You are on page 1of 10

Numerical Study of the Effect of Preloading

on Undrained Bearing Capacity


L. Zdravković1; D. M. Potts2; and C. Jackson3
Downloaded from ascelibrary.org by MARQUETTE UNIVERSITY LIBRARIES on 08/17/14. Copyright ASCE. For personal use only; all rights reserved.

Abstract: In practice the situation often arises where extra load needs to be added onto existing foundations. This can arise due to the
installation of additional equipment in a building, or to the construction of new floors, etc. It also arises in a situation where old
foundations are used for a new building, a practice likely to become increasingly popular in the future. If the foundation soil is clay, then
its undrained strength would have increased with time due to the dissipation of excess pore water pressures generated during the initial
construction of the foundations. The design problem is determining how large this increase in undrained strength is, since this will
determine the amount of new load that can be applied. This paper provides guidelines to answering this question through a numerical
study of preloaded surface footings on both soft and stiff clay foundations. Additional information resulting from this study addresses the
footing failure mechanisms, as well as the effect of a surface crust in soft clay on both the bearing capacity and the failure mechanism.
DOI: 10.1061/共ASCE兲1532-3641共2003兲3:1共1兲
CE Database subject headings: Footing; Bearing capacity; Loads; Numerical analysis; Failure modes.

Introduction many buildings undergo reconstruction, will also face this prob-
lem.
The near surface foundations on both clay and sand soils are This paper investigates this problem using the results from
usually designed using the well-established bearing capacity for- coupled finite element analyses of surface strip footings. The ap-
mulas of Terzaghi 共1943兲, Brinch-Hansen 共1970兲, Meyerhof proach adopted in the analyses is as follows: 共1兲 the strip footing
共1950兲, Davis and Booker 共1973兲, and others. These formulas take is first loaded undrained to failure to establish its initial short term
account of foundation shape, size, and depth and assume the soil (undrained) bearing capacity; 共2兲 a further series of analyses is
strength parameters to be either constant or to vary in some then performed, on the same footing, in which the footing is first
simple manner with depth. The design of new foundations is loaded undrained to a percentage 共20, 40, 60, 80, or 100%兲 of its
therefore relatively straightforward and for clay soil the short initial short term bearing capacity; 共3兲 the load is then held at this
term 共undrained兲 bearing capacity is the most critical. value until all the excess pore water pressures, generated in the
However, analysis of the current short term bearing capacity of clay during the first loading, dissipate; and 共5兲 additional load
an existing footing, which has been in place for a period of time, under undrained conditions is then applied until failure is reached
is not straightforward and the established bearing capacity formu- to determine the new ultimate undrained bearing capacity.
las mentioned above are not applicable. This is because the un- Both stiff and soft clay foundations are considered, as well as
drained strength of the clay would have changed since the con- both smooth and rough footing-soil interfaces. Two foundation
struction of the footing, due to the dissipation of excess pore widths of B⫽2 and 10 m are used. For the soft clay, the analyses
water pressures generated during its construction and initial load- were performed assuming both a stronger surface crust and no
ing. The resulting distribution of undrained strength will be com-
crust in the soil profile. For the stiff clay, the overconsolidation
plex. Examples of situations that may require the determination of
ratio 共OCR兲 was varied from 1 to 25. Some preliminary results
the current short term bearing capacity are where new or addi-
from this study were reported by Jackson et al. 共1997兲. This
tional equipment is to be added to an existing building or offshore
paper, however, extends the range of analyses and provides a
structure, or where additional floors are to be added to a building.
complete interpretation of the two main outcomes of this study:
The reuse of old foundations when constructing new buildings, a
the effect of preloading on the ultimate undrained bearing capac-
practice likely to become increasingly popular in the future as
ity and the development of failure mechanisms in the clay.
1 The soil constitutive model chosen for the analyses was a form
Lecturer, Dept. of Civil and Environmental Engineering, Imperial
of the modified Cam clay model and the parameters have been
College, London. E-mail: l.zdravkovic@imperial.ac.uk
2
Professor of Analytical Soil Mechanics, Dept. of Civil and Environ- selected to represent real soil types.
mental Engineering, Imperial College, London.
3
Ove Arup and Partners, 13 Fitzroy St., London; formerly, Dept. of
Civil and Environmental Engineering, Imperial College, London. Choice of Constitutive Model
Note. Discussion open until February 1, 2004. Separate discussions
must be submitted for individual papers. To extend the closing date by
Since the purpose of this study is to investigate the change of
one month, a written request must be filed with the ASCE Managing
Editor. The manuscript for this paper was submitted for review and pos- undrained strength during consolidation, it was necessary to select
sible publication on September 4, 2001; approved on May 28, 2002. This a constitutive model that can accurately predict this change. A
paper is part of the International Journal of Geomechanics, Vol. 3, No. form of the modified Cam clay model was used in which the
1, September 1, 2003. ©ASCE, ISSN 1532-3641/2003/1-1–10/$18.00. shape of the yield and plastic potential surfaces are given by a

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / SEPTEMBER 2003 / 1

Int. J. Geomech. 2003.3:1-10.


