Soils and Foundations 2016;56(6):1071–1081

HOSTED BY The Japanese Geotechnical Society

Soils and Foundations

www.sciencedirect.com
journal homepage: www.elsevier.com/locate/sandf

Coupled infiltration model of unsaturated porous media for steady rainfall

Yongmin Kima, Sangseom Jeongb, Jaehong Kimc,n
a
School of Civil & Environmental Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore
b
School of Civil & Environmental Engineering, Yonsei University, 50 Yonsei-ro, Seodaemun-gu, Seoul 03722, Republic of Korea
c
Department of Civil Engineering, Dongshin University, 185 Geonjae-ro, Naju 58245, Republic of Korea
Received 18 April 2016; received in revised form 10 August 2016; accepted 30 September 2016
Available online 5 December 2016

Abstract

Unsaturated soil slopes introduce complex hydro-mechanical coupled processes which greatly alter matric suction distribution of an unsaturated soil
during rainfall. In order to investigate matric suction and volume change of unsaturated soils, a two-dimensional hydro-mechanical coupled infiltration
model (YS-Slope) is developed by incorporating the hydraulic and mechanical characteristics of unsaturated soils, such as the soil-water characteristic
curve, permeability function, shear strength, and porosity. Special attention is given to the porosity-dependent permeability function of unsaturated soils.
In addition, in order to highlight effectiveness of YS-Slope and coupling effects of hydro-mechanical processes on the infiltration behavior of
unsaturated soils, a series of infiltration analyses for a soil column under various soil properties is conducted and their results are compared with those of
commercial software, GEO-SLOPE (2012). The results of the numerical analyses show good agreement with data from the analytical solution and
laboratory tests, which indicates that the proposed model is appropriate for use in the simulatfion o the infiltration of rainwater into deformable soils.
The transient seepage and rainwater flow in deformable soils are influenced by the volume change of the soil. The change in matric suction on a slope
due to rainfall infiltration influences change in effective stress while the effective stress alters seepage processes according to hydraulic properties. The
results indicate that hydro-mechanical coupled behavior of soils has a positive effect on the stability of unsaturated soil slops during rainfall.
& 2017 The Japanese Geotechnical Society. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords: Infiltration model; Coupled analysis; Unsaturated soil; Numerical analysis; Seepage, Matric suction

1. Introduction done on analytical solutions describing the infiltration process

under transient state in homogeneous and layered soils. Gardner
It is important to evaluate the potential for landslides so that (1958) proposed an analytical solution for unsaturated flows
human safety can be maintained in high-risk areas. The likelihood based on an exponential hydraulic parameter model. Srivastava
of a shallow landslide due to rainfall infiltration on slope face is and Yeh (1991) developed theoretical solutions with hydraulic
strongly dependent on the hydraulic properties of unsaturated constitutive relations describing the non-uniform initial conditions
soils (Kim et al., 2014). The pore-water pressure on the for homogeneous or continuous heterogeneous soils based on the
unsaturated soil slopes influences the infiltration rate, the wetting Richards equation. However, these solutions may be very
band depth, and the soil strength. A number of studies have been restricted for any practical applications (e.g., complex geometry
condition and stress–strain relationship).
n
For the finite element (FE) analysis of unsaturated soils, several
Corresponding author.
E-mail addresses: [email protected] (Y. Kim),
commercial software, such as GEO-SLOPE (2012); PLAXIS 2D
[email protected] (S. Jeong), [email protected] (J. Kim). (2012); Soil Vision (2009) are widely used to simulate the water
Peer review under responsibility of The Japanese Geotechnical Society. infiltration process and to evaluate the slope stability with respect

http://dx.doi.org/10.1016/j.sandf.2016.11.010
0038-0806/& 2017 The Japanese Geotechnical Society. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
1072 Y. Kim et al. / Soils and Foundations 56 (2016) 1071–1081

Nomenclature φb Angle indicating the rate of increase in shear

strength relative to matric suction
α, n, m Fitting parameters of SWCC ρ Mass density of mixture
α Pore-size distribution parameter. ρwR Intrinsic mass density of liquid
Be;u Strain-displacement matrix for element S Degree of saturation
ce Fourth-order elastic modulus tensor Sr Residual degree of saturation
D Elastic modulus matrix Sc Degree of saturation due to capillary forces
de Nodal displacement vector for element Sa Bounded degree of saturation due to adhesion
ε Small strain tensor Sw Infiltration rate
εh Strain vector s, pa  pw Matric suction
g Gravitational acceleration r Total stress
γw Unit weight of water r0 Effective stress
hcap Mean capillary rise s  pa Net normal stress
ks Saturated permeability ts Total traction
kw Unsaturated permeability τ Shear strength of unsaturated soil
krw Relative permeability θ Volumetric water content
κ Intrinsic permeability θs Saturated volumetric water content
l2 Pore geometry parameter θr Residual volumetric water content
M Total mass of mixture θe Nodal vector of pore-water pressure for element
n Porosity V Total volume of mixture
pw Pore-water pressure v Velocity of solid
pa Pore-air pressure vw True velocity of liquid
φ0 Internal friction angle associated with the net v~ w Superficial Darcy velocity of liquid
normal stress ηw Dynamic water viscosity
χ Effective stress parameter

