Engineering Geology 207 (2016) 7890

Contents lists available at ScienceDirect

Engineering Geology

journal homepage: www.elsevier.com/locate/enggeo

Effects of gravel content on shear resistance of gravelly soils

Wen-Jong Chang ,1, Thitibhorn Phantachang 1
Department of Civil Engineering, National Cheng Kung University, 1 University Road, Tainan 70101, Taiwan

a r t i c l e i n f o a b s t r a c t

Article history: Shallow landslides in colluvium, which generally consists of wide particle gradation in a loose state, occur in
Received 9 January 2016 steep-slope terrains. To perform stability analyses, shear strength parameters that represent soil constituents,
Received in revised form 14 April 2016 in-situ stress state, and shearing mode are needed. This experimental study investigates the effects of gravel con-
Accepted 17 April 2016
tent on the shearing characteristics of gravelly soils based on the intergrain state concept. A series of drained sim-
Available online 20 April 2016
ple shear tests was performed on reconstituted specimens of gap- and well-graded sands with various gravel
Keywords:
content levels. The results reveal that: (1) the shear strength of gravelly soils depends on the packing condition
Gravelly soils of dominant particles, with the intergrain granular void ratio being a more representative parameter for the soil
Colluvium packing condition than the global void ratio; (2) the drained shear strength of loose, gravelly sand is dominated
Direct simple shear test by sand matrix properties, with the inclusion of gravel content reducing the strength by no more than 20% com-
Shear strength pared to that of pure sand; (3) the strength parameters of sand-like gravelly soils can be estimated from the sand
Intergrain state concept matrix with proper reduction.
2016 Elsevier B.V. All rights reserved.

1. Introduction
Colluvium is a heterogeneous and incoherent mass of soil materi-
al and/or rock fragments moved downhill by gravity and rain, and
typically in an unconsolidated, loose state with low shear strength.
Slope failures in colluvium triggered by rainfall or earthquakes
have been reported in the literature (e.g., Carro et al., 2003;
Shakoor and Smithmyer, 2005; Chang et al., 2005; Schulz, 2007). De-
pending on the composition of materials and degree of coherence,
colluvium can be categorized as rock or soil mass, the latter of
which is more vulnerable to slope failure. This study focuses on
sand-like colluvium, which is composed of a matrix of sand particles
mixed with rock fragments or gravel. The incoherence of sand-like
colluvium allows deposits to be treated as granular materials. Fig. 1
shows the cutting slope of the colluvium deposit in central Taiwan
described in Chang et al. (2005). The oversized particles are dis-
persed in the soil matrix, which mainly consists of sand.
Due to the presence of oversized gravel particles, block sampling and
large-scale testing apparatus are required to determine the shear
strength parameters of colluvial deposits. Limited by the sampling
methods and testing apparatus, laboratory testing on soils with over-
sized particles is a challenging task. In addition, the effects of particle
Corresponding author.
E-mail addresses: wjchang@mail.ncku.edu.tw (W.-J. Chang), rockmass@gmail.com
(T. Phantachang).
1
Department of Civil Engineering, National Cheng Kung University, Tainan 70101,
Taiwan.
size on colluvium or rockll are difcult to evaluate and could be site-
dependent. Indirect approaches such as back analyses, empirical corre-
lations, and parametric studies have thus been employed to determine
the shear strength parameters of colluvium. However, the uncertainties
embedded in these techniques are difcult to evaluate. Therefore, a sim-
ple but rational approach for evaluating the strength parameters of
sand-like colluvium for stability analyses is attractive for practical engi-
neering applications.
This experimental study investigates the effects of gravel content
(GC) on the shearing characteristics of gravelly soils based on the
intergrain state concept. A systematic study is performed on
reconstituted specimens of both gap- and well-graded sands with
various GC levels, subjected to drained, simple shear condition,
which represents the stress state and shearing conditions along a
sliding surface. A practical procedure is proposed for evaluating the
strength parameters of sand-like colluvium.
2. Literature review
2.1. Shear strength of gravelly soils
Soils with gravel particles are generally called gravelly soils, which
include colluvium, rockll, and tailings. Based on in-situ investigations,
Fleming and Johnson (1994) and Iannacchione and Vallejo (2000)
found that colluvium contained approximately 10% to 50% rock frag-
ments. Laboratory tests on gravelly soils have been performed on
these materials. To facilitate laboratory testing, four common methods
have been proposed to redcue the maximum particle size of specimens,
namely (1) the scalping technique (Zeller and Wulliman, 1957), (2) the

