You are on page 1of 17

SOILS AND FOUNDATIONS Vol. 20, No. 4, Dec.

1980
Japanese Society of Soil Mechanics and Foundation Engineering

STANDARD PENETRATION TESTS AND SOIL


LIQUEFACTION POTENTIAL EVALUATION

FuMro TATSUOKA*, Tosmo IwASAKI**, KEN-rem ToKmA***, SusuMu YAsunA****,


MAKOTO HrRosE*****, TsuNEO IMAI****** and MAsASHI KoN-No*******

ABSTRACT
Several methods to evaluate cyclic undrained strengths of sandy soils from standard
penetration resistances (N-values) are reviewed and compared. It is shown that some
empirical equations which do not take account of the effects of grain size on N-values
can be rather conservative for fine or silty sands. When the effects of grain size on
N-values are adequately taken into account, strengths estimated from N-values are very
close to the measured strengths of undisturbed samples. A simplified procedures for
evaluating liquefaction potential of an element is proposed. An index showing integrated
liquefaction potential of a deposit is also proposed and examined by a liquefaction case
study.

Key words: dynamic, earthquake resistant, grain size, laboratory test, liquefaction,
penetration test, repeated load, sampling, sandy soil, sounding
JGC: D6/C3/C6

INTRODUCTION
Several practical methods to evaluate soil liquefaction potential of level grounds have
been developed. Considering that standard penetration tests have been widely conducted
for designing most of civil engineering structures, methods of soil liquefaction potential
evaluation which are based on blow counts by standard penetration tests (N-values) are
still useful tools for practical engineers. In any analytical method of liquefaction
potential evaluation, in situ cyclic .undrained strength or liquefaction strength should
be firstly evaluated. Presently available method to evaluate in situ cyclic undrained
strength based on N-values may be classified into the following three main categories
(see Fig. 1):
( 1) Method A, m which relative density is used as an essential parameter and

* Associate Professor, Institute of Industrial Science, University of Tokyo, 22-1, Roppongi 7,


Minato-ku, Tokyo.
** Head, Ground Vibration Section, Public Works Research Institute, Ministry of Construction,
Asahi 1, Toyosato-machi, Tsukuba-gun, Ibaragi.
*** Research Engineer, do.
**** Research Engineer, Kisojiban Consultants, Co., Ltd., 2-14-1, Ishikawa-cho, Ohta-ku, Tokyo.
***** Research Engineer, Soil Laboratory, Toa Harbor Works Co., Ltd., Yokohama-shi, Kanagawa.
****** Chief, Earthquake Engineering Division, Urawa Research Institute, OYO Co., Ltd., Urawa-
shi, Saitama.
******* Research Engineer, do.
Written discussions on this paper should be submitted before October 1, 1981.

This is an Open Access article under the CC-BY-NC-ND license.


NII-Electronic Library Service
96 T ATSUOKA ET AL.

1
firstly evaluated from N-values
N-value
a-:
K •
1
X100(%)11
I.--D,=--"c"~~":-
em., em.,
and other parameters. Then,
® grading*
triaxial strength (or if possible,
: static and dynamic : I
:stress-strain-time:-.,.--------
simple shear strength) is deter-
i.~_i?.~?_!~t_'______________ ,.
..

- Triaxial mined from relative density. In

' (:'' '---i-~ situ strength is obtained by cor-


recting triaxial strength. Seed
~
N-value and Idriss (1971) firstly proposed
®
~
grading
J' this method.
( 2) Method B, which is
N-va,Iue l
based on the empirical relation-
~',. J'----------------' ship between cyclic undrained
grading' h'OTE * : in general, not taken into account triaxial strength of undisturbed
Fig. 1. Three major categories for evaluating cyclic specimens and N-values (Ishihara
undrained strength from N-values and Ogawa, 1978; Tatsuoka et
al., 1978).
( 3) Method C, which is based on the empirical correlation between stress ratio causing
liquefaction in the field and N-values (Castro, 1975; Seed, 1979).
The authors have found by applying these methods to actual liquefaction potential
evaluation problems for young alluvial deposits and uncompacted hydraulic fills that the
effects of grain size on N-values are significant. They have also found that when these
effects are ignored any of above three methods can be misleading, especially for fine
sand deposits.
In this paper, the outlines of above three methods are firstly presented and then these
methods are used to evaluate cyclic undrained strengths of sand deposits at two sites,
both being uncompacted hydraulic fills underlain by young alluvial deposits. In the last
part, the previous liquefaction experiences in Japan are analyzed by these methods, and
a new parameter indicating liquefaction potential of a deposit is proposed.

THREE METHODS EXAMINED


Different methods will be compared in respect of the cyclic undrained triaxial strength
defined as
Rz=(~)
2 d/ Nc=20
(1)
which is the stress ratio (adp/2 ac') at the number of loading cycles Ne of 20 where
liquefaction takes place, where <Jdp is the single amplitude of deviator stress and ac' is
the effective confining stress.
Method A It is suggested in the simplified procedure by Seed and Idriss (1971) that the
value of Ri at a certain value of relative density Dr up to around 80% can be
evaluated as
Rz = (Ri)50 x Dr/50 ( 2)
in which (Rz) 50 is the value of Ri at Dr=50%.
The value of (Rz) 50 as a function of mean grain size (D50 ) suggested by Seed and Idriss
(1971) is shown in Fig. 2. Note that the value of (Rz) 50 shown in Fig. 2 are for Ne=
20 which were obtained by interpolating their original data. Recently, Martin et al.
(1978) have shown that the variation of Rz with grain size is mainly due to membrane
penetration effect. As shown in Fig. 2, the suggested strength by Martin et al. (1978)
is almost independent of D 50 • Therefore, it seems reasonable to estimate (Rz) 50 from
the strength curves shown in Fig. 2 as

NII-Electronic Library Service


SOIL LIQUEFACTION POTENTIAL 97

40 I I I

0.3~--------.-----.,.----.-------,

. ~- "
II
-
Saturated
Fine Sand
I Saturated
Coarse Sand -

----7--
By Seed and Idriss (1971)

