LIQUEFACTION AND UNDRAINED RESPONSE EVALUATION OF

SANDS FROM DRAINED FORMULATION

By Mohamed Ashour1 and Gary Norris,2 Members, ASCE

ABSTRACT: A general approach has been established to assess the undrained stress-strain curve and effective
stress path under monotonic loading from drained triaxial tests. An appropriate formulation of a drained and
drained rebounded (i.e., overconsolidated) triaxial test response is developed that, in turn, allows the assessment
of developing liquefaction and the undrained behavior of saturated sands. The formulation presented is based
upon reported experimental drained test results that were obtained from different investigators using different
testing techniques. This formulation is a function of the confining pressure and basic properties of the sand,
such as relative density, uniformity coefficient, and particle shape (roundness), which can be obtained from
visual inspection. The approach is verified by comparing predicted and reported (observed) undrained behavior.
The developed formulas allow one to predict the potential of sand to liquefy, the type of liquefaction, the peak
and residual strength values, as well as the whole undrained stress-strain curve and effective stress path. The
simplicity of this approach makes it an attractive general method to characterize the undrained behavior of sands
in a preliminary analysis with no need to run sophisticated experimental tests.

INTRODUCTION
Several studies have been conducted to provide a better un-
derstanding of the undrained behavior of saturated sand under
different types of loading. Some of the pioneering work in this
field has been performed by Seed and Lee (1967), Castro
(1969), Ishihara et al. (1975), Casagrande (1976), Castro and
Poulos (1977), Poulos (1981), and Castro et al. (1982). In
addition, several recent studies (Mohamad and Dobry 1986;
Guzman et al. 1988; Vaid et al. 1989; Ishihara 1993) have
made significant contributions to the understanding of the un-
drained behavior of saturated sands. The main interest in these
studies is to relate the undrained strength of sand to its initial
state in order to allow the designer to predict the potential for
a saturated sand to liquefy. In other words, most of the inves-
tigations focused on the influence of the consolidation pressure
and the associated void ratio of the sand on the undrained
behavior of the saturated sand under monotonic and cyclic
loading.
The undrained behavior of isotropically consolidated satu-
rated sand under monotonic loading is accompanied by a
change in the excess pore-water pressure, which, in turn, leads
to different forms of undrained behavior. Unfortunately, no
prior study provides a clear picture of mobilized undrained
behavior and the associated effective stress path under un-
drained monotonic loading. The available studies indicate the
potential for sand to liquefy, and characterize it as liquefiable
or nonliquefiable material. The only way to assess the mobi-
lized undrained behavior of a saturated sand (its stress-strain
and stress path) under monotonic loading is via laboratory test-
ing. By consolidating the saturated sand to different values of
confining pressure or void ratio, a series of isotopically con-
solidated undrained (ICU) tests allow one to assess the vari-
ation in the peak undrained resistance, the residual stress of
the saturated sand, and the associated levels of strain.
Recently, it has been shown that the undrained response of
sand can be assessed from its drained laboratory response
1
Asst. Prof., Civ. Engrg. Dept., El-Mansoura Univ., Egypt; Postdoc-
toral Res. Fellow, Civ. Engrg. Dept., Univ. of Nevada, Reno, NV 89557.
E-mail: ashourm@unr.edu
2
Prof., Civ. Engrg. Dept., Univ. of Nevada, Reno, NV.
Note. Discussion open until January 1, 2000. To extend the closing
date one month, a written request must be filed with the ASCE Manager
of Journals. The manuscript for this paper was submitted for review and
possible publication on March 3, 1998. This paper is part of the Journal
of Geotechnical and Geoenvironmental Engineering, Vol. 125, No. 8,
August, 1999. 䉷ASCE, ISSN 1090-0241/99/0008-0649–0658/$8.00 ⫹
$.50 per page. Paper No. 17828.
(Norris et al. 1997). The present paper deals specifically with
the formulation of drained behavior as a function of state con-
ditions and sand properties, thus reducing the need for labo-
ratory testing. Therefore, the current study provides a general
approach to assess the mobilized undrained behavior of satu-
rated sand under monotonic loading, whether the sand is loose,
medium dense, or dense. Moreover, the present study allows
one to characterize the undrained response of the saturated
sand, whether the sand is contractive and/or dilative, to define
the potential for the sand to liquefy, and to characterize the
type of the expected liquefaction (complete or limited lique-
faction), as seen in Fig. 1.
The approach presented here assesses the undrained behav-
ior of saturated sand under monotonic loading and is based on
the most basic properties of the sand, such as its void ratio,
e c, or the relative density, Drc, at the end of consolidation to
pressure, ␴3c, the roundness of sand grains, ␳, the uniformity
coefficient, Cu, the effective angle of the internal friction, ␸,
and the drained axial strain at 50% stress level, ε 50. This work
was developed to deal with most types of sand under different
levels of confining pressure. The validity of the work pre-
sented and the equations formulated are verified by several
comparisons with observed results employing Nevada, Ione,
Ottawa, Banding, and Fraser River sands.
METHOD OF ANALYSIS
The technique developed by Norris et al. (1997) employs a
series of drained tests, with volume change measurements, on
samples isotropically consolidated to the same confining pres-
sure, ␴3c, and void ratio, e c, to which the undrained test is to
be subjected. However, the drained tests are rebounded to dif-
FIG. 1. Undrained Behavior of Saturated Sand under Mono-
tonic Loading

