Galavi Groundwater Flow and Coupled Analysis
Galavi Groundwater Flow and Coupled Analysis
Galavi Groundwater Flow and Coupled Analysis
Vahid Galavi
Research department
Plaxis BV 2010
Preface
The main goal of this research was to implement groundwater flow and fully
coupled flow-deformation analysis in PLAXIS 2D and 3D to enhance the code to
simulate flow and deformation in saturated and partially saturated soils. The existing undrained analysis of PLAXIS has also been modified for partially saturated soils. The mechanical behaviour of unsaturated soil is described by the
well-known Barcelona Basic Model (developed for PLAXIS 3D via user defined
soil model option by Nubia Gonzalez), Gonzalez & Gens (2008).
It is acknowledged that the scientific part of the fully coupled flow-deformation
analysis is mainly based on the report written by Radu Schwab (2008).
The outline of the work presented in this study is the following:
Chapter 1 presents basic equations and definitions.
Chapter 2 presents the governing equations of flow and deformation. Fist of all
Darcys law is described and then the continuity equation and deformation equations are derived.
The finite element formulation of the flow and the deformation equations derived
in Chapter 2 is presented in Chapter 3.
Chapter 4 presents available boundary conditions for groundwater flow calculation.
Hydraulic models, implemented in PLAXIS 2D and 3D, are discussed in Chapter
5.
Barcelona Basic Model is briefly described in chapter 6. This chapter gives the
basic features of the model which is written by Gonzalez & Gens (2008).
Chapters 7 to 10 present numerical verification of the code. Verification of onedimensional groundwater flow problems is given chapter 7, while the two problems are discussed in Chapter 8. Fully coupled-flow deformation analysis and
unsaturated soil model are verified in Chapter 9 and 10 respectively.
In chapter undrained analysis of PLAXIS is reviewed and the bulk modulus of
water used in different types of calculation is summarised. At the end of this
chapter some examples are given.
Table of contents
List of symbols
Introduction
Governing equations
15
Boundary conditions
21
Hydraulic models
27
33
47
117
175
235
247
12 Bibliography
273
Appendices
277
1 Introduction
1 Introduction
To analyse mechanical behaviour of saturated or partially saturated soils by
means of numerical methods (e.g. finite element method) in proper manner, it is
necessary to take into account both deformation and groundwater flow. For time
dependent behaviour, this leads to mixed equations of displacement and pore
pressure, called coupled hydro-mechanical approach, which have to be solved
simultaneously. For applications which involve a horizontal phreatic surface, the
equations can be simplified by decomposing the total pore pressure into a
constant component (steady state pore pressure) and a time dependent component
(excess pore pressure). But in many practical cases the distribution of stationary
pore pressure is unknown in the beginning of the calculation stage (undrained
excavations with dewatering or simulation of wave loading in off-shore
conditions). Therefore a more general formulation according to Biots theory of
consolidation is needed which enables the user to simultaneously calculate
deformation and groundwater flow with time-dependent boundary conditions in
saturated and partially saturated soils, as presented here. The main challenge in
this case is the need to use the consolidation theory for unsaturated soil
conditions at least due to the need to simulate the phreatic line. Due to
elastoplastic behaviour of soil skeleton and suction dependency of degree of
saturation and relative permeability, all coefficients of the global stiffness matrix
in the finite element formulations of Biot theory are non-linear. This case is
completely different from the equations of saturated soils where only the
elastoplastic stiffness matrix is non-linear. Therefore efficient numerical
procedures are required, as implemented in PLAXIS. Accuracy, robustness and
efficiency of the calculation depend on the method that selects the time
increments. PLAXIS 2D and 3D utilise a fully implicit scheme which is
unconditionally stable (Booker & Small, 1975).
Another essential issue for modelling the mechanical behaviour of unsaturated
soils is the constitutive model implemented in a coupled flow-deformation
analysis. A conceptually similar model to the well-known Barcelona Basic
Model (BBM) (Alonso et al., 1990), developed by Gonzalez & Gens (2008), has
been implemented in PLAXIS via user defined soil model option. The main
features of the implemented model is that it utilises Bishop stress and suction as
state variables (Sheng et al., 2003; Gallipoli et al., 2003) instead of net stress and
suction as utilised in the original BBM. In addition to an implicit stress
integration scheme based on backward Euler algorithm, a sub-stepping scheme
proposed by Prez et al. (2001) is used to integrate the strain-stress relations. The
input variables of the constitutive model are the increment of total strain and
increment of suction.
1 Introduction
Two types of calculations, namely steady state and transient groundwater flow
calculations for saturated and unsaturated soils have been fully implemented in
PLAXIS kernels. According to the type of element used for deformation analysis,
the new kernel uses the same type of element for groundwater flow calculation.
Usually higher order elements do not behave as well as lower order elements for
groundwater flow calculation. But additional procedures are utilised in the kernel
to overcome the problems related to higher order elements. It is shown that the
model is capable of calculating groundwater flow with a good accuracy.
Five types of hydraulic models have been implemented in PLAXIS kernels,
namely Van Genuchten, Mualem (simplified Van Genuchten which has been
called Van Genuchten in PlaxFlow kernel developed by GeoDelft), linearized
Van Genuchten, spline and fully saturated.
In the following all features of the new implementations are presented.
dVv
dV
S=
dVw
dVv
(1.1)
(1.2)
The water content is the ratio of the weight (or mass) of the water and the solids:
w=
dWw
n w
=S
dWs
1 n s
(1.3)
(1.4)
where s stands for the density of the solid particles and w is the water density.
1 Introduction
pw
= z + p
(1.5)
The equations are presented in a three-dimensional space with a vertical and
upwards oriented z-axis. For two dimensional problems the y-axis is vertical and
the range of the vectors and matrices is correspondingly reduced.
The vector format of the gradient operator is:
T
(1.6)
x
L =
T
x
0
z
0
(1.7)
1 Introduction
(groundwater flow boundary conditions). In the areas where the upward flux (i.e.
evaporation and evapotranspiration) exists, suction above phreatic level increases
(and degree of saturation decreases) and the water level is lowered with time
while in the case of downward flux (i.e. precipitation) suction decreases (and
degree of saturation increases) and the water level rises with time. In the case of
zero net surface flux, the pore water pressure profile become in equilibrium at a
hydrostatic condition (Figure 1.1).
Fig. 1.1: A visualisation of soil mechanics showing the role of the surface flux
boundary conditions (Fredlund, 1996)
1.3 Suction
Water potential is the potential work of pure water relative to a reference. This
causes to flow water in porous media from an area with higher water potential to
an area with lower water potential. The total water potential can be considered as
the summation of water potential due to matric, osmotic, gas pressures and
gravity. Flow in unsaturated zones relates to total suction which summation of
matric S and osmotic suction :
St = S +
(1.8)
(1.9)
Matric suction is related to soil matrix (adsorption and capillarity due to soil
matrix) and it is the difference between soil water pressure and gas pressure:
S = pa pw
(1.10)
1 Introduction
where, pw and pa are the pore water pressure and the pore air pressure,
respectively. In most cases, the pore air pressure is constant and small enough to
be neglected. Therefore the matric suction is negative of the pore water pressure:
S = pw
(1.11)
(1.12)
where:
= ( xx yy zz xy yz zx )
(1.13a)
m = (1 1 1 0 0 0 )
(1.13b)
is the vector with total stresses, contains the effective stresses, pw and pa are
the pore water pressure and the pore air pressure, respectively, and m is a vector
containing unity terms for normal stress components and zero terms for the shear
stress components. is an effective stress parameter called matric suction
coefficient and varies from 0 to 1 covering the range from dry to fully saturated
conditions. Considering these two special cases shows that for a fully saturated
soil ( = 1), the classical effective stress equation for compressive pore pressure
is obtained as:
= + m p w
(1.14)
(1.15)
This concept can be simplified for practical application assuming that the pore air
pressure is constant and is small enough to be neglected (i.e. pa 0). Therefore
for a completely dry soil, effective and total stresses are essentially equal. The
matric suction coefficient is generally determined experimentally. This
parameter depends on the degree of saturation, porosity and on the matric suction
1 Introduction
(pa pw) (e.g. Bolzon et al., 1996; Bishop & Blight, 1963). The experimental
evidences on the matric suction coefficient are quite sparse and therefore this
parameter is often assumed to be equal to the effective saturation in PLAXIS.
Now the effective stress formulation can be simplified to
= + m (S e p w )
(1.16)
2 Governing equations
2 Governing equations
2.1 Darcys law
The flow of water in a saturated soil is commonly described using Darcys law
(1856). He postulated that the rate of water flow through a soil mass is
proportional to the hydraulic head gradient. The equations of equilibrium for
groundwater flow are:
pw + w g + = 0
(2.1)
where g = (0, -g, 0)T is the vector of gravitational acceleration and the vector
of the friction force, per unit volume, between the flowing fluid and the soil
skeleton. This force is linearly dependent on the fluid velocity and acts in
opposite direction. The relations are:
= mint q
(2.2)
mint =
0
0
(2.3)
with the dynamic viscosity of the fluid and i the intrinsic permeability of the
porous medium. From (2.1) and (2.2) results:
pw w g + m q = 0
int
(2.4)
int
( pw + w g )
(2.5)
0
0
0
0
(2.6)
2 Governing equations
sat
= w g
i = x, y , z
(2.7)
i = x, y , z
(2.8)
k rel sat
k ( p w + w g )
w g
(2.9)
sat
sat
0
0
0
sat
ky
0
0
0
kz
(2.10)
sat
(2.11)
where
=
dVw Vw
dp
(2.12)
where dVw and Vw are volume of the water and volume variation due to the
variation of the pressure.
For unsaturated groundwater flow the compressibility of water can be expressed
as follows (Bishop & Eldin, 1950; Fredlund & Rahardjo, 1993).
2 Governing equations
= S w +
1 S + hS
K air
9
(2.13)
1 S
K air
(2.14)
sat
T w rel k ( p w + w g ) = ( w n S )
t
w g
(2.15)
( w n S ) = n S w w n S w S n
t
t
t
t
(2.16)
These three terms represent the changes in water density, saturation and soil
porosity, respectively.
According to the principle of mass conservation, for different corresponding
values of pressure and volume, the mass is constant, i.e.:
m w = wVw = c
(2.17)
Thus:
dm w = w dVw + d wVw = 0
or
(2.18)
10
2 Governing equations
dVw d w
=
Vw
w
(2.19)
= dp
(2.20)
(2.21)
Now, the term containing the derivative of w with respect to time can be
expressed as:
nS
w
p
n w p w
= nS w w =
S
t
p w t
Kw
t
(2.22)
The second term of the right hand side of Eq. (2.16) has the form:
w n
S
S p w
= n w
t
p w t
(2.23)
v
T
= m
t
t
(2.24)
The compression of the solid particles due to the changes of the pore
pressure:
(1 n) pw
S
Ks
t
(2.25)
where Ks is the bulk modulus of the solid particles forming the soil skeleton and
The compression of the solid particles due to the changes in effective
stresses:
1 T
1
p
m M
S w m
3K s
t 3K s t
(2.26)
2 Governing equations
11
Substituting all the factors in Eq. (2.15) and neglecting the second order
infinitesimal terms the continuity equation is obtained as:
w S mT
k
(1 n ) p w
S p w
sat
w S
+
+ n w
+ T w rel k ( p w + w g ) = 0
t
K s t
p w t
Kw
wg
(2.27)
Equation (2.27) can be reformulated for the flow problem by neglecting the
deformations of the solid particles and the density gradients of water
(Boussinesqs approximation):
Sm
S
n
t
K w p w
p w
k
sat
+ T rel k ( p w + w g ) = 0
t
w g
(2.28)
K w p w
p w
k
sat
+ T rel k ( p w + w g ) = 0
t
w g
(2.29)
(2.30)
12
2 Governing equations
specific storage (L-1). The specific storage Ss is a material property which can be
expressed as:
1 n
n
S s = w g
+
K
K
w
s
(2.31)
n w g
Kw
(2.32)
=
(nS ) = n S
h h
h
(2.33)
By substituting equations (2.32) and (2.33) in the Richards equation (Eq. 2.30)
and changing from head based equation to pore water pressure based equation,
equation (2.29) is obtained.
For steady state groundwater flow, in which variation of pore water pressure with
respect to time is zero, the continuity condition applies:
k
sat
T rel k ( p w + w g ) = 0
w g
(2.34)
2 Governing equations
13
(2.35)
where
= (1 n) s + n S w
(2.36)
(2.37)
(2.38)
(2.39)
M represents the material stress-strain matrix. The governing equation for the
deformation model is obtained:
L M ( L d u ) + S e dp w m] + d ( g ) = 0
T
(2.40)
14
2 Governing equations
15
(3.1)
(3.2)
where B is a matrix containing the spatial derivatives of the shape functions. The
virtual work equation is:
dV = u T b dV + u T t d
(3.3)
where b is the body force vector in the volume V and t is the traction on the
boundary . The stresses can be computed incrementally:
=
i
i 1
+ =
i 1
ti
+ & dt
t
(3.4)
i 1
If Eq. (3.3) is considered for the actual state i, the unknown i can be eliminated
using Eq. (3.4), therefore:
dV = u b i dV + u t d
T
T i
i 1
dV
(3.5)
dV = N b i dV + N t d B
T
T i
i 1
dV
(3.6)
By writing the body forces and the boundary tractions in incremental form the
following equation is obtained:
B
V
dV = N b dV + N t d + rv
T
i 1
(3.7)
16
i 1
T i 1
= N b i1 dV + N t
T
d B
T
i 1
(3.8)
dV
The residual force vector should be equal zero if the solution of step i is accurate.
Plaxis uses in the consolidation analyses the same shape functions for
displacements1 and for pore pressures, i.e.:
pw = N pn
(3.9)
The principle of effective stresses Eq. (1.12) can be written in the following
form:
i 1 = i 1 + S e i 1 p w i 1 m
= + S e
i 1
(3.10)
pw m
(3.11)
( + S e p w m) dV = N b dV + N t d + rv
i
i 1
(3.12)
M B v dV +
T
T
T
i
B m p w dV = N b dV + N t d + rv (3.13)
or in matrix form:
K v + Q pw = f u + r
i
v
(3.14)
where K, Q and fu are the stiffness matrix, the coupling matrix and the
increment of the load vector, respectively.
T
K = B M B dV
(3.15)
For generality, different sets of shape functions may be used to describe the variation of the
displacements and the pore pressure rates. This implies that the nodes in the finite element mesh may
have varying degrees of freedom, with some being associated with displacements, some being associated
with pore pressure, and some being associated with both. In order for the pore pressure rates to be
consistent with the stress rates, one can choose the polynomial describing the pore pressure rates to be
one order lower than the polynomial describing the displacements. This approach leads to less accurate
estimates of the displacements but smaller oscillations in the pore pressures (see Abbo, 1997)
17
Q = S e B m N dV
(3.16)
V
T
f u = N b dV + N t dS
(3.17)
The actual changes of the degree of saturation are included in the increments of
the body forces (Eq. 3.17).
