You are on page 1of 12

Three-dimensional analysis of embankments on soft

soils incorporating vertical drains by finite element method
Jose´ Leita˜o Borges
*
Department of Civil Engineering, Faculty of Engineering, University of Porto, Rua Dr. Roberto Frias, 4200-465 Porto, Portugal
Received 22 October 2003; received in revised form 1 November 2004; accepted 3 November 2004
Available online 15 December 2004
Abstract
Three-dimensional behaviour of an embankment on soft soils incorporating vertical drains is analysed by a numerical model
based on the finite element method. The model, which incorporates the BiotÕs consolidation theory (coupled formulation of the flow
and equilibrium equations) and constitutive relations simulated by the p–q–h critical state model, is applied on both the embank-
ment with vertical drains (three-dimensional analysis) and the same problem but without vertical drains (two-dimensional analysis).
Special emphasis is given to the analysis, during and after the construction period, of the excess pore pressures, settlements, hori-
zontal displacements and stress levels.
Ó 2004 Elsevier Ltd. All rights reserved.
Keywords: Vertical drains; Embankment; Soft soils; Finite element method; Three-dimensional analysis; Consolidation
1. Introduction
The study of embankments on soft soils is one of the
permanent problems of the soil mechanics and has been
analysed by a large number of authors. At the present
time, in spite of all experience obtained over the last dec-
ades, the execution of this kind of constructions still col-
locates diverse and delicate problems that are
determined by the weak geotechnical characteristics of
the foundation soils: (i) low strength significantly limits
the load (embankment height) that is possible to apply
with adequate safety for short term stability; (ii) high
deformability and low permeability determine large set-
tlements that develop slowly as pore water flows and ex-
cess pore pressure dissipates (consolidation).
To design embankments on soft soils it is essential to
take into account the multiple constructive techniques
that allow to solve those problems. The constructive
solutions – usually based on both foundation soil prop-
erties improvement and construction procedures or fill
properties alteration – provide one or more of the fol-
lowing effects: increase of global stability, consolidation
acceleration and decrease of long term settlements [1–3].
The most used technique when the main purpose is to
accelerate the consolidation is the use of vertical drains
in the foundation soils (Fig. 1), which usually determines
drastic decreases of hydrodynamic consolidation time.
In the paper, the geotechnical behaviour of an
embankment on soft soils incorporating vertical drains
(geosynthetic band drains) is analysed during and after
the construction period by a numerical model developed
by Borges [4] for plane strain and axysymmetric analyses
(initial version) and three-dimensional analysis
(improvement included in the program in 2001). Special
emphasis is given to the three-dimensional behaviour of
this kind of constructions by the comparison of the re-
sults of the embankment incorporating the drains
(three-dimensional analysis) with the results of the plane
strain analysis of the same problem without vertical
drains.
0266-352X/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compgeo.2004.11.001
*
Tel.: +351225081928; fax: +351225081940/1440.
E-mail address: leitao@fe.up.pt.
www.elsevier.com/locate/compgeo
Computers and Geotechnics 31 (2004) 665–676
Basically, for the present applications, the model uses
the following theoretical hypotheses: (a) coupled formu-
lation of the flow and equilibrium equations considering
soil constitutive relations (elasto-plastic models) formu-
lated in effective stresses (BiotÕs consolidation theory)
[4–7]; this formulation is applied to all phases of the
problem, both during the embankment construction
and in the post-construction period; (b) utilisation of
the p–q–h critical state model [4–6,8], an associated plas-
tic flow model, to simulate constitutive behaviour of the
foundation and embankment soils.
Fig. 2(a) shows, in the principal effective stress space,
the yield and critical state surfaces of the p–q–h critical
state model. In the p–q coordinate system (where p is
the effective mean stress and q the deviatoric stress),
the yielding function is a ellipse (Fig. 2(b)) and, depend-
ing on the stress path, the p–q–h model simulates hard-
ening behaviour (as shown in Figs. 2(b) and (c) for stress
path 1–2, where e
v
is the volumetric strain and e
d
the
shear strain) or softening behaviour (stress path 3–4,
Figs. 2(b) and (d)).
In this model, the parameter that defines the slope
of the critical state line, M, is not constant as in the
Cam-Clay and Modified Cam-Clay models, but de-
pends on the angular stress invariant, h, and friction
angle of the soil defined in effective terms, /
0
, as
follows:
M ¼
3 sin /
0
ffiffiffi
3
p
cos h þ sin /
0
sin h
: ð1Þ
This defines the Mohr–Coulomb criteria (whose sur-
face in the principal effective stress space is shown in Fig.
