Sunteți pe pagina 1din 14

599

Consolidation analysis for unsaturated soils


Enrico Conte

Abstract: This paper deals with the multidimensional consolidation of unsaturated soils when both the air phase and
water phase are continuous. Following the approach proposed by D.G. Fredlund and his coworkers, the differential
equations governing the coupled and uncoupled consolidation are first derived and then solved numerically. The solu-
tion is achieved using a procedure that depends on the transformation of the field equations by using the Fourier trans-
form. This transformation has the effect of reducing a two- or three-dimensional problem to a problem involving only a
single spatial dimension. The transformed equations are solved using a finite element approximation that makes use of
simple one-dimensional elements. Once the solution in the transformed domain is obtained, the actual solution is
achieved by inversion of the Fourier transform. The time integration process is formulated in a stepwise form. Results
are presented to point out some aspects of the consolidation in unsaturated soils. Moreover, it is shown that the results
obtained using the simple uncoupled theory are of sufficient accuracy for practical purposes.
Key words: coupled consolidation, uncoupled consolidation, unsaturated soils, Fourier transform.
Résumé : Cet article traite de la consolidation multidimensionnelle des sols non saturés quand aussi bien la phase li-
quide que la phase gazeuse sont continues. A partir de la formulation proposée par D.G. Fredlund et ses collaborateurs,
les équations différentielles controlant la consolidation couplée et la consolidation désaccouplée sont d’abord établies et
ensuite sont résolues numériquement. La solution est obtenue par une procédure qui peut compter sur la transformation
des équations de base au moyen de la transformation de Fourier. Cette transformation a l’effet de réduire un problème
à deux ou trois dimensions à un problème unidimensionnel. Les équations transformées sont résolves par la méthode
des éléments finis employant de simples éléments unidimensionnels. Une fois que la solution des équations transfor-
mées est obtenue, la solution réel est dérivée au moyen de la transformation inverse de Fourier. L’intégration par rap-
port au temps est exécutée pas à pas. Des résultats sont présentés pour mettre en évidence quelques caractéristiques de
la consolidation des sols non saturés. En outre, on montre que les résultats obtenus par la simple théorie désaccouplée
de la consolidation sont suffisamment précis pour leur utilisation pratique.
Mots clés : consolidation couplée, consolidation désaccouplée, sols non saturés, transformation de Fourier.

Conte
612

Introduction In the last few decades, there has been a considerable in-
crease in the understanding of the behaviour of unsaturated
The consolidation of cohesive soils as a result of dissipa- soils; as a result, several theoretical and experimental studies
tion of the excess pore pressures generated by external load- have been conducted to analyse the consolidation processes
ing is a problem of considerable concern for geotechnical in such soils (Fredlund and Rahardjo 1993). Unsaturated
engineers. In 1925, Terzaghi presented a simple theory for soils essentially consist of three phases: solid, liquid, and
the analysis of one-dimensional consolidation in saturated gaseous. As first pointed out by Barden (1965), three differ-
soils which is still widely used in practice. Generalization to
ent classes of behaviour may be singled out on the basis of
three dimensions has given rise to two different approaches,
the continuity of the fluid phases: for high values of the de-
which are known as the uncoupled consolidation theory
gree of saturation of the soil, the water phase is continuous
(Rendulic 1936) and the coupled consolidation theory (Biot
and the air phase is discontinuous; for lower values of the
1941). The latter is preferable from a theoretical point of
degree of saturation, both the air phase and the water phase
view because it provides a coupling between the magnitude
may be considered as continuous; lastly, when the degree of
and progress of settlement. The uncoupled approach cannot
saturation is low, the air phase is continuous and the water
model all the aspects of consolidation in saturated soils
phase is discontinuous. Consolidation should be analysed us-
(Schiffman et al. 1969), but it has proved to be useful in
ing a specific approach for each of these three classes. For
practice (Davis and Poulos 1972).
example, when the soil is close to saturation, the air con-
tained in the pores is occluded and cannot flow as a continu-
ous fluid. In these circumstances, the air bubbles and pore
Received 15 April 2003. Accepted 2 February 2004. water behave as a homogeneous compressible fluid flowing
Published on the NRC Research Press Web site at under pore water pressure gradients. As a result, the case of
http://cgj@nrc.ca on 20 August 2004. occluded air may be analysed using essentially the same for-
E. Conte. Faculty of Engineering, University of Calabria, mulation as that for saturated soils provided that pore fluid
87030 Rende, Cosenza, Italy (e-mail: conte@dds.unical.it). compressibility is accounted for. Solutions were proposed by

Can. Geotech. J. 41: 599–612 (2004) doi: 10.1139/T04-017 © 2004 NRC Canada
600 Can. Geotech. J. Vol. 41, 2004