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / AUGUST 1999 / 649

FIG. 2. Interrelationships among: (a) Drained and Undrained Stress-Strain Behavior; (b) Isotropic Consolidation Rebound; (c) Effec-
tive Stress Path
ferent lower values of effective confining pressure, ␴¯ 3, before
being sheared.
During an ICU test, the application of a deviatoric stress,
␴d , causes the pore-water pressure, ⌬ud , to build up, which
results in a reduced effective confining pressure, ␴¯ 3, i.e.
␴
¯ 3 = ␴3c ⫺ ⌬ud
and an associated isotropic expansive volumetric strain ε v,iso,
the same as recorded in an isotropically rebounded drained
triaxial test. However, in the undrained test, the volumetric
change or volumetric strain must be zero. Therefore, there
must be a compressive volumetric strain component, ε v,shear,
due to the deviatoric stress, ␴d . This shear induced volumetric
strain, ε v,shear, must be equal and opposite to ε v,iso
ε v,shear = ⫺ε v,iso
so that the total volumetric strain, ε v = ε v,iso ⫹ ε v,shear, in the
undrained response is zero. In the isotropically rebounded
drained shear test, ε v,iso and then ε v,shear (to match ε v,iso) are
obtained separately and sequentially; in the undrained test,
they occur simultaneously.
During drained isotropic expansion, the resulting axial
strain, ε1, is
1 ISOTROPICALLY CONSOLIDATED SAND
ε1,iso = ε 2,iso = ε 3,iso = ε v,iso
3 UNDER DEVIATORIC STRESSES
Based on Hooke’s law and effective stress concepts (Norris
et al. 1998), the undrained axial strain due to shear (␴d) and
effective stress (␴¯ 3) changes can be related to the drained or
effective stress strains as
1 held constant). This stress-strain relationship was originally
(ε1)undrained = (ε1)␴d ⫹ (ε1)⌬␴¯ 3 = (ε1)drained ⫹ ε1,iso = (ε1)drained ⫺
3 established by Norris (1986) and then modified by Ashour et
Therefore, with isotropically consolidated-rebounded
drained triaxial tests available for different ␴¯ 3, one can assume
a value of ␴¯ 3, find ε v,iso [Fig. 2(b)], enter the ε v-ε1 drained shear
curves [Fig. 2(a)] at ε v,shear equal to ε v,iso, and find the drained
ε1 and ␴d on the same confining pressure (␴¯ 3) ε v-ε1 and ε1-␴d
curves. Then (ε1)undrained is established according to (4) and one
point on the undrained ␴d -ε1 curve can be plotted. The cor-
responding effective stress path ( p̄ = ␴¯ 3 ⫹ ␴d /2 versus q = ␴d /
650 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / AUGUST 1999
2) can also be plotted as shown in Fig. 2(c). This technique is
illustrated in detail by Norris et al. (1995, 1997).
The technique presented above requires that a series of
isotropically consolidated drained (ICD) tests be performed to
allow one to assess the undrained stress-strain curve of a sat-
urated sand consolidated to a certain confining pressure, ␴3c,
(1) at a particular value of void ratio, e c, or relative density, Drc.
This, in turn, requires a certain level of experience, effort,
time, and cost. The study presented here establishes a group
of applicable equations based on the basic properties of sand
in order to yield the following relationships:
• The stress-strain and volume change curves of the ICD
triaxial tests at different values of confining pressure
• The isotropically consolidated, rebounded, volume change
(2) curve
• The stress-strain and volume change curves of the iso-
tropically consolidated, rebounded, drained tests at differ-
ent overconsolidation ratios (OCR = ␴3c /␴¯ 3)
These drained relationships can then be used to predict the
undrained response of saturated sand as described above.
DRAINED BEHAVIOR FORMULATION OF
(3)
Drained Stress-Strain (␴d -␧1) Relationship
The stress-strain relationship presented here is employed to
assess the drained stress-strain curve of the isotropically con-
solidated sand under shear loading (confining pressure, ␴¯ 3,
ε v,iso
(4) al. (1998). The ratio of deviatoric stress, ␴d , at axial strain,
ε1, to the failing stress, ␴df , is the stress level (SL), which is
given as
␴d ␭ε1
SL = =
␴df ε 50
exp(⫺3.707SL) (5)
where ε 50 = value of ε1 at SL = 0.5; and 3.707 and ␭ represent
fitting parameters of the relationship. Parameter ␭ is equal to
3.19 at ε1 ⱕ ε 50 and then varies linearly with the stress level
between 3.19 at ε1 = ε 50 and 2.14 at ε1 = ε 80. If the stress level by Norris (1986) based on correlation of the sand’s uniformity
is greater than 80%, the stress-strain relationship is given as coefficient, Cu, and the void ratio, e c (or Drc), at the applied

冋 册
consolidation pressure.
100ε1 Given that a simple shear response for an initial isotropic
SL = 0.2 exp (6)
(mε1 ⫹ q) stress state (␴¯ m = ␴3c, K o = 1) can be viewed as an expanding
Mohr circle about the point ␴3c, (5) can be used to evaluate
where m = 59.0; and q = 95.4ε 50 = fitting parameters.
the modulus reduction relationship, G/Go, to compare it with
At constant confining pressure, the drained strength, ␴df , of
the long accepted variations (Seed and Idriss 1970), as shown,
a sand is a function of both ␴¯ 3 and the frictional angle, ␸.
for example, in Fig. 3. Such good agreement gives the pro-
Accordingly
posed formulation added credibility. As shown in Appendix I,

