Computers and Geotechnics 69 (2015) 210–218

Contents lists available at ScienceDirect

Computers and Geotechnics

journal homepage: www.elsevier.com/locate/compgeo

Research Paper

Bearing capacity of strip footings on two-layered clay under

combined loading
Pingping Rao a,⇑, Ying Liu b, Jifei Cui a
a
Department of Civil Engineering, University of Shanghai for Science and Technology, Shanghai 200093, China
b
College of Civil Engineering and Architecture, Nanchang Institute Technology, Nanchang 330099, China

a r t i c l e i n f o a b s t r a c t

Article history: In this paper, a lower bound limit analysis in conjunction with finite elements and second-order cone
Received 31 December 2014 programming (SOCP) is used to determine the bearing capacity of a rigid strip footing placed on
Received in revised form 30 May 2015 two-layered clay subjected to inclined or eccentric loading. The footing is founded on the free surface
Accepted 31 May 2015
of the soil mass with no surcharge applied. Two types of footing–soil interfaces are considered: (1) zero
tensile capacity and (2) non-zero tensile capacity. The numerical results are presented in the form of fail-
ure envelopes in the loading plane, and the corresponding failure mechanisms are also presented. The
Keywords:
size and shape of the failure envelopes are dependent on (1) the undrained shear strength ratio (cu1/cu2),
Bearing capacity
Lower bound
where cu1 and cu2 are the undrained shear strength of the top and bottom clay, respectively, and (2) the
Second-order cone programming normalized thickness D/B of the upper clay, where D is the thickness of the top clay layer and B is the
width of the footing.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction [21], and Michalowski [22], and a continuous deformation mecha-

The ultimate capacity of strip footings placed on homogeneous
soil can be estimated using the classic Terzaghi equation and its
associated bearing capacity factors. However, in reality, soil
strength profiles are not homogeneous and may consist of distinct
layers with significantly different properties. If the thickness of the
upper layer is large compared to the width of the foundation, real-
istic estimates of the bearing capacity may be obtained from classic
bearing capacity theory based on the properties of the upper layer.
If the thickness of the upper layer is small or comparable to the
foundation width, then the classic theory may not be appropriate
[4]. Because the failure mechanism will extend to the underlying
layer, the homogeneity assumption in the classic theory fails.
The bearing capacity of strip footings on a horizontally layered
soil profile has been investigated using the limit equilibrium
method [24,8] and upper-bound limit analysis with a simplified
circular failure mechanism [3,5]. Although the upper bound limit
analysis may not lead to a conservative prediction of the limit load,
it is still popular in the literature because of its simplicity. Its accu-
racy can be improved through the use of more complex and more
realistic failure mechanisms. For example, a rigid-block collapse
mechanism was used by Florkiewicz [7], Michalowski and Shi
⇑ Corresponding author. Tel./fax: +86 21 5527 5979.
E-mail address: raopingping@usst.edu.cn (P. Rao).
nism was used by Michalowski and Shi [21] and Michalowski [22].
Semi-empirical approaches have also been proposed based on a
series of model footing tests [2]. Numerical methods such as FEM
[13,4,30], which can handle layered soil profiles, have also been
applied to this problem. Recently, Merifield et al. [19] developed
rigorous lower and upper bound solutions to the bearing capacity
of two-layered clay based on the method from Sloan [25] and Sloan
and Kleeman [26].
However, most of the previous works only consider vertical
loads. In practice, foundations can be subjected to horizontal loads
and moments, for example, the wind and wave forces in offshore
environments. Therefore, the stability of strip footings on layered
soils under combined loading is of practical interest.
Unfortunately, available work on this problem is very limited in
the literature, except for the works of Meyerhof and Hanna [20],
Georgiadis and Michalopoulos [8], Youssef-Abdel Massih et al.
[29], and Zhan [31]. The solutions of Meyerhof and Hanna [20]
for the case of inclined loading were based on a series of model
footing tests from which empirical and semi-empirical bearing
capacity factors were derived. Georgiadis and Michalopoulos [8]
developed a numerical method of slip surfaces (force equilibrium)
for layered soils both for eccentric and inclined loads.
Youssef-Abdel Massih et al. [29] investigated the bearing capacity
of a strip footing resting on a two-layer foundation soil (sand and
clay) in the case of inclined and/or eccentric loads using the