0.2-
*•
--.::._.. Air-Dry
Fine Sand
?:+d µ+d I
µ-d
µ1

By Martin et al (1978); Corrected for µ-d

[~~
.~
membrane penetration effect Cl,)
I s ange of Data
~i
C1l ;
-20 - ean value Air-Dry and -
. andard deviation . Moist Coarse
D,=50%, N,=20, Initial Liquefaction 11"
Sand
~
0·8.'-o-2_ _ _ _0"""'.o=-5----,,o"":-.1---+-0.2::------;;o"".5--_.,..i.o I
-40 I I

Mean Grain Diameter Dso (mm) 0.1 0.2 0.5


Dso(mm)

Fig. 2. Cyclic undrained triaxial strength Fig. 3. Dr-Dr* and D 50 relationship by the
affected by D 5o data by Gibbs and Holtz (1957)

(3)
for the range of D 50 from 0. 06 to 1. 0 mm. By substituting Eq. (3) into Eq. (2), we get
Ri=RiDr =0. 0042 Dr ( 4)
m which Rivr means the value of Rz determined from Dr. The same equation was
proposed by Ishihara (1978). Seed and Idriss (1971) used the test results by Gibbs and
Holtz (1957) for evaluating relative density from N-values. Meyerhof (1957) showed
that the test results by Gibbs and Holtz (1957) on air-dry and moist coarse sand can
be well represented by
Dr=Dr* ( 5)
in which Dr*=210vN/(<lv'+70) (as distinguished from measured relative density Dr) and
<lv' is the effective overburden stress in kN/m 2 • Note that if <lv' is m kgf/cm 2, D 7 *=21
v'N/(<1v'+0. 7). By combining Eqs. (4) and (5), we get
Ri =RzA =0. 0042 x 210.y' N/ (<1v' +70) =0. 88.y' N/(<1v' +70) ( 6)
in which RzA means the value of Ri by Eq. (6). As discussed by Schmertman (1972),
it seems that Eq. (5) is questionable when applied to saturated fine sand which is most
susceptible to liquefaction, because Eq. (5) is based on the test results on air-dry and
moist coarse sand. It should be noted that the test results by Gibbs and Holtz (1957)
themselves shows that Eq. (5) underestimates considerably the value of Dr of saturated
fine sand as shown in Fig. 3. This fact was confirmed again for in situ conditions (see
Fig. 6 of Tatsuoka et al. (1978)).
Method B A more direct method which does not include the process of estimating
relative density was proposed by Ishihara (1977) based on the correlation between cyclic
undrained triaxial strength of undisturbed samples and N-values;
(7)
This has the same form with Eq. (6). However, it should be emphasized that in
deriving Eq. (7), relative density was not considered. Tatsuoka et al. (1978) modified
Eq. (7) by taking account of the effects of grain size on N-values. Based on the data
of cyclic undrained triaxial strength of undisturbed samples from young alluvial sandy
deposits and uncompacted hydraulic fills, Tatsuoka et al. (1978) proposed the following
equations.

NII-Electronic Library Service


98 T ATSUOKA ET AL.

Ri:Rin:O. 0042 Dr:-0. 225 logi 0 (D50 /0. 35) for 0. 04~D 50 ~0. 6 mm} ( )
8
Ri-Rin-0. 0042 Dr -0. 05 for 0. 6~D50 ~1. 5 mm
in which Rin means the value of Ri by Eq. (8) or Eq. (9). In this case, Rz is defined
for double axial strain DA of 5 or 6%. This can be rewritten as
Ri=Rin=O. 88{ JN/(av'+70)-0. 258 logi 0 (D50 /0. 35)} for 0. 04~D 50 ~0. 6 mm}
for 0.6~D ~1.5mm 50
9
Ri=Rin=0.88{JN/(av'+70)-0.0567} ( )
Eq. (9) means that N-values should be corrected at least both for av' and for D 50 • It
was emphasized by Tatsuoka et al. (1978) that Eqs. (8) and (9) are valid only for young
.alluvial sandy deposits and uncompacted hydraulic fills in which N-values are less than
around 20. It was also cited that the details of Eqs. (8) and (9) should be modified in
future by our improved knowledges.
Method C The most direct and the most empirical method is the one which is based
·on the study on liquefaction and non-liquefaction cases in the field. Seed (1979) showed
that for Magnitude 7-1/2 earthquakes the lower bound value of the stress ratio required
to cause cyclic liquefaction is approximately expressed by the equation

( Ta;) ~Ni/90
Ov l
(10)

in which Crav!av')i is the stress ratio which is 0. 65 times the ratio of the maximum
single amplitude of cyclic shear stress time history to the effective overburden pressure
<Jv' and Ni= {l-1. 25 logi 0 (av'/96) }N in which av' is in kN/m 2 • For a wide range of av',
Ni can be precisely related to Dr*=210JN/Cav'+70) as
0
Ni=CDr*/16) 2 a): 70 N (11)

On the other hand, the value of Crav!av')i in Eq. (10) can be related to triaxial strength
defined for the number of loading cycles Nc=Neq as

(~)
av
=Cr(adp/2ac')N.q
l
(12)

in which Cr is a correction factor. By the data shown by Seed (1979), Cr is 0. 63 for


K 0 =0. 5, which is the reasonable value for normally consolidated young alluvial deposits
and uncompacted hydraulic fills. By Seed (1979) Neq is equal to around 15 for M=7-1/2.
Tatsuoka et al. (1980) showed that Ri = (adp/2 a/)Nc= 20 can be approximately related to
(adp/2 ac')Nc=l5 as
(13)
The value of b of -0. 20 was obtained from the data of the large simple shear test by
De Alba et al. (1975). Hence, (20/15)b is 0. 944 in this case. Combining Eqs. (10), (11),
(12) and (13), we get

Ri =Rici =0. 944 X (aap/2 a/)N.=15 =0. 944 X Cl ( -r a1f-)


r Ov l

1 1 170 - '
=0.944X 0. 63 x 90 x av'+ 70 N-2.8N/(av +70)
(14)

m which Rici means the value of Ri by Eq. (14). It should be noted that Eq. (10) is
strongly influenced by the data from the Niigata Earthquake of 1964. It was shown by
the recent investigation by Ishihara et al. (1978) and other data that for the conditions
at Niigata, the mean grain size for the liquefied soil ranged mainly from about 0. 2 mm
to 0. 7 mm. This means that Eq. (10) is most applicable to clean medium sand deposits.
Pyke et al. (1978) has reported that Eq. (10) or Eq. (14) could significantly underestimate

NII-Electronic Library Service


SOIL LIQUEFACTION POTENTIAL 99

the cyclic undrained strength of fine sand or silty sand deposits. They showed that
there was no liquefaction in a hydraulic fill during the Long Beach Earthquake of 1933
(M=6. 3). The estimated field strength value of ('C'av!av')i from the accelerograph record
and some analyses was 0. 2 or more. The estimated averaged value of Ni was 10.
Following Seed (1979), when the effects of magnitude are taken into account, Eq. (10)
becomes

(~)
O'v'
=J!J__
78
l .
(15)

By making N 1 equal to 10 in Eq. (15), we get a strength ('C'avfav')i of around 0.128.