to the factor of safety. However, difficulties associated with triaxial shearing-infiltration test results. A series of infiltration
incorporating a soil deformation and a complex hydro- analyses for a soil column under various soil properties are
mechanical coupled process have hindered the study of the performed and their results are compared with those of the
mechanism of rainfall-induced slope failures. Recently, some commercial software, GEO-SLOPE (2012) to investigate the
researchers have developed a coupled hydro-mechanical finite coupling effect of hydro-mechanical processes on the infiltra-
element method, where the volume change of unsaturated soils is tion behavior of unsaturated soils.
included in the analysis. To observe the coupled solid-water-air
phenomenon in detail, the coupled balance equations for three 2. Mathematical frameworks for unsaturated flow
phase system consisting of a solid and two immiscible fluids (i.e.,
liquid and gas) are formulated by Khalili et al. (2004), Gray and 2.1. Governing equations
Schrefler (2007), and Borja et al. (2012). In addition, a number of
constitutive models for unsaturated soils have been developed The volume and mass of a mixture can be defined
based on a combination of experimental observations and mathematically. Based on small strain theory and homoge-
theoretical formulations (Alonso et al., 1990; Kohgo et al., neous condition, the volume of the mixture is
1993; Cui and Delage, 1996; Loret and Khalili, 2002; Borja, V ¼ V s þ V w þ V a and the corresponding total mass is
2004; Sheng et al., 2008; Nuth and Laloui, 2008; Zhang and M ¼ M s þ M w þ M a . Similarly, for the α phase (where α is
Ikariya, 2011; Zhou et al., 2012; Song and Yosef, 2015). Based on solid, liquid, and gas), M α ¼ ραR V α , where ραR is the true mass
past research work, it can be concluded that the focus of most of density of the α phase. The volume fraction occupied by the α
the investigations was on validating the models the authors phase is given by nα ¼ V α =V. Therefore, for the mixture of a
proposed. Although hydraulic properties and shear strength might solid, liquid and gas, the concept of volume fractions can be
affect the stability of soil slopes during rainfall (Cho and Lee, applied as (de Boer, 2006):
2001, Zhang et al., 2005, Kim et al., 2012; Borja et al., 2012),
ns þ nw þ n a ¼ 1 ð1Þ
coupled analyses are yet to be extensively employed in practical
engineering problems (Griffiths and Lu, 2005; Pande and where n ¼ ðV w þ V a Þ=V ¼ nw þ na is the porosity. The partial
Pietruszczak, 2015). mass density of the α phase is given by ρα ¼ nα ραR . In this
In this study, the two-dimensional hydro-mechanical model, mass exchange among the three phases is ignored (i.e.,
coupled infiltration model (called YS-Slope) is developed by chemical reactivity and phase changes are neglected). The
incorporating hydraulic and mechanical properties of unsatu- solid and liquid phases are assumed to be incompressible, and
rated soils. This model is verified with analytical solutions and the pore-air pressure is considered negligible (i.e., pa  0). The
 Y. Kim et al. / Soils and Foundations 56 (2016) 1071–1081 1073

respective equations for the balance of linear momentum and 2.2. Coupled formulation and implementation
the balance of mass for a three phases mixture are
For coupled formulations of two-dimensional plane strain
div rþ ρg ¼ 0 ð2Þ
nonlinear poromechanics, we assume the following: the iso-
tropic elasticity is given by the solid skeleton effective stress
∂S
n p_ þ Sdiv v ¼  div v~ w ð3Þ (Eq. (10)); the permeability function for the liquid phase is
∂s w described by Eq. (8); and the porosity, n, is a function of the
where r is the total (Cauchy) stress tensor, the mass density of solid skeleton volumetric strain, εv , (Eq. (11)). The coupled
the soil mixture is defined as ρ ¼ ρsR þ ρwR þ ρaR , g is the formulations are expressed in matrix form as:
gravitational acceleration vector, S is the degree of saturation, r' ¼ D U εh ¼ D U Be;u U d e ð10Þ
s ¼  pw is the matric suction, pw is the pore-water pressure, v
is the velocity of the solid skeleton, v~ w ¼ nw ðvw  vÞ is the nn þ Δεhv
n¼ ; Δεhv ¼ trðΔεh Þ ð11Þ
superficial Darcy velocity vector of liquid, and vw is the true 1 þ Δεhv
velocity vector of liquid.
The total stress can be written in terms of the solid skeleton's ρ ¼ ½1  nðde ÞρsR þ nðde ÞSðθe ÞρwR ð12Þ
effective stress, r0 , (Bishop, 1959) and the degree of saturation,
where D is the elastic modulus matrix, ε is the solid skeleton
h
S, is defined in the classical form as (van Genuchten, 1980):
strain vector, Be;u is the strain-displacement matrix for element,
r' ¼ r  pa 1þ χðpa  pw Þ1 ¼ r χpw 1 ð4Þ de is the nodal displacement vector for element, nn is the
porosity at the last converged time step, t n , θe is the nodal
S ¼ Sr þ Se ð1  Sr Þ; Se ¼ ½1 þ ðαsÞn   m ð5Þ vector of pore-water pressure, and ρ is the mass density of the
mixture. Based on these equations, under rainfall, the degree of
where χ is the effective stress parameter, Sr is the residual saturation and/or matric suction may be changed due to the
degree of saturation, s is the matric suction, and α (1/kPa), n, change of permeability function rapidly. At the same time, the
and m ( ¼ 1  1=n) are the curve fitting parameters of soil- alteration of the hydraulic condition may also influence the
water characteristic curve (SWCC). volumetric response of the soil. It means that the hydraulic and
For many quasi-static geotechnical engineering problems, mechanical properties of unsaturated soil are strongly depen-
we can assume that the pore air pressure, pa , is approximately dent on each other in this model.
equal to zero, but the soil suction still corresponds to a The weak form can be expressed in integral form to enable
negative pore-water pressure, s ¼  pw . To solve the two the implementation of YS-Slope. In YS-Slope, the weighting
unknowns (i.e., u and pw ), we use the balance of linear functions of ω and η are considered variations of the
momentum in Eq. (2) and the balance of mass for three phases displacement and pore-water pressure, respectively. The weak
mixture in Eq. (3). The constitutive equations are based on form is integrated by parts, and then, the divergence theorem is
linear isotropic elasticity for the effective stress r' and general- applied. Consequently, this results in the weak form for an
ized Darcy's law for the Darcy velocity of liquid as: unsaturated biphasic mixture, represented by
Z Z Z
r' ¼ ce : ε ð6Þ
½∇ω : ðr  Se pw 1 dV ¼ ρω U g dV þ ω U ts da ð13Þ
B B Γt


