You are on page 1of 9

Effect of Sampling Disturbance on Properties

of Singapore Clay
Thiam-Soon Tan1; Fook-Hou Lee2; Poh-Ting Chong3; and Hiroyuki Tanaka4

Abstract: This paper examines results from triaxial unconfined compression tests and undrained compression tests on reconsolidated
samples of a Singapore marine clay retrieved using two sampling methods that offer differing quality of samples. Both local internal strain
Downloaded from ascelibrary.org by New York University on 05/12/15. Copyright ASCE. For personal use only; all rights reserved.

measurements using a Hall-effect transducer and external strain measurements using LVDTs were employed in the triaxial tests. Bender
elements were embedded in some of the samples to establish the maximum shear modulus. If the samples are not reconsolidated, the shear
strength and stiffness determined from triaxial tests are found to be sensitive to the quality of the samples, and generally lower than that
determined by in situ tests. However, if the samples are subjected to isotropic or K 0 consolidation to the estimated in situ condition, there
is little difference between the shear strengths of samples retrieved using different samplers, and also consistent with results from vane
shear tests. However, for the maximum shear modulus, even with reconsolidation, there is still a 10% difference between the results from
samples retrieved using different samplers. Further, the laboratory determined maximum shear moduli are about 10% lower than the value
determined in an in situ seismic cone test.
DOI: 10.1061/共ASCE兲1090-0241共2002兲128:11共898兲
CE Database keywords: Sampling; Sample disturbance; Shear strength; Clays; Singapore.

Introduction 1984; Graham et al. 1987兲. Sampling disturbance is attributed to


two components, damage to the soil structure and a loss of effec-
In situ tests are normally used to provide information of soil at the tive stress.
in situ stress condition, with minimal disturbance to the ground. Kirkpatrick and Khan 共1984兲 and Graham et al. 共1987兲
However, as the condition around an in situ probe tends to be showed that reconsolidating a sample back to its in situ condition
nonuniform, an in situ test, therefore, measures an integrated re- could largely compensate for the effect of stress release on
sponse of the ground, and soil parameters are determined based strength, especially if the sample was stored in an ‘‘undrained’’
on empirical or semiempirical correlations. On the other hand, condition. Other researchers have suggested similar reconsolida-
stress and strain conditions are better controlled in laboratory tion procedures to determine the ‘‘in situ’’ strength 共Davis and
tests, and these are often used to study the stress–strain behavior Poulos 1971兲.
of soil elements under different strain and stress paths. However, on the effect of sampling disturbance on the stress–
Disturbance to a soil specimen is inevitable in the sampling strain behavior, there is some variance in the findings to date. For
and preparation processes. Such disturbances to the structure and instance, Kirkpatrick and Khan 共1984兲 reported that stress–strain
in situ stress state of soil may result in a sample exhibiting be- behaviors of reconsolidated samples are similar to those that were
havior in the laboratory that is different from its in situ behavior. not subjected to any stress release. On the other hand, Graham
The effects of sampling disturbance in tube-sampled soils have et al. 共1987兲 reported that Young modulus E 50 of reconsolidated
been extensively studied 共for example, Kirkpatrick and Khan samples is only about half that of control samples that were not
subjected to any stress release. The strain at failure is also con-
1 siderably different. In the two papers quoted above, stress-release
Associate Professor, Centre for Soft Ground Engineering, Dept. of
Civil Engineering, National Univ. of Singapore, Kent Ridge Crescent, disturbance is induced on reconstituted samples prepared in the
Singapore 119260. E-mail: cvetants@nus.edu.sg laboratory. Thus, their findings may not necessarily apply to in
2
Associate Professor, Centre for Soft Ground Engineering, Dept. of situ soils, in which structure often plays an important role.
Civil Engineering, National Univ. of Singapore, Kent Ridge Crescent,
According to Hight 共1985兲, as sampling modified the effective
Singapore 119260.
3
Research Engineer, Centre for Soft Ground Engineering, Dept. of stress state in an element of soil, important differences in stress–
Civil Engineering, National Univ. of Singapore, Kent Ridge Crescent, strain characteristics and peak strength could occur. Results on
Singapore 119260. young sedimented clays showed that due to sampling, the peak
4
Chief, Geotechnical Survey Laboratory, Port and Harbour Research undrained shear strength under triaxial compression may be re-
Institute, Japan. duced by as much as 30% but the ultimate strength is little af-
Note. Discussion open until April 1, 2003. Separate discussions must fected. Clayton et al. 共1992兲, using simulated strain paths on
be submitted for individual papers. To extend the closing date by one Bothkennar clay, concluded that the maximum shear modulus,
month, a written request must be filed with the ASCE Managing Editor.
usually defined as the shear modulus measured at strain level less
The manuscript for this paper was submitted for review and possible
publication on May 8, 2000; approved on February 6, 2002. This paper is than 0.001%, could not be fully recovered by reconsolidation.
part of the Journal of Geotechnical and Geoenvironmental Engineer- Many of the studies to date were on reconstituted samples,
ing, Vol. 128, No. 11, November 1, 2002. ©ASCE, ISSN 1090-0241/ with the principal focus on shear strength. There were fewer stud-
2002/11-898 –906/$8.00⫹$.50 per page. ies on shear strength of in situ clays, and little data on effect of-