2(a)) when M is introduced in the equation of the critical
state line
q ¼ M Á p: ð2Þ
Fig. 2. Yield and critical state surfaces of the p–q–h critical state model in (a) principal effective stress space; (b) p–q coordinate system; (c) stress path
1–2 (hardening behaviour); (d) stress path 3–4 (softening behaviour).
EMBANKMENT
HARD STRATUM
DRAINAGE LAYER
(0.5 to1m)
VERTICAL
DRAINS
SOFT
SOILS
Fig. 1. Vertical drains – acceleration of hydrodynamic consolidation.
666 J.L. Borges / Computers and Geotechnics 31 (2004) 665–676
This is an important characteristic of the p–q–h mod-
el because, as shown by triaxial tests [33], the critical
state of soil depends on h. (Drucker–Prager is the crite-
ria of the Cam-Clay and Modified Cam-Clay models
and does not depend on h).
About embankments on soft ground, in order to ver-
ify accuracy of the finite element program in this kind of
works, Borges [4] compared numerical and field results
of two embankments on soft soils, one constructed up
to failure [9] and the other observed until the end of
the consolidation [10–12]. The accuracy was considered
adequate in both cases, as numerical and field results are
similar, namely in terms of settlements and pore pres-
sures. Only some quantitative differences were observed
in the horizontal displacements, despite an overall qual-
itative similarity too.
In general terms, the verification of the program
was made by comparing numerical results not only
with field results, as said above (which is the adequate
way for complex problems without theoretical solu-
tions), but also with theoretical results of several
closed problems, namely one-dimensional Terzaghi
consolidation, circular load on an elastic foundation
[34], consolidation of a semi-infinite elastic medium
under an uniform surface loading over a circular area
[35], drained and undrained triaxial tests for soils sim-
ulated with critical state models. Comparisons with
numerical results presented by other authors, as Lewis
and Schrefler [5] with CRISP (critical state program),
were also verified.
It should be remembered that the first consistent and
general theory of multi-dimensional consolidation tak-
ing into account interdependence between soil effective
strains and pore water flow (coupled formulation of
the flow and equilibrium equations) was presented by
Biot [13,14]. This theory, which was initially developed
considering soil with isotropic and linear elastic behav-
iour, was posteriorly improved for more complex behav-
iours, namely anisotropy and viscosity [15–17].
After the presentation of BiotÕs consolidation theory,
several authors have applied the finite element method
on consolidation problems using mathematical formula-
tions in which some of the initial hypotheses of Biot are
reformulated (extensions of the BiotÕs theory). Small
et al. [18] considered the plasticity using Mohr–Cou-
lomb model and Desai and Siriwardane [19] and Runes-
son [20] applied critical state models. The variation of
the permeability in the course of the consolidation was
considered by Lewis et al. [21]. Norris [22] extended
the study to the use of constitutive models with kinemat-
ical hardening.
Concerning the use of vertical drains in embankments
on soft soils, Zeng et al. [23] conducted coupled consol-
idation analyses to assess the effectiveness of a method
that calculates an equivalent horizontal permeability
for plane strain analysis in which the vertical drains
are represented as sheets. Hird and Kwok [24] per-
formed a parametric study of a test embankment where
the effect of the vertical drains was considered by
increasing the subsoil permeability by an estimated fac-
tor. Poran et al. [25] defined an equivalent vertical per-
meability for the subsoils by comparing an
axisymmetric analysis of a cylindrical unit cell, with
drainage occurring both horizontally, inwards, and ver-
tically, upwards, with a plane strain unit cell with drain-
age towards the upper boundary only. This equivalent
vertical permeability was used in a plane strain analysis
of a trial embankment.
Because embankments on soft soils incorporating
vertical drains behave three-dimensionally (in terms of
stress and water flow, as shown below in Section 3),
one of the aims of the paper is to achieve a more com-
plex phenomenological interpretation of the 3D gheo-
technical behaviour of this kind of works, by
performing a three-dimensional numerical analysis with
the program developed by the author.
For three-dimensional applications, the program uses
two types of the 20-noded brick element. Fig. 3(a)
shows the type used in the foundation soils (element
with 60 displacement degrees of freedom, at the corners
and at middle of the sides, and with 8 more excess pore
pressure degrees of freedom, at the corners), where con-
solidation analysis is considered. In the fill, it is the 20-
noded brick element with only 60 displacement degrees
of freedom (at the corners and at middle of the sides)
that is used.
