International Journal of Geotechnical Engineering

ISSN: 1938-6362 (Print) 1939-7879 (Online) Journal homepage: http://www.tandfonline.com/loi/yjge20

A review on sand sample reconstitution methods

and procedures for undrained simple shear test

Wing Shun Kwan

To cite this article: Wing Shun Kwan (2018): A review on sand sample reconstitution methods and
procedures for undrained simple shear test, International Journal of Geotechnical Engineering, DOI:
10.1080/19386362.2018.1461988

To link to this article: https://doi.org/10.1080/19386362.2018.1461988

Published online: 23 Apr 2018.

Submit your article to this journal

Article views: 9

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=yjge20
International Journal of Geotechnical Engineering, 2018
https://doi.org/10.1080/19386362.2018.1461988

A review on sand sample reconstitution methods and procedures for undrained

simple shear test
Wing Shun Kwan
Department of Civil Engineering, California State University, Los Angeles, CA, USA

ABSTRACT ARTICLE HISTORY

This paper discusses the common methods available for reconstituting sand samples that are used for Received 20 February 2018
soil strength characterization. Reconstituted samples are preferred to undisturbed samples because the Accepted 3 April 2018
current field sampling procedures cannot retrieve granular soil specimens at a reasonable cost. Water
KEYWORDS
sedimentation is generally recognized as the better one than the other common methods such as moist sand; simple shear;
tamping and air pluviation. Most of the existing literature about sand specimen reconstitution is designated liquefaction; laboratory;
for triaxial set-up, and there is no study specifically for simple shear set-up in this topic. Therefore, this paper reconstitution; moist
discusses the differences in reconstituting sand specimens between triaxial and simple shear setups. Two tamping; air pluviation;
reconstitution methods particularly designed for undrained simple shear tests were introduced and tested water sedimentation; sand
through a CSS testing programme. The CSS test data are also compared with the results from published densification
literature, and the strain-based criteria for liquefaction initiation are reviewed.

Introduction Triaxial and simple shear tests

Distinct types of boundary conditions and loading mechanisms
Sand shear strength characterization
(i.e. stress paths) provide different strengths of soil. For example,
Soil strength characterization is important for civil engineering the bearing capacity that can be depicted as a circular slip surface
designs, and typically requires obtaining high-quality samples for underneath an embankment is composed of a combination of
laboratory testing. While undisturbed cohesive soil samples can compression, simple shear, and extension type of loading mecha-
be easily obtained by typical conventional tube sampling meth- nism. Therefore, several types of laboratory testing are designated
ods, it is very difficult to retrieve cohesionless (sand) samples to provide the required stress path. Triaxial testing can simulate
from field since there is no cohesion to hold the particles together the compression and extension loading for a soil element, and
after removing the confining stresses provided by the ground. Direct Simple Shear (DSS) can simulate the shear mechanism.
Moreover, the dynamic behaviour of saturated sand is important Moreover, for assessment of liquefaction resistance, Cyclic DSS is
for evaluating the potential of liquefaction initiation at a given preferred to Cyclic Triaxial test, because the simple shear mech-
sand-deposited site. To acquire and advance the understanding anism allows the principal stress axes to rotate smoothly during
of the fundamentals of sand behaviour and calibrate constitutive cyclic loading, which displays a better simulation of upward
models for numerical analysis, conducting element-level labora- propagating shear waves generated from earthquakes (Boulanger
tory sand tests is essential. et al. 1993). On the other hand, the principal axes instantaneously
Ground freezing methods allow for recovering undisturbed rotate 90 degrees upon loading reversal in Cyclic Triaxial test.
sand samples, but they are extremely expensive. A few site inves- This paper focuses on simple shear testing so only the back-
tigation projects (Robertson et al. 2000; Ghionna and Porcino ground of simple shear testing is mentioned in the following.
2006) were launched in the past by utilizing ground freezing The requirements for simple shear testing are uniform strains
technique to retrieve undisturbed sand samples, but they are very along the height of the specimen, and plane strain (no strains in
rare due to the high cost. Gel-Push method (Taylor, Cubrinovski, the plane perpendicular to the direction of shear loading), which
and Haycock 2012; Umehara et al. 2015) reduces the cost of are both typically achieved by rigid lateral confinement. Latex
retrieving undisturbed sand samples but this technique still membrane reinforced with a spiral winding of wire or Stacked
remains as an emerging technique. Because of the constraints, Rings are commonly used for lateral confinement. (Roscoe
sand specimens are commonly prepared by the reconstitution 1953) has proven that the shear stresses at the specimen’s top
of representative samples of sandy soils collected from fields for and bottom could be non-uniform during the shearing stage. The
the purposes of both practice and research. non-uniformity is severe at the edges and can be explained by the

CONTACT Wing Shun Kwan [email protected]