http://dx.doi.org/10.1016/j.compgeo.2015.05.018
0266-352X/Ó 2015 Elsevier Ltd. All rights reserved.
 P. Rao et al. / Computers and Geotechnics 69 (2015) 210–218
kinematical approach in limit analysis, in which the translational
(for the case of an inclined load) and rotational (for the case of
an eccentric load) failure mechanisms were used. The general fail-
ure mechanism used by Georgiadis and Michalopoulos [8] is rather
simple, and the accuracy of the results is questionable. In addition,
the solutions of Youssef-Abdel Massih et al. [29] are upper bounds
that may be unsafe, whereas the solutions of Meyerhof and Hanna
[20] are largely empirical. In summary, numerical analyses of the
ultimate bearing capacity of strip footings on layered soils under
combined loading can be improved in several aspects identified
above, particularly in terms of rigor and generality of the results.
In the work of Zhan [31], the bearing capacity of strip footings on
two-layered clay under inclined, eccentric and eccentric-oblique
loading is investigated using the finite element method, where
no sustained tension is assumed for the footing–soil interface.
However, the case of interface with tensile capacity is not investi-
gated. This refers to a fully bonded foundation–soil interface par-
ticularly relevant to offshore shallow foundations, which are
often equipped with a circumferential skirt to achieve embedment.
During loading, suctions develop within the soil plug, providing
tensile capacity for the duration over which undrained conditions
can be sustained [11].
The aim of this paper is to determine the ultimate bearing
capacity of strip footings placed on two-layered clay under inclined
or eccentric loading using the lower bound limit analysis in con-
junction with finite elements and SOCP, as described in Tang
et al. [27]. The results are presented in the form of load interaction
diagrams in the (H, M, V) space for different values of the thickness
of the upper clay layer and soil strength profiles, where H, M and V
are the horizontal force, the overturning moment, and the vertical
load, respectively. These curves define a region inside which all
allowable loading combinations should lie. In other words, they
are ultimate limit states that can be applied to design. The velocity
fields are also studied to show the failure of a strip footing on
two-layered clay subjected to combined loading. Some new
expressions for the ultimate vertical capacity and the failure envel-
opes are proposed to fit the lower bound results of the ultimate
capacity of strip footings on two-layered clay. The results pre-
sented in this paper are useful for offshore foundations, as several
types of offshore foundations are essentially shallow footings (e.g.,
the spudcan footings of jack-up units, mudmats for fixed jackets,
and concrete gravity bases).
2. Problem definition
Fig. 1 shows the case of a strip footing of width B resting on
two-layered clay, where the upper layer of clay is characterized
by undrained shear strength cu1 and thickness D. This layer is
underlain by a clay layer of undrained shear strength cu2 that
extends beyond the influence zone of the foundation (deeper than
10B). The bearing capacity will be a function of the two ratios D/B
and cu1/cu2. In this paper, solutions will be computed for problems
where D/B ranges from 0.125 to 2 and cu1/cu2 varies from 0.1 to 5.
Fig. 1. General layout of the problem.
211
This covers most problems of practical interest. Note that cu1/-
cu2 > 1 corresponds to the common case of a stiff clay layer over
a soft clay layer, while cu1/cu2 < 1 corresponds to the reverse
scenario.
The applied load on the strip footing is inclined (characterized
by the load inclination a) and eccentric (denoted as the load eccen-
tricity e) with respect to the central line of the footing, as shown in
Fig. 1, together with the stress boundary conditions. This loading
can also be represented by three statically equivalent forces V
(V = Qcosa), H (H = Qsina), and M (M = Qecosa), as shown in
Fig. 2a. All possible combinations of vertical load V, horizontal load
H, and overturning moment M causing a footing to fail delineate
the failure envelope, which can be written as a function
f(V, M, H). The aim of this paper is to determine the function
f(V, M, H) = 0, which is represented in terms of the following
dimensionless loads
V 0 H M
v0 ¼ ; h ¼ ; m0 ¼ ð1aÞ
Bcu1 Bcu1 B2 cu1
which indicates the absolute size of the failure envelope, or in terms
of normalized loads
V H M
v¼ ; h¼ ; m¼ ð1bÞ
V ult Hult M ult
where Vult, Hult, and Mult represent the ultimate vertical, horizontal,
and moment capacity, respectively. These equations indicate the
shape and relative size of the failure envelope.
Due to the symmetry of the problem, analyses with V P 0 and
H, M P 0 are adequate to define the complete envelope in each
of the VH and VM loading planes. Therefore, analyses for inclined
and eccentric loadings are only performed for (V P 0, H P 0) and
(V P 0, M P 0), respectively. The definition for the positive loading
direction is presented in Fig. 2b.
3. Finite element formulation of lower bound limit analysis
The following section is a brief summary of the use of special
finite element formulations and second-order cone programming
to compute lower bound solutions.
3.1. Statically admissible stress field
The lower bound theorem states that the collapse load calcu-
lated from a statically admissible stress field is a lower bound to
the actual collapse load. For a stress field to be statically admissi-
ble, the following conditions need to be satisfied: (a) equilibrium
within each element; (b) continuity of normal and shear stresses
along the interface between two adjacent soil elements; (c) equi-
librium and/or compatibility at the boundaries; and (d)
non-violation of the Tresca criterion in the soil mass. In the event
