Professional Documents
Culture Documents
Abstract: Soils are generally known not to be isotropic materials. The inherent anisotropy of clays results from the deposition process, which
tends to induce a horizontal bedding plane in the soil layer. In this paper, the anisotropy of clays was studied to obtain more accurate analytical
results for geotechnical problems, especially for deep excavations in soft clay with nearby structures. A series of K0 -consolidated undrained
triaxial compression (CK0 UC) tests thus was conducted on tube samples of natural Taipei silty clay with multidirectional bender elements.
A newly designed triaxial testing system equipped with a high-precision servo motor and local strain sensors was developed. A new soil suction-
control system was also developed to perform soil saturation for triaxial tests. The suction-control system was used to reduce the change in the
void ratio during saturation in the triaxial tests. Soil samples were retrieved from a site near a well-documented excavation case in Taipei. The
anisotropy ratios for both the shear modulus and the undrained Young’s modulus were obtained by performing small-strain triaxial tests. The test
results indicated that the anisotropy ratios of the shear moduli at the end of reconsolidation ranged from 1.15 to 1.44. Soils with higher
overconsolidation ratio values had higher anisotropy ratios. Numerical simulations for an excavation case history were performed using
a developed small-strain soil model that incorporates the anisotropy in the soil stiffness. The analysis results showed that differences between the
anisotropic and isotropic models for the wall displacements, ground-surface settlements, and lateral soil movement behind the wall ranged
from 10 to 43% at specified depths. DOI: 10.1061/(ASCE)GT.1943-5606.0001010. © 2014 American Society of Civil Engineers.
Author keywords: Anisotropy; Stiffness; Excavation.
Introduction 1991). Results from Osouli et al. (2010) have shown that an in-
clinometer placed at some distance from the excavation wall can
The stress-strain behavior of soils at small strains has become one provide important but not yet exhaustive information on small-
of the most important geotechnical engineering topics over the last strain nonlinearity in soil behavior and estimate surface
two decades. Many researchers (Jardine et al. 1986; Burland 1989; settlements.
Mair 1993) have used in situ data measurements to show that the However, soils are stiff and nonlinear as well as anisotropic in
strains from construction are usually less than 0.0005 in regions far the small-strain range. The anisotropy of undrained shear strengths
from construction areas. The accuracy of predictions for ground has been studied and shown to be important in excavation analyses
movement during underground construction, such as excavation (Whittle et al. 1994; Hashash and Whittle 1996; Hashash and
in an urban area, is greatly improved by the use of soil models that Whittle 2002; Mesri and Huvaj 2007). The importance of the initial
can simulate soil behaviors from very small strains up to failure. stress-strain strength anisotropy and its degradation has been il-
Therefore, increasing numbers of soil constitutive models on lustrated in computing ground movements induced by excavations
small-strain behaviors have been developed in recent years, e.g., (Hashash and Whittle 1996; Finno and Tu 2006). In addition,
the MIT-E3 model (Whittle and Kawadas 1994), the Hardening inherent anisotropy in the stiffness has been observed in tests on
soil model with small-strain (HS-small) model (Benz 2007), and several types of soil (Pennington et al. 1997; Ng et al. 2004;
the undrained soft-clay (USC) model (Hsieh and Ou 2011). These Nishimura et al. 2005; Gasparre et al. 2007; Cho and Finno 2010).
soil models emphasize the importance of high stiffness and The effect of stiffness anisotropy on the prediction of ground
nonlinearity in soils, as has been observed by various researchers movement, especially lateral displacement, is as important as that
(e.g., Jardine et al. 1986; Burland 1989; Atkinson and Sallfors of high stiffness and nonlinearity. Schuster et al. (2009) developed
a simplified method for estimating the lateral ground movement
1 induced by excavations based on numerical experiments per-
Postdoctoral Fellow, Dept. of Construction Engineering, National
Taiwan Univ. of Science and Technology, Taipei City 10672, Taiwan,
formed by Kung et al. (2007). However, large numbers of artificial
ROC (corresponding author). E-mail: D9405003@mail.ntust.edu.tw data were generated through finite-element analyses using an
2
Professor, Dept. of Construction Engineering, National Taiwan Univ. isotropic small-strain soil model, i.e., an earlier version of the USC
of Science and Technology, Taipei City 10672, Taiwan, ROC. E-mail: model. The accuracy of the lateral ground movement predictions
ou@mail.ntust.edu.tw may be improved by considering the stiffness anisotropy of the
3
Professor, Dept. of Assets and Property Management, Hwa Hsia In- clays in the analysis.