Similarly, for two-dimensional analyses, two types
of the six-noded triangular element are considered
(Fig. 4): (i) with 12 displacement degrees of freedom,
at the vertices and at middle of the edges (for fill ele-
ments) and (ii) with 3 more excess pore pressure de-
grees of freedom at the vertices (for foundation
elements).
2. Description of the problem
The problem concerns the construction of a 2 m
height symmetric embankment, with a 10.6 m crest
width, 2/3 (V/H) inclined slopes and very large longitu-
dinal length. The foundation is a 5 m thick saturated
clay layer lying on a rigid and impermeable soil, which
constitutes the lower boundary. The clay is lightly over-
consolidated to 1.8 m depth and normally consolidated
from 1.8 to 5 m. It will be analysed the embankment
with and without vertical band-shaped drains (geosyn-
thetic prefabricated drains) with a 200 · 5 mm
2
section
and installed in a square grid with drain spacing of 2
m. The grid limit is 1.7 m beyond the toe to take up
any lateral spread of excess pore pressures. It is not in-
tended that the band drain reproduces any commercial
product.
J.L. Borges / Computers and Geotechnics 31 (2004) 665–676 667
Fig. 5 shows the finite element mesh used in the three-
dimensional analysis of the embankment incorporating
the vertical drains.
The displacement boundary conditions were defined
taking into account that the soft clay lays on a hard
stratum (y = 0 plane, where displacements are set as
zero in the three directions, x, y and z). One the other
hand, symmetry conditions imply: (i) zero displacement
in x-direction for nodes on the x = 0 plane; (ii) zero
displacement in z-direction for nodes on the z = 0
plane, vertical plane containing one row of drains;
(iii) zero displacement in z-direction for nodes on the
z = 1 m plane, vertical plane equidistant from two rows
of drains in x-direction. Assuming that the horizontal
displacement can be defined as zero at nodes that are
enough distant from the embankment, the plane of
Fig. 3. 3-D finite element used in the program, 20-noded brick element: (a) with 60 displacement degrees of freedom and 8 excess pore pressure
degrees of freedom; (b) with 60 displacement degrees of freedom.
- displacement unknown
- excess pore pressure
unknown
Fig. 4. 2-D finite element used in the program, 6-noded triangular
element: (a) with 12 displacement degrees of freedom and 3 excess pore
pressure degrees of freedom; (b) with 12 displacement degrees of
freedom.
Fig. 5. 3D finite element mesh for the problem with vertical drains.
668 J.L. Borges / Computers and Geotechnics 31 (2004) 665–676
x = 23.9 m was considered as the lateral boundary with
zero displacement in x-direction.
With regard to drainage boundary conditions, excess
pore pressure was set as zero on the ground level (upper
drainage surface), i.e., on the y = 5 m plane, and on the
drainage surfaces defined by the drains considered as
sheets, namely on the following planes: x = 0, x = 2,
x = 4, x = 6, x = 8 and x = 10 m, with y-coordinate
varying from 0 to 5 m and z-coordinate from 0 to 0.1
m (which means that centres of the drains are on the
z = 0 boundary plane and each drain was installed with
its larger dimension, 0.20 m, in z-direction).
The embankment construction was simulated activat-
ing the elements that form the fill layers. Four 0.5 m
height layers were considered and, in order to assess
the drainage effect even during the construction period,
a discontinuous sequence of construction was defined as
indicated in Fig. 6. The first three layers were con-
structed in 3.5 days each, and the fourth in 7 days.
The pause periods, which took place after each layer
construction, were respectively 3.5, 3.5 and 38.4 days.
As said in Section 1, the coupled analysis was performed
in all phases of the problem, both during the embank-
ment construction and in the post-construction period.
The constitutive relations of both the embankment
and foundation soils were simulated using the p–q–h
critical state model [4–6,8] with the parameters indicated
in Table 1 (k, slope of normal consolidation line and
critical state line; k, slope of swelling and recompression
line; C, specific volume of soil on the critical state line at
mean normal stress equal to 1 kPa; N, specific volume of
normally consolidated soil at mean normal stress equal
to 1 kPa). Table 1 also shows other geotechnical proper-
ties: c, unit weight; m
0
, PoissonÕs ratio for drained load-
ing; c
0
and /
0
, cohesion and angle of friction defined
in effective terms; k
x
and k
y
, coefficients of permeability
in x and y directions. Table 2 indicates the variation
with depth of the at rest earth pressure coefficient, K
0
,
and over-consolidation ratio, OCR, in the foundation.