Table 1. Basic Material Properties ) 共 1⫺K nc
o 兲
A⫽
Second g 共 ⫺30° 兲共 1⫹2K nc
o 兲
elastic ␥
Soil type ␸⬘ ␬ ␭ parameter 共kN/m3兲 ␯1 k 共m/s兲
o ⫽1⫺sin ␸ ⬘
K nc
⫺10
Stiff clay 23° 0.0332 0.161 ␮⫽0.2 18 2.84 1⫻10
3.00 5⫻10⫺10 sin ␸ ⬘
o ⫽K o OCR
K oc
Soft clay 32° 0.0220 0.220 G⫽1,700 17 nc

Note: ␸ ⬘ ⫽angle of shearing resistance; ␯ 1 ⫽specific volume at p ⬘


⫽1 kPa on the isotropic virgin consolidation line; ␬⫽slope of the swell- ␪⫽Lode’s angle; OCR⫽overconsolidation ratio; ␴ ⬘v ⫽vertical ef-
ing line in ␯-ln p⬘ space; ␭⫽slope of the consolidation line in ␯-ln p⬘ fective stress; S u ⫽undrained shear strength; K nc
o ⫽normally con-
space; ␮⫽elastic Poisson’s ratio; G⫽elastic shear modulus in kPa; solidated coefficient of earth pressure at rest; K oc
Downloaded from ascelibrary.org by MARQUETTE UNIVERSITY LIBRARIES on 08/17/14. Copyright ASCE. For personal use only; all rights reserved.

o
k⫽permeability; and ␥⫽bulk unit weight of soil. ⫽overconsolidated coefficient of earth pressure at rest; ␸ ⬘
⫽angle of shearing resistance; ␬⫽slope of a swelling line in
␯-ln p⬘ space; and ␭⫽slope of the virgin consolidation line in
Mohr-Coulomb hexagon and a circle, respectively, Potts and Gens ␯-ln p⬘ space.
共1984兲, Gens and Potts 共1988兲. A full description of the model is Eq. 共1兲 indicates that, no matter what value of OCR is chosen,
given in Potts and Zdravković 共1999兲, but the basic details can the undrained strength will be zero if the effective vertical stress
also be found in Zdravkovic et al. 共2001兲. The model parameters is zero. In relation to the footing problem under investigation, this
for the two soils are given in Table 1. implies that for a finite undrained strength at the soil surface the
As can be seen from Table 1, the undrained strength is not an effective vertical stress must be greater than zero. The equation
input parameter for this model. However, it can be calculated also indicates that the undrained strength varies with the value of
from the effective stresses and input parameters using the follow- the Lode’s angle. It will therefore be different under triaxial com-
ing equation 关see Potts and Zdravković 共1999兲 for derivation兴: pression (␪⫽⫺30°) and plane strain (␪⫽0°) conditions.

Su 1
⫽ g 共 ␪ 兲 cos ␪ 共 1⫹2K o兲 OCR共 1⫹A 2 兲
␴ ⬘v 6 Soil Conditions

⫻ 冋 o 兲
2 共 1⫹2K oc
o 兲 OCR共 1⫹A 兲
共 1⫹2K nc 2 册 ␬/␭
(1) Stiff Clay

The stiff clay parameters are based on the behavior of London


where
clay. The laboratory data from Carswell et al. 共1993兲, from un-
sin ␸ ⬘ drained triaxial compression tests on samples from up to 10 m
g共 ␪ 兲⫽ depth, are shown in Fig. 1共a兲. The water table is assumed to be
1
cos ␪⫹ sin ␪ sin ␸ ⬘ 2.5 m below the ground surface and above this level the clay
) remains saturated and can sustain tensile pore water pressures.
This gives a finite value of the effective vertical stress, and there-

Fig. 1. 共a兲 Undrained strength profile for London clay and 共b兲 undrained strength profiles for stiff clay 共laboratory data from Carswell et al. 1993兲

2 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / SEPTEMBER 2003

Int. J. Geomech. 2003.3:1-10.


Table 2. S u /␴ v⬘ Values for Stiff Clay
Overconsolidation ratio 1 2 4 6 9 25


S u /␴ v 0.242 0.433 0.781 1.1 1.55 3.735

fore the undrained strength, at the ground surface. An initial OCR


of 6 is assumed for the London clay. The undrained strength
profile that varies linearly with depth, with S u /␴ ⬘v ⫽1.1 as calcu-
lated from Eq. 共1兲, is fitted through the laboratory data, Fig. 1共a兲.
Keeping the same input parameters for the stiff clay as given in
Downloaded from ascelibrary.org by MARQUETTE UNIVERSITY LIBRARIES on 08/17/14. Copyright ASCE. For personal use only; all rights reserved.

