Published June 14, 2018
Original Research
Core Ideas
• We present a model to estimate
the variation of undrained shear
strength of unsaturated soils.
• Environmental factors influenced
the undrained shear strength of
unsaturated soils.
• The proposed model can be imple-
mented into a numerical analysis.
W.T. Oh, Dep. of Civil Engineering, Univ. of
New Brunswick, Head Hall, 17 Dineen Drive,
Fredericton, NB E3B 5A3, Canada; S. Vanapalli,
Dep. of Civil Engineering, Univ. of Ottawa,
Ottawa, ON K1N 6N5, Canada. *Correspon-
ding author (woh@unb.ca).
Received 29 Jan. 2018.
Accepted 10 Apr. 2018.
Citation: Oh, W.T., and S. Vanapalli. 2018.
Undrained shear strength of unsaturated
soils under zero or low confining pressures in
the vadose zone. Vadose Zone J. 17:180024.
doi:10.2136/vzj2018.01.0024
© Soil Science Society of America.
This is an open access article distributed under
the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-
nd/4.0/).
Undrained Shear Strength of
Unsaturated Soils under Zero or Low
Confining Pressures in the Vadose Zone
Won Taek Oh* and Sai Vanapalli
In geotechnical engineering practice, many projects such as stability analysis of
slopes or trenches and pavement design involve soils in vadose zones at shallow
depths, where soils are under unsaturated conditions. For these types of proj-
ects, undrained shear strength test results obtained under zero or low confining
pressure can be effectively used to analyze the behaviors of unsaturated soils.
Experimental procedures to determine the undrained shear strength for different
soil suction values, however, is time consuming even under low confining pres-
sure. For this reason, we attempted to develop a semi-empirical model to predict
the undrained shear strength of unsaturated soils at a shallow depth assuming
zero confining pressure as a function of soil suction. In addition, existing empiri-
cal or semi-empirical models are also presented along with the characteristic
behaviors of unsaturated soils under axial forces at low or zero confining pres-
sures. Finally, the advantages and disadvantages of each model are discussed
based on the comparison between the measured undrained shear strengths
under zero or low confining pressures with those predicted using the existing
and proposed models.
Abbreviations: SWCC, soil-water characteristic curve; UC, unconfined compression; UCS, unconfined
compressive strength; USS, undrained shear strength; VG, van Genuchten.
In conventional geotechnical engineering practice, soil behavior in earth-
work projects such as foundation design, slope or trench stability analysis, and retaining
wall design are analyzed based on the assumptions that (i) soils are in a saturated condi-
tion and (ii) drainage conditions are governed by either a perfectly drained condition or
an undrained condition in terms of pore-water pressure. However, there are limitations
in adopting the conventional approaches in geotechnical engineering practice because
most earthworks predominantly occur in vadose zones where soils exist under unsaturated
conditions. When soils get desaturated, the voids that were initially occupied by fluid
are replaced with free air. This leads to a change in soil suction, which plays a key role in
controlling the mechanical behavior of soils.
For unsaturated coarse-grained soils, the shear strength can be reliably analyzed based
on the assumption that both pore air and pore water are under drained conditions. In other
words, no change in the soil suction takes place throughout the shearing process until the
soils reach failure conditions. On the other hand, there are difficulties in predicting the
shear strength of unsaturated fine-grained soils due to the low air and water permeabil-
ity. In this case, soil suction continuously changes throughout the shearing process, and
the difference in soil suction between the initial and failure conditions can be positive,
negative, or negligible. These characteristic behaviors encourage the use of a total stress
approach instead of an effective stress approach in analyzing the mechanical behavior of
unsaturated fine-grained soils.
The shear strength of unsaturated soils can be determined experimentally using the
various laboratory testing methods such as consolidated drained, constant water con-
tent, unconsolidated undrained, and unconfined compression (UC) tests. Among these
tests, the UC test can be the most reasonable choice in estimating the undrained shear
strength of unsaturated fine-grained soils, extending the total stress approach. This is
Vadose Zone Journal | Advancing Critical Zone Science
not only because the UC test is the simplest and most economi-
cal, but many projects in geotechnical and pavement engineering
involve soils at shallow depths (i.e., stability of natural, expan-
sive, or embankment slopes and pavement design) (American
Association of State Highway and Transportation Officials,
1993; Whenham et al., 2007; Li and Zhang, 2015). Oh and
Vanapalli (2013) performed model footing tests in Indian Head
till for different soil suction values to investigate the bearing
capacity of unsaturated fine-grained soils. They showed that half
of the unconfined compressive strength (UCS) of unsaturated
soils can be used as an undrained shear strength of unsaturated
fine-grained soils without introducing significant error in the
total stress approach for footings near the soil surface.
Although UC tests are performed under zero confining
pressure, it is still challenging to prepare the soil specimens at
different soil suction values without causing notable changes to
the soil structure. For this reason, we attempted to develop a
semi-empirical model to predict the undrained shear strength
of unsaturated fine-grained soils in vadose zones under zero or Fig. 1. Stress paths during shearing process under zero confining pres-
low confining pressure as a function of the soil suction based on sure for unsaturated soils (si ® sfc: no change in soil suction; si ® sfd:
decrease in soil suction) (modified after Fredlund and Rahardjo, 1993).
the UCS of unsaturated fine-grained soils. In addition, exist-
ing empirical or semi-empirical models are also presented along
with the characteristic behaviors of unsaturated soils under increase, decrease (stress path si ® sfd), or remain constant (stress
axial forces under zero or low confining pressure. Finally, the path si ® sfc).
advantages and disadvantages of each model are discussed by Fredlund et al. (1978) proposed an equation for the shear
comparing the measured undrained shear strengths with those strength of unsaturated soils based on the extended Mohr–
predicted using various existing models. Coulomb failure envelope:

t ff = c ¢ + ( sf - ua )tan f ¢ + ( ua - uw )f tan f b [1]