898 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / NOVEMBER 2002

J. Geotech. Geoenviron. Eng. 2002.128:898-906.


Downloaded from ascelibrary.org by New York University on 05/12/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. Soil profile at test site

disturbance on shear modulus at small strain, an increasingly im- Marine clay is the main constituent of this formation. Fig. 1
portant issue in urban geotechnical engineering. This paper exam- shows the soil profile at the site where the current study is con-
ines the effects of tube sampling on the undrained shear strength ducted. The marine clay is deposited in two layers, typically re-
and shear modulus of an in situ marine clay found in Singapore, ferred to as the Upper and Lower Marine Clay, separated by a
typically referred to as the Lower Marine Clay. In this study, stiffer intermediate layer, widely considered to be the desiccated
results from laboratory tests on samples collected using two dif- crust of the Lower Marine Clay. The marine clay formation varies
ferent sampling methods will be compared to each other and also in thickness; it is usually between 10 and 15 m near the estuaries,
to results from some in situ tests. but in some instances, it can be as thick as 40 m. The basic soil
The Singapore Lower Marine Clay is a member of the Kallang properties of the Singapore Lower Marine Clay are given in Table
Formation, which covers much of the coastal plain and immediate 1 and show that it is a high plasticity, inorganic clay of medium to
offshore zone in Singapore. It is a recent Holocene deposit and low sensitivity, with a friction angle much lower than many other
consists of soil of marine, alluvial, littoral, and estuarine origins. soft clays.

Table 1. Comparison of Basic Properties of Singapore Lower Marine Clay with other Clays
Singapore Lower Marine Clay Ariake Bothkennar
Liquid limit, LL 共%兲 63– 80 107–122 40– 85

Plastic limit, PL 共%兲 22–24 45–50 25– 40

Sensitivity, S r 3– 6 5–16 5–13

Activity, A c 0.9 1.2–2.0 0.25–0.75

Specific gravity, G s 2.62–2.78 2.62–2.64 2.60–2.70

Mineralogy Illite, kaolinite, smectite, and quartz Smectite, illite, kaolinite, and chlorite Illite, kaolinite, quartz, and feldpar

Compression index, C c 0.72–1.01 1.0–1.5 0.75–1.40

Swelling index, C s 0.15–0.22 ¯ ¯

K 0 ⫽␴ ho /␴ v o 0.57–0.62 0.50–0.54 0.40–0.70

OCR 1.30–1.45 1.28 –1.60 1.40–1.60

Friction angle, ␾ 22–25 35– 40 34

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / NOVEMBER 2002 / 899

J. Geotech. Geoenviron. Eng. 2002.128:898-906.


Table 2. Type and Schedule of Tests Conducted in Laboratory
Type of test Type of loading and sample’s dimension Number of tests Remark
Consolidation Load ratio⫽1 19 With and without lateral
NUS-(H⫽20 mm, D⫽50 mm) pressure measurement

Consolidation Load ratio⫽1 4 With bender element


NUS-(H⫽20 mm, D⫽50 mm)

Constant rate of strain Strain Rate⫽0.02%/min 17


PHRI-(H⫽25 mm, D⫽60 mm)

UCT Shear rate⫽1%/min 21 With local strain measurement


NUS-(H⫽76 mm, D⫽38 mm)
PHRI-(H⫽80 mm, D⫽35 mm)
Downloaded from ascelibrary.org by New York University on 05/12/15. Copyright ASCE. For personal use only; all rights reserved.

CIU Shear rate⫽0.1%/min 10 With local strain measurement


NUS-(H⫽76 mm, D⫽38 mm)
PHRI-(H⫽80 mm, D⫽35 mm)

CK0 U 共compression and extension兲 Shear rate⫽0.1%/min 5 With local strain and Bender
NUS-(H⫽76 mm, D⫽38 mm) element measurement
PHRI-(H⫽80 mm, D⫽35 mm)
Note: D⫽diameter of the soil sample; H⫽height of the soil sample.

