1 s2.0 S003808062032429X Main
1 s2.0 S003808062032429X Main
1 s2.0 S003808062032429X Main
1979
Japanese Society of Soil Mechanics and Foundation Engineering
ABSTRACT
When overconsolidation is induced in a horizontal deposit of sand by the removal of
surcharge pressure or by other events, the lateral deformation is prevented and the coef-
ficient of earth pressure at rest, K 0 , increases. To evaluate effects of the overconsolidation
ratio as well as increased K 0-value on the liquefaction resistance of sand deposits, several
series of cyclic torsion shear tests were carried out on samples of clean sand over-
consolidated to different degrees under various K 0-conditions. The results of these tests
were summarized in an empirical formula. This formula is used to estimate the cyclic
strength of sand at any given OCR- and K 0 -values, based on the cyclic strength for iso-
tropically and normally consolidated samples.
Key words: at rest pressure, dynamic, earthquake, laboratory test, liquefaction, over-
consolidation, repeated load, torsion
IGC: D6/D7
INTRODUCTION
It has been found that a sample of sand previously subjected to a higher effective con-
fining stress than that applied at the initial stage of cyclic testing exhibits higher resistance
to liquefaction than the sample without such stress history. Ogawa (1973) performed a
series of cyclic triaxial tests on samples of clean sand subjected to different degrees of over-
consolidation and found that a sample with a relative density of 55% subjected to an over-
consolidation ratio of 5 showed approximately twice as much resistance to liquefaction as
a sample at the same relative density but without any overconsolidation. Seed and Peacock
(1971), using cyclic simple shear tests, found similar increases in strength due to overcon·
solidation. While investigating the cyclic behavior of sand at the Ekofisk tank site, Lee
and Focht (1975) also found that the cyclic strength of the overconsolidated dense sand was
greater than that of the normally consolidated sand. Seed, Arango and Chan (1975) in-
troduced a correction factor to take into account the effect of overconsolidation in their
method of evaluating the liquefaction potential of sand. Ishihara, Sodekawa and Tanaka
(1978b) showed that two clean sands overconsolidated to OCR=2 exhibited approximately
30% increase in cyclic strength over the cyclic stength of the same sand in the normally
consolidated state. Ishihara and Okada (1978 a) made a detailed in vestiga ti on by means of
static triaxial tests on the mechanism by which an increase in cyclic strength is induced in
sand due to overconsolidation. Overconsolidation is, therefore, recognized as one of the more
important factors inducing stress history effects in the response of sands subjected to cyclic
loading.
Overconsolidation in a horizontal deposit of sand is generally produced by removal of an
overburden pressure or by changes in the level of the ground water table. In both cases,
loading and unloading cycles take place with no lateral displacement. In this constrained
deformation mode, the lateral stress always changes so as to produce an increase in the
ratio of lateral to vertical effective stresses, K 0 , compared to the K 0 -condition in the normally
consolidated state. Therefore, the generation of an overconsolidated state is generally
accompanied by an increase in the K 0 -value in a horizontal deposit of sand. Consequently,
in evaluating the liquefaction potential of the sand deposit brought to an overconsolidated
state, it becomes necessary to take into account not only the effect of overconsolidation but
also the effect of increased K 0 -value that is taking place concurrently. In this investigation
in order to evaluate the effects of these two factors, OCR and K 0 , separately, several
series of cyclic torsion shear tests were performed on a sand overconsolidated to different
degrees with different K 0 -conditions. The content of this investigation and the experimentally
obtained results are described below.
TEST SCHEME
Overconsolida tion of test specimens was achieved in two ways. In the first series of tests,
effective confining pressure was increased and, then, reduced to achieve the desired over-
consolidation ratio. The K 0 -values in this series of tests were kept constant during the
process of creating an overconsolidated state. The K 0 -values employed in this test series
were 0. 5, 1. 0 and 1. 5 and the overconsolidation ratios were 1, 2, 3 and 4. The stress paths
used for obtaining different overconsolidation ratios under given K 0 -conditions are shown in
Fig. 1. All the cyclic torsion tests were planned to study the effect of overconsolidation on
the cyclic strength of sand for a given K 0-value.
