You are on page 1of 15

HOSTED BY Available online at www.sciencedirect.

com

ScienceDirect
Soils and Foundations 59 (2019) 571–585
www.elsevier.com/locate/sandf

Technical Paper

Small strain dynamic behavior of two types of carbonate sands


Huan He a,1, Wei Li b,2, Kostas Senetakis a,⇑
a
City University of Hong Kong, Hong Kong Special Administrative Region
b
The University of Hong Kong, Hong Kong Special Administrative Region

Received 2 July 2017; received in revised form 21 July 2018; accepted 3 November 2018
Available online 31 January 2019

Abstract

The study reports on the small-strain dynamic behavior of two types of carbonate sands from Western Australia and the Philippines.
Basic characterization of the soils was performed in terms of specific gravity, grading information, angle of shear strength at critical state,
particle shape characterization and composition analysis. Piezo-element inserts were utilized to carry out the dynamic tests. For the Wes-
tern Australia (WA) carbonate sand, both bender and extender element tests were performed, thus the shear modulus, the Young’s mod-
ulus and the Poisson’s ratio were examined. Both vertical and lateral bender element tests were performed on a set of specimens from the
Philippines (PH) carbonate sand to study the shear modulus and, from which, no fabric anisotropy of the reconstituted specimens was
found. It was observed that the overconsolidated specimens had higher stiffness than those during the first loading stages for both car-
bonate sands. In the pressure range of the study, grain breakage was small and its effect on the behavior of the samples was almost neg-
ligible. Empirical equations in the literature proposed from quartz sands could not predict the stiffness of the carbonate soils
satisfactorily. In this regard, a preliminary study was carried out adopting the assumed void ratio that only considers the inter-
particle voids instead of the summation of inter- and intra-particle voids; based on this concept, the predicted and measured stiffness
(including both small-strain shear modulus, Gmax and small-strain Young’s modulus, Emax) were found to be satisfactorily close.
Ó 2019 Production and hosting by Elsevier B.V. on behalf of The Japanese Geotechnical Society.
This is an open access article under CC BY-NC-ND license. (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Keywords: Carbonate sand; Small-strain dynamic properties; Stiffness; Poisson’s ratio; Stiffness anisotropy; Bender/extender element test; Resonant
column test; Intra-particle voids; Inter-particle voids

1. Introduction small strains, the behavior is linear-elastic and the soil


properties, commonly expressed in terms of stiffness, mate-
Soil mechanics modeling requires the investigation of rial damping and Poisson’s ratio, are pressure-dependent.
soil behavior at different strain ranges, including the very With a focus on the very small strain behavior of unce-
small strain range with deformations to be less than mented sands, previous laboratory and theoretical studies
103%, the small-to-medium strain range with typical have indicated that the mechanical properties of soils are
strains to be between about 103% and 101% and the strongly linked to the grain characteristics. These charac-
behavior at larger strains (Atkinson, 1993; Vucetic, 1994; teristics include, for example, the shape of grains (e.g.
Ishihara, 1996; Kramer, 1996; Clayton, 2011). At very Cho et al., 2006; Payan et al., 2016), the coefficient of uni-
formity (e.g. Iwasaki and Tatsuoka, 1977; Menq, 2003;
Wichtmann and Triantafyllidis, 2009; Senetakis and
Peer review under responsibility of The Japanese Geotechnical Society.
⇑ Corresponding author. Madhusudhan, 2015), the type of the material and its com-
E-mail address: ksenetak@cityu.edu.hk (K. Senetakis). position (Cascante and Santamarina, 1996; Jovicic and
1
Formerly University of New South Wales (UNSW), Sydney, Australia. Coop, 1997; Fioravante, 2000; Senetakis et al., 2012;
2
Formerly City University of Hong Kong, Hong Kong Special Senetakis and Madhusudhan, 2015). The investigation of
Administrative Region.

https://doi.org/10.1016/j.sandf.2018.11.003
0038-0806/Ó 2019 Production and hosting by Elsevier B.V. on behalf of The Japanese Geotechnical Society.
This is an open access article under CC BY-NC-ND license. (http://creativecommons.org/licenses/by-nc-nd/4.0/)
572 H. He et al. / Soils and Foundations 59 (2019) 571–585

the very small strain range of behavior requires the applica- material Poisson’s ratio (m) as well as the comparison of
tion of dynamic test methods, and two typical methods the results between different methods in measuring soil stiff-
used in many research laboratories include the resonant ness (i.e. bender elements against resonant column). As for
column, which is a medium frequency method, and the the carbonate sand from the Philippines, the small strain
use of piezo-element testing, which is a high frequency shear modulus was investigated, with a particular focus
technique (Richart et al., 1970; Shirley and Anderson, on the comparison between different shear moduli, named
1975; Ishihara, 1996; Jovicic et al., 1996; Leong et al., Gvh, Ghv and Ghh to examine the fabric anisotropy of
2005, 2009; Gu et al., 2013, 2015). reconstituted specimens. The dynamic behavior of the car-
Previous studies on the small strain behavior of sands bonate sands was analyzed during isotropic compression
have focused, primarily, on quartz soils. However, other and additional information is provided in the paper during
types of granular soils may be of significant interest in isotropic swelling of the samples where the samples were
geotechnical engineering practice. For example, carbonate overconsolidated. The grain breakage of the materials
sands, which are found in natural deposits, comprise typi- was measured and quantified after the tests and its effect
cal foundation soils for offshore engineering structures such on the dynamic properties of the carbonate sands is briefly
as oil platforms and subsea infrastructure. Due to their discussed in the paper. Published equations proposed for
geological origin, the carbonate soils are found, predomi- quartz sands were used to predict the behavior of the car-
nantly, in calcareous sediments (i.e. calcium carbonate) bonate soils and the effects of inter- and intra-particle voids
(Fioravante, 2000; Coop and Airey, 2003; Fioravante on the prediction accuracy of these expressions are also
et al., 2013). The carbonate sands are usually found either discussed.
cemented or uncemented in different gradings, which are
closely related to the environmental factors they are 2. Materials and methods
exposed to (e.g. water temperature, depth, etc.) (Coop
and Airey, 2003). In the literature, carbonate sands, in a 2.1. Characterization of materials used
cemented or uncemented state, have been extensively exam-
ined particularly in the range of large deformations (Coop, Two types of poorly-graded carbonate sands were used
1990; Airey, 1993; Coop and Atkinson, 1993; Coop et al., in the study. Both soils are classified as SP based on the
2004; Miao and Airey, 2013, among others). In general, USCS system. The basic characteristics of the materials
the carbonate sands are found to have high shear strength are summarized in Table 1. The finer grained carbonate
but also high compressibility which is due to the irregular sand (mean grain size D50 = 0.23 mm) is a beach sand with
shape and interlocking of angular particles and presence origin from Western Australia (WA), while the coarser
of intra-particle voids. These characteristics of the carbon- grained one (D50 = 0.50 mm) was from the Philippines
ate soils result in breakable and of low strength grains (PH). The Western Australia carbonate soil has a specific
(Coop, 1990; Fioravante, 2000; Coop et al., 2004), which gravity of solids, Gs, equal to 2.68, a coefficient of unifor-
in turn may give rise to an unstable behavior of offshore mity, Cu, equal to 1.70, while the Gs and Cu values for the
structures founded on carbonate soil deposits, for example, Philippines carbonate sand are equal to 2.72 and 1.85,
low side friction and high deformation during pile driving. respectively. The grading curves for both sands are given
There have been quite a few empirical equations proposed in Fig. 1. Both soils have a higher specific gravity than that
in the literature predicting the stiffness – pressure relation- for quartz sands (typically in the range of 2.62–2.67). Con-
ship of quartz sands (e.g., Hardin and Richart, 1963; solidated drained monotonic triaxial shearing tests were
Hardin, 1978; Iwasaki et al., 1978; Chung et al., 1984; conducted on both materials and their angles of shear
Saxena and Reddy, 1989; Menq, 2003; Wichtmann and strength at critical state (ucritical) are summarized in Table 1.
Triantafyllidis, 2009, 2010; Senetakis et al., 2012; Note that the ucritical value for the PH sand was found to be
Senetakis and Madhusudhan, 2015; Payan et al., 2016, greater in magnitude than that of the WA sand (39.4°
2017 among others). However such studies for carbonate against 36.4°), which is possibly related to the more irreg-
sands are relatively limited (Jovicic and Coop, 1997; ular grain shape of the former material.
Fioravante, 2000; Fioravante et al., 2013; Senetakis and The shape of the grains from both carbonate sands was
He, 2017; He et al., 2017). examined visually through an optical microscope and SEM
In this study, a comprehensive experimental investiga- images. In total, forty grains were randomly chosen from
tion was conducted exploring the small strain dynamic the parent sand and the empirical chart proposed by
behavior of two types of uncemented carbonate sands from Krumbein and Sloss (1963) was used to quantify the
two different locations (Western Australia and the Philip- particle shape descriptors, namely sphericity, S, and round-
pines). For the carbonate soil from Western Australia, ness, R. Thereafter, the regularity, q, which represents the
experiments were carried out using bender/extender ele- mean arithmetic value of the shape descriptors S and R
ment tests and the results were compared with additional (first introduced by Cho et al., 2006), was estimated. From
tests from resonant column. The focus of this testing pro- the SEM images in Fig. 2, it can be seen that both sands
gram was the study of the small strain stiffness in terms have grains which are rather irregularly shaped. The
of shear modulus (Gmax), Young’s modulus (Emax) and particle shape descriptors and the corresponding standard
H. He et al. / Soils and Foundations 59 (2019) 571–585 573