http://dx.doi.org/10.1016/j.enggeo.2016.04.015
0013-7952/ 2016 Elsevier B.V. All rights reserved.
 W.-J. Chang, T. Phantachang / Engineering Geology 207 (2016) 7890
Fig. 1. Cutting slope of typical colluvium in Taiwan.
parallel gradation technique (Lowe, 1964), (3) the generation of a qua-
dratic grain-size distribution curve (Fumagalli, 1969), and (4) the re-
placement technique (Frost, 1973).
Holtz (1961) performed triaxial tests on scalped, reconstituted
gravelly clay and concluded that the friction angle increases as the
GC increases and that the strength increment is related to clayey
soils. Marachi et al. (1972) used the parallel gradation technique to
prepare remolded gravelly sands and concluded that the frictional
angle decreases as the maximum particle size increases and that
the major factors affecting the stressstrain curve are the oversized
particle shape and the gradation curve. Fannin et al. (2005) conduct
direct shear tests on both a eld block sample and matrix materials
and concluded that under low vertical stress (5 to 20 kPa), the resid-
ual friction angles of the mixtures and the matrix materials are
similar and that the residual strength is mainly affected by soil com-
position, particle shape, and gradation. Fragaszy et al. (1992) con-
ducted drained triaxial tests on a soil matrix mixed with oversized
particles and showed that the oversized particles oating in the
ne matrix had an insignicant inuence on soil shear strength and
deformation characteristics.
Recent studies related to shearing behavior of gravelly soil mix-
tures were carried out by Vallejo and Mawby (2000); Varadarajan
et al. (2003); Kokusho et al. (2004); Simoni and Houlsby (2006),
and Anantanasakul et al. (2012). These studies were conducted
based on the global void ratio framework. Studies of the effects of
nes content on liquefaction characteristics revealed that the
engineering properties of mixed soils, such as clayey and silty
sands, depend on the packing condition of dominant (or matrix)
particles, which is represented by the intergranular void ratio. Ex-
perimental investigations on liquefaction characteristics of mixed
soils based on the intergrain state framework were conducted by
Thevanayagam and Martin (2002); Naeini and Baziar (2004), and
Chang and Hong (2008).
Kokusho et al. (2004) performed cyclic triaxial tests on reconstituted
gravelly soil specimens and concluded that the undrained cyclic
strength of such soils is mainly dependent on relative density rather
than particle gradation. Chang et al. (2014) studied the liquefaction
characteristics of gap-graded gravelly soils based on the intergrain
state concept and categorized gap-graded, sand-gravel mixtures as
sand-like, gravel-like, and in-transition soils, each of which showed dif-
ferent liquefaction characteristics. Based on these studies on gravelly
79
soils subjected to cyclic loading, the monotonic responses of gravelly
soils can be studied based on the intergrain state concept and the effects
of GC can be quantitatively described.
So far, no systematic studies have been conducted to evaluate the ef-
fects of GC in the soil matrix while considering the packing state of ma-
trix particles. The main objective of this paper is to investigate the
effects of GC on the shearing characteristics of gravelly soils such as col-
luvium, tailings, and rockll by performing consolidated, drained simple
shear tests. The concepts of binary packing and intergrain state are
applied to simplify the particle packing system. Moreover, systematic
experiments of gap- and well-graded sand matrices with various GC
levels were performed to evaluate the applicability of these concepts
to shearing characteristics such as shear stressstrain curves, volumetric
strain variations, and drained shear strengths at various shear strain
levels.
2.2. Intergrain state concept
The intergrain state concept, which considers the microstructure of
particle packing, has been used to describe the engineering properties
of mixed soils, such as the strength of sandy gravels (Fragaszy et al.,
1992), undrained shear strength of silty sands (Georgiannou et al.,
1990), and liquefaction resistance of silty sands (Kuerbis et al., 1988).
The packing conditions and void distributions in natural soils are com-
plicated because of large variations in particle size and deposition con-
ditions. To simplify the packing conditions of mixed soils, binary
packing models have been proposed. The three major assumptions in
an idealized binary packing model are as follows: (1) only two distinct
particle sizes are considered, (2) the ratio of coarse to ne particle diam-
eters is large (no less than 20), and (3) the packing of coarse particles is
not affected by ne particles, and vice versa. Loose colluvium with over-
sized particles oating in the granular matrix satises most of these
assumptions.
Because void distribution is one of the major factors affecting the
shearing behavior of granular soils at intermediate to large strain levels,
various void parameters inferred from idealized binary packing models
have been proposed, as shown in Fig. 2. The void ratio (e), dened as the
volumetric ratio between voids and the volume of all solid particles, is
used as an overall packing index for soils (Fig. 2(a)). Skeleton void ratios
are used to describe the idealized packing conditions of the dominant
particle fraction. For soils with distinct coarse and ne particles, the
80 W.-J. Chang, T. Phantachang / Engineering Geology 207 (2016) 7890

(a) (b) (c)

Voids, e Vv Voids, e Vv

Vs + Vv
Apparent voids,
e + (1 GC)
Fine grain Vs Fine grain Vs
1 GC 1 GC

Coarse grain Coarse grain

GC
Vg GC
Vg

Global void ratio, e Sand skeleton void ratio, esk Gravel skeleton void ratio, egk
Vg Vs = 1.0 esk = e egk = e (1 GC)
1 GC GC
e = Vv
Vg V s

Fig. 2. Illustration of intergrain state concept.

