You are on page 1of 17

Bearing Capacity of Shallow Foundations on

Unsaturated Soils: Analytical Approach with 3D Numerical


Simulations and Experimental Validations
Amir Akbari Garakani 1; Hamed Sadeghi 2; Soheil Saheb 3; and Amin Lamei 4
Downloaded from ascelibrary.org by Indian Institute Of Technology, Indore on 01/16/24. Copyright ASCE. For personal use only; all rights reserved.

Abstract: The ultimate bearing capacity of shallow foundations on unsaturated soils is investigated through the concept of suction-
dependent effective stress. The first approach is a new analytical solution considering the influence of matric suction on ultimate bearing
capacity by extending Vesic’s solution for saturated soils. Accordingly, a modification factor has been introduced as a nonlinear function of
matric suction for tuning the cohesion-dependent component in the bearing capacity equation. The second approach is incorporating the
unsaturated effective stress state in conjunction with the suction-dependent cohesion into a three-dimensional (3D) finite-difference code.
In developing the 3D simulations, the variation in matric suction versus the depth of the soil was considered as well as the dependency of the
degree of saturation on the soil suction. In addition, in three-dimensional numerical analyses, the input material parameters were modified to
take into account the suction-stress concept in unsaturated soils. To assess the validity of the analytical and numerical approaches, four series
of experimental data from physical plate load tests conducted under different matric suctions and embedment depths of the footing were
selected. Accordingly, water retention curves of different test materials were considered as key input parameters used in both approaches to
improve model predictions. Results from the analytical approach show the dependency of the presented correction factor on the soil proper-
ties, geometrical aspects of the foundation, and its embedding depth. In addition, the 3D numerical simulation revealed the suitable func-
tionality of the effective stress approach on predicting the load-displacement behavior of shallow foundations on unsaturated soils.
Moreover, the comparison between analytical, numerical, and experimental data shows a good conformance between the experimental
test results, analytical solutions, and numerical predictions, especially for sandy soils. DOI: 10.1061/(ASCE)GM.1943-5622.0001589.
© 2019 American Society of Civil Engineers.
Author keywords: Ultimate bearing capacity; Unsaturated soils; Effective stress; Analytical solution; 3D numerical modeling;
Plate load tests.

Introduction and Background the surface in some semiarid to arid areas, implying that the shallow
load-bearing soil layers may never become fully saturated.
Shallow foundations are generally designed so that the loads from the Due to its utmost importance, the bearing capacity of shallow
superstructure can be safely transferred to the soil underneath in such a foundations has received a great deal of attention from numerous
way to ensure the settlement criteria enacted by the design and con- investigators over the last century. By loading a strip footing until
struction codes are met. The design of this category of structures, it penetrates into the underneath soil, Prandtl (1921) first investi-
being widely used in buildings, retaining walls, power plants, bridges, gated the bearing capacity of soils. As a result, the ultimate bearing
antenna towers, and silos, needs to be not only safe but also economi- capacity qu was defined as the applied stress at which stability
cal. Regarding the latter aspect, less attention has yet been given to the failure occurs. After Prandtl’s pioneering work, several researchers
strengthening role of interparticle capillary forces in unsaturated states investigated the bearing capacity of soils and tried to provide com-
on enhancement of bearing capacity, for the sake of a conservative prehensive understanding about it (e.g., Terzaghi 1943; Terzaghi
design. However, groundwater tables are located tens of meters below and Peck 1948; Meyerhof 1951, 1956; Larkin 1968; Vesic 1973;
Bolton and Lau 1993). The focus of all these studies was on satu-
1
Assistant Professor, Structural Research Dept., Niroo Research Insti- rated soils considering the shape, dimension, inclination, and depth
tute, Tehran 1468617151, Iran (corresponding author). ORCID: https:// of the footing. Although the bearing capacity of soils is convention-
orcid.org/0000-0002-9696-3455. Email: aakbari@nri.ac.ir
2 ally determined assuming fully saturated conditions, shallow foun-
Assistant Professor, Dept. of Civil Engineering, Sharif Univ. of Tech-
nology, Tehran 1458889694, Iran. ORCID: https://orcid.org/0000-0002 dations in many cases are located above the groundwater table
-3453-9309. Email: hsadeghi@sharif.edu where the soil is mostly in unsaturated conditions. About 33% of
3
Master of Geotechnical Engineering, Dept. of Civil Engineering, the earth’s surface is considered to have arid and semiarid climatic
Sharif Univ. of Technology, Tehran 1458889694, Iran. ORCID: https:// conditions where the soil is unsaturated (Vanapalli et al. 2010).
orcid.org/0000-0003-2574-2847. Email: ingssaheb@gmail.com Therefore, design and construction of shallow foundations in these
4
Master of Geotechnical Engineering, Dept. of Civil Engineering, vast areas require an understanding of bearing capacity and settle-
Sharif Univ. of Technology, Tehran 1458889694, Iran. Email: ment behavior of soils in unsaturated states.
aminlamei94@gmail.com
Ignoring the unsaturated state of soil leads to conservative
Note. This manuscript was submitted on December 20, 2018; approved
on July 25, 2019; published online on December 17, 2019. Discussion per- and costly designs of shallow footings because unsaturated soils
iod open until May 17, 2020; separate discussions must be submitted for tend to have greater bearing capacity in comparison to saturated
individual papers. This paper is part of the International Journal of Geo- ones. Terzaghi (1943) reported that the bearing capacity of a shal-
mechanics, © ASCE, ISSN 1532-3641. low foundation in sand that is embedded below the groundwater

© ASCE 04019181-1 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019181


table equals approximately one-half of that in dry sand. Moreover, eventually suggested that a modified total stress approach can
several researchers have stated that the bearing capacity of soils be effectively used if the unconfined compression test is conducted
increases due to the contribution of matric suction ψ to shear on specimens with average suction within the influence depth of
strength (e.g., Broms 1963; Steensen-Bach et al. 1987; Fredlund the foundation. Based on several model-footing tests conducted on
and Rahardj 1993; Schnaid et al. 1995; Miller and Muraleetharan sand with varying depths of water table, Shahriar et al. (2015) in-
1998; Oloo et al. 1997; Costa et al. 2003; Xu 2004). However, the troduced a new correction factor to the method of strain influence
role of matric suction in the bearing capacity of unsaturated soils factor to account for the extra settlement induced by groundwater
has not been fully addressed yet and is therefore neglected. Regard- rise. The proposed correction factor, however, was dependent on
less, neglecting the contribution of matric suction in the bearing soil type, and the amount of capillary force for the tested sand
capacity of unsaturated soils would be similar to not considering was measured to be insignificant.
the effect of reinforcement in the design of reinforced concrete In spite of several experimental field and laboratory investiga-
Downloaded from ascelibrary.org by Indian Institute Of Technology, Indore on 01/16/24. Copyright ASCE. For personal use only; all rights reserved.