© 2018 Informa UK Limited, trading as Taylor & Francis Group
2  W. S. KWAN

lack of complimentary shear stresses on the sides. While the shear specimens (before data interpretation), Table 1 summarizes the
stress is horizontally applied from the top or bottom platen, the pros and cons of each method. Figure 1 depicts each method,
smooth inner surface of latex membrane cannot impose a verti- and general procedures for creating a triaxial sand specimen are
cally balanced shear stress at the sides. The imbalance in forces summarized in the following.
creates a tendency for soil specimens to tilt in monotonic loading
or rock in cyclic loading. The rocking problem can be minimized Air pluviation (AP)
by using a larger diameter to height (D/H) ratio of specimen. Sandy soil particles are placed with a funnel that is initially placed
(Amer et al. 1987) compared saturated DSS sand specimens with at the bottom of a split mould. Then, the funnel is slowly raised,
D/H ratios ranging from 3 to 12 and found comparable results. and the sand particles are deposited with very small drop heights
(ASTM D6528-17, 2017) requires simple shear specimens of to form very loose specimens (generally relative density, Dr ~
a D/H ratio of at least 2.5. Many research projects (Boulanger 30% or less). If a denser specimen is desired, vibration can be
et al.1993; Kammerer, Pestana, and Seed 2002; Hazirbaba 2005; applied by tapping the split mould in a symmetrical pattern. The
Rutherford 2012; Kwan 2015) adopt a D/H ratio of four for Cyclic deposition and tapping procedures can be performed up to seven
Simple Shear (CSS) liquefaction research studies. The specimen layers (the number of layer increases with desired density), and
D/H ratio is one of the key differences in preparing triaxial verse typically allows creating sand specimens up to 85% Dr without
simple shear specimens. On the other hand, for triaxial testing, excessive effort of tapping.
(ASTM D4767-11, 2011) requires specimen to have a D/H ratio
of 0.4 to 0.5 to avoid effects of buckling (too tall) and end plates Wet sedimentation (WS)
(too short). This method is generally similar to AP, except that the processes
This paper discusses the techniques for reconstituting sand of deposition and tapping are performed under water. Firstly,
specimens specifically used for simple shear testing. There are place sand particles in a volumetric flask with water; then satu-
many studies on sand reconstitution method for triaxial speci- rate the sand–water solution by either boiling or applying vac-
mens, but one is missing for simple shear set-up. In the author’s uum. The flask is then inverted and lowered to the bottom of the
opinion, there are two reasons to explain this situation. First, split mould and raised up slowly while sand–water mixture is
it is generally believed that reconstituting a simple shear sand syphoned under water. Fine sand that is lighter than water sus-
specimen is not different from a triaxial one since both repre- pend in the volumetric flask to cause separation, and therefore,
sent a soil element. Nevertheless, this research project points out this method only works for reconstituting clean sand. Like the
that the specimen D/H ratio does matter when reconstituting procedures of AP, soil specimens can be tapped along the split
simple shear or triaxial specimens. Second, triaxial apparatus is mould to achieve desired relative density for triaxial specimens.
more commonly available than that for simple shear. Therefore,
previous studies in this subject are predominately investigated Moist tamping (MT)
through triaxial testing. Dry sand is initially mixed with water that represents 3 to 6%
water content of the specimen. Then, the moist coarse-grained
sand is compacted (or tamped) into four to seven layers inside a
Sand specimen reconstitution methods
split mould. Each layer is compacted into a designated portion
Among the many different reconstitution methods available, of the required dry unit weight of the specimen using a method
this paper only discusses three main types: Air Pluviation (AP), of ‘undercompaction’, which considers the factor that the bot-
Wet Sedimentation (WP), and Moist Tamping (MT). These three tom layers also absorb the tamping efforts from the layers above
methods are selected because they are commonly adopted in (Ladd 1978).
both research and practice, and do not require complicated and
custom-made parts such as a vacuum chamber (Kildalen and
Literature review in the effects of different preparation
Stenhamar 1977) and mechanical pluviator (Gade and Dasaka
methods
2016). From the standpoint of reconstituting triaxial sand
Many past experimental research projects show that the stress–
strain behaviour of sand highly depends on its reconstitution
Table 1. Comparison of the three popular sand reconstitution methods.
method. Different methods provide different fabrics of sand (the
Pro Cons orientation of the contacts between sand grains), which is the
Air pluviation • Simple to prepare • Sensitive to drop height primary reason for the observed differences in the stress–stain
(AP) • No fine separation • Large particles migrate to
• Can achieve a wide range of the specimen edge during
curves.
density taping (Mulilis et al. 1977) studied the effects of different prepa-
Water sedi- • Simulate the natural sedi- • Fine separation ration methods through a cyclic triaxial programme on clean
mentation mentation process • Heavy particles sediment
(WS) • Achieve better saturation faster than lighter particles
sand, Monterey Sand No. 0. They studied the effects of different
without an extra stage of CO2 methods (Vibration, Tamping, and Pluviation) at various vibra-
flushing tion frequencies and numbers of layer construction. The authors
Moist tamping • Simple to prepare • Apply high stresses
(MT) • No fine separation • Layer formation
conclude that the following factors are crucial to resistance to
• Good density control sand liquefaction: preparation method, number of layers and
• Can achieve a wide range of frequency of vibrations. The major concern for the WS method
density
is fine separation.
 INTERNATIONAL JOURNAL OF GEOTECHNICAL ENGINEERING  3

Figure 1. Schematic diagram showing sand reconstitution methods for triaxial testing: (a) Air Pluviation (AP); (b) Water Sedimentation (WS); (c) Moist Tamping (MT).