The embankment soil was considered with 0.43 for K
0
and 1 for OCR. All these parameters were defined tak-
ing into account typical experimental values for this
kind of soils [4,26].
Fig. 7 shows the 2D finite element mesh for the
embankment without the drains, problem that can be
considered as a plane strain problem, given the very
large longitudinal length of the embankment; y axis is
the symmetry line and, with exception of the boundary
conditions for excess pore pressure (set as zero only on
the upper drainage surface, i.e., at nodes with y = 5
m), all the other characteristics of the problem, when
compared with the three-dimensional problem, are
maintained.
Given the non-linearity of the constitutive model
used in the soils (p–q–h critical state model) and the
boundary conditions, in both problems (with and with-
out drains) mesh sensitivity in terms of variation of
numerical results was analysed by considering several
meshes and time increments (different geometry and
time discretizations). The meshes (Figs. 5 and 7) and
the increments used in the paper were assessed adequate
by analysing, at each calculation of the coupled analysis,
the global equilibrium of the problem (comparing exter-
nal forces with stresses at all Gauss points of the ele-
ments). The smooth geometric variation of the stress
results in the medium (presented in the following
5 10 15 20 25 30 35 40 45 50 55 60 65 time (days) 0
0.5
1
1.5
2
0
embankment
height
(m)
Fig. 6. Embankment construction sequence.
Table 1
Geotechnical properties of the foundation and embankment soils
c (kN/m
3
) m
0
c
0
(kPa) /
0
(°) k
x
(m/s) k
y
(m/s) p–q–h critical state model
k k C N
Foundation 17 0.25 0 30 10
À9
10
À9
0.22 0.02 3.26 3.40
Embankment 20 0.30 0 35 – – 0.03 0.005 1.80 1.817
Table 2
At rest earth pressure coefficient, K
0
, and over-consolidation ratio,
OCR, in the foundation
Depth (m) K
0
OCR
0–1 0.7 2.43
1–1.8 0.7–0.5 2.43–1
1.8–5 0.5 1
Fig. 7. 2D finite element mesh for the problem without vertical drains.
J.L. Borges / Computers and Geotechnics 31 (2004) 665–676 669
section) corroborated that the numerical convergence of
this non-linear problem was achieved adequately.
In the problem with the drains the total number of
increments (i.e., total number of finite element calcula-
tions) was 359 (209 during the construction and 150 dur-
ing the post-construction period). In the embankment
without the drains the same number was used during
construction, but a higher number (177) was considered
during the post-construction period (since the consolida-
tion time is longer).
3. Analysis of the results
When a load is applied on a saturated soil mass, the
distribution of the excess pore pressure has, usually, gra-
dients that determine a field of relative velocity among
different zones of the soil. Initial conditions of a tran-
sient flow process are determined and transferences of
load from the water (pore pressure) to the soil skeleton
(effective stress) take place. Therefore, until a steady
state is reached, the soil mass behaviour is determined
by the variation of the fields of stress (pore pressure
and effective and total stress), strain and displacement
(consolidation).
Figs. 8 and 9 show results of excess pore pressures for
the two analyses of the problem, with and without ver-
tical drains, at different phases, during and after the con-
struction period. For the 3D analysis, Fig. 8 shows
results both on the vertical plane that contains one
row of drain centres, z = 0 plane (on the right side),
and on the vertical plane equidistant from two rows of
drains, z = 1 m plane (on the left side).
Based on these results, and considering the founda-
tion divided into four typical zones as illustrated in
Fig. 10, one can say that, for the problem without verti-
cal drains, during construction period: (a) maximum
values happen in zone A and are approximately similar
to the vertical stress due to the embankment weight; (b)
in zone B, excess pore pressures decrease from zone A to
zone D, where their values are not significant. After con-
struction, when the problem is, above all, characterised
by the transient water flow, one can see that isovalue
curves have a very regular shape, normal to the flow
lines.