(Steensen-Bach et al. 1987). tions on the bearing capacity of unsaturated soils, less attention has
More than five decades ago, Meyerhof (1956) stated that there been given to theoretical, analytical, and numerical modeling of
was a lack of information regarding the influence of matric suction this phenomenon to be used effectively in practice. Recently, some
on the bearing capacity of unsaturated coarse-grained soils; how- analytical models have been proposed to predict the bearing capac-
ever, as time passed, several field and laboratory investigations ity of sandy and silty soils using the saturated strength parameters
were carried out to address this issue. For example, in-situ plate and soil-water retention curve (e.g., Oloo et al. 1997; Mohamed and
load tests were conducted to investigate the influence of matric suc- Vanapalli 2006; Vanapalli and Mohamed 2007; Oh and Vanapalli
tion on the bearing capacity of coarse-grained and fine-grained 2008; Vahedifard and Robinson 2015). An analytical solution was
soils (e.g., Steensen-Bach et al. 1987; Costa et al. 2003; Xu 2004; derived to consider suction as an independent stress state variable in
Oloo 1994; Larson 1997; Conciani et al. 1998; Consoli et al. 1998; calculation of Rankin’s earth pressure (Zhang et al. 2016). Based on
Rojas and Salinas 2002; Rojas et al. 2007; Mohamed and Vanapalli a parametric study, it was concluded that the factor of safety against
2012, 2015). These experiments have shown the notable contribu- overturning is too conservative if the wall is designed based on the
tion of matric suction ψ to the ultimate bearing capacity of unsatu- saturated soil assumption, though it may not be the case for most
rated soils qu . The effects of plate dimensions, variation in soil retaining walls constructed. Despite the fact that the proposed
stiffness, and loading rate, as well as the definition of ultimate model is quite straightforward, no comparisons were made with
load are mainly discussed in these studies. Mohamed and Vanapalli field or laboratory data to validate it. The influences of hydraulic
(2015) carried out in-situ plate load tests (plate dimensions of hysteresis and drainage conditions on unsaturated soil bearing
0.2 × 0.2 m) on an unsaturated poorly graded sand (SP). It was capacity were numerically investigated through an effective stress-
concluded that the bearing capacity of unsaturated sand at a matric based finite-element formulation (Tang et al. 2017). The notable
suction of 8 kPa is approximately 3.5 times greater in comparison role of both suction and hydraulic hysteresis on bearing capacity
to its bearing capacity in saturated (i.e., zero suction) conditions. was highlighted. More importantly, it was concluded that consid-
In addition to the in-situ plate load experiments, the effect of ering a constant contribution of suction to the shear strength com-
matric suction on the bearing capacity of unsaturated soil has been pared with constant water and suction conditions results in a better
studied through conducting model-footing tests (e.g., Mohamed approximation of bearing capacity. However, the outcome may not
and Vanapalli 2006; Vanapalli and Mohamed 2007; Vanapalli be conclusive for different types of soil possessing different water
et al. 2007, 2010; Lins et al. 2009; Vanapalli and Oh 2010; retention capabilities.
Mohamed et al. 2011; Vanapalli and Mohamed 2013). In these Complementary to the previous works, this study aims to
studies, square or circular model footings simulating the in-situ plate present a new closed-form solution for predicting the bearing
load tests were loaded in unsaturated coarse-grained or fine-grained capacity of unsaturated coarse-grained and fine-grained soils for
soils with different matric suctions to evaluate the ultimate bearing the design of both surficial and embedded shallow foundations
capacity. According to the results, a notable contribution of matric based on the effective stress approach for unsaturated soils. More-
suction to the bearing capacity of the studied soils was observed and over, the bearing capacity of soil is determined using numerical
reported. For instance, Mohamed and Vanapalli (2006) conducted analyses implementing the finite-difference method (FDM). The
laboratory tests to understand the bearing capacity of surface model proposed formulation and numerical simulations are eventually
footings on unsaturated sandy soil. It was revealed that even a small verified against experimental results of in-situ plate load tests
increase of matric suction to 2 and 6 kPa could enhance the bearing and model footings from the literature.
capacities to five and seven times greater, respectively, in comparison
to the corresponding value in the saturated state. Results also indi-
cated that there is a nonlinear increase in the bearing capacity of Theoretical Considerations
unsaturated soils with respect to matric suction. In effect, the
follow-up study of Mohamed et al. (2011) showed that there is a Bearing Capacity Formulations for Saturated Soils
linear increase in the bearing capacity of unsaturated sandy soils
up to the air entry suction ψa followed by a nonlinear trend in Bearing capacity equations have traditionally been developed
for saturated soils based on saturated shear strength parameters.
the transition zone, that is, the range of suction from the air entry
Terzaghi’s widely used equation assumes a rigid strip foundation
value to the residual suction. Finally, the bearing capacity decreases
under vertical load sitting on a homogeneous soil and eventually
with respect to matric suction in the residual zone.
reaches general failure along a predetermined failure surface
In addition, Oh and Vanapalli (2013) pointed out that using the
modified effective stress approach in the interpretation of bearing qu ¼ c 0 N c þ qN q þ 0.5BγN γ ð1Þ
capacity of unsaturated fine-grained soils may not be the best
choice because the drained conditions for both the pore air and where qu = ultimate bearing capacity; c 0 = effective soil cohesion;
water phases are not valid for this category of soils. Therefore, they q = surcharge; B = width of foundation; γ = bulk unit weight of the
interpreted the results of model-footing tests on unsaturated fine- soil; and N c , N q , and N γ are the bearing capacity factors as func-
grained soils following a modified total stress approach. It was tions of the soil internal friction angle ϕ 0 . Several limitations of his

© ASCE 04019181-2 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019181


pioneering work, including footing shape, buried depth, load incli- factors were adopted from different sources because better math
nation, and failure surface, were later overcome (e.g., Meyerhof was observed between the predictions and experimental results.
1951, 1956; Vesic 1973; Hansen 1970; Hanna and Meyerhof 1981). However, each bearing capacity equation was developed based
Meyerhof (1956) proposed the general bearing capacity equation to on specific assumptions varying from one model to another. As
account for all the shortcomings mentioned previously a result, combining various bearing capacity factors in this equation
can cause major concern. Terzaghi’s equation has recently been de-
qu ¼ c 0 N c sc dc ic þ qN q sq dq iq þ 0.5BγN γ sγ dγ iγ ð2Þ veloped as an unsaturated version, based on the concept of effective
stress (Vahedifard and Robinson 2015)
This was achieved by multiplying several shape (sc , sq , and sγ ),
depth (dc , dq , and dγ ), and inclination factors (ic , iq , and iγ ) by qu ¼ fc 0 þ ψb ð1 − Sw;AVE Þ tan ϕ 0 þ ½ðua − uw ÞSe AVE tan ϕ 0 gN c ξ c
each bearing component. The bearing capacity equation generally
includes three components standing for the contributions of soil þ qN q ξ q þ 0.5BγN γ ξ γ ð5Þ
Downloaded from ascelibrary.org by Indian Institute Of Technology, Indore on 01/16/24. Copyright ASCE. For personal use only; all rights reserved.

cohesion, surcharge, and foundation width to the ultimate capacity


where ψb = air entry suction; Se = effective saturation; and Sw;AVE =
(Das 2015). In addition, the internal friction angle affects all three
average degree of saturation, which is a function of average matric
bearing capacity factors.
suction in the stress bulb from the bottom of the foundation up to a
depth of 1.5B below it. Eq. (5) has been used to predict the results
Transition from Saturated to Unsaturated States of laboratory and field tests conducted by Vanapalli and Mohamed
in Previous Research (2007, 2013). Comparisons revealed that predictions are satisfac-
tory for foundations on the ground surface (i.e., q ¼ 0), though dis-
The classical bearing capacity relationships do not take into ac- crepancies between physical measurements and model predictions
count the unsaturated state of soils except some minor corrections increased for foundations buried under the ground (i.e., q > 0).
of surcharge and soil unit weight based on the location of the water Moreover, the model was introduced to be used for different soil
table (Bowles 1996). In fact, the major role of matric suction in the types, whereas comparisons were simply limited to unsaturated
enhancement of shear strength is totally ignored. To cope with this sand.
limitation, some modifications were applied to the conventional
bearing capacity equations in the last decades. Broms (1963), who
investigated the behavior of flexible pavements on saturated soils, Concept of Effective Stress in Unsaturated Soils
observed that any rise in saturation degree would decrease the bear-
There has been an extensive debate on the concept of effective
ing capacity of soil. A similar conclusion was drawn for the bearing
stress in unsaturated soils for decades, and a consensus has not
capacity of shallow foundations on two types of sand at saturated as
yet been reached. Research in this regard has mainly been con-
well as unsaturated states (Steensen-Bach et al. 1987). Capillary
ducted following either the two independent stress state variable
forces were reported to be responsible for a higher capacity of
approach (Fredlund et al. 1978) or the single effective stress ap-
unsaturated soil compared with the saturated one. Oloo et al. (1997)
proach, also known as Bishop’s effective stress (Bishop 1959).
considered the influence of matric suction on the ultimate bearing
The second approach states the effective stress in terms of the
capacity in a systematic manner by modifying Terzaghi’s equation.
net stress (σ − ua ), matric suction (ua − uw ), and effective stress
The concept of two stress state variables introduced by Fredlund
parameter χ
and Morgenstern (1977) was used to modify the cohesion term
in the first component of the equation as follows: σ 0 ¼ ðσ − ua Þ þ χðua − uw Þ ð6Þ
0
qu ¼ ½c þ ðua − uw Þ tan ϕ N c þ qN q þ 0.5BγN γ
b
ð3Þ Because of the simplicity of this approach as well as the fact that
the effective stress can easily be used for prediction of shear
where (ua −uw ) = matric suction; ψ = difference between pore air
strength, Bishop’s effective stress has received much attention in
and pore water pressure; and ϕb = parameter describing the rate of
practical geoengineering (e.g., Khalili and Khabbaz 1998; Lu
increase in shear strength with matric suction under constant net
and Likos 2004; Garakani 2013; Haeri et al. 2014, 2016, 2019;
stress. The major limitation of Eq. (3) is the monotonic increase
Garakani et al. 2015, 2018, 2019). Shear strength can indeed be
of shear strength with matric suction because of the planar failure
expressed in terms of saturated shear strength parameters of c 0
surface of extended Mohr-Coulomb (Fredlund et al. 1978). Indeed,
and ϕ 0 in addition to the effective stress parameter χ, accordingly
ϕb is not constant and varies with matric suction (Vanapalli et al.
1996). According to Oh and Vanapalli (2008), the assumption of τ ¼ c 0 þ χðua − uw Þ tan ϕ 0 þ ðσ − ua Þ tan ϕ 0 ð7Þ
linear increase in shear strength with matric suction can result in
misleading predictions, especially for suction values beyond the However, there has been a long-running debate on the nature
air entry value. Therefore, Vanapalli and Mohamed (2013) pro- and determination of χ. This parameter is a function of soil satu-
posed a new equation for estimating the bearing capacity of foun- ration and reflects the contribution of matric suction to effective
dations on the ground surface (i.e., q ¼ 0) based on the modified stress. Although χ is a simplified parameter, it is an average weigh-
shear strength models of Vanapalli et al. (1996) and Fredlund et al. ing factor reflecting the attractive interparticle capillary as well as
(1996) physicochemical forces. Khalili and Khabbaz (1998) introduced χ
as a function of suction ratio
qu ¼ fc 0 þ ψb ½tan ϕ 0 − SF tan ϕ 0  þ ðua − uw ÞAVE SF tan ϕgN c ξ c
8 −0.55
þ 0.5BγN γ ξ γ ð4Þ < ua − uw ua − uw > ψ b
χ¼ ψb ð8Þ
:
where ψb = air entry suction; ðua − uw ÞAVE = average matric suc- 1 ua − uw ≤ ψ b
tion below foundation; S = degree of saturation; F = fitting param-
eter; and N c and N γ = bearing capacity factors (Terzaghi 1943; where ψb = air entry suction along the drying path and the air ex-
Kumbhojkar 1993). In addition, ζ c and ζ γ are the shape factors that pulsion value during the wetting process. An alternative definition
were chosen from Vesic (1973). As mentioned, the bearing capacity of χ by Lu and Likos (2004) was put forward as

© ASCE 04019181-3 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019181


S − Sr 1.0
χ ¼ Se ¼ ð9Þ
1 − Sr

Degree of saturation, Sr
0.8
where Se = effective saturation; and Sr = residual degree of satu-
0.6
ration. The value of Se can be expressed as a function of matric
suction (van Genuchten 1980) as:
0.4
 m
1
Se ¼ ð10Þ 0.2
1 þ ½αðua − uw Þn After Rojas et al. (2007)
After Vanapalli & Mohamed (2007, 2013)
0.0
where α, n, and m = fitting parameters. The value of α is almost 1 10 100 1000 10000
equal to the inverse of air entry suction, n depends on the pore size
Downloaded from ascelibrary.org by Indian Institute Of Technology, Indore on 01/16/24. Copyright ASCE. For personal use only; all rights reserved.

