyright ASCE. For personal use only; all rights reserved.

0.75
1
1.25
1.5
600 0 0.5 14 BE, 10 RC, 14 CT Variable dependent on K
30
50 1
75 1.5
100

Specimen Preparation
Cyclic triaxial and resonant column tests were conducted on the
cylindrical soil specimens, which were prepared using the wet-
tamping sample preparation technique with a low initial water con-
tent and undercompaction of the lower layers during preparation
(Ladd 1974). The most important advantages of the moist-tamping
method include its simplicity, the wide range of specimen densities,
and no soil particle segregation for granular soils containing fines
(Huang et al. 2015; Ishihara 1996). Moistening makes a weak ap-
parent cohesion between finer particles and causes the smaller par-
ticles to stick to larger ones. This may prevent or at least decrease
soil particle segregation in the moist-tamping method. In the cur-
rent study, to minimize the potential of particle segregation between
the gravel and sand particles during reconstitution, the wet-tamping
Fig. 2. Grain size distribution curves of the sand-gravel mixtures.
method was adopted because other preparation methods such as air

Fig. 3. Maximum and minimum composite void ratios versus gravel content.

© ASCE 04019197-5 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019197


pluviation or water pluviation tend to intensify soil particle segre- the triaxial apparatus, after completion of the consolidation stage,
gation. Every effort was made to prevent material segregation and a wave signal was generated using a signal wave generator to the
the sand-gravel mixture was gently placed into the mold with zero transmitter bender element before cyclic loading.
drop height. During specimen preparation, dry gravel was mixed
with dry sand at various weight ratios (0%, 30%, 50%, 75%, and
100%). The soil specimens were formed by compacting soil at a Results and Discussion
low water content of 3% in 10 uniform layers with a layer thickness
of about 2 cm. Then the wet soil was poured into the mold and a
Small Strain Shear Modulus (G 0 or G max )
hand tamper was used to create a uniform density across the layer
and balance the surface of each layer. Compaction of each layer In this work, the Gmax values corresponding to the smallest shear
was performed by the hand tamper with a graduated vertical rod to strain (γ ≈ 10−6 ) were obtained from BE and RC testing. Based on
control the desired height by abutting against the edges of the mold. the results, the effects of gravel content, consolidation stress ratio,
Each layer was compacted by pressing the hand tamper with a zero
Downloaded from ascelibrary.org by University of Hong Kong on 02/03/23. Copyright ASCE. For personal use only; all rights reserved.

drop height and without blow to achieve the target density. This
specimen preparation method produced cylindrical specimens with
a length-to-diameter ratio of about 2 (100 mm in diameter and
200 mm in length). All specimens were prepared to a constant rel-
ative density of about 60% (from 58% to 63%), which refers to the
density after the consolidation stage. However, the associated cost
and the practical size of triaxial testing devices limit the specimen
size. Accordingly, several studies have proposed minimum sizes for
triaxial specimens. The maximum allowable particle size to be
tested, d, was determined by the smallest dimension of the triaxial
apparatus, D. In terms of the D=d ratio, Penman (1971) stated that
the lower limit is 4 for a broad grading, or 6 for a narrow grading
after performing triaxial tests on rockfill dam material. ASTM
D4767 (ASTM 2011c) and ASTM D7181 (ASTM 2011a) also re-
quire a minimum cylindrical specimen diameter of 33 mm and a
specimen diameter at least equal to six times the diameter of the
largest particle within the test specimen. After specimen prepara-
tion and measurement of the diameter and height, full saturation is
required. A similar method has been used to saturate the cyclic tri-
axial and resonant column specimens. When the specimen prepa-
ration was completed, the specimens were deaired by flushing CO2
and then saturated by gravitational flushing of the deaired water.
The specimen was saturated according to ASTM D5311 (ASTM
2013) so that a Skempton’s B value larger than 0.95 was attained.
Following the saturation stage, the soil specimens were consoli-
dated under anisotropic or isotropic consolidation conditions. In

Fig. 5. Variation of Gmax versus consolidation stress ratio for the


Fig. 4. Comparison between results of RC and BE tests. specimens with various gravel contents.

© ASCE 04019197-6 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019197


and mean effective stress on Gmax are presented. Fig. 4 shows a Gmax increased as the confining pressure increased for a given K
comparison of the results using both techniques. Comparison of and GC. Fig. 5 also shows that Gmax decreased as the consolidation
Gmax-BE (Gmax of the BE test) and Gmax-RC (Gmax of the RC test) stress ratio increased, which may be because of a decrease in the
shows that the Gmax values obtained by both methods were in mean effective stress. For this reason, a set of tests at different con-
acceptable agreement for the sand-gravel mixture specimens; how- solidation stress ratios were carried out for a given mean effective
ever, the Gmax-BE for the pure gravel specimens were consistently stress (300 kPa). Fig. 6 shows that in this case also a decrease in K
lower than those for Gmax-RC. This indicates that a BE test does not increased Gmax at a constant mean effective stress. This means that
perform well for uniform coarse-grained soil. decreasing the consolidation stress ratio at a given mean effective
Fig. 5 shows Gmax values obtained from RC or BE tests in the stress increased Gmax ; this could be the result of an increase in ef-
small shear strain amplitude range versus consolidation stress ratio fective stress in the direction of wave propagation (Fig. 7). The RC
in terms of gravel content. The RC and BE results are marked in the and BE results are marked in Fig. 6. The results of the tests also
figure. When the results of both tests were available, the results of indicate that an increase in mean effective stress increased Gmax at a
the RC test were used because the RC technique is recognized to given K value. In conclusion, it can be stated that Gmax was affected
Downloaded from ascelibrary.org by University of Hong Kong on 02/03/23. Copyright ASCE. For personal use only; all rights reserved.