of a general shear failure, the magnitude of the mobilized shear
stress st along the footing–soil interface should not exceed the
shear strength of the soil mass, namely, |st| 6 cu1. The stresses over
the loaded segment must satisfy the force and moment equilib-
rium in the case of eccentric and inclined loading. If no tension
can be sustained across the footing–soil interface, separation can
occur along the interface under the moment load. In this case,
the Tresca yield criterion must be modified to include tension
cut-off (i.e., rn 6 0). These considerations were discussed in
Ukritchon et al. [28] and Tang et al. [27] and then applied in this
study.
212 P. Rao et al. / Computers and Geotechnics 69 (2015) 210–218
Fig. 2. General load conditions and their equivalent terms for a surface foundation.
3.2. Mesh details
The domain is discretized into a number of three-node triangu-
lar elements. In this case, the unknown stresses with each element
vary linearly. Due to the high stress gradient that cannot be cap-
tured using the linear element and the change in the direction of
the principal stress, a very fine fan of elements at the edges of
the footing is desirable. An illustration of the finite element mesh
is shown in Fig. 3, where D/B = 0.5. This mesh is generated using
the ABAQUS software [15].
Unlike the usual form of mesh used by FE analysis, each node is
unique to a particular element and more than one node may share
the same coordinates. The interface between two clay layers is rep-
resented as a straight line. According to the mesh feature described
earlier, each node on the straight line would in fact be two nodes
with the same coordinates but different node numbers. The nodal
stresses developed at the lower and upper sides of the interface
belong to two adjacent elements and satisfy the yield criterion
for the lower and upper clay layer, respectively. Due to the conti-
nuity of the normal and shear stress along the interface [i.e., condi-
tion (b) described earlier], the equilibrium condition is satisfied
naturally. On this basis, it is easier to address a layered soil profile
than with conventional FE analysis by assigning it to each node in
the present methodology. Consequently, the stress field can be
modeled using the three-node triangular elements, as shown in
Fig. 3, to construct a statically admissible stress field.
3.3. Finite element formulation
For linear elements, the constraints arising from the conditions
(a–c) and force and moment equilibrium along the footing–soil
interface give rise to linear equality constraints on the nodal stres-
ses, and the Tresca yield criterion [i.e., condition (d)] can be
expressed as a set of second-order cone constraints (see [6]). The
tension cut-off leads to a set of inequality constraints. The objec-
tive function is to maximize the integration of the normal stress
rn along the footing–soil interface, which is also linear. After
assembling the objective function coefficients and constraints on
each node, the present formulation leads to a SOCP problem, which
can be efficiently solved using MOSEK ApS [23]. After obtaining the
vertical bearing capacity V, the horizontal load H and moment M
can be computed as H = Vtana and M = Ve, according to the force
and moment equilibrium condition. Full details of the numerical
procedures can be found in Tang et al. [27] and will not be repeated
here. The only difference between the present work and Tang et al.
[27] is the yield criteria, which are the Tresca and Mohr-Coulomb
criterion, respectively. However, the expressions for the
second-order cone constraint are similar.
In the present work, the dimension of the chosen domain is 16B
in the horizontal direction and 8B in the vertical direction. These
dimensions ensure that the failure mechanism is fully contained
within the chosen domain. The FE mesh consists of 16,162 ele-
ments and 48,486 nodes. The total number of edges between any
two adjacent elements is 23,997. Its accuracy can be evaluated
by computing the bearing capacity factor Nc of surface strip foot-
ings on homogeneous clay. The exact plasticity solution is
2 + p  5.14, while the lower bound solution is 5.11. The error is
0.6% suggesting that this mesh can give an accurate prediction of
the bearing capacity.
4. Results and discussion
4.1. Vertical bearing capacity
4.1.1. Lower bound results
The computed lower bound values of the dimensionless vertical
load v 0 for various values of D/B are illustrated graphically in
Fig. 4a–d. These plots show that the present analyses provide bet-
ter lower bound solutions than those of Merifield et al. [19]. This
can be explained by the fact that a finer fan mesh at the edges of
the footing is used in the present work. In addition, the Tresca cri-
terion is approximated by a linear function of the unknown stres-
ses in the method of Merifield et al. [19], leading to a linear
programming problem, while the yield function is treated in its
native form in the present method. The advantages of SOCP com-
pared to the method based on linear programming [25] and non-
linear programming [17] can be found in Tang et al. [27]. The
true collapse load can be bracketed within 11% or better by com-
bining the present lower bounds with the upper bounds of
Merifield et al. [19].
Fig. 4a–c clearly illustrate that the bearing capacity increases as
cu1/cu2 decreases and becomes a constant value when cu1/cu2 -
6 (cu1/cu2)crit [here (cu1/cu2)crit denotes the critical value]. The loca-
tion of the critical value is highlighted by an arrow in Fig. 4. These
constant values for different thickness of the upper layer (different
D/B) reflect that cu1/cu2 does not affect the ultimate capacity of the
foundation. This phenomenon can be explained by the fact that the
failure mechanism is completely contained within the upper soil
 P. Rao et al. / Computers and Geotechnics 69 (2015) 210–218 213