Matric suction, (kPa)


distribution in the soil, and m is a parameter that controls the gen-
eral symmetry of the soil water retention curve (SWRC). In sum- Fig. 1. SWRC of the tested soil materials.
mary, the shear strength of unsaturated soils can be expressed by
using Bishop’s effective stress approach [Eq. (7)]. The effective
stress parameter χ is also related to the retention characteristic
of different soil types through the suction-dependent effective de- surface in a model box to failure (Test #1). Four tests under matric
gree of saturation [Eq. (10)]. The relationships derived previously suctions of 0, 2, 4, and 6 kPa were conducted on a poorly graded
are the basis for development of a new analytical model for the sand (SP) with a bulk unit weight of 16 kN=m3 , and the water level
bearing capacity of unsaturated soils. was adjusted to change the unsaturated state of the soil and matric
suction. Four tensiometers were installed along the depth to mea-
sure the soil suction. In this study, the average suction values within
Development of a New Analytical Solution the stress bulb below the model footing were considered the cor-
The general bearing capacity formulation of Vesic (1973), developed responding matric suction that is shown in the form of a dashed line
for saturated soil, is selected in the current study for further develop- in Fig. 1. In addition, the saturated shear strength parameters of the
ment. The formulation is updated and modified to account for extra test material were determined from direct shear tests as c 0 ¼
suction stress arise from capillary forces [Eq. (11)]. Matric suction is 0.6 kPa and ϕ 0 ¼ 35.5°. As presented by Eq. (10), the SWRC
included in the basic formulation through an effective stress ap- has three fitting parameters, namely α, m, and n. For the data pre-
proach in which suction contribution is added to net stress through sented in Fig. 1, the fitting parameters α, m, and n have been cal-
a weighing factor χ. This weighing factor is directly correlated to culated as α ¼ 0.15, m ¼ 2.1, n ¼ 4.9, and α ¼ 0.08, m ¼ 0.12,
the water retention characteristic of any specific soil with random n ¼ 1.5 for the data reported by Vanapalli and Mohamed (2007,
distribution of pores through the effective degree of saturation, Se 2013) and Rojas et al. (2007), respectively.
Vanapalli and Mohamed (2013) conducted several other
qu ¼ ½c 0 þ χðua − uw Þ tan ϕ 0 Cf N c ξ c þ qN q ξ q þ 0.5BγN γ ξ γ model-footing tests on both the surface of a soil (Test #2) and
ð11Þ at the depth of 0.15 m below the soil surface (Test #3) to evaluate
the bearing capacity of an unsaturated sand under four different
where Cf = correction factor for tuning the cohesion-dependent matric suctions of 0, 2, 4, and 6 kPa. Tests were carried out using
component in the bearing capacity equation. The unsaturated state a 0.15-m-wide square footing inside a chamber box with dimen-
is also considered in the other components by modifying the sur- sions of 0.9 × 0.9 × 0.75 m3 . Soil water characteristics and
charge and soil unit weight, as suggested by Bowles (1996) and physical-mechanical parameters of the tested soil were identical
Vahedifard and Robinson (2015). The correction factor is introduced to those reported by Vanapalli and Mohamed (2007).
as a nonlinear function of matric suction and can be determined by The last set of laboratory data was obtained from Rojas et al.
back analysis of experimental results, accordingly (2007), who conducted a series of in-situ plate load tests regarding
ASTM D1194 (ASTM 1994) to study the bearing capacity of an
aχðua − uw Þ þ b
Cf ¼ ð12Þ unsaturated clayey soil (Test #4) under six matric suctions of 0, 10,
χðua − uw Þ þ b 48, 56, 60, and 63 kPa. For this purpose, test pits 1.2 m in width and
3.0 m in length were excavated to the depth of 1.4 m. Test pits were
where a and b = fitting parameters depending on the soil type, instrumented with four tensiometers at 0.1, 0.3, 0.6, and 0.9 m
geometrical aspects, and embedment depth of the foundation and down to the bottom to capture the induced matric suction (as shown
loading conditions. At two extreme states of the soil saturation, in Fig. 1) underneath the loading plate. Two pits were filled with
namely fully saturated or completely dry conditions, Cf becomes water to a specific level until all the tensiometers read zero suction
unity because χðua − uw Þ ¼ 0. Accordingly, for the two states of for the simulating bearing capacity test under the fully saturated
dry or fully saturated soil, Eq. (12) turns into the conventional form state. The rest of the pits were watered until the target matric suc-
of Vesic’s formula. The procedure for determining fitting parameters
tion was attained, followed by the loading test. A standard steel
a and b is explained in the next sections.
disk [in accordance with ASTM D1195 (ASTM 1993)] 31 mm
in diameter was used, and the ultimate bearing capacity was
Field and Laboratory Data Set assumed once settlement became 3.1 mm or 10% of diameter.
In addition, another criterion (namely the yield stress) was consid-
In general, experimental results on the bearing capacity of unsatu- ered for estimating the ultimate bearing capacity. Based on the cone
rated soils are divided into two main categories, model-footing tests penetration test (CPTu) test results, soil parameters were evaluated
and in-situ plate load tests. Regarding the first category, Vanapalli as c 0 ¼ 3 kPa, ϕ 0 ¼ 26.0°, and γ d ¼ 16 kN=m3 . Physical model-
and Mohamed (2007) evaluated the bearing capacity of an unsatu- ing and in-situ tests properties are summarized in Table 1 for four
rated sand by loading a square plate with dimension of 0.1 m on soil sets of tests (i.e., Test #1 to Test #4).

© ASCE 04019181-4 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019181


Table 1. Physical modeling and in-situ test properties
Property Test #1a Test #2b Test #3b Test #4c
Shape and size of the loading area Square 10 × 10 cm2 Square 15 × 15 cm2 Square 15 × 15 cm2 Circle D ¼ 31 cm
Depth of testing (m) 0 0 0.15 0d
Soil type Sand Sand Sand Clay
Soil density, γ (kN=m3 ) 16 16 16 16
Soil suctions below the loading area, ψ (kN=m2 ) 0, 2, 4, 6 0, 2, 4, 6 0, 2, 6 0, 10, 48, 56, 60, 63
Soil cohesion, c 0 (kPa) 0.6 0.6 0.6 3.0
Soil friction angle, ϕ 0 (degrees) 35.3 35.3 35.3 26.0
a
Data from Vanapalli and Mohamed (2007).
b
Data from Vanapalli and Mohamed (2013).
Downloaded from ascelibrary.org by Indian Institute Of Technology, Indore on 01/16/24. Copyright ASCE. For personal use only; all rights reserved.

c
Data from Rojas et al. (2007).
d
Test was carried out at the depth of 1.40 m below the ground surface. However, the edge of loading area was far enough from the test pit perimeter, and it was
modeled as a surface test in numerical simulation.