Table 1, further analyses were performed with initial OCRs rang-


ing from 1 to 25. The distributions of the initial undrained
strength profiles with depth for these OCRs are shown in Fig.
1共b兲, and the S u /␴ v⬘ ratios associated with these distributions 关as
calculated from Eq. 共1兲兴 are given in Table 2. Fig. 2. 共a兲 Undrained strength profile for soft clay and 共b兲 variation
of OCR and K o with depth 共laboratory data from Mair et al. 1992兲
Soft Clay
Soft clays are usually normally consolidated below the ground strength profile presented by the dashed line in Fig. 2共a兲.
water table, while developing a stronger, overconsolidated, crust Throughout this paper, when reference is made to soft clay, the
above the ground water table. This crust often provides the nec- existence of the surface crust is assumed, unless stated otherwise.
essary strength for any surface construction. In this study, the For both the soft and stiff clays in this study, the permeability
laboratory test data on samples from a soft clay site in Grimsby, parameters have been obtained from the site investigation data
Yorkshire 共Mair et al. 1992兲, are used to estimate the initial un- and are given in Table 1. However, the values do not have a major
drained strength profile. The results from the undrained triaxial influence on the analyses, as loading of the footings is performed
tests are shown in Fig. 2共a兲. At this site the water table is assumed sufficiently fast to ensure that near undrained conditions are
to be at a depth of 2 m and the clay is normally consolidated achieved and sufficient time is allowed for full dissipation of
below this level 共i.e., OCR⫽1). To obtain a realistic distribution excess pore water pressures during the rest periods. It should also
of undrained strength above the ground water table, the OCR and be noted that there is no need to specify a coefficient of consoli-
K o have been varied in Eq. 共1兲 to give an undrained strength dation, as this is essentially calculated from the permeability and
profile as shown in Fig. 2共a兲. The resulting distributions of OCR the soil compressibility 共obtained from the constitutive model兲
and K o are shown in Fig. 2共b兲. The ratio S u /␴ ⬘v for the normally during the coupled analysis.
consolidated clay is 0.3, which is typical of soft clays. Similar
undrained strength profiles, with a surface crust, have been de-
rived for other soft clay sites, see for example, Trak 共1980兲, or La Geometry and Boundary Conditions
Rochelle et al. 共1974兲 for soft Champlain clays of Canada. To
investigate the effect of the surface crust, some analyses were The majority of the analyses were performed with a 2 m wide
performed assuming the clay to be normally consolidated over the rigid strip surface footing. The geometry and finite element mesh
entire depth 共i.e., OCR⫽1), which results in the undrained for the 2 m wide footing are presented in Fig. 3. Because the

Fig. 3. Finite element mesh

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / SEPTEMBER 2003 / 3

Int. J. Geomech. 2003.3:1-10.


Table 3. Bearing Capacity q f of Soft Clay with Crust
Ultimate bearing capacity
of preload analyses
Footing Initial bearing
type capacity analysis 20% 40% 60% 80% 100%
2 m smooth 61 64 68 74 83 92
2 m rough 61 64 68 74 83 92
10 m smooth 58 62 69 77 87 96
10 m rough 64 72 82 92 103 112
Note: Values given are in kPa.
Downloaded from ascelibrary.org by MARQUETTE UNIVERSITY LIBRARIES on 08/17/14. Copyright ASCE. For personal use only; all rights reserved.

maintain undrained conditions in the soil. However, during the


consolidation stage the pore water pressures along this surface
were set and maintained to the initial values existing prior to
footing construction.
Small time steps were used to maintain undrained conditions
Fig. 4. Load-displacement curves for 2 m wide smooth strip footing during the loading stages. During the consolidation stages, suffi-
on soft clay
cient time was allowed for all excess pore water pressures to fully
dissipate. Typically this was achieved in 80 and 30 years for the
soft and stiff clay analyses, respectively.
problem is symmetric, only half of the geometry is considered. Some analyses have been performed with a 10 m wide strip
The bottom boundary of the mesh is at 10 m depth, while the far footing, with the same soil properties and initial conditions as for
boundary is at 20 m from the line of symmetry. For the 10 m wide the 2 m footing analyses. A comparison of results from these
footing a similar mesh is used, but with the dimensions increased analyses with those performed with a 2 m wide footing demon-
by a factor of 5. Eight-noded plane strain isoparametric elements strates the effect of the 2 m deep surface crust in soft clay on
were used, with the four corner nodes having pore water pressure, footing behavior.
as well as displacement, degrees of freedom. Reduced (2⫻2)
integration was used, with an accelerated modified Newton-
Raphson scheme with a substepping stress point algorithm Results of Analyses
for solving the nonlinear finite element equations 共Potts and
Ganendra 1994兲. All the analyses were performed using the Im-
General
perial College Finite Element Program.
Regarding the displacement boundary conditions, the two ver- For the soft clay case, assuming the profile with the surface crust,
tical boundaries of the mesh were restricted from moving in the both 2 and 10 m footings were analyzed, with both rough and
horizontal direction, while the bottom boundary was restricted smooth footing-soil interfaces. There are therefore four sets of
from moving in both the vertical and the horizontal directions. analyses, each having a range of preload values. Four additional
The top boundary from the edge of the footing to the far right was analyses were performed to investigate the initial short term 共un-
a stress-free boundary. The boundary conditions underneath the drained兲 bearing capacity of 2 and 10 m rough and smooth foot-
footing were chosen to represent either a rough or a smooth ings on a soft clay profile without the crust.
footing-soil interface. In the case of the rough interface the hori- The set of load-displacement curves for the 2 m wide smooth
zontal displacements at boundary nodes underneath the footing footing on soft clay is shown in Fig. 4. As outlined in the ‘‘Intro-
were set to zero, while in the case of the smooth interface a zero duction,’’ the first step of rapid loading from initial conditions to
horizontal force was set at these nodes. The footing itself was not failure produced the initial short term 共undrained兲 bearing capac-
discretized in the finite element mesh. Because it was considered ity 共61 kPa兲, for which the load-displacement curve in Fig. 4 is
to be rigid, the loading was simulated by applying increments of marked as ‘‘initial.’’ The following steps involve analyses in
uniform vertical displacements to nodes on the boundary under- which the footing has been preloaded to a percentage 共20, 40, 60,
neath the footing. The resulting reaction forces represent the 80, and 100%兲 of this initial short term bearing capacity and then
equivalent load. During the consolidation stages, these nodes allowed to consolidate with no change in load, before being sub-
were free to move vertically but by the same amount, with the jected to rapid loading to failure. It can be seen from the figure
total load on the footing remaining constant. To ensure that the that the higher the preload, the higher the ultimate bearing capac-
nodes were displacing uniformly during consolidation, the tied ity. A similar trend in the results occurred for the 2 m wide rough
degrees of freedom concept was used. The resulting displace- footing and for both the 10 m rough and smooth strip footings.
ments were therefore not prescribed, but were a result of the The ultimate undrained bearing capacity for each analysis is given
analysis. in Table 3.
Regarding the pore water pressure boundary conditions, A summary of initial short term 共undrained兲 bearing capacities
throughout the analysis no flow of water was allowed through the for the 2 and 10 m rough and smooth footings on soft clay soil
base of the mesh, immediately beneath the footing, nor through with and without the surface crust is given in Table 4. The com-
the left-hand boundary which forms the vertical plane of symme- parison gives an indication of the effect of the surface crust on the
try through the footing. On the right-hand vertical boundary the bearing capacity, in relation to the foundation width to depth of
pore water pressures were maintained equal to their original val- crust ratio.
ues determined by the position of the water table. During the For the stiff soil, analyses with OCRs ranging from 1 to 25
loading stages of the analyses no flow of water was allowed have been performed. For the OCR⫽6 case both smooth and
across the ground surface adjacent to the footing. This helps rough 2 m wide footings have been considered, whereas for the