Table 1
Basic characteristics of the Western Australia (WA) and the Philippines (PH) carbonate sands.
Material WA Carbonate PH Carbonate
Sand Sand
Specific gravity Gs 2.68 2.72
Mean grain size D50 (mm) 0.23 0.50
Coefficient of uniformity Cu 1.70 1.85
Coefficient of curvature Cc 0.90 0.88
ucritical (°) 36.4 39.4
Sphericity S (mean) 0.60 0.53
Sphericity standard deviation 0.20 0.20
Roundness R (mean) 0.53 0.30
Roundness standard deviation 0.18 0.18
Regularity q 0.565 0.415
Normalized Silicon weight percentage (%) 1.80 0.42
Normalized Calcium weight percentage (%) 41.20 44.60

PH CS WA CS
100
90
Percent finer by weight

80
70
60
50
40
30
20
10
0
0.01 0.1 1 10
Grain size (mm)
Fig. 1. Grading curves of both types of carbonate sands: the Philippines
(PH) and the Western Australia (WA) carbonate sand.

deviation among the values are summarized in Table 1. The


PH carbonate sand had slightly more angular grains than
the WA sand. The average value of the regularity for the
WA carbonate sand was found to be 0.565, while that for
the PH carbonate sand was 0.415 (Table 1).
In order to quantify the chemical composition of the
carbonate sands, scanning electron microscope (SEM) –
Energy Dispersive Spectroscopy (EDS) analysis was car-
ried out for both materials (Knight et al., 2002). The
EDS analysis indicated that both sands are dominated by
calcium compounds with a very small inclusion of silicon.
Typical plots of the EDS analysis from both sands are
given in Fig. 3.
Fig. 2. Scanning electron microscope (SEM) image of: (a) the Western
2.2. Dynamic experimental methods and testing program Australia carbonate sand; (b) the Philippines carbonate sand.

For the WA carbonate sand, fourteen specimens were can accommodate specimens of 50 mm in diameter and
prepared and tested at the Geotechnical Engineering Lab- 100 mm in length and the second is the Hardin-type which
oratory of UNSW Australia (Centre for Infrastructure can accommodate specimens of 70 mm in diameter and
Engineering and Safety) using two different resonant col- 140 mm in length. Both resonant columns are embedded
umn apparatuses. One apparatus is the Stokoe-type which with vertically placed piezo-element inserts on the top
574 H. He et al. / Soils and Foundations 59 (2019) 571–585

3500 out shear waves propagating in the horizontal direction


Calcium (a)
3000 with one of them inducing vertical vibration of the soil par-
Intensity (Counts)

Calcium ticles (measuring Vs,hv and Ghv) while the other triggers
2500
Oxygen horizontal particle motion (measuring Vs,hh and Ghh).
2000 The system gives the advantage of testing the shear wave
1500 velocity in both vertical and horizontal directions simulta-
Silicon neously and thus the fabric anisotropy of the specimen can
1000
be examined, where the corresponding wave velocity and
500 stiffness measured from the vertical bender elements are
0 denoted as Vs,vh and Gvh.
0 1 2 3 4 5 6 Thereafter, Eqs. (1) and (2) were used to quantify, in a
Energy (keV) straightforward way, the small-strain shear modulus, Gmax,
for all the experiments of the study for the WA and PH car-
3500 bonate sands and constrained modulus, Mmax, when ben-
Calcium (b)
3000 der/extender element tests were performed which was the
Intensity (Counts)

2500
case for the WA carbonate sand. In Eqs. (1) and (2), qs is
the mass density of the specimen and Eq. (3) is used to
2000 Calcium quantify the Poisson’s ratio (m) for a given specimen based
1500 Oxygen on the measurement of Vp and Vs, where the shear wave
1000 velocity corresponds in this case to Vs,vh. The small-strain
Young’s modulus Emax was quantified indirectly based on
500
both bender/extender element test results for the WA sand
0 specimens through Eq. (4).
0 1 2 3 4 5 6
2
Energy (keV) Gmax ¼ qs  ðV s Þ ð1Þ
2
Fig. 3. Energy dispersive spectroscopy (EDS) analysis of: (a) the Western M max ¼ qs  V p ð2Þ
Australia carbonate sand specimen; (b) the Philippines carbonate sand
specimen. 0:5  V  V
2
p
2
s
m¼ ð3Þ
V 2p  V 2s
M max  ð1 þ mÞ  ð1  2  mÞ
cap and the pedestal, respectively, which work as both ben- Emax ¼ ð4Þ
ð 1  mÞ
der elements and extender elements, similar to the configu-
ration described by Leong et al. (2009). This allows the Note that the soil specimens were reasonably assumed
propagation of S-waves (bender element configuration) to be homogeneous based on the specimen preparation
and P-waves (extender element configuration) through methods. Generally, the specimens were divided into multi-
the body of the specimen, thus measurements of shear ple layers of soil with the same weight and less compaction
and compression wave velocities (Vs and Vp respectively) was applied on the initial layers while greater effort was
can be conducted based on the obtained time arrivals of applied on the top ones. Moreover, no anisotropic stress
the waves with direction of the wave propagation along state was applied on any of the specimens and significant
the axis of the specimen. The Stokoe-type resonant column intrinsic anisotropy of the material was not observed as dis-
has been described in the previous studies by Payan et al. cussed later in the paper. Consequently, Eqs. (3) and (4),
(2016, 2017) and He and Senetakis (2016). The Hardin- which should be used for homogeneous, isotropic and elas-
type resonant column has been described in the recent stud- tic materials, could be applied to estimate the Poisson’s
ies by Li et al. (2018) and Senetakis and He (2017). ratio and Young’s modulus of the soil, in this case. In
The Philippines carbonate sand was tested in a Bishop Eqs. (1) and (2), dry mass density was used for the case
and Wesley type of triaxial system (Bishop and Wesley, of dry specimens, but for saturated specimens, equivalent
1975) at the Soil Mechanics Laboratory of City University density was adopted (as described by Youn et al., 2008),
of Hong Kong. The apparatus is equipped with both verti- which accounts for the relative movement of the solid
cal and lateral bender element inserts (no extender element and fluid phases during high frequency dynamic excitation
mode is available in this case). Like the vertical bender ele- (i.e. bender element tests). The use of equivalent density in
ment system, the lateral bender element inserts are in pairs, the analysis of bender element test results, provides a better
but they are placed horizontally oppose to each other to matching between bender element and resonant column
measure the horizontally propagating shear wave velocity. tests, as also described by Li and Senetakis (2017b).
There are two bender element inserts placed orthogonally The dynamic testing program and specimen details for
in a T shape on each side of the system (described in the the WA carbonate sand are summarized in Table 2, while
recent study by Li and Senetakis, 2017a). This configura- those for the PH carbonate sand are listed in Table 3. All
tion allows both pairs of lateral bender elements to send the specimens from both materials were prepared in a dry
H. He et al. / Soils and Foundations 59 (2019) 571–585 575

state into split molds of appropriate dimensions using the


dry compaction method. In order to achieve variable initial

0.14
0.14
0.06
0.16
0.14
0.06
0.13
0.07
0.15
0.07
0.06
0.06

0.10
0.21
Poisson’s ratio

densities, the specimens were compacted with a light weight


bm



tamper applying variable compaction energies. After the
constants

preparation of the specimen, prior to the setting of the


0.53
0.51
0.30
0.42
0.48
0.30
0.44
0.31
0.56
0.37
0.34
0.34

0.95
apparatus and application of the first isotropic confining

0.4
am



pressure, vacuum of about 10 kPa was applied in order
to support the specimen and measure its actual dimensions.
0.36
0.37
0.29
0.29
0.29
0.29
0.31
0.30
0.32
0.34
0.36
0.36