be the most reliable. As shown in Fig. 5, because of the inefficiency by both factors (K and σm0 ). This finding is in good agreement with
of the BE tests in the pure gravel specimens and also the lack of the findings of Ishihara (1996), Youd and Idriss (2001), Payan et al.
RC testing under extension mode, the results of Gmax for the pure (2016a), and Zhou et al. (2017), which revealed that the resistance
gravel under extension mode are not presented here. As expected, or Gmax of soil increases as the initial static shear stress increases.

Fig. 6. Variation of Gmax of the pure sand specimens versus consolidation stress ratio under constant and variable mean effective stress.

Fig. 7. Effective stress changes in the direction of wave propagation under anisotropic consolidation conditions.

© ASCE 04019197-7 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019197


Table 2. Parameters D50 , Cu , and Cc of the tested soils and the fitting
parameter A used in Eq. (2)
Gravel content (%) D50 Cu Cc A
0 0.61 3.59 0.88 4.9
30 1.15 7.02 0.63 4.2
50 4.75 15.31 0.76 3.8
75 7.07 2.67 1.03 4.3
100 0.61 1.80 0.90 6.9
Downloaded from ascelibrary.org by University of Hong Kong on 02/03/23. Copyright ASCE. For personal use only; all rights reserved.

Fig. 9. Values of Gmax predicted by Eq. (2) versus measured Gmax data.

As previously mentioned, the exact values of Gmax of the pure gravel


specimens under extension mode are ambiguous. In Fig. 10(c), Gmax
of the pure gravel is estimated only based on the CT test result. The
results indicate that the pure sand specimen behaved slightly more
linearly than the sand-gravel mixture specimens and the G=Gmax -γ
Fig. 8. Variation of n as a function of K. curves for the sand-gravel mixture or pure gravel specimens were
lower than that of the pure sand specimen at a given shear strain
amplitude.
The results indicate that the specimens that contained 50%
On the other hand, as shown in Fig. 5, Gmax increased with an in-
gravel were more nonlinear than other specimens when deformed
crease in GC up to 50% and then decreased. The pure gravel spec-
to the same strain level at a given confining pressure, consolidation
imens recorded higher Gmax values than the specimens containing stress ratio, and relative density. An increase in GC also slightly
75% gravel, but the specimen containing 50% gravel recorded the increased the damping ratio compared with that of the pure sand
highest Gmax at a given consolidation stress ratio. Similar findings specimen. Under the same conditions, however, the D-γ curves
were presented by Hubler et al. (2018). The specimens containing of the specimens at various GC values were similar, but the speci-
50% gravel had the lowest void ratio when compared with the other men containing 50% gravel exhibited the highest damping ratio.
specimens at a given relative density. The D50 , Cu , Cc values and The highest degradation of the normalized modulus curve and
the fitting parameter A of the tested soils based on gravel content the high damping ratio in the specimens containing 50% gravel in-
are presented in Table 2. The fitting parameter A was taken as a dicate that the part of the sand in contact with the surface of the
constant for each subgroup soil. However, the values of the fitting gravel grains caused separation and sliding of the gravel grains dur-
parameter A presented in Table 2 show that they are almost the ing cyclic loading. In general, the G=Gmax curves degraded more
same (with an average value of about 4.3) for various gravel con- quickly as the GC increased up to a threshold value (GCth ¼ 50%),
tents, except for the pure gravel, which has been increased. The at which point the sand grains were close to each other and the
results of the tests presented in Fig. 6 indicate that the effect of gravel grains broke away. These findings are in good agreement
σm0 on Gmax is dependent on the K value. Nonlinear least-squares with experimental results reported by previous researchers (Lin
analysis was also used to determine the n and A values. Based on et al. 2000; Tanaka et al. 1987).
the results of RC and BE testing presented in Fig. 6, as shown in The effect of K on the G=Gmax and D curves for pure sand under
Fig. 8, n in Eq. (2) was adopted as a function of consolidation stress constant σ30 (300 kPa) is presented in Fig. 11(a). As shown, an in-
ratio. Fig. 8 shows that the n value increases by changing stress crease in K decreased G=Gmax and increased D at a given shear
state from extension (K ¼ 1.5) to compression (K ¼ 0.5). This strain amplitude. This finding is consistent with the results of Sun
means that the effect of σm on Gmax depends on the stress state. The et al. (2013), which demonstrated an increase in G=Gmax as a result
results indicate that the effect of σm on the increase of Gmax under of an increase in initial shear stress. Fig. 11(b) shows the effect of
compression mode is more than the extension mode. The RC and K on the G=Gmax and D curves of pure sand with constant σm0
BE results in Fig. 8 are marked with squares and circles, respec- (300 kPa). As shown, the effect of K on the G=Gmax curves under
tively. The Gmax values obtained from the tests were used to evalu- constant mean effective stress was negligible, but on the D-γ curves
ate empirical Eq. (2). The quality of the predictions of Gmax are it was significant. Nonetheless, the effect of anisotropic stress on
shown in Fig. 9, where it can be seen that most data points plot the dynamic parameters under constant effective confining stress
close to the bisecting line. is more significant than its effect under constant mean effective
stress.
The effects of consolidation stress ratio and mean effective
G=G max -γ and D-γ Curves
stress on the nonlinear behavior and dynamic properties of sand-
The typical normalized modulus reduction and damping ratio gravel specimens are easier to discuss in terms of the reference
curves of the soil containing different amounts of gravel with threshold shear strain, γ r , which is simply the value of γ at
σ30 ¼ 300 kPa and K ¼ 0.5, 1, and 1.5 are shown in Fig. 10. G=Gmax ¼ 0.5. Parameter γ r is regarded as a soil parameter for