(Vaid, Sivathayalan, and Stedman 1999) compared the stress– (Ghionna and Porcino 2006) documented a study that com-
strain sand behaviour from AP, WS, and MT reconstituted Fraser pares reconstituted sand and gravels samples with undisturbed
sand specimens with undisturbed samples that were retrieved by ones collected by ground freezing methods. The reconstituted
ground-freezing method through simple shear and triaxial test- methods used were AP and WS, and the authors concluded that
ing. The study concluded that WS specimens provide the closest WS closely replicates the in situ fabric of the investigated deposit.
response when compared with undisturbed samples. (Frost and Park 2003) critically assessed the MT method, and
(Høeg, Dyvik, and Sandbækken 2000) compared undisturbed pointed out that the vertical stresses applied by tamping can be
silty sand specimens (Triaxial) with those reconstituted by MT higher than the typical confining stresses in triaxial testing. Also,
and WS methods and concluded that WS method likely provides the bottom layer on top of the rigid bottom platen experiences
the most promising results. The paper also mentions that the greater compaction force than the layer above it.
laboratory results from reconstituted specimens on silt and silty In summary, through the past studies, including a few that
sand should not be used in the design analyses of associated were able to retrieve undisturbed sand samples, reconstituted WS
engineering projects, because the engineering evaluations were sand specimens show closer stress–strain response to natural fab-
changed significantly when results from reconstituted specimens ric than AP and MT methods. Therefore, the experimental study
were used instead of results from undisturbed specimens. described in this paper focuses on WS method. Nevertheless,
(Wood, Yamamuro, and Lade 2008) tested on Nevada 50/200 the results from reconstituted sand specimens should always be
sand containing 10–40% of silts with several types of reconsti- treated with great care because none of them can perfectly reflect
tution methods: AP, WS, Slurry Deposition, and Mixed Dry the natural fabric of a sand deposit.
Deposition. The authors concluded that undrained triaxial tests
performed on specimens with high densities reveal no significant
Undrained cyclic simple shear testing programme
effects on depositional method. Medium densities indicated sig-
nificant differences. Specimens formed by Water Sedimentation An undrained CSS testing programme (Table 2) was set up to
exhibited stable behaviour while those formed by AP in some search for and investigate the optimal ways of reconstituting sand
cases underwent temporary liquefaction. specimens through the WS method and was performed at the
4  W. S. KWAN

Table 2. CSS testing programme for investigations of optimal procedures reconsti- Procedure for reconstituting specimens with loose to
tuting sand specimen through the water sedimentation method at two different
densities.
medium relative density

Test Reconstitution Dr Test Reconstitution Dr Reconstituting uniformly loose specimens in simple shear set-
No. Method (%) CSR No. Method (%) CSR
ups is more challenging than in triaxial set-ups because of the
1 WS 74 0.159 9 WS-M1 73 0.230
larger cross-sectional area of simple shear specimens. To cre-
2 WS 73 0.150 10 WS-M1 71 0.177 ate tall and long triaxial specimens, the sand–water mixture is
3 WS 73 0.149 11 WS-M1 69 0.176 syphoned by raising the inverted volumetric flask along the axis
4 WS 65 0.174 12 WS-M2 65 0.228
5 WSS 49 0.176 13 WS-M2 65 0.176
of symmetry of the centre of specimen. This procedure can prom-
6 WSS 39 0.200 14 WS-M2 81 0.125 isingly reconstitute a specimen with uniform density throughout
7 WSS 41 0.226 15 WS-M3 76 0.259 its length. However, for short and wide simple shear specimens,
8 WSS 44 0.101 16 WS-M3 74 0.207
17 WS-M3 77 0.202
the deposition process requires going around the specimen area
18 WS-M3 82 0.152 in circles while raising up the inverted volumetric flask, and
19 WS-M3 73 0.125 therefore it is harder to maintain a constant drop height nec-
20 WS-M4 85 0.152
21 WS-M4 85 0.202
essarily to provide uniformity. To address this shortcoming, an
23 WS-M4 79 0.304 additional screen procedure (modified after Kammerer, Pestana,
24 WS-M4 78 0.350 and Seed 2002) can improve the sand–water mixture syphon-
ing process. A screen with a diameter slightly less than the soil
specimen of which the opening is slightly larger than the size
100 of the largest grain was placed at the bottom of the split mould
90 UnWashed under water prior to placement of the sand. After syphoning the
Washed
saturated sand, the screen was then pulled up slowly to drain the
80
sand particles through the screen opening with a constant drop
70 Gs = 2.67 height (Figure 3).
3
Percent Passing (%)