S m LN
( N ) T
V
k rel
k
dv
S dp w
T S
sat
N
dV N n
dV ( N ) T rel k N p w dV
dt
K
p
dt
w
w
w
V
V
k
sat
w g dV N qdS = 0
(3.18)
and in matrix form:
H pw S
d pw
dt
+C
dv
= G +qp
dt
(3.19)
where H, Q, C and S are the permeability matrix, the coupling matrix and the
compressibility matrix. qp is the flux on boundaries. G is a vector in which effect
of gravity on flow in vertical direction is considered. This vector is a part of
external flux.
H = ( N ) T
V
k rel
sat
( N ) dV
T nS
dS
N dV
S = N
n
dpw
Kw
V
C = N S L N dV
(3.20)
(3.21)
(3.22)
G = ( N ) T
V
k rel
sat
w g dV
(3.23)
18
q p = N q w dS
(3.24)
d pw
dt
= G +qp
(3.25)
For steady state calculation, time derivative of pore pressure is zero, therefore:
H pw = G + q p
(3.26)
K
0
Q v 0
+
H p w C
dv
0 dt f u
S d p w G + q p
dt
(3.27)
The symmetry of the system (Eq. 3.27) can be restored by the time differentiation
of the first equation:
K
C
dv
Q dt
0 0 v
=
0 H p +
S d p w
w
dt
df u
dt
G + q p
(3.28)
19
dX
+CX =F
dt
where X T = v
(3.29)
dt
i +
i +1
X X X
=
=
t
t
i +
= (1 ) X + X
i
i +1
(3.30)
[B + t C ]
i +
i +1
= B (1 ) t C
i +
X + t F
i
i +
(3.31)
S *
i +
0
v
0
p =
0 t H
w
i +
fu
vi
i + t G + t (q i + q ) (3.32)
p
p
p w
with
S = ( S + t H )
*
H = ( N ) T
V
k rel
(3.33)
k
sat
( N ) dV
T nS
dS
N dV
S = N
n
dpw
Kw
V
G = ( N ) T
V
k rel
q p = N q w dS
sat
w g dV
(3.34)
(3.35)
(3.36)
(3.37)
K = B M B dV
(3.38)
V
T
Q = S B m N dV
V
(3.39)
20
(3.40)
V
T
f u = N b dV + N t dS
V
(3.41)
In the case of consolidation of unsaturated soils all matrices and the external flux
(right hand vector) are nonlinear. In this respect the following issues should be
taken into account:
The stiffness matrix K is usually stress-dependent.
The permeability in the permeability matrix H and in the vector G is
pressure dependent, due to suction dependency of relative permeability
krel.
The coupling matrices Q and C as well as the compressibility matrix S are
suction dependent. The latter is also depends on the derivative of
saturation
In addition, the boundary conditions for the seepage line and drains are
also nonlinear.
The right hand side of both equilibrium and mass conservation equations
are nonlinear terms for unsaturated soils. The nonlinearity of the first
equation is due to the weight of the soil which is a function of the degree
of saturation and the nonlinearity of the right side of the second equation
is due to the suction dependency of the relative permeability and the
variable Neumann boundary conditions.
For both equations the Cauchy BC are imposed directly in the equation system.
4 Boundary conditions
21
4 Boundary conditions
The following boundary conditions are available in PLAXIS:
4.1 Closed
This type of boundary conditions specifies a zero Darcy flux over the boundary
as
q n = q x nx + q y n y + q z nz = 0
(4.1)
where nx, ny and ny are the outward pointing normal vector components on the
boundary.
4.2 Inflow
A non-zero Darcy flux over a boundary is set by a prescribed recharge value q
and reads
q n = qx nx + q y n y + qz nz = q
(4.2)
This indicates that the Darcy flux vector and the normal vector on the boundary
are pointing in opposite directions.
4.3 Outflow
For outflow boundary conditions the direction of the prescribed Darcy flux, q ,
should equal the direction of the normal on the boundary, i.e.:
q n = qx nx + q y n y + qz nz = q
(4.3)
4.4 Head
For prescribed head boundaries the value of the head is imposed as
=
(4.4)
(4.5)
These conditions directly relate to a prescribed head boundary condition and are
implemented as such.
22
4 Boundary conditions
4.5 Infiltration/evaporation
This type of boundary conditions poses a more complex mixed boundary
condition. An inflow value q may depend on time and as in nature the amount of
inflow is limited by the capacity of the soil. If the precipitation rate exceeds this
capacity, ponding takes place at a depth max and the boundary condition
switches from inflow to prescribed head. As soon as the soil capacity meets the
infiltration rate the condition switches back.
This boundary condition simulates evaporation for negative values of q . The
outflow boundary condition takes place, when the groundwater head is higher
than the minimum head specified by the user min .
These boundary conditions are expressed as
if
= y + max
q n = q x n x + q y n y + q z n z = q if
= y + min
if
y + min
ponding
< y + < y + max (4.6)
drying
4.6 Seepage
Flow problems with a free water level may involve a seepage surface on the
downstream boundary, as shown in Figure 4.1. A seepage surface will always
occur when the water level touches an open downstream boundary. The seepage
surface is not a streamline (in contrast to the water level) or an equi-potential
line. It is a line on which the groundwater head, h, equals the elevation head y (=
vertical position). This condition arises from the fact that the water pressure is
zero on the seepage surface, which is the same condition as that exists at the
water level.
It is not necessary to know the exact length of the seepage surface before the
calculation begins, since the same boundary conditions (h = y) may be used along
the whole boundary line where seepage is expected to occur. Free boundaries
with h = y may therefore be specified for all boundaries where the hydraulic head
is unknown. Alternatively, for boundaries well above the water level where it is
obvious that a seepage surface does not occur, it may also be appropriate to
prescribe those boundaries as closed flow boundaries.
4 Boundary conditions
23
if
=
=z
if
q n = q n + q n + q n = 0 if
x x
y y
z z
(4.8)
As the source term in the governing equation simulates water flowing in the
system, the source term is positive for a recharge well.
(4.9)
The source term in the governing equation is negative for a discharge well.
4.9 Drain
Drains are handled as seepage boundaries. However, drains are located inside the
domain. In reality drains cannot work perfectly and do not permit water leaving
24
4 Boundary conditions
=
if
q n = q x n x + q y n y + q z n z = 0 if
outflow
suction
(4.10)
4.10 Interfaces
Interface elements are used to simulate impermeable structural elements. In such
an element there is no connection between both sides of the element and
therefore a zero Darcy flux over internal boundary is obtained.
Initial conditions are generated as a steady state solution for a problem with a
given set of boundary conditions.
This parameter represents the time interval for the calculation phase,
expressed in unit of time. Its value is equal to the Time interval parameter
as specified in the Parameters tab sheet of the Phase list window. The
value is fixed and cannot be changed in the Time-dependent head
window.
H0 :
This parameter represents the actual height of the water level, expressed
in unit of length. Its value is automatically calculated in the kernel based
on the initial pore pressures.
Hult:
4 Boundary conditions
25
of head for the current calculation phase. Hence, together with the time
interval this parameter determines the rate of the water level increase or
decrease.
For a linear variation of infiltration, inflow or outflow the inputs of the following
parameters are required:
Q0 :
This parameter is the initial specific discharge through the geometry line
under consideration, expressed in unit of length per unit of time.
Qult: This parameter, specified in unit of length per unit of time, represents the
ultimate specific discharge in the time interval of the current calculation
phase.
Harmonic (function 2): This option can be used when a condition varies
harmonically in time. The harmonic variation of the water level is described as:
y (t ) = y0 + 0.5 H sin(0t + 0 )
(4.11)
with
0 = 2 / T
in which H, T and 0 are the wave height in the unit of length, the wave period in
the unit of time and the initial phase angle, respectively.
In case of infiltration, inflow or outflow, the parameter QA needs to be entered
instead of H. QA represents the amplitude of the specific discharge and is
specified in unit of length per unit of time.
Table (function 3): In addition to the pre-defined functions for variations with
time, PLAXIS provides the possibility to enter user-defined time series. This
option can be useful for a back-analysis when measurements are available. In the
table, time always starts from zero which is related to the beginning of the
calculation phase.
26
4 Boundary conditions
5 Hydraulic models
27
5 Hydraulic models
5.1 Van Genuchten model
There are many material models which describe the hydraulic behaviour of
unsaturated soils. The most common in the groundwater literature is the model of
Van Genuchten (1980) relationship, which is used in PlaxFlow. This relationship
is a more general case of Mualem (1976) function. Van Genuchten function is a
three-parameter equation and relates the saturation to the suction pore pressure
head p:
gn gc
(5.1)
where
p = pw g
w
Sresidu is the residual saturation which describes the part of water that remains in
soil even at high suction heads. Ssat is the saturation when the pores are filled with
1 gn
gn
(5.2)
Figure 5.1 shows the effect of the parameter ga on the shape of retention curve.
This parameter is related to the air entry value (AEV) of the soil.
The effect of the parameter gn, which is a function of the rate of water extraction
from the soil once the AEV has been exceeded, is plotted in Figure 5.2.
The parameter gc is a function of the residual water content (related to the
curvature in the high suction range), Figure 5.3.
28
5 Hydraulic models
1.2
ga = 1.0
ga = 10
ga = 100
Saturation
0.8
0.6
0.4
0.2
0
0.1
10
100
1000
Suction (kPa)
Fig. 5.1: Effect of the parameter ga on the retention curve, (gn = 2.0 and
gc = -1.0)
1.2
gn = 0.5
gn = 1.0
gn = 2.0
Saturation
0.8
0.6
0.4
0.2
0
0.1
10
100
1000
Suction (kPa)
Fig. 5.2: Effect of the parameter gn on the retention curve, (ga = 1.0 and
gc = -1.0)
5 Hydraulic models
29
1.2
gc = -1.0
gc = -2.0
gc = -4.0
Saturation
0.8
0.6
0.4
0.2
0
0.1
10
100
1000
Suction (kPa)
Fig. 5.3: Effect of the parameter gc on the retention curve, (ga = 1.0 and gn = 2.0)
The effective saturation is defined as:
Se =
S S residu
S sat S residu
(5.3)
gn
gn
gl
k rel ( S ) = ( S e ) 1 1 S e gn 1
(5.4)
0
S ( pw )
=
1 2 gn
pw
gn
gn g
g
g
p
g
1
n
a
w
n
(S S
g
p
1
g
residu
w
n
a
sat
w
g n w
if
( pw 0 )
if
( pw > 0 )
(5.5)
30
5 Hydraulic models
Figure 5.1 and Figure 5.2 present the Mualem - Van Genuchten relations for a
sandy material with parameters Ssat = 1.0, Sres = 0.027, ga = 2.24 m-1, gl = 0:0 and
gn = 2.286 graphically.
S ( p ) = 1 + p
ps
p 0
1
if
( p w 0)
pws
0
if
p < p s
( pw > pws )
(5.6)
5 Hydraulic models
31
S ( pw )
=
pw
if
0
1
pws
0
p 0
( p w 0)
p < p s
(5.7)
( pw > pws )
The variable ps represents the threshold of the unsaturated condition and can be
derived from Van Genuchten model:
ps =
(5.8)
S p = 1, 0 m S sat
if
if
if
p 0
( p w 0)
(5.9)
32
5 Hydraulic models
33
(
)
(6.1)
34
Where, p is the mean effective stress, J is the square root of the second stress
1
invariant of deviatoric stress tensor, J = trace ij p ' ij
2
1/2
. The function
g ( ) is defined as:
g ( ) =
sin '
sin sin '
cos +
3
(6.2)
Po s* *
Pc = Pr
Pr
(6.3)
where P0' is the yield surface location at zero suction and is also the hardening
parameter, Pr is a reference mean stress, o* is the modified compression index
for saturated soil, s* is the slope modified of the NCL for unsaturated soil and
* is the modified swelling index which is assumed to be independent of suction.
(6.4)
where, s describe the increase in cohesion with suction. The slope s* is assumed
to vary with the suction according to:
s* = o (1 r ) exp ( s ) + r
(6.5)
Where r and are two additional material constants that can be determined
experimentally. The first is a constant related to the maximum stiffness of the
soil (for an infinite suction), r = s ( s )/ o , and the second controls the rate of
increase of soil stiffness with suction.
35
K=
G=
p'
(6.6)
*
3 (1 2 ) K
2 (1 + )
(6.7)
In the model a change in suction produces a volumetric elastic strain given by:
d ije,s =
s
3( s + patm )
ds ij =
1
ds ij
3K s
(6.8)
(6.9)
(6.10)
M ( M 9 )( M 3)
9(6 M )
1/ 1 * / *o
(6.11)
36
and Sture (1997) and Prez et al., (2001). The solution is sough by using the
flow direction mij = G ij at the final stress state.
The variables given as input of the mechanical constitutive subroutine are the
increment of total strain and increment of suction.
(6.12)
(6.13)
d ijp = d
d =
G ij , , s
ij
d ijp
ijp
(6.14)
(6.15)
where, d ij , d ije and d ijp are increments of the total elastic and plastic strain
tensors respectively and d ije, s is the contribution of suction to increment of elastic
strain tensor. d represents the increment of hardening parameters (in this case of
P0 ) and d is the plastic multiplier to be determined with the aid of the loading-
F ij , , s 0
d 0
(6.16)
F d = 0
During any process of loading, conditions (Eq. 6.16) must hold simultaneously.
37
(6.17)
( n+1) = ( n ) + ( n+1)
F
( n+1)
=0
where,
(6.18)
(n+1)
p ( n+1)
ij
G
= (n+1)
ij
(6.19)
(n+1)
( n+1)
= p
ij
ijp (n+1)
(6.20)
Where (n+1) is the actual load step and (n) is the converged step.
Time-integration of equation (6.17) with backward Euler scheme yields the
following nonlinear local problem (in compact notation),
(n+1) = (n) + D: ( n+1) D:m ( n+1) D: se
(n+1)
= + p
( n+1)
(n)
m ( n+1) ( n+1)
(6.21)
F ( (n+1) , (n+1) , s ) = 0
In equation (6.21), the state at time t(n) (i.e., quantities (n) and (n) ), the
increment of total strains from time t(n) to time t(n+1), , and the suction, s, are
known. The unknowns of this local problem are the stresses (n+1) and the
hardening parameters (n+1) at time t(n+1), and the plastic multiplier .