© ASCE 04019197-8 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019197


Downloaded from ascelibrary.org by University of Hong Kong on 02/03/23. Copyright ASCE. For personal use only; all rights reserved.

Fig. 10. Variation of G=Gmax and D as a function of γ in terms of gravel content: (a) K ¼ 0.5; (b) K ¼ 1; and (c) K ¼ 1.5.

hyperbolic models to define the stiffness degradation curve (Hardin of pure sand decreased as the σm0 decreased at a given stress ratio.
and Drnevich 1972). Fig. 12 shows γ r versus K for specimens with The increase in γ r with an increase in effective confining pressure is
different GC values under a constant σ30 (300 kPa) and also pure in good agreement with the results of previous studies (Hardin and
sand specimens under a constant σm0 (300 kPa). The results pre- Drnevich 1972; Ishihara 1996). The results presented for pure sand
sented for pure sand specimens show that an increase in K under specimens indicate that the consolidation stress ratio has a signifi-
a constant σm0 caused little change in γ r . On the other hand, the γ r cant effect on γ r when σm0 is variable. The results of previous

© ASCE 04019197-9 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019197


Downloaded from ascelibrary.org by University of Hong Kong on 02/03/23. Copyright ASCE. For personal use only; all rights reserved.

Fig. 11. Variation of G=Gmax and D as a function of γ for different consolidation stress ratios: (a) σ3 ¼ 300 kPa, σm ¼ variable; and
(b) σ3 ¼ variable, σm ¼ 300 kPa.

studies and the current study indicate that γ r can vary significantly equation has been proposed that includes the effect of anisotropic
with e, σm0 , GC, and Cu . As shown in Fig. 11(b), the values of γ r are consolidation as
almost independent of K. Based on the tests results, γ r can be
1
defined for granular soil as follows: G=Gmax ¼ h  α iβ ð11Þ
γ
 0 0.54 1þ γr
ð2.66 − eÞ2 −ðGCþ0.15Þ σm
γ r ð%Þ ¼ 0.13 × × cu × ð10Þ
5.70 þ e 100 where γ r = reference strain, which is obtained using Eq. (10); and α
and β = curvature parameters found to be related to the grain size
where σm0 is in kilopascals. Genetic programming (GP) was used to distribution curve parameters and applied stress condition. The ef-
develop Eq. (10). GP is an automated invention machine that is fect of these parameters on the normalized modulus reduction curve
adept at solving optimal instruction set problems. Nonlinear is presented in Fig. 14. In the current study, curvature parameters
least-squares analysis was also used to fit a set of experimental data (i.e., α and β) were considered as a function of GC, K, and σm0 .
points to a model. The parameters of the model were determined in In the current study, D–Dmin was considered as a function
such a way that the sum of the squares of the difference between a of G=Gmax , which has already been used by previous studies
predicted model value and the corresponding actual data point can (Wichtmann and Triantafyllidis 2013; Zhang et al. 2005). The value
be minimized. The results were used to evaluate the empirical equa- of (D–Dmin ) for all test data with upper and lower bounds are plot-
tion in Eq. (10). Fig. 13 shows the quality of the predictions of γ r . ted versus the corresponding G=Gmax in Fig. 15. The general damp-
The predicted γ r values based on Eq. (10) are plotted versus the γ r ing equation adopted in the current study has the following form:
values obtained from testing. As shown in Fig. 13, most data points
plot close to the bisecting line. D–Dmin ¼ fðG=Gmax Þ ¼ η1 ðG=Gmax Þ2 − η2 ðG=Gmax Þ þ η3
Despite the many models proposed to represent the dynamic ð12Þ
behavior of soil, currently there is no model for granular soil that
includes the effect of anisotropic consolidation. In the current where Dmin = small strain damping ratio; and ηi = model parameter
study, based on a set of experimental results, a modified hyperbolic in which η1 − η2 þ η3 ≈ 0 (for very small shear strains such that

© ASCE 04019197-10 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019197


Downloaded from ascelibrary.org by University of Hong Kong on 02/03/23. Copyright ASCE. For personal use only; all rights reserved.

Fig. 15. Relationships between G=Gmax and D–Dmin .