This value is significantly lower than the value of 0. 2 or more. The reported value
·of D 50 of the fill was mainly between 0. 15,...,_,0. 016 mm, which means that the hydraulic
fill consisted of mainly :fine sand or silty sand. From the facts described above, Pyke
et al. (1978) concluded that use of Eq. (10) or (15) may be rather conservative for more
silty materials.

A MODIFICATION OF EMPIRICAL METHOD C


By using the data from which Eq. (8) was derived, Method C will be modified to take
account of the effects of grain size on values of Ni or N-values. Fig. 4 shows the
relationship between mean grain diameter D 50 and the following parameter
DRi**=RzM-Rzc 1 (16)
in which RzM is measures strength (a dp/2 a c') at Ne= 20 for DA= 5 or 6 % of undisturbed
samples, Rzc 1 is equal to 2. 8N/(av'+70) by Eq. (14) and N is a representative blow
count by conventional standard penetration tests at a depth from where undisturbed
samples were obtained. The details of the methods of sand sampling and the triaxial
testing procedures are described in Tatsuoka et al. (1978). While a large scattering can
be seen in the data shown in Fig. 4, there is a trend showing the increase in DRi **
with the decrease in D 50 • The average line drawn in Fig. 4 is expressed by
DRi**=-0. 275 log io(D50 /0. 4) for 0. 04;£D50 ;£0. 6 mm}
7
DRi**= -0. 05 for 0. 6;£D50 ;£1. 5 mm (l )
Note that the value of DRi** by Eq. (17) is equal to zero for D 50 of 0. 4 mm, resulting
in RzM=Rci· D 50 of 0. 4 mm is well in the range of D 50 of the liquefied Niigata Sand
the data of which strongly influence Eq. (10). This means that the values of RzM
which were used to derive Eq. (17) well agree with Eq. (10) or (14). From Eqs. (16)
and (17) and by making RzM to
be Rz, we get Underwater Sampling
0.4
Rz=Rzc 2 =2. 8 N/(av'+70)
-0. 275 logi 0 (D50 /0. 4) 0

+
b
0.2
for 0. 04;£D 50 ;£0. 6 mm (18 a)
Ri=Rzc 2 =2. 8 N/(av' +70)-0. 05 ~
for 0. 6;£D50 ;£1. 5 mm (18 b)
"'
I o.o

in which Rzc 2 means the value Site DA(%)


oD

.
a: (xl0 2 ,kN/m2 )
of Ri by Eq. (18) and av' is m o• F 6 a; =0.37-0.77 4
{D
F 6 a; =0.5, 1.0
kN/m 2 • Eq. (18) means again G 6 a; =0.8-1.2
-o,4.___ _--1._ _-..1._ _......L._ _ _-1.-_ _...l-_ _...1-.---l
that on the basis of the data 0.02 0.05 0.1 0.2 0.5
shown in Fig. 4, N-values should Dso (mm)

he corrected at least both for av' Fig. 4. Measured values of DRz** as a function of D 50

NII-Electronic Library Service


100 T ATSUOKA ET AL.

and for D 50 • By using Eq. (11), Eq. (18 a) becomes


R 1c2=0. 0165 Ni -0. 275 log 10 (D50 /0. 4) =0. 0165{Ni -16. 7 log 10 (D50 /0. 4)} (19)
By replacing N 1 with Ni -16. 7 log 10 (D50 /0. 4) in Eq. (15) for taking account of the effect
of grain size on Nl> we get for M=6. 3

( ::~ ) i ~ 7~ [Ni -16. 7 log 10 (D50 /0. 4)] (20)

The data of the case reported by Pyke et al. (1978) are M=6. 3, Ni =10 and D 50 =0.15
.-....,Q. 016 mm, with the averaged value of D 50 being 0. 083 mm. By substituting these

Dso(mm)
0.01 0.02 0.05 0.1 0.2 0.5

N-value
10 20 o.o 0.4
I I
LEGEND
(a) Point A
\" R,T Rm
D Rm, t::,. R1c1
X RtA 'V R1c2
m RECLAIMED LAND
~
+10LYO BAY .· . ·. . / CHANNE0~.-'"-~~~ D,
(I ,x
N-value Dso
-l~l =Y&~ 2'/~~l:r;!~T ~r-!jL ~
"
f- xv~k
) "'
t :
~ I 6
~ x//
-20
-30 DILUVIAL DEPOSIT ~ -
>,
:::r: 4I I
A x o I
-
~
-40 I I
x A X
-so Pd
-
RECLAIMED LAND
100
oc m IIOOm
OB 0 0
•A
"'>
:.:;:"

1~ -
O BORING POINT WHERE N-VALUES WERE MEASURED ~ /x AX
LEGEND { e BORING POINT WHERE N-VALUES WERE MEASURED
AND SAND SAMPLES WERE TAKEN 10 -
1J . I
x
/
/(
I

I
"' x
I
Fig. 5. Cross-section and plan of Site A 0 50 100 D,(%)
I.a 1.5 .2.0 pd(g/cm 3)

(a)

Dso(mm)
Dso (mm)
0.02 o.os 0.1 0.2 0.5
o.oi 0.02 0.05 0.1 0.2 0.5
R 1 (N,=20)
R1 (N,=20) N-value
N-value 10 20 0.0 0.2 0.4 0.6
10 20 o.o 0.2 0.4
( c) Point I See a) for legend.
-R- ( b) Point F See a) for legend.