skeleton void ratio is dened as the volumetric ratio between the voids
formed by the soil skeleton and the volume of particles that make up the
skeleton. Assuming that the specic gravities of all soil particles are the
same and the soil system has a unit solid volume, for sandgravel mix-
tures with gravel particles oating in the sand matrix, the sand skeleton
void ratio (esk) is expressed as:
e under a given effective stress, respectively. Typical void ratio ranges
esk
1GC
where GC is dened as the weight percentage of the gravel fraction in all
soil particles. The phase diagram is shown in Fig. 2(b). The values of e
and GC are expressed in decimal form. If the gravel particles are in con-
tact with each other and the sand particles ll the voids formed by grav-
el particles, the gravel skeleton void ratio (egk), shown in Fig. 2(c), is
expressed as:
e 1GC
egk
GC
The two equations are derived from phase diagrams of a soil system
with a unit solid volume. The individual volumes for gravel and sand
fractions can be dened using the denition of GC at a given specic
gravity for all soil particles. The two skeleton void ratios are used in
this study to describe the packing conditions of sand- and gravel-like
soils.
Depending on the GC levels, the engineering properties of sand
gravel mixtures are dominated by either sand or gravel microstruc-
tures, resulting in sand- or gravel-like soil behaviors. For sandgravel
mixtures with gravel particles oating within the sand matrix, the
engineering properties of the mixtures are primarily controlled by
the packing conditions of the sand particles, and the mixtures be-
have as sand-like soils. Consequently, the sand skeleton void ratio
is a more representative packing index than the global void ratio
for correlating the soil characteristics with the packing conditions.
Similarly, gravelly soils with sand particles lling the voids formed
by contacting gravel particles behave as gravel-like soils, and thus
the gravel skeleton void ratio should be used for correlations.
The GC that divides sandgravel mixtures into sand- or gravel-
like soils is dened as the transitional gravel content (TGC).
Adopting the derivation by Chang and Hong (2008), the TGC is
expressed as:
1 e2
TGC 3
1 e1 e2
where e1 and e2 are the void ratios of pure gravel and pure sand
1
for gravel and clean uniform sands are 0.85 to 0.14 and 1.0 to 0.4, re-
spectively (Mitchell and Soga, 2005). Therefore, the TGC value for
gravelly soils ranges from 62 to 93%, increasing with decreasing
gravel void ratio. It should be noted that the TGC for a specic com-
position is a range rather than a single value, and that soils in the
transition range are in a metastable condition, with characteristics
different from those of either sand- or gravel-like soils. Based on
the eld investigations of Fleming and Johnson (1994) and
Iannacchione and Vallejo (2000), most colluvium can be character-
ized as sand-like soils.
2
2.3. Direct simple shear test
The direct simple shear (DSS) test has been extensively used by re-
searchers to evaluate the shear stressstrain responses and strength pa-
rameters of soils under simple shear strain conditions. Ladd et al. (1977)
stated that the DSS test is the preferred test because it represents the av-
erage shear condition along a failure surface of slope failures. Airey and
Wood (1987) found that the strength parameters from DSS tests are
close to the results from in-situ vane shear tests and back analyses. Pre-
vious studies by Airey et al. (1985); Airey and Wood (1987), and Boylan
and Long (2009) stated that the DSS test has advantages over the triax-
ial compression test because in the former, the sample assembly is sim-
ple and the in-situ failure mechanism is better represented than the
latter. In addition, simple shear requires a shorter consolidation time
and principal stress rotation is allowed during the shearing process.
Two types of DSS apparatus have been developed. The Cambridge
type simple shear apparatus, proposed by Roscoe (1953), requires a
cube-shaped specimen. The Norwegian Geotechnical Institute (NGI)
type apparatus was developed by Bjerrum and Landva (1966), who
adopted the concept proposed by Kjellman (1951). The NGI type appa-
ratus uses a cylindrical specimen surrounded by a wire-reinforced
 W.-J. Chang, T. Phantachang / Engineering Geology 207 (2016) 7890
membrane to maintain the K0 condition, at which lateral deformation is
prohibited and the coefcient of lateral earth pressure is equal to K0.
Chang et al. (2014) developed a stack-ring-reinforced membrane to re-
place the wire-reinforced membrane and conducted cyclic DSS tests on
gravelly soils. The soil container developed by Chang et al. (2014) is
used in this study.
3. Direct simple shear test on reconstituted sand-gravel mixtures
3.1. DSS soil container and testing apparatus
The soil element on the sliding surface of a slope failure is in a plane
strain state and subjected to the K0 condition. To mimic the simple shear
mode, a soil container capable of providing lateral restraint and
allowing shearing in the horizontal direction is required. The system
that models this stress state and shearing mode is referred as a K0-DSS
system, in which the soil container capable of providing lateral restraint
and shearing in the vertical plane is required. Wire-reinforced mem-
branes have been used in NGI-type DSS to create the K0 condition dur-
ing simple shearing testing despite some concerns regarding the stress
state. The most signicant concern is that wired-reinforced membranes
cannot assure sufcient lateral restraint and prevent distortion in the
horizontal plane due to the nonuniform complementary shear stress
on the vertical plane. The low stiffness and varying cross-section of a
wire-reinforced membrane make it unsuitable for gravelly soils, which
have a high stiffness and exhibit a membrane penetration effect.
To reduce these concerns, Chang et al. (2014) developed a soil con-
tainer that uses a conventional latex membrane reinforced with low-
friction stack rings. Stack rings made of stainless steel plate were
manufactured with an inner diameter of 102 mm, an outer diameter
of 112 mm, and a thickness of 2 mm. Each ring was coated with a
0.01-mm-thick Teon layer to reduce the friction between rings. A
latex membrane for a 100-mm-diameter specimen was glued with liq-
uid epoxy onto the inner face of the ring layer by layer, with layer gaps
of less than 0.01 mm. In this study, 15 rings were stacked to t soil spec-
imens with dimensions of 100 mm in diameter and 30.3 mm in height,
corresponding to a height-to-diameter ratio (H/D) of 0.3.
Vertical direct drive motor
Vertical frames
Vertical LVDT
Top cap
Teflon stack rings
Bottom platen
Shear slide table
Horizontal frames
Cell fixer
Horizontal direct drive motor
Fig. 3. NCKU direct simple shear system.
81
A modied NGI-type DSS system with a stack-ring-reinforced mem-
brane was used in this study. The DSS system was developed at National
Cheng Kung University (NCKU-DSS) and is capable of applying mono-
tonic and cyclic loading for both stress- and strain-controlled conditions
in vertical and horizontal directions using two closed-loop direct-drive
servo motors, as shown in Fig. 3. The closed-loop electric actuator of
the NCKU-DSS system is composed of a servo motor, ball screw, ball
spline, and control hardware and software. The servo motor has a
resolution of 614,400 steps per revolution, a peak torque of 110 Nm,
and a working axial load of up to 25 kN. A precision, caged ball screw
with a lead of 5 mm per revolution was connected to a ball spline to
convert the rotational motion to a linear stroke for a resolution of
8.1 10 6 mm/step. To facilitate different feedback control modes
(force, displacement, etc.), position control mode of the motor was
adopted for different feedback signals. A motion controller that sends
pulse signals based on a proportional integral derivative (PID) algo-
rithm between the target and feedback signals was used to control the
actuators.
Two closed-loop actuators were integrated into the testing platform
to perform stress- and strain-controlled cyclic shearing in both normal
and shear directions. To minimize the rocking deformation of soil spec-
imens due to the lack of complementary shear stresses on the vertical
side, the normal loading frame was reinforced with a high-stiffness
supporting system. Caged ball guides and a linear bushing were used
in both normal and shear loading mechanisms to minimize and main-
tain constant friction. The friction measured from the horizontal motion
was 5.0 N; this value was used to correct the shear stress calculations.
The NCKU-DSS system provides vertical and horizontal loading, de-
formation, and pore water measurement. Vertical loading was applied
using a closed-loop actuator connected to the top cap of the soil contain-
er. To control the vertical load, a button-type load cell with a capacity of
25 kN was installed under the bottom platen to accurately measure the
vertical force applied on the specimen. This reading was used as the
feedback signal of the vertical actuator. Shear strain was induced by
the horizontal actuator under displacement control mode. A constant
deformation rate of 0.1 mm/min was applied on the bottom platen
and a load cell with a capacity of 2.0 kN was used to measure the
force. A 24-bit automatic data acquisition system was used to record
High stiffness
support
Button load cell
Horizontal LVDT
Horizontal
load cell
82 W.-J. Chang, T. Phantachang / Engineering Geology 207 (2016) 7890

the vertical and horizontal loads, vertical and horizontal displacements,

pore pressure during undrained shearing, and volume change during
drained shearing.