66Models
Behavior of Unsaturated Soils under Zero
or Low Confining Pressures
The UC test is the most commonly used laboratory test to
determine the undrained shear strength of soils due to its simplic-
ity. For a saturated soil, UC tests are conducted with a relatively
fast shearing rate (typically 1% of a specimen height per minute)
to achieve undrained conditions in terms of the pore water, and
the results are analyzed assuming zero volume change (i.e., the
internal friction angle of a saturated soil from an unconsolidated
undrained test f u = 0 analysis). However, the soils encountered in
geotechnical projects are typically under unsaturated conditions
and the behavior of unsaturated soils during UC tests is different
from that of saturated soils. This is mainly because, unlike for satu-
rated soils, volume changes are inevitable in unsaturated soils even
under undrained conditions since volume changes in unsaturated
soils are associated with not only the volume of water but also the
volume of air. This phenomenon, in turn, leads to a change in the
soil suction during shearing under undrained conditions. Possible
stress paths during the shearing process under zero confining pres-
sure for unsaturated soils are well explained in Fig. 1. For a known
initial suction value, the stress state of a soil specimen for a UC test
can be represented as the point si under zero confining pressure
before shearing. During the shearing process, the soil suction can
where tff is the shear strength of an unsaturated soil, c¢ is effective
cohesion, f¢ is effective internal friction angle, (s f − ua) is net
normal stress at failure, (ua − uw)f is matric suction at failure, ua is
pore-air pressure, uw is pore-water pressure, and f b is the internal
friction angle due to the contribution of matric suction.
If soil suction does not change throughout the UC test, the
undrained shear strength of an unsaturated soil can be approxi-
mated as half of the UCS without introducing significant error
(Fig. 2). In the present study, half of the UCS of an unsaturated
soil, qu(unsat)/2, is defined as the undrained shear strength (USS)
of an unsaturated soil with the notation of cu(unsat):
qu(unsat)
= c u(unsat) » c ¢ + ( sf - ua )tan f ¢ + ( ua - uw )f tan fb [2]
2
where qu(unsat) is the UCS of an unsaturated soil and cu(unsat) is the
undrained shear strength of an unsaturated soil.
Characteristic behaviors of unsaturated soils subjected to
axial forces under zero or low confining pressure have been stud-
ied by several researchers. Some of the previous research available
in the literature is presented in this study with emphasis on the
stress–strain behavior and the changes in soil suction and volume
during the shearing process.