Olson (1986) for one-dimensional consolidation and by Although only soil systems under plane-strain conditions are
other authors for two- or three-dimensional conditions (Biot considered in this paper, the procedure could be extended to
1941; Verruijt 1969; Ghaboussi and Wilson 1973; Conte three-dimensional problems as well. The method is also used
1998). In this context, Chang and Duncan (1983) developed to solve the equations governing the uncoupled consolida-
a modified version of the Cam–Clay model to describe the tion in unsaturated soils owing to the application of an exter-
stress–strain behaviour of unsaturated soils. For lower val- nal load. Results are presented to ascertain whether the
ues of the degree of saturation, consolidation analysis is uncoupled approach is able to provide suitably accurate re-
more complex because air and water may flow simulta- sults, despite its greater simplicity compared with the cou-
neously and separately through the soil. A general formula- pled solution. Moreover, some interesting aspects of the
tion for one-dimensional consolidation in which the air and consolidation processes in unsaturated soils are pointed out.
water phases are assumed to be continuous was presented at
almost the same time by Fredlund and Hasan (1979) and Coupled consolidation for unsaturated soils
Lloret and Alonso (1980). This formulation is based on two
continuity equations, one for the water phase and one for the When air and water phases are both continuous, the gov-
air phase, which have to be solved simultaneously to give erning equations for the coupled consolidation in unsatu-
water and air pressures at any time and elevation. In the rated soils are the equilibrium equation and the continuity
method developed by Fredlund and Hasan, the constitutive equations for the two fluid phases. In the present section,
relations proposed by Fredlund and Morgenstern (1976) these equations are derived on the basis of the following
were incorporated. Lloret and Alonso used appropriate state assumptions: (i) small strains occur; (ii) solid particles are
surfaces to define the mechanical behaviour of the soil. In incompressible; (iii) air behaves as an ideal gas; (iv) both air
both solution procedures, the governing equations were flow and water flow are governed by a Darcy-type law;
solved using numerical techniques. Ausilio and Conte (v) the effects of temperature change, air dissolved in water,
(1999) showed that the solution for one-dimensional consol- air flow diffusion, and water vapour movement are disre-
idation may also be expressed in terms of the degree of set- garded; and (vi) dynamic and chemical effects are ignored.
tlement and the average degree of consolidation for both the Moreover, it is supposed that the soil is in a state of equilib-
water phase and the air phase. The three-dimensional con- rium prior to loading and deforms under plane-strain condi-
solidation problem was studied by Dakshanamurthy and tions. Consequently, all the field variables in the following
Fredlund (1980) using an uncoupled approach and by Dak- equations represent excesses over their existing values, and
shanamurthy et al. (1984), who first presented the differen- deformation is restricted to the x–z plane. Following
tial equations of the coupled theory. Lloret et al. (1987) Fredlund and Morgenstern (1976), the constitutive relation
extended the one-dimensional consolidation model proposed for the unsaturated soil structure may be expressed as
by Lloret and Alonso to three dimensions and developed an
dV  σ + σz 
iterative procedure for solving the field equations. This pro- [1] = m1s d  x − ua  + m 2s d (ua − uw)
cedure was organized into two stages: the first stage solves V  2 
the stress–strain equations, and the second stage solves the
air and water flow equations. An attractive method based on where dV/V is the volume change of a soil element with re-
the state surface approach was also developed by Thomas spect to its initial volume V; σx and σ z are the total normal
and He (1997), who in addition accounted for the effects of stresses in the x and z directions, respectively; m1s and m 2s are
temperature change and water vapour movement. Moreover, the coefficients of volume change of the soil with respect to
Wong et al. (1998) presented a theoretical study on coupled a change in the net normal stress ([(σx + σ z)/ 2] − ua ) and the
consolidation in unsaturated soils which is based on the matric suction (ua − uw), respectively. These latter quantities
three-dimensional formulation proposed by Dakshanamurthy are the stress state variables commonly used to describe the
et al. Nevertheless, Wong et al. assumed that pore-air pres- volume change of unsaturated soils under plane-strain condi-
sure is atmospheric and remains unchanged during consoli- tions. Moreover, ua and uw are the pore-air and pore-water
dation. As a consequence, the air continuity equation was pressure, respectively. Similarly, the constitutive relations for
not considered in the analysis and the solution was obtained the water and air phases are
by solving simultaneously the equilibrium equation and the
dVw  σ + σz 
water continuity equation via the finite element method. [2] = m1w d  x − ua  + m 2w d (ua − uw)
In this paper, following the formulation developed by V  2 
Dakshanamurthy et al. (1984), a solution is presented for
coupled consolidation in unsaturated soils when both the air dVa  σ + σz 
[3] = m1a d  x − ua  + m 2a d (ua − uw)
phase and the water phase are continuous. A numerical pro- V  2 
cedure is proposed that allows the solution to a coupled con-
solidation problem to be obtained using the finite element where dVw/ V is the volume change of water in the soil, m1w
method without demanding great computational efforts. The and m 2w are the coefficients of water volume change, dVa / V
procedure makes use of the Fourier transform to formally re- is the volume change of air in the soil, and m1a and m 2a are
move the dependence of the field variables on the spatial co- the coefficients of air volume change. A negative sign is
ordinates in the horizontal plane. As a result, the soil system usually assigned to these coefficients to indicate that an in-
can be discretized using simple one-dimensional elements. A crease in the stress state variables causes a volume decrease.
similar methodology was previously developed by Booker Moreover, the continuity requirement leads to the following
and Small (1982) to analyse consolidation in saturated soils. relations:

© 2004 NRC Canada


Conte 601

[4] m1s = m1w + m1a 2 (1 − 2ν) ms


[10] σx = ua + s
(εx + a sεv) − 2s (ua − uw)
[5] m 2s = m 2w + m 2a m1 m1

(Fredlund and Morgenstern 1976). 2 (1 − 2ν) m 2s


[11] σ z = ua + ( ε z + a s ε v) − (ua − uw)
If the soil is assumed to behave as an isotropic linear elas- m1s m1s
tic material, the normal stresses can be related to the normal
strains and matric suction using the relations proposed by In addition, as known, the shear stress is
Dakshanamurthy et al. (1984): [12] τxz = G γ xz
[6] σx − ua = 2G (εx + a sεv) − β (ua − uw)
where τxz and γ xz are the shear stress and shear strain in the
[7] σ z − ua = 2G (ε z + a sεv) − β (ua − uw) x–z plane, respectively. The shear and normal strains are re-
lated to displacement by the well-known relations
where G is the shear modulus of the soil; εx and ε z are the
normal strains in the x and z directions, respectively; εv is ∂u ∂w ∂u ∂ w
[13] εx = ; εz = ; γ xz = +
the volumetric strain (εv = εx + ε z); a s = ν/(1 − 2ν), where ν ∂x ∂z ∂z ∂x
is Poisson’s ratio of the soil; and β = 2G (1 + ν) /(1 − 2ν)Hs,
where Hs is a modulus relating the change of εv to a change in which u and w denote the components of displacement in
in matric suction. Under plane-strain conditions, we can the x and z directions, respectively. It should be noted that,
write in general, the material parameters in the aforementioned
constitutive relations are functions of the stress state vari-
1 − 2ν ables. Therefore, these relations should be used in an incre-
[8] m1s =
G mental sense. Under the assumptions made and on the basis
of the these relations, the equations governing the coupled
2 (1 + ν)
[9] m 2s = consolidation are (1) the equilibrium equations, (2) the con-
Hs tinuity equation for the water phase, and (3) the continuity
equation for the air phase.
(Fredlund and Rahardjo 1993). Substituting eqs. [8] and [9]
into eqs. [6] and [7] yields

(1) Equilibrium equations — Under plane-strain conditions, the equilibrium equations are:
 ∂σx ∂τxz
 ∂x + ∂z = 0
[14]  ∂τ ∂σ z
 xz + =0
 ∂x ∂z
which, after substituting eqs. [10]–[13] and rearranging the terms, can be written as
  2  2   2 
 2bs  ∂ u + a s  ∂ u + ∂ w  + bs  ∂ u + ∂ w  − cs ∂ (ua − uw) + ∂ua = 0
2 2

 ∂ 2  ∂ 2
∂ ∂   ∂ 2
∂ ∂  ∂x ∂x
 x  x z x   z z x 
[15] 
 2b  ∂ 2w + a  ∂ 2u + ∂ 2w  + b  ∂ 2u + ∂ 2w  − c ∂ (ua − uw) + ∂ua = 0
 s  ∂z 2 s 2 
 s 2  s
   ∂x ∂z ∂z   ∂z ∂x ∂x  ∂z ∂z

where bs = (1 − 2ν)/m1s and cs = m 2s /m1s.