3.2. Properties of gravelsand mixtures and sample preparation

Because the oversized particles in colluvium are residual materials

without severe abrasion (as shown in Fig. 1), crushed coarse aggregates
were used as gravel particles. To meet the minimum size of gravel in the
USCS classication and the required ratio of specimen height to maxi-
mum diameter particle size (H/dmax) in the DSS test, crushed coarse ag-
gregates with particle sizes ranging from 4.75 to 10 mm and a mean
grain size of 6.7 mm were used in this study as gravel particles. Clean
sand particles with a round/sub-round shape were used as ne parti-
cles. A uniform sand with a mean particle size of 0.56 mm was used to
represent poorly-graded (SP) sand. A mixture of three different sands
was mixed to meet the well-graded (SW) requirements with a mean
grain size of 0.55 mm. The basic properties of the testing materials are
shown in Table 1. The angles of repose for gravel, SP sand, and SW
sand were 23.7, 27.0, and 27.5, respectively. The gravel and sand par-
ticles were mixed to prepare specimens with various GC levels. The
grain size distributions of specimens are shown in Fig. 4.
To prepare specimens with various GC levels, the weights of gravel
and sand were calculated based on weight proportions and the mixtures
were mixed with 5% of water for moist tamping. The soil mixtures were
divided into ve layers and compacted using a lightweight steel rammer
to control the height of each soil layer with the designated void ratio.
Fig. 5(a) and (b) shows the compacted specimens with various propor-
tions of poorly- and well-graded sands with gravels.
3.3. Testing procedure and experimental program
Drained DSS tests under a constant vertical stress were conducted
on the sandgravel mixtures. The testing procedure combines ASTM
standard D7181 (consolidated drained triaxial compression test) and
D6528 (undrained direct simple shear testing of cohesive soils) to de-
termine the drained shear strength of gravelly soils. A vertical stress of
55 kPa with a back pressure of 50 kPa was applied to soil specimens
for saturation. To determine the degree of saturation, a modied B-
value in the K0 condition (BK0) was evaluated based on the derivation
by Chang et al. (2014):
u 1
BK 0
v nC v 3 maximum and minimum void ratios (emax and emin) for sand in the
1
C sk 1 2K 0
where u is the induced excess pore pressure, v is the increment of
vertical stress, n is the porosity, Cv is the compressibility of the pore
uid, and Csk is the compressibility of the soil skeleton. The B-check
was performed by applying vertical stress in a 50-kPa increment and
measuring the induced excess pore pressure under the undrained
Table 1
Physical and mechanical properties of testing materials.
Fig. 4. Grain size distribution of testing soils.
condition. Kokusho (2000) stated that the saturation process is com-
pleted when the B-value is larger than 0.90 for sand and gravel. In addi-
tion, the degree of saturation was veried by measuring the water
content of the specimens after testing. The results show that the degree
of saturation was above 99%. This process took one to four hours de-
pending on specimen proportions.
After the completion of the saturation process, K0 consolidation
under effective vertical stresses of 50, 100, and 150 kPa was per-
formed. The consolidation process typically took one to two hours.
During the shearing stage, consolidated specimens were monotoni-
cally sheared to the maximum shear strain level of 40% under a
xed strain rate of 0.1 mm/min, while the vertical stress was main-
tained constant.
Two series of consolidated, drained simple shear tests were conduct-
ed to investigate the effects of GC on the strength parameters of poorly-
graded (SP) and well-graded (SW) sands. The rst (SP) series was con-
ducted to study the effects of GC on the binary packing model, which is
applicable to two distinct particle sizes. The second (SW) series was
performed to evaluate the applicability of the intergrain state concept
to well-graded sand mixtures.
Reconstituted specimens with GC levels of 0% (pure sand) to
100% (pure gravel) were prepared to cover sand-like, gravel-like,
and in-transition gravelly soils. To represent the loose packing con-
dition observed in most of vulnerable colluvium, soil samples were
assigned a sand skeleton of around 0.55 for both testing series. The
4
SP series were 0.638 and 0.446, respectively, and those for gravel
were 0.570 and 0.465, respectively. Based on Eq. (3), the TGC range
for the gap-graded soils was about 70.6 to 79.9% and that for the
well-graded sand was about 70.6 to 77.6%. Therefore, specimens
with GC of 0 to 60% were likely to be sand-like soils; these specimens
were prepared with similar sand skeleton void ratios to evaluate the
effects of GC on the shear strength of gravelly soils. Specimens with
GC of 75% were used to represent in-transition soils. Specimens
with GC of 90 and 100% were used for batch tests for pure gravel
and the effects of sand on gravel-like soils.

Property Gravel SP sand SW sands 4. Testing results

Specic gravity 2.62 2.65 2.71
D10 (mm) 5.00 0.30 0.10 4.1. Summary of testing results
D50 (mm) 6.70 0.56 0.55
Coefcient of uniformity, CU 1.44 2.00 6.50
Twenty four tests in the SP series and 21 tests in the SW series
Coefcient of curvature, CC 1.00 1.39 2.71
USCS classication GP SP SW-ML were conducted. The specimen properties and results are summa-
emax 0.570 0.638 0.614 rized in Tables 2 and 3 for poorly- and well-graded sand matrices,
emin 0.465 0.446 0.366 respectively. The degree of saturation is above 0.95 for all tests.
Relative density, DR (%) 19.1 45.7 25.8 The skeleton void ratios were calculated based on Eqs. (1) and (2).
Angle of repose (degree) 23.7 27.0 26.7
For sand-like specimens, the consolidated sand skeleton void ratios
 W.-J. Chang, T. Phantachang / Engineering Geology 207 (2016) 7890 83

GC = 0% GC = 15% GC = 30% GC = 45%

GC = 60% GC = 75% GC = 90% GC = 100%

(a) Gravel with gap-graded sand mixtures

GC = 0% GC = 15% GC = 30% GC = 45%

GC = 60% GC = 75% GC = 90% GC = 100%

(b) Gravel with well-graded sand mixtures
Fig. 5. Photographs of compacted soil mixtures with various GC levels. (a) Gravel with gap-graded sand mixtures, (b) Gravel with well-graded sand mixtures.