60 γd, min = 15.14 kN/m To investigate into the effects of an additional screening pro-
3
50 γd, max = 17.09 kN/m cedure, three specimens were reconstituted with typical WS
emax = 0.76 procedure, and compared with four CSS specimens that went
40
emin = 0.56
through an additional screening procedure after sand–water
30 USCS = SP-SM (Unwashed)
USCS = SP (Washed) syphoning (hereby WSS). All seven tests (with and without
20
additional screening) were subjected to various undrained cyclic
10 loading. For the tested Nevada sand and syphoning at almost
0 zero drop height, the typical WS procedure reconstituted sand
10 1 0.1 0.01
Grain Size (mm)
specimens at a Dr range of 60 to 70%, while WSS produced spec-
imens of a Dr range of 40%. Figure 4 summarizes the test results
Figure 2. Grain size distribution curves and sand properties for tested Nevada in a semi-log plot of CSR vs. Nf, where liquefaction initiation is
Sand, before and after pluviation (Kwan 2015). defined as the excess pore pressure ratio equal to unity (excess
pore pressure equals to σ’vo). Even with a lower Dr range, the
WSS specimens show stronger resistance to liquefaction than
University of Texas at Austin using the modified Geotechnical WS specimens, which implies that WS specimens’ density was
Consulting and Testing System (GCTS) manufactured CSS non-uniform. Since the syphoning process requires going around
apparatus. The WS procedures discussed above are adopted for in a circular pattern to cover the large area, it gives a high poten-
reconstituting triaxial specimens. The experimental research pro- tial on syphoning under inconsistence drop heights. This renders
gramme introduced here focuses on undrained simple shear con- a global density that falsely represents the soil specimen. Yet,
figuration. Liquefaction resistance curves (Cyclic Stress Ratio, this screen procedure is not ideal for triaxial specimen because
CSR vs. No. of cycle reaching liquefaction initiation, Nf) are used of the smaller D/H ratio.
to assess the performance of proposed reconstitution procedures.
CSR is defined as the applied amplitude of applied shear stress
Procedure for reconstituting specimens with dense to very
normalized by the initial vertical initial effective stress, σ’vo. An
dense relative density
ideal reconstitution procedure should able to generate consistent
liquefaction resistance curves reflecting the effect of soil density. Densification of sand is usually achieved by some forms of
For this testing programme, Nevada sand was tested, and the vibration, and the most typical method in reconstituting triax-
sand properties are summarized in Figure 1. The Nevada sand ial specimens is by tapping along the side of split mould (longer
used in this testing programme is a fine, uniform (Cu = 2), and dimension). Tapping must be performed at the larger dimension
angular sand with the mean grain size of 0.2 mm (Figure 2). All (i.e. along the specimen height for triaxial specimen; and along
reconstituted specimens were about 1-inch high and of 4-inch diameter face for simple shear specimen) so that the vibration
diameter and consolidated to 100 kPa vertical effective stress provided by tapping can penetrate through the entire specimen.
before subjected to undrained cyclic loadings. More details about Otherwise, only the portion of sand near the edge is densified, but
the testing apparatus can be found in (Kwan 2015; Kwan et al. the centre part remains loose. For creating WS specimens, drain-
2017). age is also important while vibration is being applied. Drainage is
 INTERNATIONAL JOURNAL OF GEOTECHNICAL ENGINEERING  5

500mL Volumetric Flask

Pulling up the screen
Zero Fall Height

Bottom Platen

Figure 3. Picture on the left shows the soil syphoning; picture on the right shows the screen procedure afterward. The soil particles rain through the screen (Kwan 2015).
This method allows to reconstitute simple shear sand specimen with relative density of loose to medium.