v~ w ¼ k w ðn; SÞ  ∇pw þ ρwR g ð7Þ Z Z Z Z
∂S
ηn ð  p_ w ÞdV þ ηS div v dV  ∇ηv~ w dV ¼ ηSw ds
where the fourth-order elastic modulus tensor is B ∂s B B Γs
ce ¼ λ1  1 þ 2μI, the Lame parameters are λ and μ, the small ð14Þ
strain tensor for the solid skeleton can be defined as s
ε ¼ sym∇u, k w is the unsaturated permeability, and ρwR is where t is the total traction on Γ t and S is the infiltration rate
w

the intrinsic mass density of liquid. The unsaturated perme- and positive inward at the boundary.
ability function, kw , and the relative permeability of the pore
water, k rw , are written as: 3. Coupled finite element modeling and validation
 
κðnÞ n3 The pore-water pressure profile and the wetting band depth
kw ðn; SÞ ¼ krw ðSÞ; κðnÞ ¼ l2 ð8Þ
ηw 1  n2 are dominant design factors for unsaturated soil slopes. Thus,
the numerical model, which can rationally estimate the pore-
pffiffiffi water pressure distribution of unsaturated soil slopes subjected
Sð1  ð1 S  m Þm Þ
2
krw ðSÞ ¼ ð9Þ
to rainfall infiltration, is needed for the slope stability analysis.
where κ is the intrinsic permeability of the porous solid In this section, two sets of numerical analysis are carried out to
skeleton, which is a function of the porosity, n, l2 is the pore simulate the shearing and rainfall infiltration on a unsaturated
geometry parameter, and ηw is the dynamic water viscosity. soil column, and numerical results are compared with data
The function n3 =ð1  n2 Þ is called the Kozeny–Carman formula from analytical solution and triaxial shearing infiltration tests
(Coussy, 2004). for the verification.
1074 Y. Kim et al. / Soils and Foundations 56 (2016) 1071–1081

3.1. Coupled finite element model z ¼ H) applied to the soil column, as shown in Fig. 1.
Impermeable boundaries on the right and left sides of the soil
The two-dimensional coupled balance equations can be column are applied to the coupled analyses. The saturated/
solved using two integration schemes (i.e., fully-implicit and unsaturated permeability and SWCC strongly influence the
semi-implicit). The semi-implicit time integration scheme leads pore-water pressure distributions in unsaturated zone. There-
to a linear solution at each time step. It is note that the semi- fore, it should be defined based on the laboratory tests.
implicit time integration scheme has better conditional stability
and is computationally faster than the fully-implicit time 3.2. Validation
integration scheme under small time steps of Δt ¼ 0:1 h
(Kim et al., 2015). Therefore, this study employed the semi- 3.2.1. Comparison with analytical solution
implicit linear solution that can be used effectively and with To confirm that the YS-Slope prediction of rainfall infiltration
greater efficiency in solving the coupled nonlinear finite in unsaturated soil is reasonable, the numerical results are
element balance equations and constitutive equations. compared with analytical solutions. Srivastava and Yeh (1991)
In the YS-Slope, rainfall infiltration and deformation ana- developed the analytical solutions to the linearized Richards
lyses of the unsaturated soil are carried out under a two- equations for constant rate vertical infiltration in homogeneous
dimensional plane strain condition. The solid skeleton is soils and two-layer soils. They assumed steady state infiltration
modeled using a quadrilateral finite element which comprises water pressure profiles as the initial conditions and adopted
a biquadratic for displacement and a bilinear for pore-water exponential functional relations. In addition, they presented the
pressure. The coupled finite element mesh for the unsaturated hydraulic behavior of the soils subjected to the constant infiltra-
soil column and the initial boundary conditions are presented tion rate during wetting process. Although the analytical solutions
in Fig. 1. The overall dimensions comprise a width of B and are not capable of simulating complex geometry condition as well
height of H. The displacement is fixed at the bottom boundary as stress–strain relationship, they often serve as the means for
and the vertical boundaries on both sides of the soil column. verifying numerical model for unsaturated flows.
In general, the initial pore-water pressure or matric suction The transient flow is simulated in a non-deformable homo-
profile is of great importance to achieve the desired initial geneous soil column with a height of 1 m, represented by three
condition. The specified initial matric suction associates with elements mesh (Fig. 1), for 100 h using the same geometry and
the particular values of permeability and SWCC at the boundary conditions of analytical solution (Srivastava and Yeh
beginning of seepage analysis, and hence rainfall infiltration 1991). All simulations are conducted in two dimensions
and runoff ratio are governed by the initial hydraulic condition. (essentially one-dimensional for the flow problems) with P4
In this analysis, the initial pore-water pressure is assumed to be element (displacement nodes are constrained). For verification,
initially hydrostatic, although the stress state of the unsaturated the constitutive equations for the liquid phase are assumed to
soil in the field is rarely constant (Lu and Likos, 2004). The be described by the following relations (Srivastava and Yeh,
model incorporates an unsaturated deformable soil with a 1991):
water table (pw ¼ 0 at z ¼ 0) and an infiltration rate (Sw at
ks
k w ðsÞ ¼ expð  αs=γ w Þ ð15Þ
γw