were controlled to be in the range of 0.51 to 0.55. For gravel-like
soils, the consolidated gravel skeleton void ratios were in the
range of 0.52 to 0.55. The relative densities for pure SP sand, SW
sand, and pure gravel specimens were 45.7, 25.8, and 19.1%,
respectively.
The shear stress () was calculated as the horizontal force divided
by the cross section area of the specimen. The shear strain () was
evaluated as the horizontal displacement divided by the thickness
of the specimen. As recommended in ASTM D6528, the shear
strength in this study refers to shear stress at 20% shear strain. The
critical state is reached when shear stress remains constant under
continued shearing at large strain levels (Salgado et al., 2000).
Most specimens at 30% of shear strain had reached a constant shear
stress and volume state.
4.2. Results of poorly-graded sand series
The drained DSS results for the SP series with 0, 45, 75, and 100% GC
are shown in Fig. 6. Tests for GC of 0 and 100% are batch tests for two soil
types. The results of GC = 45% are typical responses of sand-like soils
and those of GC = 75% are typical responses of in-transition soils. The
shear stressstrain curves for all SP series specimens show typical
loose sand responses such as a monotonic increase in shear stress
with shear strain level and an increase of contractive volumetric strain
with increasing vertical effectives stress.
Volumetric strain is dened as negative when the height of the
specimen decreased. The results show that specimens contracted
under small strain and then dilated at a shear strain of approximate-
ly 10% at a low vertical effective stress of 50 kPa. The volumetric re-
sponse is possibly due to the slippage of particles in the vertical
direction then particles moved over each other under low conned
stress. Increasing the GC increased the dilation of the soil mixtures.
For vertical effective stresses of 100 and 150 kPa, volumetric strain
trends are similar to the typical contractive behavior of loose sands.
The normalized shear stressstrain curves of the SP series for various
GC levels are shown in Fig. 7. The results show that the shapes of the
shear stressstrain curves fall within a narrow range when the shear
stress is normalized with the initial vertical effective stress ('v). This
trend reects that soil mixtures are stress-dependent. Additionally,
the majority of specimens reached constant shear stress and constant
volume at a shear strain level around 30%, except those at a vertical
stress of 50 kPa.
The MohrCoulomb failure envelopes obtained using the shear
resistance at shear strain levels of 20 and 30% are shown in Fig. 8.
The failure criterion at 20% shear strain level is referred to the recom-
mendation of ASTM D6528. The majority of the specimens reached a
84 W.-J. Chang, T. Phantachang / Engineering Geology 207 (2016) 7890

Table 2 4.3. Results of well-graded sand series

Testing results of gap-graded sand matrix with gravel (SP series).

GC Behavior 'vo BKo-value Consolidated void @ = 20% The drained DSS results for the SW series with 0, 45, 75, and 90% GC
classication ratio are shown in Fig. 9. For sand-like specimens in the SW series, the sand
(%) (kPa) e esk egk (kPa) skeleton void ratio was 0.55. The results for the specimen with 90% GC
represent typical gravel-like soils with sand content. For sand-like spec-
50 0.95 0.56 0.56 27.5
0 Sand-like 100 0.98 0.55 0.55 48.9 imens, typical loose sand responses are observed and the state of con-
150 0.96 0.54 0.54 76.8 stant shear stress was reached at around the 25% shear strain level.
50 0.98 0.44 0.51 8.58 26.0 However, the volumetric strain continuously increased for shear strain
15 Sand-like 100 0.95 0.43 0.51 8.54 53.1 levels greater than 25%.
150 0.96 0.47 0.55 8.77 80.0
50 0.99 0.38 0.55 3.62 24.4
The MohrCoulomb failure envelopes of the SW series are shown in
30 Sand-like 100 0.99 0.37 0.53 3.57 51.2 Fig. 10 for strain levels of 20 and 30%. Again, the lowest friction angle
150 0.99 0.37 0.53 3.57 75.2 was obtained at a GC of 75%, at which the mixture was in the in-
50 0.97 0.30 0.55 1.89 23.8 transition state. For sand-like specimens, the soils showed contractive
45 Sand-like 100 0.96 0.28 0.51 1.85 55.2
responses with monotonically increasing shear stress and volume con-
150 0.96 0.28 0.52 1.85 77.3
50 0.98 0.22 0.54 1.03 19.1 traction for all specimens in the SW series. In general, the testing results
60 Sand-like 100 0.97 0.21 0.52 1.02 42.1 agree with those for loose granular soils under drained shearing
150 0.96 0.22 0.55 1.03 68.0 conditions.
50 0.97 0.15 0.58 0.53 19.3
75 In-transition 100 0.95 0.13 0.51 0.50 36.2
150 0.95 0.14 0.55 0.52 59.5
5. Discussion and application
50 0.97 0.39 3.87 0.54 22.0
90 Gravel-like 100 0.98 0.38 3.83 0.54 43.0 5.1. Effects of GC on stressstrain curves of sand-like soils
150 0.96 0.37 3.66 0.52 64.0
50 0.97 0.55 0.55 21.0
Because the amount of rock fragments in colluvium is approximately
100 Gravel-like 100 0.95 0.53 0.53 44.1
150 0.99 0.55 0.55 68.0 10% to 50% (Fleming and Johnson, 1994; Iannacchione and Vallejo,
2000), the majority of colluvium is sand-like soils. To investigate the ef-
fects of GC on stressstrain curves, normalized stressstrain curves
under effective vertical stresses of 50 and 150 kPa are shown in
relatively constant shear resistance at the 30% shear strain level. The Fig. 11. The curves for a vertical consolidation stress of 50 kPa represent
state at this shear level is close to the critical state. Because of mono- the shearing behavior at low conning stress, in which the gap-graded
tonic increases, the failure envelopes at the 30% shear strain level are specimens show a dilative response at large strain levels and lower vol-
greater than those at 20% shear strain for all GC levels. The lowest ef- umetric contractions than those at larger conning stress in well-
fective friction angle was observed for GC = 75%, which was the in- graded sands. The curves for a vertical consolidation stress of 150 kPa
transition state with the metastable packing condition. The drained represent the responses under high conning stress, in which all
friction angle for sand-like specimens was close to the angle of re- sand-like specimens show contractive responses during the shearing
pose of pure SP sand. process.
Fig. 11(a) shows the results of the SP series with GC of no greater
than 75%, which includes the sand-like and in-transition specimens.
The results show that the curves of in-transition specimens deviate
from those of the sand-like specimens (GC of no less than 60%). The de-
Table 3 viation is more signicant under 50-kPa effective vertical stress than
Testing results of well-graded sand matrix with gravel (SW series). under 150-kPa effective vertical stress due to the increase of TGC from
the reduction of the void ratio of pure sand under high conning stress.
GC Behavior 'vo BKo-value Consolidated void @ = 20%
classication ratio Nevertheless, the results show that the sand-like and in-transition soils
behave differently and that GC has minor effects on the normalized
(%) (kPa) e esk egk (kPa)
stressstrain curves of a gap-graded sand matrix.
50 0.99 0.54 0.54 20.3 Fig. 11(b) shows the results of the SW series with a GC of no greater
0 Sand-like 100 0.98 0.53 0.53 48.9
150 0.98 0.53 0.53 78.9
than 75%. Different responses are observed for sand-like and in-
50 0.97 0.47 0.55 8.79 16.4 transition specimens under a low consolidation stress level (50 kPa).
15 Sand-like 100 0.97 0.45 0.53 8.68 40.0 However, deviation starts to appear at a GC of greater than 60% for
150 0.99 0.44 0.52 8.61 71.6 high consolidation stress level (150 kPa). Therefore, the results for spec-
50 0.95 0.38 0.54 3.60 21.6
imens with 60% GC are excluded from the sand-like soils. The results
30 Sand-like 100 0.95 0.38 0.54 3.60 41.5
150 0.98 0.37 0.53 3.58 64.0 indicate that GC has minor effects on the stressstrain curves of the
50 0.95 0.30 0.54 1.88 17.0 well-graded sand matrix.
45 Sand-like 100 0.96 0.29 0.53 1.87 39.0
150 0.99 0.29 0.53 1.87 70.0 5.2. Effects of GC on drained resistance of sand-like soils
50 0.99 0.21 0.53 1.02 21.0
60 Sand-like 100 0.99 0.22 0.55 1.03 35.6
150 0.98 0.21 0.53 1.02 62.0 The effects of GC on drained resistance were investigated from the
50 0.99 0.13 0.54 0.51 14.3 shear resistances and drained friction angles. Shear resistances at 20%
75 In-transition 100 0.98 0.13 0.52 0.51 37.7 shear strain are shown in Fig. 12. For a gap-graded sand matrix with a
150 0.99 0.13 0.53 0.51 65.3
constant sand skeleton void ratio, the shear resistance is relatively con-
50 0.99 0.38 3.81 0.53 25.0
90 Gravel-like 100 0.99 0.40 3.98 0.55 41.8 stant for a GC of no greater than 45%. In the SP series, the particle size
150 0.98 0.39 3.90 0.54 66.3 distributions satisfy the requirements of the binary packing model.
50 0.97 0.55 0.55 21.0 Therefore, the drained shear resistance of the sand-like mixture is dom-
100 Gravel-like 100 0.95 0.53 0.53 44.1 inated by the sand matrix and should be a function of the void ratio of
150 0.99 0.55 0.55 68.0
the sand matrix. In other words, the drained shear resistances of sand-
 W.-J. Chang, T. Phantachang / Engineering Geology 207 (2016) 7890 85