(2) Continuity equation for the water phase — The continuity equation for the water phase requires that the rate of the water
volume change be equal to the net flow of water through a soil element of volume V:
∂  Vw  ∂v x ∂v z
[16] −  = +
∂t  V  ∂x ∂z
where t is time; Vw is the volume of water in the soil; and vx and vz indicate the water flow rate across a unit area of the
soil element in the x and z directions, respectively. Introducing Darcy’s law and the relation for the hydraulic head, hw =
z + uw/γw, the term on the right-hand side of eq. [16] becomes
∂v x ∂v z ∂  k ∂u  ∂  k ∂u  ∂k
[17] + = −  w w  −  w w  − w
∂x ∂z ∂x  γ w ∂x  ∂z  γ w ∂z  ∂z

where γw is the unit weight of water, and kw is the coefficient of permeability with respect to the water phase. In eq. [17]
the same coefficient of permeability is used for the water flow in the x and z directions, since the soil is assumed to be
isotropic for simplicity. The term on the left-hand side of eq. [16] can be obtained by differentiating the water phase con-
stitutive relation (eq. [2]) with respect to time, under the assumption that the coefficients of volume change remain con-

© 2004 NRC Canada


602 Can. Geotech. J. Vol. 41, 2004

stant during consolidation. This assumption may be considered acceptable for practical purposes when the stress
increments are small. Therefore, we can write
∂  Vw  ∂  σ + σz  ∂
[18]   = m1w  x − ua  + m 2w (ua − uw)
∂t  V  ∂t  2  ∂t

Substituting eqs. [10] and [11] into eq. [18] and taking into account eq. [13], after some rearrangements the following
equation is obtained:
∂  Vw  ∂  ∂u ∂w  ∂
[19]   = aw  +  + bw (ua − uw)
∂t  V  ∂t  ∂x ∂z  ∂t
where a w = m1w/ m1s and bw = m 2w − m1wm 2s / m1s. Lastly, substituting eqs. [17] and [19] into eq. [16] leads to the continuity
equation for the water phase:
∂  ∂u ∂w  ∂  ∂k ∂u ∂k ∂u   ∂ 2u ∂ 2uw  ∂kw
[20] aw  +  + bw (ua − uw) = dw  w w + w w  + fw  2w + +
∂t  ∂x ∂z  ∂t  ∂x ∂x ∂z ∂z   ∂x ∂z 2  ∂z

in which dw = 1/γw and fw = kw/γw.


(3) Continuity equation for the air phase — The continuity equation for the air phase requires that the rate of air mass change
in the soil be equal to the net air mass flow through the soil element:
∂  ρ aVa  ∂Jx ∂J z
[21] −  = +
∂t  V  ∂x ∂z
in which Va is the volume of air in the soil; and Jx and Jz denote the mass rate of air flowing across a unit area of the soil
in the x and z directions, respectively. Jx and Jz can be expressed as
ka ∂ua ka ∂ua
[22] Jx = − ; Jz = −
g ∂x g ∂x
in which g is the acceleration due to gravity; ka is the coefficient of permeability for the air phase; and ρa is the air den-
sity, which is given by the ideal gas law as
[23] ρ a = ηua
where η = ωa /RT , in which ωa is the molecular mass of air; R is the universal gas constant; T indicates the absolute tem-
perature; and ua is the absolute air pressure (i.e., ua = ua + uab + uatm, where uab is the initial air pressure existing in the
soil and uatm is the atmospheric pressure). Because of air compressibility, we can write
∂  ρ aVa  ∂  V  V ∂ρ a
[24]   = ρa  a  + a
∂t  V  ∂t  V  V ∂t
where
Va
[25] = (1 − S ) n
V
in which n is the soil porosity, and S is the degree of saturation of the soil. Substituting eqs. [22], [23], and [24] into
eq. [21] yields the following equation:
∂  Va  1 Va ∂ua 1 ∂  ka ∂ua  1 ∂  ka ∂ua 
[26] −  = − −
∂t  V  ua V ∂t ηua ∂x  g ∂x  ηua ∂z  g ∂z 

Moreover, using Boyle’s law and taking into account eq. [25], we can write
1 Va 1
[27] = 2 uab(1 − S b ) n b
ua V ua
where uab is the initial absolute air pressure in the soil; and Sb and nb are the initial degree of saturation and the initial po-
rosity of the soil, respectively. The term ∂(Va / V )/∂t can also be expressed as the derivative of the air phase constitutive re-
lation (eq. [3]) with respect to time, under the assumption that the coefficients of volume change remain constant during
consolidation:
∂  Va  ∂  σ + σz  ∂
[28]   = m1a  x − ua  + m 2a (ua − uw)
∂t  V  ∂t  2  ∂t

© 2004 NRC Canada


Conte 603

Substituting eqs. [27] and [28] into eq. [26] and taking into account eqs. [10], [11], and [13], after some rearrangements
the partial differential equation for the air phase can be obtained:
∂  ∂u ∂w  ∂ ∂u  ∂k ∂u ∂k ∂u   ∂ 2u ∂ 2u 
[29] aa  +  + ba (ua − uw) = −ca a + da  a a + a a  + fa  2a + 2a 
∂t  ∂x ∂z  ∂t ∂t  ∂x ∂x ∂z ∂z   ∂x ∂z 

in which
m1a  m am s  1 1
aa = , ba =  m 2a − 1 s 2  , ca = uab(1 − S b ) n b, da = ,
m1s  m1  ua2 ηua g

and
ka
fa =
ηua g

Equations [15], [20], and [29] form the set of differential Fig. 1. Finite element discretization scheme. B, width of founda-
equations governing the coupled consolidation in unsatu- tion; H, thickness of soil layer.
rated soils under plane-strain conditions; they are similar
to those presented by Dakshanamurthy et al. (1984) and
Fredlund and Rahardjo (1993). After solving these equa-
tions, the water and air pressures and the soil displacements
are simultaneously obtained as a function of x, z, and t. It
should be noted that the aforementioned set of equations is,
in general, nonlinear because the soil properties involved de-
pend on the current stress state variables. Moreover, some
coefficients in eq. [29] depend on the absolute air pressure
ua . As a consequence, an iterative procedure should be set up
to find the solution. Nevertheless, when the excess pore-air
pressure is small or, as often occurs, rapidly dissipates dur-
ing consolidation, ua may be considered constant and the use
of such an iterative procedure is not essential. Lastly, with
regard to the boundary conditions, when the soil is subjected
to prescribed surface tractions (Fig. 1), we may put at the Numerical solution
loaded upper surface When considering soil systems of infinite lateral extent,
from a computational point of view it is convenient to trans-
[30] σz = q z and τxz = q x
form the governing differential equations by an application
where qz and qx denote the normal and tangential surface of the Fourier transform. For the field variables involved in
tractions, respectively. Moreover, the base of the soil may be the problem at issue, this transformation and its inverse can
supposed to be rough and rigid be defined, respectively, as