0.32
buE

The initial void ratios, eo, of the WA carbonate specimens



Emax unloading

ranged from 1.17 to 1.47 (Table 2). The PH carbonate spec-


imens had lower in magnitude eo values, from 0.89 to 1.07
auE (MPa)
constants

(Table 3). Except for specimens WA-13 and WA-14, which


103.2

103.0

were tested in a fully saturated state, all the other speci-


67.4
65.0
95.0

96.0
89.6
92.8

80.2
64.5
66.7
69.7

82.8
mens were tested in a dry state. Note that the void ratio


values for both carbonate soils are relatively high. This is
0.42
0.41
0.34
0.37
0.40
0.34
0.39
0.35
0.42
0.41
0.41
0.41

0.39
0.44
because the sands have intra-particle voids, and the void
bE


ratio calculation considers the total volume of voids as


Emax loading

the summation of inter- and intra-particle voids.


aE (MPa)
constants

All the specimens were subjected to an isotropic stress


41.8
46.8
69.3
59.9
46.1
67.7
50.0
70.0
69.6
40.6
40.2
42.9

53.7
19.5

path (compression) and the dynamic tests were conducted



at progressively increasing mean effective confining stres-


ses, p0 , equal to 25 (if applicable), 50, 100, 200, 400, 600
0.37
0.37
0.28
0.28
0.27
0.30
0.28
0.29
0.29
0.33
0.34
0.31
0.42
0.37
0.32
buG

and 800 kPa (if applicable). For the four specimens tested

Gmax unloading

in the Hardin resonant column (i.e. specimens WA-11 to


WA-14 in Table 2), a greater maximum effective confining
auG (MPa)
constants

pressure, equal to 1.6 MPa was applied. For most of the


specimens, measurements of the dynamic properties were
26.7
26.6
41.3
45.9
44.6
37.6
40.2
43.3
47.0
27.7
31.0
40.0
20.7
28.3
35.8

also performed following an unloading isotropic path


(swelling) after the completion of the test at the highest


level of p0 .
0.45
0.44
0.35
0.40
0.43
0.45
0.35
0.41
0.37
0.43
0.42
0.47
0.41
0.41
0.41
0.49
bG

Additional measurements of small-strain shear modulus


Gmax loading

in torsional resonant column testing were performed for


aG (MPa)
constants

the WA carbonate specimens in order to validate the signal


analysis interpretation from the bender element tests. For
14.3
16.2
26.8
21.6
16.2
13.8
26.2
18.0
26.9
14.8
16.1
12.0
18.0
19.6
18.6
Testing program and specimen details for the Western Australia carbonate sand.

5.5

the dry specimens, the volume changes were estimated


based on the measured axial strain, ea, from a vertically
Average constant values

positioned displacement transducer and the assumption


Literature predicted
Initial inter-particle

of isotropic compression (i.e. the volumetric strain ev


void ratio (esg)

equals to three times the axial strain). The sensitivity of


stiffness measurements to possible inaccuracies to the esti-
mated volumetric strains (and void ratio) of the specimen
0.67
0.75
0.79
0.77
0.77
0.77
0.72
0.81
0.78
0.66
0.63
0.89
0.70
0.73

has been described briefly by He and Senetakis (2016). In


that study, the authors found insignificant effect of the
Initial void

inaccuracy in the estimation of the volumetric strain on


ratio (eo)

Gmax, as in their example, a 100% error in volumetric strain


1.22
1.31
1.36
1.33
1.33
1.33
1.27
1.38
1.34
1.21
1.17
1.47
1.25
1.29

resulted in no larger than 5% of over- or under-estimation


of Gmax values. For saturated specimens, the volume
Saturation

Saturated
Saturated

changes were measured in a straightforward way from a


status

back volume controller.


Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry
Dry

3. Results and discussion


Specimen Code

3.1. Piezo-elements signal analysis


WA-01
WA-02
WA-03
WA-04
WA-05
WA-06
WA-07
WA-08
WA-09
WA-10
WA-11
WA-12
WA-13
WA-14

The estimation of the wave velocities from the bender/


Table 2

extender element tests was performed adopting the first


No.

10
11
12
13
14

time of arrival method applying a sinusoidal type input


1
2
3
4
5
6
7
8
9
576 H. He et al. / Soils and Foundations 59 (2019) 571–585

Table 3
Testing program and specimen details for the Philippines carbonate sand.
Gmax loading constants Gmax unloading constants
No. Specimen Saturation Initial void Initial inter-particle aG (MPa) bG auG (MPa) buG
Code status ratio (eo) void ratio (esg)
1 PH-01 Dry 0.89 0.56 10.6 0.45 33.4 0.27
2 PH-02 Dry 0.93 0.60 8.0 0.49 34.1 0.27
3 PH-03 Dry 0.96 0.66 9.1 0.47 23.2 0.32
4 PH-04 Dry 1.07 0.82 8.1 0.51 39.3 0.27
5 PH-05 Dry 0.92 0.59 9.4 0.43 – –
6 PH-06 Dry 0.95 0.64 7.7 0.50 38.2 0.27
7 PH-07 Dry 0.96 0.65 9.8 0.46 32.2 0.29
8 PH-08 Dry 0.95 0.64 8.7 0.49 38.3 0.26
9 PH-09 Dry 1.00 0.71 8.3 0.49 – –
10 PH-10 Dry 0.92 0.59 7.1 0.52 – –
Average constant values 8.7 0.48 34.1 0.28
Literature predicted 3.7 0.53 – –

excitation of 10 kHz for both S-waves and P-waves. During Output signal for VS
the experiments, bender/extender element tests with 15 kHz p'=200kPa
source waves were also performed to verify and eliminate p'=400kPa
possible mistakes during the interpretation, especially for
the bender element tests at high confining pressure, and Peak to peak

the difference in wave velocity measurements between the


Amplitude

two frequencies was found to be small (less than 5%). Typ-


ical plots of the input and output waves from bender and
0 0.0001 0.0002 0.0003 0.0004 0.0005 0.0006
extender element tests for the specimen with code name
WA-06 are given in Fig. 4(a) and (b). Additional measure-
First Time Arrival
ments of Vs and Vp were performed adopting the peak-to-
peak method for the time arrival estimation, however, it is
worth mentioning that the results from the first-time of
arrival method were adopted for further analysis through- (a)
out the study. A comparison between the two different
Time (s)
interpretation methods of shear and compression wave
velocities is given in Fig. 5, where the data are clustered
around the 45° line indicating an excellent agreement Output signal for Vp
between the two methods (5% error). Even though there p'=200kPa

is a discussion in the literature on the signal analysis inter- p'=400kPa

pretation from bender element tests using additional meth-


Peak to peak
ods (for example in the studies by Lee and Santamarina,
Amplitude

2005; Gu et al., 2015; Ogino et al., 2015; Kawaguchi


et al., 2016 among others), the authors have been working
mostly in the time domain particularly using the first time 0 0.0001 0.0002 0.0003 0.0004 0.0005 0.0006
of arrival and peak-to-peak methods (e.g. He and
Senetakis, 2016; Li et al., 2018; Li and Senetakis, 2017a, First Time Arrival
2017b). Typically, the signal analysis from previous
research works by the authors has been validated based
on comparisons with resonant column tests. Fig. 6 provides
(b)
a comparison of the shear moduli (Gmax) obtained from the
bender elements exercise against the torsional resonant col- Time (s)
umn tests for the WA sand specimens tested at UNSW,
Fig. 4. Example of the signal analysis-interpretation for: (a) the bender
Australia. The differences between the Gmax measured from
element tests for specimen WA-06; (b) the extender element tests for
the two methods were mostly below ±10%, with a small specimen WA-06.
scatter when the stiffness was in the lower range. These
results indicated a satisfactory agreement between piezo- nal analysis. Note that the Gmax values in Fig. 6 corre-
element tests and resonant column tests and validated the sponded to the values measured during both the isotropic
interpretation method adopted for the bender element sig- loading (compression) and unloading (swelling) stages.
H. He et al. / Soils and Foundations 59 (2019) 571–585 577

800 400
Vs Vp 350
700
300

Gmax /f(e) (MPa)