Numerical Modeling of the constructed domain, as well as the soil properties and
water table level.
The finite-difference method has been implemented in this study 5. An initial rate of loading (in terms of the displacement rate) was
for numerical investigation of the bearing capacity of unsaturated selected.
soils. For this purpose, FLAC3D software (version 5.00; 2012) 6. The foundation area on the constructed domain was loaded until
from Itasca Consulting Group Inc. was implemented as the numeri- the soil reached its yielding state. Then, the load-displacement
cal modeling toolbox. curve was obtained.
FLAC3D is a three-dimensional explicit finite-difference pro- 7. On the basis of the obtained results from step 6, sensitivity and
gram for computational mechanics. It can simulate the behavior convergence analysis on the constructed model was carried out
of three-dimensional structures built of soil, rock, or other materials as follows:
that undergo plastic flow when their yield limits are reached. • The mesh size was changed (especially under the loading
Materials in FLAC3D are represented by polyhedral elements area) to ensure reaching the highest possible accuracy within
within a three-dimensional grid that is adjusted to fit the shape a rational time for running the model.
of the object to be modeled. Each element behaves according to • The size of the domain was changed to ensure that boundaries
a prescribed linear or nonlinear stress-strain law in response to ap- were far enough from the interactions of the loading area.
plied forces or boundary restraints. The material can yield and flow, • The rate of loading was changed so that convergence could
and the grid can deform (in large-strain mode) and move with the be achieved in a suitable running time.
material that is represented. The explicit Lagrangian calculation • During all analyses, the obtained profiles for the matric suc-
scheme and the mixed-discretization zoning technique used in tion, degree of saturation, and effective stress in the depth of
FLAC3D ensure that plastic collapse and flow are modeled accu- the soil were checked and controlled to be consistent with the
rately (FLAC3D 2012). initial calculations.
The main advantage of implementing FLAC3D in modeling 8. The optimized conditions for the mesh size, domain size, and
unsaturated soil behavior is its capability of defining any state rate of loading were selected regarding data from step 7.
variables such as stress and suction tensors by the user. Accord- 9. The optimized model was simulated to obtain the load-
ingly, the effective stress tensor for an unsaturated soil element displacement curves.
can be defined in terms of the mechanical and hydraulic variables To build up the numerical domain for Tests #1 to #3, “Radial
(i.e., net stress, matric suction, and degree of saturation), and Brick elements” were used and cubic domains were constructed.
the coupled hydromechanical behavior of the unsaturated soils For Test #4, a cylindrical domain was constructed because a plate
can be simulated by considering the proper constitutive function. disk was used in the experiments for loading. In addition, a quarter
On this basis, many researchers have implemented FLAC3D of the geometry from the center of loading only was modeled
to simulate different engineering problems dealing with soil behav- because the axisymmetric conditions are valid for all four sets of
ior under unsaturated conditions (e.g., Wu and Huang 2006; experiments. In order to avoid boundary effects on the results, nu-
Rajeev and Kodikara 2011; Rutqvist et al. 2011; Li et al. 2012; merical model boundaries were considered far enough from the
Ni et al. 2018). center of loading. In this regard, a distance in excess of 3–5 times
In this study, the following steps were conducted during the the disk diameter or dimension was taken into account from the
numerical simulations: corner of the foundation to the boundaries for different model foun-
1. The matric suction profile in the depth of the soil was calculated dations. Actually, model dimensions were selected such that the
on the basis of the water table level and the data provided in maximum possible accuracy was reached with the minimum pos-
experimental cases (Tests #1 to #4). sible required time to run the models. Fig. 2 demonstrates the
2. The profile of the degree of saturation in the depth of the soil geometrical aspects of the constructed domains, along with three-
was calculated following the corresponding SWRC. dimensional perspective views for each numerical model used.
3. The primary geometry of the problem was constructed in The soil parameters used in numerical modeling were the same
FLAC3D by taking into account the initial considerations for as those reported by Vanapalli and Mohamed (2007, 2013) and
the size of the elements and domain and applying the proper Rojas et al. (2007) for laboratory Tests #1 to #4 (as presented in
fixities at the boundaries of the domain. Table 1). In numerical modeling, some parameters, like internal
4. By preparing the suitable FISH code in FLAC3D, the tensors for friction angle, cohesion, and unit weight, have been chosen as di-
the matric suction, degree of saturation, effective stress, and rectly reported in the aforementioned references, whereas some
suction-dependent soil cohesion were introduced to all elements others (e.g., dilation angle and modulus of elasticity) have been

© ASCE 04019181-5 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019181


Downloaded from ascelibrary.org by Indian Institute Of Technology, Indore on 01/16/24. Copyright ASCE. For personal use only; all rights reserved.

Fig. 2. Geometrical aspects of the 3D numerical models: (a) Test #1; (b) Test #2; (c) Test #3; and (d) Test #4.

estimated. For instance, dilation angle was considered to be 0.1ϕ 0 , element based on its current suction and degree of saturation.
and the modulus of elasticity was determined from load-settlement The following equation by Lu and Likos (2004) was used to ex-
curves reported for the conducted tests. In addition, reported press the suction-dependent cohesion cunsat for unsaturated cases:
SWRC data (Fig. 1) in conjunction with a hydrostatic distribution
of matric suction with depth have been considered to define the cunsat ¼ c 0 þ χðua − uw Þ tan ϕ 0 ð13Þ
matric suction and soil saturation relationship in all numerical sim-
According to Eq. (13), cohesion takes constant values for
ulations. Although a nonlinear suction distribution is closer to real-
the soil under the ground water table (i.e., zero matric suction)
ity, assuming a linear trend is beneficial in the following ways:
and, for conditions above the water table, increases to a peak, fol-
1. Direct or indirect measurements of suction in soil are expensive.
lowed by a reduction with a further increase of suction. Fig. 4
2. Under zero water infiltrations or evaporations and the existence
shows sample unsaturated cohesion contours for the numerical
of a steady water table condition, the assumption of linear suc-
models at the matric suction of 2 and 15 kPa for Tests #1 to #3
tion change along the soil profile is accurate enough, specifi-
and #4, respectively. Table 2 summarizes parameters considered
cally in sandy soils (Lu and Likos 2004).
in numerical simulations.
3. According to the concept of cohesion in the effective stress the-
ory for unsaturated soils, suction increase has a limiting effect
on the increase of cohesion. In other words, cohesion increases Interpretation of the Results
with suction to a peak and falls afterward. Therefore, a nonlinear
assumption for suction distribution results in overestimation of Results are explained for the four tests described in the previous
soil cohesion, especially near the ground surface. sections. Comparisons are made between physical measurements
4. For conditions where the ground water table is located very and predictions of both the analytical model and numerical simu-
deep, nonlinear distribution of suction makes the modulus of lations, accordingly.
elasticity unrealistically large.
Fig. 3 shows sample pore pressure contours for the numerical
models at the matric suction of 2 and 15 kPa for Tests #1 to #3 and Test #1 [Performed by Vanapalli and Mohamed (2007)]
#4, respectively. In Fig. 3, negative pore pressure values stand for Analytical and numerical results of Test #1 reported by Vanapalli
the associated matric suction within the soil regions. and Mohamed (2007) are discussed in this part. Contours of effec-
In the current numerical study, the effective stress tensor in each tive vertical stress, maximum shear strain rate, and total displace-
soil element was defined by considering Eq. (6). It was also as- ments are plotted in Figs. 5(a), 6(a), and 7(a), respectively, for
sumed that the pore air phase was continuous and connected to matric suction of 2 kPa at the ultimate shear failure state. Regarding
the atmosphere, so the atmospheric pressure was assumed for ua. data presented in these figures, generation of the shear failure
In addition, the cohesion was numerically modeled for each soil wedge below the loading area is obvious, implying the soil has

© ASCE 04019181-6 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019181


Downloaded from ascelibrary.org by Indian Institute Of Technology, Indore on 01/16/24. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Sample contours of pore water pressure (positive values, N=m2 ) and matric suction (negative values, N=m2 ) in soil profile for (a) Test #1;
(b) Test #2; (c) Test #3; and (d) Test #4.

reached its ultimate bearing capacity. Numerical simulation for the analytical correction factor Cf for different matric suctions.
variations in loading with settlement is compared with experimen- Accordingly, a back calculation on the in-situ recorded bearing
tal records in Fig. 8(a) for a matric suction of 2 kPa. This suction capacity values in conjunction with Eq. (11) was performed to ob-
value was modeled by lowering the water table from the initial fully tain the exact values of Cf , and results were compared to those
saturated state so that the target suction was obtained [Fig. 3(a)]. As obtained from Eq. (12) for different matric suctions [Fig. 9(a)].
shown in Fig. 8(a), a monotonic rise can be observed for the bearing Fig. 9(a) shows a very good conformance between the exact values
capacity as settlement increases until the ultimate bearing capacity of Cf (obtained from field test data) and those evaluated from the
is reached at a pick point and then descends. Fig. 8(a) shows good analytical solution for Test #1. The corresponding ultimate bearing
conformance between the numerical and experimental data. The capacities are also predicted implementing Eq. (12) and consider-
ultimate bearing capacity, in fact, corresponds to the maximum co- ing the predicted Cf values. The best estimated fitting parameters
hesion. In addition to numerical modeling, Eqs. (11) and (12) were of the correction factor function (i.e., a and b) were calculated, as
implemented along with the experimental data to calculate the shown in Table 3. The correction factor Cf is physically dependent

© ASCE 04019181-7 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019181


Downloaded from ascelibrary.org by Indian Institute Of Technology, Indore on 01/16/24. Copyright ASCE. For personal use only; all rights reserved.

Fig. 4. Sample contours of cohesion (N=m2 ) in soil profile for (a) Test #1; (b) Test #2; (c) Test #3; and (d) Test #4.

Table 2. Numerical modeling parameters


Property Test #1 Test #2 Test #3 Test #4
Domain size (cm3 ) 15 × 15 × 30 25 × 25 × 50 25 × 25 × 50 D ¼ 77.5 and H ¼ 77.5
Soil density, γ (kN=m3 ) 16 16 16 16
Saturated soil cohesion, c 0 (kPa) 0.6 0.6 0.6 3.0
Soil friction angle, ϕ 0 (degrees) 35.3 35.3 35.3 26.0
Soil dilation angle, Ψd (degrees) 3.53 3.53 3.53 0
Soil suction below the loading area, ψ (kPa) 0, 2, 4, 6 0, 2, 4, 6 0, 2, 4, 6 0, 10, 15, 48, 63

on the soil type, geometrical aspects, and embedment depth of the Similar numerical analyses to that explained for matric suction
foundation and loading conditions. Accordingly, the corresponding of 2 kPa have been run for other matric suctions, and the variations
fitting parameters (a and b) can be different for different aforemen- in the ultimate bearing capacity with matric suction are shown in
tioned conditions. Fig. 10(a). In Fig. 10(a), experimental test data along with the

© ASCE 04019181-8 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019181


Downloaded from ascelibrary.org by Indian Institute Of Technology, Indore on 01/16/24. Copyright ASCE. For personal use only; all rights reserved.