0.26 (horizontal direction) does not allow the vibration to penetrate

0.24
D r = 39-49% (WSS)
the four-inch specimens thoroughly and uniformly. Method 3
with screen
D r = 65-74% (WS) provides vibration vertically (i.e. shorter direction) using an
0.22
without screen inverted small vibratory table, and meanwhile, allows excess
0.2 pore pressure to dissipate through a simple custom-made plate,
0.18 which contains small holes for drainage as shown in Figure 5.
Method 4 is very similar to Method 3 but requires two layers of
0.16
reconstitution. After syphoning half of the designated soil mass
0.14 by the WS procedures, vertical vibration is applied as in Method
0.12
3. The same procedure is repeated for the second layer.
Figure 6 shows the CSS test results from all different densi-
0.1
fication methods (Dr = 65 to 85%) and are compared with the
0.08 results of those with relative density of loose to medium speci-
10 0 10 1 10 2 10 3
mens (reconstituted by WSS method, Dr = 39 to 49%). The test
results show that Method 4, ‘surface vibration with two layers’,
provides a significantly higher resistance than the loose sand
Figure 4. Illustration of ‘Surface vibration with two layers (WS-MP4)’ preparation
procedure (Kwan 2015). This method allows to reconstitute simple shear sand curve, which is an indication that it effectively densifies the sand.
specimen with relative density of dense to very dense. On the other hand, the other three methods produce specimens
with only a slightly higher resistance than the loose samples,
which indicates that the applied vibration could not uniformly
needed to dissipate the excess pore pressure induced by vibration,
penetrate the whole sample.
so the sand particles can rearrange into denser configurations.
To vibrate (tap) a simple shear specimen reconstituted by WS
method, one feasible way is to tap the underside face of bottom Data analysis and discussion
platen after syphoning sand–water mixture and before placing
Liquefaction initiation criteria
the top platen (Kammerer, Pestana, and Seed 2002). In such way,
the applied vibration can penetrate the specimens through the Liquefaction initiation is commonly defined as the generated
shorter dimension and induced excess pore pressure can be dis- excess pore pressure (Δu) is equal to (100% of) the initial vertical
sipated through the open face at the top of specimen. However, effective stress 𝜎vo
′
. Δu can also be expressed as the excess pore
this method only works if the underside face of the bottom platen pressure ratio (ru), which is a ratio of Δu∕𝜎vo�
. This criterion is
is accessible. For some simple shear apparatus, the bottom platen theoretically sound because when ru is very close to unity, the
required to be mounted on the shaking table before pluviation, shear strength of the sand becomes almost zero and liquefaction
and therefore tapping cannot be applied. is initiated causing ground failure that is associated with lateral
The CSS apparatus used in this study does not allow access spreading and settlement. However, some past studies show that
to the underside face of the bottom platen after sand syphoning. liquefaction can be initiated at lower ru values for some circum-
The experimental programme investigated into four different stances. (Ishihara 1993) reports that the highest achievable ru
densification methods following typical WS procedure. Table 3 values could be levelled out at 0.9 for sandy silts and silty sands.
describes the methods and comments on the performance of (Kammerer er al., 2004) shows that for simulation of sloping
each method. Method 1 utilizes the shear actuator to provide ground condition (Kα effect) using bi-directional simple shear
vibrations; however, there is a lack of channel for excess pore loadings, the maximum ru value can be limited to 0.7 or less,
pressure dissipation. Method 2 densifies the sand specimen by because the superposed static shear stress suppresses the genera-
tapping the side of split mould (like the tapping procedure of tion of γ while excessive deformations develop. Moreover, a ru of
triaxial specimens); however, tapping the longer dimension 0.95 or greater may not be achievable in very dense sand deposit.
6  W. S. KWAN

Shear strain (γ) can also be served as a criteria for accessing

tion during vibration. However, this set-up was found to be ineffective, because
specimen was heavily disturbed at the circular edge, but remained undensified
‘undrained’ condition. Since the pore pressure cannot be dissipated, this meth-

The ½″ layers allowed vibration to penetrate through the entire thickness, when
liquefaction initiation because it closely relates to pore pressure

The drilled holes on the plastic plate were designed for pore pressures dissipa-
The vibration cannot penetrate through the longer dimension (4″ Dia). So, the
generation (Ladd et al. 1989). Wu et al. (2004) recommends a

With the applied top platen, the sand specimen was vibrating almost at an

∆u dissipated through the holes. This method was found to be effective

single (SA) or double amplitude (DA) shear strain of 6% (here
by γDA = 6%) as the failure criteria of the initiation of flow-type
deformation after examination from an uni-directional CSS
study on Monterey #0/30 sand. (Ishihara 1993) recommends

the vibration could not penetrate the one-inch thickness

using 5% DA axial strain (εa) after studying many results from
cyclic triaxial tests, which correspond to 3.75% SA (hereby
Comments/Remarks

γSA = 3.75%) in simple shear set-up because γ = 1.5εa. This study

utilizes the CSS test data from this study and compares the per-
formances of the two liquefaction triggering criteria. An ideal
od cannot densify the sand efficiently

criterion should provide liquefaction resistance curves at both

loose and dense densities that are very close to the ones that are
provided by a high ru range such as 0.8 to 1.0. Figure 7 shows the
liquefaction resistance curves (WSS and WS-M4) determined
from both criteria of γDA = 6% and γSA = 3.75% and compared
with liquefaction resistance curves determined from ru = 0.8, 0.9,
at the centre

and 1.0. Figure 7 provides the following insights:

(1) The comparison shows that the two shear strain crite-
ria provide liquefaction resistance slightly less than the
ones determined from ru = 1.0 and are generally within
device was placed on the top sand surface, and an inverted vibratory table was
After depositing sand under water, vibration was applied by tapping the side of

drainage holes, the vibratory table was removed. The escaped sand was dried
turned on and attached to the tip of metal thread, to provide vibration to the

the ranges of ru = 0.8 to 1.0, which validates the two

The method is like the one above, except reconstituting the soil specimen with
was applied (before the split mould was dissembled), a dead weight (5.5 kg)
was added on top of it. The specimen was then vibrated by applying a 2 Hz,
strain-controlled vibration of 2 mm peak-to-peak amplitude from the shear
First, the sand was syphoned into the mould/membrane. After the top platen

soil specimen. After the water–sand mixture escaped from the plastic plate
A custom device (Figure 5) was created for applying uniform vibration to the

plastic plate. Holes were drilled on the plastic plate to allow drainage. The

liquefaction initiation criteria.