With regard to the results of the problem incorporat-
ing the vertical drains (Fig. 8), the shape of the isovalue
curves clearly shows the three-dimensional condition of
the problem, with drainage occurring both horizontally
and vertically towards the several drainage surfaces
(band drains and upper drainage surface). Maximum
values also occur in zone A and also with similar values
to the vertical stress determined by the embankment
weight. However, the most important fact (see below
the analysis of the settlements) concerns the significantly
effect of consolidation (mainly in zone A) due to the ver-
Fig. 8. Excess pore pressure (u) for the embankment with vertical drains. (a) 1 m height embankment (time = 14 days); u
max
= 20.15 kPa. (b) 2 m
height embankment (end of construction); u
max
= 38.16 kPa. (c) 64 days after construction; u
max
= 29.29 kPa.
670 J.L. Borges / Computers and Geotechnics 31 (2004) 665–676
tical drains, in comparison with embankment without
the drains.
Figs. 11 and 12 show the deformed meshes for both
analyses, at the end of construction and at the end of
consolidation (displacements are represented in a scale
three times larger than the scale used for lengths). These
results are complemented by the results shown in Figs.
13–16, namely, settlements at the embankment base at
different phases for the analysis with drains (Fig. 13),
and for both analyses: (i) settlements at the embankment
base at the end of construction and at the end of consol-
idation (Fig. 14); (ii) variation in time of settlement at
the middle point under the embankment on the ground
level, i.e., point with x = 0 and y = 5 m (and z = 0, for
three-dimensional case) (Fig. 15); (iii) horizontal dis-
placements at the end of construction and at the end
of consolidation, along vertical line under the toe, i.e.,
along points with x = 8.3 m (and z = 0, for three-dimen-
sional case) (Fig. 16).
Based on these results, one can point out that, for
both analyses, during the construction period: (i) there
are settlements in the central loaded zone (zone A of
Fig. 10); (ii) there are important upward vertical dis-
placements near the embankment toe (Figs. 11–14);
(iii) horizontal displacements are outwards (Fig. 16).
As explained by Borges [4] and Borges and Cardoso
[27], if the consolidation is not significant during the
construction period (which is valid, in most cases, for
embankments without vertical drains, given the usually
low permeability of the soft soils), these displacements
are related to a strain field characterised by volumetric
strains that are practically zero (undrained conditions)
and shear strains that have very expressive values. If
the consolidation effect is significant during the con-
struction period (which really happens in the problem
with the band drains), this effect is essentially expressed
by the increase of the settlements (associated with the in-
crease of volumetric strains by consolidation), as shown
in Fig. 14 comparing maximum values of settlements at
the embankment base for both analyses (13.8 and 8.4
cm).
During the post-construction period, the main char-
acteristic of the problem in terms of displacements is
Fig. 9. Excess pore pressure (u) for the embankment without vertical
drains. (a) 1 m height embankment (time = 14 days); u
max
= 19.62 kPa.
(b) 2 m height embankment (end of construction); u
max
= 40.16 kPa.
(c) 925 days after construction; u
max
= 25.13 kPa.
A B D
C
Fig. 10. Four typical zones in the foundation.
Fig. 11. Deformed meshes for the embankment with vertical drains.
(a) End of construction. (b) End of consolidation.
Fig. 12. Deformed meshes for the embankment without vertical
drains. (a) End of construction. (b) End of consolidation.
J.L. Borges / Computers and Geotechnics 31 (2004) 665–676 671
expressed by generalised settlements and horizontal dis-
placements that can be outwards too, as shown in Fig.
16 for the embankment without the vertical drains. As
explained by Borges [4], these outward horizontal dis-
placements, in consonance with experimental results ob-
served in real works [9–12,28–31], are associated with
shear strains during the consolidation process which
are properly simulated only by elastoplastic models with
closed yielding surfaces, which is the case of the p–q–h
critical state model used in this study.
Three more important effects about the use of the ver-
tical drains, obtained from the analysis of the results, are
pointed out below.
The first effect, as expected, is the very expressive de-
crease of the consolidation time (reduction from approx-
imately 5000 to 500 days, as shown in Fig. 15).
The second effect is the reduction of the maximum va-
lue of the long term settlements (about 16%, from 48.7
to 40.9 cm, as illustrated in Fig. 14). This effect is asso-
ciated with a certain improvement of the foundation soil
properties (decrease of voids ratio) by consolidation
-0.60
-0.50
-0.40
-0.30
-0.20
-0.10
0.00
0.10
0.00 2.00 4.00 6.00 8.00 10.00
Distance from x=0 plane (m)
S
e
t
t
l
e
m
e
n
t

(
m
)
h = 0.5 m (t = 7 days)
h = 1 m (t = 14 days)
h= 1.5 m (t= 17.5 days)
h =1.5 m (t = 55.9 days)
h = 2 m (end of construction)
h = 2 m (end of cosolidation)
h – embankment height
t – time after beginning
of construction
Fig. 13. Settlements at the embankment base for the embankment with vertical drains.