600
Vs or Vp (m/s) PP

250 WA-01
WA-03
500 200
WA-07
±5% range 150
400 WA-11
100 PH-01
300 50 PH-02
PH-03
200 0
0 500 1000 1500 2000
100 p' (kPa)
100 200 300 400 500 600 700 800
Fig. 7. Typical plots of small-strain shear modulus measured from bender
Vs or Vp (m/s) FTA element tests against the confining pressure during the isotropic loading
for both WA and PH carbonate sand specimens (Note that modulus is
Fig. 5. Comparisons of S-wave and P-wave velocities between the first-
normalized with respect to a void ratio function).
time of arrival (FTA) and peak-to-peak (PP) methods of signal analysis
(note: Vs results include both the Philippines and the Western Australia
carbonate sand tests while Vp results are only from the Western Australia
carbonate sand). the coefficient of uniformity, Cu, and the particle shape
(expressed here with the regularity q) with respect to unce-
mented sands (e.g. Santamarina et al., 2001; Menq, 2003;
Wichtmann and Triantafyllidis, 2009, 2010; Payan et al.,
Gmax (MPa): Bender element tests

300
2016, 2017, among others). The term f(e) corresponds to
a void ratio function used for normalization purposes. In
this study, the widely adopted void ratio function f(e)
= e1.3 proposed by Jamiolkowski et al. (1991) was used
200
in order to eliminate the effect of initial specimen density
and determine the model parameters aG and bG. In Eq.
(5), the pressure p0 is normalized with respect to a reference
pressure of 1 kPa to unify the units. Note that the carbon-
ate sands have intra-particle voids; in Eq. (5) and the sub-
100
sequent analysis, the void ratio, e, corresponded to a global
void ratio accounting for both inter- and intra-particle
voids. It means that the development of stiffness expres-
sions is based on the combined inter- and intra-particle
0
0 100 200 300 voids, which would be the typical approach to be adopted
Gmax (MPa): Resonant column tests to develop design parameters for carbonate sands. How-
ever, as it will be discussed later in the paper, there can
Fig. 6. Comparison between shear stiffness (Gmax) of the WA carbonate be an alternative approach in the development of stiffness
sand measured from the piezo-element inserts and the torsional resonant expressions, adopting an equivalent void ratio which con-
column tests.
siders only inter-particle voids.
Based on the analysis of the results from the carbonate
3.2. Small strain shear modulus Gmax against effective sand specimens, the constants aG and bG were obtained.
isotropic confining pressure (p0 ) relationship for both soils The values of the elastic constants for the individual Wes-
tern Australia carbonate sand specimens as well as their
Typical plots of Gmax/f(e)  p0 from a representative set average values for all the experiments are summarized in
of specimens for both soils during the isotropic loading Table 2 and those for the Philippines carbonate sand spec-
stages are given in Fig. 7. The stiffness development trends imens are given in Table 3. It is noticed from Fig. 7 that
followed the general power law expression in Eq. (5) for after normalization, at a given pressure, the PH carbonate
small strain shear modulus: sand specimens yielded lower Gmax/f(e) values. The aG val-
 0 bG ues for the WA carbonate sand ranged between 12.0 and
p 26.9 (MPa) with an average value of 18.6 (MPa), while
Gmax ¼ aG  f ðeÞ  ð5Þ
1kPa those for the PH carbonate sand were observed to be lower,
between 7.1 and 10.6 (MPa) and the average value was
where aG (MPa) and bG are elastic constants (model equal to 8.7 (MPa). On the other hand, the Philippines car-
parameters) which depend, primarily, on material type, bonate sand appeared to be more sensitive to the increase
578 H. He et al. / Soils and Foundations 59 (2019) 571–585

of the pressure since the average power law exponent bG 1000


for the PH carbonate sand was equal to 0.48 (Table 3) 900 (a)
while that for the WA carbonate sand was equal to 0.41 800
(Table 2). Therefore, adopting the computed average val-

Emax/f(e) (MPa)
700
ues of model parameters aG (given in MPa) and bG, Eqs. 600
(6a) and (6b) can be used for the estimation of the small- 500
strain shear modulus of the WA and the PH sands, respec- 400 WA-01
tively (considering isotropic compression): 300 WA-03
 0 0:41
p 200 WA-07
Gmax ¼ 18:6  f ðeÞ  ð6aÞ 100
1kPa WA-11
 0 0:48 0
p 0 500 1000 1500 2000
Gmax ¼ 8:7  f ðeÞ  ð6bÞ p' (kPa)
1kPa
0.40
where Gmax is given in MPa. (b)
The differences in Gmax constants between the WA and
0.35
PH sands may be attributed, predominantly, to the differ-
ent particle shapes, the slight differences of the coefficients

Poisson's ratio ν
0.30
of uniformity as well as possible differences of the overall
grain morphologies between the two sands. With respect
0.25
to quartz sands, it has been reported in the literature that
the increase of Cu and the decrease of the regularity, q,
would result in a drop in aG value and an increase in bG 0.20
value (Menq, 2003; Senetakis et al., 2012; Payan et al.,
0.15 WA-01 WA-02 WA-03
2016), which explains the observed differences in the model
parameters between the two sands. WA-04 WA-08 WA-10
0.10
0 200 400 600 800 1000
3.3. Small strain Young’s modulus Emax and Poisson’s ratio
(m) against effective isotropic confining stress (p0 ) p' (kPa)
relationships for WA carbonate sand Fig. 8. Typical plots of: (a) small strain Young’s modulus from bender/
extender element tests against the confining pressure during the isotropic
Similar to the small strain shear modulus, the Emax/f(e)  loading of the WA carbonate sand specimens; (b) the Poisson’s ratio from
p0 and m  p0 plots for representative specimens from the WA bender/extender element tests against the confining pressure during the
isotropic loading of the WA carbonate sand specimens (Note that
carbonate sand are given in Fig. 8(a) and (b), respectively.
modulus is normalized with respect to a void ratio function).
The variation of the stiffness with respect to pressure can
be depicted by the power law type equations as follows:
 0 bE constant bG (equal to 0.41). Based on the listed average val-
p
Emax ¼ aE  f ðeÞ  ð7Þ ues of the model parameters for Young’s modulus and
1kPa Poisson’s ratio in Table 2, Eqs. (7) and (8) are re-written as:
 0 bm  0 0:39
p p
m ¼ am  ð8Þ Emax ¼ 53:7  e 1:3
 ð9Þ
1kPa 1kPa
where like in Eq. (5), aE (expressed in MPa), bE, am and bm  0 0:10
p
are model parameters. The same void ratio function, f(e) m ¼ 0:40  ð10Þ
1kPa
= e1.3, was adopted in the normalization of the small-
strain Young’s modulus (i.e. considering inter- and intra- Note that in Eq. (9) Emax is expressed in MPa and that
particle voids). Since no specific trend of Poisson’s ratio the expressions above correspond to isotropic compression
against void ratio (e) was observed in the study, the Pois- (loading process).
son’s ratio is not normalized by the void ratio function. The Poisson’s ratio was found to decrease in magnitude
Through the analysis of the results with respect to the as the isotropic confining pressure increased (power law
WA sand, the model parameters for both small strain exponent bm values are negative), which is in agreement
Young’s modulus and Poisson’s ratio were quantified and with the previous experimental and numerical studies by
the resulted values are listed in Table 2. The aE values Yang et al. (2008), Kumar and Madhusudhan (2010), Gu
are more scattered compared to the aG values with a min- and Yang (2013), He and Senetakis (2016) and Payan
imum value of 40.2 and a maximum value of 70.0 et al. (2017). Through this set of tests, the Poisson’s ratio
(expressed in MPa). The average power law exponent bE dropped from ranging between about 0.25 to 0.35 at (p0 )
value (equal to 0.39) is close to the average Gmax power equal to 25 kPa, to about 0.18 to 0.23 at (p0 ) equal to
H. He et al. / Soils and Foundations 59 (2019) 571–585 579