Test Program Laboratory Tests

Triaxial tests in this study were conducted at two institutions. At


In situ Tests and Sampling the National Univ. of Singapore, soil samples with 76 mm height
The site where the clay samples were retrieved is located at the and 38 mm diam were used, while at the Port and Harbour Re-
southern coast of Singapore and is the site for the Singapore search Institute, Japan, samples with 80 mm height and 35 mm
National Arts Center. The Lower Marine Clay samples were taken diam were used. Only the central portion of the extruded sample
from a depth of about 18 m below ground surface, between PH-2 was tested. A digital controller that has a minimum increment of
and PH-3, where the thickness of the stratum is about 10 m, as 0.1 ␮m is used to actuate the axial loading on the sample, housed
shown in Fig. 1. Since the water table is located at a depth of 2.5 within a Bishop-Wesley triaxial cell. For a 76-mm-long triaxial
m, this clay is likely to be saturated. Undisturbed samples were sample, this is equivalent to a minimum strain increment of 1.3
retrieved using a Japanese thin-walled piston sampling tube ⫻10⫺6 . Stress-controlled ram loading, and cell and pore pres-
共JSSMFE 1977兲 and the Shelby tube 共ASTM D1587 1994兲. The sures are applied using pneumatic regulators with a minimum
Japanese thin-walled piston sampling tube has an internal diam- pressure increment of about 0.01 kPa.
eter of 75 mm, an area ratio of 7.5%, inside clearance ratio of 0%, A pair of bender elements were embedded into the opposing
inside cutting-edge angle of 0°, outside cutting-edge angle of 6 ends of each sample and wired in a transmitter–receiver configu-
⫾1°, and a diameter to thickness ratio of 52. The Shelby tube has ration as recommended by Dyvik and Madshus 共1985兲 to measure
an internal diameter of 73 mm, an area ratio of 11.5%, inside G max , the maximum shear modulus. This is typically defined as
clearance ratio of 1%, inside cutting-edge angle of 0°, outside the shear modulus measured at strain level below 0.001%. Each
cutting-edge angle of 20°, and a diameter to thickness ratio of element was mounted in such a way that 3 mm of its length
46.2. protruded through the filter disk, into the soil sample as recom-
For an independent measure of the in situ shear strength and mended by Viggiani and Atkinson 共1995兲. Since the dynamic
shear modulus at very low strain, field vane tests and seismic stresses generated by the shear waves are much smaller than those
cone test tests were also conducted. These two tests are chosen as applied by the triaxial loading, the former may be considered to
they do not depend on some factors that need calibration, such as be small shear stress perturbations about the mean value applied
in a cone penetration test. A summary of the tests conducted and by the triaxial loading. Thus, the measured shear modulus may be
samples collected from this site is given in Table 2. regarded as a ‘‘local’’ value corresponding to the current stress
Baligh et al. 共1987兲 and Clayton et al. 共1995兲 had shown that, state of the sample at which the measurement was made.
in tube sampling, disturbance increases with smaller diameter-to- In reality, the situation is more complex since the soil may be
thickness ratio and larger outside cutting edge angle. Tanaka et al. anisotropic. Furthermore, the mobilized shear modulus also de-
共1996兲 reported that unconfined compressive strengths of samples pends on the stress path of the shear perturbation and its relation
collected by the Japanese thin-walled piston sampling tube were to recent stress history. The stress path caused by the dynamic
comparable with those collected by the Laval sampler and higher shear wave from a bender element is likely to be different from
than those collected by the Shelby tube, and suggested that the that experienced by a sample in a static triaxial compression test.
Japanese sampling tube is likely to inflict a lower degree of dis- According to Atkinson and Stallebrass 共1991兲, whenever there
turbance on soil samples than the Shelby tube. Tanaka et al. was a drastic change in stress path direction, the shear modulus
共1996兲 also suggested that the presence of the piston had an im- was able to recover its initial maximum value. Thus, using a
portant influence on sample quality. bender element within a static triaxial test would ensure that the

900 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / NOVEMBER 2002

J. Geotech. Geoenviron. Eng. 2002.128:898-906.


Downloaded from ascelibrary.org by New York University on 05/12/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 2. Stress–strain curves for samples retrieved using different


samplers under unconfined compression test

G max measured is likely to reflect the initial maximum value at


Fig. 3. Undrained shear strength from laboratory and in situ tests
that stress state. Jardine et al. 共1985兲 had also similarly observed
that the initial shear modulus determined from triaxial tests was
similar to the G max obtained from high quality in situ tests.
Besides the unconfined compression test 共UCT兲, isotropically
consolidated 共CIU兲 and K 0 -consolidated (CK0 U) undrained tri- Thus, because of its dimensions, the Shelby tube generally inflicts
axial compression tests were also conducted. For isotropic con- a higher degree of disturbance to the soil sample than the Japa-
solidation, step loading with a load ratio equal to unity was ap- nese piston sampler.
plied until the all around effective stress p ⬘ of the sample is equal It can also be observed from Fig. 3 that there is a small dif-
to its estimated in situ mean effective stress. For K 0 consolida- ference of less than 5% between the undrained shear strength
tion, an axial stress rate of 0.1 kPa/min was applied until the determined from the CIU and CK0 U tests, but both values are
estimated in situ vertical effective stress is reached, while a con- significantly higher than the corresponding undrained shear
dition of zero radial strain was maintained using a volume change strength determined by unconfined compression tests. Thus, with
feedback loop. During the shearing stage, a shear strain rate of reconsolidation, the undrained shear strengths for samples re-
1%/min was used in the unconfined compression tests and 0.1%/ trieved using the two different samplers show a significantly
min was chosen for consolidation compression tests as suggested smaller difference. This suggests that the effect of sample quality
by Head 共1986兲. on strength is largely eliminated through isotropic or K 0 recon-
solidation to the in situ state. This observation is in line with
Clayton et al. 共1992兲, who showed that the in situ undrained shear
Test Results strength can be approximately recovered, to within about 10%, by
reconsolidation to in situ stress levels, an observation also made
by others 共Chen and Kulhawy 1993兲. Figure 3 also shows the
Undrained Shear Strength
results from the field vane shear tests. In general, the strength
A principal consideration in the study of shear strength is the determined by the field vane shear test is higher than that from
quality of the sample on which the test is carried out. Figure 2 unconfined compression tests, but much less than those subjected
shows typical stress–strain curves from unconfined compression to reconsolidation. This lower value for shear strength determined
tests of samples retrieved by the two different samplers used in by the field vane shear tests is likely to be caused by the slightly
this study. As can be seen, both samples appear to reach the peak higher disturbance during the preboring process.
strength at nearly the same strain level, but with obviously differ-
ent values. However, the ultimate strengths seem to converge. Shear Modulus
This suggests that sampling disturbance affects the peak strength In addition to the LVDT, a Hall-effect strain transducer 共HET兲
more than the ultimate strength. This result is in agreement with 共Clayton et al. 1989兲, was also employed to measure the defor-
Hight’s 共1985兲 finding on the young sedimented marine clay. mation at the middle one third of the sample. Through appropriate
Figure 3 shows that the undrained shear strength determined amplification of the analog signal and using a 16 bit analog–
from unconfined compression tests for samples retrieved by the digital converter, a quantization error of less than 0.8 ␮m was
Japanese piston sampling tube are generally higher than the cor- achieved. Figure 4 shows the stress–strain curves for a typical
responding results from the Shelby tube samples, the average dif- sample, with the strain measured using the two different transduc-
ference being roughly 30%, consistent with the findings of Hight ers. As can be seen, a significant difference in the measured strain
共1985兲, and also the findings of Tanaka et al. 共1996兲, who had was obtained between the two transducers. In particular, the HET
shown that for the Ariake clay samples collected by Japanese reflects a much stiffer response than the LVDT in the early part of
thin-walled piston samplers, the undrained shear strengths were, the loading, an observation noted by others 共Jardine et al. 1984;
on average, 40% higher than those collected by the Shelby tube. Tatsuoka et al. 1994兲, and also consistent with the finding in the