Fig. 5. Sample contours of effective vertical stress (N=m2 ) at ultimate state in soil profile obtained from numerical simulations for (a) Test #1;
(b) Test #2; (c) Test #3; and (d) Test #4.

results of analytical modeling are also plotted for comparison. Test #2 [Performed by Vanapalli and Mohamed (2013)]
According to the results, the maximum bearing capacity is achieved
The second validation tests were the physical model tests carried
at matric suction between 4 and 6 kPa. Afterward, a further increase out by Vanapalli and Mohamed (2013) on a model footing on
in suction results in a reduction in bearing capacity as water me- soil surface (Test #2). The same soil as the one used in Test #1
niscus vanishes and capillary forces reduce. There is a qualitative was used in their follow-up study. Contours of effective vertical
good agreement between numerical simulations, analytical model- stress, maximum shear strain rate, and total displacements are plot-
ing, and physical measurements. However, there is a maximum of ted in Figs. 5(b), 6(b), and 7(b), respectively, for a matric suction of
19% difference between the numerical and experimental results at 2 kPa at the ultimate shear failure state, which shows the soil has
4 kPa matric suction. The measured bearing capacity at this suction reached its ultimate bearing capacity. The variations in loading with
was 715 kPa, whereas the computed value of numerical analysis settlement were compared with experimental measurements in
was 850 kPa. Nevertheless, analytical and numerical predictions Fig. 8(b) for a matric suction of 2 kPa, which follows a monotonic
were better at other suctions within the studied range of suction. rise as settlement increases until the steady state corresponding to

© ASCE 04019181-9 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019181


Downloaded from ascelibrary.org by Indian Institute Of Technology, Indore on 01/16/24. Copyright ASCE. For personal use only; all rights reserved.

Fig. 6. Sample contours of the maximum shear strain rate at ultimate state in soil profile obtained from numerical simulations for (a) Test #1;
(b) Test #2; (c) Test #3; and (d) Test #4.

the ultimate bearing capacity is reached. Exact and predicted values Comparisons reveal that in this case, the maximum difference is
of the analytical correction factor Cf were calculated as explained between measurements and analytical results at a matric suction
before and compared for different matric suctions in Fig. 9(b). of 0 kPa (fully saturated condition). The measured ultimate bearing
Fig. 9(b) shows very good conformance between the exact values capacity of 180 kPa was 70 and 47 kPa higher than the numerical
of Cf (obtained from field test data) and those evaluated from the and analytical predictions, respectively. However, the differences
analytical solution, for Test #2. The corresponding ultimate bearing significantly decreased at the other three suction levels of 2, 4,
capacities were also predicted implementing Eq. (12) and consid- and 6 kPa. More importantly, it can be inferred from the analytical
ering the predicted Cf values. The best estimated fitting parameters results that bearing capacity descends with suction after 6 kPa. In
of the correction factor function (i.e., a and b) were calculated, as other words, the influence of matric suction on shear strength and
shown in Table 3. hence bearing capacity decreases as suction goes beyond 6 kPa,
Field measurements and analytical and numerical values of the which is possibly within the residual domain of the water retention
ultimate bearing capacity for Test #2 are plotted in Fig. 10(b). curve for the test material. At residual water contents, the degree

© ASCE 04019181-10 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019181


Downloaded from ascelibrary.org by Indian Institute Of Technology, Indore on 01/16/24. Copyright ASCE. For personal use only; all rights reserved.

Fig. 7. Sample contours of total displacements (m) at ultimate state in soil profile obtained from numerical simulations for (a) Test #1; (b) Test #2;
(c) Test #3; and (d) Test #4.

of saturation of the soil is low, leading to significant reduction surcharge should be considered in numerical modeling. To achieve
in the effective stress parameter χ. Thus, despite the fact that that, the complete soil block was first modeled and reached equilib-
the matric suction is significant in this region, the suction stress rium under body weight. Thereafter, soil elements up to 0.15 m
ðPs ¼ χ½ua − uw Þ can be small and consequently reduce the effec- depth were removed, according to Fig. 2, to reach the level of foun-
tive stress, shear strength, and soil bearing capacity. dation base and loading. The final soil model was brought to equi-
librium again before proceeding to the loading stage. Similar to
previous tests, numerical results for bearing capacity under suction
Test #3 [Performed by Vanapalli and Mohamed (2013)] of 2 kPa are presented in Figs. 5(c), 6(c), 7(c), and 8(c) for the sake
In the third series of tests, also reported by Vanapalli and Mohamed of completeness. Fig. 8(c) compares the load-settlement curves
(2013), model-footing loading was conducted at 0.15 m depth. obtained from experimental and numerical data sets. As illustrated
Because the distance between footing corner and soil face was in Fig. 8(c), compared to the previous model test without the
less than four times that of the plate diameter, the influence of soil surcharge effect (i.e., D=B = zero in Test #2), a significant increase

© ASCE 04019181-11 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019181


1500 1500
Laboratory tests by Vanapalli and Mohamed (2007) Field tests by Vanapalli and Mohamed (2013)
(Surface tests)
1200 1200
Matric suction just Matric suction just
at the level of the at the level of the

Load (kPa)

Load (kPa)
900 loading area 2kPa 900 loading area 2kPa

600 600

300 Lab test data 300 Field test data


Numerical data Numerical data
0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
(a) Displacement (mm) (b) Displacement (mm)
Downloaded from ascelibrary.org by Indian Institute Of Technology, Indore on 01/16/24. Copyright ASCE. For personal use only; all rights reserved.

1500 1500
Field tests by Vanapalli and Mohamed (2013) Field tests by Rojas et al. (2007)
(Embedded tests)
1200 1200 Field test data
Matric suction just
at the level of the Numerical data
Load (kPa)

Load (kPa)
900 loading area 2kPa 900 Matric suction just at
the level of the
600 600 loading area 10kPa

300 Field test data 300


Numerical data
0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
(c) Displacement (mm) (d) Displacement (mm)

Fig. 8. Sample load-displacement curves obtained from numerical simulations for (a) Test #1; (b) Test #2; (c) Test #3; and (d) Test #4.

6.0 6.0
Laboratory tests by Vanapalli and Mohamed (2007) Field tests by Vanapalli and Mohamed (2013)
(Surface tests)
Correction Factor, CF
Correction Factor, CF

4.0 4.0

2.0 2.0

Lab test data Field test data


Analytical data Analytical data
0.0 0.0
0 2 4 6 8 10 0 2 4 6 8 10
(a) Matric suction, (kPa) (b) Matric suction, (kPa)
6.0 6.0
Field tests by Vanapalli and Mohamed (2013) Field tests by Rojas et al. (2007)
(Embedded tests)
Correction Factor, C F
Correction Factor, C F

4.0 4.0

2.0 2.0
Field test data Field test data
Analytical data Analytical data
0.0 0.0
0 2 4 6 8 10 0 20 40 60 80 100
(c) Matric suction, (kPa) (d) Matric suction, (kPa)

Fig. 9. Variation of the correction factor Cf versus matric suction for (a) Test #1; (b) Test #2; (c) Test #3; and (d) Test #4.

can be observed in ultimate bearing capacity by burring the foun- shown in Table 2. Comparisons between calculated, computed,
dation at D=B ¼ 1. In addition to the numerical results, the changes and measured data are made in Fig. 10(c). An excellent match be-
of the correction factor Cf against matric suction are depicted for tween calculated and measured data can be found at the three suction
Test #3 in Fig. 9(c), which shows an excellent conformance between levels considered. However, the computed data deviated from the
the real and predicted values. In spite of the previous tests, this series measured at 6 kPa suction, though the agreement is very good at
of tests was only conducted at three different suction states, as 2 kPa suction. The variations in bearing capacity with suction

© ASCE 04019181-12 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019181


Table 3. Analytical solution parameters
Property Test #1 Test #2 Test #3 Test #4
Soil suction below the loading area, ψ (kPa) 0, 2, 4, 6 0, 2, 4, 6 0, 2, 4, 6 0, 10, 15, 48, 63
Correction factor parameter, a 1.72 1.80 5.70 1.40
Correction factor parameter, b 1.05 0.90 1.40 −0.85
Shape and size of the loading area Square 10 × 10 cm2 Square 15 × 15 cm2 Square 15 × 15 cm2 Circle D ¼ 31 cm
Depth of testing (m) 0 0 0.15 0
Soil type Sand Sand Sand Clay

1500 1500
Downloaded from ascelibrary.org by Indian Institute Of Technology, Indore on 01/16/24. Copyright ASCE. For personal use only; all rights reserved.

