Professional Documents
Culture Documents
2011
Japanese Geotechnical Society
ABSTRACT
The eŠects of sample disturbance on undrained shear strength were investigated from samples with various qualities,
retrieved by diŠerent types of samplers at the Takuhoku site, Sapporo. Sample quality was evaluated by three types of
shear tests: unconˆned compression, fall cone and triaxial recompression tests. Similar to a previous study conducted
by Horng et al. (2010), in which the sample quality was evaluated by two nondestructive tests, the residual eŠective
stress (suction) by ceramic disc and the shear modulus by bender element under unconˆned conditions, the present
study shows that the small edge angle of a tube sampler is important to obtain high quality sample. In addition, the ex-
istence of a piston does not have a signiˆcant eŠect on the strength properties. The recompression technique in the
triaxial test, where the specimen is consolidated back to the in situ stresses, was able to duplicate undisturbed soil be-
havior except when the structure of a soil sample was signiˆcantly destroyed. The unconˆned compressive strength is
apparently governed by the residual eŠective stress. From the two types of tests simulating sampling processes,
however, it was found that a reduction in the unconˆned compressive strengths of low quality sample was brought by
the loss of the residual eŠective stress as well as destruction of soil structures.
Key words: residual eŠective stress, sample disturbance, sample quality, sampling, soil structures, undrained shear
strengths (IGC: C6)
203
Table 1. Main characteristics and dimensions of tube samplers used in this study
Field Edge angle Tube thickness Area ratio Piston Sampling depths Sampling depths
samplers (9) (mm) (z) (upper layer) (m) (lower layer) (m)
Sampling Site
SAMPLERS USED IN THIS STUDY AND Sampling was carried out at the Takuhoku site, Sap-
SAMPLING SITE poro, Japan. For details of the properties, refer to Horng
Samplers et al. (2010). They are otherwise only brie‰y mentioned in
The samplers used in this study are the same as those this paper. The main geotechnical properties of this site
used in our previous study (Horng et al., 2010) and their are shown in Fig. 1. The deposits consist of 5 m ˆll and
geometric dimensions are indicated in Table 1. The ˆrst peat followed by a 4.5 m silty sand deposit, overlying the
sampler, which is the standard tube currently used in clay layers investigated in this study. A sandy silt layer at
Japan, consists of an inside diameter of 75 mm, a length a depth of 15 to 18.5 m separates the soil proˆle into the
of 1.0 m (the sample length is 0.8 m), and an edge angle upper and lower clay layers. Sampling was carried out at
of 69. The thickness of the tube wall is 1.5 mm, which two diŠerent depths: the upper (10¿15 m) and lower
corresponds to an area ratio of 8.2z. The sampling tube (20¿24 m) clay layers as indicated in Fig. 1. The ground
is made of stainless steel. More details of this sampler water table is located about 3 m below the ground sur-
may be found in the JGS article (1998) and in Tanaka et face. The natural water content varies between 60 and
al. (1996). In this study, more geometrically diŠerent tube 70z and the plasticity index (Ip) is about 45¿53 and
samplers consisting of the same inside diameter of 75 mm 50¿63 for the upper and lower clay layers, respectively.
were designed. The fourth tube sampler, 909F10, has an The yield consolidation pressure ( p?y), which was meas-
edge angle of 909and a wall thickness of 10 mm, result- ured by CRS oedometer at a strain rate of 0.02z/min
ing in an area ratio of 60.4z. The last sampler, (3.3×10-6/s), is somewhat lower than the in situ eŠective
69F1.5(O), is the 69F1.5 without using ˆxed piston (an overburden pressure (s?vo), which is calculated by assum-
open drive sampler) during sampling. The sampling ing that the pore water pressure distribution is hydrostat-
method and procedures in this study were reported by ic. Since the ˆll material at ground surface was placed in
Horng et al. (2010). the 1960's, it is believed that the sampling clayey soil is
still undergoing consolidation. For more details of this
investigation, readers can refer to Horng et al. (2010).