␴d = SL(␴d)f = SL␴
¯3 冋 冉
tan2 45 ⫹
␸
2 冊 册
⫺1 (7)
the position of the modulus reduction curve will shift with
ε 50, which, as judged from (8), varies with the confining pres-
sure, ␴3c.
During the undrained test, ␴¯ 3 will vary with the changing
pore-water pressure, but the stress level, SL, at the current Volume Change in Drained Response due to Shear
␴¯ 3, and shear induced axial strain, ε1, are given by (5). Note Stress (␧v,shear-␧1 Relationship)
that ε1 in (5) is (ε1)drained of (4).
The drained axial strain at 50% stress level, ε 50, is given as The prediction of the volume change of sand through the
volumetric strain, ε v,shear, due to shear loading is based upon
冉 冊
0.2
␴3c basic information such as ␴¯ 3, Drc, ε 50, and ␳. The roundness,
(ε 50)␴3c = (ε 50)42.5 (8) ␳, of the sand is determined by using a comparator chart (Pow-
42.5
ers 1953). See Table 1 for numerical values for descriptive
where ␴3c represents applied consolidation pressure (kPa); and adjectives such as ‘‘angular.’’ A number of drained tests per-
(ε 50)42.5 denotes drained axial strain at 50% stress level under formed on different sands (Table 2) are used to formulate a
a confining pressure of 42.5 kPa. Parameter (ε 50)42.5 is given series of empirical equations that describe the drained behavior
of isotropically consolidated sands.
As seen in Fig. 4, the main features that control the shape
of the ε v,shear-ε1 relationship are the coordinates and the slopes
of the ε v,shear-ε1 curve at points A, B, and C. The initial slope
of the ε v,shear-ε1 curve at point A is SA, which is equal to
dε v,shear /dε1, where ε1 and ε v,shear are equal to zero. The coor-
dinates of peak volumetric strain at point B are (ε1)B and
(ε v,shear)B, where (ε v,shear)B represents (ε v,shear)max. The slope of the
ε v,shear-ε1 curve at point B is SB and is equal to zero. Finally,
at failure, the slope and the coordinates at point C on the
ε v,shear-ε1 curve are expressed by Sf and (ε1, ε v,shear)C, respec-
tively. At point C, Sf is equal to (dε v,shear /dε1)f . Beyond point
C, the ε v,shear-ε1 curve extends linearly at a constant slope equal
to Sf .
To plot the ε v,shear-ε1 curve, a fifth-order binomial equation
FIG. 3. Variation of Shear Modulus with Shear Strain for
Sands

TABLE 1. Roundness Classes (Powers 1953)

Grade terms Class intervals Geometric means

(1) (2) (3)
Very angular 0.12–0.17 0.14
Angular 0.17–0.25 0.21
Subangular 0.25–0.35 0.30
Subrounded 0.35–0.49 0.41
Rounded 0.49–0.70 0.59
Well rounded 0.70–1.00 0.84 FIG. 4. Volumetric Strain Curve due to Shear versus Axial
Strain and Its Major Points

TABLE 2. Properties of Sands Employed to Develop Approach Presented

Roundness
Material (␳) e max e min Cu Reference
(1) (2) (3) (4) (5) (6)
Nevada sand (subrounded, clean, fine, white quartz, 0.45 0.856 0.548 1.6 Norris et al. (1995, 1997)
foundry sand)
Ione sand (subangular, clean, fine, grayish quartz) 0.29 1.000 0.717 1.4 Norris et al. (1995, 1997)
Toyoura sand (angular, uniform, Japanese sand) 0.21 0.977 0.605 1.46 Fukushima and Tatsuoka (1984)
Sacramento River sand (SRS) (subangular to sub- 0.35 1.03 0.61 1.4 Lee and Seed (1967)
rounded, uniform, feldspar and quartz sand)
Banding sand (subrounded to subangular, uniform, 0.33 0.84 0.50 1.80 Castro (1969)
clean, fine quartz sand)
Kogyuk sand (350/2) (subrounded to subangular, uni- 0.40 0.829 0.47 1.80 Been and Jefferies (1985)
form, medium quartz sand)
Ottawa sand (rounded, medium uniform quartz 0.55, 0.55 0.82, 0.95 0.50, 0.57 1.50, 1.66 Negussey and Vaid (1990), Dakoulas
sand—ASTM C-109) and Yuanhui (1992)