Fig. 3. Illustration of the finite element mesh used for the lower-bound limit analysis.

Fig. 4. Comparison of the dimensionless vertical limit load V/Bcu1 with those of Merifield et al. [19].

layer and the horizontal and vertical extent of the failure mecha-
nism remain unchanged, as shown in Fig. 5 for cu1/cu2 = 0.2 with
D/B = 0.25. In this case, the dimensionless vertical load v 0 = V/Bcu1
is equal to 6.16. The existence of a stiff clay underneath a thin soft
clay leads to the increased bearing capacity of foundation.
Particularly, for cu1/cu2 = 0, which corresponds to the special case
of the bearing capacity of soil on a rigid base, the bearing capacity
of shallow foundations under vertical loading has been discussed
by Mandel and Salençon [18].
The reduction of the bearing capacity for cu1/cu2 P (cu1/cu2)crit
with increased cu1/cu2 can be easily understood with the failure
mechanism for cu1/cu2 = 0.8 and 2 with D/B = 0.25, as illustrated
in Fig. 6a and b. The failure surface penetrates into the soft clay,
which leads to a reduced bearing capacity, indicating the possibil-
ity of punching shear failure. When D/B = 2, Fig. 4d shows that the
bearing capacity for all values of cu1/cu2 between 0 and 4 is close to
that of the homogeneous soil profile. Some values of (cu1/cu2)crit for
214 P. Rao et al. / Computers and Geotechnics 69 (2015) 210–218

Table 1
Polynomial coefficients for the dimensionless vertical capacity and critical value of
cu1/cu2.