Laboratory tests by Vanapalli and Mohamed (2007) Field tests by Vanapalli and Mohamed (2013)
Bearing capacity, q u (kPa)

Bearing capacity, qu (kPa)


(Surface tests)
1200 1200

900 900

600 600
Lab test data Field test data
300 Analytical data 300 Analytical data
Numerical data Numerical data
0 0
0 2 4 6 8 10 0 2 4 6 8 10
(a) Matric suction, (kPa) (b) Matric suction, (kPa)

1500 1500
Field tests by Vanapalli and Mohamed (2013) Field tests by Rojas et al. (2007)
Bearing capacity, q u (kPa)

Bearing capacity, qu (kPa)


(Embedded tests)
1200 1200

900 900

600 600
Field test data Field test data
300 Analytical data 300 Analytical data
Numerical data Numerical data
0 0
0 2 4 6 8 10 0 20 40 60 80 100
(c) Matric suction, (kPa) (d) Matric suction, (kPa)

Fig. 10. Comparison between analytical and numerical solutions with test data for (a) Test #1; (b) Test #2; (c) Test #3; and (d) Test #4.

followed a monotonic trend and seemed to become close to its maxi- depicted in Figs. 5(d), 6(d), 7(d), and 8(d). In Fig. 8(d), the load-
mum value at suction of 6 kPa. However, more measurements be- settlement curves for experimental and numerical data sets are com-
yond this suction value are required to justify this postulation. pared at the matric suction of 10 kPa because the field data at ψ ¼
In the three series of verification tests explained so far, similar 15 kPa were not presented in Rojas et al. (2007). The correction
patterns of stress bulbs were developed beneath the foundation factor change with matric suction is also shown in Fig. 9(d) for
base, according to Figs. 5(a–c). In addition, relatively good matches both real and predicted values. As shown in Fig. 9(d), the predicted
were observed between computed and calculated data and mea- and real Cf values show fair agreement. Both analytical and numeri-
sured experimental values for different suctions and test conditions. cal analyses were run for four other suctions, namely 0, 10, 48, and
In terms of failure mode, all of the computed results follow a 63 kPa, and results are superimposed on field data in Fig. 10(d).
slightly strain-hardening mode corresponding to a medium dense Comparing the results with measurements reveals that both predic-
state of relative density in sands. tive methods deliver satisfactory results for all suction levels but
zero. In fact, measurements are significantly higher than predictions
Test #4 [Performed by Rojas et al. (2007)] for the saturated state. More importantly, measured data indicate an
The last series of field tests conducted on an unsaturated clay by increase in bearing capacity with suction, but the amount is not so
Rojas et al. (2007) was chosen for verification of analytical and significant compared with the observed trends for the previous tests
numerical modeling, namely Test #4. Although the field experi- [Fig. 10(a–c)]. In other words, this comparison with other physical
ments were carried out at 1.4 m depth, the influence of soil surcharge measurements casts some doubt on measured bearing capacity at the
should be ignored because the distance between the foundation cor- saturated state because the field data at zero suction show diversity
ner and soil face is large enough. Therefore, the numerical model from those obtained from a conventional Vesic’s formulation
was constructed so that loading was applied on the ground surface. [Eq. (11), χ ¼ 1 and ua − uw ¼ 0]. Nevertheless, predictions were
Fig. 2(d) shows the numerical model as a quarter of a solid soil much improved by shifting from fully saturated to unsaturated
cylinder because the axisymmetric assumption is valid. Because conditions.
loading was applied at the model centerline, the mesh dimensions Data presented in Table 3 show that the a and b parameters are
enlarged as they became farther apart from the center. Results of highly dependent on the soil type, depth of embedment, and shape
the numerical analysis for bearing capacity at 15 kPa suction are or size of the loading area.

© ASCE 04019181-13 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019181


Table 4. Quantitative comparison between experimental and predicted (analytical or numerical) values of the ultimate bearing capacity on Tests #1 to #4
Comparison Parameter Test #1 Test #2 Test #3 Test #4
Experimental and analytical data RMSE (kPa) 69.9 86.5 25.7 176.3
NRMSE (%) 10.3 13.3 2.8 24.0
Experimental and numerical data RMSE (kPa) 80.3 42.8 78.5 168.7
NRMSE (%) 11.8 6.6 8.5 22.9

Table 5. Quantitative comparison between experimental and predicted in interparticle interactions, especially at high suctions. Therefore,
(analytical or numerical) values of the ultimate bearing capacity on the linear assumption for variation of the matric suction along the
Downloaded from ascelibrary.org by Indian Institute Of Technology, Indore on 01/16/24. Copyright ASCE. For personal use only; all rights reserved.

different soil types fine soil profile may not be as accurate as for coarse-grained soils.
Comparison Parameter Sandy soil Clayey soil
Experimental and RMSE (kPa) 68.1 176.3
analytical data NRMSE (%) 6.9 24.0 Limitations of the Current Study
Experimental and RMSE (kPa) 68.5 168.7 Similar to any research, this study is subjected to some limitations;
numerical data NRMSE (%) 7.1 22.9 among them the following seem to be more important:
1. The modification factor in the analytical approach Cf has been
considered in the form of a simple mathematical equation. By
considering more complicated function for Cf with more fitting
For a more precise quantitative comparison of the data presented
parameters, the accuracy of the model in prediction may in-
in Fig. 10, the root-mean-square error (RMSE) and normalized
crease, but the complexity of model will increase as well.
root-mean-square error (NRMSE) of the experimental and pre-
2. In analytical and numerical solutions, implementing a more ad-
dicted ultimate bearing capacity values (obtained from analytical
vanced constitutive relationship for defining the shear strength
or numerical analyses) were calculated as presented by Eqs. (14)
can improve the accuracy of the predictions. For example, con-
and (15) and shown in Table 4
sidering suction-induced dilation (Ng et al. 2019) can give more
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Pk realistic values for peak shear strength.
2
i¼1 ðquexperimental − qupredicted Þ 3. It is needed to consider more experimental data for verification
RMSE ¼ i i
ð14Þ of the proposed analytical and numerical approaches, which is
k
suggested to be performed in future studies.
4. The effects of infiltration and evaporation of the water phase in
RMSE
NRMSE ¼ ð15Þ unsaturated soil conditions and changing the soil temperature
Maxðquexperimental Þ − Minðquexperimental Þ can considerably affect the suction profile beneath the founda-
tion and consequently change the bearing capacity of the soil.
where quexperimental and qupredicted = experimental and predicted values of Considering the aforementioned factors for future studies is
the ultimate bearing capacity, respectively; k = number of exper- highly suggested.
imental data; and i = numeral parameter.
Referring to Fig. 10 and considering the RMSE and NRMSE
values presented in Table 4, it is evident that the ultimate bearing Conclusions
capacity values calculated using the analytical and numerical analy-
ses compare very well with the experimental ones, specifically for In this paper, the bearing capacity of shallow foundations on
sandy soils (the cases of Tests #1, #2, and #3), for which the unsaturated strata was investigated via a new simple analytical
NRMSE values for both analytical and numerical solutions vary solution and a series of three-dimensional numerical simulations.
between 2.8% to a maximum of 13.3%. However, for clayey soil To validate the presented analytical solution and corresponding
types (Test #4), the deviation between the measured and predicted numerical modeling, four sets of experimental results obtained
values is higher (i.e., 24.0% and 22.9% for analytical and numerical from both physical modeling and in-situ plate loading tests on
analyses, respectively). unsaturated soils were implemented. In this research, Vesic’s equa-
In addition, from another point of view, the magnitudes of the tion for calculating the ultimate bearing capacity of the foundation
RMSE and NRMSE parameters were calculated by considering all laid on saturated soil was modified for unsaturated soil conditions
data on sandy soils (i.e., Tests #1, #2, and #3 together) and all data by using a suction-dependent correction factor and implementing
on clayey soil (i.e., Test #4) to make a comparison for the function- the suction-dependent cohesion parameter. It was found that the
ality of the proposed analytical and numerical approaches on differ- introduced simple analytical solution could predict the ultimate
ent soil types (Table 5). In accordance with the information reported bearing capacity of shallow foundations laid on unsaturated soils
in Table 5, one can postulate that for sandy soils, the predicted data with an appropriate level of accuracy, especially for sandy soils. In
obtained by implementing the suggested analytical and numerical addition, studies show that the correction factor is generally depen-
solutions in this study are more accurate than those calculated for dent on the soil type and geometrical aspects of the loading area
clayey soils because the NRMSE values for the analytical and (i.e., foundation).
numerical approaches have been calculated as 6.9% and 7.1% In addition to analytical studies and to provide a prospective
for sandy soil, respectively, and 24.0% and 22.9% for clayey soil, ground for simulating more complex unsaturated bearing capacity
respectively. This can be attributed to the suction regime in fine problems beyond the analytical scope, three-dimensional numerical
soils that is completely different from sandy soils. In fine soils, ma- simulations on the studied experimental data sets were performed
tric suction does not only originate from capillary mechanisms; the using a finite-difference code by applying conceptual modifications
short-range interactions between soil particles play a governing role to the input material parameters. To this end, the stress tensors

© ASCE 04019181-14 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019181


were defined in accordance with the effective stress approach in ua = pore air pressure;
unsaturated soil condition, and the SWRC of the soil was intro- uw = pore water pressure;
duced to the model by considering the simple hydrostatic variation ðua − uw ÞAVE = average matric suction (Vanapalli and
of the pore water pressure in conjunction with the experimental Mohamed 2013);
data. Moreover, the shear strength parameters of the soil were de- α; n, and m = fitting parameters of van Genuchten (1980)’s
fined in terms of the matric suction. Accordingly, variations of the model;
suction-dependent soil parameters and state variables were inves-
γ = bulk unit weight of soil;
tigated. As a result, experimental trends for variations in bearing
γ d = dry unit weight of soil;
capacity with suction and load-settlement behavior were reason-
ably captured. ζ c and ζ γ = shape factors proposed by Vesic (1973);
To make a better investigation on the functionality of the ϕ 0 = effective friction angle;
ϕb = parameter describing rate of increase in shear
Downloaded from ascelibrary.org by Indian Institute Of Technology, Indore on 01/16/24. Copyright ASCE. For personal use only; all rights reserved.