1

SðsÞ ¼ θr þ ðθs  θr Þexpð αs=γ w Þ ð16Þ
n
where k s is the saturated permeability, γ w is the unit weight of
water, s is the matric suction, α is a soil pore-size distribution
parameter, θr is the residual volumetric water content, θs is the
saturated volumetric water content, Sw0 is the initial (t ¼ 0)
infiltration rate, and Sw is the final (held constant for t 4 0)
infiltration rate. The saturated permeability is assumed to be
2.8  10  6 m/s. The α is assumed to be 1.0, which means that
hysteresis effect (wetting and drying cycles) can be neglected.
The saturated and residual volumetric water contents of the soil
are taken as 0.45 and 0.2, respectively. The initial infiltration
rate (Sw0 ) is equal to 2.8  10  7 m/s, and then, at t greater than
0, the infiltration is increased to 9Sw0 . The material properties
used in YS-Slope are summarized in Table 1.
Fig. 2 shows the comparison between the analytical solution
and the numerical analyses results at different time steps for the
transient flow with time increment of Δt ¼ 0:1 h. With the
exclusion of the soil deformation, the simulated matric suctions
Fig. 1. Coupled finite element mesh and boundary conditions. have fitted perfectly into the analytical solutions. The presence of
 Y. Kim et al. / Soils and Foundations 56 (2016) 1071–1081 1075

Table 1
Material properties of soil used in comparison with analytical solution.

Material property Value

Saturated permeability, ks (m/s) 2.8  10  6

Pore-size distribution parameter, α(1 /m) 1.0
Saturated volumetric water content, θs 0.45
Residual volumetric water content, θr 0.20
Initial infiltration rate, Sw0 (m/s) 2.8  10  7
Final infiltration rate, Sw (m/s) 9Sw0

Fig. 2. Comparison between analytical solution and simulated matric suction

Fig. 3. Comparison between analytical solution and simulated pore-water
obtained from YS-Slope at different time steps.
pressure obtained from YS-Slope at different depths.

soil deformation might have no significant effect on the transient 1997). The shearing is done under the drained condition.
flow since there is no loading condition in the analysis. Variations During infiltration stage, water is injected from the base at a
in the pore-water pressure at different depths are also shown in constant rate of 0.04 mm3/s while the top drainage is kept
Fig. 3. The node numbers 1, 7, and 13 refer to pore-water closed. The deviator stress is simultaneously maintained
pressure nodes of the mesh from top, middle and bottom as constant and therefore, there is no shear strain applied to the
shown in Fig. 1, respectively. The simulated pore-water pressure specimen. A fixed displacement rate of 1 mm/min is applied by
tends to be slightly lower than the analytical solution at the a compression machine to obtain a soil specimen with a
beginning of the simulation and has fitted conspicuously into the diameter of 50 mm and a height of 100 mm. In this study, the
analytical solution at the end of the simulation. Overall, a net normal stress applied is 200 kPa and the matric suctions
reasonably good agreement of transient flows is obtained between applied are 50 kPa, 100 kPa and 200 kPa for soils tested under
the numerical analysis results and analytical solutions. the unsaturated condition.
In the simulation, the full size of the 2D plane strain is
3.2.2. Comparison with triaxial shearing infiltration tests considered with the fixed boundary condition against displace-
Triaxial tests on unsaturated soil specimens under shearing- ment. The overall dimensions of the triaxial model boundaries
infiltration conditions (Rahardjo et al., 2009) are simulated comprise a width (B) of 50 mm and height (H) of 100 mm
using the YS-Slope. Triaxial shearing-infiltration tests are represented by a three finite element mesh. For the consolida-
conducted using a modified triaxial apparatus designed and tion stage, isotropic normal stress is applied on the top and the
constructed by Wong et al. (2001). Prior to the tests, the soil right side of the soil specimen equal to initial net confining
specimens are first saturated under zero net normal stress. pressure of 200 kPa. For the matric suction equalization stage,
When the soil specimens have been saturated, they are the desired matric suction is isotopically applied at the pore-
consolidated under a specified net confining pressure and then, water pressure nodes (nodes 1, 3, 7, 9, 13, 15, 19, and 21 in
matric suction equalization stage is carried out under a Fig. 1). The consolidation stage is set as the initial condition
specified matric suction. After the soil specimens reach for the subsequent simulations involving shearing and infiltra-
equilibrium, the soil specimens are then sheared using a strain tion. For the shearing stage, a predetermined vertical shear
rate of 0.0008 mm/min until the shear stress of the soil stress obtained from consolidated drained test (Meilani, 2004)
specimens reached 85–90% of the peak shear stress (Han, is applied to the top surface of the soil specimen as a stress
1076 Y. Kim et al. / Soils and Foundations 56 (2016) 1071–1081