~ - ~·alue D, 0
""\\ T ~ b, X0\70

II 'I
I
\ I
I I I I
I \ D
\ I "" x
'i' ~
I I
fI "" x 'VO\
\
I \
\ I \
\ I \
I \ I I
I
'T
I

r r
I
x ¥v
I I I I
pd I I
I I
I / I
I
f
I
I

\,/
x "" x 0

I /\
I
I /// \ 10
10 -- / \
x •
"" x
0 50 100 D,(%)
0 50 100 D,(%) 1.0 1.5 2.0 pd(g/cm 3)
J.0 1.5 2.0 pd (g/cm 3 )

(b) (c)
Fig. 6. Soil profiles examined at Site A

NII-Electronic Library Service


SOIL LIQUEFACTION POTENTIAL 101

values into Eq. (20), we get (T'av/a,/)i of around 0. 274, which is much larger than 0.128
estimated by Eq. (15) and is well comparable with the value of 0. 2 or more which was
the strength estimated from the recorded acceleration.
It should be emphasized here that Eq. (18) is a result of one trial to improve Eq. (14)
for taking account of the effects of grading or grain size on N-values. Eq. (18) should
be improved in future based on our improved knowledges.

SOME APPLICATIONS
In applying the methods described in the previous part to several projects, the authors
found that the differences in estimated Table 1. Summary of methods for estimat-
liquefaction strength of sand layers by ing Rt
different methods were so large that we
could derive completely different answers •
RiM : Measured by cyclic undrained triaxial
tests on undisturbed samples
for soil liquefaction potential by the differ-
ent methods. Example shown in the fol- 0 Ri Dr =0. 0042 Dr ( Dr emax-e
X 100 : e is
Cmax-emin
lowing are from the analyses of two sites,
Site A and Site B, for which measured from Pa in the liner of sand sampler)
values of Ri are available. Site A is an
uncompacted hydraulic fill made around ten
x RiA=0.0042Dr*=0.88~ l1v'~ 70
years ago using dredged alluvial sand. This

~ ""·~:~ :4 ~~:g~'H!~}
2
fill is underlain by a young alluvial deposit
(see Fig. 5). For the design purpose, a 0
= o. {
88
0 R,, l
series of field geotechnical survey was per-
formed as shown in Fig. 5. It was found
from these surveys that the hydraulic fill
0. 88 70 -0. 0567} l
for 0. 6~D 50 ~1. 5 mm
{la;;·~
and the shallow portion of alluvial deposit /:;,.
2.8N
Rici
consisted mainly of fine sandy materials. l1v'+70
In Fig 6 (a) through 6 (c) are shown the 2.BN
o. 21s 1og1 0 ( TI
D5o)
representative soil profiles at Site A. In
R
l1v 1
+70
_j
'"-l
for 0. 04~D 50 ~0. 6mm
'V
these figures are shown only the data at
depths from which undisturbed samples 2.BN
0.05 for 0. 6~D 50 ~1. 5 mm
O'v'+70
0.5~----=----~--------,

Siies A and G
2 2
q; =0.20-1.2 x10 ,kN/m

0.4

*II
<
A
o.3 +
-to
-40

-60
2nd Diluvial Deposit
-80
0 Bedrock (granite)
-100
R,,,=0.0042D.' -120'------------------------

r + Dso<0.074mm
PLAN
0.1 jO o.074:;;Dso<0.15

F-r·~·--c--•H-•p-H,, ~, .·.-~
[ e 0.15:;;Dso<0.3

O.OO"'----------,,c50--------:-'IOO 0 BORING POINT WHERE N-VALUES MEASURED


.. cr:+70
~
D, =210/
LEGEND
{
e BORING POINT WHERE N-VALUES MEASURED
AND SAND SAMPLES WERE TAKEN

Fig. 7. RtM and Dr* relationship Fi~. 8. Cross-section and plan of Site B
for Sites A and G

NII-Electronic Library Service


102 T ATSUOKA ET AL.

were secured. In these figures, six different estimated values of Ri by different methods.
are compared (see Table 1 for legend). In Figs. 9(a) through 9(c) are shown similar
comparisons for Site B whose simplified profile is shown in Fig. 8. At this site, an around
ten years old hydraulic fill of medium to coarse sand is underlain by a young alluvial
deposit of medium to coarse sand. By comparing the values of Ri estimated by different
methods, the followings can be derived:
( 1) The variation of N-values with depth is much larger than that of the measured
undrained cyclic triaxial strength of undisturbed specimen RtM· For the soil profiles
shown here, it seems that N-values can be more highly correlated to D 50 than RtM·

Dso(mm)
0.02 o.o5 0.1 0.2 o.5
R, (N,=20) Dso(mm)
' N-value 0.05 0.02 0.1 0.2 0.5
10 20 o.o ,0.2 o.4 o.6 R 1 (N,=20}
N-value
• R 1M(DA=5%) 10 20 30 o.o 0.2 0.4
(a )Point A
LEGENDD Rm,
{
-~-- X R1A,0Rrn
( b) Point B See a) for legend.

-¥-
N-value
,.,x
~
u Dso / /
~
,.,..--
5 -c~
~
x/ D
,., ·'lI' ' I
I
::r: I I I
D
~ \n, ,, /'
I

D ""'g. '' ) ,, /
x, D

0
Q
I '-.. .....
I .....
.....
I ' ..... I
I
I
IO - -;; D
'
_....x V i,x D
j 10 *I\, ..........
.......... ....
< !:,, x" 'l,O'VO

]
50 100 D,(%)

m All/
'(
ff'

r
1.0 2.0 pd(g/cm) 'X D
1.5

I l
15 - 20 0 50 100 D,(%)
0 Doo &S ;x;
1.0 1.5 2.0 pd(g/cm 3)

(a) (b)
Dso(mm)
0.05 0.1 0.2 0.5
N-value Ri(N,=20)
IO 20 30 0.0 0.2 0.4

( c) Point D See a) for legend.