VZJ | Advancing Critical Zone Science p. 2 of 13


Fig. 2. Justification of using half of the unconfined compressive
strength (UCS) in estimating the shear strength of unsaturated soils
assuming no change in soil suction during the shearing process under
zero confining pressure (modified after Fredlund and Rahardjo, 1993).
Pineda and Colmenares (2005)
Pineda and Colmenares (2005) performed axial load tests on
statically compacted kaolinite specimens (plasticity index Ip = 38%;
diameter = 50 mm, height = 100 mm) under low confining pres-
sure (net normal stress [s3 − ua] = 5 kPa) to study the behaviors
of unsaturated fine-grained soils for different soil suction values. A
conventional triaxial cell apparatus was modified for the tests such
that the pore-water and pore-air pressures in the specimens could
be controlled and measured. Suction values in the specimens before
shearing ranged between 25 and 400 kPa, which corresponds to
a degree of saturation in the range of 75.10 to 90.35%. Changes
in the soil suction during shearing were monitored, extending the
axis-translation technique (Hilf, 1956). Stress–strain relationships
showed that specimens behaved like either normally consolidated
soils or overconsolidated soils for the initial soil suction values less
than or greater than 100 kPa, respectively. The ratios of soil suction
values at failure to the initial values were approximately in the range
of 0.9 to 1.6, which decrease as the initial soil suction increases.
Chae et al. (2010)
Chae et al. (2010) conducted UC tests on non-plastic silty soil
samples prepared by both dynamic and static compaction (diam-
eter = 50 mm, height = 120 mm). The UC test results only for
dynamically compacted soils (optimum, drier than optimum, and
wetter than optimum) are presented here. The rate of strain used
for the UC tests was 0.1% per minute. Initial soil suction values
in the specimens ranged between 0 and 40 kPa (corresponding
degree of saturation = 40–80%). The specimen compacted drier
than optimum exhibited negligible change in suction at failure
(less than about 5 kPa); however, soil suction in the specimens
compacted at optimum and wetter than optimum increased up to
VZJ | Advancing Critical Zone Science
about 23 kPa at failure. Volumetric strains at failure were approxi-
mately between −3 and 2%.
Park and Kwon (2011)
Park and Kwon (2011) performed UC tests for a statically
compacted silty sand sample (Ip = 4.1%; diameter = 50 mm, height
= 100 mm). The UC test apparatus used by Kato et al. (2005) and
Chae et al. (2010) were adopted in the testing program to mea-
sure UCS and soil suction. An axial loading rate of 0.1 mm/min
was used to shear the specimens. Volume change in the specimens
was measured by monitoring the change in water level in a double
burette that was connected to the cell. The range of initial suction
values used in the testing program was between 20.1 and 93.1 kPa,
and the maximum change in suction at failure was <6 kPa. The
volumetric strains at failure was <2%.
Li and Zhang (2015)
Li and Zhang (2015) investigated the stress–strain behavior of
unsaturated soils (compacted Fairbank silt, Ip = 3.1%) under low
confining pressures (s3 − ua = 5 and 40 kPa) at a constant rate of
strain (i.e., 0.3 mm/min). Two high-suction tensiometers (Ridley
and Burland, 1993) were used to monitor the soil suction during
shearing. Initial suction values used in the testing program ranged
from 51.1 to 604.9 kPa. Most specimens experienced a continuous
decrease in suction, with a maximum change of 17% at failure.
All tested specimens failed at axial strains of <2%. Initial specific
volumes of the specimens (1.711–1.734 m3/m3) and those at fail-
ure (1.704–1.752 m3/m3) indicated negligible volume changes
throughout the tests.
These UC test results for unsaturated soils clearly indicate
that (i) matric suction can increase, decrease, or remain constant
at failure, and (ii) volume change at failure may not be significant.
Estimating Undrained Shear Strength of
Unsaturated Soils Based on Unconfined
Compressive Strength
Figure 3 shows the USS of unsaturated soils for different
suction values and best-fitting curves using hyperbola and power
models (Table 1). Most data sets can be well plotted using the
hyperbola model (Eq. [3]) with high R2 values. In the case of the
data of Vanapalli et al. (2000; Fig. 3b) and Babu et al. (2005; Fig.
3c), although R 2 values are relatively high, the hyperbola model
does not capture the behaviors of the USS in the low suction range;
instead, a power model (three-parameter function; Eq. [4]) pro-
vides better fits including the USS at low suction range:
h1y
c u(unsat) = y0 + [3]
1 + h2 y
c u(unsat) = y0 + p1y p2 [4]
where y0 is the y intercept; h1, h2 , p1, and p2 are parameters; and
y is soil suction.
p. 3 of 13
 However, the parameters h1 and h2 (Eq. [3]) and p1 and p2 (Eq.
[4]) summarized in Table 1 do not show any rules or trends, which
are the keys in developing empirical or semi-empirical models.
Fig. 3. Undrained shear strength of unsaturated soils for different suc-
tion values and best-fitting curves using hyperbola and power models.
VZJ | Advancing Critical Zone Science
Here, the existing methodologies and approaches to predict the
UCS for different suction values are introduced.
Aitchison (1957)
Aitchison conducted UC tests for unsaturated heavy clays and
proposed an equation to predict the UCS of unsaturated soils by
extending the research of Bjerrum (1954):
D( log 10 C r )
= constant [5]
e
where D(log10C­r) is the change in true cohesion Cr, and e is the
void ratio. Equation [5] indicates that true cohesion and water
content in any soil can be represented as a linear relationship on
a semi-logarithmic scale. Equation [6] shows the pressure vs. void
ratio relationship of the consolidation process:
D( log 10 p )
= constant [6]
e
where D(log10p) is the change in effective consolidation pressure,
p. Combining Eq. [5] and [6] yields
C r = kp M [7]
where k and M are constants.
Extending Eq. [7], the UCS of an unsaturated soil can be
predicted using
m m
qu(unsat) = 2c u(unsat) = 2 K ( sa¢ ) = 2 K ( y ) [8]
where sa ¢ is ambient effective stress, y is soil suction, and K and
m are coefficients.
Aitchison (1957) investigated the coefficients K and m using
UC test results for seven heavy clays; however, no trends were
observed (see Fig. 4). To further investigate Eq. [8], the six sets
of UC test results for unsaturated soils (data in Table 1) avail-
able in the literature were analyzed in this study. The K and m
coefficients of Aitchison (1957) for each date set are summarized
in Table 2. No correlations were observed when K and m values
were plotted against Ip (Fig. 5). The coefficient m seems to be
decreasing with increasing K, but the trend is too weak to develop
any correlations (Fig. 6).
Vanapalli and Fredlund (1997)
The shear strength equation for unsaturated soils (i.e., Eq. [1])
can be rewritten as Eq. [9] and then modified as Eq. [10] to esti-
mate the UCS of unsaturated soils by setting s3 = 0 (UC test) and
replacing tanf b with Qk tanf¢ (Vanapalli et al., 1996; Vanapalli
and Fredlund, 1997):
s1 -s3
=
2
æ s +s3 ö [9]
c ¢ cos f ¢ + çç 1 - ua ÷÷÷sin f ¢ + ( ua - uw )tan fb cos f ¢
çè 2 ø
p. 4 of 13
Table 1. Models and parameters used for the best-fitting analysis for the unconfined compression test results for unsaturated soils.
Reference Ip yo h1 h2 p1 p2 R2
%
Hyperbola model (Eq. [3])
Vanapalli et al. (2000) 8 78.8 1370 66,100 0.95
Vanapalli et al. (2007) 15.5 12.5 87.8 147 0.98
Ridley (1993) 32 7.32 1210 4004 0.98
Chen (1984) 38 63.1 1.07 ´ 109 3.02 ´ 109 0.99
Pineda and Colmenares (2005) 38 19.9 272 1105 0.99
Babu et al. (2005) 60 14.6 75.2 4106 0.95
Power model (Eq. [4])
Vanapalli et al. (2000) 8 15.6 6.53 0.425 0.98
Vanapalli et al. (2007) 15.5 12.0 2.68 0.559 0.97
Ridley (1993) 32 3.81 0.892 0.811 0.98
Chen (1984) 38 64.0 0.272 1.043 0.99
Pineda and Colmenares (2005) 38 19.4 0.485 0.839 0.99
Babu et al. (2005) 60 13.0 1.52 0.388 0.98

Babu et al. (2005) examined the UCS of Red and Black

s1 c ¢ cos f ¢ + ( ua - uw ) êë(Q )tan f ¢ úû cos f ¢
é k ù
Cotton soils and suggested that the UCS of unsaturated soils can
c u(unsat) = = [10]
2 1- sin f ¢ be written in the form

where Q is the normalized volumetric water content (= q/q s, where
q is the volumetric water content and q s is the volumetric water
content under saturated conditions), and k is a fitting parameter
for shear strength (Garven and Vanapalli, 2006; Fig. 7).
Equation [10] was successfully used to predict the UCS of an
unsaturated silty soil (Vanapalli et al., 2000) for suction values in
the rage of 0 to 106 kPa. Equation [10] can also be used to estimate
f b of an unsaturated soil based on UC test results (Vanapalli and
Fredlund, 1997).
a 0 c u -a1 = ( ua - uw )(Qk )a 2 [11]
where a 0, a1, and a 2 are functions of the strength parameters of
soils (i.e., c¢ and f¢).
To further investigate the validity of Eq. [10], UC test results
presented by Cunningham et al. (2003) were analyzed (Ip = 18%)
in the present study. The soil-water characteristic curve (SWCC)
of the soil is shown in Fig. 8. Best-fit analysis was conducted using
the van Genuchten (1980) equation (i.e., the VG model):

q-q r éê 1 ùm
S re = = ú [12]
q s -q r êê 1 + (ay )n úú
ë û

where Sre is the effective degree of saturation, q s is the volumet-

ric water content under saturated conditions, q r is the residual

Table 2. Plasticity index Ip and coefficients K and m of Aitchison

(1957) for various soils based on unconfined compression test results.
Reference Ip K m
%
Vanapalli et al. (2000) 8 16.5 0.41
Vanapalli et al. (2007) 15.5 25.3 0.3
Ridley (1993) 32 1.77 0.81
Chen (1984) 38 0.88 0.63
Fig. 4. Summary of coefficients K and m (Eq. [8]) for various soils used Pineda and Colmenares (2005) 38 11.15 0.46
by Aitchison (1957) (modified after Aitchison, 1957). Babu et al. (2005) 60 11.8 0.26