The best-fit curve between the tests results for Dmin and the soil
Fig. 12. Variation of reference strain γ r versus consolidation stress ra- characteristics is expressed as
tio for the specimens with various gravel contents (confining pressure =
300 kPa).   0 −0.1 
1 −0.68 0.15 σm
Dmin ¼ ð1 þ eÞ × cu × ð13Þ
72 100

The precision of the proposed model [Eq. (11)] was investigated


by plotting the measured versus predicted values of G=Gmax
[Fig. 16(a)]. Good agreement between the prediction of the ex-
tended empirical equations and the tests results is demonstrated in
Fig. 16(a). Fig. 16(b) shows the D-γ data for all specimens versus
the predicted values obtained with the proposed model [Eqs. (12)
and (13)]. As seen, the predicted D values are in good agreement
with the measured values. However, between Figs. 16(a and b)
under compression, it is clear that prediction of the normalized
shear modulus was more accurate than prediction of the damping
ratio. This occurred because accurate measurement of the damping
ratio is more difficult than for the normalized shear modulus.
Values of the parameters of the model (α, β, and ηi ) were
calculated based on nonlinear least-squares analysis that lead to
Fig. 13. Values of γ r predicted by Eq. (10) versus measured γ r values. the best fit of the model to the data. Fig. 17 shows the values used
for parameters α and β versus K for all the soil subgroups. The
results indicate that the parameters α and β are not only dependent
on K value but are also related to the mean effective confining
G=Gmax ¼ 1 and D ¼ Dmin ), where η1 , η2 , and η3 are considered as stress and grading characteristics. There is no obvious regularity
a function of K and σm0 and are independent of soil types (or GC). between either parameter and K. Parameter α varied from 0.8 to
The effect of soil type (or GC) has been applied on damping 1.54 with an average of 1.11, whereas parameter β varied from
ratio indirectly via G=Gmax . The preceding equation has already 0.63 to 1.61 with an average of 1.05. The results show that α
been used by previous researchers (Zhang et al. 2005). Using has less sensitivity to K, while β has higher sensitivity to K. This
the best-fit curves, the value of ηi can be calculated for each test. result can also be obtained by comparing Figs. 11(a) and 14(b). In

Fig. 14. Effect of the curvature parameters on the normalized modulus reduction curve.

© ASCE 04019197-11 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019197


Downloaded from ascelibrary.org by University of Hong Kong on 02/03/23. Copyright ASCE. For personal use only; all rights reserved.

Fig. 16. Comparison of measured: (a) G=Gmax values predicted by Eq. (11); and (b) D values predicted by Eqs. (12) and (13).

Fig. 17. Relationships between α or β and K.

Fig. 18. Variation of ηi as a function of gravel content.

other words, K and β have the same effect on the normalized modu- verifying the model (digitization of the test data was undertaken).
lus reduction curve. The specimens were derived from various countries and were
Fig. 18 shows ηi versus GC. The aim of presenting Figs. 17 tested under a variety of conditions. Table 3 also details values
and 18 is to find a relationship between the model parameters of specimen properties (materials type, D50 , Cu , and e) and stress
and loading characteristics or soil grading characteristics. As pre- state (σ30 and K) that were reported in the original publications. As
viously mentioned, ηi is considered a function of σm0 and GC and shown in Table 3, for the soils in the database the consolidation
is independent of K. There is no obvious regularity between ηi pressure varies from 20 to 3,000 kPa; void ratio, e0 , ranges from
and GC. 0.179 to 0.643. The model was fitted to the entire data set. The best-
fit values of parameters (α, β, and ηi ) for each of the soils studied
Validation of the Models with Previous Studies are listed in Table 3 together with the corresponding coefficients of
The data from Kokusho (1980) for sand, Tanaka et al. (1987) for determination, R2 . Fig. 19 shows a plot of predicted and measured
sandy gravel, Araei et al. (2010) for gravel, and Zhou et al. (2017) values of G=Gmax and D for the 29 tests detailed in Table 3. As
for the anisotropic condition were used to validate the ability of the shown in the figure, the correlation coefficient between the mea-
proposed models. Table 3 summarizes the sources of data used in sured and predicted values is 0.994 and 0.967 for normalized shear

© ASCE 04019197-12 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019197


Table 3. Summary of database used to validate G=Gmax and D equations
Shear modulus
Confining
Soil properties parameters Damping ratio parameters
Testing Materials stress
Publication apparatus type Materials symbol D50 (mm) Cu e0 (kPa) K α β R2 η1 η2 η3 R2
Zhou et al. Large-scale Limestone HZY-1 17.34 7.52 0.235 1,000 0.67 0.694 1.29 0.945 0.204 0.353 0.150 0.984
(2017) cyclic rock fills HZY-1 17.34 7.52 0.235 2,000 0.4 0.665 0.972 0.128 0.326 0.201
triaxial HZY-2 14.2 30.25 0.195 2,000 0.4 0.783 0.567 0.120 0.310 0.188
HZY-2 14.2 30.25 0.195 3,000 0.67 0.772 0.649 0.141 0.309 0.169
HZY-3 7.85 85.29 0.205 1,000 0.67 0.797 0.685 0.181 0.381 0.200
HZY-3 7.85 85.29 0.205 3,000 0.67 0.822 0.527 0.409 0.677 0.271
HZY-4 17.34 7.52 0.235 1,000 0.67 0.785 1.306 0.422 0.655 0.270
HZY-4 17.34 7.52 0.235 2,000 0.4 0.682 0.998 0.423 0.720 0.300
HZY-5 17.34 17.76 0.179 2,000 0.4 0.747 0.702 0.399 0.688 0.302
Downloaded from ascelibrary.org by University of Hong Kong on 02/03/23. Copyright ASCE. For personal use only; all rights reserved.