SAMPLE DISTURBANCE SHEAR STRENGTHS 205
The ˆeld vane test (FVT), using a vane blade of 40 mm the soil proˆle in Fig. 1 shows that the objective sampling
in diameter and 80 mm in height and the piezocone test clay layers are under consolidation caused by the ˆlling.
(CPT) were carried out to measure mechanical properties Thus, consolidation was done in the vertical direction by
of the site. From the CPT test, the undrained shear a pressure 0.8 times the yield consolidation pressure, p?y,
strength was calculated using the relation su(CPT)=(qt-svo)/ which was measured by the constant rate of the strain
Nkt, where qt is the point resistance of the piezocone, svo oedometer (CRS) test under a strain rate of 0.02z/min.
is the total overburden pressure, and Nkt is the cone fac- A coe‹cient of 0.8 was speciˆcally chosen to avoid the
tor. By equating the undrained shear strengths of CPT overestimation of p?y from the CRS under a relatively high
and FVT, the cone factor Nkt was able to be calculated. In strain rate and possibility overestimating the strength due
this site Nkt was estimated to be 11.5. The undrained to large consolidation pressure. The coe‹cient of lateral
shear strengths from the unconˆned compression test earth pressure at rest (Ko) was estimated to be 0.55 from
(UCT) are also plotted in this ˆgure, where the soil sam- the Ko-consolidation triaxial test at the normally consoli-
ples were retrieved by the Japanese standard ˆxed piston dated state. The consolidation stresses were kept for
sampler (69 F1.5 in Table 1). The mean undrained shear about 24 hours. After complete consolidation, the speci-
strengths for the upper and lower sampling depths are ap- men was sheared under the undrained condition at an axi-
proximately 20 kPa and 40 kPa, respectively. al strain rate of 0.1z/min. The shearing was done until
the axial strain reached 15z.
Other types of triaxial tests were also conducted to
LABORATORY TEST METHODS simulate the eŠects of the residual eŠective stresses and
Unconˆned Compression Test (UCT) soil structures. More details of these triaxial tests will be
The undrained shear strength (qu/2) from the UCT is discussed later.
widely used for stability designs in geotechnical engineer-
ing, and this type of test is believed to be the most sensi-
tive to sample disturbance (Lacasse et al., 1985; Tanaka EVALUATION OF SAMPLE QUALITY BY THE UCT
et al., 1996; Mitachi et al., 2001). The UCT was carried AND FCT
out by the Japanese Industrial Standard (JIS A 1216: Test Results from the UCT
2009): the specimen was trimmed by a wire saw to a di- Figure 2 shows the typical test results of the UCT on
ameter and height of 35 and 80 mm, respectively. The samples from various types of samplers. The vertical axis
specimen was compressed at a constant strain rate of 1z/ (s/2/p?y) is the unconˆned compressive stress normalized
min. by the yield consolidation pressure at the depth of the
specimen. The normalizations are done to take into ac-
Fall Cone Test (FCT) count the slightly diŠerent depths of the samples for com-
The FCT has been used to measure the undrained shear parative purposes. The horizontal axis is the axial strain,
strength mostly in Scandinavian countries. The FCT is ea(z). From both the upper and lower layers in the ˆgure
not used in Japan for measuring the undrained shear it can be seen that the stress-strain curves of the samples
strength, but for determining the liquid limit, also ob- retrieved by the 69F1.5, 69F10, and 69F1.5(O) tube sam-
tained by the ``Casagrande test''. In this study, the FCT plers are very similar and their peak shear strengths are
was used to measure undrained shear strength, following larger than those of the other two tubes, 909F1.5 and
the standard of the Japanese Geotechnical Society (JGS 909 F10. The 909 F10 samples show unusual stress-strain
0142–2009) by using a cone with a tip angle of 609and patterns, implying severe disturbance by the geometric
mass of 60 g. The cone was allowed to fall freely under its design of the blunt edge angle and thick wall thickness
own weight from a position at rest with the cone tip just (909and 10 mm). Their stress-strain curves move to the
touching the surface of the soil sample. The undrained right hand side and harden up to a ˆnal axial strain of
shear strength of FCT was calculated following the equa- 15z without showing any peak strength and the curves
tion of su=ka(mg/d 2), where m is the mass of the cone are completely diŠerent from other tube samples. The
(= 60 g), g is the earth gravity acceleration, d is the depth 909 F1.5 samples show stress-strain curves similar to those
of penetration of the cone tip into a soil specimen after 5 of the other 69samplers, but their peak strengths are
seconds, and ka is the cone factor depending on the angle somewhat smaller than those of the 69 F1.5, 69F10, and
of the cone tip. The value ka=0.29 was assumed accord- 69F1.5(O) samplers.