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / NOVEMBER 2002 / 901

J. Geotech. Geoenviron. Eng. 2002.128:898-906.


Downloaded from ascelibrary.org by New York University on 05/12/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 6. Ratio of maximum shear modulus from laboratory tests over


Fig. 4. Stress–strain curves from UCT using external LVDT and
in situ shear modulus versus normalized mean effective pressure
local HET

1960s by Rowe and Barden 共1964兲 and Kirkpatrick and Younger this test, a cone with a receiver was pushed down a hole while a
共1970兲 that strains and stresses may not be uniform in the vicinity heavy roughened platen resting on the ground surface is hit to
of the end platens owing to the platen–soil friction. send a pulse of shear wave propagating down the soil. The ad-
The shear modulus of the Lower Marine Clay was determined vantage of this approach is that no correlation factor is needed,
under a number of different conditions in the laboratory, just as in and the shear modulus is measured directly from the differences
the case of shear strength 共Fig. 5兲. The maximum shear moduli of in arrival times of waves emitting from the surface when the cone
two samples, retrieved using the two different sampling tech- moves down a hole.
niques used in this study, were first determined in unconfined Fig. 5 shows the changes in G max with depth, measured by the
compression tests. The strains were determined using Hall-effect two laboratory tests and the seismic cone penetration test. Fig. 5
strain transducers, and the shear modulus obtained corresponds to shows that G max estimated from local strain measurements in un-
that at the start of loading when the strain in the sample is around confined compression tests conducted on soil samples retrieved
0.001%. Subsequently, to evaluate the effectiveness of the recon- using both samplers is both lower than the field-measured G max at
solidation technique to ameliorate the negative effect of poor the corresponding depth. This suggests that G max has also been
sample quality, K 0 consolidation of the samples, retrieved using affected by sampling disturbance in much the same way as the
the two different sampling methods used in this study, were first undrained shear strength. Equally important is the observation
carried out before undrained compression tests were carried out. that G max obtained for the sample retrieved using the Shelby tube
In these K 0 -consolidation tests, the maximum shear modulus was is clearly lower than that determined for the sample retrieved
measured using a pair of Bender elements. Finally, to provide an using the Japanese tube. This indicates that the sample quality
independent value for the maximum shear modulus, G max in the does play an important role in the determination of G max when
in situ condition, a seismic cone penetration test was carried. In tested in unconfined compression tests.

Fig. 5. Maximum shear modulus from in situ and laboratory tests Fig. 7. Variation of secant shear modulus, G s with axial strain

902 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / NOVEMBER 2002

J. Geotech. Geoenviron. Eng. 2002.128:898-906.


To evaluate the effectiveness of reconsolidation, the values of
G max , measured by Bender elements for K 0 -consolidated samples
are shown in Fig. 6. This shows that with reconsolidation, the
G max values are to a large degree recovered, although they remain
consistently about 10% lower than those from the in situ cone
penetration test.
To examine in greater detail the effect of reconsolidation, und-
rained compression tests were carried out on samples, retrieved
using the two different methods used in this study, and subjected
to isotropic consolidation to different mean effective stress, p ⬘ .
G max was measured using a pair of Bender elements and the re-
sults are plotted against the mean effective stress applied during
consolidation, p ⬘ , normalized by the estimated in situ mean ef-
fective stress, p 0⬘ , and shown in Fig. 6. The results show that
Downloaded from ascelibrary.org by New York University on 05/12/15. Copyright ASCE. For personal use only; all rights reserved.