HZY-5 17.34 17.76 0.179 3,000 0.67 0.718 0.818 0.418 0.665 0.251
Araei et al. Large-scale Rock-fill S-3BMES 10.743 652.4 0.28 7,00 1 0.825 1.326 0.998 0.071 0.253 0.172 0.976
(2010) cyclic materials S-3BMES 10.743 652.4 0.28 200 1 0.780 1.619 0.035 0.239 0.205
triaxial S-3AMES 4.114 139.1 0.281 700 1 0.920 0.990 0.087 0.243 0.143
S-3AMES 4.114 139.1 0.281 200 1 0.918 1.198 0.094 0.255 0.154
S-SC 4.100 95.53 0.227 800 1 0.842 0.756 0.040 0.227 0.199
S-SC 4.100 95.53 0.227 200 1 0.761 1.184 0.001 0.275 0.278
S-SK 4.166 142.6 0.335 500 1 0.697 0.960 0.189 0.361 0.231
S-SK 4.166 142.6 0.335 300 1 0.758 0.952 0.068 0.274 0.255
Tanaka et al. Large-scale Sand- CH-1 1 5.65 0.404 400 1 0.954 1.089 0.985 0.154 0.378 0.216 0.996
(1987) cyclic gravel CH-2 1 5.65 0.402 200 1 0.891 1.154 0.169 0.386 0.202
triaxial mixtures CH-3 1 5.65 0.406 100 1 0.912 1.2 0.263 0.501 0.232
CH-10 2 11.25 0.285 400 1 0.944 1.022 0.161 0.383 0.179
CH-11 2 11.25 0.288 200 1 0.841 1.101 0.067 0.237 0.153
CH-12 2 11.25 0.288 100 1 1 1.414 0.256 0.470 0.208
Kokusho Cyclic Toyoura T-300 0.198 1.34 0.643 300 1 0.809 0.989 0.995 0.454 0.803 0.362 0.962
(1980) triaxial test sand T-200 0.198 1.34 0.643 200 1 0.795 0.957 0.439 0.78 0.354
T-100 0.198 1.34 0.643 100 1 0.863 0.826 0.455 0.819 0.372
T-50 0.198 1.34 0.643 50 1 0.959 0.672 0.450 0.808 0.369
T-20 0.198 1.34 0.643 20 1 1.044 0.561 0.412 0.752 0.342

Fig. 19. Comparison of measured and calculated: (a) G=Gmax ; and (b) D.

modulus reduction and damping ratio curves, respectively, which curve. Based on the results, the following conclusions can be
verifies the feasibility of the proposed models. drawn:
1. The Gmax of the gravel-sand mixtures mainly depended on the
mean effective confining stress, void ratio, consolidation stress
Conclusions ratio, and gravel content. The increase in mean effective confin-
ing stress or decrease in the consolidation stress ratio led to an
An experimental study was performed on sand-gravel mixtures increase in Gmax that is in good agreement with previous studies.
to investigate the effect of gravel content, confining stress, and con- The increase in Gmax owing to the decrease in consolidation
solidation stress ratio on the dynamic deformation properties of stress ratio under constant mean effective confining stress could
the mixtures using triaxial, bender element, and resonant column be the result of an increase in effective stress in the vertical
testing. New models were developed for maximum shear modulus, direction. However, the variation in Gmax with a decrease in
reference strain, shear modulus reduction, and material damping consolidation stress ratio was more impressive under constant

© ASCE 04019197-13 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019197


effective confining stresses. In addition, pure sand exhibited sig- 5. Eqs. (12) and (13) were used to estimate (D−Dmin ) and Dmin for
nificantly lower Gmax values in comparison with the sand-gravel granular soil. The (D–Dmin ) is considered a function of G=Gmax ,
mixture and pure gravel, such that Gmax increased nonlinearly as which has already been used by previous researchers. The mod-
GC increased. This continued until a maximum value for Gmax el parameters (η1 , η2 , and η3 ) were found to be related to mean
was reached at a GC of 50%. effective stress and grading characteristics. The Dmin is a func-
2. Eq. (2) proposed by Hardin and Black (1966) was used to es- tion of void ratio, uniformity, and mean effective stress. The pro-
timate Gmax for granular soil and has been used extensively in posed equations were validated by the tests results from the
previous studies. The Gmax value was shown to be best predicted current study and those of previously published studies.
as a power function of two variables, void ratio function fðeÞ ¼
ð2.17 − eÞ2 =ð1 þ eÞ and mean effective stress (σm0 ) and para-
meters A and n, which are functions of grading characteristics Data Availability Statement
and the consolidation stress ratio, respectively. It was shown
that the proposed equations are capable of predicting the All data, models, and code generated or used during the study ap-
Downloaded from ascelibrary.org by University of Hong Kong on 02/03/23. Copyright ASCE. For personal use only; all rights reserved.