Dso
6
Dpvx
/
/
v Llc(ox
~
-c ',,
~ 5 ;;;· A xv'q

""'
p"'
'

10 Fig. 9. Soil profiles examined at Site B

0 50 100 D.(%)
1.0 1.5 2.0 pd (g/cm 3)

( c)

NII-Electronic Library Service


SOIL LIQUEFACTION POTENTIAL 103

Figs. 7 and 10 show the relationships between RiM and Dr* for both sites. It can be seen
from these figures that the increase in RiM with the increase in Dr* is not as large as
expected from Eq. (6): Ri =0. 0042 Dr*. Castro (1975) reported that the small increase in
Ri with the increase in Dr* or N-values seamed due to larger disturbances of specimens
for larger Dr*. Nevertheless, it should be noted that the data by Castro (1975) included
the ones having Dr* of as large as 225 and <Jc' of as large as 10. 5 kgf/cm 2 (around 1000
kN/m 2) and that any analysis about the effects of grain size was not reported. When
data are limited to those from shallow uncompacted hydraulic fills and young alluvial
deposits, another interpretation about the relationship between RiM and Dr* seems
possible to the present authors as follows. It seems that values of Ri of sandy layers
which were deposited in a similar circumstance have a small variation. However, it is
likely that N-values or values of Dr* are considerably affected by the variation of grain
size. By examining the data shown in Figs. 7 and 10, it can be seen that the data
points for larger values of D 50 are located at larger values of Dr*. Based on this
observation, Eqs. (9) and (18) were derived.
( 2) The values by RiDr estimated by measured relative density (the symbols D in
the figures) are close to or slightly smaller than the measured values for Site A, but,
in general, considerably larger than the measured values for Site B. It seems that a
slight amount of shell involved in the samples from Site B increased the values of emin
determined in dry condition under no surface pressure, thereby increased Dr unreasonably.
( 3) The values of RiA by Eq. (6) (the symbols x in the figures) are considerably
smaller than the measured values for Site A. This means that for fine sand deposits,
this method is rather conservative. For Site B, the values of RiA can be reasonably
compared with the measured value. This means that this method can be reasonably
applied to medium to coarse sand deposits.
In the original simplified procedure by Seed and Idriss (1971), the effects of D 50 was
taken into account as the variation of Ri with the variation of D 50 for an identical value
of Dr as shown in Fig. 2. It can be seen from Fig. 2 that for an identical value of Dr
or Dr*, Ri becomes smaller for a smaller value of D 50 • Therefore, if the original
simplified procedure by Seed and Idriss (1971) is applied to Site A, the discrepancy of
Ri by the original simplified procedure from measured values becomes larger than it is
m Fig. 6.

o.s,~-------------~ 0.5

Site B a: =0.45-1.75
0 0.074~Dso<0.!5 mm 2 2

!
xl0 ,kN/m
/::; 0.15 ~Dso<0.3
0.4 e 0.3 ~D,o<0.7
0.4
A 0.7 ~Dso<2.0

~
'ii' 0.3
<
Q
.
... 0
0.3

II
~ ~
z ~
~ 0.2 0.2
=
O.J
=(D,') 2/15,090

50 100 50 100

o; =21Dh:~?o D,"=21y
a: N+7 --O
Fig. 10. RiM and Dr* relationship for Fig. 11. Comparison of Ri by two methods
Site B

NII-Electronic Library Service


104 TATSUOKA ET AL.

Table 2. Cases of liquefaction and non-liquefaction investigated in this study

Jindo·ji

Kogane Cho

Shinano River
Railroad Bridge
I+
I Br.!

Higashi- I~
Kosen Bridge I Br. 4
Br.5
Bandai Bridge I Br. 6

l
~Niigata
Yachiyo Br.5 City
Bridge
Br.7

Shii1-Mntsuhama1~
Bridge Br. 2

Taihci Bridi;c 1~
Br.2
-------
Niigata city Omiya i

,.....,,~ I::: ;
. :::: :

Br.3
I
Nishi Oh-Hata Cho
Niigata Airport

Seki ya

Br.1
Niigata
Br.2
Railroad
Hospital Br. I

N"nac Beach,
1Iakodatc

* B: Soil data before earthquake, A: Soil dat~ after earthquake, U: Unknown.


** E : Estimated by using Table 1, M : Measured,
*"'* x : Non-liquefaction, O : Liquefaction
The numbers in ( ) are estimated.depths of liquefaction in meter.
CD Recorded value at Kawagishi-cho.
@ Estimated from recorded val:.ics at other sit('S,
@ Recorded value at Ua.chinolic,
(i) Estimated by the empirical equation for alluvial deposits (16) :
a 5 max(gals) =32.1 x ioo.z~HI x ,J-C1.1~1
in which M is the Richter's magnitude and J is the cpiccntral distar:ce in km.
® Estimated from damage to structures.

NII-Electronic Library Service


SOIL LIQUEFACTION POTENTIAL 105

( 4) Both the values of Rls by Eq. (9) (the symbols O m the figures) and the ones
of Rzc 2 by Eq. (18) (the symbols v in the figures) are, in general, in well agreement
with the measured values for both sites. This means that for both sites, it is necessary
to take account of the effects of grain size on N-values to obtain reasonable values of
R 1 from N-values. At the same time, it can be seen by examining the data shown in
Figs. 6 and 9 carefully that the values of RlB (the symbols O) are slightly closer to the
measured values than the values of R1c 2 (the symbols v). This is because Rlc 2 is more
sensitive to the variation of N-values than Rzs·
( 5) The values of Rici by Eq. (14) (the symbols D. in the figures) are considerably
smaller than the measured values for smaller N-val ues and are larger than measured
values for larger N-values. This is because Rici is very sensitive to the variation of
N-values. Therefore, for fine sand deposits which have smaller N-values, Rici is
extremely small. It is also noted that Rici is much smaller than RzA (the symbols x)
for Site A. The difference between Rici and RtA is illustrated in Fig. 11. It can be
seen that for the value of Dr* smaller than 50, R1c 1 is considerably smaller than RzA· By
two reasons; no considerations about the effects of grain size on N-values and high
sensitivity to the variation of N-values, R 1c1 seems significantly conservative for fine or
silty sand deposits. In summary, it seems essential to the present authors to take account
of the effects of grading or grain size on N-values in evaluating in situ cyclic undrained
strength from N-values, especially for fine sands.