presented solutions, quantitative comparisons were performed be-


tween the experimental and predicted results. Studies showed that strength with matric suction;
deviations in analytical and numerical predications from physical σ = total stress;
measurements were below 13.3% for three sets of experimental σ 0 = effective stress;
data conducted on sandy soils, but it was about 24.0% for clayey ðσ − ua Þ = net stress;
soil, suggesting that more studies are still necessary for this field of τ = shear strength;
research to be conducted on fine soil materials. χ = effective stress parameter;
Ψd = dilation angle;
ψ = matric suction; and
Data Availability Statement ψa = air entry suction.
Some or all data, models, or code generated or used during the
study are available from the corresponding author by request, as:
• Generated codes with FLAC3D. References
• Constructed models with FLAC3D.
ASTM. 1993. Standard test method for repetitive static plate load
tests of soils and flexible pavement components, for use in evaluation
and design of airport and highway pavements. ASTM D1195. West
Acknowledgments Conshohocken, PA: ASTM.
ASTM. 1994. Standard test method for bearing capacity of soil for static
This research has been financially supported by Sharif University load and spread footings. ASTM D1194. West Conshohocken, PA:
of Technology, which is greatly acknowledged. In addition, the first ASTM.
author would like to appreciate the Niroo Research Institute for Bishop, A. W. 1959. “The principle of effective stress.” Teknisk Ukeblad
providing him with the opportunity to contribute in this research. 106 (39): 859–863.
Bolton, M. D., and C. K. Lau. 1993. “Vertical bearing capacity factors for
circular footings on Mohr-Coulomb soil.” Can. Geotech. J. 30 (6):
Notation 1024–1033. https://doi.org/10.1139/t93-099.
Bowles, J. E. 1996. Foundation analysis and design. 5th ed. New York:
The following symbols are used in this paper: McGraw-Hill Companies.
a and b = fitting parameters of correction factor Cf ; Broms, B. B. 1963. “The effect of degree of saturation on the bearing
B = width of foundation; capacity of flexible pavements.” Highway Res. Rec. 71 (1): 1–14.
Conciani, W., M. M. Soares, J. M. Naime, and S. Crestana. 1998.
Cf = correction factor;
“Plate load test with CT.” In Proc., 2nd Int. Conf. on Unsaturated Soils,
c 0 = effective cohesion; 333–337. Beijing: International Academic Publishers.
cunsat = suction-dependent cohesion; Consoli, N. C., F. Schnaid, and J. Milititsky. 1998. “Interpretation of plate
D = depth of foundation; load tests on residual soil site.” J. Geotech. Geoenviron. Eng. 124 (9):
dc , dq , and dγ = depth factors; 857–867. https://doi.org/10.1061/(ASCE)1090-0241(1998)124:9(857).
F = fitting parameter proposed by Vanapalli and Costa, Y. D., J. C. Cintra, and J. C. Zornberg. 2003. “Influence of matric
suction on the results of plate load tests performed on a lateritic soil
Mohamed (2013);
deposit.” Geotech. Test. J. 26 (2): 219–227.
i = numeral; Das, B. M. 2015. Principles of foundation engineering. 8th ed. Boston:
ic , iq , and iγ = inclination factors; CENGAGE Learning.
k = number of experimental data in each test set; FLAC3D. 2012. User’s guide. Minneapolis, MN: Itasca Consulting Group.
N c , N q , and N γ = bearing capacity factors in terms of ϕ 0 ; Fredlund, D. G., and N. R. Morgenstern. 1977. “Stress state variables for
Ps = suction stress; unsaturated soils.” J. Geotech. Eng. Div. 103 (GT5): 447–466.
Fredlund, D. G., N. R. Morgenstern, and R. A. Widger. 1978. “Shear
q = surcharge load;
strength of unsaturated soils.” Can. Geotech. J. 15 (3): 313–321.
qu = ultimate bearing capacity; https://doi.org/10.1139/t78-029.
quexperimental = experimentally measured qu ; Fredlund, D. G., and H. Rahardj. 1993. Soil mechanics for unsaturated
qupredicted = predicted qu ; soils. 1st ed. New York: Wiley.
S = the degree of saturation; Fredlund, D. G., A. Xing, M. D. Fredlund, and S. L. Barbour. 1996. “The
Se = effective saturation; relationship of the unsaturated soil shear to the soil-water characteristic
curve.” Can. Geotech. J. 33 (3): 440–448. https://doi.org/10.1139/t96
Sr = residual degree of saturation; -065.
Sw;AVE = average degree of saturation (Vahedifard and Garakani, A. A. 2013. Laboratory assessment of the hydro-mechanical
Robinson 2015); behavior of unsaturated undisturbed collapsible soils—Case study:
sc , sq , and sγ = shape factors; Gorgan loess. Tehran, Iran: Sharif Univ. of Technology.

© ASCE 04019181-15 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019181


Garakani, A. A., S. M. Haeri, D. Y. Cherati, F. A. Givi, M. K. Tadi, A. H. correlations.” Int. J. Geotech. Eng. 9 (1): 2–12. https://doi.org/10.1179
Hashemi, N. Chiti, and F. Qahremani. 2018. “Effect of road salts on /1939787914Y.0000000082.
the hydro-mechanical behavior of unsaturated collapsible soils.” Mohamed, F. M. O., S. K. Vanapalli, and M. Saatcioglu. 2011. “Bearing
Transp. Geotech. 17 (Dec): 77–90. https://doi.org/10.1016/j.trgeo.2018 capacity and settlement behavior of footings in an unsaturated sand.”
.09.005. In Proc., 14th Pan American on Soil Mechanics and Geotechnical
Garakani, A. A., S. M. Haeri, C. S. Desai, S. M. Hosein Seyed Ghafouri, Engineering, and the 64th Canadian Geotechnical Conf. Richmond,
B. Sadollahzadeh, and H. Hashemi Senejani. 2019. “Testing and con- Canada: Canadian Geotechnical Society.
stitutive modeling of lime-stabilized collapsible loess. II: Modeling and Ng, C. W. W., H. Sadeghi, F. Jafarzadeh, M. Sadeghi, C. Zhou, and S.
validations.” Int. J. Geomech. 19 (4): 04019007. https://doi.org/10.1061 Baghbanrezvan. 2019. “Effect of microstructure on shear strength
/(ASCE)GM.1943-5622.0001386. and dilatancy of unsaturated loess at high suctions.” Can. Geotech.
Garakani, A. A., S. M. Haeri, A. Khosravi, and G. Habibagahi. 2015. J. https://doi.org/10.1139/cgj-2018-0592.
“Hydro-mechanical behavior of undisturbed collapsible loessial soils Ni, P., G. Mei, and Y. Zhao. 2018. “Influence of raised groundwater level
Downloaded from ascelibrary.org by Indian Institute Of Technology, Indore on 01/16/24. Copyright ASCE. For personal use only; all rights reserved.