Formulating the residual of the three non-linear equations (6.21), the local
Newton-Raphson solver may be stated as follows:
( n+1) + D:m (n+1) + D: e ( n) D: = 0
s
n+1)
(
R { (n+1) , (n+1) , } = ( n+1) p
m ( n+1) ( n) = 0
F (k ) , ( k ) , s = 0
n+1
n+1
(6.22)
38
The non-linear system of eight equations is solved by linearizing the residual and
expanding it into a Taylor series:
R { , , }
R { + , + , + } = R { , , } +
+ O 2
( , , )
(6.23)
R { , , }
renders the Jacobian of the residual R :
( , , )
m
m
D:
+ D:
m
m
J { , , } = - p
1 - p
F
F
D:m
-
m
p
(6.24)
Truncation after the first order terms, O 2 0 , and letting the residual equation
(6.23) go to zero, it obtains a set of linear equations for the corresponding
increments of [ , , ] , that simultaneously reduces all three residuals to zero:
k+1
0 = R { k , k , k } + J { k , k , k } k+1
k+1
(6.25)
Indices k and k+1 denote the iteration cycle. Solving the linearized system of
equations the new iterative update of the eight variables is obtained:
k+1
= J , , 1 R , ,
{ k k k}
k }
{ k k
k+1
k+1
(6.26)
Addition of the iterative corrector to the old values of the independent variables
yields the eight updates:
k+1 k k+1
= +
k+1 k k+1
k+1 k k+1
(6.27)
39
For starting the iteration, an initial solution is required. This solution is chosen to
be the elastic solution at the contact point with the yield surface given by:
0 = c = h + (1 ) D:
0 = h
(6.28)
s0 = s c
0 = 0
( Trial )
The trial stress state n+1
= D: and the elastic strain vector due to suction
where
PT = (In s 0n s , n c + 1 )
note that
0n s , n c + 1 is
(6.29)
nc
ns
rows and
n c + 1 columns, n s
40
Input/output variables:
iMod = 1 : Number model
nProps = 18 : Model parameters
ParamName (1): Poisson ratio ()
ParamName (2): Slope of the unload/reload line () of saturated soil
ParamName (3): Slope of the normal compression line (0) of saturated
soil
ParamName (4): Elastic stiffness due to suction (s)
ParamName (5): Parameter to control the tensile strength due to suction
(k)
ParamName (6): Slope of the Critical state line (M)
ParamName (7): Friction angle at CS ( cs [degrees])
ParamName (8): Initial void ratio (e0)
ParamName (9): Preconsolidation pressure of saturated soil (Po [kPa])
ParamName (10): Reference mean stress ( Pr [kPa])
ParamName (11): Parameter to control infinite suction (r)
ParamName (12): Parameter to control soil stiffness with suction ( [kPa1
])
ParamName (13): Van Genuchten Parameter (a [kPa])
ParamName (14): Van Genuchten Parameter (b)
ParamName (15): Van Genuchten Parameter (c)
ParamName (16): Parameter of non associated flow rule (g)
ParamName (17): Coefficient of earth pressure at rest (Ko)
ParamName (18): Overconsolidation ratio (OCR)
41
Description
Subroutine
101
102
103
Note: Internally the subroutine Unsat_Model groups some input data into a
derived data type named Prop_model which has the follow fields:
Prop_model%Imod
Prop_model%npar
Prop_model%n
Prop_model%nhv
Prop_model%par
= Imod
= Nprops
= Nstress
= Nstat
= Props()
Nstress is the number of stress components (6) plus the number of the hardening
parameters in order to build a vector of generalized stress.
42
Model
parameters
Pr [F/L2]
Definition
Reference
mean stress
at which
one may
reach the
saturated
virgin state,
starting at a
partially
saturated
condition,
through a
wetting
path which
involves
only
(elastic)
swelling.
Typical
values
When Pr = Po
(saturated
condition) the
LC yield curve
becomes a
straight line.
In this case,
changes in s
do not result in
plastic
deformations.
Reference
Alonso et
al (1990).
See Table 1:
Some values
from literature.
Suction, s
Influence of Pr on LC:
r* :
Constants
Pr = 0.2 * Po
Pr = 0.4 * Po
Pr = 0.6 * Po
Pr = 0.8 * Po
Pr = Po
Po
Parameter
defining
the
maximum
soil
stiffness
(for an
infinite
suction)
See Table 1:
Some values
from literature.
Alonso, et
al (1990).
Barrera, et
al (2002)
Definition
Typical
values
Reference
Influence of r on LC:
Suction, s
No
43
P r , :
Constants
r = 0.75
r = 0.80
r = 0.85
r = 0.90
r = 0.95
r=1
Po
[F/L2]-1
Parameter
controlling
the rate of
increase of
soil
stiffness
with
suction
12
P r , r*:
Constants
[MPa-1]
= 2.5
=5
= 10
= 15
= 20
Po
See Table 1:
Some values
from literature.
Alonso, et
al (1990)
Barrera, et
al (2002).
44
Tab. 6.2: Parameters for the Barcelona Basic Model for different soil types
Soil type
BCN Silt
0.3
3
pr
(0)
1.155
0.005
0.073
0.782
155
0.65
0.02
(MPa)
-1
(MPa)
Reference
7.0 E-
Barrera
05
(2002)
0.032:
p:100
Sion silt
0.4
0
1.33
0.007
400 kPa
0.047:
Geiser et al
(2001)
p>400
kPa
G=
Compacted
3.3
kaolin, ML
MP
0.821
0.015
0.14
0.26
16.4
0.043
0.011
0.065
0.75
20
0.01
Josa
(1988)
a
Compacted kaolin
[w = 6% ,
Sr= 0.627]
Karube
(1986)
Variab
le
Compacted kaolin
0.027
0.063
0.32
Thu, et al.
10
(2007)
0.58
0.972
Sandy Clay
(Lower Cromer
till)
Lower Cromer till
Lambeth Sand
(London City)
G=
7
1.2
0.0077
0.066
0.25
20
0.012
0.2
1.2
0.0077
0.066
0.35
16.4
0.012
0.2
0.9
0.005
0.06
0.25
20
MP
Sand
45
MP
Georgiadis
(2003)
Georgiadis
(2003)
Variab
le
1.54
0.0056
0.022
0.68
Rampino
24
0.001
0.91
et al
(2000)
0.978
Boom Clay
Pellets
0.015
0.16
0.57
0.05
0.080
0.90
0.50
0.11
0.29
18.1
FEBEX Bentonite
1.24
0.005
1.32
0.009
the Riverside
(1985)
G
Metramo silty
Maswoswe
0.041
Snchez
(2004)
Lloret et al
(2003)
Mun B-J
(2004)
Soil type
45
pr
(0)
0.008
0.25
0.85
0.05
0.0085
0.18
0.95
8.0
0.004
0.052
0.17
1.8
0.05
0.15
0.75
0.05
0.1
0.26
0.564
0.544
0.06
0.37
0.27
10
(MPa)
-1
(MPa)
Reference
Campus
Bentonite Sand
mixtures
Lixhe chalk
0.2
8.0E-
Alonso et
06
al (2005)
0.003
Collin et al
(2002)
Compacted silt.
(clayey, slightly
sandy Silt)
[w = 23.1%,
Vasallo et
al (2007)
d = 15.6 kN/m3]
Serrate bentonite
Boom Clay
0.4
0.3
33
1.5
1.0
Reconstituid
0.0026
5
0.014
Kaolin
Zhang et al
(2003)
Zhang et al
(2003)
Slatter et al
(2006)
Earth fill
Cordao
compacted with a
soil with a lower
0.005
0.085
and Farias
than optimum
(2006)
water content
Jossigny's Silt
(silt of low
0.3
1.0
0.015
0.108
0.911
5.75
0.895
0.14
0.29
0.24
19.69
plasticity)
6.55E-
Vaunat et
06
al (2000)
G=
Residual granite
8.9
soil (CH)
MP
a
0.045
Mofiz et al
(2005)
46
47
Fig. 7.1: finite element meshes used in 2D and 3D calculations; Left: 2D; Right
3D
48
49
Symbol
kx , ky , kz
einit
Kw,ref/n
Ssat
Sres
gn
ga
gl
Unit
[m/day]
[-]
[kN/m2]
[-]
[-]
[-]
[m-1]
[-]
Time (day)
Steady state
0.232
0.463
0.694
0.926
1.16
1.39
2.31
3.47
Steady state
Value
0.1521
0.5625
4.875105
1.0
0.06203
2.286
2.24
0
50
51
1E1
Time 0
Time 0.232
Time 0.463
Time 0.695
Time 0.926
Time 1.16
Time 1.39
Time 2.31
Time 3.47
Time 3.47
9.04E0
8.04E0
7.04E0
6.04E0
5.04E0
4.04E0
3.04E0
2.04E0
1.04E0
4.37E-2
0
1
[m]
Fig. A1.6:
1E2
9.83E1
Time 0
Time 0.232
Time 0.463
Time 0.695
Time 0.926
Time 1.16
Time 1.39
Time 2.31
Time 3.47
Time 3.47
9.63E1
9.43E1
9.23E1
9.03E1
8.83E1
8.63E1
8.43E1
8.23E1
8.03E1
7.83E1
7.63E1
Degree of saturation [%]
7.43E1
7.23E1
7.03E1
6.83E1
6.63E1
6.43E1
6.23E1
6.03E1
5.83E1
5.63E1
5.43E1
5.23E1
5.03E1
4.83E1
4.63E1
4.43E1
4.23E1
4.03E1
3.83E1
3.63E1
3.43E1
0
1
[m]
Fig. A1.7:
52
Summary:
As seen the results from PLAXIS 2D, PLAXIS 3D and PlaxFlow are similar.
53
54
Symbol
kx , ky , kz
einit
Kw,ref/n
Ssat
Sres
gn
ga
gl
Unit
[m/day]
[-]
[kN/m2]
[-]
[-]
[-]
[m-1]
[-]
Time (day)
Steady state
0.232
0.463
0.694
0.926
1.16
1.39
2.31
3.47
Steady state
Value
0.1521
0.5625
4.875105
1.0
0.06203
2.286
2.24
0
55
56
1E1
Time 0
Time 0.232
Time 0.463
Time 0.695
Time 0.926
Time 1.16
Time 1.39
Time 2.31
Time 3.47
9.04E0
8.04E0
7.04E0
6.04E0
5.04E0
4.04E0
3.04E0
2.04E0
1.04E0
4.35E-2
0
1
[m]
1E2
9.83E1
Time 0
Time 0.232
Time 0.463
Time 0.695
Time 0.926
Time 1.16
Time 1.39
Time 2.31
Time 3.47
9.63E1
9.43E1
9.23E1
9.03E1
8.83E1
8.63E1
8.43E1
8.23E1
8.03E1
7.83E1
7.63E1
Degree of saturation [%]
7.43E1
7.23E1
7.03E1
6.83E1
6.63E1
6.43E1
6.23E1
6.03E1
5.83E1
5.63E1
5.43E1
5.23E1
5.03E1
4.83E1
4.63E1
4.43E1
4.23E1
4.03E1
3.83E1
3.63E1
3.43E1
0
1
[m]
57
58
59
Symbol
kx , ky , kz
einit
Kw,ref/n
Ssat
Sres
gn
ga
gl
Unit
[m/day]
[-]
[kN/m2]
[-]
[-]
[-]
[m-1]
[-]
Time (day)
Steady state
0.00579
0.0116
0.0694
0.127
0.706
1.28
7.07
12.9
Steady state
Value
0.1521
0.5625
4.875105
1.0
0.06203
2.286
2.24
0
60
61
2.02E1
1.92E1
1.82E1
Time 0
Time 0.00579
Time 0.0116
Time 0.0694
Time 0.127
Time 0.706
Time 1.28
Time 7.07
Time 12.9
Time 12.9
1.72E1
1.62E1
1.52E1
1.42E1
1.32E1
1.22E1
1.12E1
1.02E1
9.15E0
8.15E0
7.15E0
6.15E0
5.15E0
4.15E0
3.15E0
2.15E0
1.15E0
1.53E-1
-8.47E-1
-1.85E0
-2.85E0
-3.85E0
-4.85E0
-5.85E0
-6.85E0
-7.85E0
-8.85E0
-9.85E0
-1.08E1
-1.18E1
-1.28E1
-1.38E1
-1.48E1
0
1
[m]
Fig. B1.6:
1E2
Time 0
Time 0.00579
Time 0.0116
Time 0.0694
Time 0.127
Time 0.706
Time 1.28
Time 7.07
Time 12.9
Time 12.9
9.54E1
9.04E1
8.54E1
8.04E1
7.54E1
7.04E1
6.54E1
6.04E1
5.54E1
5.04E1
4.54E1
4.04E1
3.54E1
3.04E1
2.54E1
2.04E1
1
[m]
62
Summary:
As seen the results from PLAXIS 2D, PLAXIS 3D and PlaxFlow are similar.