[31] u=0 and w =0 1
[35] (U, W , Uw, Ua ) = ∫ (u, iw, iuw, iua) e−iax dx
2π −∞
or smooth and rigid
[32] τxz = 0 and w =0 ∞

In addition, the upper and lower surfaces may be considered


[36] (u, w, uw, ua ) = ∫ (U,−iW,−iUw,−iUa) eiax dα
−∞
permeable
where α is the wave number, which describes the variation
[33] uw = 0 and ua = 0 of the variables in the x direction; and i = (–1)1/2. In
or impermeable eqs. [35] and [36] and in the following equations, the trans-
formed field variables are indicated by an upper case letter.
∂uw ∂ua Moreover, a different transformation formula is used for u
[34] =0 and =0
∂z ∂z and for the other variables for the sake of convenience.
Under the assumption that the coefficients of permeability
In reality, the boundary conditions for the fluid phases are do not change with changes in x and z, after substituting
in general more complex than those expressed by eqs. [33] eq. [36] into eqs. [15], [20], and [29] and rearranging the
and [34]. For example, at the upper surface the influence terms, the equations governing the coupled consolidation in
of the specific climatic conditions should be accounted unsaturated soils are transformed into the following differen-
for. This has not been considered in the present study. tial equations:
© 2004 NRC Canada
604 Can. Geotech. J. Vol. 41, 2004

  ∂W   ∂ 2U ∂W 
 2bs −α 2(1 + α s) U + α a s  + bs  2 + α − α cs(Ua − Uw) + αUa = 0
  ∂z   ∂z ∂z 
      
 2bs −(1 + α s) ∂ W + α a s ∂U  + bs  α 2W + α ∂U  + cs  ∂Ua − ∂Uw  − ∂Ua = 0
2

  ∂z 2
∂z   ∂z   ∂z ∂z  ∂z
[37] 
 a w  α ∂U − ∂ W  + bw  ∂Uw − ∂Ua  − fw  α 2Uw − ∂ Uw  = 0
2 2

  ∂t ∂z∂t   ∂t ∂t  
 ∂z 2 

  ∂U ∂ 2W   
 + ba  ∂Uw − ∂Ua  − fa  α 2Ua − ∂ Ua  − ca ∂Ua = 0
2
 a a  α −   2 
  ∂t ∂z∂t   ∂t ∂t   ∂z  ∂t

The boundary conditions (eqs. [30]–[34]) must also be ex- technique in conjunction with a time integration process,
pressed in terms of Fourier transform. To this end, the fol- where the solution is found from that at the previous time.
lowing transformation formulas can be used for the surface Such a time integration process is appropriate, especially
tractions: when the soil properties have to be updated during consoli-
∞ dation in accordance with the current stress state. In this
1 work, simple quadratic shape functions have been adopted
[38] (Qx , Qz) = ∫ (q x, iq z) e−iax dx
2π −∞ along with a central difference approximation as a time-
marching scheme (Booker and Small 1975). The finite ele-
and ment formulation for eq. [37] leads to a set of algebraic

equations that may be written in matrix form as (Aromataris
2002)
( q x , q z) = ∫ (Qx,−iQz) eiax dα
−∞ [45] CF = R
where Qx and Qz denote the direct Fourier transforms of the where C is a matrix, the terms of which depend on the soil
prescribed surface tractions. Therefore, taking into account properties involved in eq. [37], the thickness of the elements,
eqs. [10]–[13] and [38], eq. [30] gives the wave number α, and the time step ∆t; F is the vector of
 ∂W the unknown variables that are the nodal transformed dis-
 ∂W 
[39] Qz = Ua + 2bs  + a s  −αU +  − cs(Ua − Uw) placements U and W and the nodal transformed pressures Uw
 ∂z  ∂z  and Ua at time t; and R is the vector of the known terms con-
taining the nodal transformed variables at the previous time
and and the prescribed transformed surface tractions. It should
 ∂U  be emphasized that the solution obtained from eq. [45] is a
[40] Qx = bs  + αW  function of the transform variable α; inversion is therefore
 ∂z  required to recover the original physical variables. In other
In addition, the other boundary conditions (eqs. [31]–[34]) words, the present method works in the wave number do-
take the form main, and consequently the solution requires first that the
external loads be expanded in harmonic components and
[41] U=0 and W =0 then that eq. [45] be solved for each harmonic component to
obtain the nodal transformed displacements and pore pres-
∂U
[42] + αW = 0 and W =0 sures as a discrete function of α. Once these transformed
∂z variables are found, the actual displacements and pore pres-
sures at any desired position can be calculated by numerical
[43] Uw = 0 and Ua = 0
inversion of the Fourier transform (eq. [36]). This procedure
∂Uw ∂Ua has to be used for any time step, as shown in the flow chart
[44] =0 and =0 of Fig. 2. The soil properties involved in the governing equa-
∂z ∂z
tions may be supposed to be constant during consolidation
The main advantage of eq. [37] over the starting equations or be updated at each time step in accordance with the cur-
(eqs. [15], [20], and [29]) is that for a given value of α the rent stress state using appropriate relations and experimental
unknown quantities U, W, Uw, and Ua only depend on z and t. data (Fredlund and Rahardjo 1993; van Genuchten 1980;
From a computational point of view, this makes the solution Fredlund et al. 1994). Although the flow chart in Fig. 2 re-
easier to obtain, since the problem, originally two dimen- fers to an external load that is supposed constant in time,
sional, in effect is reduced to a one-dimensional problem. As loading of more general time–history could be considered.
a consequence, to solve eq. [37] numerically, only the soil In addition, the use of the Fourier transform allows us to ac-
thickness needs to be discretized by a finite number of one- count for an arbitrary spatial distribution of the surface load-
dimensional elements, as shown in Fig. 1. The solution may ing. To simplify the solution, it is also possible to express
be advantageously obtained using the standard finite element the field variables as periodically spaced functions in the

© 2004 NRC Canada


Conte 605

Fig. 2. Solution flow chart for coupled consolidation. Fig. 3. Representation of a strip load by Fourier series.