CASE STUDIES
To examine the validity of Eq. (9), which can be recommended at present, field data
of liquefaction and non-liquefaction cases in Japan will be analyzed (see Table 2). In
these analyses, the following index will be used to represent the degree of resistance of
a soil element against liquefaction,
(21)
in which FL is the liquefaction resistance factor, R is the resistance of a soil element
and L is the cyclic load applied to the soil element. Both R and L will be defined
later. In this method, the effects of diffusion of excess pore pressure during earthquake
are not directly accounted for. The value of R can be related to Rias (Iwasaki et al., 1978)
(22)
C 1 is the correction factor for the difference of in situ confining pressures from those
in triaxial tests and has been proposed as (1 +2 K 0 )/3 by Ishihara and Li (1972). C 2 is
that for the difference between in situ "random" loading forms during earthquake motions
and the sinusoidal loading form in triaxial tests and has been proposed as 1/0. 55,........,1/0. 7
with the average being 1. 62 by Ishihara and Yasuda (1975). The value of C 2 can be
related to the parameter b in Eq. (13) and the randomness of earthquake motions or
earthquake magnitude and epicentral distance (Tatsuoka et al., 1980). As this relationship
is not well established yet, the average value of 1. 62 will be used in this study. C 3 is
that for the effects of soil disturbances in the process of sampling and handling. c4 is
that for the effects of densification in the process of sampling and handling. As samples
tested in this study are relatively loose, it is likely that some increases in strength due
to densification occured. Because densification of samples is caused by disturbances during
sampling and handling, these effects may cancel. Thus, the value of the product C 3 x C 4
was taken to be unity in this study. C 5 is the correction factor for the multi-direction
shaking which may be anticipated in actual in situ loading conditions. According to Seed
(1979) this factor is equal to 0. 9. Using the values of C 1 through C 5 described above,

NII-Electronic Library Service


106 T ATSUOKA ET AL.

Eq. (22) becomes


R=2/3x1. 62x1. 0 x 0. 9 x Ri~Ri (23)
for K 0 =0. 5. This value of K 0 can be considered adequate for uncompacted reclaimed
lands and alluvial deposits which will be analyzed later.
The cyclic load L can be estimated as

(24)

in which (-r/av')max is the maximum stress ratio during an earthquake motion, asmax is the
estimated maximum acceleration on the ground surface in gals, g is the acceleration of
gravity (980 gals), <Iv is the total overburden pressure, av' is the effective overburden pres-
sure and rd is a stress reduction coefficient with a value less than 1 (Seed and Idriss, 1971).
Iwasaki et al. (1978) has reported that it was found by a comprehensive series of
response analyses of soil layers that rd is a function of depth, the predominant period
of input motion (Te) and the natural period (T g) of the site concerned. As a simple
correlation among rd• Te (or M, LI), T g or so was not established yet, it was considered
reasonable to estimate rd for the simplified method from the average relationship as
rd= 1-0. 015 z (z: depth in meter) (25)
While the effects of earthquake magnitude M on FL are not taken into account in this
method, these effects are taken into account in the method by Seed and Idriss (1971).
By following their method, FL can be obtained as
FL= Cr(ddp/2d/)Neq ( 26 )
0. 65·asmaxfg·dvfdv' ·rd
Cr is 0. 63 for K 0=0. 5 by Seed (1979). As in Eq. (13), (adp/2 ac')N•q can be related to
Ri for Nc=20 as
(27)
The values of Neq were suggested by Seed (1979) as Neq=2l for M=8, Neq=lO for M=7
and Neq=4 for M=6. For the same values of Ri and rd, the ratio of FL by the method
proposed here (Eqs. (23) and (24)) to that based on the Seed-Idriss method (Eqs. (26)
and (27)) can be obtained as
___!'_L in this investigation 0.65
(28)
FL based on Seed and Idriss (1971)
The ratio for Cr=O. 63 and b= -0. 2 is shown against M in Fig. 12. It can be seen from
Fig. 12 that for the same values of Ri and rd, the difference between the value of FL by
the method proposed by the authors and the value of FL based on the method by Seed
and Idriss (1971) is not significant for the range of M in this case study (see Table 2).
Therefore, it is likely that errors due to that the effect of M, LI or so is not taken into
account in the method proposed in this paper may not be significant in this case study.
The main reason why the effects of M is not incorporated in the method proposed here
is that it seems to the authors that the effects of M on FL is not as simple as expressed
in Eqs. (26) and (27). Further investigations are necessary to clarify the effects of M, LI
or so on FL.
It is obvious that the parameter FL is not enough to express the severeness of
liquefaction of a sand deposit. It was found by Iwasaki et al. (1978) that severe damage
to foundations was found only when the value of FL was considerably less than 1. 0, say
0. 6, and the zone where FL was considerably less than 1. 0 was not thin, say, thicker
than 5 m. To express the degree of severeness of liquefaction of a soil mass, the
liquefaction potential indexes will be defined as

NII-Electronic Library Service


SOIL LIQUEFACTION POTENTIAL 107

The range
of M
in this
study.

0 6'--~-~__._7~~~~~­

Fig. 12. Ratio of FL by two methods Fig. 13. Integration of FL


as a function of earthquake magni-
tude

PL= i 2
°F·w(z)dz (29a), PL*= i 2
°F·w*(z)dz (29b)

in which F=l-FL for FL~l. 0 and F=O for FL>l. 0 and w(z) =10-0. 5 z (z in meter)
and w*(z) =5 as illustrated in Fig.13. For the case of FL=O. 0 for the entire range from
z=O to 20 m, both PL and PL* become 100. And for the case of FL~l. 0 for the entire
range from z=O to 20 m, both PL and PL* become 0. 0. In other cases where FL varies
between 0 to 1. 0 in some zones, PL and PL* have values ranging from 0 to 100. The
difference between PL and PL* is due to the difference in the weight function, w(z) and
w*(z). This is based on such an idea that shallower deposits can be more susceptible to
liquefaction than deeper deposits for an equal value of FL due to the seepage of pore
water from deeper portions. Further investigation on the weight function will be necessary.
The sites where liquefaction or non-liquefaction were clearly observed during earthqua-
kes and good soil data were also available were selected as listed in Table 2. The non-
liquefied sites listed in Table 2 are hydraulic fills or young alluvial sandy deposits which
can be considered to have liquefied if earthquake motion had been sufficiently large. To
calculate the values of PL and PL*, two different estimated values of Ri were adopted,
namely RzA and Rz 8 • RzA by Eq. (6) was adopted as a representative one which does
not take account of the effects of grain size on Ri and RzB by Eq. (9) was adopted as a
representative one which takes account of the effects of grading on Rz. When only the
name of soil type were available, the values of density and D 50 were estimated
following Table 3. The values of FL
Table 3. Pt and D 50 used when these values
were calculated by the methods described
were not measured
above for the soil layers whose mean
Soil Type
diameters ranged from 0. 02 to 2. 0 mm. It Reported Pt (t/m 3 ) D50 (mm)

was postulated that the soil layers whose Surface soil 1. 7 0.02
D 50 are less than 0. 02 mm or larger than Gravel 2.1 >0.6
2. 0 mm do not liquefy. Except one case
Sand 1. 9 0.25
for the Fukui Earthquake, soil layers with
Coarse sand 1. 9 0.3
D 50 larger than 2. 0 mm were not included in
Medium sand 1. 9 0.25
the surface deposits above the depth of 20
Fine sand 1. 9 0.2
m. The values of asmax were estimated by
Silt 1. 7 0.02
several methods as explained in the foot-
Silty sand 1. 9 0.1
note of Table 2.
Sandy silt 1. 7 0.04
Fig. 14 shows the relationship between

