You are on page 1of 8

Hird, C. C., Pyrah, I. C., & Russell,D. (1990).G&technique40, No.

4, 633-640

TECHNICAL NOTE

Finite element analysis of the collapse of reinforced


embankments on soft ground

C. C. HIRD. I. C. PYRAH and D. RUSSELL

KEYWORDS: analysis; bearing capacity; finite ele- PLASTICITY SOLUTIONS


ments; footings/foundations; plasticity; reinforced soil. The problem analysed involves the collapse of
a rigid strip footing resting on idealized soil. Two
idealized shear strength profiles are considered:
INTRODUCTION uniform strength over a limited depth with a hard
The realistic prediction of displacements caused layer beneath (Fig. l(a)) and strength increasing
by the construction of a reinforced embankment linearly and indefinitely with depth (Fig. l(b)).
on a soft foundation requires the use of modern For each strength profile, both smooth and rough
numerical analysis techniques, such as the finite footings are considered, making a total of four
element method. The technique has the capacity cases in all. For the smooth footing no horizontal
to model the behaviour of the embankment, shear stress is allowed at the base of the footing,
reinforcement and foundation soil both during while for the rough footing a shear stress, up to a
construction and during the subsequent life of the maximum value of SueI can be generated at its
structure, providing the relevant material proper- base to resist relative horizontal movement
ties and their variation in space and time can be between the soil and the footing. The bearing
established. However, before the method can be capacity factors for the above conditions are
used with confidence, either directly in design or shown in Fig. 2 and the influence of the shear
in the validation of simplified design methods, it stress in increasing the bearing capacity at all but
must be shown to be capable of modelling known the lowest geometric ratios (defined in the figure)
behaviour correctly. While the main purpose of is clear.
finite element analysis is usually to examine In addition to the bearing capacity factors,
stresses, strains and displacements under working plasticity theory gives the distributions of vertical
conditions, little confidence can be placed in the stress on the underside of the rigid footing
method if it fails to predict collapse adequately. (Fig. 3). These are ideal distributions of loading
The stability of an embankment reinforced at which make the best possible use of the founda-
its base has been discussed by Hird & Jewel1 tion strength and provide a means of checking
(1990). Three classes of failure can be identified: the finite element analysis. Jewel1 (1988) has
internal instability of the embankment, instability shown how they may also be used directly to
of the foundation soil and overall instability design a reinforced embankment.
involving both the embankment and the founda-
tion. Rigorous plasticity solutions exist for the
second of these and have been selected for com-
FINITE ELEMENT ANALYSES
parison with finite element results. For the
The finite element program CRISPLY (Gunn &
analysis of foundation stability, the loads applied
Britto, 1984; Britto & Gunn, 1987) has been used
to the subsoil by the reinforced embankment can
to carry out plane strain analyses to determine
be considered similar to those applied by a
the ideal distribution of loading on soft ground
footing and the problem treated as one of bearing
referred to above. The double precision version of
capacity. The load is applied at the ground
the program was used and all analyses were
surface and, although this does not allow the
carried out on the Sheffield University IBM 3083
effect of embankment stiffness to be modelled, it
computer.
does allow the finite element analysis to be
checked against the plasticity solutions (Davis &
Booker, 1973; Mandel & Salencon, 1972). Element types and constitutive models
Eight element types and five constitutive
Discussion on this Technical Note closes 5 April 1991; models are available in CRISPLY. The six-noded
for further details see p. ii. linear strain triangle was used to model the soil,
* Department of Civil and Structural Engineering, Uni- as this element is capable of correctly modelling
versity of Sheffield, Mappin Street, Sheffield Sl 3JD. collapse behaviour under undrained conditions

633
Downloaded by [ TUFTS UNIVERSITY] on [18/10/16]. Copyright © ICE Publishing, all rights reserved.
634 HIRD, PYRAH AND RUSSELL

Shear strength S, Shear strength S,

&IO S”0

Fig. 1. Idealized profiles of shear strength with depth

0 ’ ’ ’ ’ ’ ’ ’ ’ ’ ’
XID
(a)
Rough footmg Smooth footing
30 -
I I I I I I
I--LLLi i i i I
m/7/ , I “-
I
E 20 -

lo-

t3-

6 -5.14

4-

2-
OL
0
’ ’
2

4
’ ’ ’
6
’ ’
a
’ ’
10
WS,,
01 ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ (b)
Geometric ratlo BID
Fig. 3. Ideal load distribution on an ideal&d soil profile:
(a)
(a) uniform strength/limited depth (after Mandel &
Salengon, 1972, with adaptations); (b) increasing strength
with depth (after Davis & Booker, 1973)