Fig. 4. Time (Ts) – settlement (ws) behaviour of strip loading on


a saturated soil layer (adapted from Booker and Small 1982).

x direction using the Fourier series instead of the Fourier


transform (Fig. 3). In this case, however, the spatial period
must be chosen wide enough to minimize the effects of the
interaction among the load areas and of the rigid and imper-
vious vertical walls fictitiously introduced whenever the
functions periodically repeat themselves, on the solution.
The writer has found that using a value of the semi-period,
L, greater than five times the foundation width, B (Fig. 3),
and at least 20 terms of the Fourier series usually provides
satisfactory results for practical purposes. The results shown
in the following have been obtained using such an approxi-
mation.
The accuracy of the proposed procedure can be assessed
by comparing the results with those obtained using other (1970) and Booker (1974) under the assumption that the
theoretical solutions. Nevertheless, the writer is not aware of base is also smooth or rough, respectively. Specifically,
existing problems concerning the coupled consolidation in Fig. 4 shows the dimensionless vertical displacement, ws =
unsaturated soils induced by external loading, in which the 2Gw/qB, beneath the central point of the strip loading,
same constitutive relations considered in the present study against a time factor defined as Ts = 2Gkwt/γwH2. As shown
have been employed. Therefore, comparisons have been car- in Fig. 4, the present procedure provides results in very
ried out considering only examples involving saturated soils. close agreement with those obtained by Gibson et al. and
It should be noted that the solution to such a problem can be Booker.
obtained by eq. [37], setting S b = 100% and m1s = m 2s = m1w =
m 2w = m v, where mv denotes the coefficient of volume change Example of coupled consolidation in
for the saturated soil, which is given by eq. [8]. As an exam- unsaturated soils
ple, the results presented in Fig. 4 concern a saturated soil
layer subjected to a uniform strip load, q, which is constant To point out some features of the coupled consolidation in
in time. The ratio of the layer thickness to the foundation unsaturated soils, a soil layer resting on a rigid, rough, and
width is H/B = 0.5. Free drainage is allowed across the up- impervious base is considered as an example. The layer is
per surface of the layer, whereas the base is impervious and subjected to an external strip load of uniform intensity, q,
rigid. Poisson’s ratio is assumed to be zero. The solution which is held constant in time; the foundation width, B, is
to this problem was obtained analytically by Gibson et al. set equal to 3H/2, where H is the layer thickness. The soil

© 2004 NRC Canada


606 Can. Geotech. J. Vol. 41, 2004

Fig. 5. Effect of ka/kw on the time (T) – settlement (w*) behav- Fig. 6. Effect of ka/kw on pore-water pressure dissipation with
iour of a strip foundation on an unsaturated soil layer. time.

properties are as follows: nb = 50%, Sb = 80%, and ν = 0.33;


moreover, it is assumed that m1s = –2.5 × 10–4 kPa–1,
m 2s /m1s = 0.4, m1w /m1s = 0.2, and m 2w /m1w = 4. Lastly, four dif-
ferent values of the ratio ka/kw are considered, namely 0.1, 1, Fig. 7. Effect of ka/kw on pore-air pressure dissipation with time.
10, and 105. For the sake of simplicity, all the aforemen-
tioned soil parameters are assumed to remain constant dur-
ing consolidation, and the initial absolute pore-air pressure
is supposed to be atmospheric. Figure 5 shows the time his-
tory of the settlement at the centre of the loaded area. Both
settlement and time are expressed in a dimensionless form
as w* = w/m1sqB and T = kw t/ m1sγ wH 2, respectively. As can
be noted, when ka is less than kw (ka/kw = 0.1) the progress
of settlement with time presents a classic S-shaped curve
similar to that predicted by the consolidation theory for satu-
rated soils. On the contrary, when ka > kw, which most often
is the case (Rahardjo and Fredlund 1995), a double S-shaped
settlement curve can be observed. This is because the early
stage of consolidation is mainly governed by the dissipation
of air pressures faster than that of water pressures, whereas
the final stage is due entirely to the gradual dissipation of
excess water pressure. Similar behaviour was also observed
during oedometer tests on compacted soil samples (Ausilio
and Conte 2003). For the highest value of ka/kw considered,
air pressures dissipate very rapidly and, as a result, a higher
instantaneous settlement occurs, replacing in practice the
first S-shaped settlement curve. This physical interpretation
is also supported by the results of Figs. 6 and 7, where the and 9, where the pore-water pressure at a given depth is ex-
dissipation curves for the water and air pressures calculated pressed as a ratio of its initial value arising at the same
at depth z = 0.5B are shown. Moreover, from Fig. 6 it can depth. Pore-air pressure also exhibits a similar behaviour
also be observed that the Mandel–Cryer effect occurring (Figs. 10 and 11).
during the consolidation in saturated soils (Mandel 1953;
Cryer 1963), which as known is characterized by an increase
in pore-water pressure before dissipation starts, is minimized Uncoupled consolidation for unsaturated
in unsaturated soils. This aspect was pointed out by Wong et soils
al. (1998). In addition, contrary to the case of saturated soils
(Schiffman et al. 1969), the effect at issue appears to be The governing equations of uncoupled consolidation in
much attenuated even when depth increases or Poisson’s ra- unsaturated soils are given by the continuity equations for
tio decreases. These influences are documented in Figs. 8 the water phase and the air phase. Under the same assump-

© 2004 NRC Canada


Conte 607

Fig. 8. Effect of depth on pore-water pressure dissipation with Fig. 10. Effect of depth on pore-air pressure dissipation with
time. uw, pore-water pressure; uwo, initial pore-water pressure at time. ua, pore-air pressure; uao, initial pore-air pressure at the
the same depth. same depth.