63
64
Symbol
kx , ky , kz
einit
Kw,ref/n
Ssat
Sres
gn
ga
gl
Unit
[m/day]
[-]
[kN/m2]
[-]
[-]
[-]
[m-1]
[-]
Value
0.1521
0.5625
4.875105
1.0
0.06203
2.286
2.24
0
Time (day)
Steady state
2
4
8
16
32
64
128
Steady state
65
66
2.01E1
1.91E1
Time 0
Time 2
Time 4
Time 8
Time 16
Time 32
Time 64
Time 128
Time 128
1.81E1
1.71E1
1.61E1
1.51E1
1.41E1
1.31E1
1.21E1
1.11E1
1.01E1
9.09E0
8.09E0
7.09E0
6.09E0
5.09E0
4.09E0
3.09E0
2.09E0
1.09E0
8.75E-2
0
1
[m]
Cross-section
1E2
Time 0
Time 2
Time 4
Time 8
Time 16
Time 32
Time 64
Time 128
Time 128
9.54E1
9.04E1
8.54E1
8.04E1
7.54E1
7.04E1
6.54E1
6.04E1
5.54E1
5.04E1
4.54E1
4.04E1
3.54E1
3.04E1
2.54E1
2.04E1
1
[m]
67
68
Symbol
kx , ky , kz
einit
Kw,ref/n
Ssat
Sres
gn
ga
gl
Unit
[m/day]
[-]
[kN/m2]
[-]
[-]
[-]
[m-1]
[-]
Value
0.1521
0.5625
10.84
1.0
0.06203
2.286
2.24
0
Time (day)
Steady state
2
4
8
16
32
64
128
Steady state
69
Cross-section
8.71E-2
-9.13E-1
Time 0
Time 2
Time 4
Time 8
Time 16
Time 32
Time 64
Time 128
Time 128
-1.91E0
-2.91E0
-3.91E0
-4.91E0
-5.91E0
-6.91E0
-7.91E0
-8.91E0
-9.91E0
-1.09E1
-1.19E1
-1.29E1
-1.39E1
-1.49E1
-1.59E1
-1.69E1
-1.79E1
-1.89E1
-1.99E1
0
1
[m]
70
71
Symbol
kx , ky , kz
einit
Kw,ref/n
Unit
[m/day]
[-]
[kN/m2]
Value
8.156
0.403
4.875105
kr
0.00E+00
1.0
1.0
2.00E-01
0.44448
0.94019
4.00E-01
2.91E-02
0.57286
6.00E-01
4.36E-03
0.35567
8.00E-01
1.12E-03
0.29439
1.00E+00
3.89E-04
0.27536
1.20E+00
1.64E-04
0.26821
1.40E+00
7.90E-05
0.26508
1.60E+00
4.19E-05
0.26354
1.80E+00
2.40E-05
0.26272
2.00E+00
1.46E-05
0.26224
2.20E+00
9.27E-06
0.26195
2.40E+00
6.14E-06
0.26177
2.60E+00
4.20E-06
0.26165
2.80E+00
2.95E-06
0.26157
3.00E+00
2.13E-06
0.26151
3.20E+00
1.57E-06
0.26147
3.40E+00
1.18E-06
0.26144
3.60E+00
8.98E-07
0.26141
3.80E+00
6.95E-07
0.2614
72
Time (day)
Steady state
188
0.00463
312
0.00926
436
0.0139
560
0.0185
684
0.0232
808
0.0278
932
0.0324
1259
0.0463
1393
Steady state
73
74
1E1
Time 0
Time 0.00463
Time 0.00926
Time 0.0139
Time 0.0185
Time 0.0232
Time 0.0278
Time 0.0324
Time 0.0463
Time 0.0463
9.04E0
8.04E0
7.04E0
6.04E0
5.04E0
4.04E0
3.04E0
2.04E0
1.04E0
4.37E-2
0
1
[m]
75
Cross-section
1E2
Time 0
Time 0.00463
Time 0.00926
Time 0.0139
Time 0.0185
Time 0.0232
Time 0.0278
Time 0.0324
Time 0.0463
Time 0.0463
9.53E1
9.03E1
8.53E1
8.03E1
7.53E1
7.03E1
6.53E1
6.03E1
5.53E1
5.03E1
4.53E1
4.03E1
3.53E1
3.03E1
0
1
[m]
76
77
Symbol
kx , ky , kz
einit
Kw,ref/n
Unit
[m/day]
[-]
[kN/m2]
Value
0.0106
0.98
4.875105
kr
0.00E+00
1.0
1.0
2.00E-01
0.3829
0.9263
4.00E-01
0.1539
0.8498
6.00E-01
8.15E-02
0.7935
8.00E-01
0.05064
0.7505
1.00E+00
0.03469
0.7164
1.20E+00
0.02536
0.6886
1.40E+00
0.01942
0.6653
1.60E+00
0.0154
0.6454
1.80E+00
0.01254
0.6283
2.00E+00
0.01043
0.6132
2.20E+00
0.008822
0.5999
2.40E+00
0.004573
0.588
2.60E+00
0.006579
0.5773
2.80E+00
0.005775
0.5675
3.00E+00
0.005114
0.5587
3.20E+00
0.004565
0.5505
3.40E+00
0.004102
0.543
3.60E+00
0.003709
0.5361
3.80E+00
0.003372
0.5297
78
Time (day)
Steady state
107
2.32
150
4.63
193
6.95
236
9.26
279
11.58
322
13.89
365
16.21
623
32.41
655
Steady state
79
80
1E1
Time 0
Time 2.31
Time 4.62
Time 6.93
Time 9.24
Time 11.6
Time 13.9
Time 16.2
Time 32.4
Time 32.4
9.04E0
8.04E0
7.04E0
6.04E0
5.04E0
4.04E0
3.04E0
2.04E0
1.04E0
4.37E-2
0
1
[m]
Fig. C2.8:
1E2
9.91E1
Time 0
Time 2.31
Time 4.62
Time 6.93
Time 9.24
Time 11.6
Time 13.9
Time 16.2
Time 32.4
Time 32.4
9.81E1
9.71E1
9.61E1
9.51E1
9.41E1
9.31E1
9.21E1
9.11E1
9.01E1
Degree of saturation [%]
8.91E1
8.81E1
8.71E1
8.61E1
8.51E1
8.41E1
8.31E1
8.21E1
8.11E1
8.01E1
7.91E1
7.81E1
7.71E1
7.61E1
7.51E1
7.41E1
7.31E1
7.21E1
0
1
[m]
81
82
Sand
Sand
Clay
Clay
83
Time (day)
14
Steady state
40
0.00463
58
0.00926
76
0.0139
94
0.0185
107
0.0232
141
1.07
177
2.22
240
4.54
261
Steady state
84
85
86
1.01E1
Time 0
Time 0.00463
Time 0.00926
Time 0.0139
Time 0.0185
Time 0.0232
Time 1.06
Time 2.22
Time 4.54
Time 4.54
9.09E0
8.09E0
7.09E0
6.09E0
5.09E0
4.09E0
3.09E0
2.09E0
1.09E0
8.73E-2
-9.13E-1
-1.91E0
-2.91E0
-3.91E0
-4.91E0
-5.91E0
-6.91E0
-7.91E0
-8.91E0
-9.91E0
0
1
[m]
Fig. C3.8:
1E2
Time 0
Time 0.00463
Time 0.00926
Time 0.0139
Time 0.0185
Time 0.0232
Time 1.06
Time 2.22
Time 4.54
Time 4.54
9.53E1
9.03E1
8.53E1
8.03E1
7.53E1
7.03E1
6.53E1
6.03E1
5.53E1
5.03E1
4.53E1
4.03E1
3.53E1
3.03E1
0
1
[m]
87
88
Clay
Clay
Sand
Sand
89
Time (day)
Steady state
174
1.157
207
2.314
240
3.471
273
4.628
306
5.785
339
6.942
372
8.099
439
11.571
476
Steady state
90
91
1.58E1
Time 0
Time 1.16
Time 2.31
Time 3.47
Time 4.63
Time 5.79
Time 6.94
Time 8.1
Time 11.6
Time 11.6
1.48E1
1.38E1
1.28E1
1.18E1
1.08E1
9.83E0
8.83E0
7.83E0
6.83E0
5.83E0
4.83E0
3.83E0
2.83E0
1.83E0
8.33E-1
1
[m]
Fig. C4.8:
1E2
Time 0
Time 1.16
Time 2.31
Time 3.47
Time 4.63
Time 5.79
Time 6.94
Time 8.1
Time 11.6
Time 11.6
9.53E1
9.03E1
8.53E1
8.03E1
7.53E1
7.03E1
6.53E1
6.03E1
5.53E1
5.03E1
4.53E1
4.03E1
3.53E1
3.03E1
1
[m]
92
Summary:
As seen the results from PLAXIS 2D, PLAXIS 3D and PlaxFlow are similar.
93
94
Symbol
kx , ky , kz
einit
Kw,ref/n
Ssat
Sres
gn
ga
gl
Unit
[m/day]
[-]
[kN/m2]
[-]
[-]
[-]
[m-1]
[-]
Value
0.1521
0.5625
98.39
1.0
0.06203
2.286
2.24
0
Time (day)
Steady state
0.579
1.158
1.737
2.316
2.895
3.474
4.053
4.632
5.211
5.790
Fig. D1.4: Pore pressure at the Gauss point at the top of the column in time in
PLAXIS 2D
95
96
Fig. D1.7: Pore pressure at the Gauss point at the top of the column in time in
PLAXIS 3D
97
Cross-section
2.02E1
1.82E1
Time 0
Time 0.579
Time 1.16
Time 2.31
Time 2.9
Time 3.48
Time 4.06
Time 4.64
Time 5.22
Time 5.79
1.62E1
1.42E1
1.22E1
1.02E1
8.22E0
6.22E0
4.22E0
2.22E0
2.18E-1
-1.78E0
-3.78E0
-5.78E0
-7.78E0
-9.78E0
-1.18E1
-1.38E1
-1.58E1
-1.78E1
-1.98E1
-2.18E1
-2.38E1
-2.58E1
-2.78E1
-2.98E1
0
1
[m]
1E2
Time 0
Time 0.579
Time 1.16
Time 2.31
Time 2.9
Time 3.48
Time 4.06
Time 4.64
Time 5.22
Time 5.79
9.54E1
9.04E1
8.54E1
8.04E1
7.54E1
7.04E1
6.54E1
6.04E1
5.54E1
5.04E1
4.54E1
4.04E1
3.54E1
3.04E1
2.54E1
2.04E1
1
[m]
Fig. D1.9:
20.072
Point K
19.072
18.072
17.072
16.072
15.072
14.072
13.072
Pressure [kN/m^2]
12.072
11.072
10.072
9.072
8.072
7.072
6.072
5.072
4.072
3.072
2.072
1.072
.072
0
[day]
Fig. D1.10: Pore pressure at the Gauss point at the top of the column in time
(existing PlaxFlow)
98
Summary:
As seen the results from PLAXIS 2D, PLAXIS 3D and PlaxFlow are similar.
99
100
Symbol
kx , ky , kz
einit
Kw,ref/n
Ssat
Sres
gn
ga
gl
Unit
[m/day]
[-]
[kN/m2]
[-]
[-]
[-]
[m-1]
[-]
Time (day)
Steady state
0.579
1.158
1.737
2.316
2.895
3.474
4.053
4.632
5.211
5.790
Value
0.1521
0.5625
98.39
1.0
0.06203
2.286
2.24
0
Fig. D2.4: Pore pressure the top of the column in time in PLAXIS 2D
101
102
Fig. D2.7: Pore pressure the top of the column in time in PLAXIS 3D
103
Summary:
For this case, the existing PlaxFlow does not generate proper results and its
results are different from the results of PlaxFlow verification report. However, it
seems that PLAXIS 2D and PLAXIS 3D are capable of simulating this example
and provide similar results to the above-mentioned report. In the case of PLAXIS
3D results at 4.632 day and 5.211 day are not shown because the harmonic phase
is simulated in one phase.
104
105
Q (m/day)
0.0432
0.2315
0.0432
0.2315
0.0864
0.463
0.0864
0.463
0.1296
0.6944
0.1296
0.6944
0.1728
0.9259
0.1728
0.9259
0.216
1.1574
0.216
1.1574
0.1728
2.3148
0.1728
2.3148
0.1296
3.4722
0.1296
0.25
Q (m/Day)
0.2
0.15
0.1
0.05
0
0
0.5
1.5
2.5
3.5
Time (day)
Symbol
kx , ky , kz
Unit
[m/day]
Value
0.1521
einit
[-]
0.5625
Elastic storage
Kw,ref/n
[kN/m2]
487.5e3
Saturated saturation
Ssat
[-]
1.0
Residual saturation
Sres
[-]
0.06203
Van Genuchten
gn
[-]
2.286
Van Genuchten
ga
-1
[m ]
2.24
Van Genuchten
gl
[-]
106
Time (day)
Steady state
0.23
0.465
0.695
0.925
1.16
1.39
2.31
3.45
3.494
107
Fig. D3.5: Pore pressure at node and Gauss point at the top of the column in time
in PLAXIS 2D
108
Fig. D3.8: Pore pressure at node and Gauss point at the top of the column in time
in PLAXIS 3D
Cross-section
1E1
Time 0
Time 0.232
Time 0.463
Time 0.695
Time 0.926
Time 1.16
Time 1.39
Time 2.31
Time 3.47
9.05E0
8.05E0
7.05E0
6.05E0
5.05E0
4.05E0
3.05E0
2.05E0
1.05E0
4.81E-2
-9.52E-1
0
1
[m]
Fig. D3.9:
109
Cross-section
1E2
9.83E1
Time 0
Time 0.232
Time 0.463
Time 0.695
Time 0.926
Time 1.16
Time 1.39
Time 2.31
Time 3.47
9.63E1
9.43E1
9.23E1
9.03E1
8.83E1
8.63E1
8.43E1
8.23E1
8.03E1
7.83E1
7.63E1
Degree of saturation [%]
7.43E1
7.23E1
7.03E1
6.83E1
6.63E1
6.43E1
6.23E1
6.03E1
5.83E1
5.63E1
5.43E1
5.23E1
5.03E1
4.83E1
4.63E1
4.43E1
4.23E1
4.03E1
3.83E1
3.63E1
3.43E1
0
1
[m]
9.597
Point K
8.597
7.597
6.597
Pressure [kN/m^2]
5.597
4.597
3.597
2.597
1.597
.597
-.403
[day]
Fig. D3.11: Pore pressure at the Gauss point at the top of the column in time
(existing PlaxFlow)
Summary:
As seen the results from PLAXIS 2D, PLAXIS 3D and PlaxFlow are similar.
110
111
Q (m/day)
-0.0432
0.6944
-0.0432
0.6944
-0.00043
1.1574
-0.00043
-0.05
Q (m/day)
-0.04
-0.03
-0.02
-0.01
0
0
0.2
0.4
0.6
0.8
1.2
1.4
Time (day)
Unit
[m/day]
[-]
[kN/m2]
[-]
[-]
[-]
[m-1]
[-]
Time (day)
Steady state
0.23
0.465
0.695
0.925
1.16
Value
0.1521
0.5625
487.5e3
1.0
0.06203
2.286
2.24
0
112
Fig. D4.5: Pore pressure at node and Gauss point at the top of the column in time
in PLAXIS 2D
113
Fig. D4.8: Pore pressure at node and Gauss point at the top of the column in time
in PLAXIS 3D
114
1E2
Time 0
Time 0.232
Time 0.463
Time 0.695
Time 0.926
Time 1.16
9.54E1
9.04E1
8.54E1
8.04E1
7.54E1
7.04E1
6.54E1
6.04E1
5.54E1
5.04E1
4.54E1
4.04E1
3.54E1
3.04E1
2.54E1
2.04E1
1.54E1
1.04E1
0
1
[m]
Fig. D4.9:
3.41E1
3.31E1
Time 0
Time 0.232
Time 0.463
Time 0.695
Time 0.926
Time 1.16
3.21E1
3.11E1
3.01E1
2.91E1
2.81E1
2.71E1
2.61E1
2.51E1
2.41E1
2.31E1
2.21E1
2.11E1
2.01E1
1.91E1
1.81E1
1.71E1
1.61E1
1.51E1
1.41E1
1.31E1
1.21E1
1.11E1
1.01E1
9.06E0
8.06E0
7.06E0
6.06E0
5.06E0
4.06E0
0
1
[m]
99.635
Point K
94.635
89.635
84.635
79.635
74.635
69.635
64.635
Pressure [kN/m^2]
59.635
54.635
49.635
44.635
39.635
34.635
29.635
24.635
19.635
14.635
9.635
4.635
-.365
0
1
[day]
Fig. D4.8: Pore pressure at the Gauss point at the top of the column in time
(existing PlaxFlow)
115
Summary:
The example given in this section is not a practical problem. In this example, it is
intended to extract water from a very dry zone. By imposing a prescribed outflow
boundary conditions, soil in the near of the boundary gets more unsaturated and
the relative permeability becomes smaller and consequently extracting water
becomes more difficult.
116
7.14 Conclusions
Results of 13 one dimensional flow problems, solved by the new PLAXIS 2D
and 3D kernels, are shown in this chapter and most of them are compared with
PlaxFlow results. As seen, results are in agreement in most cases with the results
from PlaxFlow kernel.
For all examples very fine meshes with 15 noded elements in 2D and 10-noded
tetrahedral elements in 3D are used.
The following features have been tested:
1. Hydraulic models. Van Genuchten and spline models have been tested.
It has been found that the predefined parameters of linearised Van
Genuchten relationship do not produce similar results to Van
Genuchten relationship. Therefore, it is suggested to use Van
Genuchten model and not the approximated one.
2. Different material set. The capability of the code to simulate hydraulic
behaviour of soils has been shown for two types of material, namely
sands and clay. It has been found that the models usually need finer
mesh for sands compared to clays, as variation of relative permeability
is higher for sands.
3. Boundary conditions. Almost all boundary conditions have been tested
in this chapter, namely prescribed boundary head, seepage boundary
condition, varying head (in PLAXIS 2D and 3D codes, this is a part of
seepage boundary condition), inflow, outflow, precipitation and
evaporation. The rest boundary conditions, namely wells and drains
are tested in chapter 10 where groundwater flow analyses are verified
against two and three dimensional problems.
4. Automatic time stepping. PLAXIS 2D and 3D kernels use an automatic
time stepping. The kernel calculates the first (critical) time, maximum
and minimum time steps in the beginning of calculation based on
elements size and material parameters.
5. Steady state and transient calculation. Both steady state and transient
types of calculations have been tested.
6. Permeability contrasts. Results of two cases, namely low-high and
high-low permeability contrasts have been shown.