VZJ | Advancing Critical Zone Science p. 5 of 13

 Fig. 5. Aitchison’s coefficients K and m against the plasticity index for
the soils summarized in Table 1.
volumetric water content, and a, n, and m (=1 − 1/n) are fitting
parameters. The parameters are shown in Fig. 8.
Figure 9 shows the measured and predicted (Eq. [10]) UCS
for different suction values using the data of Cunningham et al.
(2003). According to Garven and Vanapalli (2006), k ranges from
1 to 3 and k = 2.23 can be used for Ip = 18%. The value k = 1 pro-
vided better estimates when compared with the measured UCS
than those obtained using k = 2.23 or 3. This indicates that the
predicted UCS using Eq. [10] is sensitive to the fitting parameter k.
Chae et al. (2010)
Chae et al. (2010) suggested that there is a unique relationship
between the UCS and suction stress at failure. Stress components
resulting from bulk water and meniscus water are termed bulk
(pb) and meniscus (pm) stresses, respectively. The sum of these
two stresses is denoted as the suction stress (ps) (Karube and
Kato, 1996). The suction stress for different suction values can be
Fig. 6. Coefficients K vs. m for the soils used by Aitchison (1957) and
those summarized in Table 1.
VZJ | Advancing Critical Zone Science
Fig. 7. Variation of the fitting parameter k (Eq. [10]) with respect to
the plasticity index Ip (after Garven and Vanapalli, 2006).
estimated by adopting the VG model (Eq. [12]; Lu and Griffiths,
2004) and using
é 1 ùm
ps = pb + pm = êê nú
ú y [13]
êë 1 + (ay ) úû
Suction stress is located on the axis of net normal stress (i.e.,
s – ua) as an intercept (Karube et al., 1996; Lu and Likos, 2006).
Therefore, if it is assumed that suction stress at failure acts as a
confining pressure during a UC test, an equation can be derived
based on Mohr’s circle at failure (Fig. 10):
4sin f ¢
qu(unsat) = psf [14]
1- sin f ¢
where psf is the suction stress at failure.
Fig. 8. Soil-water characteristic curve for the soil used by Cunningham
et al. (2003).
p. 6 of 13
 Fig. 9. Measured and predicted (Eq. [10]) for different suction values
(data from Cunningham et al. 2003).
The internal friction angle in Eq. [14] is based on the net
normal stress. There is a limitation to the use of Eq. [14] because
predicting the suction stress at failure is still challenging.
Zhou et al. (2016)
A similar approach to that of Chae et al. (2010) was also used
by Zhou et al. (2016) to predict the UCS of unsaturated soils. The
unsaturated soil effective stress theory of Bishop (1959) is
s¢ = ( s- ua ) +c ( ua - uw ) [15]
Because s is zero during UC tests, the effective horizontal and ver-
tical stresses at failure conditions become sh ¢ and sv ¢, respectively:
s¢h = cy f ; s¢v = cy f + qu [16]
where y f is the soil suction at failure.
The effective stress parameter, c, in Eq. [15] can be evaluated
using the approach of Khalili and Khabbaz (1998):
ì
ï ( u - uw )
ï
ï c= a if ( ua - uw ) > ( ua - uw )b
ï
í ( ua - uw )b [17]
ï
ï
ï if ( ua - uw ) £ ( ua - uw )b
îc = 1
ï
where (ua − uw)b is the air-entry value of a soil.
By replacing psf and psf + qu in Fig. 10 with cy f and cy f +
qu, respectively, the UCS for different matric suction values can
be obtained using
-cy f sin f ¢ - c ¢ cos f ¢
qu(unsat) = [18]
0.5( sin f ¢ -1)
For the soil tested, these researchers suggested that a linear rela-
tionship can be used to predict the soil suction at failure based on
the initial suction value. Incorporating this linear relationship into
Eq. [18], the UCS of an unsaturated soil can also can expressed as
VZJ | Advancing Critical Zone Science
Fig. 10. Geometric relationship between the failure line and Mohr’s
stress circle for unsaturated soil at failure (after Chae et al., 2010).
-c ( d1y i +d 2 )sin f ¢ - c ¢ cos f ¢
qu(unsat) = [19]
0.5( sin f ¢ -1)
where d1 and d 2 are linear relationship fitting parameters and y i
is the initial soil suction.
The disadvantage of the approaches proposed by Chae et al.
(2010) and Zhou et al. (2016) are that they require not only shear
strength parameters for saturated conditions (i.e., c¢ and f¢) but
also the suction value at failure.
Proposed Model
According to Han and Vanapalli (2016), various types of
equations are available to predict the stiffness and shear strength
properties of unsaturated soils, and one of them can be written in
the form (Vanapalli et al., 1996; Gupta et al., 2007; Sawangsuriya
et al., 2009; Oh et al., 2009)
x
W= Wsat +Gy ( S r ) [20]
where W is the stiffness and shear strength properties of an unsat-
urated soil, W sat is the stiffness and shear strength properties of
unsaturated soils at saturation, G is a suction-related variable, y
is soil suction, Sr is the degree of saturation, and x is an exponent.
The exponent x in Eq. [20] is empirical in nature, and regres-
sion analysis is necessary to determine the value. Some researchers
have suggested using the effective degree of saturation instead of the
degree of saturation, as shown in Eq. [12], to avoid the regression
analysis procedure. The degree of saturation can replace the effective
degree of saturation in the case where a soil sample experiences rela-
tively small volume change during tests (Romero and Vaunat, 2000;
Tarantino and Tombolato, 2005; Alonso et al., 2010; Lu et al., 2010):
S r - S r0
S re = [21]
100 - S r0
p. 7 of 13
where Sre is the effective degree of saturation, Sr is the degree of q = M a ( p - ua ) + M w ( ua - uw ) [27]
saturation, and Sr0 is degree of residual saturation.
Adopting the form of Eq. [20] and extending the concept in where (p − ua) is the net mean stress, Ma is the total stress ratio,
Eq. [2] and Fig. 2, the USS of an unsaturated soil can be estimated and Mw is the suction ratio (both Ma and Mw are functions of the
using either degree of saturation).
Wheeler (1991) stated that Eq. [27] is useful to analyze exist-
qu(unsat) n ing experimental data but not practical since there is no way of
c u(unsat) = = c u(sat) +GUC y( S r ) [22]
2 predicting the variation of Sr during tests. He then suggested
that the shear strength of an unsaturated soil can be estimated by
or uncoupling two variables, (p − ua) and (ua − uw):