for a plane strain problem (Sloan & Randolph,


1982). An elastic-perfectly plastic material with a
Tresca yield criterion was assumed; this is the
same yield criterion as adopted in the plasticity
4-
solutions.
The original version of CRISPS~ has been
2- extended by incorporating an additional element
to model the slip which may occur between the
0’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’
0 2 4 6 6 IO soil and the footing or along other boundaries.
Geometric ratio oB/S,, The interface element is a six-noded quadrilateral
0) of the type developed by Goodman et al. (1968)
Fig. 2. Bearing capacity of a rigid footing on an idealized and has the constitutive relationships shown in
soil profile: (a) uniform strength/limited depth (after Fig. 4. This element models interface behaviour
Msndel & Salenqon, 1972, with adaptations); (b) increes- efficiently (Wilson, 1977) and has previously been
ing strength with depth (after Davis & Booker, 1973) used to represent the interface between the soil

Downloaded by [ TUFTS UNIVERSITY] on [18/10/16]. Copyright © ICE Publishing, all rights reserved.
ANALYSIS OF COLLAPSE OF REINFORCED EMBANKMENTS 635

and the reinforcement in the analysis of rein-


forced embankments (Hird & Kwok, 1990). In the
work being described the maximum shear resist-
ance of this element was equal to the shear
strength of the soil.

Meshes for equivalent loading problems


In all cases advantage was taken of symmetry
and only half the loaded width was modelled.
Relatwe shear disrelacement
For soil of uniform strength and limited depth
D, the mesh for the smooth loading case is shown
in Fig. 5(a). The width of the loading B was rela-
tively large (B = 2OD). Sufficient distance was
allowed beyond the edge of the loading that the
conditions at the left-hand boundary of the mesh
had no significant effect on the solution. It was
assumed that the soil rested on a perfectly rough
and rigid surface; a row of interface elements was
included along this boundary. In the rough
loading case the same mesh was used except that
a row of interface elements was also included
along the loaded surface, which was restrained
horizontally.
R&We normal displacement
For soil of increasing strength with depth, the
Fig. 4. Constitutive behaviour of interface element mesh for the smooth loading case is shown in Fig.

\ Interface elements on boundary

(a)

ii
c 0.756 (1 Em) Loaded area 0.56 (12m) _I

(b)
Fig. 5. Finite element meshes: (a) uniform strength/limited depth; (b) strength increasing with
depth

Downloaded by [ TUFTS UNIVERSITY] on [18/10/16]. Copyright © ICE Publishing, all rights reserved.
636 HIRD, PYRAH AND RUSSELL

5(b). As the soil is of unlimited depth, the mesh adjacent soil (e.g. either 2.5 kPa or 30 kPa at the
had to be sufficiently extensive to ensure that foundation surface). To limit relative displace-
none of the boundaries affected the solution sig- ments prior to slip and normal to the interface,
nificantly. The extent of the failure zone was ini- the stiffness parameters (Fig. 4) were set to high
tially estimated from charts produced by Houlsby values, k, = 6 x IO3 kPa/m and k, = 5 x lo5
& Jewel1 (1988). As before, in the case of rough kPa/m.
loading a row of interface elements was included
along the loaded surface, where no horizontal
movement was allowed. RESULTS
Displacement controlled analyses
Because the plasticity solutions assume a rigid
Material properties footing, the finite element analyses were initially
In working with an elastic-perfectly plastic carried out by applying increments of uniform
material to obtain comparisons with the plasticity vertical displacement along the loaded surface.
solutions, it is important to restrict the elastic Failure was deemed to occur when a further
strains as much as possible. A high value of shear increase in the displacement caused no significant
modulus (G = 6OCOkPa) was therefore used. increase in the vertical stress at any location on
Approximately undrained conditions were model- the displaced boundary (e.g. see Fig. 7). Failure
led by specifying a large bulk modulus, while rea- was reached in approximately 100 increments.
lizing that, if this parameter is too large, The vertical stress distribution on the ground
numerical problems may occur (Griffiths 1985). A surface at failure was compared with the plasticity
bulk modulus K = 5 x lo5 kPa (equivalent to a solution for each of the four cases.
total stress Poisson’s ratio Y = 0.4925) was found For soil of uniform strength and limited depth,
to be satisfactory. the distributions of total vertical stress for both
The soil yielded according to the Tresca cri- smooth and rough footings are shown in non-
terion when the yield strength equalled either a dimensional form in Fig. 8(a) together with the
constant value, 5, = 30 kPa (limited depth case) distributions predicted by plasticity theory. The
or a value at the foundation surface, S,, = 2.5 corresponding stress distributions for the case of
kPa, which increased at a rate of p = 2 kPa/m strength increasing with depth are shown in
with depth. For the case of strength increasing Fig. 8(b).
with depth the strength prohle could not be mod-
elled exactly, but it was possible to divide the
mesh into several horizontal layers and to specify Load controlled analyses
a different strength for each layer. The resulting In addition to the displacement controlled
stepped distribution (Fig. 6) was found to provide analyses, load controlled analyses were later
sufficiently accurate results. carried out in which the ideal plasticity load dis-
When interface elements were used their shear tributions were applied in increments at the
strength was set equal to the strength of the ground surface. The load was applied in 100