-0.600
-0.500
-0.400
-0.300
-0.200
-0.100
0.000
0.100
0.000 2.000 4.000 6.000 8.000 10.000
Distance from x=0 plane (m)
S
e
t
t
l
e
m
e
n
t

(
m
)
Embankment without
vertical drains (end of
construction)
Embankment without
vertical drains (end of
consolidation)
Embankment with
vertical drains (end of
construction)
Embankment with
vertical drains (end of
consolidation)
Fig. 14. Settlements at the embankment base for the embankment with and without vertical drains.
S
e
t
t
l
e
m
e
n
t

(
m
)
-0.60
-0.50
-0.40
-0.30
-0.20
-0.10
0.00
0 1000 2000 3000 4000 5000 6000
Time after the end of construction (days)
Embankment without
vertical drains
Embankment with
vertical drains
Fig. 15. Settlement in time at the middle point under the embankment
on the ground level (point with x = 0, y = 5 and, for three-dimensional
case, z = 0) for the embankment with and without vertical drains.
672 J.L. Borges / Computers and Geotechnics 31 (2004) 665–676
during the construction period. This is a kind of a soil
‘‘hardening’’ effect that influences the decrease of long
term settlements (as well as the reduction of long term
horizontal displacements, as shown in Fig. 16). This ef-
fect only happens because shear stress increases during
the load periods and can reach higher values in the prob-
lem with the vertical drains (the consolidation effect dur-
ing the pause periods increases undrained strength of the
soil). If the problem was one-dimensional (as in the
oedometer tests), this effect would not take place be-
cause in the load periods there is no variation of effective
stress (in undrained conditions) and it is indifferent
whether there is pause periods or not, if total load is
the same.
The third effect is the uniformity of settlements along
z-direction for the embankment with band drains, de-
spite its three-dimensional behaviour in terms of stresses
and water flow. This point is justified by the existence of
‘‘arch effect’’ inside the fill, which is expressed by the
application of a non-uniform vertical load on the foun-
dation surface (see Fig. 17). The vertical load, at a deter-
mined phase, is smaller on zones that tend, by
consolidation, to settle more than the others, i.e., near
the vertical drains. This effect, as shown by field results
[10–12], tends to be equilibrated in a non-uniform distri-
bution of the vertical load that approximately unifor-
mizes the corresponding settlements.
Finally, Figs. 18 and 19 show values of the stress lev-
els (which vary from 0 to 1, being 1 the critical state le-
vel) at different phases of the problem; Fig. 20 presents
the effective principal stresses for the embankment with-
out vertical drains at the ends of construction and
consolidation.
The definition of stress level, SL, is given in Fig. 21
for an arbitrary stress state represented by the point A
in the p–q plane, where p is the effective mean stress
and q the deviatoric stress.
For the embankment without vertical drains (Fig.
18), one can point out that: (i) during the construction
period, the main characteristic of the problem is
Fig. 16. Horizontal displacements at the end of construction and at the end of consolidation along vertical line under the toe (points with x = 8.3 m
and, for three-dimensional case, z = 0) for the embankment with and without vertical drains.
Fig. 17. Increment of vertical effective stress (kPa) at the end of
construction for the embankment with vertical drains.
J.L. Borges / Computers and Geotechnics 31 (2004) 665–676 673
expressed by the increase of the stress levels in the foun-
dation (and therefore by the decrease of the problem
safety [27,32]), especially in Zone B (Fig. 10); as shown
in Fig. 20(a), this effect is associated with an effective
stress path basically characterised by the rotation of
the principal stresses directions (i.e., low variations of
effective mean stress and significant increases of devia-
toric stress); (ii) during the post-construction period,
the effective stress path (associated with the dissipation
of the excess pore pressures) is characterized by expres-
sive increases of effective mean stress and low variations
of deviatoric stress (for in Fig. 20 the magnitude of the
principal stresses increases without their directions sig-
nificantly changing), which implies a generalised reduc-
tion of the stress levels, as illustrated in Fig. 18.
With regard to the results of stress levels of the
embankment with vertical drains (Fig. 19), the main
difference in relation to the embankment without
drains concerns the very significant reduction of the
stress levels at all phases of the problem (and particu-
Fig. 18. Stress levels for the embankment without vertical drains. (a) 1
m height embankment (time = 14 days). (b) 2 m height embankment
(end of construction). (c) End of consolidation.