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / AUGUST 1999 / 651

is established based on the slopes and the coordinates at the
three major points on the ε v,shear-ε1 curve (A, B, and C). The
slopes and coordinates at the points A, B, and C represent the
boundary conditions of (9)
ε v,shear = a ⫹ bε1 ⫹ cε ⫹ dε ⫹ eε ⫹ f ε
2
1
3
1
4
1
5
1 (9)
Terms a–f = constants needed to satisfy these boundary
conditions; and ε1 and ε v,shear symbolize axial strain and the
associated volumetric strain due to deviatoric stress of any
point on the ε v,shear-ε1 curve, respectively. The derivation of (9)
is presented in Appendix I.
The slopes and coordinates at points A, B, and C on the
ε v,shear-ε1 curve at ␴3c are obtained using (10)–(15). These
equations are empirically developed based on data of isotrop-
ically consolidated drained tests on seven different sands from
different environments and obtained by different investigators
(Table 2). It should be mentioned here that the volume change
behavior of any sand is influenced by the method of sample
From Fig. 5(a), the initial slope, (SA)␴3c, at point A on the
ε v,shear-ε1 curve is given as
1
(SA)␴3c = ␭1 = (10)
exp(␳2 ⫹ Drc)
Note that Drc in this and the following equations is a dec-
imal value. As noticed in the data recorded by Dakoulas and
Yuanhui (1992), Negussey and Vaid (1990), and Been and Jef-
feries (1985), the variation of the consolidation pressure has
no significant effect on the slope SA.
The maximum value of the volumetric strain due to applied
deviatoric stress, (ε v,shear)max, is located at point B and is equal
to (ε v,shear)B. According to the experimental results shown in
Fig. 5(b), (ε v,shear)B at ␴3c is obtained as
 ε 50 
2
(ε v,shear)B,␴3c = (ε v,shear)max,␴3c = 2␭ 2 = 2 (11)
exp(Drc) ␴
3c

preparation (Been and Jefferies 1985; Ishihara 1993). The

drained tests were performed on loose, medium, and dense From Fig. 5(c), the axial strain (ε1)B that is associated with
sands prepared by different methods. Therefore, the following (ε v,shear)max is given as
formulation represents an average for different sample prepa-
ration techniques. The sands were isotropically consolidated 6(ε v,shear)max,␴3c
(ε1)B,␴3c = 6␭ 3 = (12)
to different confining pressures (␴3c). exp(␳Drc)

FIG. 5. Sand Properties versus Axial and Volumetric Strains, and Slopes at Major Points on Volumetric Strain Curve due to Shear
Stresses

652 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / AUGUST 1999

FIG. 6. Sand Properties versus Isotropic Volumetric Strain at
Consolidation Pressure

Similar to point B, the slope and coordinates at point C on

the ε v,shear-ε1 curve can be evaluated based on the data pre-
sented in Figs. 5(d–f ). The value of the volumetric strain at
point C is assessed using the data plotted in Fig. 5(d), and is
obtained as

冋 册
0.2
(ε1)B,␴3c
(ε v,shear)C,␴3c = ␭ 4 = (ε v,shear)max,␴3c [1 ⫹ (Sf )␴3c] (13)
(ε1)C,␴3c
The value of the axial strain at point C [Fig. 5(e)] is related
to (ε v,shear)max, and ␸, and is given by
(ε1)C,␴3c = 6␭ 5 = 6(ε v,shear)max,␴3c exp(tan2␸) (14)
FIG. 7. Isotropic Consolidated-Rebounded Volumetric
Fig. 5(f ) indicates the variation of the slope Sf at point C, Change of Saturated Sands
which is expressed as
(Sf)␴3c = ⫺␭ 6 = ⫺␳0.5Drc tan2␸ (15)
ically rebounded sand under different values of the OCR. The
Based on the experimental data of the isotropically consol- drained behavior of consolidated-rebounded sand during shear
idated drained tests presented in Fig. 6, the isotropic volu- loading is expressed by the stress-axial strain-volumetric strain
metric strain, (ε v)c, to pressure ␴3c is given as relationship due to shear, ␴d-ε1-ε v,shear. The following equations
provide the terms of the boundary conditions that are required
(ε v)c = ␭ 7 = ε 50 exp[Drc(1 ⫹ ␳)] (16) to plot the ε1-ε v,shear relationship at OCR greater than 1. These
equations are a function of sand properties and the ␴d -ε1-
FORMULATION OF DRAINED ε v,shear relationship at OCR equal to 1, as discussed in the pre-
REBOUNDED BEHAVIOR vious section.
Isotropic Rebounded Volume Change of Saturated To assess the drained stress-strain relationship of the re-
Sand (␴¯ 3-␧v,iso Relationship) bounded sand under a reduced confining pressure, ␴¯ 3, the axial
strain at SL of 50% is given as
Fig. 7 presents the observed isotropically consolidated-re-
bounded behavior of Nevada, Ione (Norris et al. 1997), and (ε 50)␴3c
(ε 50)␴¯ 3 = ␳0.25 (19)
Ottawa sands (Dakoulas and Yuanhui 1992) for different val- OCR 0.5
ues of the consolidation pressure, ␴3c (400, 800, and 300 kPa),
and OCR. The observed data of the sands considered are em- where ε 50 at ␴3c (i.e., OCR = 1) is given by (8). Using (ε 50)␴¯ 3
ployed to assess an empirical relationship that expresses the in (5)–(7), the drained stress-strain curve of a sand at a re-
rebounded behavior of the sand under decreasing values of the bounded confining pressure, ␴¯ 3 (OCR > 1), can be evaluated.
confining pressure (Fig. 7). The isotropic rebounded volumet- The empirically calculated slopes and coordinates at points
ric strain of the saturated sand (ε v,iso) is given as A, B, and C on the ε1-ε v,shear curve of the isotropically consol-
idated-rebounded drained test at ␴¯ 3 (OCR > 1) are used in the
(ε v)c determination of constants a–f in (9), as follows:
ε v,iso = (ε v)c ⫺ (17)
OCR ␩
where • The initial slope at point A