"''"'
.§
5.0
---.
Ko=1-5
z 4.0
::s:::
'b
2.
~ 3.0
(./)
(/)
(lJ
.....
.....
(/)
2.0
Cl
c0
N
·;:::
I
0 1.0
0
Vertical stress , (Jv' (x1a2KN/m2 )
Fig. 1. Stress paths for obtaining different overconsolidation
ratios under identical K 0 -conditions
'"' OCR =2
1z K~ : Ko-value in normally
::s::: consolidated state
'b
B
3.1.0 1- - - -
6
~
.....(/)I-
Cl 0-5
.....
c Ko= K~ (OCR)0· 42
0
.!::! Schmartman(1975)
l-
o
I
0 1.0 2.0
Vertical stress , <JV' (x1dKN/m 2
)
In the second series of tests, the specimen was first consolidated anisotropically to an
effective vertical stress <1v'=200kN/m 2 and an effective horizontal stress ah'=100kN/m 2 ;
producing a condition of K 0 =0. 5, as represented by a stress path from point A to B in Fig.
2. The effective vertical stress was then reduced to av'= 100 kN/m 2 in drained conditions
from point B to C in Fig. 2, while preventing any lateral displacement from occurring. The
value of lateral stress at point C corresponding to the end of the axial stress reduction was
measured and the K 0 -value was determined. The K 0 -value thus obtained with OCR=2
ranged between 0. 71 and 0. 89 as indicated in Fig. 2. Also shown in Fig. 2 are the empirical
relationships, proposed by Sherif et al. (1974) and Schmertmann (1975), for K 0 -values of
. 0·6
( Q) [']: OCR= 1
~6
~
0.5 Ko =0·5 .
0: ,, =2
: ,, =3
Dr :::55 °lo
0.4
0
.....
0 0.3 ~• I OCR:3
~ ~
L-
(fl
(/)
(J)
~
.....
L- 0.2
(fl
OCR=1/ m ·
u
u;:>.
0.1 0-
u
0 1 10 100
Number of cycles to initial liquefaction
( b) Ko =1·0 G : OCR= 1
0 : ~ =2
~ 0.51----- -~ =3
0
~
~ vr--OCR:4 •:
e,; ~ :4
~" 0.4 \ ""''\_A Dr =55 °lo
~~~.
~~·.0··~
0 G
:;::;
~ 0.3
(fl
(/)
(J) ~. 0
~ 0.2 . G
u OCR :3
u;:>. 0.1 ! - - - - - -
u
0 1 10 100
Number of cycles to ini tia[ liquefaction
o.s
(C) Ko=1·5
0.5
6
15- Q.4
0
....c1
L- 0.3
(/)
lll
(J)
G :ocR=1
.....(/)L- 0.2 0: ,, =2
• : ~ :3
\..)
Dr =55 °lo
u
»
0.1
u
0
10 100
Number of cycles to initial liquefaction
Fig. 3. Cyclic stress ratio versus number of cycles
required to cause initial liquefaction
TEST RESULTS
The results of the first series of tests in which the samples were tested with varying
degree of overconsolidation under a constant K 0 -value are shown in Fig. 3. The cyclic stress
ratio, defined as the amplitude of cyclic torsional shear stress, 7:', divided by the initial
vertical effective stress, av' is plotted as the ordinate. The number of cycles required to
~I
0
:;:;
E 0.3
~ / \ocR=2
OJ
~ 0.2
-~
~· 0.1
u u
u
g, 0.1
1___
{},
~ _ \ocR=1 °
u
0·3
0.2
UG ~·5
~ •
u OCR:1
0.1 u 20 cycles of
(.) » 0.1 uniform loading
>. u
u Dr :55 °lo
cause initial liquefaction under given values of cyclic stress ratio is plotted as the abscissa.