sand surface. The device included a metal thread attached to a 3.96″ Dia

(2) The tail part of the dense liquefaction resistance curves

(high density and small CSR values) show that the two
shear strain criteria provide liquefaction resistances
greater than the one determined from ru = 1.0. At high
two layers. Figure 5 illustrates this preparation method

density, pore pressure is generated with limited strain

(cyclic mobility), and this effect is especially mani-
fested under smaller loading amplitude. Therefore, for
Procedures

the combination of small CSR values and high density,

Table 3. Different methods of preparing CSS water-pluviated dense to very dense specimens, Kwan (2015).

the threshold values for liquefaction initiation should

the slit mould with a rubber hammer

be decreased and further investigation is needed.

(3) The liquefaction resistance curves provided by
γSA = 3.75% are slightly higher than those provided by
γDA = 6%.
and weighed

CSS test data comparison and discussion

actuator

The undrained CSS test results obtained from the optimal recon-
stitution methods (WSS and WS-M4) that are described above
are compared with other published undrained CSS test results.
[M1] Vibration with the shear actuator under a normal load

Since there are no available test result acquired under the exact
testing conditions as this study, three past studies (Arulmoli
et al., 1992, Kammerer et al. 2000, and Kammerer et al. 2004)
with similar testing conditions are selected for comparison.
These three studies have the same conditions of no initial static
[M2] Tapping on the side of split mould

shear stress (i.e. kα = 0), no pre-shearing effect, and under uni-

[M4] Surface vibration with two layers
[M3] Surface vibration with one layer

directional loading. The stress level of selected tests from the other
research projects is similar, about 100 kPa. However, Arulmoli
et al. (1992) and Kammerer et al. (2000) adopted AP as the spec-
imen reconstitution method, and Kammerer et al. (2004) tested
on Monterey #0/30 sand. Figure 8 compares the results from
the four projects (the author’s study and other comparable three
projects) and summarizes the testing conditions.
Methods

For loose to medium density tests (Dr = 30 to 50%), the recon-

stitution method has the most significant effect on liquefaction
 INTERNATIONAL JOURNAL OF GEOTECHNICAL ENGINEERING  7

vibration
vibration

Plastic Plate used for soil

Second lift Pluviation (12.7 mm) specimen densification,
First lift Pluviation (12.7 mm) 25.4 mm small holes were created for
First lift Pluviation (12.7 mm)
drainage.

101.6 mm

Figure 5. Liquefaction resistance curves from the sand specimens reconstituted by the methods of Water Sedimentation (WS) and Water Sedimentation with Screen
(WSS).

0.4
Loose D = 39-49%
r
0.35 Method 1 D r = 65-81%
Method 2 D r = 69-73%
0.3 Method 3 D r = 73-76%
Method 4 D r = 78-85%
0.25

0.2

0.15

0.1

0.05

0
10 0 10 1 10 2

Figure 6. Comparison of four different densification methods for reconstituting

water-pluviated sand specimens for Simple Shear testing. The green curve
represents test results from specimens reconstituted by water sedimentation with
screen (WSS). Descriptions and procedures of each method are listed in Table 3. Figure 7. Liquefaction resistance curves obtain from the two initiation criteria:
γDA = 6% and γSA = 3.75%, and compared with establishments based on ru of 0.8,
0.9 and 1.0. The CSS data are from specimens that are reconstituted by the WSS
resistances. The tests reconstituted with WSS method (Kammerer and WS-M4 methods.
et al. 2004 and Kwan 2015) provide higher liquefaction resistance
than those by AP (Arulmoli el al., 1992). Typical WS method cre- density configurations because the sand particles are packed
ates soil fabric that is more stable than those reconstituted by AP much closer to each other. Wood, Yamamuro, and Lade (2008)
method (Wood, Yamamuro, and Lade 2008), and (Mulilis et al. also reports that dense specimens show no significant effect of
1977; Wood, Yamamuro, and Lade 2008) showed experimental reconstitution method. However, significant differences in terms
results that WS specimens provide higher liquefaction resistance of liquefaction resistance are found in both Kwan (2015) and
than AP specimens. Despite the differences in sand type (Nevada Kammerer et al. (2004), which adopted similar WS procedures
sand vs. Monterey #0/30 sand), test results from Kwan (2015) (WS-M4 and WS with tapping at the bottom face) and a similar
agree very well with those from Kammerer et al. (2004). At loose vertical effective stress (100 vs. 80 kPa). The key different testing
to medium density, reconstitution method provides significant condition between the two studies is the sand type (Nevada vs.
effects in liquefaction resistance, while the effect of mean grain Monterey #0/30). While both sand types are very uniform (Cu = 2
size (D50) could be minor. and 1.29), Nevada sand is angular with D50 = 0.2 mm whereas
For dense to very dense density tests (Dr > 75%), the tests Monterey #0/30 sand is rounded with D50 = 0.36 mm. (Seed
reconstituted with WS-M4 method (Kwan 2015) agree very well & Idriss, 1971) also reports that coarser sand provides higher
with the test with the same sand type (Nevada sand) but was liquefaction resistance than finer sand. At dense configuration,
reconstituted by AP method. It implies that the differences in D50 provides significant effect in liquefaction resistance, while
fabric created by AP vs. WS have an insignificant effect in denser the method of reconstitution could be minor.
8  W. S. KWAN