qu(unsat) q = M ( p - ua ) + f ( ua - uw ) [28]
c u(unsat) = = c u(sat) +GUC y( S re ) [23]
2
where M is the critical state stress ratio for saturated soils and f
where G UC is a suction-related variable based on UC test results, (ua − uw) is a function of suction.
and n is an exponent. Pineda and Colmenares (2005) analyzed the UCS results for
If the variable G UC in Eq. [22] and [23] can be written in a compacted kaolin samples and showed that the UCS can be pre-
form [cu(sat)/m](Pa/101.3), where m is a function of the plasticity dicted by extending the approach of Wheeler (1991) (i.e., Eq. [28]):
index of soil properties, Pa is the atmospheric pressure (in kPa),
and (Pa/101.3) is used to maintain the consistency in units, Eq. qu(unsat) = M ( p - ua ) + f ( ua - uw )
[22] and [23] can be rewritten, respectively, as [29]
= 0.81( p - ua )
+ éê 0.4208( ua - uw )- 0.0003( ua - uw ) + 23.35 ùú
2
æ öæ ö÷
c u(unsat) = c u(sat) + c u(sat) çç ÷÷÷çç ÷( S ) ë û
çè ÷øçè 101.3 ø÷ r
[24] The use of M (i.e., 0.81 for saturated conditions) and the
é æ öæ ö÷ ù
ê ç ÷ ç ú
= c u(sat) ê + ç ÷÷ç
ç ÷ç ÷÷( S r ) ú y intercept (i.e., 23.35) indicate that, although the behaviors of
ë è øè 101.3 ø û saturated and unsaturated soils during UC tests are different,
unconfined compression test results for saturated conditions can
be used as a reference strength to estimate the shear strength of an
æ y öæ P ö
c u(unsat) = c u(sat) + c u(sat) ççç ÷÷÷çç a ÷÷÷( S re ) unsaturated soil for different suction values.
è m ø÷çè 101.3 ø
[25]
é æ y öæ P ö ù
= c u(sat) êê1 + ççç ÷÷÷çç a ÷÷÷( S re )úú
÷ ç
66Validation of the Proposed Model
ë è m øè 101.3 ø û
To check the validity of the proposed model, six sets of UC
Although Eq. [25] requires one fewer parameter (i.e., n), more test results for compacted fine-grained soils (Table 1) were ana-
suction measurements are necessary to reliably estimate the resid- lyzed. Figure 11 shows the SWCCs of the soils. Best-fit analyses
ual degree of saturation based on the entire range of the SWCC.
Hence, in the present study, Eq. [24] was used to develop a model
to estimate the variation in the USS with respect to soil suction.
There is a fundamental flaw in Eq. [24] because, as explained
above, unsaturated soils undergo volume changes due to the
compressibility of an air–water mixture even under undrained
conditions while the volume change during UC tests for saturated
soils is neglected. This critical difference in the behaviors of satu-
rated and unsaturated soils during UC tests indicates that the USS
under saturated conditions, cu(sat) should be replaced with a certain
reference value, cu(ref):

é æ y öæ P ö nù
c u(unsat) = c u(ref) êê1 + çç ÷÷÷çç a ÷÷÷( S r ) úú [26]
çè m ÷øçè 101.3 ø
ë û

Toll (1990) proposed that the shear strength of unsaturated

soils can be estimated based on the critical states using Fig. 11. Soil-water characteristic curves of the soils used in the analysis.

VZJ | Advancing Critical Zone Science p. 8 of 13

were conducted using the VG model (Eq. [12]), and the fitting 10) provided good estimates, and then m increased with increasing
parameters for each SWCC are summarized in Table 3. Ip value. This behavior is shown in

Determination of the Fitting Parameter n ì

The fitting parameter n controls the variation in the USS of
unsaturated soils in a nonlinear pattern. To investigate the fitting
parameter, n values for two sets of UC test results were used—one
for the lowest Ip (= 8%, Vanapalli et al., 2000) and another one
for the highest Ip (= 60%, Babu et al., 2005) from Table 1. Figure
12a shows the measured and predicted undrained shear strengths
for different suction values using the UC test results of Vanapalli
et al. (2000). Combination of n = 2 and m = 10 provided the most
reasonable estimates. On the other hand, at the same m value, und-
rained shear strengths were either significantly overestimated or
underestimated when n = 1 or n = 3 was used for suction values
greater than about 40 kPa. This suction value is close to the air-
entry value of the soil. Due to this reason, no combination of n and
m provides good estimates for n = 1 or n = 3 (Fig. 12b). Figure 13
shows the measured and predicted USS using the UC test results
of Babu et al. (2005) and those predicted using the proposed model
with different sets of n and m. Combinations of both n = 2 and m
= 490 or n = 3 and m = 200 provided reasonable estimates. These
comparisons indicate that n = 2 can be used for soils with a wide
range of Ip values to estimate the variation in USS with respect to
suction. Using a constant n value also minimizes the number of
fitting parameters required in the proposed model. For the data of
Vanapalli et al. (2000) and Babu et al. (2005), the degree of satura-
tion for different suction values were obtained from the SWCCs.
ï m = 10 for 8 £ I p (%) £15.5
ï
ï
í [30]
ï
ï m = 1.77 exp éê 0.0937 ( I p )ùú for 15.5 £ I p (%) £ 60
ï
î ë û
The point obtained using the UC test results of Cunningham et
al. (2003) is explained below.
Initial Soil Suction vs. Soil Suction at Failure
As pointed out by Chae et al. (2010) and Zhou et al. (2016),
the UCS of unsaturated soils can be more reliably predicted when
soil suction at failure is used. Figure 19 shows five different plots
of soil suction vs. USS behaviors considering both initial and
failure conditions as detailed below using the UC test results of
Cunningham et al. (2003): the measured USS vs. y i, the measured
USS vs. y f, the predicted USS using Sri and y i, the predicted USS
using Srf and y f, and the predicted USS using Sri and y f, where
subscript i = initial and f = failure.
The results show that the USS values plotted against soil suc-
tion at failure are in good agreement when compared with the USS