controlled test. The 85% peak shear stress is chosen since the Table 3
shear stress in steep slopes in the field can be as high 85% to Mechanical properties of soil used in Modified Cam-Clay model.
90% of peak shear stress (Han, 1997). During the shearing Mechanical property Value
stage, applied matric suction at the pore-water pressure nodes
in matric suction equalization stage is maintained to incorpo- Virgin compressibility soil index, λ 0.13
rate the constant matric suction condition. For the infiltration Elastic compressibility index, κ 0.018
Slope of critical state line, M 1.05
stage, the positive value of total flux boundary condition is
Dimensionless parameter of suction, k 0.60
applied to the bottom nodes of the soil specimen (nodes 19, 20, Poisson's ratio, ν 0.21
and 21 in Fig. 1), while maintained matric suction boundary at Initial preconsolidation pressure, pc (kPa) 90
the bottom nodes is removed. This process causes the water Reference void ratio at pc , N 2.76
infiltration and the constant matric suction to be released. Parameter of overconsolidation, β' 1.00
The response of the unsaturated soil specimen is prescribed
by an isotropic, elasto-plastic constitutive model using the
Modified Cam-Clay (MCC) criterion developed by Borja 4. Numerical analyses of rainfall infiltration in unsaturated
(2004) with normally consolidated condition. The YS-Slope soils
can employ the various constitutive model as well as isotropic
elasticity according to numerical purpose. The hydraulic The YS-Slope analysis of rainfall infiltration in unsaturated
properties involved in the YS-Slope are listed in Table 2. soils under various soil conditions is carried out to obtain
The mechanical properties of the unsaturated soil specimen are matric suction profiles. The numerical results obtained from
listed in Table 3. The material properties used in the numerical YS-Slope are compared with those of commercial software
analyses are determined based on the soil properties used in the GEO-SLOPE (2012) to investigate the coupling effects of
triaxial consolidated drained tests by Meilani (2004). hydro-mechanical process on the behavior of unsaturated soils
Fig. 4 shows a comparison between the laboratory test and under rainfall infiltration. The numerical technique for simulat-
simulated results of the unsaturated soil specimen under ing shearing-infiltration behavior of unsaturated soils described
shearing-infiltration conditions. The dashed lines are the test in previous section is used in YS-Slope analysis.
results and the solid lines are the simulations. By taking into
account the coupled process and soil deformation, the YS-
Slope can describe the triaxial shearing-infiltration tests with 4.1. Finite element model
overall satisfactory accuracy. As can be seen in the stress–
strain relationships, the deviator stress could not be main- For the finite element analysis of unsaturated soil, GEO-
tained, and dropped gradually after failure due to the water SLOPE (2012) is widely used to evaluate the unsaturated flow,
infiltration. As can be seen in the matric suction variations, the soil deformation, and the slope stability. The two finite
decrease in the matric suction in response to changes in axial element software, namely Sigma/W and Seep/W, are numerical
strain is observed during the water infiltration stage, reaching methods used to estimate the flux of water through a vadose
minimum values at the end of the infiltration stage. As zone for the calculation of flow processes in unsaturated soils.
expected, in general, the test and simulated results describe The numerical analyses of the rainfall infiltration and soil
the triggering mechanism of unsaturated soil slope failure. deformation are conducted under a two-dimensional plane
Note that the complicated transient behavior of an unsatu- strain condition using Seep/W and Sigma/W, respectively. The
rated soil, which depends on types of soil, geometry of the soil has been discretized using nine-noded quadrilateral finite
irregular slope, types of the flux boundary condition, cannot be element with pore-water pressure and displacement nodal
simply simulated through YS-Slope in an idealized condition. degree of freedom. The dimensions of the numerical model
Although the proposed model has such limitations, it reason- comprise a width (B) of 1.0 m and a height (H) of 3.0 m. The
ably represents the behavior of the transient flow and initial hydraulic and fixed boundary conditions are the same as
deformation of unsaturated soils based on the comparisons the YS-Slope analysis, as shown in Fig. 1.
studies with results of analytical solution and laboratory tests. The YS-Slope simulates rainfall infiltration using a two-step
process. First, the infiltration is set equal to the precipitation
rate during each time step. Second, if the surface of the soil
Table 2
Hydraulic properties of soil used in comparison with triaxial shearing- saturates, the solution for that time step is repeated using a
infiltration test. Dirichlet boundary condition (with the surface node saturated).
The flux resulting from the surface profile is defined as the
Hydraulic property Value infiltration rate. Because the flow rate of infiltration across the
Saturated volumetric water content, θs 0.51 edge of an element is uniform, the total flow across the edge is
Residual volumetric water content, θr 0.21 the flow rate multiplied by the length of the element edge. To
Fitting parameter for SWCC, α(1/kPa) 0.024 calculate the unit rate of flow across the edge of an element, it
Fitting parameter for SWCC, n 1.77 is necessary to integrate along the edge of the element and
Saturated permeability, ks (m/s) 8.0x10  6
Residual degree of saturation, Sr 0.151
convert the unit rate of flow, q, into the nodal flow, which is
the Neumann boundary condition, Q.
 Y. Kim et al. / Soils and Foundations 56 (2016) 1071–1081 1077