stitute of Technology, New Taipei City 23568, Taiwan, ROC. E-mail: The stiffness anisotropy of clays is generally assumed to be
spg@cc.hwh.edu.tw cross-anisotropic because there is a horizontal plane of isotropy,
Note. This manuscript was submitted on January 20, 2012; approved on
i.e., the bedding plane. A cross-anisotropic constitutive equation is
July 19, 2013; published online on July 25, 2013. Discussion period open
until June 1, 2014; separate discussions must be submitted for individual usually used to simulate the anisotropy of soils. Cross-anisotropic
papers. This paper is part of the Journal of Geotechnical and Geoenvir- soil models have been used in numerical analyses to predict
onmental Engineering, Vol. 140, No. 1, January 1, 2014. ©ASCE, ISSN ground settlement from tunneling (Lee and Rowe 1989; Simpson
1090-0241/2014/1-237–250/$25.00. and Ng 1995). The use of cross-anisotropic elastic parameters has
Eh
Cross-Anisotropic Stiffness Matrix Ghh ¼ ðhomogeneous on the bedding planeÞ (3)
2ð1 þ nhh Þ
The notation for a cross-anisotropic material used hererin is: the The five independent parameters are Ev , Eh , nvh , Ghh , and Gvh .
y-axis (the vertical direction) represents the direction of the an- Kuwano and Jardine (1998) and Pennington et al. (2001) have
isotropy, and the x,z-plane is the plane of isotropy. The stress-strain shown how Ghh and Gvh or Ghv may be obtained directly from bender
increment equation for a cross-anisotropic material can be written as element tests. The undrained Young’s moduli Ev and Eh were
follows: obtained by performing an undrained triaxial test. In the undrained
condition, there is an additional constraint on the relationship be-
2 3 tween the Poisson’s ratios
1 y y
2 vh 2 hh 0 0 0
6 Eh Ev Eh 7 y hh þ y hv ¼ 2
dɛ z dɛ y
2 ¼ 2
dɛ z þ dɛ y
¼ 2
2dɛ x
¼1
6 yhv 7 (4)
9 6 78 9
1 y dɛ x dɛ x dɛ x dɛ x
8
> dɛ x > 6 2 Eh Ev
2 hv
Eh
0 0 0
7> dsx >
>
> > 6
> 7>
7>
>
>
> 6
>
> dɛ y >> >
> dsy > >
The following stress-strain relation is obtained from Eq. (1):
> y > >
= 6
y
>
< dɛ > 2 hh 2 vh
1
7> >
7 dsz =
<
0 0 0
6 Eh Ev Eh v v
dɛ x ¼ 1 dsx 2 vh dsy 2 hh dsz
¼6 7> dt >
z (5)
>
>
>
dg xy >
>
> 6 0 1 7> xy >
Eh Ev Eh
>
> >
>
> 6
6 0 0 0 0 7>
>
>
> yz >
>
>
> yz >
>
:
dg
; 6
Ghv 7> dt >
> Under conventional triaxial conditions (i.e., sx 5 sz ), and the
dg zx 7 dtzx ;
:
6 0 1 7 constraint in Eq. (2), Eq. (5) reduces to
6 0 0 0
Gvh
0
7
4 5
1 dɛ x ¼ 1 ð1 2 vhh Þdsx 2 vhv dsy (6)
0 0 0 0 0 Eh
Ghh
(1) Eq. (6) can be rearranged to yield the following equation:
where Ev and Eh 5 Young’s moduli in the vertical and horizontal ð1 2 y hh Þdsx 2 y hv dsy
directions, respectively; yhh and y vh 5 Poisson’s ratios for horizontal Eh ¼ (7)
dɛ x
strains from a horizontal and vertical strain, respectively; y hv
5 Poisson’s ratio for vertical strains from a horizontal strain; Gvh Combining Eqs. (4) and (7) yields
and Ghv 5 shear moduli in the vertical plane; and Ghh 5 shear
modulus in the horizontal plane. Fig. 1 shows the shear moduli
ð1 2 y hh Þ dsx 2 dsy
measured by bender elements. Eh ¼ (8)
dɛ x
In addition to Gvh 5 Ghv , the following two constraints lead to
only five independent parameters that must be identified
Next, y hh from Eq. (3) is substituted into Eq. (8), and the equation is
vhv vvh rearranged. The undrained horizontal Young’s modulus is then
¼ (2)
Eh Ev found from the following equation:
and in situ cross-hole tests (Ou et al. 1998; Kung 2007). The
geometry of the excavation site and the location of the sampling reduced by saturating the soil specimen at its residual effective
borehole are shown in Fig. 4. In the following section, numerical stress, which was measured by pressure sensors over a range from
simulations for the lateral soil movements are compared with a field 21 to 0 bar and porous ceramic filters in the suction-control system.