4 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / SEPTEMBER 2003

Int. J. Geomech. 2003.3:1-10.


Table 4. Initial Bearing Capacity q f of Soft Clay With and Without
Crust
Footing type Soft clay with crust Soft clay without crust
2 m smooth 61 29
2 m rough 61 34
10 m smooth 58 38
10 m rough 64 52
Note: Values given are in kPa.
Downloaded from ascelibrary.org by MARQUETTE UNIVERSITY LIBRARIES on 08/17/14. Copyright ASCE. For personal use only; all rights reserved.

other OCRs analyses with only rough 2 m wide footings have


been performed. In addition, a set of analyses with a 10 m wide
rough footing has been performed for the OCR⫽6 soil. The ulti-
mate bearing capacities for each analysis are given in Table 5.
The load-displacement curves for the 2 m wide smooth footing on
the OCR⫽6 soil, following the same steps as in the soft clay
case, are shown in Fig. 5. These results again indicate that the Fig. 5. Load-displacement curves for 2 m wide smooth strip footing
ultimate bearing capacity increases with the level of preload. on stiff clay with OCR⫽6
However, the relative magnitude of the increase is smaller than
for the soft soil.
Stiff Clay
Because the initial distribution of S u with depth is linear for
the stiff clay, it is possible to obtain an estimate of the initial Results from 2 m wide smooth and rough footings on stiff clay
undrained bearing capacity using results from Davis and Booker with an OCR⫽6 are shown in Figs. 6共a and b兲, respectively. The
共1973兲. This gives a value for the smooth footing on the OCR failure mechanism for the smooth footing shows the soil imme-
⫽6 soil of q f⫽145 kPa. To obtain this value interpolation and diately under the footing having both vertical and horizontal com-
scaling from the figures supplied in the above paper are necessary. ponents of displacement. In contrast, the failure mechanism for
The Davis and Booker figures are based on calculations which the rough footing shows the soil below the foundation moving
involve some finite difference approximations and have been predominantly vertically. The difference in mechanisms occurs
drawn to provide conservative estimates for design purposes. The because of the horizontal restraints imposed at the soil-footing
above result is therefore subject to error. Nevertheless, this value interface. Clearly, the failure mechanism for the rough footing
of q f is within 5% of the finite element result 共138 kPa兲. penetrates deeper below the foundation and involves a consider-
ably larger volume of soil than that for the smooth footing. As the
undrained shear strength increases with depth, see Fig. 1 and
Failure Mechanisms
insets in Fig. 6, the deeper mechanism of the rough footing en-
Vectors of incremental displacement at failure for selected analy- gages higher undrained strength at depth and hence results in a
ses are shown in Fig. 6. Note that the plots only show vectors for larger bearing capacity 共158 kPa兲 than the smooth footing 共138
that part of the mesh in the vicinity of the footing. These vectors kPa兲, see Figs. 6共a and b兲 and Table 5. For all the stiff clay
indicate the relative magnitudes and directions of displacements analyses with different OCRs the smooth footings gave shallower
within the soil during the last increment of the analyses, when the failure mechanisms than the rough footings. The only effect of
foundation has failed. It is the relative magnitudes and the direc- OCR was to change the depth and lateral extent of the mecha-
tions of the vectors which provide an insight into the failure nisms.
mechanism. The absolute values of the displacements are not im-
portant and therefore a scale is not given on the figures. The
Soft Clay
vectors clearly indicate the nature and extent of the failure mecha-
nism for each analysis. All the results presented in Fig. 6 are from Similar distinct mechanisms are obtained for 10 m wide smooth
analyses which were performed to obtain the initial undrained and rough footings on soft clay, see Figs. 6共c and d兲. These results
bearing capacities prior to performing the preload analyses. imply that the surface crust does not have a major influence on