Fig. 4. Comparison between triaxial tests and simulated results of soil specimen under shearing-infiltration condition.
1078 Y. Kim et al. / Soils and Foundations 56 (2016) 1071–1081

Table 4 Table 5
Hydraulic properties of soils used in rainfall infiltration analysis. Mechanical properties of soils used in SIGMA/W analysis.

Hydraulic property Loam Silt loam Clay Mechanical property Value

Saturated volumetric water content, θs 0.434 0.396 0.446 Young's modulus, E (kPa) 15,000
Residual volumetric water content, θr 0.218 0.131 0.0 Poisson's ratio, υ 0.33
Fitting parameter for SWCC, α(1/kPa) 0.02 0.00432 0.00152 Effective cohesion, c0 (kPa) 27
Fitting parameter for SWCC, n 2.76 2.06 1.17 Effective internal friction angle, φ0 (deg) 30
Saturated permeability, ks (m/s) 3.7x10  6 5.7x10  7 9.5x10  9
Residual degree of saturation, Sr 0.5 0.33 0.0

4.2. Material properties

The hydraulic properties of soils used in infiltration analysis

are summarized in Table 4. The terminology of θs and θr are
the saturated and residual volumetric water content, respec-
tively, Sr is the residual degree of saturation, ks is the saturated
permeability, α and n are fitting parameters of van Genuchten
equation for the SWCC. The hydraulic properties of the
subsurface, such as the rainfall intensity and duration, as well
as the initial water content affect the process of unsaturated
flow into the soil. At the beginning of rainfall event, the
instantaneous infiltration rate is estimated under the assump-
tion of saturated soil. During the rainfall duration, if the rate is
higher than the saturated permeability of the surface soil, then
the wetted depth of the soil increases. The seepage force is Sr =0.5
based on the infiltration rates for loam (3.710  5 m/s), silt
loam (5.710  7 m/s), and clay (9.510  9 m/s), which is
presented by van Genuchten (1980).
In Sigma/W analysis, the soil is modeled using an elastic-
plastic constitutive model (i.e., Mohr–Coulomb model). The
shear strength equation for unsaturated soils (Fredlund et al.,
1978) is as follows:
τ ¼ c0 þ ðs  ua Þ tan φ0 þ ðua  uw Þ tan φb ð17Þ
where c0 is the intercept of the extended Mohr-Coulomb failure
envelope on the shear stress axis where the net normal stress
and the matric suction are equal to zero, s  ua is the net
normal stress at failure, ua  uw is the matric suction at failure,
φ0 is the internal friction angle associated with the net normal
stress, and φb is the angle indicating the rate of increase in
shear strength relative to matric suction. Table 5 summaries the
mechanical properties of the soils used in Sigma/W analysis.

4.3. Results and discussions Sr =0.0

Fig. 5. Pressure head profiles of loam.
Fig. 5 demonstrates the pressure head differences in the
results from Seep/W and YS-Slope analyses with respect to the
residual degree of saturation, Sr , which uses different calcula-
tions for the residual degree of saturation in the infiltration soil. It is found from the numerical results that the increase in
simulation. There are two important reasons why the residual the capillary pressure due to the absence of the residual degree
degree of saturation is considered in numerical simulation. The of saturation slows down the infiltration rate. The influence of
first reason is that the residual degree of saturation of loam soil this effect is shown by Eqs. (8) and (9) for YS-Slope and
tends toward zero, but the value from van Genuchten (1980) is Eqs. (18) and (19) for Seep/W. In addition, the results of YS-
excessively high. The second reason is due to the effect of the Slope, which uses a porosity-dependent permeability, exhibit a
residual degree of saturation on the infiltration in unsaturated lower infiltration rate than the results of Seep/W because the
 Y. Kim et al. / Soils and Foundations 56 (2016) 1071–1081 1079

water flow rate is reduced in the deformable soil, where the

porosity is reduced due to the seepage force.

2
1  ðαsÞn  1 ð1þ ðαsÞn Þ  m
kw ðsÞ ¼ ð18Þ
½1þ ðαsÞn m=2

θ
S¼ ¼ Sc þ Sa ð1  Sc Þ ð19Þ
n
where θ is the volumetric water content, Sc is the degree of
saturation due to capillary forces, and Sa is the bounded degree
of saturation due to adhesion (Sa ), where Sa ¼ o 1  Sa 4 þ 1.
The adhesive component is a bounded value since Sa can be
greater than 1 at low matric suction. The bounded value
ensures that for Sa greater or equal to 1, Sa is equal to 1, and if
Sa is less than 1, Sa is equal to Sa . The capillary saturation,
which depends on the pore diameter and pore size distribution,
is given by Geo-Slope:
"  #m "   #
hcap 2 hcap 2
Sc ¼ 1 þ 1 exp  m ð20Þ
s s

where hcap is the mean capillary rise.