Gmax value of granular soil under isotropic and anisotropic stress pear in the published article.
states.
3. The normalized modulus reduction and material damping
curves of the saturated sand-gravel mixtures were mainly af- Notation
fected by the mean effective confining stress, consolidation
stress ratio, and gravel content, which is confirmed by previous The following symbols are used in this paper:
studies. The proposed model facilitates estimation of shear mod- Cu = uniformity coefficient;
ulus degradation and damping ratio for routine geotechnical pro- D = damping ratio;
jects for a wide range of granular soil. More specifically, Dmin = small strain damping ratio;
• Pure sand experienced less normalized modulus reduction Dr = relative density;
with an increase in shear strain than the sand-gravel mixture e = void ratio;
and pure gravel. The sand with an intermediate gravel con- fðeÞ = void ratio function;
tent exhibited the greatest decrease in the normalized mod- fR = resonant frequency;
ulus. This indicates that, under this condition, the part of the GC = gravel content;
sand that is in contact with the surface of the gravel grains Gmax = maximum shear modulus;
caused sliding and rolling of the gravel grains during cyclic G=Gmax = normalized shear modulus;
loading. K = consolidation stress ratio;
• Pure sand exhibited slightly lower damping values in V s = shear wave velocity;
comparison with the sand-gravel mixtures. Increasing the
γ = shear strain;
gravel content up to about 50% slightly increased the damp-
γ r = reference shear strain;
ing values, followed by a slight decrease.
• The mean effective confining stress and consolidation stress η, α, β = model parameters; and
ratio influenced the normalized modulus reduction and ρ = soil density.
damping ratio, especially at large strains, such that an in-
crease in the effective confining stress basically shifted
the modulus reduction curves to the right and either upward References
for G=Gmax or downward for D. An increase in consolidation
Amir-Faryar, B., M. S. Aggour, and R. H. McCuen. 2017. “Universal
stress ratio moved the modulus reduction curves to the left
model forms for predicting the shear modulus and material damping
and either downward for G=Gmax or upward for D. The effect of soils.” Geomech. Geoeng. 12 (1): 60–71. https://doi.org/10.1080
of consolidation stress ratio on G=Gmax was negligible under /17486025.2016.1162332.
a constant mean effective confining stress, although this ef- Araei, A. A., H. R. Razeghi, S. H. Tabatabaei, and A. Ghalandarzadeh.
fect was more impressive when the mean effective confining 2010. “Dynamic properties of gravelly materials.” Sci. Iran. Trans.
stress was varied. In other words, the effect of mean effective A: Civ. Eng. 17 (4): 245.
confining stress was more significant than the effect of K on ASTM. 2011a. Method for consolidated drained triaxial compression test
the dynamic properties of granular soil. for soils. ASTM D7181. West Conshohocken, PA: ASTM.
4. A relationship was fitted to the experimental shear modulus ASTM. 2011b. Standard test methods for the determination of the modulus
and damping properties of soils using the cyclic triaxial apparatus.
degradation curves in the form of Eq. (11) that features three
ASTM D3999/D3999M-11e1. West Conshohocken, PA: ASTM.
curve-fitting parameters: reference strain γ r and curvature para- ASTM. 2011c. Standard test method for consolidated undrained triaxial
meters α and β. The nonlinear curve fitting was used to deter- compression test for cohesive soils. ASTM D4767. West Consho-
mine the coefficients of the model. Curvature parameters α hocken, PA: ASTM.
and β were found to be related to grading characteristics and K. ASTM. 2011d. Standard practice for classification of soils for engineering
It was found that the γ r of granular soil depended on the mean purposes (Unified Soil Classification System). ASTM D2487. West
effective stress, void ratio, and grading characteristics. Here, the Conshohocken, PA: ASTM.
value of γ r can be predicted by Eq. (10), which offers the best ASTM. 2013. Standard test method for load controlled cyclic triaxial
strength of soil. ASTM D5311. West Conshohocken, PA: ASTM.
fit to the tests results. Comparisons between the predictive equa-
Atkinson, J. H., and G. Sallfors. 1991. “Experimental determination of soil
tions developed in this study and the test results of the current
properties.” In Vol. 3 of Proc., 10th European Conf. on Soil Mechanics
study and previously published studies show that the modified and Foundation Engineering, 915–956. Florence, Italy.
model is useful for describing the shear modulus degradation Bayat, M., and A. Ghalandarzadeh. 2018. “Stiffness degradation and damp-
of granular soil for a wide strain range under isotropic and ani- ing ratio of sand-gravel mixtures under saturated state.” Int. J. Civ. Eng.
sotropic consolidation states. 16 (10): 1261–1277. https://doi.org/10.1007/s40999-017-0274-8.