when little or no reconsolidation stress is applied, G max for a


sample retrieved using a Shelby tube is lower than the value of
G max of a sample retrieved using a Japanese tube by as much as
30%. However, this difference reduces as the reconsolidation
stress increases. Nevertheless, a difference of about 10% still per-
sists between the two samples even when the reconsolidation Fig. 8. Stress–strain curves for CIU and CK0 U tests
stress reaches the in situ level (p ⬘ /p ⬘0 ⫽1). However, when plot-
ted against the shear modulus obtained from the seismic cone
penetration tests, the results show that G max for samples retrieved solidation. Thus, it is important to examine the nature of this loss
using the Shelby and Japanese tube samples are lower than the in structure that allows the shear strength to be recovered by
seismic cone value by about 30 and 20%, respectively. Clearly, reconsolidation.
for determining G max , the above result means that reconsolidation The structure of a soil can be classified into two main compo-
can recover, to a large extent, the difference in disturbances be- nents, termed bonding and fabric. Bonding enhances the strength
tween the two types of samplers, but not totally as indicated by and stiffness of a soil by the formation of physical attachments
the difference when the samples are reconsolidated to the in situ between soil grains. Jardine et al. 共1991兲 suggested that if a soil is
level. But more importantly, the results indicate that G max cannot loaded monotonically along a stress path, the stress space sur-
be fully recovered in the laboratory even when the samples are rounding the initial point is divided into three zones:
reconsolidated, when compared to G max measured in situ using 1. Zone I corresponds to the region where behavior is perfectly
seismic cone tests. linear elastic;
2. Zone II corresponds to the region where the stress-strain be-
havior is nonlinear but complete load–unload cycles show
Discussion of Sampling Disturbance on Undrained fully recoverable behavior, that is, nonlinear elastic; and
Shear Strength and Shear Modulus 3. Zone III corresponds to the region where the soil will expe-
rience irrecoverable strain.
Undrained Shear Strength Jardine 共1992兲 also suggested that the linear elastic zone I behav-
ior might be extensive for strongly cemented materials, indicating
Two factors are known to contribute to sampling disturbance,
that the linear elastic zone is connected to bonding. Hence, dam-
namely, effective stress release and destructuring 共Hight 1985;
age to bonding is likely to be exhibited by the fact that zone I has
Graham et al. 1987兲. ‘‘Destructuring’’ is taken to mean the ero-
been exceeded. Fig. 8 shows the changes in secant shear modulus
sion of that component of resistance that is related to the bonding
with strain. The estimated position of the limit of zone I is
and fabric of the soil. Clayton et al. 共1992兲 had shown that sig-
nificant loss of effective stress and some loss of structure occurred
even in very high quality samples of Bothkennar clay. Therefore,
in the present study, it is likely that some loss of structure would
have occurred, including in Japanese thin-walled piston tube
samples.
Graham et al. 共1987兲 had evaluated the effectiveness of the
reconsolidation method using reconstituted samples, which have
no structure. Without the influence of structure, their results
showed that the difference in undrained shear strength after re-
consolidation is very small, about 4 –9%. For the Singapore
Lower Marine Clay, the previous section has shown that shear
strengths determined from undrained compression tests on isotro-
pically and K 0 -consolidated samples were nearly the same for
samples retrieved using the two different samplers. However, the
shear strengths determined from unconfined compression tests,
which do not recover the loss of effective stress, are markedly
lower. What is interesting is that although the samples would have
Fig. 9. One-dimensional compression behavior of remoulded sample
suffered some degree of loss of structure, in the case of Singapore
and one retrieved using Japanese sampler
Lower Marine Clay, such loss is still recoverable through recon-

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / NOVEMBER 2002 / 903

J. Geotech. Geoenviron. Eng. 2002.128:898-906.


Table 3. Water Contents of Japanese and Shelby Samples sample fabric remains largely intact, then differences in the shear
Average Water strength observed between the two methods of sampling can only
Content % be attributed to differences in the residual effective stresses. Table
Difference
Japanese Shelby in water
3 shows the water contents of Japanese and Shelby samples,
Depth, M sampler tube content, % which show that, with one exception, the water content of Shelby
tube samples is, on average, about 0.6% higher. This trend is also
18.5–19.5 58.15 共6兲 58.54 共8兲 0.39 evident in the data on St-Simon and Nicolet clays 共Milovic 1971兲,
20.0–21.0 56.90 共7兲 57.22 共7兲 0.32 in which tube samples consistently show slightly higher water
content than block samples; the latter is considered the least dis-
21.5–22.5 54.77 共5兲 54.37 共9兲 0.40 turbed. It thus appears that the loss of strength in UCTs may be
23.0–24.0 61.42 共7兲 62.78 共11兲 1.36 related to an increase in water content, that is, to the swelling in
‘‘undisturbed’’ samples.
24.5–25.5 53.73 共7兲 54.85 共9兲 1.12 For the Singapore Lower Marine Clay, a water content differ-
Downloaded from ascelibrary.org by New York University on 05/12/15. Copyright ASCE. For personal use only; all rights reserved.