Determination of the Fitting Parameter m

After the determination of n (i.e., 2) for the two sets of UC
test results in Table 1 (i.e., Vanapalli et al., 2000; Babu et al., 2005),
the remaining four sets of UC test results were analyzed to deter-
mine the fitting parameter m with n = 2. Figures 14 to 17 show
measured and predicted USS using the UC test results in Table 1
except for those of Vanapalli et al. (2000) and Babu et al. (2005).
The m values required to have best-fitting curves are summarized
in Table 4. Figure 18 shows the relationship between the fitting
parameter m and the plasticity index for the soils used in the analy-
ses. The relationship can be represented as a bilinear curve on a
semi-logarithmic scale. For the low Ip values, a constant m value (i.e.,

Table 3. Van Genuchten model parameters for the various soils ana-
lyzed in this study.
Reference a n m (= 1 − 1/n)
Vanapalli et al. (2000) 1.0023 ´ 10−2 1.300 0.23077
Vanapalli et al. (2007) 7.6576 ´ 10−2 1.511 0.33819
Ridley (1993) 6.2100 ´ 10−4 8.911 0.88778
Chen (1984) 5.7531 ´ 10−3 1.522 0.34297
Pineda and Colmenares 5.5388 ´ 10−2 1.251 0.20045
(2005) Fig. 12. Measured and predicted (Eq. [24]) undrained shear strength
Babu et al. (2005) 4.4403 ´ 10−3 1.220 0.18033 (USS) with different sets of n and m (data from Vanapalli et al., 2000).

VZJ | Advancing Critical Zone Science p. 9 of 13

Fig. 13. Measured and predicted (Eq. [24]) undrained shear strength
(USS) with different sets of n and m (data from Babu et al., 2005).
Fig. 14. Measured and predicted (Eq. [24]) undrained shear strength
(USS) (data from Vanapalli et al., 2007).
Fig. 15. Measured and predicted (Eq. [24]) undrained shear strength
(USS) (data from Ridley, 1993).
VZJ | Advancing Critical Zone Science
Fig. 16. Measured and predicted (Eq. [24]) undrained shear strength
(USS) (data from Chen, 1984).
Fig. 17. Measured and predicted (Eq. [24]) undrained shear strength
(USS) (data from Pineda and Colmenares, 2005).
predicted using Sri and y i or Srf and y f. This indicates that the
proposed model that uses Sri + y i can predict the variation of the
USS with respect to soil suction without knowing Srf and y f. This
is attributed to the reason that the increase (or decrease) in soil
suction at failure leading to a decrease (or increase) in the degree
of saturation, which minimizes the difference in USS predicted
using Sri and y i or Srf and y f.
Limitations of the Proposed Model
The model proposed in the present study was validated using
limited sets of UC test results for unsaturated fine-grained soils
whose Ip values are between 8 and 60% for drying paths. Hence,
more UC test results need to be analyzed to increase the reliability
of the proposed model. The measured and predicted USS values
for various suction values using the data of Ridley (1993; Fig. 15)
showed that the discrepancy increases as soil suction approaches
the residual suction value. Similar behavior was observed for the
p. 10 of 13
Table 4. Plasticity index Ip and the fitting parameter m of the proposed
model using an exponent n = 2 for various soils based on unconfined
compression test results.
Reference Ip m
%
Vanapalli et al. (2000) 8 10
Vanapalli et al. (2007) 15.5 10
Ridley (1993) 32 35
Chen (1984) 38 60
Pineda and Colmenares (2005) 38 65
Babu et al. (2005) 60 490

data of Aitchison (1957; Fig. 20). This is because, theoretically, the

Fig. 19. Measured and predicted (Eq. [24]) undrained shear strength
net contribution of soil suction toward the shear strength of soils (USS) considering both initial (yi and degree of saturation Sri) and
starts decreasing as the soil suction approaches the residual suction failure (yf and Srf ) conditions (data from Cunningham et al., 2003).
value (Vanapalli et al., 1996). In other words, the proposed model
may not provide good estimates of the USS for suction values
greater than the residual suction value.
It also should be noted that the sample volume change during
wetting–drying cycles or the shearing process was not taken into
account in developing the proposed model (i.e., Eq. [24]). As men-
tioned above, volume changes are inevitable in unsaturated soils
even under undrained conditions (i.e., in spite of a constant water
content condition). Measurement of the volume change during
wetting–drying cycles or the shearing process is one of the major
challenges in unsaturated soil testing, which can be achieved with
double cell or on-sample strain transducers. However, as described
above, the proposed model provides reasonable estimation of the
USS of unsaturated cohesive soils using information on only the
initial soil suction and degree of saturation.

Fig. 20. (a) Soil-water characteristic curve and (b) measured and
Fig. 18. Relationship between the plasticity index Ip and the fitting predicted (Eq. [24]) undrained shear strength (USS) (data from
parameter m. Aitchison, 1957).