冋 册
0.25
␳0.1 ␴3c (SA)␴3c (ε v,shear)max,␴¯ 3
␩= exp(0.5␳Drc) and OCR = (18) (SA)␴¯ 3 = (20)
4 ␴
¯3 OCR 0.5 (ε v,shear)max,␴3c

Drained Behavior of Isotropically Consolidated- • The volumetric strain due to shear at point B
Rebounded Saturated Sand (ε v,shear)max,␴3c
(ε v,shear)max,␴¯ 3 = (21)
Two series of isotropically consolidated rebounded drained OCR ␮

冉 冊
tests (Norris et al. 1997) performed on two different sands
(Nevada and Ione sand) are employed to develop empirical Dr␳
␮ = ␳0.8 exp (22)
equations in order to describe the drained behavior of isotrop- OCR

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / AUGUST 1999 / 653

 • The axial strain at point B change curves exhibit the same slopes, Sf , as that at ␴3c

冋 册
0.5
(ε v,shear)max,␴¯ 3
(ε1)B,␴¯ 3 = (ε1)B,␴3c
(ε v,shear)max,␴3c
• Volumetric and axial strains, and the slope at point C
(ε v,shear)C,␴3c
(ε v,shear)C,␴¯ 3 = (ε v,shear)max,␴¯ 3
(ε v,shear)max,␴3c
(OCR = 1)
(23) (Sf)␴¯ 3 = (Sf)␴3c (26)
This approach is more accurate if the value of ε 50 at con-
solidation pressure, ␴3c, is determined from testing rather than
(24) from reliance on (8). This, in turn, affects the determination
of the undrained response from the drained behavior.

冋 册
0.25
(ε1)B,␴¯ 3 Based on the approach presented, the drained behavior of
(ε1)C,␴¯ 3 = (ε1)B,␴3c (25) consolidated and overconsolidated sands can be assessed. Figs.
(ε1)B,␴3c
8 and 9 exhibit a comparison between observed and predicted
An interesting phenomenon reported by Norris et al. (1997) behavior for Nevada and Ione sand. The assessed stress-strain
is that all drained rebounded (i.e., overconsolidated) volume relationship and volume change curves show good agreement
with the experimental results performed by Norris et al.
(1997). In addition, the predicted and observed isotropically
consolidated and then rebounded volume change responses for
different sands exhibit a good match, as shown in Fig. 7.

LIMITATION FOR DILATIVE BEHAVIOR

The undrained strength of saturated sand under monotonic
loading in its dilative mode is limited, in the present paper, to
its drained strength because negative pore-water pressure (i.e.,
a drained Poisson’s ratio, ␯, >0.5) is not considered in this
study [Fig. 10(a)]. Once the excess pore-water pressure during
the dilative response becomes equal to zero, the undrained
response of the sand will be limited by its drained behavior,
as indicated in Fig. 10(b). Therefore, the approach presented
exhibits a conservative interpretation under monotonic loading
when dense sand generates negative excess pore-water pres-
sure. The incorporation of negative excess pore-water pressure
in dense sands using the approach presented is under investi-
gation and will be considered in a separate paper.

VERIFICATION OF PROPOSED APPROACH

FIG. 8. Predicted and Observed Drained Response of Isotrop-
ically Consolidated and Overconsolidated Nevada Sand
FIG. 9. Predicted and Observed Drained Response of Isotrop-
ically Consolidated and Overconsolidated Ione Sand
654 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / AUGUST 1999
The approach developed has been verified through various
comparisons to experimental results of different types of
sands. Some of these comparisons are presented in the current
paper, and the properties of these sands are presented in Table
2. The comparison study covers the density of sand from very
loose to dense sand under different values of the confining
pressure. Most of the undrained stress-strain responses and
stress paths assessed, using a computer program, exhibit good
agreement with the observed results, as seen in the accompa-
nying figures.
Figs. 11 and 12 show a comparison between the predicted
and measured undrained response of Nevada and Ione sand
(undrained stress-strain curve and effective stress path) under
monotonic loading. Predicted curves plotted in Figs. 11 and
12 are obtained using the equations given previously. Each
curve is represented by a large number of calculated points.
The assessed response shows good agreement with the un-
drained response observed by Norris et al. (1997).
The test results of the monotonic undrained behavior of Ot-
tawa sand (Vaid et al. 1989) are shown in Fig. 13(a). Fig. 13(a)
presents a comparison between the measured and assessed un-
drained stress-strain relationship of Ottawa sand using the
same consolidation pressure (196.2 kPa) and different relative
densities. The stress paths assessed for these tests using the
proposed approach are shown in Fig. 13(b). Fig. 13(b) shows
the capability of the approach in assessing the effective stress
path for undrained conditions exhibiting both contractive and
dilative behavior.
Data for Banding sand under undrained monotonic loading
as obtained by Castro (1969) are shown in Fig. 14 along with
the curves predicted by the proposed approach. The values of
ε 50 and the angle of internal friction of each tested sample were
obtained from reported drained stress-strain curves under the
 FIG. 10. Excess Pore-Water Pressure with Undrained Response of Contractive and Dilative Sands

