100 .....-�---T"""---,r---"'?'--.----,,--.----, 0.

4
Shaking table tests conducted by Fraser Delta sand
De Alba et al (1976); - _ viat_ha_yalan_
_(V_aid_&S 996)_
_ 1_.._
� 80 ,__ +- ------+--------"'-' cr'3c =
A
0
Relative Density= 90%
Relative Density= 82%
CQCO
�i � 0.3 ...__-+----+----+-�50 kPa
100 kPa
"" '-
60 -- -
-----+- --+---""--+��__, .s 200 kPa
111 Relative Density= 68%
£� .s 400 kPa
0.3
8. Relative Density= 54%
initial confining
��
� � 40 i--- -----t----;,"'-----c;a-"7
-
0.2

pressure = 8 psi (55 kPa)

S
�- .8 0.1

, 6 20 1-------0,LA--:77"""---1----+- ----l
- 0::
0::
Cl)
Cl)
0
0.2 100 200 300 400 0 20 40 60 80 100
Cl) Effective consolidation stress, a'3c (kPa) Relative density, DR (%)

0
Figure 20. Cyclic triaxial test results for clean Fraser Delta sand, showing
(.) that cyclic stress and the CRR cause 3% shear strain in 10 uniform cycles at
0.1 DR values of 31-72% and effective consolidation stresses of 50-400 kPa
(original data from Vaid and Sivathayalan 1996).

The CRR of sand increases with increasing relative density, as

0.0 L-----1..-..L-...J_..l--l...J....1...!.-!--.....L..--'---....._,1--1....J-1,.......i..--........--'-................,
1 10 100 illustrated by the shaking table test results in Figure 19.
Number of cycles to cause initial liquefaction
The CRR of sand also depends on the effective confining stress,
which reflects the fact that the tendency of sand to dilate or contract
Figure 19. The CSR required to reach initial liquefaction (ru = 100%), depends on confining stress. This aspect of behavior is illustrated by
from shaking table tests by De Alba et al. (1976). the ICU cyclic triaxial test results on Fraser Delta sand by Vaid and
Sivathayalan (1996), shown in Figure 20. The plot at left shows the
whereas the CSR for isotropically consolidated cyclic triaxial tests is cyclic stress (qcyc /2) required to reach 3% shear strain in 10 cycles
defined as the maximum cyclic shear stress (qc y c /2) divided by the versus the effective consolidation stress for specimens at D R values
isotropic consolidation stress (q3 J, of 31, 40, 59, and 72%. This plot shows that cyclic strength increased
qc c with increasing consolidation stress for all values of D R , but the
CSR= y (10) shape of these relationships ranged from being nearly linear at the
2o-3 c lowest D R to being most strongly concave at the highest D R . The plot
The CSR that is required to reach liquefaction in a specified num­ at right shows the ratio of qcyc /2 divided by o-3 c that is required to
ber of loading cycles may also be called the sand's cyclic resistance cause 3% axial strain in 10 cycles (i.e., the CRR at 10 cycles) versus
ratio (CRR), for notational convenience. The relationship between D R for different consolidation stresses. This plot shows that the CRR
the CRR and N, within the range of cycles of interest for earthquake increased with increasing D R but that it also decreased as the effective
engineering, can generally be approximated with a power function as consolidation stress was increased from 50 kPa to 400 kPa for D R
values greater than 30%.
CRR=a· N-b (11)
Seed (1983) introduced the overburden correction factor (Ka- )
where the parameters a and b are determined by regression against as a way to represent the dependence of the CRR on consolidation
the experimental data. Note that the CRR versus N relationship plots stress, with Ka- defined as
as a straight line on a log-log plot, whereas it plots as a curve on the CRRa-c'
semilog plot format of Figure 19. The parameter b for clean sands Ka- = (12)
CRRa-�=l
is typically about 0.34, whereas the parameter a depends on a wide
range of factors. The dependence of the CRR on N means that any where CRRa- ' is the CRR of a soil under a specific value of effective
reference to a sand's CRR must specify N. consolidatio� stress o-�, and CRRo-�=l is the CRR of the same soil