Fig. 6. Testing results of SP series with various GC levels.

like mixtures should be a constant for a given packing of the sand matrix shown in Fig. 11(a). The '30 values decrease linearly with GC. The
(i.e., xed esk). The insensitivity of drained shear resistance to GC for the '30 value of gap-graded, sand-like soils with GC, denoted as '30(GC),
sand-like specimens of the SP series validates the applicability of the bi- can be approximated as:
nary packing model to gap-graded, sand-like gravelly soils.
0 0
In the well-graded sand matrix, the inclusion of gravel shows a 30 GC 30 00:22GC for GC 60% 5
slightly different effect. The inclusion of gravel particles generally re-
duces the shear resistance with increasing GC, as shown in Fig. 12(b). where '30(0) is the drained friction angle of the pure sand matrix at
The amount of reduction increases as the effective vertical stress shear strain level of 30% and GC is a percentage. The reduction in
increases. The lowest resistance was obtained at GC of 60 to 75% in shear resistance could be attributed to the decrease of force chains de-
the SW series. veloped among particles due to the smaller specic surface area of
To avoid the effect of vertical stress on drained resistance, the gravel than that of sand.
drained friction angle () from both test series are shown in Fig. 13, In the well-graded sand matrix with a GC of no greater than 45% or
along with the variations of from the values of pure sands (GC = sand-like soils, the '20 values decrease nonlinearly with the inclusion
0%). The values at shear strain levels of 20 and 30% are presented of GC, as shown in Fig. 13(b). The nonlinear relationship can be approx-
for various failure criteria. Because of the contractive response during imated as:
shearing, the values at a shear strain of 30% (denoted as '30) are
greater than those at a shear strain of 20% (denoted as '20). In the 0 0
h i
20 GC 20 0 10:9GC 0:014GC 2 for GC 45% 6
gap-graded sand matrix with a GC of no less than 45%, the '20 values
are relatively constant with a percentage difference of 0.8 to 3.7%.
The '20 value decreased by 12% for a GC of 60%, which is the upper where '20(0) is the drained friction angle of the pure sand matrix at
bound of GC for sand-like soils based on the stressstrain curves shear strain level of 20%.
86 W.-J. Chang, T. Phantachang / Engineering Geology 207 (2016) 7890

Fig. 7. Normalized shear stressstrain curves of SP series.

However, the '30 values of sand-like soils with the well-graded sand
matrix and a GC of no greater than 45% decrease less than 3%, which in-
dicates that the behavior of the binary packing model might be more ap-
plicable at the critical state for the well-graded sand matrix. The
similarity between the drained friction angles of the sand matrix and
sand-like gravelly soils agrees with the majority of back analysis cases,
in which the drained friction angle of colluvium is close to or lower
than that of sand.
colluvium soils might be feasible without using large-scale element
tests. A tentative procedure is proposed based on the preliminary re-
sults. First, the GC and void ratio are prerequisite parameters for esti-
mating TGC and the friction angle. In practice, the GC and void ratio of
a site can be determined from open trenches or obtained from pub-
lished data (e.g., Kokusho et al., 2004). Alternatively, the void ratio can
be estimated from empirical correlations among the maximum shear
modulus (Gmax), void ratio, and effective conning stress ('0). The
general form can be expressed as (Ishihara, 1996):

5.3. Proposed procedure for evaluating colluvium strength parameters  0 n

The testing results reveal that the intergrain state concept is applica-
ble to categorizing and characterizing colluvium soils with minor ad-
justments. Combining the testing results of the gap- and well-graded
sand matrices with a specic sand-skeleton void ratio and various GC
levels, evaluation of drained shearing characteristics of sand-like
G max V 2s AF e 0 7
where is the total density, A and n are material constants, F(e) is the
void ratio function, and the values of Gmax and '0 are in kPa. The void
ratio can be back-calculated from the in-situ Vs, soil prole, and typical
material constants listed in the study of Ishihara (1996).