FIG. 11. Predicted and Observed Undrained Response of Ne-

vada Sand

FIG. 13. Undrained Response of Ottawa Sand under Mono-

tonic Loading

associated ␴3c (Castro 1969). These tests were performed on

loose, medium dense, and dense sand at different confining
pressures varying from 30 kPa to 1,000 kPa. It was noticed
that a large number of the tested samples experienced a com-
plete liquefaction (steady state) response as a result of excess
pore-water pressure concentration at the middle zone of the
tested sample. Therefore, failure planes developed in the mid-
dle of the sample, creating a local zone of complete liquefac-
tion that occurs only in the lab. This explains the complete
liquefaction in the samples presented in Figs. 14(a) and 14(b)
as compared to the predicted (uniform) limited liquefaction
response.
The observed and predicted undrained response of Fraser
River sand is presented in Fig. 15. The experimental tests were
performed by Vaid and Thomas (1995). The data presented
describe the stress-strain relationship and effective stress path
of the sand at relative density (Drc) of 19% under different
values of consolidation pressure (␴3c = 1,200, 400, and 200
FIG. 12. Predicted and Observed Undrained Response of Ione kPa). Fig. 15 shows reasonable agreement between the pre-
Sand dicted and observed response.
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / AUGUST 1999 / 655
 FIG. 15. Undrained Response of Fraser River Sand under
Monotonic Loading

FIG. 14. Undrained Response of Banding Sand under Mono- representation of an increasing simple shear stress state. While
tonic Loading the normal stress on horizontal and vertical planes remains
constant (equal to ␴3c), the corresponding ␴¯ 3 and ␴¯ 1 become

SUMMARY AND CONCLUSIONS ␴

¯ 1 = ␴3c ⫹ ␶ max; ␴
¯ 3 = ␴3c ⫺ ␶ max (27)

The present paper has provided the formulation for assess- and
ing the drained and drained rebounded response (ε v-␴¯ 3 and
␴d -ε1-ε v) for sand for use in assessment of its undrained be-
havior (␴d -ε1 and stress path). The wide range in diversity of
␶ max =
␴d
2
; ␴d = SL␴df = SL␴
¯3 冋 冉
tan2 45 ⫹
␸
2
冊 册
⫺1 (28)

the responses as well as properties of the sands required that ␶ max ␴d /2 E

the dependence upon important sand properties and conditions ␥max = ε1(1 ⫹ ␯); G= = = (29)
␥max ε1(1 ⫹ ␯) 2(1 ⫹ ␯)
be evaluated. The formulation presented is a function of the
void ratio, confining pressure, and basic properties of sand, where ␯, ␥, and G represent Poisson’s ratio, shear strain, and
such as relative density, uniformity coefficient, and roundness, shear modulus, respectively.
which are typically available to the designer or can be obtained From (5)
from visual inspection. The developed formulas allow one to
␴d ␴df
predict the potential of sand to liquefy, the type of liquefaction E= = ␭ exp(⫺3.707SL) (30a)
(limited or complete liquefaction) or dilative behavior, the ε1 ε 50
peak and residual strength values, as well as the whole un- and
drained stress-strain curve and effective stress path. The de-
veloped approach has been validated using several compari- ␴df
sons with published results of undrained triaxial tests on Ei = ␭ at SL = 0 (30b)
ε 50
different sands (Nevada, Ione, Banding, Ottawa, and Fraser
River sand). The simplicity of this approach makes it an Therefore, the Young’s modulus reduction, E/E i, is simply
attractive general method (in a computer program) to charac-
E
terize the undrained behavior of sands (loose, medium, or = exp(⫺3.707SL) (31)
dense). Ei
Corresponding to this, the shear modulus reduction,
APPENDIX I. SIMPLE SHEAR CHARACTERIZATION G/G o, is

An expanding Mohr circle about an initial isotropic stress G E/[2(1 ⫹ ␯)] E (1 ⫹ ␯)

= = (32)
state (␴¯ m = ␴3c, K o = 1) can be considered as a two-dimensional Go E i /[2(1 ⫹ ␯ i)] E i (1 ⫹ ␯ i)

656 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / AUGUST 1999

where E/E i is given by (31). If one takes ␯ to vary linearly Boundary Conditions at Point C
with SL from ␯ i = 0.1 at SL = 0 to, say, ␯ = 1/2 at SL = 1/2,
then G/Go versus ␥ can be obtained in the following fashion: x = xC; y = yC ; ȳ = S C = S f (47)
Therefore
1. Choose SL.
2. Assess ε 1, ␯, and ␥max. yC = SA x C ⫹ cx 2C ⫹ dx 3C ⫹ ex 4C ⫹ fx 5C (48)
3. Assess E/E i and G/Go. By substituting for constants c and d [(45) and (46)] in (48)
Plott G/Go versus ␥max.
冉 冊
4.
5. Repeat steps 1–4. 3yB 2SA
yC = SA x C ⫹ 2 ⫺ ⫹ ex 2B ⫹ 2 fx 3B x 2C
xB xB
Fig. 3 was constructed in this fashion for a value of ε 50 =
0.003 at ␴3c = 100 kPa, and ␯ = 0.1 ⫹ 0.8SL, for SL from
0.0 to 0.5.
⫹ 冉 SA
x 2B
2yB
⫺ 3 ⫺ 2ex B ⫺ 3 fx B2
xB
冊 x C3 ⫹ ex C4 ⫹ fx C5
(49)