Each diagram of Fig. 3 shows that, under a constant K 0 -condition, the cyclic stress ratio
causing initial liquefaction in a given number of cycles increases with an increase in OCR-
value. It is also noted that the cyclic strength is increased with an increase in K 0-value if
the overconsolidation ratio is kept unchanged. Similar findings have been reported by
Ishihara et al. (1977).
The amount of pore water pressure that has built up in 20 uniform load cycles may be
read off for the stress ratio still small enough not to induce initial liqufaction. This pore water
pressure may be called residual pore water pressure, because it is the pore water pressure
that remains in the sample after it is subjected to 20 cycles of uniform loading with a given
stress ratio. The cyclic stress ratio, 7:/<Jv', versus the residual pore water pressure, Ur, which
has been produced in the tests by applying the 20 load cycles, is plotted in Fig. 4. In this
figure the residual pore water pressure was normalized to the initial effective vertical
stress. This ratio will be referred to as the residual pore water pressure ratio. Fig. 4 shows
that the cyclic stress ratio required in order to generate a certain amount of residual pore
water pressure in 20 cycles of uniform loading under a constant K 0-condition always
increases with increases in the OCR-value. Similarly the cyclic stress ratio required to
produce a certain a amount of residual pore water pressure under a constant OCR-condition
increases with increases in the K 0 -value.
The results of the second series of tests are shown in Fig. 5, as plots of the stress ratio
required to generate various levels of residual pore water pressure in 20 cycles, together
with the K 0 -values employed in each of the tests. As explained previously, the K 0-test
condition was induced in the sample by reducing the effective vertical stress while preventing
any lateral strain. For comparison, curves from the first test series for K 0 =0. 5 and K 0 =
1. 0 which bound the test values of K 0 are also shown. It may be seen in Fig. 5 that the
measured data points fall approximately within the limits bounded by the two curves per-
taining to the tests in which the K 0 -values were controlled to be 0. 5 and 1. 0.
0 2 3 4 5 6 7
Overconsolidation ratio • OCR
Fig. 6. Relationship between cyclic strength and OCR-value
(/)
(]) (/)
u>."O c
(]) Fuji river sand
UCE •: Ko=0·5
o 3
N
O·-
U • : Ko=1·0
c
·- "O (/)
~ • : Ko=1·5
..c.2
+' 0 "O
Or= 55 °lo
Ol"O (])
c ·- +'
~-o.g
......
Ill
(/)
c:=
2
u8g
·-I.. c
u (]) 0
~~u
..,_
0
0
c~
(])-
<lJ 0
(~)QC I (~)NC= ~-0-cR-
:;:;~ E
&~g
0 2 3 4 5 6 7
Overconsolidation ratio
Fig. 7. Cyclic strength increase due to overconsolidation
(5J,)
dv OC
= .JOCR (.!i,)
dv NC,
( 1)
where 7:'z is the amplitude of cyclic shear stress required to cause initial liquefaction under
20 cycles of uniform loading and av' is the vertical effective stress just prior to cyclic test-
ing. This relationship is shown in Fig. 7, where it may be noted that Eq. (1) fits the tests
data rather well in the range of K 0 -values between 0. 5 and 1. 5. It has also been shown
that Eq. (1) holds valid for any given number of cycles other than 20 for the cyclic stress
ratio causing initial liquefaction.
A previous study has shown that the cyclic strengths of isotropically and anisotropically
normally consolidated specimens are correlated by the formula (Ishihara .et al., 1977).