The residual effective stress is the effective stress remaining in the
soil specimen after sampling, storage, and handling (Ladd and
DeGroot 2003; Cho et al. 2007). The suction-control system con-
sisted of vacuum generators and electropneumatic valves that
produced the needed suction in the triaxial test [detailed information
on the suction-control system can be found in Teng and Ou (2011)].
Therefore, all the specimens tested in the study were saturated with
suction control to prevent swelling. An example of the suction-
control process is shown in Fig. 5. The suction of the soil sample was
measured at 26 kPa before saturation. Fig. 5 shows that the suction
was then removed in 13 stages by the suction-control system. The
cell pressure was increased simultaneously by the same amount of
removed suction. The effective mean stress of the sample thus was
kept constant during saturation, and the swelling effect from satu-
ration was greatly reduced.
The tests in this study were performed to investigate the anisot-
ropy of the stiffness of natural Taipei clay for strains ranging from
very small strains up to failure. Thus bender-element tests were
carried out continually during K0 consolidation and undrained
shearing. The trigger point of the bender-element test followed every
two steps of the K0 consolidation and occurred at specific strains
during undrained shearing. The consolidation and shearing rates
were 1 kPa=h and 0:002=h, respectively. Odometer tests were per-
formed to obtain the preconsolidation stress, the overconsolidation
Fig. 3. Local measurement assembly for a soil sample ratio, the compressibility index, and the swelling index of the sampled
soil.
AC-6 19 186.67
AC-7 21 192.36
AC-8 23 217.83
AC-9 25 242.99
AC-10 27 253.31
Odometer test O1–O10 9–27 — — — — —
Note: BE 5 bender-element test.
Table 3. Summary of Anisotropy Ratios for All the AC Tests at Very Small
Strains
Test name Depth (m) ARE ARG
AC-1 9 1.38 1.39
AC-2 11 1.41 1.42
AC-3 13 1.54 1.34
AC-4 15 1.48 1.44
AC-5 17 1.45 1.35
AC-6 19 1.46 1.34
AC-7 21 1.46 1.20
AC-8 23 1.42 1.23
Fig. 7. Anisotropic shear moduli versus effective mean stresses
AC-9 25 1.41 1.18
AC-10 27 1.40 1.15
Fig. 8. Anisotropy ratios for shear modulus measured in K0 Fig. 9. Anisotropy ratios for the stiffness during undrained shearing
consolidation for AC-1 test
modulus differed considerably from that of the shear modulus. The decreased gradually for depths greater than a GL of 217 m. The
reason for this behavior is discussed in subsequent paragraphs. swelling index was approximately 1:7.5 to 1:12 of the compressibility
Table 4. Parameters Used in USC Model and Their Physical Interpretations extension was used to simulate the stress path of a soil element
outside the excavation zone. An axial extension path was used to
Definition or physical Method for obtaining
simulate the behavior of a soil element within the excavation zone.
Parameter interpretation parameters
Fig. 11 shows that the degradation behavior of the Young’s modulus
suc Undrained shear strength Triaxial CK0 UC test for soft Taipei clay for different paths could be simulated using
for the triaxial condition a single set of degradation parameters.
(axial compression) Only seven parameters are required to fully describe the USC
Ei Initial Young’s modulus Bender-element tests or model: the undrained shear strength suc , the Young’s modulus at
or the maximum elastic small-strain triaxial test at small strain Ei , the failure ratio Rf , Poisson’s ratio y, the anisotropy
Young’s modulus strains of approximately ratio for the undrained strength Ks , and the degradation parameters m
1025 to 1026 and n. Among these parameters, suc and Rf are exactly the same as
Rf Failure ratio Same as Duncan and Chang those in the original Duncan-Chang model (Duncan and Chang
model: Rf 0:9 for soft clay 1970). The Young’s modulus at small strains Ei can be obtained from
Ks Anisotropic strength ratio, Triaxial CK0 UC test and small-strain tests, bender-element tests, or empirical correlations.