APPENDIX II
The ε 1-ε v,shear relationship (Fig. 4) is expressed by a fifth-
yC = 冉
SA x C ⫹
3yB x
xB2
2
C
⫺
2SA x
xB
2
SA x
⫹ 2 ⫺
xB
C 2yB x
x B3
3
C
冊
3
C

order binomial equation as follows: ⫹ e(x 2B x 2C ⫺ 2x B x 3C ⫹ x 4C) ⫹ f (2x 3B x 2C ⫺ 3x 2B x 3C ⫹ x 5C) (50)

y = a ⫹ bx ⫹ cx ⫹ dx ⫹ ex ⫹ fx
2 3 4 5
(33) Eq. (50) can be expressed in shorthand as
The slope of any point on the ε 1-ε v,shear relationship is given yC = M ⫹ eN ⫹ f O (51)
as
where M, N, and O symbolize parenthetic terms in (50).
ȳ = b ⫹ 2cx ⫹ 3dx 2 ⫹ 4ex 3 ⫹ 5fx 4 (34) Consequently, constant e is given as
where x represents axial strain ε 1; y symbolizes volumetric ( yC ⫺ M ⫺ f O)
e= (52)
strain due to shear ε v,shear; and ȳ indicates slope of a point on N
the curve of the ε 1-ε v,shear relationship (dε 1 /dε v,shear).
Using (52), constants c and d [(45) and (46)] can be ex-
Boundary Conditions at Point A pressed as

x = 0; y = 0; ȳ = SA (35) 3yB 2SA ( yC ⫺ M ⫺ f O)

c= ⫺ ⫹ x B2 ⫹ 2 fx B3 (53)
x 2B xB N
By substituting the above conditions in (33) and (34)
and
a = 0; b = SA (36)
SA 2yB ( yC ⫺ M ⫺ f O)
Therefore, (33) and (34) can be rewritten as d= ⫺ 3 ⫺ 2x B ⫺ 3 fx B2 (54)
x 2B xB N
y = SA x ⫹ cx 2 ⫹ dx 3 ⫹ ex 4 ⫹ fx 5 (37) The slope at point C is defined as S f and represents the sixth
ȳ = SA ⫹ 2cx ⫹ 3dx ⫹ 4ex ⫹ 5fx2 3 4
(38) boundary condition [(47)]. By substituting into (38)
S f = SA ⫹ 2cx C ⫹ 3dx 2C ⫹ 4ex 3C ⫹ 5fx 4C (55)
Boundary Conditions at Point B
By substituting for constants c, d, and e [(52)–(54)] in (55)
x = xB; y = yB ; ȳ = 0 (39)
Therefore
yB = SA x B ⫹ cx ⫹ dx ⫹ ex ⫹ fx
2
B
3
B
4
B
5
B (40)
S f = SA ⫹ 冋 6yB x C 4SA x C
xB2 ⫺
xB
⫹ 2x C x 2B
( yC ⫺ M ⫺ f O)
N
⫹ 4fx 3B x C 册
c=
yB
x 2B
⫺
SA
xB
⫺ dx B ⫺ ex B2 ⫺ fx B3 (41) ⫹ 冋
3SA x 2C 6yB x 2C
x 2B
⫺
x 3B
⫺ 6x B x C2
( yC ⫺ M ⫺ f O)
N
⫺ 9fx B2 x C2 册
and
yB = SA ⫹ 2cx B ⫹ 3dx ⫹ 4ex ⫹ 5fx = 0
2 3 4
(42)
冋
⫹ 4x 3C
( yC ⫺ M ⫺ f O)
N 册
⫹ 5fx 4C
(56)

冉 冊
B B B
2 2
6yB x C 4SA x C 3SA x 6yB x
SA ⫹ ⫺ ⫹ ⫺
C C
2yB Sf =
SA ⫹ ⫺2SA ⫺ 2dx 2B ⫺ 2ex 3B ⫺ 2 fx 4B ⫹ 3dx 2B x 2B xB x B2 x C3
xB
( yC ⫺ M ⫺ f O) ( yC ⫺ M ⫺ f O)
⫹ 4ex B3 ⫹ 5fx B4 = 0 (43) ⫹ 2x C x 2B ⫺ 6x B x 2C
N N