Fig. 8. Failure envelopes of SP series.

 W.-J. Chang, T. Phantachang / Engineering Geology 207 (2016) 7890
Fig. 9. Testing results of SW series with various GC levels.
The next step is to categorize the colluvium into sand-like or
gravel-like soils based on the GC and the estimated TGC value. The
TGC value can be evaluated using Eq. (3) if the void ratios of the
pure sand and gravel particles are available. Alternatively, loose col-
luvium with a GC of less than 60% for the gap-graded sand matrix or
45% for the well-graded sand matrix can be categorized as a sand-
like soil based on the testing results. The proposed procedure is not
applicable to in-transition and gravel-like soils. However, a 20% re-
duction of the pure sand matrix strength for in-transition soils
seems a rational approximation based on current testing data.
Because the sand/silt matrix dominates the shearing behaviors of
sand-like colluvium, the shear responses of the sand/silt matrix must
be characterized. A modied scalping technique, in which the gravel
particles are removed and the sand skeleton void ratio of sand/silt com-
ponents is maintained, is recommended for laboratory element tests. In
addition, empirical correlations among shear strength and the in-situ
index or mechanical properties can be used to estimate the shearing
characteristics of the sand/silt matrix under a specic sand skeleton
void ratio.
87
The nal step is to perform GC correction of the shearing charac-
teristics for different types of sand matrix. A series of DSS tests for
various GC levels under the in-situ sand skeleton void ratio is pre-
ferred. However, limited by access to testing equipment, a practical
approach is recommended here based on the testing results of this
study. For the gap-graded sand matrix with a GC of no greater than
45%, no correction is required for '20 and the normalized stress
strain curves. Eq. (5) can be used for correcting '30 values of speci-
mens with a GC of less than 60%. For the well-graded sand matrix
with a GC of no greater than 45%, no correction is required for
'30 and the normalized stressstrain curves. The '20 values for
the well-graded sand matrix can be corrected using Eq. (6) for a GC
of no greater than 45%.
6. Conclusion
To investigate the effects of gravel particles on the shearing charac-
teristics of gravelly soils, a series of drained DSS tests was performed
on gap- and well-graded sand matrices with various GC levels. The
88 W.-J. Chang, T. Phantachang / Engineering Geology 207 (2016) 7890

Fig. 10. Failure envelopes of SW series.

intergrain state concept was introduced to describe the packing condi-
tions of mixed soils and the binary packing model was used to catego-
rize sand-like, in-transition, or gravel-like soils. The testing results and
ndings are summarized as follows:
1. Based on the variations of stressstrain curves and drained shear re-
sistances, gravelly soils consisting of a sand matrix and rock frag-
ments can be categorized as sand-like or in-transition soils
depending on the GC and TGC. GC in sand-like gravelly soils has
minor effects on the normalized stressstrain curve for both gap-
and well-graded sand matrices.
2. The insensitivity of drained shear resistances to GC, as determined
from the results of SP series, validates the applicability of the bina-
ry packing model to gap-graded, sand-like gravelly soils. In the
well-graded sand matrix, the inclusion of gravel particles general-
ly reduces the shear resistance as the GC increases. Nevertheless,
the GC in loose, gravelly soils will not signicantly increase the
drained shear resistance.
3. The lowest shear resistance of sandgravel mixtures was obtained
when the mixture was in the in-transition condition. The reduction
was less than 20% compared to the pure sand matrix.
4. A preliminary procedure for evaluating the drained shear resistance
of loose, sand-like colluvial soils was proposed. The procedure in-
cludes screening based on GC, evaluating the drained resistance of
the sand matrix, and conducting GC correction.
5. To determine the shearing characteristics of colluvial soils from labo-
ratory element tests, a modied scalping technique, in which the
gravel particles are removed and the sand skeleton void ratio of

(a)

(b)
Fig. 11. Effects of GC on stressstrain curves: (a) SP series and (b) SW series.
 W.-J. Chang, T. Phantachang / Engineering Geology 207 (2016) 7890 89

Acknowledgment

This study was supported by the Ministry of Science and Technology,

Taiwan, under grant NSC 100-2625-M-006-008-MY3, which is grateful-
ly acknowledged. Any opinions, ndings, and conclusions or recom-
mendations expressed in this material are those of the authors, and do
not necessarily reect the views of the Ministry of Science and Technol-
ogy, Taiwan.