Ks 5 sue =suc ; sue is the CK0 UE test The anisotropy factor Ks is obtained from triaxial compression tests
undrained shear strength and extension tests. The degradation parameters m and n can be
for the triaxial condition obtained from unloading-reloading triaxial tests. Table 4 provides
(axial extension) the USC model parameters and their physical interpretations.
m, n Stiffness degradation Multiple unloading and The aforementioned soil constitutive models consider only iso-
parameters reflecting the reloading test tropic soil properties but incorporate anisotropy into the strength.
variation in the pore-water However, soils generally exhibit anisotropic stiffness in addition to
pressure generation with
strain Table 5. TNEC Excavation Process
Stage Day Excavation activities
1 156–162 Excavated to a GL depth of 22:8 m
2 164–169 Installed struts H300 3 300 3 10 3 15 at a GL depth
of 22:0 m; the preload for each strut was 784.8 kN
3 181–188 Excavated to a GL depth of 24:9 m
4A 217 Constructed B1F floor slab at a GL depth of 23:5 m
4B 222–238 Dismantled the first level of strut and constructed
the 1F floor slab; started the construction of the
superstructure
5 233–255 Excavated to a GL depth of 28:6 m
6 279 Constructed B2F floor slab at a GL depth of 27:1 m
7 318–337 Excavated to a GL depth of 211:8 m
8 352 Constructed B3F floor slab at a GL depth of 210:3 m
9 363–378 Excavated to a GL depth of 215:2 m
10 400 Constructed B4F floor slab at a GL depth of 213:7 m
11A 419–423 Excavated the central zone to a GL depth of 217:3 m
12A 425–429 Installed struts H400 3 400 3 13 3 21 at a GL
depth of 216:5 m in the central zone; the preload
of each strut was 1,177 kN
11B 430–436 Excavated the side zones to a GL depth of 217:3 m
12B 437–444 Install struts H400 3 400 3 13 3 21 in the two side
zones at a GL depth of 216:5 m; the preload
of each strut was 1,177 kN
13 445–460 Excavated to a GL depth of 219:7 m
457 Completed the superstructure
14 464–468 Cast the foundation slab
15 506–520 Constructed B5F floor slab at a GL depth of 217:1 m
16 528 Dismantled the second level struts
Fig. 12. Profile of excavation sequences and subsurface soil layers
Note: GL 5 ground level.
for the cross-anisotropic soil model. The anisotropy ratios of the generated by the bender elements, as shown in Fig. 9. The inherent
Young’s modulus and the shear modulus were chosen as the required anisotropy ratios of the shear moduli, the ARG values, were obtained
input parameters. The derivation is given in subsequent paragraphs. directly from the bender-element tests or from other empirical
Eqs. (2)–(4) can be used to obtain the following relations: expressions. All the anisotropy ratios at strains of 0.00001 for Tests
Eh
y hv ¼ × y vh ¼ ARE × yvh (10)
Ev Table 6. Input Parameters for USC Model
Soil layer Depth (m) g t ðkN=m3 Þ suc (kPa) K0 ARE ARG
y hh ¼ 1 2 y hv ¼ 1 2 ARE × y vh (11)
1-a 0–2 18.25 5.44 1.06 1.40 1.41
1-b 2–4 18.25 13.18 1.04 1.40 1.41
ARE × Ev
Ghh ¼ (12) 1-c 4–5.6 18.25 13.72 1.04 1.40 1.41
2ð2 2 ARE × y vh Þ 3-a 8–12 18.15 40.00 0.74 1.40 1.41
3-b 12–16 18.15 45.00 0.74 1.51 1.39
1 ARE × Ev 3-c 16–20 18.15 53.00 0.74 1.46 1.35
Gvh ¼ (13)
ARG 2ð2 2 ARE × y vh Þ 3-d 20–24 18.15 68.00 0.74 1.44 1.22
3-e 24–27 18.15 93.00 0.74 1.41 1.18
Eqs. (10)–(13) show that a simplified stress-strain matrix can be 3-f 27–30 18.15 113.00 0.74 1.40 1.15
used for Eq. (1) involving only four parameters, i.e., Ev , y vh , ARE , 3-g 30–33 18.15 118.00 0.74 1.40 1.15
and ARG . That is, the anisotropic stiffness can be determined using 5 35–37.5 19.13 150.00 0.81 1.40 1.15
the anisotropy ratios, which are all based on the familiar stiffness
parameters, i.e., Ev and yvh .