NII-Electronic Library Service


108 TA TSUOKA ET AL.

f,O

40r-----r---~--~--~

0 Liquefied .Sites
{ X Non- Liquefied Sites

30

~ 20
Liquefaction

10. "'
u"'"'

0 ~~'---_,!,..10----,L20----'30---""'40

PL
Non-Liquefaction
\7
Fine or Silty Sand Deposits

Fig. 14. PL* and PL relationship for


RlB Fig. 15. Distribution of cases against PL

PL and PL* which were obtained by using RzB· It can be seen from this figure that
while the values of PL* are somewhat less than those of PL, there is an almost
linear relationship between PL and PL*. Therefore, since similar results can be expected
from analyses by using PL or PL*, only PL will be used in the following.
Distribution of the number of the cases of liquefaction and non-liquefaction are
displayed against the value of PL in Fig. 15. From the values of PL both for RzB and
RiA, cumulative curves of cases of liquefaction and non-liquefaction were constructed
as shown in Fig. 16. For example, the point A in Fig. 16 means that when RiB is
used, the case of liquefaction with PL less than 14 is 40%. It can also be seen from
Fig. 16 that there is a clear separation between the curve for liquefaction and that for
non-liquefaction for the case of RlB· This separation is clearer than for the case of
RzA· It is apparent that the larger the separation between these two curves, the better
the method. One of the reasons why there is not a good separation between two
cumulative curves for liquefaction case and for non-liquefaction for the case of RzA is
due to the fact that the values of PL for some of the non-liquefied sites of fine or silty
sand deposits are not small. This fact can be seen by carefully examining the data
shown in Table 2 and Fig. 15.
In summary, it seems that the parameter PL calculated by using RiB can well distingush
liquefaction cases and non-liquefaction cases listed in Table 2. Furthermore, it can be
seen from the curves for the case of RzB shown in Fig. 16 that PL for all the non-
liquefaction sites is less than 20 and PL for 50% of liquefaction sites is larger than 15.
It is also seen in Fig. 16 that for the case of RzB, PL for 65% of non-liquefaction sites
is less than 5 and there are only around 10% of liquefaction sites whose PL is less than
5. From the facts shown above, it is likely that if PL less than 5 is obtained for a
site under consideration, there may be low probability of liquefaction, but if PL larger
than 15 is estimated for a site under consideration, there may be high probability of
liquefaction.
From the above discussions, it is apparent that the parameter PL is a ·useful tool for
evaluating liquefaction potential of a site. In Fig. 17 are shown the values of PL by the

NII-Electronic Library Service


SOIL LIQUEFACTION POTENTIAL 109

!Site A, Point F j

40

100
30
_....- ,
Non- Liquefaction _ 4 _,
p..
;:>~---
~
(
1
20
/:f/J,
/
~ (
,P
..
>
/ 10 -
~
8 R,=R1A
OX---X
00
!
0 50 250 300
10 20 30 40 0 10 20 30 40 50
P1.. P1..

Fig. 16. Cumulative curves of cases against PL Fig. 17. PL versus asmax by different methods

different methods for estimating Ri against asmax for the point F at Site A (see Fig.
6(b)). Note that Rzc 1 was originally de.fined for M=7-1/2. It can be seen that PL
increases with the increase in asmax for any case. However, the value of PL by RzB
is very close to that by RzM· It is also seen that value of PL by Rici is much larger
than that by RlM· If the design value of as max is 200 gals, the value of p L 2 by RlB•
This means that we will have very low liquefaction potential when Rin is used.
Based on the above analyses, the following procedure can be suggested for design
purposes.
( 1) Perform undisturbed sand samplings of high quality at several representative
points, and perform standard penetration tests for a whole area under consideration.
( 2) Perform cyclic undrained triaxial tests of high quality.
( 3) Examine Eq. (9) with the data obtained by step (2).
( 4) Modify Eq. (9) if necessary.
( 5) Construct PL (or PL*) versus asmax curves for representative points.
( 6) Obtain the values of asmax for values PL of 5 and 15 and compare with the
design value of asmax•
( 7) If asmax for PL=5 is larger than the design value of asmax' severe liquefaction
would not take place. If asmax for PL=15 is smaller than the design value of asmax'
moderate or severe liquefaction would take place.

CONCLUSIONS
Several methods to evaluate cyclic undrained strength of sands from standard penetration
resistances (N-values) have been compared by applying these methods to actual soil
pro.files. It was found that there were large differences among the results by the
different methods. It was found that an empirical equation for evaluating strength from
N-values which takes account of the effects of grain size on N-values could provide
reasonable results. It was also found that when the effects of grain size on N-values were
not taken into account, values of strength evaluated from N-values could be considerably
smaller for fine or silty sand.
A simplified procedure for assessing soil liquefaction potential has been proposed. In
this method, liquefaction potential for a layer at a depth is expressed by the liquefaction

NII-Electronic Library Service


110 T ATSUOKA ET AL.

resistance factor FL and liquefaction potential for a soil deposit from the surface to
the depth of 20 m is expressed by the liquefaction potential index PL or PL* which
can be obtained by integrating FL with a weight function w(z) or w*(z). The values
of FL and PL or PL* were calculated for the sites of liquefaction and non-liquefaction
during previous earthquakes in Japan. It was found that when the effects of grain size
on N-values are taken into account, the value of PL or PL* are quite different between
the sites of liquefaction and those of non-liquefaction. The index PL or PL* can
be useful in liquefaction potential evaluation of soil deposits.