800 kPa. No systematic effect of the void ratio on the Pois- value to increase from 8.7 to 34.1 (MPa). On the other
son’s ratio was observed through this set of tests, which hand, during the unloading stages, the sensitivity of stiff-
coincides with the observations by He and Senetakis ness (for both Gmax and Emax) to pressure was less pro-
(2016) and Payan et al. (2017). nounced in comparison to that during the loading stages,
which was mirrored from the lower bG and bE values dur-
3.4. Behavior during isotropic unloading ing the swelling process. The drop of the average bG and bE
values for the WA sand and the average bG values for the
During the isotropic unloading (swelling) process, both PH sand, between the compression and swelling stages, are
the WA and PH carbonate sands showed to have greater 0.09, 0.07 and 0.20, respectively. A summary of the unload-
stiffness in comparison to the loading process. In particu- ing stiffness constants (denoted as auG, buG, auE and buE) of
lar, based on power-law type fitting, the constants aG and the individual specimens, are given in Table 2 for the WA
aE were found to be greater, whereas the constants bG specimens and in Table 3 for the PH specimens. These
and bE were found to be lower during the unloading pro- observations are in contrast with the general trend
cess (swelling) in comparison to the compression stage. observed in quartz sands, where for a typical range of pres-
Typical plots for specimens WA-01 and PH-02 during the sures, for example from 25 to 800 kPa, quartz sands would
isotropic loading and unloading paths are given in Fig. 9 show the same sensitivity to pressure during the loading or
(Gmax/f(e)-p0 relationships in Fig. 9(a) and Emax/f(e)-p0 unloading processes resulting in almost unchanged values
relationships in Fig. 9(b)). Compared to the loading stages, of the power law exponent. However, these observations
the average aG and aE values for the WA carbonate sand of the study coincide with previous research works on other
increased from 18.6 to 35.8 MPa (for aG) and from 53.7 types of carbonate sands studied by Fioravante et al. (1994,
to 82.8 MPa (for aE) during unloading. For the PH carbon- 1998, 2013), Jovicic and Coop (1997) and Yamashita et al.
ate sand, the average aG had an increase of about 290% (2000), as well as the recent study by He and Senetakis
from the loading to the unloading stage, with the average (2016) on an engineered crushable sand. From the current
study, as well as the previous study by Fioravante et al.
350 (2013) where the authors included the over-consolidation
(a) ratio factor into the model, the increase in stiffness during
300 the unloading process could not be solely normalized by
250 the void ratio function.
Gmax /f(e) (MPa)

200 WA-01 3.5. Particle breakage analysis and its effect on dynamic
properties
150 WA-01
unloading
100 PH-02 The change of the stiffness from the loading to the
unloading process is hypothesized to be mainly attributed
50 PH-02
to the elastoplastic to plastic in nature contact response
unloading
0 between the particles (Cascante and Santamarina, 1996),
0 200 400 600 800 1000 and, perhaps, due to particle breakage during the compres-
p' (kPa) sion of the specimens (Jovicic and Coop, 1997). Therefore,
the grading of the materials was measured after the speci-
800 mens were subjected to a maximum confining pressure of
(b)
700 800 kPa during compression and thereafter the specimens
600
were unloaded. The comparisons between the grading
Emax /f(e ) (MPa)

curves before and after the tests are given in Fig. 10(a)
500
for a WA carbonate sand specimen and in Fig. 10(b) for
400 a PH carbonate sand specimen. Through the relative
300 breakage analysis of the grading curves adopting Hardin
WA-01 (1985) method, it was found that the WA carbonate sand
200
(Fig. 10(a)) had a relative breakage of 3.7% while the PH
100 WA-01 carbonate sand (Fig. 10(b)) yielded almost identical grad-
unloading
0 ing curves before and after the compression indicating neg-
0 200 400 600 800 1000 ligible breakage. These breakage values are representative
p' (kPa) for the whole study.
Since there was observable breakage particularly for the
Fig. 9. (a) Small strain shear modulus of specimens WA-01 and PH-02
WA sand, though small in magnitude, an analysis of the
against the confining pressure during isotropic loading and unloading; (b)
small strain Young’s modulus of specimen WA-01 against the confining power-law exponent bG was carried out to examine
pressure during isotropic loading and unloading (Note that modulus is whether there is any dependency of model parameter bG
normalized with respect to a void ratio function). on the grain breakage during the isotropic loading process,
580 H. He et al. / Soils and Foundations 59 (2019) 571–585

Original WA CS 800kPa tested WA CS 0.50


(a)
100

bG for WA carbonate sand


(a) 0.40
90
Percent finer by weight

80
70 0.30
60
50 0.20
40
30 0.10
WA-01 WA-08 WA-11 WA 12
20
10
0.00
0 200 400 600 800 1000
0
0. 05 0. 5 Maximum pressure (p') reached (kPa)
Grain size (mm)

0.6
Original PH CS 800kPa tested PH CS (b)

bG for PH carbonate sand


100 0.5
(b)
90
80 0.4
Percent finer by weight

70
0.3
60
50 0.2
40
0.1
30
PH-02 PH-03 PH-06 PH-04
20 0
10 0 200 400 600 800
0 Maximum pressure (p') reached (kPa)
0. 07 0. 7
Grain size (mm) Fig. 11. The power for shear modulus (bG) at different maximum isotropic
confining pressures for representative specimens: (a) the WA carbonate
Fig. 10. Grading curves before and after the specimens being isotropically
sand specimens; (b) the PH carbonate sand specimens.
compressed to 800 kPa for: (a) the Western Australia carbonate sand; (b)
the Philippines carbonate sand.
be a dominant factor in the range of pressures in the cur-
rent study, even though it caused a slight increase of the
where the breakage could happen progressively as the pres- power-law exponent bG for the WA sand of the order of
sure increases. If there is a massive grain crushing, it is 8% as an averaged value throughout the four specimens
expected that the crushing would have an impact on the presented in Fig. 11(a).
bG values. For example, Cascante and Santamarina This analysis is in agreement with the previous work on
(1996) found that for brittle materials, the power-law expo- Kenya carbonate sand by Bellotti et al. (1991) and
nent bG at higher stress levels was lower, which was attrib- Fioravante et al. (2013), where they neglected the effect
uted, mainly to the breakage of the grains. In this study, for of breakage on the dynamic test results since they did not
a set of four specimens from each material, the power law find more than 2.5% of relative breakage until the vertically
fitting regression analysis was conducted at different maxi- applied stress yielded 11 MPa. Thus, the higher values of
mum pressures and the bG values are plotted against the stiffness exhibited during the unloading compared to the
maximum pressure in Fig. 11(a) and (b) for the WA and loading process could possibly be due to the breakage of
PH carbonate sands, respectively. Note that there was no the micro asperities on the grain surfaces and a higher level
breakage found through sieving analysis for both materials of interlocking caused by the compression of the specimen
after the soil specimens were confined isotropically to (Fioravante et al., 2013).
200 kPa, thus comparing the power-law exponent bG val-
ues regressed from 25 to 200 kPa and to higher pressures 3.6. Lateral bender element test results and study of fabric
would give an indication of whether there is an effect of anisotropy of carbonate sand
grain breakage on the modulus – pressure relationship.
For both materials, there is observed a slight increase of For the Philippines carbonate sand, four specimens were
the power-law exponent bG as the maximum pressure tested with both vertical and lateral (horizontal) bender ele-
increases. Thus, the minor particle crushing should not ments (PH-07, PH-08, PH-09 and PH-10). Note that the
H. He et al. / Soils and Foundations 59 (2019) 571–585 581

frequency used for the lateral bender element tests was 3.7. Comparison with published models: the use of inter-
20 kHz instead of 10 kHz to ensure that the distance particle void ratio
between the bender element tips over the wavelength is at
least 3.33 (Leong et al., 2005, 2009). The interpretation of There are extensive studies in the literature aiming at
the lateral bender element test results is identical to the ver- predicting the small strain stiffness of granular materials,
tical ones given in Fig. 4(a). Three shear wave velocity com- especially for quartz sands. It has been reported in the lit-
ponents are recorded at each stress level, denoted as Vs,vh, erature that apart from the confining pressure and void
Vs,hv and Vs,hh, from which, the corresponding stiffness is ratio (or density), the dominant factors that affect the elas-
derived from Eq. (1) in a straightforward way, denoted tic stiffness of granular soils are material type, coefficient of
as Gvh, Ghv and Ghh, respectively. Isotropic stress paths uniformity and particle shape. In the recent works by
were followed, thus no stress anisotropy was induced to Payan et al. (2016, 2017), the shear moduli and Young’s
the specimen. The ratios between Ghv and Gvh and between moduli of a wide range of quartz sands with a variety of
Ghh and Ghv are both plotted against the confining pres- particle shapes and coefficients of uniformity were exam-
sure in Fig. 12 to examine the intrinsic anisotropy of the ined and the following models for the prediction of elastic
reconstituted specimen. The ratio Ghh/Ghv was scattered modulus of quartz sands were proposed:
around unity (ranging from 0.90 to 1.05), thus no system- 
atic anisotropy could be identified from the current set of Gmax ¼ 84C u0:14  q0:68  e1:29
tests. In contrast to the cross-anisotropy observed in the lit-  0 ðC0:12
u Þð0:23qþ0:59Þ
p
erature (Bellotti et al., 1996; Fioravante, 2000; Kuwano  ð11Þ
pa
et al., 2000; Gu et al., 2017), the PH carbonate sand sam-
ples were found to be fairly isotropic from the experiments 
Emax ¼ 245C u0:09  q0:82  e1:29
performed in the study. However, the Gvh appears to be  0 ðC0:11
u Þð0:44qþ0:66Þ
slightly higher than the other two components for most p
 ð12Þ
tests. The reason of the small difference between the verti- pa
cal and horizontal bender element test results is believed to
be affected, partly, by the different boundary conditions. where Cu is the coefficient of uniformity, q is the regularity
In their recent study examining tailings of silt to sand of the soil particles, e is the void ratio, pa is the atmospheric
size, Li and Senetakis (2017a) reported a fairly isotropic pressure (taken as 100 kPa), while Gmax and Emax are
behavior in terms of similar values of the different compo- expressed in MPa. These empirical equations are adopted
nents of stiffness. Based on reported trends in the literature to predict the behavior of the carbonate sands and compare
(Santamarina and Cho, 2004; Yang et al., 2008), Li and the estimated with the measured values. The reason to use
Senetakis (2017a) highlighted that even though the soils these particular expressions in the present study, among
they tested were very angular, the observed isotropic other formulae proposed in the literature, for comparison
behavior was perhaps due to the moderate grain sphericity, purposes with the results on the WA and PH sands, is that
i.e. the grains were not too elongated in shape to give rise these expressions incorporate the important effects of both
to a notable intrinsic anisotropy. Similarly, the PH carbon- grading and particle shape characteristics of sands. Note
ate sand of the study was very angular but the grains were that in Eqs. (11) and (12), the particle shape influence has
not too elongated in shape. been incorporated through the regularity, q, which is the
arithmetic mean of sphericity and roundness.
The predicted stiffness constants are given in Tables 2
and 3 for both carbonate sands. For representative speci-
1.5 mens, based on the Cu and q values of the carbonate sands,
1.4 the predicted against the measured shear moduli of the WA
1.3 carbonate sand and the PH carbonate sand and the
Ghv/Gvh or Ghh/Ghv