© ASCE 04019197-14 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019197


Bayat, M., and A. Ghalandarzadeh. 2019. “Influence of depositional Payan, M., K. Senetakis, A. Khoshghalb, and N. Khalili. 2016b. “Influence
method on dynamic properties of granular soil.” Int. J. Civ. Eng. 17 (6): of particle shape on small-strain damping ratio of dry sands.” Géotech-
907–920. https://doi.org/10.1007/s40999-019-00412-7. nique 66 (7): 610–616. https://doi.org/10.1680/jgeot.15.T.035.
Bellotti, R., M. Jamiolkowski, D. C. F. Lo Presti, and D. A. O’Neill. 1996. Payan, M., K. Senetakis, A. Khoshghalb, and N. Khalili. 2017. “Charac-
“Anisotropy of small strain stiffness in Ticino sand.” Géotechnique terization of the small-strain dynamic behaviour of silty sands; contri-
46 (1): 115–131. https://doi.org/10.1680/geot.1996.46.1.115. bution of silica non-plastic fines content.” Soil Dyn. Earthquake Eng.
Cai, Y., Q. Dong, J. Wang, C. Gu, and C. Xu. 2015. “Measurement of small 102 (Nov): 232–240. https://doi.org/10.1016/j.soildyn.2017.08.008.
strain shear modulus of clean and natural sands in saturated condition Penman, A. D. M. 1971. Rockflll. Watford, UK: Building Research Station.
using bender element test.” Soil Dyn. Earthquake Eng. 76 (Sep): Roesler, S. K. 1979. “Anisotropic shear modulus due to stress anisotropy.”
100–110. https://doi.org/10.1016/j.soildyn.2014.12.013. J. Geotech. Eng. Div. 105 (7): 871–880.
Carlton, B. D., and J. M. Pestana. 2016. “A unified model for estimating Seed, H. B., and I. M. Idriss. 1982. Ground motions and soil liquefaction
the in-situ small strain shear modulus of clays, silts, sands, and gravels.” during earthquakes. Oakland, CA: Earthquake Engineering Research
Soil Dyn. Earthquake Eng. 88 (Sep): 345–355. https://doi.org/10.1016/j Institute.
.soildyn.2016.01.019. Seed, H. B., R. T. Wong, I. M. Idriss, and K. Tokimatsu. 1984. Moduli and
Casagrande, A., and N. Carrillo. 1944. “Shear failure of anisotropic
Downloaded from ascelibrary.org by University of Hong Kong on 02/03/23. Copyright ASCE. For personal use only; all rights reserved.