117
118
15
precipitation
precipitation
0.75
0.50
0.75
loam
sand
Symbol
kx , ky , kz
einit
Kw,ref/n
gn
ga
gl
Unit
[m/day]
[-]
[kN/m2]
[-]
[m-1]
[-]
Value
0.0154
0.754
4.875105
1.325
0.650
-2.161
119
Symbol
kx , ky , kz
einit
Kw,ref/n
gn
ga
gl
Unit
[m/day]
[-]
[kN/m2]
[-]
[m-1]
[-]
Value
0.1270
0.62
4.875105
1.951
2.13
0.168
The precipitation fluxes are given in Table G1.3. The threshold values for
ponding and precipitation (evapotranspiration) are chosen as 1 m and 0 at the top
of the boundary, respectively.
Tab. G1.3: Prescribed flux (min = 0 and max= 1 m)
Time (day)
Q (m/day)
0.01
0.01
0.03
0.03
0.03
0.03
0.01
0.01
0.01
120
0.035
0.03
Q (m/day)
0.025
0.02
0.015
0.01
0.005
0
0
10
Time (day)
121
Fig. G1.4.2D: Active pore pressure after steady state calculation (PLAXIS 2D)
Fig. G1.5.2D: Degree of saturation after steady state calculation (PLAXIS 2D)
Fig. G1.6.2D: Active pore pressure after 4.5 days (PLAXIS 2D)
122
123
Fig. G1.4.3D: Active pore pressure after steady state calculation (PLAXIS 3D)
Fig. G1.5.3D: Degree of saturation after steady state calculation (PLAXIS 3D)
124
Fig. G1.6.3D: Active pore pressure after 4.5 days (PLAXIS 3D)
125
126
Fig. G2.1: Triangular earth dam with a constant head on the left side and seepage
on the right side
In this case the entire slope CB is a seepage surface and no free surface exist,
leading to a confined groundwater flow. Such a geometry and boundary
conditions allow for solving the problem analytically. The solution of the
problem was first introduced by Davison (see Harr, 1962). Here, the total
discharge calculated by the analytical solution will be compared with that
computed by PLAXIS. The total flow through the dam can be calculated as
follows:
Boundary conditions:
Side AC : Constant pressure head (h = constant)
Side AB : No normal flow, specific discharge qy = 0
Side CB : Seepage h = y
A(x,y) = (0,0), B(x,y) = (2L,0), C(x,y) = (L,L)
In ns-system:
h=L
ns
2L
q n = k
dh
s
=k
dn
2L
q s = k
dh
n
=k
ds
2L
127
n = ( x y) / 2
s = ( x + y) / 2
In xy-system:
h=L
x2 y2
4L
q x = k
dh
x
=k
dx
2L
q y = k
y
dh
=k
dy
2L
dq x dq y dq n dq s
+
=
+
=0
dx
dy
dn
ds
k
2
k
2
total discharge Q x = L
Line AC: q n = k
s
2L
total discharge Qn =
Line BC: q s = k
1 kL 2
k
L 2= L
2 2L
2
n
2L
total discharge Qs =
1 kL 2
k
L 2= L
2 2L
2
To compare results obtained from the analytical solution with that obtained from
PLAXIS, a triangular dam, 2 m wide and 1 m high is considered. The head at the
left side is assumed to be 1 m and the coefficient of permeability is 1 m/day. The
geometry of the problem and the finite element mesh are presented in Figure
G2.2. The width of the model is 1 m in the 3D calculations.
Following the analytical solution, the total discharge through the dam at sides CD
and BC is:
Q = 0.50 m3/day/m
128
129
130
131
Analytical
discharge
error
0.5000
0
PLAXIS
2D
(A-A)
PLAXIS
2D
(B-B)
0.5000
0%
0.5010
0.2 %
PLAXIS
3D
PLAXIS
3D
(A-A)
0.5001
0.02 %
(B-B)
0.4982
0.36 %
Summary:
Apparently the results of both analytical and PLAXIS calculations are similar. It
should be noted that some error may occur during numerical integration in output
program.
132
133
134
135
136
discharge
error
0.800
0
PlaxFlow
PLAXIS
2D
15 noded
PLAXIS
2D
6 noded
PLAXIS
3D
10 noded
0.818
2.25 %
0.815
1.87 %
0.822
2.75 %
0.819
2.32 %
Summary:
PLAXIS 2D with 15 noded elements produces the most accurate result. However,
the accuracy can be improved by refining the mesh.
137
138
12 22
2L
where k is the permeability, L is the length of the layer and 1 and 2 are the
ground water head at the left and right boundary, respectively. For the current
situation this results in a theoretical solution of 0.150 m3/day/m.
139
140
141
0.152
1.3 %
PLAXIS 2D
15 noded
PLAXIS 2D
6 noded
PLAXIS 3D
10 noded
0.1497
0.2 %
0.1546
3.0 %
0.1531
2.06 %
Summary:
Since the total discharge is calculated in the output program which uses linear
interpolation, the accuracy strongly depends on the size of elements. Therefore
the results can be improved by refining the mesh.
142
143
144
145
146
25.5 m
0
PLAXIS 2D
15 noded
25.73 m
0.9 %
PLAXIS 3D
10 noded
25.74 m
0.94 %
Summary:
PLAXIS 2D with 15 noded elements produces the most accurate result.
However, the accuracy can be improved by refining the mesh.
147
3.22 m
hupstream
s
0.48 m
hdownstream
1.62 m
148
Fig. G6.2: Monograph for Muskat problem (KangKun Lee and Darrell I. Leap,
1997)
149
150
Phreatic line
151
1.54 m
0
PlaxFlow
1.68 m
9.1 %
PLAXIS 2D
PLAXIS 3D
15 noded
10 noded
1.63 m
4.6 %
1.69 m
9.7 %
Summary:
As seen the deviation from the analytical solution is pretty high in all cases.
Calculation of seepage face is usually affected by the element size of seepage
boundaries. In this example, where the seepage boundary is vertical, the effect of
the element size is higher.
152
153
154
Fig. G7.3.2D.1: Total discharge leaving the domain; Q=1.0 m3/day/m (PLAXIS
2D plane strain)
Fig. G7.3.2D.2: Total discharge leaving the domain; Q=1.0 m3/day/rad (PLAXIS
2D axi-symmetric)
155
Fig. G7.3.3D: Total discharge leaving the domain; Q=1.0 m3/day (PLAXIS 3D)
Summary:
Figure G7.2 and Figure G7.3 show flow filed and total discharge after activation
of the well. As seen the total discharge is exactly the same as total discharge
imposed in the well.
156
Fig. G8.1.2D:
Fig. G8.1.3D:
157
158
159
160
12 22
2L
where k is the permeability, L is the distance between the left boundary (or the
right boundary) and the drain and 1 and 2 are the ground water head at the left
(or right) and at the drain boundary, respectively. For the current situation this
161
4.48
8.96
PLAXIS 2D
15 noded
4.473 (0.16 % error)
8.975 (0.17 % error)
PLAXIS 3D
10 noded
4.476 (0.09 % error)
8.966 (0.07 % error)
Summary:
It can be seen that both PLAXIS 2D and PLAXIS 3D are capable of simulating
drains with a given head. The errors in the total discharge leaving the domain are
quite low in both 2D and 3D simulations.
162
163
164
Fig. G9.4.2D6: Total discharge at the middle of the earth dam for 6-noded
elements mesh (Q = 4.787 m3/day/m)
Fig. G9.4.2D15: Total discharge at the middle of the earth dam for 15-noded
elements mesh (Q = 4.749 m3/day/m)
165
166
Fig. G9.4.3D: Total discharge at the middle of the earth dam; Q = 4.773
m3/day/m (PLAXIS 3D)
167
Fig. G9.5.3D: Flow field and calculated phreatic surface (PLAXIS 3D)
Tab. G9.1: Total discharge
Analytical
Total discharge
error
4.708
m /day/m
0
3
PLAXIS
2D
6 noded
4.787
3
m /day/m
1.68 %
PLAXIS
2D
15 noded
4.749
3
m /day/m
0.87 %
PLAXIS
3D
10 noded
4.773
3
m /day/m
1.38 %
Summary:
Apparently the results of both analytical and PLAXIS calculations are similar.
168
2.5 m
K1
K2=10K1
10 m
2m
169
Ssat
Sres
ga
-1
gl
gn
[-]
[-]
[m ]
[-]
[-]
1.0
1.0
1.0
0.02
0.02
0.02
2.240
3.830
14.500
0.000
1.250
0.500
2.286
1.3774
2.680
Figure G10.2 shows a comparison of the calculated phreatic levels for the
problem from different codes (Bardet & Tobita, 2002).
The location of phreatic level calculated with PLAXIS 2D is shown in Figure
G10.3 and G10.4 for different types of material. As seen the shape of the phreatic
line (or surface) is affected by the type of material used for the soils. The phreatic
levels in Figure G10.3b and G10.3d are similar to the results of Bradet & Tobita
(2002) and Oden and Kikuchi (1980) while by using Staring sand (O1) material,
PLAXIS 2D provides results similar to the results of Borja & Kishnani (1991).
170
(a)
(b)
(c)
(d)
Fig. G10.3.2D: Active pore pressures calculated with PLAXIS 2D. In all cases
line L is the phreatic line; (a) fully saturated; (b) Hypres coarse top
soil; (c) Staring sand (O1); (d) USDA sand
171
(a)
(b)
(c)
(d)
Fig. G10.3.3D: Active pore pressures calculated with PLAXIS 3D; (a) fully
saturated; (b) Hypres coarse top soil; (c) Staring sand (O1); (d) USDA
sand
172
(a)
(b)
(c)
(d)
Fig. G10.4.3D: Phreatic surface calculated with PLAXIS 3D; (a) fully saturated;
(b) Hypres coarse top soil; (c) Staring sand (O1); (d) USDA sand
173
8.11 Conclusions
Results of 10 two and three dimensional flow problems, solved by the new
PLAXIS 2D and 3D kernels, are shown in this chapter and most of them are
compared with analytical solutions or PlaxFlow results.
The following features have been tested:
1. Boundary conditions. Almost all boundary conditions have been tested
in the chapter of one dimensional groundwater flow. However,
seepage boundary condition, inflow, precipitation and well are tested
here.
2. Steady state and transient calculation. Both steady state and transient
types of calculations have been tested.
3. Confined and unconfined groundwater flow calculations. Confined
groundwater flow calculations are linear and unconfined groundwater
flow calculations are usually nonlinear. It has been shown that
PLAXIS is capable of calculating both types of calculations.
4. Seepage face calculation. Calculation of seepage face is highly
nonlinear and needs additional procedures. In the case of higher order
elements, as used in PLAXIS, the problem is more difficult due to
irregular distribution of discharge at nodes. As shown in several
problems, PLAXIS is capable of calculating seepage face with a
reasonable accuracy.
5. Drain. The kind of boundary condition has been tested and validated
with PLAXIS 2D and 3D.
6. Element types. Both types of elements, namely 15 noded with a fourth
order of integration for pore pressure and 6-noded with a second order
of integration for pore pressure can be used for groundwater flow
calculation.
174
175
176
Verfication:
Bishop stress reads:
= + m (S p w )
(1.12)
As the degree of saturation is constant in entire the model and is equal to 0.3398,
Bishop effective stresses at the top and bottom are:
177
bottom
= bottom S p w = 40 0.3398 10 = 43.398 kPa
Figure CA1.2 shows the effective stresses calculated by PLAXIS 2D and 3D. As
seen the maximum and minimum effective stresses are -3.398 kPa and -43.40
kPa which are very close to the analytical solution.
Tab. CA1.1: Input data (Linear elastic model)
Description
Elastic modulus
Poissons ratio
initial void ratio
Saturated saturation
Residual saturation
Van Genuchten
Van Genuchten
Van Genuchten
Water weight
Soil weight (sat)
Soil weight (dry)
Symbol
Eref
einit
Ssat
Sres
gn
ga
gl
w
sat
unsat
Unit
[kN/m2]
[-]
[-]
[-]
[-]
[-]
[m-1]
[-]
[kN/m3]
[kN/m3]
[kN/m3]
Value
10000
0.2
0.5625
1.0
0.02
2.286
2.24
0
10.0
20.0
20.0
178
xx =
which provides -0.8495 kPa for the top and -10.85 for the bottom.
179
180
181
Summary:
As seen the results from PLAXIS 2D and 3D kernels and the output (total
stresses) are correct.
182
183
kEoed
(9.1)
; Eoed =
(1 v) E
; z = h y
(1 + v)(1 2v)
(9.2)
cv t
p
4 (1) j 1
y
2
( z, t ) =
cos (2 j 1)
exp
(2
j
1)
(9.3)
p0
2H
4 H2
j =1 2 j 1
T=
cv t
H2
(9.4)
-5
t= 0.5 day
-10
t= 50 day
t= 0.1 day
t= 10 day
-15
t= 5.0 day
EPP analysis
TPP analysis
t= 2.0 day
Analytical
t= 1.0 day
-20
0
0.25
0.5
distance (m)
0.75
184
1m
Symbol
E
sat
unsat
kx , ky , kz
einit
Kw,ref/n
Unit
[kN/m2]
[-]
[kN/m3]
[kN/m3]
[m/day]
[-]
[kN/m2]
Value
10000
0
10
0
0.001
0.5
4.95105
185
0 + ( 1 0 )t / t 1 if
c =
1
if
t < t1
t t1
(9.5)
-60
load (kPa)
-50
-40
-30
-20
-10
0
0
time (day)
t 1 = 1 day
0 = 0 kPa
1
= -50 kPa
top boundary open
Case 1b:
t 1 = 1 day
0 = 0 kPa
1 = -50 kPa
This example has also been calculated with PLAXIS 3D Foundation by John Van
Esch. Here the results of all calculations are given.
186
0
-2
-4
0 day
0.49 day
1.00 day
1.30 day
1.52 day
1.91 day
2.60 day
3.10 day
3.76 day
5.00 day
-6
-8
-10
-12
-14
-16
0.2
0.4
0.6
0.8
0.001
0 day
0.49 day
1.00 day
1.30 day
1.52 day
1.91 day
2.60 day
3.10 day
3.76 day
5.00 day
0.002
0.003
0.004
0.005
0
0.2
0.4
0.6
0.8
Height (m)
187
188
-2
t=0.0 day
t=0.5 day
t=1.0 day
t=1.5 day
t=2.0 day
t=2.5 day
t=3.0 day
t=4.0 day
t=5.0 day
-4
-6
-8
-10
-12
-14
-16
0
0.2
0.4
0.6
0.8
Uy (m)
-0.001
t=0.0 day
t=0.5 day
t=1.0 day
t=1.5 day
t=2.0 day
t=2.5 day
t=3.0 day
t=4.0 day
t=5.0 day
-0.002
-0.003
-0.004
-0.005
0
0.2
0.4
0.6
0.8
189
-5
0 day
0.66 day
1.00 day
1.27 day
1.42 day
1.99 day
2.75 day
5.00 day
-10
-15
-20
-25
0.2
0.4
0.6
0.8
0.001
0 day
0.66 day
1.0 day
1.27 day
1.99 day
2.75 day
5.00 day
0.002
0.003
0.004
0.005
0
0.2
0.4
0.6
0.8
190
191
-5
t=0.0 day
t=0.5 day
t=1.0 day
t=1.5 day
t=2.0 day
t=2.5 day
t=3.0 day
t=4.0 day
t=5.0 day
-10
-15
-20
-25
-30
0
0.2
0.4
0.6
0.8
Uy (m)
-0.001
t=0.0 day
t=0.5 day
t=1.0 day
t=1.5 day
t=2.0 day
t=2.5 day
t=3.0 day
t=4.0 day
t=5.0 day
-0.002
-0.003
-0.004
-0.005
0
0.2
0.4
0.6
0.8
192
1m
Symbol
E
sat
unsat
kx , ky , kz
einit
Kw,ref/n
Unit
[kN/m2]
[-]
[kN/m3]
[kN/m3]
[m/day]
[-]
[kN/m2]
Value
10000
0
10
0
0.001
0.5
4.95105
p 0 + p1 p 0 t / t 1
pc =
p1
193
if
if
t < t1
t t1
(9.6)
7
6
5
4
3
2
1
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
time (day)
0
pbottom
= 12.0 kPa
1
ptop
= 52.0 kPa
1
pbottom
= 62.0 kPa
0
pbottom
= 12.0 kPa
194
This example has also been calculated with PLAXIS 3D by John Van Esch. Here
the results of both calculations are given.