1.2 Young’s moduli of the WA carbonate sand are plotted


1.1 against the confining pressure in Fig. 13(a), (b) and (c),
1 respectively. These figures and the constants in the tables
0.9 reveal the significant underestimation of the stiffness of
0.8 the carbonate sands from the empirical equations. When
0.7 plotting the predicted values against the measured values
0.6 for both materials (Gmax and Emax), the overall under-
0.5
prediction falls in a range of about 30% to 60% (as shown
0 100 200 300 400 500 600 700
in Fig. 14).
p' (kPa)
As discussed previously, the void ratio is a key parame-
Ghv/Gvh Ghh/Ghv
ter that influences the stiffness of the soil. Therefore, con-
Fig. 12. Ratios of Ghv/Gvh and Ghh/Gvh against p0 for the Philippines sidering the porosity of the material, it is hypothesized
carbonate sand (from specimens: PH-07, PH-08, PH-09 and PH-10). that one of the main reasons for the under-prediction of
582 H. He et al. / Soils and Foundations 59 (2019) 571–585

350 WA CS Gmax PH CS Gmax WA CS Emax


(a)
300 600
WA-01 Measured
250 WA-03 Measured
500 ±60%
Gmax (MPa)

200 WA-07 Measured

Gmax or Emax predicted


WA-11 Measured
150 400 -30%
WA-01 Predicted
100 WA-03 Predicted
50-63% WA-07 Predicted 300
50
Underpredicted WA-11 Predicted
0
200
0 1000 2000 3000
p' (kPa)
100
250
(b)
0
200
0 100 200 300 400 500 600
Gmax or Emax measured
Gmax (MPa)

PH-01 Measured
150
PH-02 Measured
Fig. 14. Predicted versus measured stiffness for both materials (including
100 PH-03 Measured Gvh and Emax).
PH-01 Predicted
40-52%
50 PH-02 Predicted
Underpredicted role of the intra-particle voids on the engineering behavior
PH-03 Predicted of those soils (Coop, 1990). However, when proposing Eqs.
0
(11) and (12), the authors tested quartz sands that consist
0 500 1000
of solid grains, thus only inter-particle voids were consid-
p' (kPa)
ered. The high void ratio values would lead to a lower stiff-
900 ness prediction. This arouses the interest to study whether
800
(C) the empirical equations could predict the carbonate sands
WA-01 Measured
behavior if only inter-particle voids are considered in the
700
calculation of the void ratio.
600 WA-03 Measured
The maximum and minimum void ratio of a granular
Emax (MPa)

WA-07 Measured
500 soil with solid particles (i.e. without intra-particle voids)
WA-11 Measured
400 has been related, predominantly, with the coefficient of uni-
WA-01 Predicted
300 formity (Winterkorn and Fang, 1975; Kokusho et al., 1995;
WA-03 Predicted Menq, 2003) as well as the shape of particles (Santamarina
200
40-50% WA-07 Predicted et al., 2001). From a large number of literature and exper-
100
Underpredicted WA-11 Predicted imental results, Menq (2003) derived the following equa-
0
tions for the estimation of the maximum and minimum
0 1000 2000 3000
void ratios of reconstituted granular materials for different
p' (kPa)
coefficients of uniformity as follows:
Fig. 13. Predicted and measured stiffness against pressure for represen-
emax ¼ 0:95  ð1=C u Þ þ 0:43; 200 P C u P 1:0 ð13Þ
tative specimens: (a) predicted and measured shear moduli against
pressure for the WA carbonate sand specimens (empirical equation given emin ¼ 0:60  ð1=C u Þ þ 0:22; 200 P C u P 1:0 ð14Þ
in Eq. (11), proposed by Payan et al. (2016)); (b) predicted and measured
shear moduli against pressure for the PH carbonate sand specimens; (c) According to these equations, a sand of solid grains (i.e.
predicted and measured Young’s moduli against pressure for the WA
carbonate sand specimens (empirical equation given in Eq. (12), proposed
no intra-particle voids) with a coefficient of uniformity of
by Payan et al. (2017)). the WA carbonate sand (=1.70) would have maximum
and minimum void ratios of around emax, sgWA = 0.99
and emin,sgWA = 0.57, respectively. Similar, a sand of solid
the stiffness from the models in Eqs. (11) and (12) is the grains with a coefficient of uniformity of the PH carbonate
high void ratio values of the carbonate sand. From the sand (=1.85) would have maximum and minimum void
SEM images in Fig. 2, there was observed a significant por- ratios of around emax,sgPH = 0.94 and emin,sgPH = 0.54,
tion of intra-particle voids for the grains of both carbonate respectively. These upper and lower bounds could corre-
sands, which resulted in high void ratios of the specimens. spond to the carbonate sands if intra-particle voids were
This is true for any carbonate soil as it has been noticed not considered, i.e. these bounds give a rough estimation
since early 1990s in the literature the possible important of the inter-particle voids assuming solid grains. On the
H. He et al. / Soils and Foundations 59 (2019) 571–585 583

other hand, the measured maximum and minimum void 300


ratio (which includes inter- and intra-particle voids) for

Gmax predicted (with inter-particle e)


the Western Australian carbonate sand are emax, 250
WA = 1.60 and emin,WA = 1.10, while those for The Philip- ±20%
pines carbonate sand are emax,PH = 1.19 and emin, 200
PH = 0.86. The much higher range of void ratios in com-
parison to the predicted from Eqs. (13) and (14) were, pre-

(MPa)
150
dominantly, a result of the intra-particle voids. It is worth
mentioning that this estimation of limiting void ratio val-
100
ues, based on Eqs. (13) and (14), gives only some general
view of possible values. Cho et al. (2006) have shown that
particle shape, apart from the coefficient of uniformity, 50

also affects the packing of a granular assembly. However, PH CS WA CS


the effect of particle shape was not included in Eqs. (13) 0
and (14). Therefore, the use of Eqs. (13) and (14) in this 0 50 100 150 200 250 300

study is only approximate, for a rough estimation of limit- Gmax measured (MPa)
ing emax and emin values. To be more precise, in future Fig. 15. Predicted shear moduli (based on assumed inter-particle void
works it is recommended that more rigorous estimation is ratios) versus measured shear moduli for all the specimens from both the
conducted to de-couple inter- and intra-particle voids Western Australia and the Philippines carbonate sands.
based, for example, on micro-CT scanning or other ade-
quate techniques. 700
Based on this, the relative density concept is adopted to WA CS Ema x predicted (with inter-particle e)
assume the inter-particle void ratio of the carbonate sands 600
(i.e. ignoring the intra-particle voids). Relative density is ±20%
defined as: 500
emax  e
Dr ¼ ð15Þ 400
emax  emin
(MPa)

where emax is the maximum void ratio, and emin is the min- 300
imum void ratio. With the calculated relative density of a
carbonate sand specimen, its corresponding inter-particle 200
void ratio could be assumed (as a rough estimation) by
the following equation: 100