28 29
 120 --------..---�-"'T"'""-..,.....--,--""T'"-...,..--, 1.4 .--,---,r-...,....-,---:::::z=::::z=:::c:::::::z:=:z=:::::z:::,
Vaid & Sivathayalan (1996)
Fraser Delta sand, DR = 72%, o DR =31%, simple shear
cr '3c = 50, 100, 200, 400 kPa □ DR = 72%, simple shear
(Vaid & Sivathaya/an 1996)
• DR= 31%, triaxial
11111 DR= 72%, triaxial
A

0.8
'::!t:._b

0.6
A A

0.4
0 u:;;..___......__......________________

0 100 200 300 400 500

Effective consolidation stress, cr'3c (kPa) 0.2 Field samples summarized
A
in Seed & Harder (1990)
Figure 21. The amount of cyclic stress required to cause 3% strain in
10 uniform cycles versus effective consolidation stress in ICU cyclic 2 4 6 8 10
triaxial tests on Fraser Delta sand.
Vertical effective stress, cr' JP8

Figure 22. Comparison of Kc, relationships with data

when er� = 1 atm (""' 100 kPa). The definition of the Krr factor is illus­ from reconstituted Fraser Delta sand specimens and
trated in Figure 21, which shows the cyclic stress required to reach 3% various field samples.
strain in 10 cycles versus the effective consolidation stress for Fraser
Delta sand at DR = 72%. The relationship between cyclic stress and arrive at a relative state parameter that provides improved correlations
consolidation stress is curved, thus its secant slope (which is the CRR) with the shear behavior of sand. Determining the state or relative state
decreases with increasing consolidation stress. For example, the data parameters requires detailed experimental testing to define a sand's
in Figure 21 show that the CRR at an effective consolidation stress CSL, its minimum and maximum void ratios, and its in-situ void
of 400 kPa is 18% smaller than at 100 kPa. This slight curvature is ratio. The natural heterogeneity of sand deposits, however, makes it
analogous to the slight curvature in the failure envelopes for drained impractical to perform sufficient experimental tests to define CSLs
shearing tests and the associated observation that peak (secant) effec­ for all the different zones within a natural deposit.
tive friction angles decrease with increasing confining stress. Boulanger (2003a) introduced an index for representing the rela­
The Krr factor varies with D R and with the test device (i.e., tive state parameter. This index, �R, is defined in Figure 23 and repre­
simple shear versus triaxial) and appears to be different for freshly sents the difference between the current DR and an empirical critical
reconstituted laboratory specimens as compared with tube samples state D R (denoted DR , cs ) for the same mean effective normal stress
of natural soils, as shown in Figure 22. (p'). The empirical CSL and the expression for DR , cs in Figure 23
The dependence of the CRR on both DR and confining stress can were derived by using Bolton's (1986) relative dilatancy index URD).
be related to the sand's "state" (position) with respect to its CSL. Been The parameter Q in Figure 23 determines the stress at which the CSL
and Jefferies (1985) introduced the state parameter(�) as a measure of curves sharply downward-which indicates the onset of significant
state, where � is the difference between the current void ratio (e) and particle crushing-and the parameter's value depends on grain type,
the critical-state void ratio (ecs) for the current value of p'. Komad with Q � 10 for quartz and feldspar, 8 for limestone, 7 for anthracite,
(1988) later showed that the value of � can be normalized by the and 5.5 for chalk (Bolton 1986). The resulting �R parameter pro­
difference in the maximum and minimum void ratios (emax - em in) to vides a rational means for including the interaction between relative