Results of case 2a:
0
-10
t=0.000 day
t=0.100 day
t=0.105 day
t=0.110 day
t=0.150 day
t=0.200 day
t=0.300 day
t=0.500 day
t=1.000 day
-20
-30
-40
-50
-60
-70
0
0.2
0.4
0.6
0.8
-0.00001
t=0.000 day
t=0.100 day
t=0.105 day
t=0.110 day
t=0.150 day
t=0.200 day
t=0.300 day
t=0.500 day
t=1.000 day
Uy (m)
-0.00002
-0.00003
-0.00004
-0.00005
-0.00006
0
0.2
0.4
0.6
0.8
195
196
-10
t=0.000 day
t=0.100 day
t=0.105 day
t=0.110 day
t=0.150 day
t=0.200 day
t=0.300 day
t=0.500 day
t=1.000 day
-20
-30
-40
-50
-60
-70
0
0.2
0.4
0.6
0.8
t=0.000 day
t=0.100 day
t=0.105 day
t=0.110 day
t=0.150 day
t=0.200 day
t=0.300 day
t=0.500 day
t=1.000 day
Uy (m)
-0.00003
-0.00004
-0.00005
-0.00006
-0.00007
-0.00008
-0.00009
0
0.2
0.4
0.6
0.8
197
-10
t=0.000 day
t=0.100 day
t=0.105 day
t=0.110 day
t=0.150 day
t=0.200 day
t=0.300 day
t=0.500 day
t=1.000 day
-20
-30
-40
-50
-60
0
0.2
0.4
0.6
0.8
-0.0005
t=0.000 day
t=0.100 day
t=0.105 day
t=0.110 day
t=0.150 day
t=0.200 day
t=0.300 day
t=0.500 day
t=1.000 day
Uy (m)
-0.001
-0.0015
-0.002
-0.0025
-0.003
0
0.2
0.4
0.6
0.8
198
199
-10
t=0.000 day
t=0.100 day
t=0.105 day
t=0.110 day
t=0.150 day
t=0.200 day
t=0.300 day
t=0.500 day
t=1.000 day
-20
-30
-40
-50
-60
0
0.2
0.4
0.6
0.8
-0.0005
t=0.000 day
t=0.100 day
t=0.105 day
t=0.110 day
t=0.150 day
t=0.200 day
t=0.300 day
t=0.500 day
t=1.000 day
Uy (m)
-0.001
-0.0015
-0.002
-0.0025
-0.003
0
0.2
0.4
0.6
0.8
Summary:
As seen the results from PLAXIS 2D and 3D kernels are similar. The differences
are due to different time steps used during the calculations.
200
15
precipitation
precipitation
0.75
0.50
loam
0.75
sand
Symbol
kx , ky , kz
einit
Kw,ref/n
Unit
[m/day]
[-]
[kN/m2]
Value
0.0154
0.754
4.875105
Saturation (saturated)
Ssat
[-]
1.0
Saturation (residual)
Sres
[-]
0.06831
Van Genuchten
gn
[-]
1.325
Van Genuchten
Van Genuchten
Youngs modulus
Poissons ratio
Soil weight (sat)
Soil weight (dry)
ga
gl
E
[m-1]
[-]
[kN/m2]
[-]
[kN/m3]
[kN/m3]
0.650
-2.161
20000
0.2
19
17
sat
unsat
201
Symbol
kx , ky , kz
einit
Kw,ref/n
Unit
[m/day]
[-]
[kN/m2]
Value
0.1270
0.62
4.875105
Saturation (saturated)
Ssat
[-]
1.0
Saturation (residual)
Sres
[-]
0.06203
Van Genuchten
gn
[-]
1.951
ga
gl
E
-1
2.13
0.168
13000
0.2
19
17
Van Genuchten
Van Genuchten
Youngs modulus
Poissons ratio
Soil weight (sat)
Soil weight (dry)
[m ]
[-]
[kN/m2]
[-]
[kN/m3]
[kN/m3]
sat
unsat
The precipitation fluxes are given in Table CA5.3. The threshold values for
ponding and precipitation (evapotranspiration) are chosen as 1 m and 0 at the top
of the boundary, respectively.
Tab. CA5.3: Prescribed flux (min = 0 and max = 1 m)
Time (day)
Q (m/day)
0.01
0.01
0.03
0.03
0.03
0.03
0.01
0.01
202
Q (m/day)
0.025
0.02
0.015
0.01
0.005
0
0
10
Time (day)
203
Fig. CA5.4.2D: Active pore pressure after steady state phase (imposing head)
204
Fig. CA5.5.2D: Degree of saturation after steady state phase (imposing head)
Fig. CA5.6.2D: Flow field after steady state phase (imposing head)
Fig. CA5.7.2D: Active pore pressure and external water load after 5 days
205
206
207
208
209
Fig. CA5.4.3D: Active pore pressure after steady state phase (imposing head)
210
Fig. CA5.5.3D: Degree of saturation after steady state phase (imposing head)
Fig. CA5.6.3D: Flow field after steady state phase (imposing head)
Fig. CA5.7.3D: Active pore pressure and external water load after 5 days
211
212
213
214
215
77.5 m
25-5
m
5m
90 m
30
m
40 m
The sub-soil consists of overconsolidated silty sand. The data of the dam
materials and the sub-soil are given in Table CA6.1
Geometry Model:
The situation can be modelled with a geometry model in which the sub-soil is
modelled to a depth of 30 m. The left hand boundary can be taken 50 m left of
the dam toe and the right hand boundary can be taken 37.5 m right of the other
dam toe. The proposed geometry model is presented in Figure CA6.2 for both 2D
and 3D calculations. For 3D calculations the width of the geometry is taken 50
m.
216
Fig. CA6.2.2D: Geometry and finite element mesh of the dam and sub-soil
(PLAXIS 2D 6 noded elements)
Fig. CA6.2.3D: Geometry and finite element mesh of the dam and sub-soil
(PLAXIS 3D)
The calculation consists of nine phases. In the first phase the dam is constructed.
In the second phase the reservoir is filled up, bringing the water level to the
standard level of 25m. For this situation the water pressure distribution is
calculated using a steady-state groundwater flow calculation. The third and
fourth phase both start from this standard situation (i.e. a dam with a reservoir
level at 25 m) and the water level is lowered to 5 m. A distinction is made in the
time interval at which this is done (i.e. different speeds of water level reduction;
rapid drawdown and slow drawdown). In both cases the water pressure
distribution is calculated using a coupled flow deformation calculation. The
fifth calculation phase also starts from the second phase and considers the longterm behaviour of the dam at the low reservoir level of 5 m, which involves a
steady-state groundwater flow calculation to calculate the water pressure
distribution.
217
Finally, for all four water pressure situations the safety factor of the dam is
calculated by means of phi-c reduction. This leads to the following cases being
considered:
water level remains at 25m.
water level drops quickly from 25 to 5m.
water level drops slowly from 25 to 5m.
water level drops extremely slowly to 5m and remains there.
Material properties:
The material data sets of the clay core, the fill material and the sub-soil are
shown in table CA6.1.
Tab. CA6.1: Dam and soil properties (Mohr-Coulomb model)
Parameter
Symbol
Core
Fill
Sub-soil
Unit
Material model
Model
Mohr-Coulomb
Mohr-Coulomb
Mohr-Coulomb
Type of behaviour
Type
Undrained
Undrained
Undrained
Unsaturated weight
unsat
16.0
16.0
17.0
kN/m3
sat
20.0
20.0
21.0
kN/m3
kx , ky , kz
1.010-4
0.25
0.02
m/d
Young's modulus
8000
20000
50000
kN/m2
Poisson's ratio
0.35
0.33
0.3
Cohesion
5.0
5.0
10.0
kN/m2
Friction angle
25.0
30.0
32.0
Dilatancy angle
0.0
0.0
0.0
einit
0.5
0.5
0.5
Standard
Standard
Standard
(very fine)
(coarse)
(Medium)
n/a
n/a
0.47
Saturated weight
Permeability
Void ratio
K0
Calculation phases:
Phase 1: Steady state groundwater flow calculation.
Create a closed flow boundary at the bottom of the model
Generate groundwater head conditions at the other model boundaries by
creating a general water level. The left head must be at a level of 25 m
above the ground surface and the right boundary at a level of 10 m below
the ground surface.
Phase 2: Gravity loading
218
Fig. CA6.3.2D: Steady-state pore pressure for high reservoir level (PLAXIS 2D)
Fig. CA6.4.2D: Degree of saturation for high reservoir level (PLAXIS 2D)
219
Fig. CA6.5.2D: Flow field for high reservoir level (PLAXIS 2D)
Fig. CA6.3.3D: Steady-state pore pressure for high reservoir level (PLAXIS 3D)
220
Fig. CA6.4.3D: Degree of saturation for high reservoir level (PLAXIS 3D)
Fig. CA6.5.3D: Flow field for high reservoir level (PLAXIS 2D)
Phase 3: Rapid drawdown:
In this phase rapid drawdown of the reservoir level is considered.
In the Parameters tab, select Reset displacements to zero and set the Loading
input to Stage construction. Enter a value of 5 days for the Time interval.
Figure CA6.4 shows active pore pressures. It can be seen that, although the
reservoir level has dropped down to 5 m, there are still high pore pressures in the
dam, especially in the clay core. Other output options may be used to view the
groundwater head, the degree of saturation and the flow field. The development
221
of various quantities in time may be viewed using the Animation option or the
Curves option in the view menu.
Fig. CA6.6.2D: Active pore pressure after rapid drawdown (PLAXIS 2D)
222
Fig. CA6.6.3D: Active pore pressure after rapid drawdown (PLAXIS 3D)
223
In the Parameters tab, select Reset displacements to zero and set the Loading
input to Stage construction. Enter a value of 50 days for the Time interval.
Figure CA6.5 shows active pore pressures.
224
Fig. CA6.9.2D: Active pore pressure after slow drawdown (PLAXIS 2D)
Fig. CA6.9.3D: Active pore pressure after slow drawdown (PLAXIS 3D)
225
226
227
Fig. CA6.12.2D: Steady-state pore pressure for low reservoir level (PLAXIS 2D)
Fig. CA6.13.2D: Degree of saturation for low reservoir level (PLAXIS 2D)
228
Fig. CA6.14.2D: Flow field for low reservoir level (PLAXIS 2D)
Fig. CA6.12.3D: Steady-state pore pressure for low reservoir level (PLAXIS 3D)
229
Fig. CA6.13.3D: Degree of saturation for low reservoir level (PLAXIS 3D)
Fig. CA6.14.3D: Flow field for low reservoir level (PLAXIS 3D)
Phase 5 to 8: Stability calculations:
In Phases 5 to 8 stability calculations are defined for the phases 1 to 4
respectively. Therefore, select the corresponding phase in the Start from phase
parameter and set the Calculation type to Phi-c reduction. In the Parameter tab
set the number of additional steps to 50 and select Reset displacement to zero.
230
Output:
The results of the four groundwater flow calculations in terms of pore pressure
distribution have been shown in previous figures. Four different situations were
considered:
1.
2.
3.
4.
231
232
Fig. CA6.16.PF: Steady-state pore pressure distribution for high reservoir level
(PlaxFlow)
Fig. CA6.17.PF: Pore pressure distribution after rapid draw down (PlaxFlow)
Fig. CA6.18.PF: Pore pressure distribution after slower draw down (PlaxFlow)
Fig. CA6.19.PF: Steady-state pore pressure distribution for low reservoir level
(PlaxFlow)
233
9.7 Conclusions
Results of 5 coupled flow deformation analyses and one gravity loading with
Bishop stress, solved by the new PLAXIS 2D and 3D kernels, are shown in this
chapter. Some of them have been verified against analytical solution.
The following features have been tested:
1. Boundary conditions. All boundary conditions have been tested in the
chapters of groundwater flow. However, seepage boundary condition,
inflow, precipitation are tested here.
2. Bishop stress: Bishop effective stress has been tested in this chapter
and it has been shown that PLAXIS is capable of calculating Bishop
effective stress.
3. One dimensional consolidation. Results of one-dimensional
consolidation are very close to the analytical ones
4. Safety factor with suction. It has been shown that if Bishop stress is
used, the factor of safety might be higher than the factor of safety if
Terzaghi stress without suction is used. For practical application,
this should be changed and suction should not be considered in
phi/c reduction! For transient from Bishop to Terzaghi, a nil phase
may be needed because of the out of balance force.
234
235
236
Unit
Value
0
s
[-]
[-]
[-]
[-]
[-]
[-]
[-]
[kPa]
[kPa]
[kPa]
[kPa-1]
[m-1]
[-]
[-]
0.3
0.05
0.25
0
0
0.772
1.21
1.0
24
0.75
0.012
1.0
0.5
-1.0
ks
e0
Pr
P0
r
ga
gn
gc
237
Fig. USM1.1: Stress path in triaxial compression test (After Sheng et al, 2003)
70
0 Kpa
60
100 Kpa
200 Kpa
q[Kpa]
50
40
30
20
10
0
0
0.1
0.2
0.3
0.4
0.5
Axial strain
Fig. USM1.2: Shear stress versus axial strain for different suction (Plaxis)
250
200 Kpa
0 kpa
100 kpa
s [kpa]
200
150
100
50
0
0
20
40
60
q [kpa]
80
238
20
40
60
80
p ' [ kp a]
200 kpa
100 kpa
0 kpa
f _line
2.2
2.15
2.1
2.05
2
0
10
20
30
40
50
60
70
80
p'[ kpa]
200 kpa
100 kpa
0 kpa
s=0 kPa
s=100 kPa
s=200 kPa
60
q (kPa)
50
40
30
20
10
0
0
0.1
0.2
0.3
0.4
0.5
Axial strain
Fig. USM1.6: Shear stress versus axial strain for different suction (Gonzalez &
Gens, 2008)
239
250
B3
s (kPa)
200
C3
s=0 kPa
150
s=100 kPa
B2
100
s=200 kPa
C2
50
A
C1
0
0
10
20
30
40
50
60
70
q (kPa)
Fig. USM1.7: Suction versus deviatoric stress q (Gonzalez & Gens, 2008)
60
s=0 kPa
s=100 kPa
C2
40
q (kPa)
C3
s=200 kPa
50
CSL
30
C1
20
10
A
0
0
B2
20
B3
40
60
80
p' (kPa)
Fig. USM1.8: Stress path in p-q space (Gonzalez & Gens, 2008)
2.3
s=0 kPa
s=100 kPa
s=200 kPa
2.25
A
=1+e
2.2
B2
B3
2.15
2.1
2.05
C3
C2
C1
2
0
10
20
30
40
50
60
70
80
p' (kPa)
Fig. USM1.9: Specific volume versus mean effective stress p (Gonzalez &
Gens, 2008)
240
Fig. USM1.10: Shear stress versus axial strain for different suction (Sheng et al.,
2003)
241
USM1.2b. It follows from the figure that the stress paths reach the critical state
line at the axial strain 50%.