ACKNOWLEDGEMENTS
Extensive in situ geotechnical surveys and laboratory tests have been conducted by the
Kanto Regional Construction Bureau of the Ministry of Construction, Honshu-Shikoku
Bridge Authority and Yokohama City. The whole authors express their cordial apprecia-
tion to the staff members concerned. The authors are also grateful to Miss M. Torimitsu for
her help in typing the manuscript.

NOTATIONS
Dr=relative density
Dr*=210vN/(av'+70) (av' m kN/m 2)
D 50 =mean grain size in mm
FL=R/L; liquefaction resistance factor
L =cyclic load in stress ratio
N =standard penetration resistances
Nc=number of loading cycles
PL> PL*= liquefaction po ten ti al indexes
R=resistance of a soil element against liquefaction
Ri=stress ratio which causes liquefaction in triaxial specimen at Nc=20
Rz Dr , RlA, RiB, Rici and Rzc 2 =estimated values of Ri
rd=l-0.015z (z: depth in m)
asmax=maximum acceleration at ground surface m gals
pa, Pt =dry and total unit densities
av'=effective overburden stress in kN/m 2
O"t=total overburden stress in kN/m 2

REFERENCES
1) BRI (1965) : "Report on damage to buildings during the Niigata Earthquake," The Building
Research Institute, Ministry of Construction, No. 42, 1965 (in Japanese).
2) BRI (1969) : "Investigation on liquefaction of saturated sand and some problems on soil-struc-
ture-interaction," The Building Research Institute, Ministry of Construction, No. 55 (in Japa-
nese).
3) Castro, G. (1975) : "Liquefaction and cyclic mobility of saturated sands," Journal of the Geot.
Eng. Div., ASCE, Vol.101, No. GT 6, pp. 551-569.
4) De Alba, P., Chan, C. K. and Seed, H. B. (1975) : "Determination of soil liquefaction character-
istics by large scale laboratory tests," Report No. EERC 75-14, Earthquake Engineering Research
Center, University of California, Berkeley.
5) Gibbs, H. J. and Holtz, W. G. (1957) : "Research on determining the density of sand by spoon
penetration test," Proc. 4th International Conference on Soil Mechanics and Foundation Engineer-
ing, London, Vol. I, pp. 35-39.

NII-Electronic Library Service


SOIL LIQUEFACTION POTENTIAL 111

6) Ishihara, K. and Li, S. (1972) : "Liquefaction of saturated sand in triaxial torsion shear test,"
Soils and Foundations, Vol.12, No. 2, pp. 19-39.
7) Ishihara, K. and Yasuda, S. (1975) : "Sand liquefaction in hollow cylinder torsion under irregular
excitation," Soils and Foundations, Vol.15, No. 1, pp. 29-45.
8) Ishihara, K. (1977) : "Simple method of analysis for liquefaction of sand deposits during earth-
quake," Soils and Foundations, Vol.17, No. 3, pp.1-17.
9) Ishihara, K., Silver, M. L. and Kitagawa, H. (1978) : "Cyclic strength of undisturbed sands obtain-
ed by large diameter sampling," Soils and Foundations, Vol.18, No. 4, pp. 61-76.
10) Ishihara, K. and Ogawa, K. (1978) : "Liquefaction susceptibility map of downtown Tokyo," Proc.
of the 2nd Int. Conf. on Microzonation, San Francisco, Vol. II, pp. 897-910.
11) lshizawa, M., Nakagawa, S. and Kurohara, I. (1977) : "Liquefaction test of undisturbed samples
containing fine content," Proc. the 12th Annual Meeting of JSSMFE, pp. 397-400 (in Japanese).
12) Iwasaki, T., Tatsuoka, F., Tokida, K. and Yasuda, S. (1978) : "A practical method for assessing
soil liquefaction potential based on case studies at various sites in Japan," Proc. of the 2nd Int.
Conf. on Microzonation, San Francisco, Vol. II, pp. 885-896.
13) Japanese Society of Soil Mechanics and Foundation Engineering (1976) : "Report on earthquake
damage of subground streets and structures," March (in Japanese).
14) Japanese Society of Civil Engineers : "The report on the damage during the Niigata Earthquake
of 1964," (in Japanese).
15) Kishida, H. (1970) : Characteristics of liquefaction of level sandy ground during the Tokachioki
earthquake, Soils and Foundations, Vol. X, No. 2, pp. 103-111.
16) Martin, G. R., Finn, W. D. L. and Seed, H.B. (1978) : "Effects of system compliance on liquefac-
tion tests," Proc. of the Geotechnical Div., ASCE, Vol.104, No. GT4, Proc. Paper 13667, April,
pp. 463-479.
17) Meyerhof, G. G. (1957) : "Discussion of Session 1," Proc. 4th International Conference on Soil
Mechanics and Foundation Engineering, London, Vol.3.
18) Ohsaki, Y. (1970) : "Effects of sand compaction on liquefaction during the Tokachioki Earth-
quake," Soils and Foundations, Vol.X, No. 2, pp.112-128.
19) Pyke, R. M., Knuppel, L.A. and Lee, K. L. (1978) : "Liquefaction potential of hydraulic fills,"
Journal of the Geotechnical Engineering Division, ASCE, Vol.104, No. GTll, Proc., Paper 14133,
Nov., pp.1335-1354.
20) Seed, H.B. and Idriss, I. M. (1971) : "A simplified procedure for evaluating soil liquefaction
potential," Journal of the SMFE Div., ASCE, Vol.97, No. SM9, Sept., pp.249-274.
21) Seed, H.B. (1979) : "Evaluation of soil liquefaction effects on level ground during earthquakes,"
S-0-A Report, Reprint of ASCE, Annual Convention and Exposition on Liquefaction Problems
in Geotechnical Engineering, Philadelphia.
22) Tatsuoka, F., Iwasaki, T., Tokida, K., Yasuda, S., Hirose, M., Imai, T. and M. Kon-no (1978) : "A
method for estimating undrained cyclic strength of sandy soils using standard penetration N-
values, Soils and Foundations, Vol.18, No. 3, pp. 43-58.
23) Tatsuoka, F., Yasuda, S., Iwasaki, T. and Tokida, K. (1980) : "Normalized dynamic undrained
strength of sands subjected to cyclic and random loading," Soils and Foundations, Vol. 20, No.
3, pp.1-16.
(Received December 3, 1979)

NII-Electronic Library Service

You might also like