VZJ | Advancing Critical Zone Science p. 11 of 13

66Summary and Conclusions
Unconfined compressive strengths are widely used in geotech-
nical practice to estimate the USS of soils. Under field conditions,
environmental factors such as water infiltration or evaporation
through the surface of a soil can cause variations in soil suction,
which in turn leads to changes in the USS. Hence, to numerically
model soil behaviors under wetting and drying conditions, it is nec-
essary to use a model that can predict the variation in the USS with
respect to soil suction. In the present study, four different empirical
or semi-empirical models available in the literature were revisited
and their disadvantages discussed. To overcome the disadvantages
of the existing models, a new model is proposed to estimate the
USS of unsaturated soils for different suction values under zero
or low confining pressures following drying paths. The findings
from this study can be summarized as follows:
1. The suction value in a compacted soil can increase, decrease, or
remain constant during the shearing process under axial forces
in an unconfined condition, which makes it more challenging
to develop a model to predict the variation of USS with respect
to soil suction.
2. The model proposed by Aitchison (1957) has difficulties in
correlating the coefficients K and m to soil properties such
as the plasticity index. The predicted USS using the equation
proposed by Vanapalli and Fredlund (1997) is too sensitive to
the fitting parameter k. Lastly, the methodologies proposed
by Chae et al. (2010) and Zhou et al. (2016) require the soil
suction at failure, which is impossible to predict.
3. The proposed model requires the USS under saturated
conditions and the soil-water characteristic curve along with
two fitting parameters, n and m. However, since n = 2 provides
good comparisons when compared with the measured USS for
a wide range of Ip, only one fitting parameter (i.e., m) is required.
4. The predicted USS from the proposed model showed good
agreement with the measured USS using the degree of
saturation and soil suction under both initial and failure
conditions. In other words, the proposed model does not
require the degree of saturation and soil suction at failure. The
reason for the good agreement between the predicted USS
under both initial and failure conditions is that the change in
the soil suction at failure is compensated by the change in the
degree of saturation.
5. The USS predicted using the proposed model decreases as soil
suction approaches the residual suction value even though the
USS gradually increases or remains constant for suction values
greater than the residual suction value. Hence, the proposed
model may not be applicable to the suction values greater than
the residual suction value.
References
Aitchison, G.D. 1957. The strength of quasi-saturated and unsaturated
soils in relation to the pressure deficiency in the pore water. In:
Proceedings of the 4th International Conference on Soil Mechanics
and Foundation Engineering, London. 12–24 Aug. 1957. Butterworths,
London. p. 135–139.
Alonso, E.E., J.M. Pereira, J. Vaunat, and S. Olivella. 2010. A microstruc-
VZJ | Advancing Critical Zone Science
turally based effective stress for unsaturated soils. Geotechnique
60:913–925. doi:10.1680/geot.8.P.002
American Association of State Highway and Transportation Officials. 1993.
Guide for design of pavement structures. AASHTO, Washington, DC.
Babu, G.L.S., R.S. Rao, and J. Peter. 2005. Evaluation of shear strength func-
tions based on soil water characteristic curves. J. Test. Eval. 33:461–465.
Bishop, A.W. 1959. The principle of effective stress. Tek. Ukebl. 106:859–863.
Bjerrum, L. 1954. Theoretical and experimental investigations on the shear
strength of soils. Ph.D. diss. Swiss Fed. Inst. Technol., Zurich, Switzerland.
Chae, J., B. Kim, S.-W. Park, and S. Kato. 2010. Effect of suction on uncon-
fined compressive strength in partly saturated soils. KSCE J. Civ. Eng.
14:281–290. doi:10.1007/s12205-010-0281-7
Chen, J.C. 1984. Evaluation of strength parameters of partially saturated
soils on the basis of initial suction and unconfined compression
strength. M.S. thesis. Asian Inst. Technol., Bangkok, Thailand.
Cunningham, M.R., A.M. Ridley, K. Dineen, and J.B. Burland. 2003. The me-
chanical behavior of a reconstituted unsaturated silty clay. Geotech-
nique 53:183–194. doi:10.1680/geot.2003.53.2.183
Fredlund, D.G., N.R. Morgenstern, and R.A. Widger. 1978. The shear
strength of unsaturated soils. Can. Geotech. J. 15:313–321.
doi:10.1139/t78-029
Fredlund, D.G., and H. Rahardjo. 1993. Soil mechanics for unsaturated soils.
John Wiley & Sons, New York. doi:10.1002/9780470172759
Garven, E., and S.K. Vanapalli. 2006. Evaluation of empirical procedures
for predicting the shear strength of unsaturated soils. In: G.A. Miller
et al., editors, Proceedings of the 4th International Conference
on Unsaturated Soils, Carefree, AZ. 2–6 Apr. 2006. ASCE Geotech.
Spec. Publ. 147. Am. Soc. Civ. Eng., Washington, DC. p. 2570–2581.
doi:10.1061/40802(189)219
Gupta, S.C., A. Ranaivoson, T.B. Edil, C.H. Benson, and A. Sawangsuriya.
2007. Pavement design using unsaturated soil technology, Rep. MN/
RC-2007-11. Minnesota Dep. of Transp., St. Paul.
Han, Z., and S.K. Vanapalli. 2016. Stiffness and shear strength of unsatu-
rated soils in relation to soil-water characteristic curve. Geotechnique
66:627–647. doi:10.1680/jgeot.15.P.104
Hilf, J.W. 1956. An investigation of pore-water pressure in compacted co-
hesive soils. Ph.D. diss. US Dep. Interior, Bur. Reclam., Design Construct.
Div. Tech. Mem. 654. Univ. of Colorado, Denver.