ing to Wood (1990) for the cone angle of 609in this The UCT results for all samples are shown in Fig. 3,
study. where Figs. 3(a), (b) and (c) show the eŠect of diŠerent
geometric designs of the tube samplers on qu/2, E50
Triaxial Recompression Test (CKoUC) (secant moduli at 50z strength), and ef (the axial strain at
The triaxial test specimen size was the same size as that the peak stress), respectively. These test results show the
of the UCT in this test. A back pressure of 200 kPa was same features as those in previous studies: if a sample is
applied in order to obtain high saturation. To estimate disturbed, its stress-strain curve exhibits a small peak
the in situ strength, a recompression test was employed, strength, small E50, and large ef (Lefebvre and Poulin,
where the specimen was consolidated under the same 1979; Oka et al., 1996). It is interesting to note that E50 is
eŠective stress condition as that of the in situ. Note that much more sensitive to disturbance than the peak
206 HORNG ET AL.
the present test results, especially at small p?r/p?y, for ex- samples show the behaviors of strain softening after well-
ample, less than 0.05. deˆned peak shear stresses at small strains; 2) 909F10
samples show lower shear stresses at strains where the
other samples reach the peak stress, but higher shear
TRIAXIAL RECOMPRESSION TEST (CKOUC) stresses at large axial strains (À5z) than those of other
Stress-strain Relation and Stress Path samples. Tanaka (2000) also compared test results from
As already indicated, strengths from the UCT and FCT the recompression triaxial test for Bothkennar clay
are strongly in‰uenced by the current eŠective conˆning retrieved by JPN standard and ELE 100 samplers and he
stress, p?r, because these tests are carried out under uncon- made the same conclusions: i.e., the behavior of the ELE
ˆned conditions. Therefore, when the in situ eŠective 100 samples is the same as that of 909F10 in the present
conˆning pressure is applied to the specimen, the strength study, while the JPN sample shows the same tendency of
should be recovered. In triaxial test, this idea is called the 69F1.5, 909F1.5, 69 F10, and 69F1.5(O) samples. Similar
``Recompression method'' and was introduced by Berre hardening behavior can also be observed from test results
and Bjerrum (1973). Lacasse et al. (1985) conˆrmed that by Lunne et al. (2006) for Scandinavian clays retrieved by
the triaxial technique is able to correct large portions of NGI 54 mm.
sampling disturbance. Of course, there is also criticism The stress paths are also plotted in Fig. 7, using dia-
that the recompression technique overestimates the gram [(s?a+s?r)/2, (s?a-s?r)/2], where s?a and s?r represent
strength caused by decreasing the void ratio due to the the axial and radial eŠective stresses, respectively. Similar
recompression. This may become signiˆcant when the to the stress-strain curves above, the stress paths of
sample is heavily disturbed. 69F1.5, 909 F1.5, 69 F10, and 69F1.5(O) show no
Comparisons of typical stress-strain curves for speci- profound diŠerence in sample quality. On the other
mens collected by diŠerent types of tube samplers are hand, the stress paths of the low quality samples, i.e.,
shown in Fig. 7. It can be seen that the stress-strain 909 F10, rise up without any peaks until the end of test
curves and stress paths of all samples from the tube sam- and excess pore water pressure does not signiˆcantly
plers, 909F1.5, 69F10, and 69 F1.5(O) do not diŠer build up. These patterns of the stress path are somewhat
remarkably from those collected by the standard diŠerent from those of the poor quality samples of
Japanese sampler. On the other hand, the stress-strain Bothkennar retrieved by ELE 100 sampler or of Nor-
curves of the samples collected by the thick wall and large wegian marine clays retrieved by NGI 54 mm as reported
edge angle, 909 F10, diŠer radically from those of other by Tanaka (2000) and Lunne et al. (2006), respectively.