Shear strength S, : kPa


10 20 30 x = 1OD
1

~
- - ldealn?d

Fmlte element
approxlmatlon m
60°/T------j
\ 9 400 - x = 50
E D
m
0E e 200 x=0

5 10 15 20 25
Settlement : mm
2-: ,2_-
6
10
8 ~ ~-+TIL ~ ~ 3
9s OO
,/:-I ” ~
Fig. 7. Load-settlement curves from displacement con-
Fig. 6. Finite element approximation for strength in- trolled analysis (uniform strength/limited depth, rough
creasing with depth loading

Downloaded by [ TUFTS UNIVERSITY] on [18/10/16]. Copyright © ICE Publishing, all rights reserved.
ANALYSIS OF COLLAPSE OF REINFORCED EMBANKMENTS 631

,iO 2 4
xii3

la)
6 6 10

0 FE, rough

I I I I
0
0 2 4 6 a 10
QXIS,,

0)
Fig. 8. Comparison of finite element and plasticity solutions: (a) uniform
strength/limited depth; (h) strength increasing with depth

equal increments, i.e. 1% of the failure load pre- are shown in Fig 10(a) and (b) for soil of uniform
dicted by the plasticity solution was applied in strength and limited depth and soil of increasing
each increment. The surface settlement profiles strength with depth respectively.
from these analyses were compared with the pre-
viously assumed uniform settlement profile. Only
the rough cases were examined. FURTHER ANALYSIS USING
As the applied load approached the plasticity MODIFIED CAM CLAY
load, the incremental displacements increased Finite element solutions using an elastic-
rapidly. For example, Fig. 9 shows the develop- perfectly plastic constitutive model for the soil
ment of vertical displacement at three points can be compared directly with the theoretical
along the loaded surface in one of the analyses. plasticity solutions. However, this model is not
An attempt was made to increase the load by a necessarily satisfactory for representing soft clays
further 5%, applied in 50 increments, but the and in further studies of reinforced embankments
analyses failed due to numerical problems. The on soft ground modified Cam clay (Roscoe &
settlement profiles under the full plasticity load Burland, 1968) will be used to model the subsoil.

Downloaded by [ TUFTS UNIVERSITY] on [18/10/16]. Copyright © ICE Publishing, all rights reserved.
638 HIRD, PYRAH AND RUSSELL

Settlement: mm

Fig. 9. Load-settlement curves from load controlled analysis (uniform


strength/limited depth, rough loading)

For this reason a comparison has also been made used in the manner of Jewel1 (1988) to design
between the load distributions from plasticity embankments with any degree of flexibility. In
theory and from finite element analysis using the Fig. 10(a) larger settlements near the edge and the
modified Camclay model. The analysis was con- centre of the loaded area may be a function of the
ducted for a rough footing resting on a soil with numerical procedures used in the presence of high
strength increasing linearly with depth. The initial stress gradients or singularities.
stresses and modified Camclay parameters were The prediction made using the modified Cam
chosen to model such a soil profile (Fig. 1 I). With clay mode1 (Fig. 12) is also in close agreement
standard notation, the parameters adopted were with the relevant plasticity solution. Therefore, if
i = 0.25, K = O-07, I = 3.0, M = 1.0 and this widely applied model is selected in order to
v’ = 0.3. A displacement controlled analysis was improve the modelling of the pre-failure behav-
carried out using the same mesh as before. iour, the prediction of collapse need not be
Failure (as defined above) was reached in about adversely affected.
600 increments, when the displacement was In all cases the extent of the plastic zones at
0.06 m and the results are shown in Fig. 12. failure was in accordance with expectation (e.g.
Houlsby & Jewell, 1988) and vindicated the deci-
sions made in relation to mesh geometry.
However, when analyses were repeated without
DISCUSSION OF RESULTS including the interface elements at rough bound-
For a rigid footing being pushed into the soil, aries, the collapse loads were invariably over-
the predictions of the vertical stress distribution predicted. The role of the interface elements in
at collapse, both for the uniform strength/limited permitting superior results to be obtained will be
depth case (Fig. 8(a)) and for strength increasing discussed in a future publication.
with depth (Fig. 8(b)), are in close agreement with
the plasticity solutions and give confidence in the
ability of the finite element program to predict
collapse accurately. This was confirmed when the CONCLUSIONS
ideal load distribution was applied incrementally The finite element program CRISPLY has been
to the ground surface, since failure occurred as used successfully to predict collapse load distribu-
the theoretical limit was reached (e.g. Fig. 9). tions beneath smooth and rough footings. For
The settlement profiles shown in Fig. 10 differ both a soil of uniform strength and limited depth
significantly from the uniform settlement beneath and a soil of increasing strength with unlimited
a rigid footing, but the shape of the settlement depth, the results agree with theoretical plasticity
profile appears to have little influence on the col- solutions. The suitability of the modified Cam
lapse load distribution. This is a useful finding clay soil mode1 for the latter case has been con-
since it means that the plasticity solutions can be firmed.