The anisotropy ratios of the undrained Young’s modulus and the
shear modulus for the anisotropic soil model were determined from
the test results presented herein. The anisotropy ratios of the un-
drained Young’s moduli, the ARE values, were calculated using the
following equation:
2 dsx 2 dsy dɛ x
Eh ¼
1 þ dsx 2 dsy dɛ x 2Ghh
2 dsx 2 dsy 2dɛ y =2
¼
1 þ dsx 2 dsy 2dɛ y =2 2Ghh
2dɛ x 1
qyvh ¼ ¼ at undrained conditions
dɛ y 2
4 dsy 2 dsx dɛ y
¼
1 þ dsy 2 dsx dɛ y Ghh
4Ev dsy 2 dsx
¼ qEv ¼
1 þ ðEv =Ghh Þ dɛ y
Fig. 9 shows that the anisotropy ratio of the undrained Young’s moduli
depended on the shear strain. There was such a great difference
between ARE and ARG with increasing strain because the Ghh value Fig. 13. Variation in undrained shear strength with depth for TNEC
used to calculate the ARE was obtained from the bender-element excavation
tests, which are always very small-strain tests. However, the value of
accurately and completely (Ou et al. 1998, 2000a, b). Fig. 4 shows i.e., the drained condition, partial drainage generally increases the
the trapezoidal excavation site. The length and width of the exca- magnitude of soil deformations (Hashash and Whittle 1996).
vation site were approximately 61–105 and 43 m, respectively. The
depth of the excavation was 19.7 m, and the diaphragm wall was
0.9 m thick and 35 m deep. The basement was completed using a top-
Soil Parameter Inputs
down construction method in which the diaphragm wall was sup-
ported by 0.15-m-thick solid concrete floor slabs. There were five The numerical analysis was performed in PLAXIS 2D, corre-
inclinometers and several surface-settlement measurements on the sponding to a plane-strain condition. This assumption was appro-
main observation section. Inclinometer I-1 was used to measure the priate because the corner effect was not pronounced at the main
deflection of the diaphragm wall during the excavation. Inclin- observation section, as mentioned in the case-history introduction.
ometers (SI-1 to SI-4) were installed at 2, 8, 16, and 22 m away from Table 6 lists the input parameters for the isotropic and anisotropic
the diaphragm wall. This arrangement allowed for lateral soil models. The anisotropy ratios used in the anisotropic model were
Fig. 14. Comparisons between isotropic and anisotropic model predictions for wall deflection and ground-surface settlement
showed that the ratio of the undrained shear strength in extension and Analysis Results
compression Ks was equal to 0.75 for both models. The degradation Figs. 14 and 15 compare the isotropic and anisotropic analysis results
parameters m and n were 0.225 and 1.299, respectively. The failure with the field measurements. Figs. 14(a) and 15(a) compare the an-
ratio Rf was equal to 0.9, which was identical to the value used in the isotropic model prediction with the field measurement; Figs. 14(b) and
Duncan-Chang model (Duncan and Chang 1970). 15(b) show the ratio of the simulation results from the isotropic and
The sandy layers were treated as drained material in the analysis. anisotropic models.
Sandy soil was simulated with the Mohr-Coulomb effective drained Fig. 14(a) shows that the wall deflections of the first two excava-
analysis using the following input parameters: the effective cohesion c9, tion stages (Stages 1 and 3 in Table 5) predicted by the anisotropic
Fig. 15. Comparison between isotropic and anisotropic model predictions for lateral soil movement outside excavation
model were slightly smaller than the field measurements, implying Conclusion
that the predicted small-strain stiffness may have been slightly
higher than the actual value. However, the computed deformations In this study, the stiffness anisotropy of soft silty clay was measured
for the first two stages were satisfactory. From Stage 5 to Stage 13, by performing CK0 UC triaxial tests on tube samples. The tests
the predictions of the anisotropic model were near to the field showed that both the Young’s modulus and the shear modulus of
observations for both the maximum wall displacement and the shape Taipei silty clay were anisotropic. All the anisotropy stiffness ratios,
of wall-deformation curves. The ground-surface settlements pre- i.e., Ghh =Gvh and Eh =Ev , were larger than unity. The anisotropy
dicted by the anisotropic model were larger than the field meas- ratios for the shear modulus ranged between 1.15 and 1.44. The
urements in the early stages, i.e., Stages 1–9. However, the anisotropy ratios for the Young’s modulus ranged between 1.38 and
settlements predicted by the anisotropic model in the last two stages 1.54 at a strain of 0.00001. Thus the stiffness anisotropy is sig-
were generally near the field measurements. The predictions for the nificant, and incorporating it into the soil model can improve nu-
maximum settlement and the settlement beyond the location of merical predictions on excavations and tunneling.