tube samplers. The noteworthy unusual patterns of Reasons for this diŠerence may be attributed to soil char-
909 F10 samples compared with those of other tubes are acteristics of the sampling site or the features of samplers.
observed as follows: 1) 909F10 samples show the behav- At present, it is not possible to identify which factor is
iors of strain hardening up to axial strain of 10z, which dominant.
are similar to their UCT counterparts, whereas other tube
Volume Change due to Recompression
Lunne et al. (1997) used the parameter De/e0 to quanti-
fy sample disturbance as shown in Fig. 8, where De is the
change of void ratio during the reconsolidation back to
the in situ eŠective stress and e0 is the in situ void ratio.
Figure 8 shows the relation between De/e0 and p?r from
the recompression test, indicating the existence of a
strong correlation between them, i.e., De/e0 decreases
Fig. 7. Typical results of recompression CKoUC triaxial Fig. 8. Relations between p?r and volume change De/e0
SAMPLE DISTURBANCE SHEAR STRENGTHS 209
p?1 (kPa) p?y (kPa) su1 (kPa) De1 p?y1 (kPa) p?1/p?y1 su1/p?y1
6 132 22.9 -0.144 85.4 0.07 0.27
(b) Test 2
p?max (kPa) p?2 (kPa) su2 (kPa) De2 p?y2 (kPa) p?2/p?y2 su2/p?y2
290 15 34.9 -0.216 150.8 0.10 0.23
diŠerent eŠective conˆning pressures, from which the seriously destroyed the soil structures, which are natural-
UCT was conducted under atmosphere; whereas Test 1 ly inherent by ageing, chemical bonding, and cementa-
was conducted under large constant eŠective conˆning tion. On the other hand, the stress-strain curve of the
pressures during shearing (the back pressure of 200 kPa sample from the Test 2, i.e., its structures were destroyed
was applied). If the soil structures of samples are and by keeping the same p?r/p?y ratio as those of the other
damaged ( see Test 2 in the ˆgures), even though the soil two tests, its stress-strain curve shows more disturbance
specimens are kept under the same eŠective conˆning in terms of soil structures.
pressures, p?r/p?y, their stress-strain curves are moderate Moreover, the eŠects of sampling disturbance by de-
and also their undrained shear strengths are lower, which struction of soil structures from various types of tube
manifests the eŠect of soil structures on soil behaviors. samplers were also investigated, as shown in Fig. 17. It
Similarly, the same series of testing were applied to the should be noted from the ˆgure that all samples have the
poorest quality samples whose measured p?r/p?y=0.05 as same p?r/p?y=0.05. Their stress-strain curves show no
shown in Fig. 16. It should be kept in mind that this ratio diŠerence in shapes or in their peaks. This means that the
corresponds to the poor quality samples of the 909F10 structures of the 69F10, 909 F1.5, and 69F1.5(O) samples
tube, which were disturbed due to the geometric design of were not destroyed by sampling, compared with those of
the tube sampler during sampling. It clearly shows that the standard tube sampler. These results once again pro-
the stress-strain curve of Test 1, has a well-deˆned peak vide more evidence that, by using 69F10, 909 F1.5, and
strength even under the small p?r/p?y ratio, whereas the 69F1.5(O) samples, the recompression technique
curve of 909F10 (UCT) sample is to the far right of Test 1 (CKoUC) has the ability to duplicate the undisturbed soil
and strain hardens up to axial strain of 15z. The results behavior as mentioned earlier.
demonstrate that sampling using 909F10 tube samplers
SAMPLE DISTURBANCE SHEAR STRENGTHS 213