damping factor for dynamic analyses of cohesionless soils. Geotechni-


materials.” J. Boston Soc. Civ. Eng. 31 (4): 74–87. cal Engineering Rep. No. UCB/EERC-84/14. Berkeley, CA: Univ. of
Darendeli, M. B. 2001. “Development of a new family of normalized California at Berkeley.
modulus reduction and material damping curves.” Ph.D. dissertation, Seed, H. B., R. T. Wong, I. M. Idriss, and K. Tokimatsu. 1986. “Moduli and
Dept. of Civil Engineering, Univ. of Texas at Austin. damping factors for dynamic analysis of cohesionless soils.” J. Geotech.
Delfosse-Ribay, E., I. Djeran-Maigre, R. Cabrillac, and D. Gouvenot. 2004. Eng. 112 (11): 1016–1032.
“Shear modulus and damping ratio of grouted sand.” Soil Dyn. Earthquake Senetakis, K., A. Anastasiadis, and K. Pitilakis. 2012. “The small-strain
Eng. 24 (6): 461–471. https://doi.org/10.1016/j.soildyn.2004.02.004. shear modulus and damping ratio of quartz and volcanic sands.” Geo-
Doanh, T., Z. Finge, and S. Boucq. 2012. “Effects of previous deviatoric strain tech. Test. J. 35 (6): 20120073. https://doi.org/10.1520/GTJ20120073.
histories on the undrained behaviour of Hostun RF loose sand.” Geotech. Senetakis, K., and M. Payan. 2018. “Small strain damping ratio of sands
Geol. Eng. 30 (4): 697–712. https://doi.org/10.1007/s10706-011-9487-9. and silty sands subjected to flexural and torsional resonant column
Gu, X., and J. Yang. 2011. “Laboratory measurement of shear stiffness of excitation.” Soil Dyn. Earthquake Eng. 114 (Nov): 448–459. https://doi
decomposed granite.” In Proc., 15th European Conf. on Soil Mechanics .org/10.1016/j.soildyn.2018.06.010.
and Geotechnical Engineering. Amsterdam, Netherlands: IOS Press.
Sivathayalan, S. 2011. “Hollow cylinder torsional shear tests to evaluate
Hardin, B. O., and W. L. Black. 1966. “Sand stiffness under various triaxial
the role of principal stress directions on cyclic resistance.” In Proc.,
stresses.” J. Soil Mech. Found. Div. 92 (2): 27–42.
Pan-Am Canadian Geotechnical Conf. Richmond, BC, Canada:
Hardin, B. O., and V. V. P. Drnevich. 1972. “Shear modulus and damping in
Canadian Geotechnical Society.
soils: Measurement and parameter effects.” J. Soil Mech. Found. Div.
Sivathayalan, S., and Y. P. Vaid. 2002. “Influence of generalized initial state
98 (6): 603–624.
and principal stress rotation on the undrained response of sands.” Can.
Huang, A. B., W. J. Chang, H. H. Hsu, and Y. J. Huang. 2015. “A mist
Geotech. J. 39 (1): 63–76. https://doi.org/10.1139/t01-078.
pluviation method for reconstituting silty sand specimens.” Eng. Geol.
Sun, J., M. Gong, and X. Tao. 2013. “Dynamic shear modulus of undis-
188 (Apr): 1–9. https://doi.org/10.1016/j.enggeo.2015.01.015.
turbed soil under different consolidation ratios and its effects on surface
Hubler, J. F., A. Athanasopoulos-Zekkos, and D. Zekkos. 2018. “Mono-
ground motion.” Earthquake Eng. Eng. Vib. 12 (4): 561–568. https://doi
tonic and cyclic simple shear response of gravel-sand mixtures.” Soil
.org/10.1007/s11803-013-0197-6.
Dyn. Earthquake Eng. 115: 291–304.
Ishihara, K. 1996. Soil behaviour in earthquake geotechnics. New York: Tanaka, Y., Y. Kudo, Y. Yoshida, and M. A. Ikemi. 1987. A study on the
Clarendon Press. mechanical properties of sandy gravel-dynamic properties of reconsti-
Kokusho, T. 1980. “Cyclic triaxial test of dynamic soil properties for wide tuted sample. Rep. No. U87019. Tokyo: Central Research Institute of
strain range.” Soils Found. 20 (2): 45–60. https://doi.org/10.3208 Electric Power Industry.
/sandf1972.20.2_45. Vaid, Y. P., and J. C. Chern. 1983. “Effect of static shear on resistance to
Kumar, J., and B. N. Madhusudhan. 2010. “A note on the measurement of liquefaction.” Soils Found. 23 (1): 47–60. https://doi.org/10.3208
travel times using bender and extender elements.” Soil Dyn. Earthquake /sandf1972.23.47.
Eng. 30 (7): 630–634. https://doi.org/10.1016/j.soildyn.2010.02.003. Wang, Y. H., and C. M. Mok. 2008. “Mechanisms of small-strain shear-
Kumar, S. S., A. M. Krishna, and A. Dey. 2017. “Evaluation of dynamic modulus anisotropy in soils.” J. Geotech. Geoenviron. Eng. 134 (10):
properties of sandy soil at high cyclic strains.” Soil Dyn. Earthquake 1516–1530. https://doi.org/10.1061/(ASCE)1090-0241(2008)134:10
Eng. 99 (May): 157–167. https://doi.org/10.1016/j.soildyn.2017.05.016. (1516).
Kumruzzaman, M. D., and J.-H. Yin. 2010. “Influence of principal stress Wichtmann, T., and T. Triantafyllidis. 2013. “Effect of uniformity coeffi-
direction on the stress-strain-strength behaviour of completely decom- cient on G=Gmax and damping ratio of uniform to well-graded quartz
posed granite.” Facta Universitatis-Ser.: Archit. Civ. Eng. 8 (1): 79–97. sands.” J. Geotech. Geoenviron. Eng. 139 (1): 59–72. https://doi.org/10
Ladd, R. S. 1974. “Specimen preparation and liquefaction of sands.” .1061/(ASCE)GT.1943-5606.0000735.
J. Geotech. Eng. Div. 100 (10): 1180–1184. Yang, J., and H. Y. Sze. 2011. “Cyclic behaviour and resistance of saturated
Lin, S.-Y., P. S. Lin, H.-S. Luo, and C. H. Juang. 2000. “Shear modulus and sand under non-symmetrical loading conditions.” Géotechnique 61 (1):
damping ratio characteristics of gravelly deposits.” Can. Geotech. J. 59–73. https://doi.org/10.1680/geot.9.P.019.
37 (3): 638–651. https://doi.org/10.1139/t99-133. Youd, T. L., and I. M. Idriss. 2001. “Liquefaction resistance of soils: Sum-
Lo Presti, D. C. F., M. Jamiolkowski, O. Pallara, A. Cavallaro, and mary report from the 1996 NCEER and 1998 NCEER/NSF workshops
S. Pedroni. 1997. “Shear modulus and damping of soils.” Geotechnique on evaluation of liquefaction resistance of soils.” J. Geotech. Geoen-
47 (3): 603–617. https://doi.org/10.1680/geot.1997.47.3.603. viron. Eng. 127 (4): 297–313. https://doi.org/10.1061/(ASCE)1090
Menq, F.-Y. 2003. “Dynamic properties of sandy and gravelly soils.” -0241(2001)127:4(297).
Ph.D. dissertation, Dept. of Civil, Architectural, and Environmental Zhang, J., R. D. Andrus, and C. H. Juang. 2005. “Normalized shear modu-
Engineering, Univ. of Texas. lus and material damping ratio relationships.” J. Geotech. Geoenviron.
Oztoprak, S., and M. D. Bolton. 2013. “Stiffness of sands through a labo- Eng. 131 (4): 453–464. https://doi.org/10.1061/(ASCE)1090-0241
ratory test database.” Géotechnique 63 (1): 54–70. https://doi.org/10 (2005)131:4(453).
.1680/geot.10.P.078. Zhou, W., Y. Chen, G. Ma, L. Yang, and X. Chang. 2017. “A modified
Payan, M., A. Khoshghalb, K. Senetakis, and N. Khalili. 2016a. “Small-strain dynamic shear modulus model for rockfill materials under a wide range
stiffness of sand subjected to stress anisotropy.” Soil Dyn. Earthquake of shear strain amplitudes.” Soil Dyn. Earthquake Eng. 92 (Jan): 229–238.
Eng. 88 (Sep): 143–151. https://doi.org/10.1016/j.soildyn.2016.06.004. https://doi.org/10.1016/j.soildyn.2016.10.027.

© ASCE 04019197-15 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019197

You might also like