⫺SA ⫹
2yB
⫹ dx 2B ⫹ 2ex 3B ⫹ 3 fx 4B = 0 ( yC ⫺ M ⫺ f O)
xB
(44) ⫹ 4x 3C ⫹ f (4x 3B x C ⫺ 9x 2B x 2C ⫹ 5x 4C)
N (57)
SA 2yB ( yC ⫺ M )
d= ⫺ 3 ⫺ 2ex B ⫺ 3 fx B2 (45) Sf = Q ⫹ (2x 2B x C ⫺ 6x B x 2C ⫹ 4x 3C)
x 2B xB N
By substituting for constant d [(45)] in (41), constant c can O
be written as ⫹ f (6x B x 2C ⫺ 2x 2B x C ⫺ 4x 3C) ⫹ fP
N (58)
3yB 2SA where Q and P denote first and last parenthetic terms in (57).
c= ⫺ ⫹ ex B2 ⫹ 2 fx B3 (46)
x 2B xB Therefore
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / AUGUST 1999 / 657
 ( yC ⫺ M )
Sf ⫺ Q ⫺ (2x 2B x C ⫺ 6x B x 2C ⫹ 4x 3C)
N Found., 24(4), 30–48.
f= (59)
O Ishihara, K. (1993). ‘‘Liquefaction and flow failure during earthquakes.’’
P ⫹ (6x B x 2C ⫺ 2x 2B x C ⫺ 4x 3C)
N
According to (59), constant f can be evaluated as a function
of the slopes and coordinates of points A, B, and C on the
ε 1-ε v,shear relationship (Fig. 4). Having constant f, all other con-
stants (c, d, and e) of (37) can be determined using (52)–(54).
ACKNOWLEDGMENTS
The work described herein was supported by the U.S. Army Corps of
Engineers (USACE) and the California Department of Transportation
(Caltrans). The writers would like to acknowledge Richard Ledbetter, the
contract manager of the project (USACE), for his encouragement. Also,
the writers would like to thank the Caltrans group (Dr. Abbas Abghari,
Ken Jackura, Tim Leahy, Chris Campbell, Angel Perez-Cobo, and Tho-
mas Shantz) for their interest and encouragement.
APPENDIX III. REFERENCES
Ashour, M., Norris, G., and Pilling, P. (1998). ‘‘Lateral loading of a pile
in layered soil using the strain wedge model.’’ J. Geotech. and Geoen-
vir. Engrg., ASCE, 124(4), 303–315.
Been, K., and Jefferies, M. G. (1985). ‘‘A state of parameter for sands.’’
Géotechnique, London, 35(2), 99–112.
Casagrande, A. (1976). ‘‘Liquefaction and cyclic deformation of sands:
A critical review.’’ 5th Panamerican Conf. on Soil Mech. and Found.
Engrg.
Castro, G. (1969). ‘‘Liquefaction of sands,’’ PhD thesis, Division of En-
gineering and Applied Physics, Harvard University, Cambridge, Mass.
Castro, G., and Poulos, S. (1977). ‘‘Factors affecting liquefaction and
cyclic mobility.’’ J. Geotech. Engrg. Div., ASCE, 103(6), 501–516.
Castro, G., Poulos, S., France, J., and Enos, J. (1982). ‘‘Liquefaction
induced by cyclic loading.’’ Rep. NSF/SEE-82018 Prepared for Nat.
Sci. Found.
Dakoulas, P., and Yuanhui, S. (1992). ‘‘Fine Ottawa sand: Experimental
behavior and theoretical predictions.’’ J. Geotech. Engrg., ASCE,
118(12), 1906–1923.
658 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / AUGUST 1999
Fukushima, S., and Tatsuoka, F. (1984). ‘‘Strength and deformation char-
acteristics of saturated sand at extremely low pressures.’’ J. Soils and
Géotechnique, London, 43(3), 351–415.
Ishihara, K., Tatasuoka, F., and Yasuda, S. (1975). ‘‘Undrained defor-
mation and liquefaction of sand under cyclic stresses.’’ J. Soils and
Found., (1), 29–44.
Lee, K. L., and Seed, H. B. (1967). ‘‘Drained strength characteristics of
sands.’’ J. Soil Mech. and Found. Div., ASCE, 93(6), 117–141.
Mohamad, R., and Dobry, M. (1986). ‘‘Undrained monotonic and cyclic
triaxial strength of sand.’’ J. Geotech. Engrg., ASCE, 112(10), 941–
958.
Negussey, D., and Vaid, Y. P. (1990). ‘‘Stress dilatancy of sand at stress
ratio states.’’ J. Soils and Found., 30(1), 155–166.
Norris, G., Madhu, R., Valceschini, R., and Ashour, M. (1995). ‘‘Lique-
faction and residual strength of loose sands from drained triaxial tests.’’
Rep. No. CCEER-95-2 Prepared for U.S. Army Corps of Engrs., Civil
Engineering Department, University of Nevada at Reno.
Norris, G., Siddharthan, R., Zafir, Z., and Madhu, R. (1997). ‘‘Liquefac-
tion and residual strength of sands from drained triaxial tests.’’ J. Geo-
tech. Engrg., ASCE, 123(3), 220–228.
Norris, G., Zafir, Z., and Siddharthan, R. (1998). ‘‘An effective stress
understanding of liquefaction behavior.’’ J. Envir. & Engrg. Geosci.,
4(1), 93–101.
Norris, G. M. (1986). ‘‘Theoretically based BEF laterally loaded pile
analysis.’’ Proc., 3rd Int. Conf. on Numer. Methods in Offshore Piling,
361–386.
Powers, M. C. (1953). ‘‘A new roundness scale for sedimentary parti-
cles.’’ J. Sedimentary Petrology, 23(2), 117–119.
Seed, H. B., and Idriss, I. M. (1970). ‘‘Soil moduli and damping factors
for dynamic response analyses.’’ Rep. No. EERC 70-10, College of
Engineering, University of California, Berkeley, Calif.
Seed, H. B., and Lee, K. L. (1967). ‘‘Undrained strength characteristics
of cohesionless soil.’’ J. Soil Mech. and Found. Div., ASCE, 93(6),
333–360.
Vaid, Y. P., Chung, E. K. F., and Kuerbis, R. H. (1989). ‘‘Preshearing and
undrained response of sand.’’ J. Soils and Found., 29(4), 49–61.
Vaid, Y. P., and Thomas, J. (1995). ‘‘Liquefaction and postliquefaction
behavior of sand.’’ J. Geotech. Engrg., ASCE, 121(2), 163–173.