Summary:
As seen, results from Plaxis are in agreement with the results provided by
Gonzalez & Gens (2008) and Sheng et al. (2003).
242
243
The results of the fully coupled flow-deformation analysis are shown in Figure 4.
The predicted settlements of the ground surface are plotted in Figure 4a. Figure
4b shows the amount of suction at the centre of the footing. During drying phase
the ground surface settles roughly 9 cm. Applying the flexible footing (the
distributed load) causes the centre of the footing settles 13.6 cm. Wetting of the
soil leads to significantly increase the displacements of the nodes below the
footing while the ground surface at x=10 m settles and rises a little. The
maximum displacement occurs at the centre of the footing, 17 cm. Figure 5
shows the deformed mesh after the loading and wetting phases.
Tab. USM2.: Material properties used for triaxial tests (After Sheng et al.,
2003)
Parameter
sat
dry
kx & ky
0
s
ks
e0
Pr
P0
r
ga
gn
gc
gl
Unit
Value
3
[kN/m ]
[kN/m3]
[m/day]
[-]
[-]
[-]
[-]
[-]
[-]
[-]
[kPa]
[kPa]
[kPa]
[kPa-1]
[m-1]
[-]
[-]
[-]
18.0
16.0
0.086
0.3
0.02
0.20
0
0
0.984
1.17
1.0
1.0
0.70
0.012
1.0
0.5
-1.0
0.0
244
Fig. USM2.3: Deformed mesh; a) after loading phase; b) after wetting phase
245
10.3 Conclusions
Results of one plastic calculation and one coupled flow deformation analysis
using unsaturated soil model (Barcelona Basic Model), solved by the new Plaxis
2D kernel are shown in this chapter. The former one has been verified against
numerical results provided by Gonzalez & Gens (2008) and Sheng et al. (2003).
The following features have been tested:
1. Effect of suction. The effect of suction has been investigated by
numerical modelling of drained triaxial tests. As shown, results are
very similar to the results of Gonzalez & Gens (2008) and Sheng et al.
(2003).
2. Drying and wetting: Drying and wetting have been tested in the
footing problem. It can be seen that the model is capable of simulating
collapse upon wetting.
3. Different calculation types. The model has been tested with both
plastic and fully coupled flow-deformation analyses.
246
247
(1)
248
(3)
The types of calculations which can be done in this mode are as follows:
Plastic undrained
Plastic drained
Consolidation based on Total Pore Pressure (TPP)
Dynamics
Free vibration
Phi/C reduction
3. Flow mode: This mode is for calculating pure groundwater flow
calculations.
The types of calculations in this mode are:
Steady state groundwater flow
Transient groundwater flow
249
(4)
Kw
2 G 1 + u
1 +
= Ku K =
n
3 1 2 u 1 2
(5)
where Ku and K denote the undrained and drained bulk moduli of the soil
respectively. G is the elastic shear modulus, is the drained Poissons ratio and
u is the undrained Poissons. n is porosity of the soil.
In PLAXIS, it is possible to use effective parameters for undrained calculations.
The undrained elastic moduli of soil can be related to the effective parameters
according to
E u = 2G (1 + u )
(6.1)
Eu
3(1 2 u )
(6.2)
Ku =
E oed ,u =
(1 u )Eu
(1 2 u )(1 + u )
(6.3)
in which
u =
+ (1 + )
1 + 2 (1 + )
(7)
with
1 Kw
3 n K
E
K =
3(1 2 )
(8)
(9)
Eq. (7) shows the relationship between the undrained Poissons ratio, the
effective Poissons ratio, the bulk modulus of water, the bulk modulus of soil
skeleton and porosity of the soil. It follows form Eq. (7) that if water is assumed
to be incompressible (Kw), then (u0.5) which leads to singularity of
stiffness matrix. Therefore it is assumed u to be 0.495 for fully saturated. In
reality, the bulk modulus of water is very large, but not infinite. The bulk
modulus of pure water (without bubbles of air) is K0w =2106 kPa).
The generation of excess pore water pressure can be studied by means of the
Skempton B-parameter which is defined as the ratio of excess pore water pore
pressure increment to the mean total stress increment:
250
dp w
dp
(10)
K w d vol
n dp
(11)
(12)
where Ku is the undrained bulk modulus of soil which can be obtained from
Ku =
2G (1 + u )
3(1 2 u )
(13)
(1 + )(1 2 u )
Kw
= 1
(1 + u )(1 2 )
nK u
(14)
u =
3 + B(1 2 )
3 B(1 2 )
(15)
or
As seen, B varies between 1 (for fully saturated conditions) and 0 (for fully dry
condition) when u=0.5 and u= , respectively. Experimental data show that the
parameter B decreases by decreasing the degree of saturation (Figure 1).
In the advanced mode of PLAXIS, the value of Skemptons B-parameter is
unknown but the degree of saturation is known. Therefore the bulk modulus of
water is estimated from:
K wunsat =
K wsat K air
SK air + (1 S ) K wsat
(16)
in which Kair is the bulk modulus of air which is about 100 kPa under
atmospheric pressure. In PLAXIS, it is assumed that pa (air pore pressure) is
equal to 0 for practical application and therefore an artificial and small value (1
kPa) is used for the bulk modulus of air. Kwsat and Kwunsat are bulk moduli of
water in saturated and unsaturated conditions, respectively. Kwsat is calculated
251
based on u which is equal to 0.495 when the standard setting is being used.
PLAXIS always checks the value of Kwsat to ensure that Kwsat is less than the bulk
modulus of pure water (Kw0 =2106).
252
Tables 1 shows the summary of the bulk moduli of water used in the classical
mode of PLAXIS. As seen, in this mode, the same Kw is utilised for below and
above the phreatic level. This means that, in this mode, it is possible to generate
high excess pore water pressure in the area where lower amount of water exists.
S *
i +
0
v
0
p =
0 t H
w
i +
fu
vi
i + t G + t (q i + q ) (17)
p
p
p w
with
S = ( S + t H )
*
H = ( N ) T
V
k rel
(17.1)
k
sat
( N ) dV
T nS
dS
N dV
S = N
n
dpw
Kw
V
G = ( N ) T
V
k rel
q p = N q w dS
sat
w g dV
(17.2)
(17.3)
(17.4)
(17.5)
K = B M B dV
(17.6)
V
T
Q = S B m N dV
(17.7)
C = N S L N dV
V
(17.8)
f u = N b dV + N t dS
V
253
(17.9)
The bulk modulus of water appears in Eq (17.3). Here, the bulk modulus of water
is not reduced because the saturation is involved in the matrix S. Therefore, for
partially saturated zones, the storage is reduced according to the degree of
saturation.
Similar to the classical mode, the bulk modulus of water is decreased in all types
of calculations for materials which are just switched on to decrease the generation
of excess water pore pressure.
Tables 2 shows the summary of the bulk moduli of water used in the advanced
mode of PLAXIS.
254
Plastic
(drained)
Kw = 0
Plastic (undrained)
Kw =
2 G 1 + u
1 +
3 1 2 u 1 2
Kw = 0
Kw = 0
Kw = 0
Kw = 0
Kw = 0
Kw = 0
Consolidation
Safety or
dynamics
(drained)
Kw =
2 G 1 + u
1 +
3 1 2 u 1 2
Kw = 0
Kw =
2 G 1 + u
1 +
3 1 2 u 1 2
Kw = 0
Kw = 0
2 G 1 + u
1 +
1 10 8
3 1 2 u 1 2
Not relevant.
Not relevant.
Kw = 0
Kw = 0
Kw =
Kw = 0
Kw =
2 G 1 + u
1 +
3 1 2 u 1 2
255
Plastic
(drained)
Undrained
(pw<=0)
Kw = 0
Undrained
(pw>0)
Kw = 0
Drained
(pw<=0)
Kw = 0
Drained
(pw>0)
Kw = 0
Plastic (undrained)
K wsat =
Consolidation
Safety or
dynamics
(drained)
2 G 1 + u
1 +
3 1 2 u 1 2
K wsat =
2 G 1 + u
1 +
3 1 2 u 1 2
Kw = 0
K wsat K air
SK air + (1 S ) K wsat
K wsat =
2 G 1 + u
1 +
3 1 2 u 1 2
Kw = 0
Kw = 0
K wsat =
2 G 1 + u
1 +
3 1 2 u 1 2
Kw = 0
Kw = 0
Kw = 0
K wsat =
2 G 1 + u
1 +
3 1 2 u 1 2
Kw = 0
Kw = 0
K wunsat =
K wsat =
2 G 1 + u
1 +
3 1 2 u 1 2
K wunsat =
K wsat K air
SK air + (1 S ) K wsat
256
Material
just
switched
on
(below
and
above
phreatic
level)
Kw = 0
Kw = 0
K w = K wsat 1 10 8
Not
relevant.
Not relevant.
Nonporous or
dry
cluster
Kw = 0
Kw = 0
Kw = 0
Kw = 0
Kw = 0
257
Type of material
Transient
Undrained
(pw<=0)
Kw =
2 G 1 + u
1 +
3 1 2 u 1 2
K wsat =
2 G 1 + u
1 +
3 1 2 u 1 2
Undrained
(pw>0)
Kw =
2 G 1 + u
1 +
3 1 2 u 1 2
K wsat =
2 G 1 + u
1 +
3 1 2 u 1 2
Drained
(pw<=0)
Kw =
2 G 1 + u
1 +
3 1 2 u 1 2
K wsat =
2 G 1 + u
1 +
3 1 2 u 1 2
Drained
(pw>0)
Kw =
2 G 1 + u
1 +
3 1 2 u 1 2
K wsat =
2 G 1 + u
1 +
3 1 2 u 1 2
Kw =
2 G 1 + u
1 +
3 1 2 u 1 2
Kw = 0
K w = K wsat 1 10 8
Kw = 0
258
11.6 Examples
In this section one example has been chosen to show effects of the new
formulation in generating excess water pore pressures in the advanced mode. It is
attempted to show how the Soil Water Characteristic Curve affects the undrained
behaviour of soil.
Figure 2 shows the geometry of the problem which is 1 m wide and 2 m high.
The initial phreatic level is at 1 m high. This example is performed in two phases
as follows:
1. Phase 0: Gravity loading
2. Phase 1: Activating a distributed load of 10 kPa on top of the model and
doing plastic analysis
Phreatic level
Description
Elastic modulus
Poissons ratio
initial void ratio
Water weight
Soil weight (sat)
Soil weight (dry)
Tab. 5:
Symbol
Eref
einit
w
sat
unsat
Unit
[kN/m2]
[-]
[-]
[kN/m3]
[kN/m3]
[kN/m3]
Value
1000
0
0.50
10.0
20.0
20.0
Tab. 6:
259
Symbol
Eref
einit
gn
ga
w
sat
unsat
Unit
[kN/m2]
[-]
[-]
[-]
[m-1]
[kN/m3]
[kN/m3]
[kN/m3]
Value
1000
0
0.50
1.3774
3.830
10.0
20.0
20.0
Symbol
Eref
einit
gn
ga
w
sat
unsat
Unit
[kN/m2]
[-]
[-]
[-]
[m-1]
[kN/m3]
[kN/m3]
[kN/m3]
Value
1000
0
0.50
1.1012
3.670
10.0
20.0
20.0
260
agreement with Eq. (2). It should be noted that in this example, everywhere is
fully saturated (S=1.0).
Fig. 3: Results after the initial phase for fully saturated soil: left) vertical total
stress; middle) vertical effective stress; right) active pore pressure
The external load is applied in phase 1 which is a distributed load of 10 kPa on
top of the model. As the material is defined as undrained, this causes to generate
excess pore pressure. Eq. 16 is used for calculating the bulk modulus of water. It
follows from this equation when S=1 the Kwusat = Kwsat. Therefore, there is no
distinction between below and above the phreatic level. The bulk modulus of
water in this example is Kesat=Kwsat/n = 49.5e3 kPa (Eq. 5).
Figure 4 shows the results of phase 1. By applying 10 kPa on top of the model,
the vertical total stress increases by 10 kPa as expected. From the undrained
elastic moduli (Eqs. 6), it can be found that the Eoed,u=50499.99 kPa and therefore
y=v=0.19810-3 and Pexcess=9.802 kPa (Figure 5). The difference between the
total stress and the excess pore pressure is added to the effective stress (10 -9.802
= 0.2 kPa). As seen, the results are the same as analytical solution.
261
Fig. 4: Results after the phase 1 for fully saturated soil: left) vertical total stress;
middle) vertical effective stress; right) active pore pressure
Fig. 5: Excess water pore pressure after the phase 1 for fully saturated soil
262
Fig. 6: Results after the initial phase for coarse material: left) vertical total stress;
middle) vertical effective stress; right) active pore pressure
Verification:
Bishop stress reads:
= + m (S p w )
As the same weight for dry and saturated soil is used, the vertical total stress at
the bottom can be simply calculated by:
= h = 2 20 = 40 kPa
(2)
263
The degree of saturation is 1 at the bottom and 0.6049 at the top. Therefore
Bishops effective stresses at the top and bottom are:
= top S p w = 0 0.6049 10 = 6.049 kPa
top
bottom
= bottom S pw = 40 1 10 = 30.0 kPa
It can be seen that the calculated results from PLAXIS are the same as the
analytical solution.
It should be noted that the degree of saturation is not linear in the unsaturated
area and consequently the resulting effective stress is not linear in this area
(Figure 6).
Fig. 7: Results after the initial phase for coarse material: Degree of saturation
Figure 8 and 9 show the results of the fine material after the initial phase. In the
same manner done for the coarse material, the results can be verified. As seen in
this case, higher degree of saturation and higher effective stresses are generated
in the unsaturated area.
264
Fig. 8: Results after the initial phase for fine material: left) vertical total stress;
middle) vertical effective stress; right) active pore pressure
Fig. 9: Results after the initial phase for fine material: Degree of saturation
In phase 1, undrained behaviour is assumed and consequently excess water pore
pressure is generated. As the bulk modulus of water, weight of soil, degree of
saturation and the effective stress are suction dependent, the undrained
calculations are always non-linear in the advanced mode even for linear elastic
materials as used in this example.
The total and effective stresses as well as the active pore pressures of phase 1 in
coarse material are plotted in Figure 10.