Table 5. Bearing Capacity q f of Stiff Clay


Ultimate bearing capacity of preload analyses
Overconsolidation ratio Footing type Initial bearing capacity analysis 20% 40% 60% 80% 100%
1 2 m rough 35 39 45 49 54 60
2 2 m rough 63 65 66 70 77 85
4 2 m rough 113 117 120 124 128 135
6 2 m smooth 138 145 151 158 158 159
6 2 m rough 158 164 172 176 179 180
6 10 m rough 272 283 294 302 311 315
9 2 m smooth 194 205 214 219 221 211
9 2 m rough 219 232 244 250 252 243
25 2 m rough 585 644 664 690 653 ⬍585
Note: Values given are in kPa.

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / SEPTEMBER 2003 / 5

Int. J. Geomech. 2003.3:1-10.


rough footings, the bearing capacity is also the same 共61 kPa兲, as
is evident from the results shown on Figs. 6共e and f兲 and in Table
4. The existence of the surface crust in this case has masked the
effect of footing roughness on the bearing capacity. However, if
the surface crust in the soft clay profile was ignored and the
undrained strength profile was one of a linear increase with depth
from the ground surface 共see dashed line in Fig. 2共a兲 and insets in
Figs. 6共g and h兲, then the failure mechanisms for 2 m wide
smooth and rough footings will be different, i.e., shallow and
deep, respectively, with the deeper mechanism again generating a
Downloaded from ascelibrary.org by MARQUETTE UNIVERSITY LIBRARIES on 08/17/14. Copyright ASCE. For personal use only; all rights reserved.

larger bearing capacity 关see Figs. 6共g and h兲 and Table 4兴.
For the analyses involving preload, the failure mechanisms
were similar to those occurring during the initial loading analysis
and discussed above. The change in the spatial distribution of S u
which occurred during the consolidation stage of these analyses,
although not changing the basic type of mechanism, did change
its vertical and lateral extent.

Effect of Preload
When the footing is initially loaded rapidly, under undrained con-
ditions, excess pore water pressures ⌬u are generated in the soil.
These excess pore water pressures can be considered to have two
components. The first 共positive兲 being due to an increase in mean
total stress 共the so-called spherical part ⌬u sph), and the second
共positive or negative兲 due to an increase in the deviatoric 共shear兲
stress ⌬u dev . The magnitudes of the two components will depend
on the footing load, but the latter will also depend on the type of
clay and its overconsolidation ratio. The interaction between the
two components is demonstrated in Fig. 7, which presents total
and effective stress paths 共TSP and ESP兲 for three undrained tri-
axial compression tests on normally and overconsolidated clay, as
predicted with the modified Cam clay model and stiff clay param-
eters given in Table 1. All three tests commence from an isotropic
stress state, i.e., p ⬘ ⫽p⫽200 kPa, u⫽0. Fig. 7共a兲 shows the ESP
and TSP for clay with an OCR⫽1. The ESP is contractant, gen-
erating positive both deviatoric and spherical pore water pressure
components. As the OCR increases 关 OCR⫽6 in Fig. 7共b兲兴, the
Fig. 6. 共a兲 2 m smooth footing on stiff clay, OCR⫽6; 共b兲 2 m rough ESP becomes more dilatant and the positive deviatoric component
footing on stiff clay, OCR⫽6; 共c兲 10 m smooth footing on soft clay; of pore water pressure reduces. In this case it is predicted as zero
共d兲 10 m rough footing on soft clay; 共e兲 2 m smooth footing on soft because modified Cam clay assumes elasticity below the yield
clay; 共f兲 2 m rough footing on soft clay; 共g兲 2 m smooth footing on surface. For real soil it would probably still be positive, but
soft clay, without crust; and 共h兲 2 m rough footing on soft clay, smaller than in the normally consolidated case. Eventually, at
without crust very high OCR 关 OCR⫽25 in Fig. 7共c兲兴, the ESP becomes com-
pletely dilatant and the deviatoric component of excess pore water
pressure becomes tensile.
The changes in the undrained shear strength of any clay during
the failure mechanism. The reason for this is that the width of the
the consolidation stages of the analyses depend directly on the
footing is much larger than the depth of the crust. Again, because
the undrained strength changes linearly with depth underneath the magnitude of the excess pore water pressures and the soil prop-
crust, the deeper rough footing mechanism engages larger erties. For a given soil 共i.e., OCR value兲 the greater the positive
strength at depth than the shallower smooth footing mechanism magnitude of the excess pore water pressure, the greater the gain
and hence results in a larger bearing capacity 共64 kPa for the in strength. If the excess pore water pressures are negative, the
former compared to 58 kPa for the latter兲. undrained strength will reduce as water will be sucked into the
This is not so for the 2 m wide footing on soft clay, as can be soil. This is clearly demonstrated in Fig. 8, which shows the dis-
seen in Figs. 6共e and f兲. In this case a deep mechanism is pre- tribution of undrained strength with depth for both the normally
dicted for both the smooth and rough footing. Here the failure consolidated and overconsolidated stiff clay after a 2 m wide
mechanism, the extent of which depends partly on the footing rough footing has been loaded to its initial bearing capacity and
width, is controlled by the 2 m deep surface crust which has a then left to consolidate. For comparison, the initial undrained
strength reducing with depth. The failure mechanisms for both the strength profile, before any loading, is also shown. The gain in
smooth and rough footings are forced into the weaker soil, as this undrained strength is larger for normally consolidated clay 关Fig.
provides a failure mechanism involving the least resistance. Be- 8共a兲兴, because it has generated a larger amount of positive pore
cause the failure mechanism is the same for both smooth and water pressure during the initial loading stage.