Karube, D., and S. Kato. 1996. An ideal unsaturated soil and the Bishop’s
soil. In: Proceedings of the 13th International Conference on Soil Me-
chanics and Foundation Engineering, New Delhi, India. 5–10 Jan. 1994.
CRC Press, Boca Raton, FL. p. 43–46.
Karube, D., S. Kato, K. Hamada, and M. Hond. 1996. The relationship
between the mechanical behavior and the state of porewater in
unsaturated soil. (In Japanese, with English abstract online.) J. Jpn. Soc.
Civ. Eng. 535:83–92. doi:10.2208/jscej.1996.535_83
Kato, S., Y. Yoshimura, and D.G. Fredlund. 2005. Role of matric suction in
the interpretations of unconfined compression tests. In: Proceedings
of the 58th Canadian Geotechnical Conference, Saskatoon, SK. 18–21
Sept. 2005. Can. Geotech. Soc., Alliston, ON. p. 410–415.
Khalili, N., and M.H. Khabbaz. 1998. A unique relationship for c for the de-
termination of the shear strength of unsaturated soils. Geotechnique
48:681–687. doi:10.1680/geot.1998.48.5.681
Li, L., and X. Zhang. 2015. Modified unconfined compression testing sys-
tem to characterize stress-strain behavior of unsaturated soils at low
confining stresses. Transp. Res. Rec. 2510:54–64. doi:10.3141/2510-07
Lu, N., J.W. Godt, and D.T. Wu. 2010. A closed-form equation for ef-
fective stress in unsaturated soil. Water Resour. Res. 46:W05515.
doi:10.1029/2009WR008646
Lu, N., and D.V. Griffiths. 2004. Profiles of steady-state suction stress
in unsaturated soils. J. Geotech. Geoenviron. Eng. 130:1063–1076.
doi:10.1061/(ASCE)1090-0241(2004)130:10(1063)
Lu, N., and W.J. Likos. 2006. Suction stress characteristic curve for
unsaturated soil. J. Geotech. Geoenviron. Eng. 132:131–142.
doi:10.1061/(ASCE)1090-0241(2006)132:2(131)
Oh, W.T., and S.K. Vanapalli. 2013. Interpreting the bearing capac-
ity of unsaturated fine-grained soils using modified effec-
p. 12 of 13
 tive and total stress approaches. Int. J. Geomech. 13:769–778.
doi:10.1061/(ASCE)GM.1943-5622.0000263
Oh, W.T., S.K. Vanapalli, and A.J. Puppala. 2009. Semi-empirical model
for the prediction of modulus of elasticity for unsaturated soils. Can.
Geotech. J. 46:903–914. doi:10.1139/T09-030
Park, S.-W. and H.-K. Kwon. 2011. Suction stress and unconfined
compressive strength of compacted unsaturated silty sand. (In
Korean, with English abstract.) J. Korean Geotech. Soc. 27(8):31–37.
doi:10.7843/kgs.2011.27.8.031
Pineda, J.A., and J.E. Colmenares. 2005. Influence of matric suction on the
shear strength of compacted kaolin under unconfined conditions: 1.
Experimental results. In: A. Tarantino et al., editors, Proceedings of the
International Symposium on Advanced Experimental Unsaturated Soil
Mechanics, Trento, Italy, 27–29 June 2005. CRC Press, Boca Raton, FL.
p. 215–220.
Ridley, A.M. 1993. The measurement of soil moisture suction. Ph.D. diss.
Univ. of London.
Ridley, A.M., and J.B. Burland. 1993. A new instrument for the measure-
ment of soil moisture suction. Geotechnique 43:321–324. doi:10.1680/
geot.1993.43.2.321
Romero, E., and J. Vaunat. 2000. Retention curves in deformable clays.
In: A. Tarantino and C. Mancuso, editors, Experimental evidence and
theoretical approaches in unsaturated soils. Balkema, Rotterdam, the
Netherlands. p. 91–106.
Sawangsuriya, A., T.B. Edil, and P.J. Bosscher. 2009. Modulus–suc-
tion–moisture relationship for compacted soils in postcom-
paction state. J. Geotech. Geoenviron. Eng. 135:1390–1403.
doi:10.1061/(ASCE)GT.1943-5606.0000108
Tarantino, A., and S. Tombolato. 2005. Coupling of hydraulic and me-
chanical behaviour in unsaturated compacted clay. Geotechnique
55:307–317. doi:10.1680/geot.2005.55.4.307
VZJ | Advancing Critical Zone Science
Toll, D.G. 1990. A framework of unsaturated soil behaviour. Geotechnique
40:31–44. doi:10.1680/geot.1990.40.1.31
Vanapalli, S.K., and D.G. Fredlund. 1997. Interpretation of undrained shear
strength of unsaturated soils in terms of stress state variables. In: T.M.P.
de Campos and E.A. Vargas, Jr., editors, Proceedings of the 3rd Brazil-
ian Symposium on Unsaturated Soils, Rio de Janeiro. 22–25 Apr. 1997.
Vol. 1. Freitas Bastos, Rio de Janeiro. p. 35–45.
Vanapalli, S.K., D.G. Fredlund, D.E. Pufahl, and A.W. Clifton. 1996. Model
for the prediction of shear strength with respect to soil suction. Can.
Geotech. J. 33:379–392. doi:10.1139/t96-060
Vanapalli, S.K., W.T. Oh, and A.J. Puppala. 2007. Determination of the
bearing capacity of unsaturated soils under undrained loading condi-
tions. In: Proceedings of the 60th Canadian Geotechnical Conference,
Ottawa, ON, Canada. 21–24 Oct. 2007. Can. Geotech. Soc., Alliston, ON.
p. 1002–1009.
Vanapalli, S.K., A. Wright, and D.G. Fredlund. 2000. Shear strength of two
unsaturated silty soils over the suction range from 0 to 1,000,000 kPa.
In: Proceedings of the 53rd Canadian Geotechnical Conference, Mon-
treal. 15–18 Oct. 2000. Can. Geotech. Soc., Alliston, ON. p. 1161–1168.
van Genuchten, M.Th. 1980. A closed-form equation of predicting the hy-
draulic conductivity of unsaturated soils. Soil Sci. Soc. Am. J. 44:892–
898. doi:10.2136/sssaj1980.03615995004400050002x
Wheeler, S.J. 1991. An alternative framework for unsaturated soil behav-
iour. Geotechnique 41:257–261. doi:10.1680/geot.1991.41.2.257
Whenham, V., M. De Vos, C. Legrand, R. Charlier, J. Maertens, and J.-C.
Verbrugge. 2007. Influence of soil suction on trench stability. Springer
Proc. Phys. 112:495-501. doi:10.1007/3-540-69873-6_49
Zhou, W.-H., X. Xu, and A. Garg. 2016. Measurement of unsaturated
shear strength parameters of silty sand and its correlation with
unconfined compressive strength. Measurement 93:351–358.
doi:10.1016/j.measurement.2016.07.049
p. 13 of 13