265
Fig. 10: Results after the phase 1 for coarse material: left) vertical total stress;
middle) vertical effective stress; right) active pore pressure
Fig. 11: Excess water pore pressure in coarse material at the end of phase 1.
In the same way done in for phase 0, the stresses can be verified in this phase.
Due to the highly non-linearity behaviour, it is difficult to verify water excess
pore pressure analytically as done for the saturated case.
In Figure 10, it can be observed that the vertical total stress on top of the model is
9.968 kPa which slightly less than the analytical solution, 10 kPa (error = 0.32%).
This difference is due to the non-linearity mentioned above.
266
Fig. 12: Results after phase 1 for coarse material: Degree of saturation
Results of the fine material at the end of phase 1 are plotted in Figures 13-15. By
comparing the results from the coarse and fine materials, it can be seen that more
excess pore pressure (and consequently less effective stress) is developed in the
fine material which is in agreement with reality.
Fig. 13: Results after the phase 1 for fine material: left) vertical total stress;
middle) vertical effective stress; right) active pore pressure
267
Fig. 14: Excess water pore pressure in fine material at the end of phase 1.
Fig. 15: Results after phase 1 for fine material: Degree of saturation
The results of the coarse material (effective stress, total stress, degree of
saturation, active and excess pore pressures versus depth) are plotted in Figures
16-20 and the results of the fine material in Figures 21-25. It can be seen that,
according to the SWCC used in the calculation, different effective stresses are
developed in the model. As the degradation of the degree of saturation increases
with increasing suction, effective stress in the unsaturated zone are increases and
the problem becomes more non-linear (because the properties of the soil changes
more by changing the suction).
268
Fig. 16: Effective stresses at the end of phase 0 (step 1) and phase 1 (step 104) in
coarse material
Fig. 17: Vertical total stress at the end of phase 0 (step 1) and phase 1 (step 104)
in coarse material
269
Fig. 18: Degree of saturation at the end of phase 0 (step 1) and phase 1 (step 104)
in coarse material
Fig. 19: Active pore pressure at the end of phase 0 (step 1) and phase 1 (step 104)
in coarse material
270
Fig. 20: Excess pore pressure at the end of phase 0 (step 1) and phase 1 (step
104) in coarse material
Fig. 21: Effective stresses at the end of phase 0 (step 1) and phase 1 (step 113) in
fine material
271
Fig. 22: Vertical total stress at the end of phase 0 (step 1) and phase 1 (step 113)
in fine material
Fig. 23: Degree of saturation at the end of phase 0 (step 1) and phase 1 (step 113)
in fine material
272
Fig. 24: Active pore pressure at the end of phase 0 (step 1) and phase 1 (step 113)
in fine material
Fig. 25: Excess pore pressure at the end of phase 0 (step 1) and phase 1 (step
113) in fine material
12 Bibliography
273
12 Bibliography
Alonso, E.E., Gens, A. & Josa, A. (1990), A constitutive model for partially
saturated soils. Gotechnique, 40: 405-430.
Barrera, M., Romero, E., Snchez, M. & Lloret, A. (2002). Laboratory tests to
validate and determine parameters of an elastoplastic model for unsaturated
soils. International Symposium on identification and determination of soil
and rock parameters for geotechnical design. Jean-Pierre MAGNAN.
LCPC.
Bardet, J-P & Tobita, T. (2002). A practical method for solving free-surface
seepage problems. Computer and Geotechnics, 29: 451-475.
Biot, M.A. (1941). General theory of three dimensional consolidation. Journal
of Applied Physics, 12, 155-164.
Bishop, A.W. & Blight, A.K.G. (1963). Some aspects of effective stress in
saturated and partially saturated soils. Gotechnique, 13: 177-197.
Bishop, A. W. & Eldin, A. K. G. (1950). Undrained triaxial tests on saturated
sands and their significance in the general theory of shear strength.
Gotechnique, 2: 13-32.
Bolzon, G., Screfler, B.A. & Zienkiewicz, O.C. (1996). Elastoplastic soil
constitutive laws generalised to partially saturated states. Gotechnique,
46(2), 279-289.
Booker, J.R. & Small, J.C. (1975). An investigation of the stability of numerical
solutions of Biots equations of consolidation. Int. J. Solids Struct., 11,
907-917.
Borja, R.I., Kishnani, S.S. (1991). On the solution of elliptic free-boundary
problems via Newtons method. Computer Methods in Applied Mechanics
and Engineering, 88: 341-361.
Brinkgreve, R.B.J, Broere, W. & Waterman, D. (2006), Plaxis, Finite element
code for soil and rock analyses, users manual. The Netherlands.
Darcy, H. (1856). Histoire Des Foundataines Publique de Dijon. Paris, Dalmont,
pp. 590594.
Dogan, A. & Motz, L. H. (2005). Saturated-unsaturated 3D groundwater model.
I: Development. Journal of Hydrological Engineering, ASCE, Vol. 10, No.
6, 492-504.
274
12 Bibliography
12 Bibliography
275
Roscoe, K.H. & Burland, J.B. (1968). On the generalised stress-strain behaviour
of wet clay. Engineering Plasticity. Cambridge University Press:
Cambridge, MA, 535-609.
Schwab, R. (2008). Report: Implementation of total pore pressure approach in
Plaxis.
Sheng, D., Sloan, S.W., Gens, A., & Smith, D.W. (2003). Finite element
formulation and algorithms for unsaturated soils. Part II: Verification and
application. Int. J. for Numerical and Analytical Meth. in Geomech., 27:
767- 790.
Song, E.X. (1990). Elasto-plastic consolidation under steady state and cyclic
loads. Dissertation, Delft University of Technology, Delft, Netherlands.
Strack, O.D. & Asgian, M.I. (1978). A new function for use in the hodograph
method. Water Resource Research, No. 14(6), 1045-1058.
Van Genuchten, M.T. (1980). A closed-form equation for predicting the
hydraulic conductivity of unsaturated soil. Soil Science Society of
America Journal, 44: 892-898.
Verruijt, A. (2001). Soil mechanics, Delft University of Technology.
Zhang, C.-L., Krhn, K.-P., & Rothfuchs, T. (2003). Applications of CODEBRIGHT to thermal-hydromechanical experiments on clays. Proceedings of
the International Conference from Experimental Evidence towards
Numerical Modelling. Numerical and Theoretical Approaches Unsaturated.
Volume II. T. Schanz (Ed.). pp: 341-358.
276
12 Bibliography
Appendix A
277
Appendix A
In the following definitions of invariants and first and second derivatives of yield
function and plastic potential used in Barcelona Basic Model are given.
Definitions of invariants
Stress tensor:
ij = p ij + s ij
ij
3
kk
3
2
6
1
2
2
2
J 2 = ( 1 2 ) + ( 1 3 ) + ( 2 3 )
6
J2 =
s12
s13
s 22
s 32
s 23
s 33
1 3 3 3
( s 1 + s 2 + s 3 ) = s1 s 2 s 3
3
Lode angle:
1
3
3 3 J 3
3
2 J
= sin 1
Derivatives of invariants
First derivatives of invariants:
J = J2
278
Appendix A
p 1
r =
j 3
1
3
J s1
r =
j 2 J
0 0 0
3
s2
2J
s3
2J
s4
J
s5
J
s6
J 3 J J
3 r3 3 r
j
J j
1 J 3 1 J
r = tan ( 3 )
r
r =
j
4 J 6 27 J 33
3 J 3 j J j
1 2
1 2
2 2
2
2
2
3 ( s11 s23 ) 3 ( s22 s13 ) 3 ( s33 s12 )
2 ( s 2 s 2 ) 1 ( s 2 s 2 ) 1 ( s 2 s 2 )
3 22 13 3 11 23 3 33 12
J 3 2 2
1 2
1 2
2
2
2
r = ( s33 s12 ) ( s11 s23 ) ( s22 s13 )
j 3
3
3
mij kl
2Q
r r
1 2
2Q
r r
1 3
2Q
r r
1 4
2Q
r r
1 5
2Q
r r
2 2
2Q
r r
2 3
2Q
r r
2 4
2Q
r r
2 5
2Q
r r
3 2
2Q
r r
3 3
2Q
r r
3 4
2Q
r r
3 5
2Q
r r
4 2
2Q
r r
4 3
2Q
r r
4 4
2Q
r r
4 5
2Q
r r
5 2
2Q
r r
5 3
2Q
r r
5 4
2Q
r r
5 5
2Q
r r
6 2
2Q
r r
6 3
2Q
r r
6 4
2Q
r r
6 5
2Q
r r
1 6
2Q
r r
2 6
2Q
r r
3 6
2Q
r r
4 6
2Q
r r
5 6
2Q
r r
6 6
Appendix A
279
where:
Qi
2Q
Q 2 p p
2Q Q 2 J J
2Q Q 2
2Q
r = r r =
r2 + r
r +
r2 + r
r +
r2 + r
r
j
i j p j i p j J j i J j j i j
1
424
3
0
1
s12
3
3J 4 J
2J
=
i j
1 s1 s 2
6J 4 J 3
1
s2
23
3J 4 J
1 s1 s 3
6J 4 J 3
1 s 2 s3
6J 4 J 3
s2
1
33
3J 4 J
s1s12
2J 3
s s
2 123
2J
ss
3 123
2J
1 s122
J J3
SYM
s1s13
2J 3
ss
2 133
2J
ss
3 133
2J
s s
12 13
2J 3
1 s132
J J3
s1s 23
2J 3
ss
2 233
2J
s 3 s 23
2J 3
s12 s 23
2J 3
s13 s 23
2J 3
2
1 s 23
3
J J
sin 2 J 3 J 3 1 + cos 2 J J
2
2
= tan 2
+
r =
2
j2 ij kl
cos 2 ij kl
J 3 cos ij kl
J J 3 J 3 J
+
2
cos JJ 3 ij kl ij kl
1
1 2 J3
1 2 J
+
J 3 ij kl J ij kl
1 J 3 1 J
r
r
2
3 J 3 j J j 1 J 3 1 J tan 3
+
r = tan 3
r
r
r
r
j
j
j2
3 J 3 j J j
144424443 14442444
424
3
3 1
A66
B61
where:
1
1
2
3J3 1 J
1 J 3 J 3
J
J
=
r2 + r r
r2 r2 r
3 J 3 j j
j
J j j
j
2
A66
C61
280
Appendix A
2
3 s1
2s
3 3
2
s
2J3 3 2
r =
j2 2
s12
3
2
3 s13
4 s
23
3
2
s3
3
2
s2
3
2
s1
3
2
s12
3
4
s13
3
2
s 23
3
2
s2
3
2
s1
3
2
s3
3
4
s12
3
2
s13
3
2
s 23
3
2
s12
3
2
s12
3
4
s12
3
2
s13
3
4
s13
3
2
s13
3
2 s 3
2 s 23
2 s 23
2 s 2
2 s13
2 s12
4 T
s 23
3
2
s 23
2
s 23
3
2 s13
2 s12
2 s1
J 3
J 3
1
r
r
1
j
j
3J 3 ( 3J 3 )
= 3
= 2
r =
r
3J3
j
j
( 3 J 3 )2
tan 3
2
r = 3 (1 + tan 3 ) r
j
j
11
2Q
r
2 1
2Q
r
31
Qi
2Q
= r
=
j
i j 2Q
r
4 1
2
Q
r5 1
2
Q
r
6 1
where,
2Q
Q 2 p
p
2Q Q 2 J
J
2Q Q 2
2Q
=
+ r
+
+ r
+
+ r
r
r
r
r
i j p i j i p j J i j i J j i j i j
123
123
123
0
Appendix A
281
F
= 6J ;
J
G
= 6 J
J
2 g ( ) M 2 ( p + ps )( Pc p )
F
G
=
=
2
g ( ) g ( )
g ( 30 )
cos sin
sin sin +
g ( )
3
=
2
sin sin
cos +
g ( )
F G
=
=
M 2 ( p + ps )
2
Pc Pc
g ( 30 )
o* *
P s* * *
Pr o
o * )
(
P
Pc
= r *
Po
( s * ) Po
P
Pc ( o* * ) o* (1 r ) exp ( s ) ln o
Pc
Pr
=
2
s
( s* * )
g ( )
F G
2
=
=
M ( Pc p )
ps ps
g ( 30 )
2
ps
= s
s
Second derivatives:
g ( )
2 F
2
= 2
M
2
p
g ( 30 )
2
a) p:
2 F
=0
pJ
282
Appendix A
g ( )
F
= 2
p
g ( 30 )
2
cos sin
M 2 ( Pc 2 p ps ) sin +
sin sin
cos +
g ( )
2 F
2 Pc
=
M
pPo
Po
g ( 30 )
2
b) J
2 F
=0
J p
2 F
2G
=
6;
= 6
J 2
J 2
2 F
=0
J
2 F
=0
J Pc
c)
2 F
2 F
=
p p
2 F
=0
J
g ( )
F
= 6
g ( 30 )
2
g ( )
2
g ( 30 )
cos sin
M 2 ( p + ps )( Pc p ) sin +
+
2
sin sin
cos
+
sin sin
M 2 ( p + ps )( Pc p ) cos
sin sin
cos +
cos sin
M 2 ( p + ps ) sin +
g ( )
F
3
Pc
= 2
sin sin
Pc
Po
g ( 30 )
cos +
Appendix B
283
Appendix B
Here it is shown how Skempton B-parameter, the undrained Poissons ratio u
and the equivalent bulk modulus of pore fluid Kw/n can be derived if one of them
is known.
Note: In all cases, the effective Poissons ratio ' and shear stiffness G are
known.
Case 1:
Parameters
Conditions
Known
B
0 B 0.995
Unknown
u, Kw/n
' u 0.4988
3 + B(1 2 )
3 B(1 2 )
(15)
n
3 1 2 u 1 2
(5)
Case 2:
Known
Parameters
Conditions
' u 0.4988
Unknown
B, Kw/n
0 B 0.995
(1 + )(1 2 u )
(1 + u )(1 2 )
(14)
3 1 2 u 1 2
n
(5)
284
Appendix B
Case 3:
Parameters
Conditions
Known
Unknown
Kw/n
B, u
Kw 2 G
1 +
624.5
n
3
1 2
0 B 0.995
' u 0.4988
1
2 + 1
where
=
2 G K w 1 +
+
3
n 1 2
(1 + )(1 2 u )
(1 + u )(1 2 )
(14)
Appendix C
Here it is shown how to convert old PlaxFlow material data set to the new one. In
the new material data set, Csat is removed because it can simply be derived from
the equivalent bulk modulus of water Kw/n and water weight w:
C sat =
n w
Kw
(1)
and
K w 2 G 1 + u
1 +
n
3 1 2 u 1 2
(2)
or
Kw
E 1 + u
1 +
n
3(1 + ) 1 2 u 1 2
(3)
Therefore:
E =
(4)
C sat
in which
=
1 + u
1
1 +
3(1 + ) 1 2 u 1 2
(5)
How to convert:
Step 1: Choose linear elastic material.
Step 2: Choose Undrained (A) as type of drainage.
Step 3: Set =0.3 and u=0.495 (standard setting of undrained behaviour).
Step 4: Calculate :
=
1 + u
1
1 +
3(1 + ) 1 2 u 1 2
Step 5: Calculate E:
E =
C sat