(a)
References

(b)
Fig. 12. Effects of GC on shear resistance (a) SP series and (b) SW series.
sand/silt components is maintained, is recommended. Alternatively,
empirical correlations among the shear strength and in-situ index
or mechanical properties might be used to estimate the shearing
characteristics of the sand/silt matrix under a specic sand skeleton
void ratio.
(a)
(b)
Fig. 13. Variation of drained friction angle of sand-like soils with GC: (a) SP series and
(b) SW series.
Airey, D.W., Wood, D.M., 1987. An evaluation of direct simple shear tests on clay.
Gotechnique 37 (1), 2535.
Airey, D.W., Budhu, M., Wood, D.M., 1985. Some Aspects of the Behaviour of Soils in Sim-
ple Shear. In: Banerjee, P.K., Buttereld, R. (Eds.), Developments in Soil Mechanics
and Foundation EngineeringStressStrain Modelling of Soils vol. 2. Elsevier Applied
Science Publishers, London, pp. 185213.
Anantanasakul, P., Yamamuro, J., Kaliakin, V., 2012. Stressstrain and strength char-
acteristics of siltclay transition soils. J. Geotech. Geoenviron. Eng. ASCE 138
(10), 12571265.
ASTM standard D 6528, 2007. Standard Test Method for Consolidated Undrained Direct
Simple Shear Testing of Cohesive Soils.
ASTM standard D 718111, 2011. Standard Test Method for Consolidated Drained Triaxial
Compression Test for Soils.
Bjerrum, L., Landva, A., 1966. Direct simple-shear tests on a Norwegian quick clay.
Gotechnique 16 (1), 120.
Boylan, N., Long, M., 2009. Development of a direct simple shear apparatus for peat soils.
Geotech. Test. J. ASTM 32 (2), 126138.
Carro, M., De Amicis, M., Luzi, L., Marzorati, S., 2003. The application of predictive model-
ing techniques to landslides induced by earthquakes: the case study of the 26 Sep-
tember 1997 UmbriaMarche earthquake (Italy). Eng. Geol. 69 (12), 139159.
Chang, W.J., Hong, M.L., 2008. Effects of clay content on liquefaction characteristics of gap-
graded clayey sands. Soils Found. 48 (1), 101114.
Chang, M., Chiu, Y., Lin, S., Ke, T.C., 2005. Preliminary study on the 2003 slope failure in
Woo-wan-chai Area, Mt. Ali Road, Taiwan. Eng. Geol. 80 (12), 93114.
Chang, W.J., Chang, C.W., Zeng, J.K., 2014. Liquefaction characteristics of gap-graded grav-
elly soils in K0 condition. Soil Dyn. Earthq. Eng. 56, 7485.
Fannin, R.J., Eliadorani, A., Wilkinson, J.M.T., 2005. Shear strength of cohesionless soils at
low stress. Gotechnique 55 (6), 467478.
Fleming, R.W., Johnson, A.M., 1994. Landslides in colluvium. U.S. Geol. Surv. Bull. 2059-B,
B1B24.
Fragaszy, R., Su, J., Siddiqi, F., Ho, C., 1992. Modeling strength of sandy gravel. J. Geotech.
Engrg. ASCE 118 (6), 920935.
Frost, R.J., 1973. Some testing experiences and characteristics of bouldergravel lls in
earth dams. ASTM, STP 523, 207233.
Fumagalli, E., 1969. Tests on cohesionless materials for rockll dams. J. Soil. Mech. Found.
Div. ASCE 95 (1), 313332.
Georgiannou, V.N., Burland, J.B., Hight, D.W., 1990. The undrained behaviour of clayey
sands in triaxial compression and extension. Gotechnique 40 (3), 431449.
Holtz, W.G., 1961. Triaxial Shear Characteristics of Clayey Gravel Soils: U.S. Bureau of
Reclamation.
Iannacchione, A., Vallejo, L., 2000. Shear Strength Evaluation of Clay-Rock Mixtures. Slope
Stability 2000. Proceedings of Sessions of Geo-Denver 2000 Denver. Colorado,
American Society of Civil Engineers, pp. 209223.
Ishihara, K., 1996. Soil Behaviour in Earthquake Geotechnics. Oxford Science Publications,
Oxford, UK.
Kjellman, W., 1951. Testing the shear strength of clay in Sweden. Gotechnique 2 (3),
225232.
Kokusho, T., 2000. Correlation of pore-pressure B-value with P-wave velocity and
poisson's ratio for imperfectly saturated sand and gravel. Soils Found. 40 (4),
95102.
Kokusho, T., Hara, T., Hiraoka, R., 2004. Undrained shear strength of granular soils
with different particle gradations. J. Geotech. Geoenviron. Eng. ASCE 130 (6),
621629.
Kuerbis, R., Negussey, D., Vaid, Y.P., 1988. Effect of Gradation and Fines Content on the Un-
drained Response of Sand, Hydraulic Fill Structures, Geotechnical Special Publication
No. 21. ASCE, New York, pp. 330345.
Ladd, C.C., Foott, R., Ishihara, K., Schlosser, F., Poulos, H.G., 1977. Stress Deformation and
Strength Characteristics. Proceedings of the 9th International Conference on Soil Me-
chanics and Foundation Engineering, Tokyo, Japan. pp. 421494.
Lowe, J., 1964. Shear Strength of Coarse Embankment Dam Materials. Proceedings of the
8th International Congress on Large Dams, Edinburgh, UK, pp. 745761.
Marachi, N.D., Chan, C.K., Seed, H.B., 1972. Evaluation of properties of rockll materials.
J. Soil. Mech. Found. Div. ASCE 98 (1), 95114.
Mitchell, J.K., Soga, K., 2005. Fundamentals of Soil Behavior. third ed. Wiley, Hoboken, NJ.
Naeini, S.A., Baziar, M.H., 2004. Effect of nes content on steady-state strength of mixed
and layered samples of a sand. Soil Dyn. Earthq. Eng. 24 (3), 181187.
Roscoe, K.H., 1953. An Apparatus for the Application of Simple Shear to Soil Samples. Pro-
ceedings of the 3rd International Conference on Soil Mechanics and Foundation Engi-
neering, Vol. 1, ICOSOMEF, London, pp. 186191.
Salgado, R., Bandini, P., Karim, A., 2000. Shear strength and stiffness of silty sand.
J. Geotech. Geoenviron. Eng. ASCE 126 (5), 451462.
90 W.-J. Chang, T. Phantachang / Engineering Geology 207 (2016) 7890

Schulz, W.H., 2007. Landslide susceptibility revealed by LIDAR imagery and historical re-
cords, Seattle, Washington. Eng. Geol. 89 (12), 6787.
Shakoor, A., Smithmyer, A.J., 2005. An analysis of storm-induced landslides in colluvial soils
overlying mudrock sequences, southeastern Ohio. USA. Eng. Geol. 78 (34), 257274.
Simoni, A., Houlsby, G., 2006. The direct shear strength and dilatancy of sandgravel mix-
tures. Geotech. Geol. Eng. 24 (3), 523549.
Thevanayagam, S., Martin, G.R., 2002. Liquefaction in silty soilsscreening and remedia-
tion issues. Soil Dyn. Earthq. Eng. 22 (912), 10351042.
Vallejo, L., Mawby, R., 2000. Porosity inuence on the shear strength of granular material
clay mixtures. Eng. Geol. 58 (2), 125136.
Varadarajan, A., Sharma, K.G., Venkatachalam, K., Gupta, A.K., 2003. Testing and modeling
two rockll materials. J. Geotech. Geoenviron. Eng. ASCE 129 (3), 206218.
Zeller, J., Wulliman, R., 1957. The Shear Strength of the Shell Materials for the
Goschenenalp Dam, Switzerland. Proceedings of the 4th Conference on Soil Mechan-
ics and Foundation Engineering, Vol. II, Switzerland, pp. 399404.