(~) _
l+2Ko
dv A 3 dv I
(-'!J,-) 2) (
where (-r:ifav')r and (-r:ifav') A represent the cyclic strengths in isotropic and anisotropic states
of consolidation, respectively. In order to show whether Eq. (2) can be applicable for over-
consolidated sand as well, the cyclic strengths data presented in Fig. 6 were replotted as in
Fig. 8. The ratio of cyclic strength between anisotropically and isotropically consolidated
specimens is plotted versus K 0 -value. Fig. 8 shows that, for all overconsolidation ratios
employed in the test program, the cyclic strengths of anisotropically consolidated samples
(K0 ~l. 0) were correlated with the cyclic strength of isotropically consolidated sample (K0 =
1. O) by Eq. (2).
Eq. (1) has been established only on the basis of the results of the present test program
in which the triaxial torsion shear test device was used. In order to see whether Eq. (1)
holds valid for the test data obtained from other test devices, the results of cyclic triaxial
tests on isotropically consolidated clean sands were collected from referenced sources and
plotted together in Fig. 9. The cyclic stress ratio causing initial liquefaction in 20 cycles is
~
(/) 0.. 2.0
(]) u E
~co • : OCR=1
~~ ~ Fuji river sand _. : OCR =2
o .- (])
N .j./' d 1 . 5 1 - - - - - - - + - - - - - - ;
• : OCR=3
c 0 lJ Dr=55 °/o
·-~:.::;
:52g
010 c
cu 0
~·a.;:. l.Qi-------
u; _g 'ij
-~ g~
-u·c:2
~ 0 8 0.5 1+2Ko
'+- ~·a_ 3
0 (]) 0
.2 3: .!::;
(5Q; ~
0::: ..O·-
0 0-5 1-5
Ko - value
Fig. 8. Cyclic strength increase due to K 0 -value
Q.5 >-----+----+--
---- --------0---
01 lll
c (])
'iii u
6 ~ Q.4 1-----+--
t:\',,,,,,,,"-
u o
ON
/"
,~,,-
.- Ko= 1·0
:g c 0.3 ,,..f{- .
L- ··-
zf/"
/
8l c 0: Ogawa et al (1969)
~ .~ 0-2 Dr=55°1o
u; uo r:i : Ishihara et al (1978)
ua;
:= :::i Q. 1 1-----4----+-----+-1
8: ,,
Dr =50-'5 5 °lo
'"
~rr ,
6':.::: • : Fuji river sand D,.=55%
0 2 3 4 5 6 7
Overconsolidation ratio
Fig. 9. Relationship between cyclic strength and OCR-value
plotted in this figure versus the overconsolidation ratio. Note that the cyclic stress ratio
in the case of the cyclic triaxial test is defined as the amplitude of the cyclic axial stress
divided by twice the initial confining pressure. This definition is equivalent to the cyclic
stress ratio, 7:z/av', at K 0 =1. 0. The test results of the present investigation are also presented
in Fig. 9. Fig. 9 shows that the cyclic strength determined both by cyclic triaxial shear
tests and cyclic torsional shear tests increases with the OCR-value in the same manner.
Each set of the test data presented in Fig. 9 was normalized by dividing the cyclic strengths
at given OCR-values by the corresponding cyclic strength of normally consolidated sample.
The results of the normalization, shown in Fig. 10, show that the cyclic strength of over-
consolidated sand is correlated to a reasonable degree of accuracy with the cyclic strength
of normally consolidated sand by the relationship of Eq. (1).
For most practical testing programs, the cyclic triaxial test on isotropically normally con-
solidated specimen is the most convenient and widely used procedure to determine the cyclic
strength characteristics of sand. When the cyclic strength for a given sand is known from
this testing procedure the cyclic strength of the same sand with the same density, but with
different degrees of overconsolidation, may be estimated by Eq. (1). When it is further
0 2 3 4 5 6 7
Overconsolidation ratio , OOR
Fig. 10. Cyclic strength increase due to overconsolidation
required to determine the cyclic strength of this overconsolidated sand in a state of anisotropic
consolidation with Ka-value different from the unity, Eq. (2) may used, because this
equation is valid, as indicated in Fig. 8, irrespective of the degree of overconsolidation to
which the sand has been previously subjected. Therefore, Eq. (1) and (2) may be combined
to give an equation correlating the cyclic strengths of isotropically normally consolidated sand
specimens with those of anisotropically overconsolidated specimens.