Fig. 6 shows the pressure head profiles for the soil column
of silt loam and clay. The differences in the pressure head
between Seep/W and YS-Slope are observed. The discrepancy
could be due to the fact that the permeability function in Seep/
W is based on matric suction function, while the permeability
function in YS-Slope is based on both matric suction and
porosity functions. In addition, Seep/W does not consider the
effect of changes of porosity when simulating the rainfall
infiltration process. Therefore, the pressure head profiles from
YS-Slope might predict different patterns to those of Seep/W.
Fig. 7 shows variations in the vertical displacement at a depth
of 1 m (node 13) and 2 m (node 7) due to water infiltration in the
deformable soil. A sequential analysis is composed of Seep/W and
Sigma/W analyses. Differences in the vertical displacement from
the sequential and coupled analysis are observed. It is found that Fig. 6. Pressure head profiles of two types of soil.
the vertical displacement from coupled analysis gradually
increases and exhibits nonlinear behavior after the rainfall began.
On the other hand, the vertical displacement from sequential
analysis more rapidly increases with time. These results confirm
the fact that displacement of deformable soil depends on the
changes in porosity. Fig. 8 shows a variation in porosity obtained
from coupled and sequential analysis. The porosity in coupled
analysis decreases, reaching a minimum value after one hour,
while the porosity in sequential analysis remains fairly constant.
The time histories of the mean effective stress from coupled
and sequential analysis are shown in Fig. 9. At the beginning of
simulation, the mean effective stresses rapidly increase due to
gravity loading in both analyses. After gravity loading, the mean
effective stresses tend to gradually increase due to increases in
unit weight with increases in the moisture content at node 13 in
the coupled analysis. However, the mean effective stresses of the
sequential analysis at node 13 maintain a constant value with
time. It can be said that the rate of effective stress change depends
on the increase in unit weight and pore-water pressure. As a
result, the YS-Slope is capable of considering variations in the Fig. 7. Variations in vertical displacement of soil column.
1080 Y. Kim et al. / Soils and Foundations 56 (2016) 1071–1081

deformation problem during rainfall. The proposed model is

validated on the matric suction and stress–strain relationship
predictions using data from analytical solution and triaxial
shearing-infiltration tests. In addition, the behavior of porosity-
dependent infiltration obtained from YS-Slope is verified
through comparison with commercial software of Geo-Slope.
The transient seepage analyses in deformable soils are influ-
enced by the volume change of the soil. The change in matric
suction on a slope due to rainfall infiltration results in changes
in the effective stress, while the effective stress alters the
seepage processes according to the hydraulic properties. The
consideration of porosity in calculating the permeability
function alters the matric suction distribution, and hence the
infiltration rates of the YS-Slope are slower than those of the
Geo-Slope. These trends indicate that the hydro-mechanical
coupled behavior of soils has a positive effect on the stability
Fig. 8. Variation in porosity of soil column.
of unsaturated soil slops during rainfall. Consequently, it is
important to analyze seepage in deformable soils to identify
the changes in the matric suction in the unsaturated zones
because the shear strength of geotechnical structures, which
can affect their likelihood of their failure, is closely related to
the matric suction.

Acknowledgments

The authors acknowledge the support in this research for the

National Research Foundation of Korea (NRF) (Grant nos.
2011-0030040 and 2013R1A1A2062218).

References

Alonso, E.E., Gens, A., Josa, A., 1990. A constitutive model for partially
saturated soils. Géotechnique 40 (3), 405–430.
Bishop, A.W., 1959. The Effective Stress Principle, 106. Norwegian Geo-
Fig. 9. Time histories of mean effective stress of soil. technical Institute, Teknisk Ukeblad859–863.
de Boer, R., 2006. Trends in Continuum Mechanics of Porous Media.
mass density, which is a function of porosity and the degree of Springer.
Borja, R.I., 2004. Cam-Clay plasticity. Part V: a mathematical framework for
saturation. In addition, when the pore-water pressure becomes
three-phase deformation and strain localization analyses of partially
positive (saturated), unloading of the mean effective stress at node saturated porous media. Comput. Methods Appl. Mech. Eng. 193 (48–51),
7 in YS-Slope occurs because the increase in the pore-water 5301–5338.
pressure becomes larger than the stress increase resulting from the Borja, R.I., White, J.A., Liu, X., Wu, W., 2012. Factor of safety in a partially
load increase (Fig. 9). Similar results of unloading behavior for saturated slope inferred from hydro‐mechanical continuum modeling. Int.
J. Numer. Anal. Methods Geomech. 36 (2), 236–248.
the mean stress were found previously by Cho and Lee (2001) in
Cho, S.E., Lee, S.R., 2001. Instability of unsaturated soil slopes due to
their numerical study. infiltration. Comput. Geotech. 28 (3), 185–208.
It should be noted that different use of constitutive equations Coussy, O., 2004. Poromechanics. John Wiley & Sons.
results in discrepancies in the matric suction distribution, Cui, Y.J., Delage, P., 1996. Yielding and plastic behaviour of an unsaturated
vertical displacement, and mean effective stress between the compacted silt. Géotechnique 46 (2), 291–311.
Fredlund, D.G., Morgenstern, N.R., Widger, R.A., 1978. The shear strength of
coupled and sequential analysis. Therefore, it is found that the unsaturated soils. Can. Geotech. J. 15 (3), 313–321.
coupled analysis can use the porosity function to calculate the Gardner, W., 1958. Some steady-state solutions of the unsaturated moisture
unsaturated permeability and to incorporate stress–strain rela- flow equation with application to evaporation from a water table. Soil Sci.
tionship, which distinguishes hydraulic-mechanical behavior 85 (4), 228–232.
more accurately than the sequential analysis. GEO-SLOPE, 2012. User's Manual. Geo-Slope International Ltd, Calgary,
Alberta, Canada.
Gray, G.W., Schrefler, B.A., 2007. Analysis of the solid phase stress tensor in
5. Conclusions multiphase porous media. Int. J. Numer. Anal. Methods Geomech. 31 (4),
541–581.
Griffiths, D., Lu, N., 2005. Unsaturated slope stability analysis with steady
A coupled hydro-mechanical coupled infiltration model infiltration or evaporation using elasto‐plastic finite elements. Int. J. Numer.
(YS-Slope) is proposed to simulate transient seepage and Anal. Methods Geomech. 29 (3), 249–267.
 Y. Kim et al. / Soils and Foundations 56 (2016) 1071–1081 1081