esg ¼ emax;sg  Dr  emax;sg  emin;sg ð16Þ 0
0 100 200 300 400 500 600 700
where esg is the solid grain void ratio, which is an assumed
WA CS Ema x measured (MPa)
inter-particle void ratio (i.e. corresponds to a rough estima-
tion based on Eqs. (13) and (14)) for the carbonate sand. Fig. 16. Predicted Young’s moduli (based on assumed inter-particle void
emax,sg and emin,sg are the maximum and minimum void ratios) versus measured Young’s moduli for all the specimens from the
ratios of a solid grained sand with the uniformity coeffi- Western Australia carbonate sand.
cient of the carbonate sand.
Adopting the assumed inter-particle void ratio, the shear assumptions and consideration that the grains are solid. It
moduli and Young’s moduli of the carbonate sands are re- is believed however, as mentioned before, that more precise
predicted using the empirical equation proposed for quartz de-coupling of inter- and intra-particle voids could be per-
sand (Eqs. (11) and (12)). The comparison between the re- formed in future works based on micro-CT scanning anal-
predicted and measured shear moduli for both carbonate ysis and image processing and/or the combined use of
sands is given in Fig. 15 and the re-predicted Young’s mod- micro-CT scanning with other adequate techniques. The
uli are compared with measured Young’s moduli in Fig. 16. results of the study presented in this Section 3.7, can give
Despite the small scatter (within a ±20% error range), from an indication that the concept of inter-particle voids could
the comparisons in Figs. 15 and 16, the predicted stiffness work effectively for carbonate sands (as well as other soils)
values based on the assumed inter-particle voids are satis- which have intra-particle voids. It is needed to be men-
factorily close to the measured values without systematic tioned however, that the application of the general expres-
under- or over-prediction. In practice, the results of Figs. 15 sions of Eq. (5) and the development of model parameters
and 16 reveal that empirical expressions developed on the for stiffness prediction considering global void ratio, is the
basis of quartz sands of solid grains work satisfactorily also mainstream method to produce design parameters to be
for carbonate sands which have intra-particle voids, if only used in engineering practice. Depending on the type of soil,
inter-particle voids are assumed based on some reasonable possible grain breakage could add some complication in
584 H. He et al. / Soils and Foundations 59 (2019) 571–585

the use of inter-particle void ratio, excluding the intra- work well to predict the stiffness of a carbonate soil when
particle voids from the void ratio function, which is a topic the intra-particle voids are ignored. It was however stressed
that is worth further investigation. in our study that the basic method to compute void ratio
and derive design values of soil stiffness is based on the glo-
4. Conclusions bal void ratio which accounts for both inter- and intra-
particle voids. The use of inter-particle voids into a stiffness
The paper studied and compared two poorly graded car- expression, de-coupling inter- and intra-particle voids for
bonate sands from two distinct origins, namely Western high porosity materials, would worth further investigation
Australia (WA) and the Philippines (PH) with a focus on using adequate image analysis.
the small strain stiffness. Basic characterization of the
materials was performed and the results showed that the Acknowledgements
two sands are composed of irregularly shaped grains based
on the quantification of the particle shape descriptors with The authors would like to thank the anonymous review-
the help of SEM analysis. Piezo-element inserts were used ers for their kind comments and detailed suggestions that
to carry out the dynamic tests during the progressive iso- helped us to improve the quality of our manuscript. The
tropic confinement of the specimens. The interpretation grant No. 7200533 (ACE) from the City University of
method for bender element test results was validated by Hong Kong is acknowledged for supporting partly the
making comparisons between the shear wave velocities research work described in this paper as well as the grant
measured from bender element and torsional resonant col- from the Research Grants Council of the Hong Kong Spe-
umn tests. For the WA carbonate sand, the small strain cial Administrative Region, China (Project No. CityU
shear modulus, the Young’s modulus and the Poisson’s 112813).
ratio were examined using bender/extender elements. For
the PH carbonate sand, the small strain shear modulus References
was studied using bender elements with a focus on the com-
parison between the vertical and the lateral bender element Airey, D., 1993. Triaxial testing of naturally cemented carbonate soil. J.
test results, thus possible stiffness anisotropy could be iden- Geotech. Eng. 119 (9), 1379–1398.
tified. The dynamic tests were also performed during the Atkinson, J., 1993. An Introduction to the Mechanics of Soils and
isotropic unloading (swelling) stage and both carbonate Foundations: Through Critical State Soil Mechanics. McGraw-Hill
Book Company (UK) Ltd..
sands were found to be noticeably affected by the over- Bellotti, R., Fretti, C., Ghionna, V., Pedroni, S., 1991. Compressibility
consolidation stress history which is aligned with previous and crushability of sands at high stresses. Proceedings of the
works published in the literature on different carbonate International Symposium on Calibration Chamber Testing, Potsdam,
sands. No massive particle breakage was noticed through New York, pp. 79–90.
the comparison between the grading curves before and Bellotti, R., Jamiolkowski, M., Lo Presti, D.C.F., O’Neill, D.A., 1996.
Geotechnique 46 (1), 115–131.
after the isotropic confinement up to 800 kPa. No signifi- Bishop, A.W., Wesley, L.D., 1975. A hydraulic triaxial apparatus for
cant fabric anisotropy was observed for the reconstituted controlled stress path testing. Geotechnique 25 (4), 657–670.
PH carbonate sand specimens since the shear moduli Ghh Cascante, G., Santamarina, J.C., 1996. Interparticle contact behavior and
and Ghv measured from the lateral bender elements were wave propagation. J. Geotech. Eng. 122 (10), 831–839.
almost identical. It was assumed that the fairly isotropic Cho, G.-C., Dodds, J., Santamarina, J.C., 2006. Particle shape effects on
packing density, stiffness, and strength: natural and crushed sands. J.
behavior was majorly related to the moderate sphericity Geotech. Geoenviron. Eng. 132 (5), 591–602.
of the grains, since both sands consisted of grains which Chung, R., Yokel, F., Drnevich, V., 1984. Evaluation of dynamic
were not too elongated and that the angularity of the grains properties of sands by resonant column testing. Geotech. Test. J. 7
might play a secondary role on the intrinsic isotropy/ (2), 60–69.
anisotropy. Clayton, C., 2011. Stiffness at small strain: research and practice.
Géotechnique 61 (1), 5–37.
Published empirical equations in the literature proposed Coop, M., 1990. The mechanics of uncemented carbonate sands.
from quartz sands gave unsuccessful prediction of the car- Géotechnique 40 (4), 607–626.
bonate sands stiffness. From a novel approach, the intra- Coop, M., Airey, D., 2003. Carbonate sands. In: Tan, E.a.E. (Ed.),
particle voids in the carbonate sands were eliminated in Proceedings of Characterization and Engineering Properties of Nat-
the void ratio calculation assuming the grains as solid, thus ural Soils, pp. 1049–1086.
Coop, M., Sorensen, K., Freitas, T.B., Georgoutsos, G., 2004. Particle
equivalent inter-particle void ratio was derived based on breakage during shearing of a carbonate sand. Géotechnique 54 (3),
some rough estimation from empirical expressions. When 157–164.
taking the equivalent inter-particle void ratio into the Coop, M.R., Atkinson, J.H., 1993. The mechanics of cemented carbonate
empirical literature equations, the stiffness prediction was sands. Géotechnique 43 (1), 53–67.
found to be satisfactory. This means that for sands with Fioravante, V., Cappferri, R., Hameury, O., Jamiolkowski, M., 1994.
Deformational characteristics of uncemented carbonate Quiou sand.
intra-particle voids, particularly for carbonate soils which Proceedings of the International Symposium on Pre-failure Deforma-
are very important geo-materials in critical infrastructure tion Characteristics of Geomaterials, Sapporo, Japan, pp. 55–61.
such as offshore structures and subsea facilities, empirical Fioravante, V., Jamiolkowski, M., Ghionna, V.N., Pedroni, S., 1998.
expressions developed on the basis of quartz sands may Stiffness of carbonate Quiou sand from CPT. In: Roberston, Mayne
H. He et al. / Soils and Foundations 59 (2019) 571–585 585