30 31
 Seed and Chan (1966) gave the highest ratios of Teye /s u . These speci­ Empirically Estimating CRR by Using a Consolidation Stress
mens were partially saturated and tested in unconsolidated-undrained History Profile
conditions, so their state of effective stress was not known. Cyclic strengths may be similarly computed by combining the
The triaxial and direct simple shear (DSS) tests gave comparable empirical (reye /s u )M= 7.5 relationships with the empirical relation­
Teye /Su ratios for the natural silts and clays, whereas the triaxial tests ships previously described (Section 6.2) that relate Su to consolidation
on tailings materials appeared to give Teye /S u ratios that were about stress history. The resulting expression for CRRM= 7.5 is
,15-20% lower than those obtained in DSS tests on tailings materials.
It is clear that the data summarized in Figure 133a are insufficient
CRRM= 7.s = 0.80 · S · OCRm (107)
to clearly define all the various factors that may affect the Teye /S u For homogenous sedimentary clays (CL and CH), the simplest
ratio, such as age, PI, soil type, OCR, and test type. Despite these representation may be to use S = 0.22 and m = 0.8 (Ladd 1991), so
uncertainties, the data for natural soils do tend to fall within relatively that the CRRM= ?.s is estimated as
narrow ranges, with (reye /su )N= 30 = 0.83 representing a reasonable
CRRM=7.s = 0.18 · OCR0 ·8 (108)
average for natural clay-like soils subjected to direct simple shear
loading conditions. Measured values of (Teye /aiJN =30 for the several normally con­
The CRR for clay-like fine-grained soils beneath level-ground solidated soils from Figure 127 are plotted versus PI in Figure 133b,
sites can then be estimated as from which the following observations can be made. The tailings
materials had (reye /aiJN=30 values that are similar to those for the
natural clays, despite their differences in PI and age. The cyclic DSS
(105) tests appear to give (reye /aiJN= 30 values that are about 20% smaller
than those obtained in cyclic triaxial tests. The monotonic and cyclic
shear stresses in triaxial tests were computed as Teye = qeye /2, which
where Cw is an adjustment factor for the effects of two-directional means that, if the shear stresses were instead computed for the even­
cyclic loading, as produced by the two horizontal components of tual shear plane as Teye = (q /2) · cos(¢/), then the (reye /aiJN=30
earthquake shaking. The value of Cw was estimated to be 0.96 values for triaxial tests would have been about 15% smaller (i.e.,
for clays, versus the value of 0.90 recommended for sands by Seed ¢/ � 32° ), and the difference between the DSS and triaxial test re­
(1979b). These relative Cw values reflect the fact that the second sults in Figure 133b would have been very small (note that the ratio
direction of cyclic loading is less damaging for clays than for sands, Teye /s u for triaxial tests is the same for either interpretation of shear
which is directly related to the fact that the CSR versus the number of stresses). The one natural silt (MH) had the highest (Teye /aiJ N =30
cycles to failure relationship is flatter for clays than for sands. With value, which may be attributable to its very high PI. However, the
this value for Cw , the CRRM=7.5 for clays can then be estimated as other data show no apparent trend with PI.
For practical purposes at this time, the (Teye /aiJ N=30 ratio might
reasonably be estimated as O .183, independent of PI, for normally con­
(106) solidated clay-like fine-grained soils subjected to one-dimensional
direct simple shear loading (as shown in Figure 133b). The corre­
sponding CRRM= 7.5 value for two-dimensional shaking would then
For tailings materials, the above CRR estimate should tentatively be approximately 0.18, independent of PI. This value is consistent
be reduced by about 20%, as suggested by the data in Figure 133a. with the above derivation based on S = 0.22, (reye /su )N= 30 = 0.83,
In many situations, the uncertainty in the Su profiles will be and Cw = 0.96.
greater than the uncertainty in the (reye /su )M= 7.5 ratio, but when the For sedimentary deposits ofplastic silts and organic soils that plot
uncertainty in the (reye /s u )M= 7.5 ratio is important, a detailed cyclic below the A-line on the Atterberg limits chart, Ladd (1991) suggested
laboratory testing program would be beneficial. that S would more typically be about 0.25, rather than the 0.22 value

198 199