26.0–27.0 54.51 共7兲 55.3 共6兲 0.79 ence of 0.6% is equivalent to an increase in void ratio of about
0.016 using an average specific gravity of 2.7 for the estimation
Average water 56.58 共39兲 57.18 共50兲 0.60 共see Table 1兲. Using an average swelling compression index of
content 0.18 共see Table 1兲, this implies a loss of effective stress of roughly
Note: Parentheses show the number of samples. 20%. Although the relation between effective stress and peak
strength is unlikely to be linear, it is not unreasonable to expect
the peak strength to be reduced by roughly the same order of
marked in Fig. 8. This shows that the yielding of zone I occurs at magnitude. Examination of Fig. 2 shows that this is indeed the
strains of about 0.001–0.002% for Singapore Marine Clay. trend. Hence, it appears that the loss of strength in unconfined
The second component of structure is fabric, and this is asso- compression tests can be accounted for by the increase in water
ciated with particle arrangement and packing. Jardine et al. content due to swelling of the samples.
共1991兲 suggested that the end of zone II is related to the initiation The next question that needs to be addressed is what can cause
of particle slippage while the end of zone III is related to the this swelling? Clayton et al. 共1998兲 showed that the compressive
initiation of large scale changes in packing. Clayton et al. 共1998兲 axial strain ahead of a sampling tube is significantly affected by
suggested that the large scale yielding of zone III occurred at a changes in AR and OCA. However, it is unlikely this larger
strain of just less than 1%. This means that fabric damage is compression–extension cycle imposed by the larger AR and OCA
initiated at a much higher strain level than bond damage. Fig. 9
of the Shelby tube can account for the swelling. As shown in Figs.
shows the typical stress–strain curves of CIU and CK0 U samples.
5 and 9, peak strength occurs only after the strain level exceeds
As can be seen, the peak strength, which is taken to be the und-
1%, a level unlikely to occur during the sampling process, at least
rained shear strength, occurs at a strain level of about 1–2%.
not along the centerline of the sample 共Clayton et al. 1998兲. Fur-
Thus, in the case of such a weakly cemented clay, attainment of
thermore, positive, not negative, excess pore pressure is gener-
peak strength is likely to be associated with significant large scale
ated; the dissipation of which is likely to lead to lower, not higher,
fabric distortion and loss of packing. If the above premises are
water content especially at the outer part of the sample where the
indeed correct, then bonding would have been largely destroyed
shearing is more significant.
by the time the strain level reaches that which is required to
mobilize the peak strength. Thus, much of the peak strength is Another possibility that can account for the swelling of the
likely to arise from the fabric rather than the bonding component sample within the Shelby tube is its higher ICR. ICR is incorpo-
of structure. rated into samplers to prevent sample jamming due to build up of
From this perspective, it is important to examine the strain shear stresses along the inside wall of the sample. Clayton et al.
level induced during sampling for the two types of tubes. Clayton 共1998兲 results showed that increasing ICR will reduce the axial
et al. 共1998兲 have shown that with an AR of 10.14%, ICR of extension strain at the sample centerline, by allowing the sample
0.99%, and OCA of 9.9°, the maximum compression and exten- to expand sideways instead of axially. In other words, increasing
sion strains experienced at the sample centerline are about 0.25 the ICR probably leads to a reduction in confinement of the
and 0.5%. For the Shelby tube, with AR⫽11.5%, ICR⫽1%, and sample. If this occurs, the water content along the center of a
OCR⫽20%, the corresponding strains at the sample centerline sample will increase while a thin desiccated crust will be formed
may be somewhat higher owing to the higher OCA, but is un- adjacent to the sampler wall. Inspection of the Shelby samples
likely to exceed the 1–2% level. On the other hand, for the Japa- confirmed the presence of this thin crust. Indirect evidence of this
nese tube, the strain induced is likely to be lower owing to the may also be found in the results of some previous researchers.
lower AR, OCA, and ICR, as well as the presence of a piston that Tanaka et al. 共1996兲 compared the unconfined compression
helps to hold the sample. Thus, the strains induced during sam- strength of samples obtained by the Japanese, Laval, NGI54,
pling in both cases, are sufficient to cause a loss of bonding and, Shelby, and ELE100. The highest unconfined compression
perhaps, even some localized yielding of interparticle contacts, strength came from samples obtained using the Japanese and
but is unlikely to result in significant large scale fabric distortion Laval samplers, both of which have ICR⫽0. Bozozuk 共1971兲 also
and loss of packing, associated with peak strength. With the fabric noted that the Osterberg sampler (ICR⫽0.42%) provided less
largely intact, the tested samples are able to recover their peak disturbed samples than the NGI54 sampler (ICR⫽1%).
strengths upon reconsolidation. In summary, when tested in unconfined compression tests,
However, what remains interesting is the significant difference Shelby tube samples yield lower shear strengths than Japanese
between the Japanese and Shelby tube samples when tested in tube samples. The differences can be attributed to the different
unconfined compression tests. If it is indeed true that sampling- degree of disturbance induced by the two samplers. However, the
induced destructuring is largely due to loss of bonding, whereas loss in shear strength in unconfined compression tests can be

904 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / NOVEMBER 2002

J. Geotech. Geoenviron. Eng. 2002.128:898-906.


largely recovered by reconsolidating the samples to in situ effec- sistent with the results presented that reconsolidation can help to
tive stress levels. This indicates that sample disturbance is un- recover the maximum shear modulus substantially but not com-
likely to have resulted in large scale fabric distortion although pletely.
loss of bonding could have occurred. However, bonding is un-
likely to contribute significantly to the undrained shear strength
for Singapore Lower Marine Clay as this material is a young and Conclusion
lightly cemented clay. Finally, differences in water content indi-
cate that the loss of shear strength in unconfined compression It is important to recognize that the sampling process causes two
tests appears to be caused by a slight swelling of the clay within different disturbances, namely, stress release and loss of structure.
the sampler. This swelling is likely to be caused by the higher Reconsolidation methods are used to negate the effect of stress
ICR of the Shelby sampler. release whereas it is not clear whether it can compensate for the
loss of structure. There is also a recent recognition that structure
of the soil is composed of two components, bonding and fabric.
Maximum Shear Modulus, Gmax These are in turn related to different types of yielding, as argued
Downloaded from ascelibrary.org by New York University on 05/12/15. Copyright ASCE. For personal use only; all rights reserved.

by Jardine et al. 共1992兲.