This equation has the form
( ;:,-) A,OC
1+2Ka
3
JocR(;i,) v I.NC,
(3)
where ( -rif <J v') A,oc denotes the cyclic stress ratio causing initial liquefaction in a given number
of cycles for anisotropically overconsolidated sand. (-rif<Jv') 1 ,Nc represents the cyclic stress
ratio similarly defined but for isotropically normally consolidated state of the same sand
compacted to the same density.
CONCLUSIONS
Using a triaxial torsion shear test apparatus, effect of overconsolidation on the cyclic
strength of clean sand was investigated under varying Ka-conditions. It was found that
the cyclic stress ratio causing initial liquefaction in a given number of cycles increases ap-
proximately in proportion to the square root of overconsolidation ratio in the range of OCR-
value of 1. 0 and 4. 0, irrespective of the Ka-value. By taking into account the effect of
Ka-conditions previously investigated, the result of the present study was incorporated in an
empirical formula which may be used to assess the cyclic strength of anisotropically over-
consolidated sands through the cyclic triaxial test results on isotropically nomally consoli-
dated specimen.
ACKNOWLEDGEMENT
The cooperation of Mr. Y. Ohkochi, graduate student of the University of Tokyo, in
running the experiments, was very helpful for the completion of this investigation. Dr.
K. Mori, Research Associate of the University of Tokyo, reviewed and commented on the
original draft of this paper. The authors wish to express their appreciation to these persons.
REFERENCES
1) Ishihara, K. and Yasuda, S. (1975): "Sand liquefaction in hollow cylinder torsion tests under
irregular excitation," Soils and Foundations, Vol.15, No. 1, pp. 45-59.
2) Ishihara, K., Iwamoto, S., Yasuda, S. and Takatsu, H. (1977): "Liquefaction of anisotropically
consolidated sand," Proc., 9th International Conference on Soil Mechanics and Foundation Engi-
neering, Vol. II, pp. 261-264.
3) Ishihara, K. and Okada, S. (1978a): "Yielding of overconsolidated sand and liquefaction model
under cyclic stresses," Soils and Foundations, Vol.18, No. 1, pp. 57-72.
4) Ishihara, K., Sodekawa, M. and Tanaka, Y. (1978b): "Effects of overconsolidation on liquefaction
characteristics of sands containing fines," Dynamic Geotechnical Testing, ASTM, STP 654, pp.
246-264.
5) Lee, K. L. and Focht, J. A (1975): "Liquefaction potential at Ekofisk tank in North Sea," Proc.,
ASCE, Vol.101, GT, pp.1-8.
6) Ogawa, S. (1973): "Dynamic shear test results and its application," Tsuchi-to-Kiso, Japanese
Society of Soil Mechanics and Foundation Engineering, pp. 33-40 (in Japanese).
7) Seed, H.B., Arango, I. and Chan, C. K. (1975): "Evaluation of soil liquefaction potential during
earthquakes," Report No. EERC 75-28, University of California.
8) Seed, H.B. and Peacock, W. H. (1971): "Test procedures for measuring soil liquefaction chara-
cteristics," ASCE, SM 8, Vol. 97, pp. 1099-1119.
9) Sherif, M.A., Ishibashi, I. and Ryden, D. E. (197 4): "Coefficient of lateral earth pressue at rest,"
Soil Engineering Research Report, No. 9, University of Washington, Seattle, U.S. A.
10) Schmertman, J. S. (1975): "Measurement of in situ shear strength," Proc. of the Conference
on In Situ Measurement of Soil Properties, Specialty Conference of the Geotechnical Engineering
Division, ASCE, Vol. II, pp. 57-138.
(Received January 22, 1979)