under different stress state conditions.” J. Eng. Geol. 195 (Sep): 28–41. on the stability of unsaturated soil slopes.” Int. J. Geomech. 18 (12):
https://doi.org/10.1016/j.enggeo.2015.05.026. 04018168. https://doi.org/10.1061/(ASCE)GM.1943-5622.0001316.
Haeri, S. M., A. A. Garakani, A. Khosravi, and C. L. Meehan. 2014. Oh, W. T., and S. K. Vanapalli. 2008. “Modeling the stress versus settle-
“Assessing the hydro-mechanical behavior of collapsible soils using ment behavior of model footings in saturated and unsaturated sandy
a modified triaxial test device.” Geotech. Test. J. 37 (2): 20130034. soils.” In Proc., 12th Int. Conf. of Int. Association for Computer Meth-
https://doi.org/10.1520/GTJ20130034. ods and Advances in Geomechanics (IACMAG). Red Hook, NY: Curran
Haeri, S. M., A. A. Garakani, H. R. Roohparvar, C. S. Desai, S. M. H. Associates, Inc.
Seyed Ghafouri, and K. Salemi Kouchesfahani. 2019. “Testing and con- Oh, W. T., and S. K. Vanapalli. 2013. “Interpretation of the bearing capacity
stitutive modeling of lime-stabilized collapsible loess. I: Experimental of unsaturated fine-grained soil using the modified effective and the
investigations.” Int. J. Geomech. 19 (4): 04019006. https://doi.org/10 modified total stress approaches.” Int. J. Geomech. 13 (6): 769–778.
.1061/(ASCE)GM.1943-5622.0001364. https://doi.org/10.1061/(ASCE)GM.1943-5622.0000263.
Haeri, S. M., A. Khosravi, A. A. Garakani, and S. Ghazizadeh. 2016. Oloo, S. Y. 1994. “A bearing capacity approach to the design of low volume
“Effect of soil structure and disturbance on hydromechanical behavior traffic roads.” Ph.D. thesis, Dept. of Civil Engineering, Univ. of
of collapsible loessial soils.” Int. J. Geomech. 17 (1): 04016021. https:// Saskatchewan.
doi.org/10.1061/(ASCE)GM.1943-5622.0000656. Oloo, S. Y., D. G. Fredlund, and J. K. M. Gan. 1997. “Bearing capacity of
Hanna, A. M., and G. G. Meyerhof. 1981. “Experimental evaluation unpaved roads.” Can. Geotech. J. 34 (3): 398–407. https://doi.org/10
of bearing capacity of footings subjected to inclined loads.” Can. .1139/t96-084.
Geotech. J. 18 (4): 599–603. https://doi.org/10.1139/t81-072. Prandtl, L. 1921. “Über die eindringungsfestigkeit (härte) plastischer baus-
Hansen, B. J. A. 1970. Revised and extended formula for bearing capacity. toffe und die festigkeit von schneiden.” [In German.] Z. Angew. Math.
Bulletin No. 28. Copenhagen, Denmark: Danish Geotechnical Institute. Mech. 1 (1): 15–20. https://doi.org/10.1002/zamm.19210010102.
Khalili, N., and M. H. Khabbaz. 1998. “A unique relationship for the Rajeev, P., and J. Kodikara. 2011. “Numerical analysis of an experimental
pipe buried in swelling soil.” Comput. Geotech. 38 (7): 897–904.
determination of the shear strength of unsaturated soils.” Géotechnique
https://doi.org/10.1016/j.compgeo.2011.06.005.
48 (5): 681–687. https://doi.org/10.1680/geot.1998.48.5.681.
Rojas, J. C., and L. M. Salinas. 2002. “Bearing pressure and settlement for a
Kumbhojkar, A. S. 1993. “Numerical evaluation of Terzaghi’s Nγ.” J. Geo-
lean clay in saturated and unsaturated conditions.” In Vol. 2 of Proc.,
tech. Eng. 119 (3): 598. https://doi.org/10.1061/(ASCE)0733-9410
3rd Unsaturated Soils Conf., 703–708. Rotterdam, Netherlands: A.A.
(1993)119:3(598).
Balkema.
Larkin, L. A. 1968. “Theoretical bearing capacity of very shallow
Rojas, J. C., L. M. Salinas, and C. Sejas. 2007. “Plate-load tests on an
footings.” J. Soil Mech. Found. Div. 94 (6): 1347–1360.
unsaturated lean clay.” In Experimental unsaturated soil mechanics,
Larson, L. 1997. Investigations and load tests in silty soils. Rep. No. 54.
edited by T. Schanz. Berlin: Springer.
Linköping, Sweden: Swedish Geotechnical Institute.
Rutqvist, J., Y. Ijiri, and H. Yamamoto. 2011. “Implementation of the
Li, Y., J. Wu, and K. Li. 2012. “Saturated-unsaturated seepage analysis Barcelona basic model into TOUGH–FLAC for simulations of the geo-
based on FLAC3D.” Rock. Soil. Mech. 33 (2): 617–622. mechanical behavior of unsaturated soils.” Comput. Geosci. 37 (6):
Lins, Y., T. Schanz, and S. K. Vanapalli. 2009. “Bearing capacity and set- 751–762. https://doi.org/10.1016/j.cageo.2010.10.011.
tlement behavior of a strip footing on an unsaturated coarse-grained Schnaid, F., N. C. Consoli, R. Cumdani, and J. Milititsky. 1995. “Load-
soil.” In Unsaturated soils, edited by O. Buzzi, S. Fityus, and settlement response of hallow foundations in structured unsaturated
D. Sheng. London: CRC Press. soils.” In Proc., 1st Int. Conf. on Unsaturated Soils, 999–1004. Rotter-
Lu, N., and W. J. Likos. 2004. Unsaturated soil mechanics. New York: dam, Netherlands: A.A. Balkema.
Wiley. Shahriar, M. A., N. Sivakugan, B. M. Das, A. Urquhart, and M. Tapiolas.
Meyerhof, G. G. 1951. “The ultimate bearing capacity of foundations.” 2015. “Water table correction factors for settlements of shallow foun-
Géotechnique 2 (4): 301–332. https://doi.org/10.1680/geot.1951.2.4 dations in granular soils.” Int. J. Geomech. 15 (1): 06014015. https://doi
.301. .org/10.1061/(ASCE)GM.1943-5622.0000391.
Meyerhof, G. G. 1956. “Penetration tests and bearing capacity of cohesion- Steensen-Bach, J. O., N. Foged, and J. S. Steenfelt. 1987. “Capillary in-
less soils.” J. Soil Mech. Found. Div. 82 (1): 1–19. duced stresses—Fact or fiction?” In Proc., 9th ECSMFE, Groundwater
Miller, G. A., and K. K. Muraleetharan. 1998. “In situ testing in unsaturated Effects in Geotechnical Engineering, 83–89. Rotterdam, Netherlands:
soil.” In Vol. 1 of Proc., 2nd Int. Conf. on Unsaturated Soils, 416–421. A.A. Balkema.
Beijing: International Academic Publishers. Tang, Y., H. A. Taiebat, and A. R. Russell. 2017. “Bearing capacity of shal-
Mohamed, F. M. O., and S. K. Vanapalli. 2006. “Laboratory investigations low foundations in unsaturated soil considering hydraulic hysteresis and
for the measurement of the bearing capacity of an unsaturated coarse- three drainage conditions.” Int. J. Geomech. 17 (6): 04016142. https://
grained soil.” In Proc., 59th Canadian Geotechnical Conf. Richmond, doi.org/10.1061/(ASCE)GM.1943-5622.0000845.
Canada: Canadian Geotechnical Society. Terzaghi, K. 1943. Theoretical soil mechanics. New York: Wiley.
Mohamed, F. M. O., and S. K. Vanapalli. 2012. “Estimation of bearing Terzaghi, K., and R. B. Peck. 1948. Soil mechanics in engineering practice.
capacity of saturated and unsaturated sands from the SPT and CPT cor- New York: Wiley.
relations.” In Proc., 2012 World Congress on Advances in Civil, Envi- Vahedifard, F., and J. D. Robinson. 2015. “Unified method for esti-
ronmental, and Materials Research (ACEM’ 12). Daejeon, Korea: mating the ultimate bearing capacity of shallow foundations in vari-
Techno-Press. ably saturated soils under steady flow.” J. Geotech. Geoenviron. Eng.
Mohamed, F. M. O., and S. K. Vanapalli. 2015. “Bearing capacity of shal- 142 (4): 04015095. https://doi.org/10.1061/(ASCE)GT.1943-5606
low foundations in saturated and unsaturated sands from SPT–CPT .0001445.

© ASCE 04019181-16 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019181


Vanapalli, S. K., K. D. Eigenbrod, Z. N. Taylan, C. Catana, W. T. Oh, and Vanapalli, S. K., W. T. Oh, and A. J. Puppala. 2007. “Determination of
E. Garven. 2010. “A technique for estimating the shaft resistance of test the bearing capacity of unsaturated soils under undrained loading con-
piles in unsaturated soils.” In Vol. 2 of Proc., 5th Int. Conf. Unsaturated ditions.” In Proc., 60th Canadian Geotechnical Conf. & 8th Joint
Soils, UNSAT 2010, 1209–1216. Boca Raton, FL: CRC Press. CGS/IAH-CNC Groundwater Conf. Richmond, Canada: Canadian
Vanapalli, S. K., D. G. Fredlund, D. E. Pufahl, and A. W. Clifton. Geotechnical Society.
1996. “Model for the prediction of shear strength with respect to soil van Genuchten, M. T. 1980. “A closed-form equation for predicting the
suction.” Can. Geotech. J. 33 (3): 379–392. https://doi.org/10.1139/t96 hydraulic conductivity of unsaturated soils.” Soil Sci. Soc. Am. J.
-060. 44 (5): 892–898. https://doi.org/10.2136/sssaj1980.03615995004400
Vanapalli, S. K., and F. M. O. Mohamed. 2007. “Bearing capacity of model 050002x.
footings in unsaturated soils, experimental unsaturated soil mechanics.” Vesic, A. S. 1973. “Analysis of ultimate loads of shallow foundations.”
In Proc., 2nd Int. Conf. on Unsaturated Soils, 483–493. Weimar, J. Soil Mech. Found. Div. 99 (SM1): 45–73.
Germany: Springer. Wu, L. Z., and R. Q. Huang. 2006. “Numerical simulation and optimum
Vanapalli, S. K., and F. M. O. Mohamed. 2013. “Bearing capacity and
Downloaded from ascelibrary.org by Indian Institute Of Technology, Indore on 01/16/24. Copyright ASCE. For personal use only; all rights reserved.

design of anchor frame beam strengthening expansive soil road cut


settlement of footings in an unsaturated sand.” Int. J. Geomate 5 (1): slope.” Rock Soil Mech. Wuhan 27 (4): 605.
595–604. Xu, Y. 2004. “Bearing capacity of unsaturated expansive soils.” J. Geotech.
Vanapalli, S. K., and W. T. Oh. 2010. “Mechanics of unsaturated soils for Geol. Eng. 22 (4): 611–625. https://doi.org/10.1023/B:GEGE
the design of foundation structures.” In Proc., 3rd WSEAS Int. Conf. on .0000047043.29898.17.
Engineering Mechanics, Structures, Engineering Geology, 363–377. Zhang, C., X. Chen, and W. Fan. 2016. “Overturning stability of a rigid
Stevens Point, WI: World Scientific and Engineering Academy and retaining wall for foundation.” Int. J. Geomech. 16 (4): 06015013.
Society. https://doi.org/10.1061/(ASCE)GM.1943-5622.0000613.

© ASCE 04019181-17 Int. J. Geomech.

Int. J. Geomech., 2020, 20(3): 04019181

You might also like