6 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / SEPTEMBER 2003

Int. J. Geomech. 2003.3:1-10.


Downloaded from ascelibrary.org by MARQUETTE UNIVERSITY LIBRARIES on 08/17/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 7. Interaction between spherical and deviatoric pore water pressure components in undrained triaxial compression test 共as predicted with
modified Cam clay兲

Soft Clay centage of preload the excess pore water pressures are relatively
smaller than for the 10 m wide footing and, therefore, so is the
The ultimate undrained bearing capacity, expressed as a percent-
gain in undrained strength.
age of the initial undrained bearing capacity, is plotted against the
level of preload in Fig. 9 for the 2 and 10 m smooth and rough
footings on soft clay. The gain in bearing capacity with preload is Stiff Clay
the largest for the 10 m wide rough footing 共up to a 74% increase
A similar summary plot of the ultimate undrained bearing capac-
for a 100% preload兲. This is not surprising as the deep failure
ity, expressed as a percentage of the initial undrained bearing
mechanism for this footing involves a considerable amount of
capacity, against the level of preload, is produced in Fig. 10 for all
normally consolidated clay 关see Fig. 6共d兲兴. The gain in bearing
the stiff clay analyses with a 2 m wide rough footing. The nor-
capacity for the 10 m wide smooth footing is slightly smaller,
mally consolidated stiff clay (OCR⫽1) shows a significant gain
because its failure mechanism is slightly shallower 关see Fig. 6共c兲兴.
in undrained strength due to preload 共up to 74%兲, however, as
The failure mechanisms for both the 2 m wide rough and smooth
would be expected from the above discussion based on Fig. 7, the
footing on the soft soil involve a smaller proportion of clay with
gain in ultimate undrained bearing capacity reduces with an in-
OCR⫽1 and a greater proportion of the surface crust which has a
crease in the clay’s OCR. The analyses with OCRs less than 6
high OCR 关see Figs. 6共e and f兲兴. Consequently, for a given per-
follow the same trend of an increasing gain in bearing capacity

Fig. 8. Change in undrained strength along center line for 2 m wide Fig. 9. Gain in bearing capacity due to preloading for rough footings
rough footings on 共a兲 OCR⫽1 and 共b兲 OCR⫽6 stiff clay on soft clay

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / SEPTEMBER 2003 / 7

Int. J. Geomech. 2003.3:1-10.


Downloaded from ascelibrary.org by MARQUETTE UNIVERSITY LIBRARIES on 08/17/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 10. Gain in bearing capacity due to preloading for rough foot-
ings on stiff clay