The fact that samples retrieved using the Japanese sampler are Bonding enhances the strength and stiffness of a soil by the
less disturbed than those retrieved using the Shelby tube is also formation of physical attachments between soil particles and is
manifested in the measurement of maximum shear modulus in associated with yielding of the linear elastic zone, which occurs at
unconfined compression tests. The results consistently show that strain of about 0.002% for Singapore Marine Clay. Fabric is as-
samples from the Japanese thin-walled piston sampler have sociated with particle arrangement and packing. It has been sug-
higher maximum shear stiffness than samples from the Shelby gested that the end of the zone of the recoverable but elastic zone
tube. What is interesting is that reconsolidation to the in situ stress is related to the initiation of particle slippage while yielding in
state indicates significant recovery, especially when the G max val- zone III is related to the initiation of large scale changes in pack-
ues for samples retrieved using the two different samplers are ing.
compared with each other. However, consistently, the values ob- With these ideas as background, the following conclusions can
tained in the laboratory, even with reconsolidation, are lower than be drawn:
that from values obtained from in situ seismic cone tests. 1. In general, samples retrieved using the Shelby tube show a
Thus, the effect of sampling disturbance on G max is substan- higher degree of disturbance compared with those using the
tially, but not entirely, recovered by reconsolidation. As reconsoli- Japanese thin-walled piston sampler. The undrained shear
dation can only recover effective stress, the above results suggest strength determined by unconfined compression tests on
that for the Singapore Lower Marine Clay, G max depends to a samples retrieved using the Japanese sampler are about 30%
large extent on the effective stress levels and, to a lesser extent, higher than those using the Shelby tube. However, in this
on the bonding of the soil. This is consistent with the earlier type of tests where the stress release is not restored, the shear
suggestion that partial destructuring, in the form of loss of bond- strengths obtained are consistently lower than those deter-
ing and local yielding at interparticle contacts, is likely to have mined by in situ tests.
occurred under the strains imposed by the sampling process. The 2. The reconsolidation method can effectively recover the ef-
notion that G max of the marine clay being studied is often strongly fect of stress release on the peak strength value. This also
dependent on the effective stress level and less so on the stress 共or means that the loss of bonding, which would inevitably
strain兲 history is consistent with the results and findings of previ- occur, does not affect the peak strength in the case of the
ous researchers. For normally consolidated soils, Hardin and Singapore Lower Marine Clay, a lightly cemented clay. It is
Black 共1968兲 suggested that argued in this paper that peak strength is affected more by
fabric, which yields at a strain level not likely to be sustained
G max⬀F 共 e 兲 p ⬘ 0.5 (1) during the sampling process, and not by loss of bonding,
which occurs at very small strain.
where p ⬘ ⫽mean effective stress and F(e) is a function of the
3. However, the reconsolidation method is only substantially
void ratio e. Hardin and Black 共1968兲 also noted that, for nor-
but not completely effective in recovering the G max . The
mally consolidated clays, e is related to p ⬘ , and this has been
G max value for a sample reconsolidated to the in situ level is
used by several researchers to relate G max to p ⬘ alone for a nor-
still about 10% smaller than that determined from the seis-
mally consolidated clay. For example, Dasari 共1996兲 showed that,
mic cone penetration tests. It is argued in this paper that
for normally consolidated kaolin,
G max for a young and lightly overconsolidated marine clay is
G max⬀p ⬘ 0.69 (2) strongly dependent on the effective stress and less strongly
on recent stress history, which is related to bonding. This last
Comparing Eq. 共2兲 with Eq. 共1兲 shows that the higher exponent of aspect is likely to be disturbed when either of the samplers is
p ⬘ in Eq. 共2兲 probably accounts for the effect of the reduction in used.
void ratio with increasing p ⬘ . For overconsolidated clays, Dasari
共1996兲 further suggested that the effects of the overconsolidation
ratio 共OCR兲 can be accounted for by the relationship References
G max⬀OCRn p ⬘ 0.69 (3) ASTM D1587. 共1994兲. Standard methods for thin-walled tube sampling
of soils, American Society for Testing and Materials, Philadelphia.
However, Dasari 共1996兲 adopted a value of approximately 0.2 for Atkinson, J. H., and Stallebrass, S. E. 共1991兲. ‘‘Model of recent stress
n. This suggests that for a lightly overconsolidated clay, as is the history and nonlinearity in the stress–strain behavior of overconsoli-
case of the Singapore Lower Marine Clay, G max is strongly de- dated soil.’’ Proc., 7th Int. Conf. of Computer Methods Advance Geo-
pendent on p ⬘ and to a much lesser extent on OCR. This is con- mechanics, Beer, Booker, and Carter, eds., 1, 555–560.