0.5

0.4

0.3
CSR

0.2

0.1

0
1 10 100 1000
Nf

Figure 8. CSS data from reconstitution methods of WSS and WS-M4 compared with other published CSS data (Arulmoli et al.1992 ; Kammerer et al. 2000; and Kammerer
et al. 2004). The liquefaction initiation criteria for Kammerer et al. (2004) and Kammerer et al. (2000) is γDA = 6%, and Kwan (2015) is ru = 1.0.

Summary and conclusion loading is small (CSR < 0.2) and relative density is high (dense
to very dense).
Sand specimens are commonly reconstituted in geotechnical
laboratories instead of retrieving undisturbed ones. There are
a few popular methods available, including (Air Pluviation,
Water Sedimentation, and Moist Tamping), and they are briefly Acknowledgements
reviewed in this paper. Past studies have shown that the water This testing program was conducted during the author’s PhD study at the
sedimentation methods provide the closest simulation to results University of Texas at Austin (UT Austin), and supported by the National
from undisturbed granular sand specimens retrieved from Science Foundation, NEES Research programme, under grant number
0936408 and the Department of Civil, Architectural and Environmental
ground freezing techniques. Nevertheless, the typical reconsti-
Engineering at UT Austin. This support is gratefully acknowledged.
tution methods are set up for Triaxial specimens, and there is no
evaluation on these procedures when applying to simple shear
specimens, which have a very different diameter to height ratio. Disclosure statement
Through an undrained CSS testing programme, two opti- No potential conflict of interest was reported by the author.
mum reconstitution procedures for sand specimens, ‘Water
Sedimentation with Screen (WSS)’, and ‘Vertical Surface
Vibration with Two Layers (WS-M4)’ are established, and the Funding
results are compared with other published CSS data that have This testing program was conducted during the author’s PhD study
similar testing conditions. The results show general agreements, at the University of Texas at Austin (UT Austin), and supported by the
National Science Foundation, NEES Research programme [grant number
and the discrepancies are due to different testing conditions. 0936408] and the Department of Civil, Architectural and Environmental
Reconstitution method is the major factor that affects liquefac- Engineering at UT Austin.
tion resistance for low density tests, and mean grain size is the
major factor for high density tests. Moreover, the performance of
two popular strain-based liquefaction initiation criteria, 6% shear Notes on Contributor
strain double amplitude and 3.75% shear strain single amplitude Wing Shun Kwan is an assistant professor of Civil Engineering,
were evaluated. In terms of liquefaction resistance curve, the two California State University, Los Angeles and formerly a gradu-
criteria generally agree with those developed from high excess ate research assistant of Civil, Architectural and Eviornmental
pore pressure (ru = 0.8 to 1.0), except that when the amplitude Engineeirng, the University of Texas at Austin.
 INTERNATIONAL JOURNAL OF GEOTECHNICAL ENGINEERING  9