Fig. 9. Effect of Poisson’s ratio on pore-water pressure dissipa- Fig. 11. Effect of Poisson’s ratio on pore-air pressure dissipation
tion with time. with time.

tion introduced for the coupled theory, these equations can and equating eq. [26] to eq. [28]:
be obtained substituting eqs. [17] and [18] into eq. [16]:
∂  σx + σ z  ∂ ∂  σx + σ z  ∂
[46] m1w  − ua  + m 2w (ua − uw) [47] m1a  − ua  + m 2a (ua − uw)
∂t  ∂ ∂t  2  ∂t
2  t
 ∂k ∂u ∂k ∂u  ∂ua  ∂k ∂u ∂k ∂u 
= dw  w w + w w  = −ca + da  a a + a a 
 ∂x ∂x ∂z ∂z  ∂t  ∂x ∂x ∂z ∂z 
 ∂ 2u ∂ 2uw  ∂kw  ∂ 2u ∂ 2u 
+ fw  2w + + + fa  2a + 2a 
 ∂x ∂z 2  ∂z  ∂x ∂z 

© 2004 NRC Canada


608 Can. Geotech. J. Vol. 41, 2004

Moreover, assuming, as an approximation, that the total stresses remain constant during consolidation, eqs. [46] and [47] be-
come
 ∂u ∂u  ∂k ∂u ∂k ∂u   ∂ 2u ∂ 2uw  ∂kw
 gw a = hw w + dw  w w + w w  + fw  2w + +
 ∂t ∂t  ∂x ∂x ∂z ∂z   ∂x ∂z 2  ∂z
[48] 
 g ∂ua = h ∂uw + d  ∂ka ∂ua + ∂ka ∂ua  + f  ∂ ua + ∂ ua 
2 2
a
 a ∂t ∂z 2 
a a
 ∂t  ∂x ∂x ∂z ∂z   ∂x
2

where gw = (m 2w − m1w), hw = m 2w, ga = (ca + m 2a − m1a ), and Fig. 12. Lateral distributions of the initial excess pore pressures
ha = m 2a . Solving eq. [48], water and air pressures are calcu- resulting from the application of strip loading on an unsaturated
lated at any time and position. The boundary conditions may soil.
be expressed by eqs. [33] and [34], whereas the excess pore
pressures generated at the instant when the external load is
applied represent the initial conditions (Lloret and Alonso
1980; Fredlund and Rahardjo 1993). To this purpose, equa-
tions are given in Appendix A to evaluate in an approximate
manner the initial excess pore-air and pore-water pressures
in response to a total stress change resulting from the appli-
cation of strip loading. Typical lateral distributions of the
pore pressures obtained using these equations are shown in
Fig. 12, as an example. If it is again assumed that the coeffi-
cients of permeability do not change with changes in x and
z, after applying a Fourier transform to the field variables in
eq. [48], the transformed equations governing the uncoupled
consolidation in unsaturated soils take the following form:
 ∂U  ∂ 2Uw  ∂U
 gw a + fw  α 2Uw − 2 
− hw w = 0
 ∂ t  ∂z  ∂t
[49] 
 ∂Ua  2 
 α Ua − ∂ Ua  − ha ∂Uw = 0
2
 ga + fa  2 
 ∂t  ∂z  ∂t

This set of equations has to be solved taking into account the


aforesaid boundary and initial conditions, both expressed in
terms of transformed quantities. Application of the finite ele-
ment formulation for eq. [49] yields a set of equations that 1
may be written in matrix form as (Aromataris 2002) [51] εz = [(σ z − ua ) + a s(σ z − σx ) + cs(ua − uw)]
2bs(1 + 2a s)
[50] AD = S
in which the total stresses σx and σz can be evaluated using
where A is a matrix, the terms of which depend on the soil eqs. [A12] and [A13] in Appendix A. Lastly, the settlement
properties involved in eq. [49], the thickness of the elements, is given by
the wave number α, and the time step ∆t; the vector D con- H
tains the unknown variables that are the nodal transformed [52] w = ∫ ε z dz
pressures Uw and Ua at time t; and S is the vector of the 0
known terms containing the nodal transformed variables at
the previous time. Once the transformed variables are found, where H denotes the thickness of the compressible soil layer.
the actual pore pressures at any desired position can be cal-
culated by numerical inversion of the Fourier transform. The Comparison between coupled and
calculating procedure is shown in the flow chart of Fig. 13. uncoupled consolidation
As can be noted, the variables that have to be expanded in
harmonic components by Fourier transform at the beginning The main difference emerging when the results from cou-
of this procedure are different from those indicated in Fig. 2. pled and uncoupled theories of the consolidation in saturated
For the uncoupled analysis, in fact, these quantities are the soils are compared is the early increase in pore-water pres-
initial excess pore pressures resulting from the loading ap- sure due to the Mandel–Cryer effect, which cannot be repro-
plication. After finding the actual field variables, the normal duced using the simple uncoupled approach (Schiffman et
strain ε z is calculated in uncoupled manner using the follow- al. 1969). Nevertheless, as pointed out by Wong et al. (1998)
ing expression that derives from eqs. [10] and [11]: and supported by the results previously shown in this paper,

© 2004 NRC Canada


Conte 609

Fig. 13. Solution flow chart for uncoupled consolidation. Fig. 14. Comparison of coupled and uncoupled consolidation.
Pore-water pressure dissipation at z = 0.5B.

Fig. 15. Comparison of coupled and uncoupled consolidation.


Pore-air pressure dissipation at z = 0.5B.

such an effect should be minimized in unsaturated soils.


Consequently, it is worth ascertaining whether the uncou-
pled approach can provide a valid approximation to the cou-
pled theory, for both pore pressure dissipation and
settlement rate occurring during consolidation. To this end,
the same example as that in the previous parametric study is
again considered. Figures 14–16 present a comparison be-
tween the results derived from the coupled and uncoupled
theories in terms of excess pore pressure dissipation and set-
tlement evolution with time for the case when ka/kw = 10.
Specifically, Figs. 14 and 15 show the time – pore pressure
curves at a given depth for the air and water phases, and
Fig. 16 shows the degree of settlement Us versus time T
which is defined as
w(t) − w o
[53] Us =
wc − wo
where w(t) is the settlement at time t, wo is the immediate equations in Appendix A. Examination of Figs. 14–16
settlement, and wc is the long-term settlement. Settlement is shows that the simple uncoupled theory predicts very well
calculated for the centre of the loaded area. In other words, the pore pressures obtained by the more rigorous coupled so-
Us represents the settlement rate for the foundation consid- lution for both the air and water phases. The comparison be-
ered. For the uncoupled approach, two solutions are given; tween the two theories in terms of degree of settlement is
they are pointed out in Figs. 14–16 as uncoupled consolida- also satisfactory (Fig. 16). This should also support the as-
tion and approximate uncoupled consolidation and differ, sumption, on which the uncoupled approach is based, that
essentially, for the initial conditions used to solve eq. [50]. the total stresses induced by the considered load remain
In the former solution, the pore pressures calculated at the essentially unchanged during consolidation. Moreover, the
first time instant by the coupled theory have been used as results in Figs. 14–16 show that even the approximate un-
initial conditions for the uncoupled analysis; in the latter so- coupled solution gives, in general, results in good agreement
lution, on the contrary, the initial pore pressure at any depth with those from the coupled theory, especially with regard to
has been evaluated, in an approximate manner, using the settlement rate.