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / NOVEMBER 2002 / 905

J. Geotech. Geoenviron. Eng. 2002.128:898-906.


Baligh, M. M., Azzouz, A. S., and Chin, C. T. 共1987兲. ‘‘Disturbances due Hight, D. W. 共1985兲. ‘‘Evaluation of geotechnical parameters from tri-
to ‘ideal’ tube sampling.’’ J. Geotech. Eng., 113共7兲, 739–757. axial tests on offshore clay.’’ Offshore Site Investigation–Advances in
Bozozuk, M. 共1971兲. ‘‘Effect of sampling, size, and storage on test results Underwater technology and Offshore Engineering, 3, 253–268.
for marine clay.’’ Sampling of soil and rock, ASTM STP 483, Ameri- Japanese Society of Soil Mechanics and Foundation Engineering
can Society for Testing and Materials, Philadelphia, 121–131. 共JSSMFE兲. 共1977兲. Japanese sampling manual, Japanese Committee
Chen, Y. J., and Kulhawy, F. H. 共1993兲. ‘‘Undrained strength interrela- of Soil Sampling 共in Japanese兲.
tionships among CIUC UU, and UC tests.’’ J. Geotech. Eng., 119共11兲, Jardine, R. J., Symes, N. J., and Burland, J. B. 共1984兲. ‘‘Measurement of
1732–1750. soil stiffness in triaxial apparatus.’’ Geotechnique, 34共3兲, 323–340.
Clayton, C. R. I., Khatrust, S. A., Bica, A. V. D., and Siddique, A. 共1989兲. Jardine, R. J., Fourie, A., Maswoswe, J. and Burland, J. B. 共1985兲. ‘‘Field
‘‘Use of Hall-effect semiconductors in geotechnical instrumentation.’’ and laboratory measurements of soil stiffness.’’ Proc., 11th Int. Conf.
Geotech. Test. J., 12共1兲, 69–76. on Soil Mechanics and Foundation Engineering, 2, 511–514.
Clayton, C. R. I., Hight, D. W., and Hopper, R. J. 共1992兲. ‘‘Progressive Jardine, R. J. John, St., Hight, H. W. and Potts, D. M. 共1991兲. ‘‘Some
destructuring of Bothkennar clay: Implications for sampling and re- practical applications of a nonlinear ground model.’’ Proc., 10th Eu-
consolidation procedures.’’ Geotechnique, 42共2兲, 219–239. ropean Conf. Soil Mechanics and Foundation Engineering, Florence,
Clayton, C. R. I., Matthews, M. C. and Simons, N. E. 共1995兲. Site inves-
Downloaded from ascelibrary.org by New York University on 05/12/15. Copyright ASCE. For personal use only; all rights reserved.

1, 223–228.
tigation, Blackwell, London.
Jardine, R. J. 共1992兲. ‘‘Some observations on the kinematic nature of soil
Clayton, C. R. I., Siddique, A., and Hopper, R. J. 共1998兲. ‘‘Effect of
stiffness.’’ Soils Found., 32共2兲, 111–124.
sampler design on tube sampling disturbance—Numerical and ana-
Kirkpatrick, W. M., and Younger, J. S. 共1970兲. ‘‘Strain conditions in a
lytical investigations.’’ Geotechnique, 48共6兲, 847– 867.
compression cylinder.’’ J. Soil Mech. Found. Div., Am. Soc. Civ. En-
Dasari, G. R. 共1996兲. ‘‘Modeling the variation of soil stiffness during
gineering, 96共5兲, 1683–1695.
sequential construction.’’ PhD thesis, Cambridge Univ., U.K.
Davis, E. H., and Poulus, H. G. 共1971兲. ‘‘Laboratory investigation of the Kirkpatrick, W. M., and Khan, A. J. 共1984兲. ‘‘Reaction of clays to sam-
effect of sampling.’’ Trans., Inst. of Engineers, Australia, GE.9, No. 1, pling stress relief.’’ Geotechnique, 34共1兲, 29– 42.
86 –94. Milovic, D. M. 共1971兲. ‘‘Effect of sampling on some soil characteristics.’’
Dyvik, R., and Madshus, C. 共1985兲. ‘‘Laboratory measurements of G max ; Sampling of soil and rock, ASTM STP 483, American Society for
using Bender elements. Advances in the art of testing soils under Testing and Materials, Philadelphia, 164 –179.
cyclic conditions.’’ Proc., ASCE Convention, Detroit, Mich., 186 – Rowe, P. W., and Barden, L. 共1964兲. ‘‘Importance of Free Ends in Triaxial
196. Testing.’’ J. Soil Mech. Found. Div., Am. Soc. Civ. Eng., 90共1兲, 1–27.
Graham, J., Kwok, C. K., and Ambrosie, R. W. 共1987兲. ‘‘Stress release, Tanaka, H., Sharma, P., Tsuchida, T., and Tanaka, M. 共1996兲. ‘‘Compari-
undrained storage, and reconsolidation in simulated underwater clay.’’ son study on sample quality using several types of samplers.’’ Soils
Can. Geotech. J., 24, 279–288. Found., 36共2兲, 57– 68.
Hardin, B. O., and Black, W. L. 共1968兲. ‘‘Vibration modulus of normally Tatsuoka, F., Sato, T., Park, C. S., Kim, Y. S., Makubi, J. N., and Kohata,
consolidated clay.’’ J. Soil Mech. Found. Div., Am. Soc. Civ. Eng., Y. 共1994兲. ‘‘Measurements of elastic properties of geomaterials in
94共2兲, 353–369. laboratory compression test.’’ Geotech. Test. J., 17共1兲, 80–94.
Head, K. H. 共1986兲, Manual of soil laboratory testing, Pentech, London, Viggiani, G., and Atkinson, J. H. 共1995兲. ‘‘Interpretation of bender ele-
3. ment tests.’’ Geotechnique, 45共1兲, 149–154.

906 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / NOVEMBER 2002

J. Geotech. Geoenviron. Eng. 2002.128:898-906.

You might also like