References Kildalen, S., and P. Stenhamar (1977). “NGI Laboratory Rainer.” NGI
Internal Report 51505-15 (Unpulished), Oslo, Norway.
Amer, M. I., W. D. Kovacs, and M. S. Aggour. 1987. “Cyclic Simple Shear Kwan, W. S. (2015). “Laboratory Investigation into Evaluation of Sand
Size Effect Correction Factors.” Journal of Geotechnical Engineering 113 Liquefaction under Transient Loadings.” The University of Texas at
(7): 708–718. Austin.
Arulmoli, K., K. Muraleetharan, M. M. Hossain, and L. S. Fruth. 1992. Kwan, W. S., S. S. Sideras, S. L. Kramer, and C. El Mohtar. 2017.
VELACS: Verification of Liquefaction Analyses by Centrifuge Studies: “Experimental Database of Cyclic Simple Shear Tests under Transient
Laboratory Testing Program. Soil Data Rep. Project No. 90-0562. Irvine, Loadings.” Earthquake Spectra 33 (3): 1219–1239.
CA: The Earth Technology Corporation. Ladd, R. S. 1978. “Preparing Test Specimens Using Undercompaction.”
ASTM D4767-11. 2011. Standard Test Method for Consolidated Undrained Geotechnical Testing Journal 1 (1): 16–23.
Triaxial Compression Test for Cohesive Soils.West Conshohocken, PA: Ladd, R. S., R. Dobry, P. Dutko, F. Y. Yokel, and R. M. Chung. 1989. “Pore-
ASTM International. Accessed from: www.astm.org Water Pressure Buildup in Clean Sands because of Cyclic Straining.”
ASTM D6528-17. 2017. Standard Test Method for Consolidated Undrained Geotechnical Testing Journal 12 (1): 77–86.
Direct Simple Shear Testing of Fine Grain Soils. West Conshohocken, Mulilis, J. P., H. B. Seed, C. K. Chan, J. K. Mitchell, and K. Arulanandan.
PA: ASTM International. Accessed from: www.astm.org 1977. “Effects of Sample Preparation on Sand Liquefaction.” Journal of
Boulanger, R. W., C. K. Chan, H. Bolton Seed, R. B. Seed, and J. B. Geotechnical Engineering Division 103 (GT2): 91–108.
Sousa. 1993. “A Low-Compliance Bi-Directional Cyclic Simple Shear Robertson, P. K., C. E. (Fear) Wride, B. R. List, U. Atukorala, K. W. Biggar,
Apparatus.” Geotechnical Testing Journal 16 (1): 36–45. P. M. Byrne, R. G. Campanella, et al. 2000. “The CANLEX Project:
Frost, J. D., and J. Y. Park. 2003. “A Critical Assessment of the Moist Summary and Conclusions.” Canadian Geotechnical Journal 37 (3):
Tamping Technique.” Geotechnical Testing Journal 26 (1): 57–70. 563–591.
Gade, V. K., and S. M. Dasaka. 2016. “Development of a Mechanized Roscoe, K. (1953). “An Apparatus for the Application of Simple Shear to
Traveling Pluviator to Prepare Reconstituted Uniform Sand Specimens.” Soil Samples.” 3rd International Conference on Soil Mechanics and
Journal of Materials in Civil Engineering 28 (2): 1–9. Foundation Engineering, 186–191.
Ghionna, V. N., and D. Porcino. 2006. “Liquefaction Resistance of Rutherford, C. J. 2012. “Development of a Multi-Directional Direct Simple
Undisturbed and Reconstituted Samples of a Natural Coarse Sand Shear Testing Device for Characterization of the Cyclic Shear Response
from Undrained Cyclic Triaxial Tests.” Journal of Geotechnical and of Marine Clays.” PhD diss., Texas A&M University.
Geoenvironmental Engineering 132 (2): 194–202. Seed, H. B., and I. M. Idriss. 1971. “Simplified Procedure for Evaluating
Hazirbaba, K. 2005. “Pore Pressure Generation Characteristics of Sands Soil Liquefaction Potential.” Journal of Soil Mechanics & Foundations
and Silty Sands: A Strain Approach.” PhD diss., The University of Texas Division 97 (9): 1249–1273.
at Austin. Taylor, M. L., M. Cubrinovski, and I. Haycock (2012). “Application of New ‘
Høeg, K., R. Dyvik, and G. Sandbækken. 2000. “Strength of Undisturbed Gel-Push ‘ Sampling Procedure to Obtain High Quality Laboratory Test
versus Reconstituted Silt and Silty Sand Specimens.” Journal of Data for Advanced Geotechnical Analyses.” New Zealand Society for
Geotechnical and Geoenvironmental Engineering 126 (7): 606–617. Earthquake Engineering Conference.
Ishihara, K. 1993. “Liquefaction and Flow Failure during Earthquakes.” Umehara, Y., G., Chiaro, T. Kiyota, Y. Hosono, Y. Yagiura, and H. Chiba
Géotechnique 43 (3): 351–451. (2015). “Effectiveness of ‘Gel-Push’ Sampling Technique to Retrieve
Kammerer, A., J. Wu, J. Pestana, M. Riemer, and R. Seed. 2000. Cyclic Undisturbed Sandy Specimens for Liquefaction Test.” 6th International
Simple Shear Testing of Nevada Sand for PEER Center: Project 2051999. Conference on Earthquake Geotechnical Engineering.
Geotechnical Engineering Research Rep. UCB/GT/00-01. Berkeley, CA: Vaid, Y. P., S. Sivathayalan, and D. Stedman. 1999. “Influence of Specimen-
Univ. of California at Berkeley. Reconstituting Method on the Undrained Response of Sand.”
Kammerer, A. M., J. M. Pestana, and R. B. Seed (2002). Undrained Geotechnical Testing Journal 22 (3): 187–195.
Response of Monterey 0/30 Sand under Multidirectional Cyclic Simple Wood, F. M., J. A. Yamamuro, and P. V. Lade. 2008. “Effect of Depositional
Shear Loading Conditions. Report No UCB/GT/02-01, Geotechnical Method on the Undrained Response of Silty Sand.” Canadian
Engineering. Geotechnical Journal 45: 1525–1537.
Kammerer, A., J. Wu, M. F. Riemer, and J. Pestana (2004). “Pore Pressure Wu, J., A. M. Kammerer, M. F. Riemer, R. B. Seed, and J. M. Pestana (2004).
Development in Liquefiable Soils under Bi-Directional Loading “Laboratory Study of Liquefaction Triggering Criteria.” 13th World
Conditions.” Proceedings of the 11th International Conference on Conference on Earthquake Engineering, Vancouver, Paper No. 2580.
Soil Dynamics and Earthquake Engineering and the 3rd International
Conference on Earthquake Geotechnical Engineering, 2(1980), 697.