Fig. 19. Stress levels for the embankment with vertical drains. (a) 1 m height embankment (time = 14 days). (b) 2 m height embankment (end of
construction). (c) End of consolidation.
Fig. 20. Effective principal stresses for the embankment without
vertical drains. (a) End of construction. (b) End of consolidation.
674 J.L. Borges / Computers and Geotechnics 31 (2004) 665–676
larly at the end of the construction period) due to the
consolidation acceleration determined by use of the
vertical drains.
4. Conclusions
In the paper, a numerical model based on the finite
element method was used to analyse the structural
behaviour of an embankment on soft soils incorporating
vertical drains. The model, which incorporates the BiotÕs
consolidation theory and constitutive relations simu-
lated by the p–q–h critical state model, was applied on
both the embankment with vertical drains (three-dimen-
sional analysis) and the same problem without vertical
drains (two-dimensional analysis). The analysis of the
results (excess pore pressures, settlements, horizontal
displacements and stress levels) allows to point out the
following conclusions on the effects of the use of vertical
drains in embankments on soft soils:
1. The effect on the total time of consolidation is very
expressive (reduction of about 10 times, from approx-
imately 5000 to 500 days).
2. This fact is obviously associated with the faster dissi-
pation of the excess pore pressures (and consequent
decrease of the stress levels) at all phases of the prob-
lem, during and after the construction period.
3. The increase of the maximum settlement value at the
end of the construction is significant (about 64%,
from 8.4 to 13.8 cm), which implies a certain
improvement of the foundation soil properties
(decrease of voids ratio) by consolidation during the
construction period.
4. This effect, which is a kind of a ‘‘hardening’’ effect of
the soil, influences the decrease of the long term settle-
ments (about 16%, from 48.7 to 40.9 cm), as well as
the reduction of long term horizontal displacements.
5. In spite of its three-dimensional behaviour in terms of
stresses and water flow, settlements of the embank-
ment with vertical drains are approximately uniform
along longitudinal direction. This is justified by the
existence of ‘‘arch effect’’ inside the fill, which is
expressed by the application of a non-uniform verti-
cal load on the foundation surface.
References
[1] Terzaghi K, Peck RB. Soil mechanics in engineering practi-
ce. New York: Wiley; 1948.
[2] Johnson SJ. Treatment of soft foundations for highway embank-
ments. Synthesis of Highway Practice 29, USA: National Coop-
erative Highway Program; 1975. p. 25.
[3] Holtz RD. Treatment of problem foundations for highway
embankments. Synthesis of highway practice 147, Washington
(DC): National Cooperative Highway Research Program; 1989. p.
72.
[4] Borges JL. Geosynthetic-reinforced embankments on soft soils.
Analysis and design. Ph.D. thesis in civil engineering, Faculty of
Engineering, University of Porto, Portugal; 1995 [in Portuguese].
[5] Lewis RW, Schrefler BA. The finite element method in the
deformation and consolidation of porous media. New
York: Wiley; 1987.
[6] Britto AM, Gunn MJ. Critical soil mechanics via finite ele-
ments. England: Ellis Horwood Limited; 1987.
[7] Borges JL, Cardoso AS. Numerical simulation of the consolida-
tion processes in embankments on soft soils. R Geotecnia
2000;89:57–75. [in Portuguese].
[8] Borges JL, Cardoso AS. Numerical simulation of the p–q–h
critical state model in embankments on soft soils. R Geotecnia
1998;84:39–63 (in Portuguese).
[9] Quaresma MG. Behaviour and modelling of an embankment over
soft soils reinforced by geotextile. Ph.D. thesis, Universite Joseph
Fourier, Grenoble I, 1992 [in French].
[10] Yeo KC. Simplified foundation data to predictors. Proceedings of
the prediction symposium on a reinforced embankment on soft
ground. London: KingÕs College; 1986.
[11] Basset RH. The instrumentation of the trial embankment and of
the Tensar SR2 grid. Proceedings of the prediction symposium on
a reinforced embankment on soft ground. London: KingÕs Col-
lege; 1986.
[12] Basset RH. Presentation of instrumentation data. Proceedings of
the prediction symposium on a reinforced embankment on soft
ground. London: KingÕs College; 1986.
[13] Biot MA. The consolidation problems of clayey materials under a
load. Annales de la Socie´te´ Scientifique de Bruxelles 1935;55(Ser-
ies B):110–3. [in French].