(Ed.), 2008, Proceeding of the First International Conference on Site Leong, E., Yeo, S., Rahardjo, H., 2005. Measuring shear wave velocity
Characterisation, pp. 1039–1049. using bender elements. Geotech. Test. J. 28 (5), 488–498.
Fioravante, V., 2000. Anisotropy of small strain stiffness of ticino and Leong, E.C., Cahyadi, J., Rahardjo, H., 2009. Measuring shear and
kenya sands from seismic wave propagation measured in triaxial compression wave velocities of soil using bender-extender elements.
testing. Soils Found. 40 (4), 129–142. Can. Geotech. J. 46 (7), 792–812.
Fioravante, V., Giretti, D., Jamiolkowski, M., 2013. Small strain stiffness Li, H., He, H., Senetakis, K., 2018. Calibration exercise of a hardin-type
of carbonate Kenya sand. Eng. Geol. 161, 65–80. resonant column. Géotechnique 68 (2), 171–176.
Gu, X., Yang, J., 2013. A discrete element analysis of elastic properties of Li, W., Senetakis, K., 2017a. Dynamic shear modulus of three reconsti-
granular materials. Granular Matter 15 (2), 139–147. tuted soils from panzhihua iron tailing dam. J. Geo-Eng. 12 (3), 129–
Gu, X., Yang, J., Huang, M., 2013. Laboratory measurements of small 135.
strain properties of dry sands by bender element. Soils Found. 53 (5), Li, H., Senetakis, K., 2017b. Dynamic properties of polypropylene fibre-
735–745. reinforced silica quarry sand. Soil Dyn. Earthquake Eng. 100, 224–232.
Gu, X., Yang, J., Huang, M., Gao, G., 2015. Bender element tests in dry Menq, F.-Y., 2003. Dynamic Properties of Sandy and Gravelly Soils.
and saturated sand: signal interpretation and result comparison. Soils University of Texas at Austin, USA.
Found. 55 (5), 951–962. Miao, G., Airey, D., 2013. Breakage and ultimate states for a carbonate
Gu, X., Hu, J., Huang, M., 2017. Anisotropy of elasticity and fabric of sand. Géotechnique 63 (14), 1221–1229.
granular soils. Granular Matter 19 (2). https://doi.org/10.1007/s10035- Ogino, T., Kawaguchi, T., Yamashita, S., Kawajiri, S., 2015. Measure-
017-0717-6. Article 33. ment deviations for shear wave velocity of bender element test using
Hardin, B., Richart, F., 1963. Elastic wave velocities in granural soils. J. time domain, cross-correlation, and frequency domain approaches.
Soil Mech. Found., ASCE 89 (SM1), 33–65. Soils Found. 55 (2), 329–342.
Hardin, B., 1978. The nature of stress strain behavior of soils. Proceed- Payan, M., Khoshghalb, A., Senetakis, K., Khalili, N., 2016. Effect of
ings, Geotechnical Division Speciality Conference on Earthquake particle shape and validity of gmax models for sand: a critical review
Engineering and Soil Dynamics. ASCE, Pasadena, C.A., pp. 3–90. and a new expression. Comput. Geotech. 72, 28–41.
Hardin, B.O., 1985. Crushing of soil particles. J. Geotech. Eng. 111 (10), Payan, M., Senetakis, K., Khoshghalb, A., Khalili, N., 2017. Effect of
1177–1192. gradation and particle shape on small-strain young’s ’s modulus and
He, H., Senetakis, K., 2016. A study of wave velocities and poisson ratio poisson’s ratio of sands. Int. J. Geomech. 17 (5).
of recycled concrete aggregate. Soils Found. 56 (4), 593–607. Richart, F.E., Hall, J.R., Woods, R.D., 1970. Vibrations of Soils and
He, H., Senetakis, K., Ranjith, P., 2017. The behavior of a carbonate sand Foundations. Prentice Hall, Englewood Cliffs, p. 414.
subjected to a wide strain range of medium-frequency flexural Santamarina, C., Klein, K., Fam, M., 2001. Soils and Waves. John Wiley
excitation. Geomech. Geophys. Geo-Energy Geo-Resour. 3 (1), 51–60. and Sons, New York.
Kuwano, R., Connolly, T.M., Jardine, R.J., 2000. Anisotropic stiffness Santamarina, J.C., Cho, G.C., 2004. Soil behavior: the role of particle
measurements in a stress-path triaxial cell. Geotech. Test. J. 23 (2), shape. Proc. Skempton Conf., March, London, UK..
141–157. Saxena, S.K., Reddy, K.R., 1989. Dynamic moduli and damping ratios for
Ishihara, K., 1996. Soil Behaviour in Earthquake Geotechnics. Clarendon monterey no. 0 sand by resonant column tests. Soils Found. 29 (2), 37–
Press; Oxford University Press. 51.
Iwasaki, T., Tatsuoka, F., 1977. Effects of grain size and grading on Senetakis, K., Anastasiadis, A., Pitilakis, K., 2012. The small-strain shear
dynamic shear moduli of sands. Soils Found. 17 (3), 19–35. modulus and damping ratio of quartz and volcanic sands. Geotech.
Iwasaki, T., Tatsuoka, F., Takagi, Y., 1978. Shear moduli of sands under Test. J. 35 (6), 964–980.
cyclic torsional shear loading. Soils Found. 18 (1), 39–56. Senetakis, K., He, H., 2017. Dynamic characterization of a biogenic sand
Jamiolkowski, M., Leroueil, S., Lo Presti, D., 1991. Design parameters with a resonant column of fixed-partly fixed boundary conditions. Soil
from theory to practice. Proceedings of the International Conference Dyn. Earthquake Eng. 95, 180–187.
on Geotechnical Engineering for Coastal Development. Geo-Coast Senetakis, K., Madhusudhan, B.N., 2015. Dynamics of potential fill–
1991. Coastal Development Institute of Technology, Yokohama, backfill material at very small strains. Soils Found. 55 (5), 1196–1210.
Japan, pp. 877–917. Shirley, D.J., Anderson, A.L., 1975. In situ measurement of marine
Jovicic, V., Coop, M.R., 1997. Stiffness of coarse-grained soils at small sediment acoustical properties during coring in deep water. IEEE
strains. Géotechnique 47 (3), 545–561. Trans. Geosci. Electron. 13 (4), 163–169.
Jovicic, V., Coop, M.R., Simic, M., 1996. Objective criteria for determin- Vucetic, M., 1994. Cyclic threshold shear strains in soils. J. Geotech. Eng.
ing gmax from bender element tests. Géotechnique 46 (2), 357–362. 120 (12), 2208–2228.
Kawaguchi, T., Ogino, T., Yamashita, S., Kawajiri, S., 2016. Identifica- Wichtmann, T., Triantafyllidis, T., 2009. Influence of the grain-size
tion method for travel time based on the time domain technique in distribution curve of quartz sand on the small strain shear modulus
bender element tests on sandy and clayey soils. Soils Found. 56 (5), Gmax. J. Geotech. Geoenviron. Eng. 135 (10), 1404–1418.
937–946. Wichtmann, T., Triantafyllidis, T., 2010. On the influence of the grain size
Knight, R., Klassen, R., Hunt, P., 2002. Mineralogy of fine-grained distribution curve on p-wave velocity, constrained elastic modulus m
sediment by energy-dispersive spectrometry (eds) image analysis – a max and Poisson’s ratio of quartz sands. Soil Dyn. Earthquake Eng.
methodology. Environ. Geol. 42 (1), 32–40. 30 (8), 757–766.
Kokusho, T., Yoshida, Y., Tanaka, Y., 1995. Shear wave velocity in Winterkorn, H.F., Fang, H.-Y., 1975. Foundation Engineering Hand-
gravelly soils with different particle gradings. Static Dynam. Propert. book. Van Nostrand Reinhold Reinhiold Company, New York, New
Gravelly Soils: ASCE 56, 92–106. York.
Kramer, S.L., 1996. Geotechnical Earthquake Engineering. Pearson Yamashita, S., Jamiolkowski, M., Lo Presti, D.C.F., 2000. Stiffness
Education India. nonlinearity of three sands. J. Geotech. Geoenviron. Eng. 126 (10),
Krumbein, W.C., Sloss, L.L., 1963. Stratigraphy and Sedimentation, 929–938.
second ed. W.H. Freeman and Company. Yang, Z.X., Li, X.S., Yang, J., 2008. Quantifying and modelling fabric
Kumar, J., Madhusudhan, B.N., 2010. Effect of relative density and anisotropy of granular soils. Géotechnique 58 (4), 237–248.
confining pressure on poisson ratio from bender and extender elements Youn, J.-U., Choo, Y.-W., Kim, D.-S., 2008. Measurement of small-strain
tests. Géotechnique 60 (7), 561–567. shear modulus Gmax of dry and saturated sands by bender element,
Lee, J.-S., Santamarina, J.C., 2005. Bender elements: performance and resonant column and torsional shear tests. Can. Geotech. J. 45, 1426–
signal interpretation. J. Geotech. Geoenviron. Eng. 131 (9), 1063–1070. 1438.

You might also like