Han, K.K., 1997. Effect of Hysteresis, Infiltration and Tensile Stress on the PLAXIS 2D, 2012. User Manual, Brinkgreve, R., Swolfs, W., Plaxis.
Strength of an Unsaturated Soil (Ph.D. thesis). Nanyang Technological Rahardjo, H., Meilani, I., Leong, E.C., 2009. Shear strength characteristics of a
University, Singapore. compacted soil under infiltration conditions. Geomech. Eng. 1 (1), 35–52.
Khalili, N., Geiser, F., Blight, G., 2004. Effective stress in unsaturated soils: Sheng, D., Gens, A., Fredlund, D.G., Sloan, S.W., 2008. Unsaturated soils:
review with new evidence. Int. J. Geomech. 4 (2), 115–126. from constitutive modelling to numerical algorithms. Comput. Geotech. 35,
Kim, J., Jeong, S., Regueiro, R.A., 2012. Instability of partially saturated soil 810–824.
slopes due to alteration of rainfall pattern. Eng. Geol. 147, 28–36. Soil Vision, 2009. SVFlux User Manual: Saturated/Unsaturated Finite Element
Kim, J., Jeong, S., Kim, K., 2014. GIS-based prediction method of landslide 2D/3D Seepage Modeling. Soil Vision Ltd., Saskatoon, Canada.
susceptibility using a rainfall infiltration-groundwater flow model. Eng. Song, C.R. and Yosef, T., 2015. Seepage-heat coupled analysis for estimating
Geol. 182, 63–78. phreatic line of an earth dam from temperature profile. SAGEEP, Extended
Kim, J., Wang, W., Regueiro, R.A., 2015. Hybrid time integration and coupled paper, Austin, Texas USA.
solution methods for nonlinear finite element analysis of partially saturated Srivastava, R., Yeh, T.C., 1991. Analytical solutions for one‐dimensional,
deformable porous media at small strain. Int. J. Numer. Anal. Methods transient infiltration toward the water table in homogeneous and layered
Geomech. 39 (10), 1073–1103. soils. Water Resour. Res. 27 (5), 753–762.
Kohgo, Y., Nakano, M., Miyazaki, T., 1993. Verification of the generalized van Genuchten, M.T., 1980. A closed-form equation for predicting the
elastoplastic model for unsaturated soil. Soils Found. 33 (4), 64–73. hydraulic conductivity of unsaturated soils. Soil Sci. Soc. Am. J. 44 (5),
Loret, B., Khalili, N., 2002. An effective stress elastic–plastic model for 892–898.
unsaturated porous media. Mech. Mater. 34 (2), 97–116. Wong, J.C., Rahardjo, H., Toll, D.G., Leong, E.C., 2001. Modified triaxial
Lu, N., Likos, W.J., 2004. Unsaturated Soil Mechanics. Wiley. apparatus for shearing-infiltration test. Geotech. Test. J. 24 (4), 370–380.
Meilani, I., 2004. Shear Strength Characteristics of a Compacted Kaolin Under Zhang, F., Ikariya, T., 2011. A new model for unsaturated soil using skeleton
Infiltration Conditions (Ph.D. thesis). Nanyang Technological University, and degree of saturation as state variables. Soil Found. 51 (1), 67–81.
Singapore. Zhang, L.L., Zhang, L., Tang, W.H., 2005. Rainfall-induced slope failure
Nuth, M., Laloui, L., 2008. Effective stress concept in unsaturated soils: considering variability of soil properties. Géotechnique 55 (2), 183–188.
clarification and validation of a unified framework. Int. J. Numer. Anal. Zhou, A.N., Sheng, D.C., Sloan, S.W., Gens, A., 2012. Interpretation of
Methods Geomech. 32 (7), 771–801. unsaturated soil behaviour in the stress-saturation space, II: constitutive
Pande, G.N., Pietruszczak, S., 2015. On unsaturated soil mechanics – personal relationships and validations. Comput. Geotech. 43, 111–123.
views on current research. Studia. Geotech. Mech. 37 (3), 73–84.