the maximum settlement were both accurate. Fig. 14(a) shows that A soil model that incorporates anisotropy into the stiffness was
the location of the maximum ground-surface settlement was also developed in this study. For simplicity, the anisotropy of clays was
accurately predicted. generally assumed to be cross-anisotropic. The developed aniso-
Fig. 14(b) compares the predictions for the wall deflection d and tropic soil model is only applicable to soft clays under undrained
the ground-surface settlement S from the isotropic model and the conditions. The predictions of the anisotropic model were nearer to
anisotropic model. Both ratios diso =daniso and Siso =Saniso were always the field observations than those of the isotropic model. However,
higher than 1.0 at the different stages. Fig. 14(b) shows that the ratios from a practical perspective, both the anisotropic model and the
ranged mainly from 1.10 to 1.25, implying that the isotropic model isotropic model make reasonable predictions for wall deflection. The
overestimated the ratios by 10–25% relative to the anisotropic model difference between the anisotropic and isotropic models for wall
predictions. The ratios were near 1.10 at the positions where the displacements at the final excavation stage was 12% of the wall
maximum movement occurred. deformation predicted by the An-USC model.
Fig. 15(a) shows the lateral soil movement predicted by the A soil model that incorporates small-strain behavior and stiffness
anisotropic model for the main observation section outside anisotropy was used to predict the geometry of the ground-surface-
the excavation zone. For the soil movement at SI-1 and SI-2, the settlement troughs. The results corresponded closely with the field
simulated results were slightly smaller than the measured data in measurements at the TNEC site. The analysis indicated that both
the first two excavation stages, similar to what was observed for the nonlinear stress-strain behavior at small strains and the stiff-
the wall deflection. Thereafter, the maximum lateral soil move- ness anisotropy should be considered in simulating the surface-
ments at SI-1 were accurately predicted by the anisotropic model. settlement troughs. Considering the stiffness anisotropy of soils
For SI-3 and SI-4, the soil movements were relatively small, and can also help to improve the accuracy of simulations on lateral soil
the maximum movement was generally less than 4 cm. The de- movement beyond the diaphragm wall. The differences between the
veloped anisotropic model satisfactorily predicted the maximum predictions of the anisotropic and isotropic models for the ground-
magnitude and shape of the lateral soil movement at SI-3 and SI-4. surface settlements and the lateral soil movements behind the wall
The simulated results for the lateral soil movements dh from both ranged between 10 and 43% of the ground movements predicted by
models were compared and plotted in Fig. 15(b). The ratios of the the An-USC model.
predictions of the isotropic model to those of the anisotropic model
dh,iso =dh,aniso ranged from 1.0 to 1.25 at SI-1 and SI-2. For SI-3 and
SI-4, the ratios ranged between 1.1 and 1.43. That is, the predictions
Notation
for dh by the isotropic model were higher than those predicted by the
anisotropic model by 10–43%. Table 7 summarizes the maximum
The following symbols are used in this paper:
difference between the numerical results and the field observations
of the movements, i.e., the wall deflections, the ground settlements, ARE 5 anisotropic ratio for Young’s modulus;
and the lateral soil movements. The table shows that the anisotropic ARG 5 anisotropic ratio for shear modulus;
model predictions for all the movements were closer to the field CK0 UC 5 triaxial tests of K0 consolidation and axial
measurements than the isotropic model predictions. The table also compression;
shows the ratio of the difference between the predictions of the two Cc 5 compression index;
models to the results from the anisotropic model. The ratio ranged Cs 5 swelling index;
from 0.11 to 0.28. The maximum ratio was obtained at SI-4 because c9 5 cohesion;
the smallest movement occurred at this position. These ratios are Eh 5 undrained Young’s modulus in horizontal
a quantitative measure of the improvement provided by the an- direction;
isotropic model. Eu,sec 5 undrained secant Young’s modulus;