© 2004 NRC Canada


610 Can. Geotech. J. Vol. 41, 2004

Fig. 16. Comparison of coupled and uncoupled consolidation. Ausilio, E., and Conte, E. 1999. Settlement rate of foundations on
Degree of settlement versus time. unsaturated soils. Canadian Geotechnical Journal, 36: 940–946.
Ausilio, E., and Conte, E. 2003. Remarks on consolidation in un-
saturated soils from experimental results. In Proceedings of the
International Conference: From Experimental Evidence towards
Numerical Modelling of Unsaturated Soils, Weimar, Germany,
18–19 September 2003. Springer-Verlag, Berlin. In press.
Barden, L. 1965. Consolidation of compacted and unsaturated
clays. Géotechnique, 15(3): 267–286.
Biot, M.A. 1941. General theory of three-dimensional consolida-
tion. Journal of Applied Physics, 12(2): 155–168.
Booker, J.R. 1974. The consolidation of a finite layer subject to
surface loading. International Journal of Solids and Structures,
10: 1053–1065.
Booker, J.R., and Small, J.C. 1975. An investigation of the stability
of numerical solution of Biot’s equations of consolidation. Inter-
national Journal of Solids and Structures, 11: 907–917.
Booker, J.R., and Small, J.C. 1982. Finite layer analysis of consoli-
dation. I. International Journal for Numerical and Analytical
Methods in Geomechanics, 6: 151–171.
Chang, C.S., and Duncan, J.M. 1983. Consolidation analysis for
partly saturated clay by using an elastic–plastic effective stress –
strain model. International Journal for Numerical and Analytical
Methods in Geomechanics, 7: 39–55.
Conte, E. 1998. Consolidation of anisotropic soil deposits. Soils
and Foundations, 38(4): 227–237.
Conclusions Cryer, C.W. 1963. A comparison of the three-dimensional consoli-
A method has been presented for the analysis of coupled dation theories of Biot and Terzaghi. Quarterly Journal of Me-
and uncoupled consolidation in unsaturated soils under chanics and Applied Mathematics, 16: 401–412.
plane-strain conditions. The method makes use of the Fou- Dakshanamurthy, V., and Fredlund, D.G. 1980. Moisture and air
rier transform to represent the field variables in the horizon- flow in an unsaturated soil. In Proceedings of the 4th Interna-
tal plane. The solution to the transformed differential tional Conference on Expansive Soils, Denver, Colo., 16–
equations is obtained in the finite element fashion, dis- 18 June 1980. Edited by D.R. Snethen. American Society of
Civil Engineering, New York. Vol. 1, pp. 514–532.
cretizing the soil system by simple one-dimensional ele-
Dakshanamurthy, V., Fredlund, D.G., and Rahardjo, H. 1984. Cou-
ments. Moreover, a procedure has been suggested to
pled three-dimensional consolidation theory of unsaturated po-
evaluate, in an approximate manner, the excess pore pres-
rous media. In Proceedings of the 5th International Conference
sures in response to the application of an external load. This on Expansive Soils, Adelaide, South Australia, 21–23 May
latter procedure may be employed to determine the initial 1984. American Society of Civil Engineering, New York.
conditions for solving the equations governing the uncou- pp. 99–103.
pled consolidation. The theoretical analyses carried out in Davis, E.H., and Poulos, H.G. 1972. Rate of settlement under two-
this study have pointed out that the settlement evolution with and three-dimensional conditions. Géotechnique, 22(1): 95–114.
time during a consolidation process in unsaturated soils Fredlund, D.G., and Hasan, J.U. 1979. One-dimensional consolida-
could present a double S-shaped curve that is different from tion theory: unsaturated soils. Canadian Geotechnical Journal,
that predicted by the consolidation theory in saturated soils. 16: 521–531.
This may occur when the permeability coefficient for the air Fredlund, D.G., and Morgenstern, N.R. 1976. Constitutive relations
phase is greater than that for the water phase, which most for volume change in unsaturated soils. Canadian Geotechnical
often is the case. In these circumstances, in fact, the early Journal, 13: 386–396.
stage of consolidation is affected by the faster dissipation Fredlund, D.G., and Rahardjo, H. 1993. Soil mechanics for unsatu-
of the air pressures, and the final stage is due entirely to rated soils. John Wiley & Sons, Inc., New York.
the gradual dissipation of the excess pore-water pressures. Fredlund, D.G., Xing, A., and Huang, S. 1994. Predicting the per-
Moreover, it has been shown that the early increase in pore- meability function for unsaturated soils using the soil-water
water pressures prior to dissipation, which in the case of characteristic curve. Canadian Geotechnical Journal, 31: 533–
saturated soils is known as the Mandel–Cryer effect, is atten- 546.
uated in unsaturated soils. Lastly, the results have demon- Ghaboussi, J., and Wilson, E.L. 1973. Flow of compressible fluid
strated that the simple uncoupled approach may, in general, in porous elastic solids. International Journal for Numerical
provide a valid approximation to the coupled theory. Methods in Engineering, 5: 419–442.
Gibson, R.E., Schiffman, R.L., and Pu, S.L. 1970. Plane strain and
axially symmetric consolidation of a clay layer on a smooth
References
impervious base. Quarterly Journal of Mechanics and Applied
Aromataris, T. 2002. Coupled and uncoupled consolidation in par- Mathematics, 23: 505–520.
tially saturated soils. Dissertation, Faculty of Engineering, Uni- Jumikis, A.R. 1969. Theoretical soil mechanics. Van Nostrand
versity of Calabria, Cosenza, Italy. [In Italian.] Reinhold Company, New York.