Downloaded by [ TUFTS UNIVERSITY] on [18/10/16]. Copyright © ICE Publishing, all rights reserved.
ANALYSIS OF COLLAPSE OF REINFORCED EMBANKMENTS 639

I
01 I I I I
-10

:
L
5 10

E
%
[I)
I
2oc
(4

(b)
Fig. 10. Settlement profile beneath ideal load distribution for rough loading: (a)
uniform strength/limited depth; (b) strength increasing with depth

Stress.kPa Overconsolldatlon rat,a Calculated shear strength: kPa


O0 40 80 120 160 0 1 2 3 4 0 10 20 30 40

Effecf~ve verttcal

2- \ - -~~:water
pWS*“riZ
\

4- \

E \
5 6-
D \
0

a- \

IO- \

I \ :I:
iz-

Fig. 11. Stress and strength profiles for modified Cam clay analysis

Downloaded by [ TUFTS UNIVERSITY] on [18/10/16]. Copyright © ICE Publishing, all rights reserved.
640 HIRD, PYRAH AND RUSSELL

Mech. Fndns Div. Am. Sot. Ciu. Engrs 94,637-659.


Grifflths, D. V. (1985). The effect of pore fluid compress-
ibility on failure loads in elasto-plastic soil. Int. J.
Numer. Analyt. Meth. Geomechanics 9,253-259.
GUNN, M. & BRITTO,A. (1984). CRISP. Users’ and Pro-
grammers’ Guide, Engineering Department, Cam-
bridge University.
Hird, C. C., & Jewell, R. A. (1990). Theory of reinforced
embankments. Reinforced Embankments: Theory and
Practice in the British Isles, pp. 115-139. London:
- Plastlclty solution Thomas Telford.
0 FE, rough
Hird, C. C. & Kwok, C. M. (1990). Finite element
studies of interface behaviour in reinforced embank-
i ments on soft ground. Computers and Geotechnics 8,
OO 2 4 6 8
p/s,, 111-131.
Houlsby, G. T. & Jewell, R. A. (1988). Analysis of unre-
Fig. 12. Comparison of finite element and plasticity solu-
inforced and reinforced embankments on soft clays
tions with strength increasing with depth (modified Cam
by plasticity theory. Proc. 6th Int. Conf: on Numer.
clay) Meth. Geomechanics, Innsbruck, 2, 1443-1448.
Jewell, R. A. (1988). The mechanics of reinforced
embankments on soft soils. Geotextiles and Geo-
ACKNOWLEDGEMENT
membranes I, 237-273.
The authors are grateful for the helpful dis- Mandel, J. & SalenCon, J. (1972). Force portante d’un
cussions held with Dr R. A. Jewel1 during the sol sur assise rigide (ttude thkoretique). Giotechnique
course of the work, which was funded by the 22,79-93.
Science and Engineering Research Council. Roscoe, K. H. & Burland, J. B. (1968). On the gener-
alized stress-strain behaviour of ‘wet’ clay. Engineer-
ing Plasticity (eds J. Heyman and F. A. Leckie), pp.
REFERENCES 535-609. Cambridge University Press.
Britto, A. & Gunn, M. (1987). Critical StateSoil Mecha- Sloan, S. W. & Randolph, M. F. (1982). Numerical pre-
nics via Finite Elements. Chichester: Ellis Horwood. diction of collapse loads using finite element
Davis, E. H., SC Booker, J. R. (1973). Effect of increasing methods. Int. J. Numer. Analyt. Meth. Geomechanics
strength with depth on the bearing capacity of clays. 6, 47-76.
G&technique 23,551-563. Wilson, E. L. (1977). Finite element for foundations,
Goodman, R. E., Taylor, R. L. & Brekke, T. L. (1968). A joints and fluids. Finite Elements in Geomechanics
model for the mechanics of jointed rock. J. Soil (ed. G. Gudehus), pp. 319-350. New York: Wiley.

Downloaded by [ TUFTS UNIVERSITY] on [18/10/16]. Copyright © ICE Publishing, all rights reserved.

You might also like