[14] Biot MA. General theory of three-dimensional consolidation. J
Appl Phys 1941;12:155–64.
[15] Biot MA. Theory of elasticity and consolidation for a porous
anisotropic solid. J Appl Phys 1955;26:182–5.
[16] Biot MA. Theory of deformation of a porous viscoelastic
anisotropic solid. J Appl Phys 1956;27:459–67.
[17] Biot MA. General solutions of the equations of elasticity and
consolidation for a porous material. Transactions. J Appl Mech
1956;78:91–6.
[18] Small JC, Booker JR, Davis EH. Elastoplastic consolidation of
soil. Int J Solids Struct 1976;12:431–48.
[19] Desai CS, Siriwardane THJ. Subsidence due to consolida-
tion including nonlinear behaviour. In: Saxena SK, editor.
p
A
γ
ψ
critical
state line
SL=
tg
γ
ψ
tg
q
Fig. 21. Definition of stress level, SL, in the p–q plane.
J.L. Borges / Computers and Geotechnics 31 (2004) 665–676 675
Evaluation and prediction of subsidence. New York: ASCE;
1979. p. 500–15.
[20] Runesson K. On nonlinear consolidation of soft clays. Thesis,
Dept. of Struct. Mech., Chalmers University of Technology,
Goteborg; 1978.
[21] Lewis RW, Roberts GK, Zienkiewicz OC. A non-linear flow and
deformation analysis of consolidation problems. In: Desai CS,
editor. Numer Meth Geomech, 2. ASCE; 1976. p. 1106–18.
[22] Norris VA. The elasto-plastic analysis of soil consolidation with
special reference to kinematic hardening. Ph.D. thesis, University
College of Swansea; 1980.
[23] Zeng GX, Xie KH, Shi ZY. Consolidation analysis of sand-
drained ground by F.E.M. In: Proceedings of the eighth
regional conference on soil mechanics (vol. 1). Kyoto; 1987.
p. 139–42.
[24] Hird CC, Kwok CK. Predictions for the Stanstead Abbotts trial
embankment. Predictions symposium on a reinforced embank-
ment. London: KingÕs College; 1986.
[25] Poran CJ, Kaliakin VN, Hermenn LR, Romstadd KM, Lee DF,
Shen CK. Prediction of trial embankment behaviour Hertford-
shire county council – Stanstead Abbotts. Prediction symposium
on a reinforced embankment. London: KingÕs College; 1986.
[26] Lambe TW, Whitman RV. Soil mechanics. New York: Wiley;
1969.
[27] Borges JL, Cardoso AS. Structural behaviour and parametric
study of reinforced embankments on soft clays. Comput Geotech
2001;28/3:209–33.
[28] Soderman KL. The Behaviour of Geotextile Reinforced Embank-
ments. Ph.D. thesis, Ontario: University of Western Ontario;
1986.
[29] Litwinowicz A, Wijeyakulasuriya CV, Brandon AN. Perfor-
mance of a reinforced embankment on sensitive soft clay
foundation. In: Proceedings of the fifth international conference
on geotextiles, Geomembranes and Related Products (vol. 1).
Singapore; 1994. p. 11–6.
[30] Loke KH, Ganeshan V, Werner G, Bergado DT. Composite
behaviour of geotextile reinforcement embankment on soft clay.
In: Proceedings of the fifth international conference on geotex-
tiles, geomembranes and related products (vol. 1). Singapore;
1994; vol. 1 p. 25–8.
[31] Matichard Y, Tanays E, Carbon M. Geotextile reinforced
embankment to cross a peat-bog. Proceedings of the fifth
international conference on geotextiles, geomembranes and
related products (vol. 1). Singapore; 1994. p. 37–40.
[32] Borges JL, Cardoso AS. Overall stability of geosynthetic-rein-
forced embankments on soft soils. Geotext Geomembranes
2002;20/6:395–421.
[33] Lade PV, Duncan JM. Cubical triaxial tests on cohesionless soils,
1973, J Soil Mech. Found Div. ASCE, vol. 99, SM10.
[34] Poulos HG. Stresses and displacements in an elastic layer
underlain by a rough rigid base. Ge´otechnique 1967;17:
378–410.
[35] Booker JR, Randolph MF. Consolidation of a cross-anisotropic
soil medium. Quart J Mech Appl Math 1984;37:479–95.
676 J.L. Borges / Computers and Geotechnics 31 (2004) 665–676