© 2004 NRC Canada


Conte 611

Lloret, A., and Alonso, E.E. 1980. Consolidation of unsaturated Vwb ∆Vw  ∆Vw  Vvb
[A3] S= + =  Sb +
Vvb  Vv
soils including swelling and collapse behaviour. Géotechnique,
30(4): 449–477.
Vv Vv 
Lloret, A., Gens, A., Batlle, F., and Alonso, E.E. 1987. Flow and
deformation analysis of partially saturated soils. In Groundwater Substituting eqs. [A2] and [A3] into eq. [A1] and setting
Effects in Geotechnical Engineering: Proceedings of the 9th Eu- Vv = Vvb + ∆Vv leads to
ropean Conference on Soil Mechanics and Foundation Engi-
neering, Dublin, Ireland, 31 August – 9 September 1987. A.A. [A4] uab(1 − S b + hS b) Vvb = ua (Vvb + ∆Vv)
Balkema, Rotterdam, The Netherlands. Vol. 2, pp. 565–568.  ∆Vw 
Mandel, J. 1953. Consolidation des sols (étude mathématique). −  S b + uaVvb(1 − h)
Géotechnique, 3: 287–299.  Vvb 
Olson, R.E. 1986. State-of-the-art: consolidation testing. In Con-
solidation of soils, testing and evaluation. Edited by R.N. Yong Dividing all the terms by the total volume of the soil, V, and
and F.C. Townsend. American Society for Testing and Materials introducing the soil porosity nb = Vvb/V, eq. [A4] becomes
(ASTM), Philadelphia, Pa., Special Technical Publication STP
892, pp. 7–70.  ∆Vv 
[A5] uab(1 − S b + hS b) n b = ua  n b + 
Rahardjo, H., and Fredlund, D.G. 1995. Experimental verification  V 
of the theory of consolidation for unsaturated soils. Canadian
Geotechnical Journal, 32: 749–766.  ∆Vw 
−  S b + uan b(1 − h)
Rendulic, L. 1936. Porenziffer und porenwasserdruck in tonen. Der
 Vvb 
Bauingenieur, 17: 559–564.
Schiffman, R.L., Chen, A.T.-F., and Jordan, J.C. 1969. An analysis
of consolidation theories. Journal of the Soil Mechanics and
Under the assumption that soil particles are incompressible,
Foundations Division, ASCE, 95(SM1): 285–312. the term ∆Vv/V coincides with the volumetric strain of the
Terzaghi, K. 1925. Erdbaumechanik auf Bodenphysikalischer soil. Therefore, this term may be expressed by the constitu-
Grundlage. Leipzig Deuticke, Vienna. tive equation for the soil (Fredlund and Morgenstern 1976)
Thomas, H.R., and He, Y. 1997. A coupled heat–moisture transfer as
theory for deformable unsaturated soils and its algorithmic im- ∆Vv
plementation. International Journal for Numerical Methods in [A6] = m1s∆(σ − ua ) + m 2s ∆(ua − uw)
Engineering, 40: 3421–3441. V
van Genuchten, M.T. 1980. A closed-form equation of predicting
where, under plane-strain conditions, ∆σ is given by (σx +
the hydraulic conductivity of unsaturated soils. Soil Science So-
σz)/2; and m1s and m2s are the coefficients of volume change
ciety of America Journal, 44: 892–898.
of the soil with respect to a change in total stress [(σx +
Verruijt, A. 1969. Elastic storage of aquifers. In Flow through po-
rous media. Edited by R.J.M. de Wiest. Academic Press, New
σz)/2] – ua and matric suction (ua – uw), respectively. More-
York. pp. 331–375. over, owing to water compressibility we can write
Wong, T.T., Fredlund, D.G., and Krahn, J. 1998. A numerical study ∆Vw
of coupled consolidation in unsaturated soils. Canadian Geo- [A7] = −n b S bC w∆uw
technical Journal, 35: 926–937.
V
and
Appendix A: Equations for an evaluation of ∆Vw
[A8] = −S bC w∆uw
excess pore pressures in response to a Vvb
total stress change
where Cw is water compressibility, and ∆uw is the pore-water
Referring to the condition before loading and that at the pressure change. In these latter equations the negative sign is
instant when an external load is applied, Boyle’s law gives used to indicate that volume decreases as pore-water pres-
sure increases. Substituting eqs. [A6] and [A8] into eq. [A5]
[A1] uabVab = uaVa and setting ∆ua = ∆ua = ua − uab, after some rearrangements
the following equation is found to evaluate the change in
where ua and Va are the absolute air pressure and the volume pore-water pressure ∆uw and pore-air pressure ∆ua in re-
of air in the soil, respectively; and the subscript b indicates sponse to a total stress change ∆σ:
the condition before loading. The volume of air can be ex-
pressed as n b 
 (1 − S b + hS b) + m 2 − m1  ∆ua
s s
[A9]
[A2] Va = (1 − S + hS ) Vv  ua 
− [m 2s − S bn bC w( 1 − h)] ∆uw = −m1s∆σ
in which h is the volumetric coefficient of solubility for air
in water in accordance with Henry’s law; Vv is the volume of To derive the second equation, the constitutive relation for
voids; and S is the degree of saturation of the soil, which is the water phase (Fredlund and Morgenstern 1976) is used:
given by S = Vw/ Vv, where Vw is the volume of water in the
soil. Setting Vw = Vwb + ∆Vw, the degree of saturation after ∆Vw
[A10] = m1w∆(σ − ua ) + m 2w∆(ua − uw)
loading can be written as V

© 2004 NRC Canada


612 Can. Geotech. J. Vol. 41, 2004

where m1w and m 2w are the coefficients of water volume Fig. A1. Scheme for calculating σx and σz.
change. Equating this equation to eq. [A7] and rearranging
the terms yields

[A11] (m1w − m 2w) ∆ua + (m 2w − S b n bC w) ∆uw = m1w∆σ

A solution to eqs. [A9] and [A11], which are similar to


equations presented by Fredlund and Rahardjo (1993), gives
the excess pore pressures ∆uw and ∆ua resulting from a
change in the total stresses due to the application of an ex-
ternal load. These variables are also indicated in the paper as
uwo and uao, respectively. Since the first term on the left-
hand side of eq. [A9] contains the absolute pore-air pressure,
an iterative procedure is required for the solution. For a strip
load of uniform intensity, q, the total stresses σx and σz ap-
pearing on the right-hand side of eqs. [A9] and [A11] can by
evaluated, as an approximation, using the following expres- q
sions that derive from elasticity theory (Jumikis 1969): [A13] σ z = (2ε + sin 2ε cos 2δ )
π
q where angles ε and δ are indicated in Fig. A1.
[A12] σx = (2ε − sin 2ε cos 2δ )
π

© 2004 NRC Canada

S-ar putea să vă placă și