with the amount of preload. For the OCR⫽9 analyses the gain in
bearing capacity actually decreases when the preload increases
above 80%, while for the OCR⫽25 analyses the gain in bearing
capacity reduces when the preload exceeds 60%. In fact, for this
soil, when an analysis with a 100% preload was attempted, the
footing failed during the consolidation stage before it could be
loaded any further.
This behavior can be explained by considering the excess pore
water pressures developed during preloading. As noted above,
this consists of two components. For soil with a high OCR the
relative magnitude of these two components changes with the
amount of preload. For the OCR⫽9 and OCR⫽25 analyses the Fig. 11. Contours of excess pore water pressures at different pre-
component due to a change in total mean stress is compressive, loads for a 2 m rough footing on OCR⫽25 soil
whereas that due to deviatoric shearing is tensile 关see Fig. 7共c兲兴.
As the amount of preload increases above 80% for the OCR⫽9
and 60% for the OCR⫽25 analyses, the tensile component in-
creases in magnitude relative to the compressive component and, OCRs there will be a greater increase in S u for a given change in
consequently, the total magnitude of the excess pore water pres- ␴ ⬘v . This effect therefore acts in the opposite sense to that asso-
sure drops and results in a smaller increase in undrained strength ciated with the magnitude of the excess pore water pressure and
during the consolidation stage. This hypothesis is illustrated in explains the behavior observed in Fig. 12. For a 100% preload the
Fig. 11, which shows contours of excess pore water pressures at excess pore pressure effect dominates and the gain in bearing
different stages of preloading, from the analysis of a 2 m rough capacity falls continuously with an increase in OCR, see Fig. 12.
footing resting on OCR⫽25 soil. In the vicinity of the eventual In contrast, for the 80% preload the gain in bearing capacity first
failure surface the excess pore water pressures can be seen to be reduces 共excess pore water pressure effect dominating兲, then in-
more compressive as the footing preload is increased up to 60%
of the initial bearing capacity. With further preloading the excess
pore water pressures reduce and become tensile. This results in a
drop in undrained strength and therefore a reduction in ultimate
bearing capacity during swelling. Further investigations indicated
that for preloads in excess of 93%, a footing on a stiff OCR
⫽25 soil failed during the consolidation stage, before any further
load could be added.
To show more clearly the effect of OCR, the results presented
in Fig. 10 have been replotted in Fig. 12 as gain in undrained
bearing capacity against OCR 共plotted on a logarithmic scale兲 for
different amounts of preload. For preload values less than 50%
the gain in undrained bearing capacity first reduces with increas-
ing OCR, but later recovers slowly as OCR is increased still fur-
ther. This may at first seem to contradict the arguments put for-
ward above, which suggest that as the OCR increases, the relative
magnitude of the excess pore water pressures which are available
to dissipate is likely to reduce, and therefore lead to smaller gains
in undrained bearing capacity. However, as the OCR increases, so Fig. 12. Gain in bearing capacity due to preloading against OCR, for
rough footings on stiff clay
does the S u /␴ ⬘v ratio, see Table 2. Consequently, at the higher

8 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / SEPTEMBER 2003

Int. J. Geomech. 2003.3:1-10.


• An increase of bearing capacity of up to 74% is possible for
preloaded foundations on normally consolidated clays 共see
Figs. 9 and 10兲;
• For moderately overconsolidated clays (2⭐OCR⭐6) the
maximum increase of bearing capacity due to preload is only
18% and may occur at preloads less than 100% 共see Fig. 10兲;
• For heavily overconsolidated clays high levels of preloading
may cause a reduction in bearing capacity 共see Fig. 10兲;
• The reason for the difference in behavior of normally and
overconsolidated clays is the interaction between the spherical
Downloaded from ascelibrary.org by MARQUETTE UNIVERSITY LIBRARIES on 08/17/14. Copyright ASCE. For personal use only; all rights reserved.

and deviatoric components of the excess pore water pressures


generated during the preloading of the footing 共see Figs. 7 and
11兲; and
• In practice most foundations will be designed with a factor of
safety on load of at least 2. This means that the maximum
preload is therefore likely to be only 50%. The results in Fig.
Fig. 13. Effective stress paths from ideal triaxial compression tests, 10 indicate that such a preload will cause at most an increase
as predicted with modified Cam clay in undrained bearing capacity for subsequent loading of 35%
for a normally consolidated soil, but only up to 15% for a soil
creases slightly (S u /␴ ⬘v effect dominating兲 before reducing again with an OCR greater than 4. Consequently, the answer to the
共excess pore water pressure effect dominating兲 as the OCR in- question set out at the beginning of this study is that, for the
creases. majority of real situations, it is unlikely that underneath exist-
It is also of interest to note that for the stiff clay analyses in ing foundations there will be a significant improvement in
which the initial undrained shear strength increases linearly with undrained bearing capacity, capable of sustaining any substan-
depth, the gain in bearing capacity due to preload, expressed as a tial new loading.
percentage of the initial bearing capacity, is independent of foot-
ing width. This can be seen from Table 5 by comparing the results Notation
from the 2 and 10 m wide rough footings on OCR⫽6 soil. Figs.
10 and 12 are therefore applicable to a footing of any width on The following symbols are used in this paper:
stiff clay. G ⫽ elastic shear modulus;
k ⫽ coefficient of permeability;
Comment on Constitutive Model o ⫽ normally consolidated coefficient of earth
K nc
Whether or not the effect of OCR as discussed above is realistic pressure at rest;
depends on the ability of the modified Cam clay model to reflect o ⫽ overconsolidated coefficient of earth pressure
K oc
the behavior of real soils at high OCRs. The effective stress paths at rest;
predicted by the model in ideal 共no end effects兲 K o consolidated p ⫽ mean total stress 关 ⫽(␴ 1 ⫹␴ 2 ⫹␴ 3 )/3兴 ;
undrained triaxial compression tests, on an element of soil with an p ⬘ ⫽ mean effective stress 关 ⫽(␴ ⬘1 ⫹␴ ⬘2 ⫹␴ ⬘3 )/3兴 ;
initial vertical effective stress of 300 kPa and with a range of q ⫽ deviatoric stress 关 ⫽(␴ 1⬘ ⫺␴ 3⬘ ) 兴 ;
OCRs from 1 to 25, are shown in Fig. 13. Based on comparison q f ⫽ bearing capacity;
with real soil data it is likely that the model will either underes- S u ⫽ undrained strength;
timate or overestimate the tensile pore pressures generated due to ␯ 1 ⫽ specific volume at p ⬘ ⫽1 kPa on isotropic
shearing at high OCRs for low or high levels of loading, respec- virgin consolidation line;
tively. Consequently, the model is likely to overestimate the re- ␥ ⫽ bulk unit weight;
duction in strength for high levels of loading. The results pre- ⌬u ⫽ excess pore water pressure;
sented in Figs. 10 and 12 for high OCRs are therefore likely to be ⌬u dev ⫽ deviatoric component of excess pore water
conservative for design purposes for high level of preloading. pressure;
⌬u sph ⫽ spherical component of excess pore water
pressure;
Conclusions ␪ ⫽ Lode’s angle;
␬ ⫽ slope of swelling line in ␯-ln p⬘ space;
The following conclusions result from this numerical parametric
␭ ⫽ slope of virgin consolidation line in ␯-ln p⬘
study:
space;
• The failure mechanisms for the smooth and rough footings on
␮ ⫽ Poisson’s ratio;
a clay soil with a linear distribution of undrained strength with
␴ v⬘ ,␴ h⬘ ⫽ vertical and horizontal effective stress,
depth are different, the former also being shallower than the
respectively;
latter 关see Figs. 6共a, b, g, and h兲兴;
␴ 1 ,␴ 2 ,␴ 3 ⫽ principal total stresses;
• In the case of a soft clay soil, which usually has a stronger
␴ ⬘1 ,␴ ⬘2 ,␴ ⬘3 ⫽ principal effective stresses; and
surface crust above the ground water table, the shape of the
␸ ⬘ ⫽ angle of shearing resistance.
failure mechanism will depend on the size of the footing rela-
tive to the depth of the crust 关see Figs. 6共c, d, e, and f兲兴;
• Ignoring the existence of the surface crust in soft clays can References
reduce the bearing capacity of a footing significantly 共up to Brinch-Hansen, J. 共1970兲. ‘‘A revised and extended formula for bearing
50%, see Table 4兲, which may have serious implications on the capacity.’’ Bulletin No. 28, Danish Geotechnical Institute, Lyngby,
choice of foundation type; Denmark.

INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / SEPTEMBER 2003 / 9

Int. J. Geomech. 2003.3:1-10.


Carswell, I., Carder, D. R., and Gent, A. J. C. 共1993兲. ‘‘Behaviour during 265, Transport and Road Research Laboratory, Crowthorne, England.
construction of a propped contiguous bored pile wall in stiff clay at Meyerhof, G. G. 共1950兲. ‘‘A general theory of bearing capacity.’’ Building
Walthamstow.’’ TRL Project Rep. 10 (E468A/BG), Crowthorne, En- Research Station Note No. C143, Watford, England.
gland. Potts, D. M., and Ganendra, D. 共1994兲. ‘‘An evaluation of substepping
Davis, E. H., and Booker, J. R. 共1973兲. ‘‘The effect of increasing strength and implicit stress point algorithms.’’ Comput. Methods Appl. Mech.
with depth on the bearing capacity of clays.’’ Geotechnique, 18共1兲, Eng., 119, 341–354.
67–91. Potts, D. M., and Gens, A. 共1984兲. ‘‘The effect of the plastic potential in
Gens, A., and Potts, D. M. 共1988兲. ‘‘Critical state models in computa- boundary value problems involving plane strain deformations.’’ Int. J.
tional geomechanics.’’ Eng. Comput., 5共3兲, 178 –197. Numer. Analyt. Meth. Geomech., 8, 259–286.
Jackson, C., Zdravković, L., and Potts, D. M. 共1997兲. ‘‘Bearing
Potts, D. M., and Zdravković, L. 共1999兲. Finite element analysis in geo-
capacity of pre-loaded surface foundations on clay.’’ 9th Int. Conf.
technical engineering: Theory, Thomas Telford, London.
Downloaded from ascelibrary.org by MARQUETTE UNIVERSITY LIBRARIES on 08/17/14. Copyright ASCE. For personal use only; all rights reserved.

Computer Methods and Advances Geomechanics, Wuhan, China,


Terzaghi, K. 共1943兲. Theoretical soil mechanics, Wiley, New York.
745–750.
La Rochelle, P., Trak, B., Tavenas, F., and Roy, M. 共1974兲. ‘‘Failure of Trak, B. 共1980兲. ‘‘De la stabilite des remblais sur sols mous.’’ PhD thesis,
test embankment on a sensitive Champlain clay deposit.’’ Can. Geo- Univ. of Laval, Que., Canada.
tech. J., 11, 142–164. Zdravković, L., Potts, D. M., and Jardine, R. J. 共2001兲. ‘‘A parametric
Mair, R. J., Hight, D. W., and Potts, D. M. 共1992兲. ‘‘Finite element analy- study of the pull-out capacity of bucket foundations in soft clay.’’
ses of settlements above a tunnel in soft ground.’’ Contractors Rep. Geotechnique, 51共1兲, 55– 67.

10 / INTERNATIONAL JOURNAL OF GEOMECHANICS © ASCE / SEPTEMBER 2003

Int. J. Geomech. 2003.3:1-10.

You might also like