D/B (cu1/cu2)crit Coefficients

a1 a0
0.25 0.7 0.69 5.05
0.5 0.9 0.51 5.21
0.75 1 0.4 5.42
1 1.3 0.33 5.75
Fig. 5. Failure mechanism for D/B = 0.25 and cu1/cu2 = 0.2. 1.25 1.8 0.28 6.11
1.5 2.5 0.23 6.36
1.75 3.4 0.19 6.42

different values of D/B are presented in Table 1, showing that (cu1/-

cu2)crit increases with an increasing D/B.

4.1.2. Interpolation R2, which is almost 1 for all cases, which indicates that Eq. (2)
For cu1/cu2 > (cu1/cu2)crit, the variation of the bearing capacity can provide a reasonable prediction of the vertical bearing capacity
with the strength ratio cu1/cu2 can be written as the following expo- of surface footings on two-layered clay.
nential function

v 0 ¼ a0 ðcu1 =cu2 Þa 1
ð2aÞ 4.2. VH loading plane

where v 0 = V/Bcu1. The values of the coefficients a1 and a0 are
obtained from the linear regression analysis for v 0 and cu1/cu2 on a
log-log scale using the Matlab function ‘‘polyfit’’. Some results for
the coefficients a0 and a1 are summarized in Table 1.
When cu1/cu2 6 (cu1/cu2)crit, Fig. 4a–d show that the bearing
capacity becomes a constant that is a function of D/B only.
Therefore, the bearing capacity can be approximated as
v ¼ b1 expðb2 D=BÞ þ b3
where the coefficients b1, b2 and b3 are determined using the Matlab
function ‘‘lsqcurvefit’’ in the least-square sense. Our lower bound
values are best fitted using b1 = 8.39, b2 = 8.29, and b3 = 5.11.
Fig. 7 shows a good agreement between the predictions of Eq.
(2) and the lower bound results, where cu1/cu2 6 5. For the sake
of brevity, only the results for the cases of D/B = 0.125, 0.25, 0.5
and 1 are shown in Fig. 7. The accuracy of the regression analysis
can also be measured using the coefficient of determination or
Fig. 8 shows the failure envelope in terms of the dimensionless
0
loads h = H/Bcu1 and v 0 = V/Bcu1, which were calculated for all com-
binations of parameters: cu1/cu2 = 0.2, 1, 2, 3, and 4; and D/B = 0.25,
0.5, 1 and 2. The expansion of the failure envelope along the verti-
cal load axis suggests that the load-carrying capacity increased
with the decreasing strength ratio cu1/cu2, as illustrated in
Fig. 8a–c. For a foundation resting on a stiff-over-soft clay deposit
ð2bÞ (i.e., cu1/cu2 P 1), the soft bottom clay reduces the bearing capacity
for D/B 6 1, in which a punching shear failure may occur. In con-
trast, the stiff bottom clay increases the bearing capacity. In addi-
tion, all failure envelopes for obliquely loaded footings are
constrained at h = 1, owing to the sliding failure criterion. The
strength ratio cu1/cu2 does not affect the failure envelope for
D/B = 2, which is close to the exact plasticity solution provided
by Green [12], as shown in Fig. 8d.
Fig. 9 shows the failure envelope in terms of the normalized
loads h = H/Hult and v = V/Vult. Here, Vult is the vertical bearing
capacity, which varies for each curve. Hult is the ultimate horizontal
load capacity, which depends only on the footing width B and the
shear strength cu1 of the upper soil layer. Therefore, Hult is equal to
Bcu1 irrespective of the degree of soil strength non-homogeneity in
the vertical direction. When the thickness D of the upper clay is
small or comparable to the footing width B, the shape of the failure
envelope is not unique and the size of the failure envelope
increases with increasing strength ratio cu1/cu2, as shown in

Fig. 6. Failure mechanisms for D/B = 0.25: (a) cu1/cu2 = 0.8; and (b) cu1/cu2 = 2. Fig. 7. The variation of the ultimate bearing capacity and its interpolation.
 P. Rao et al. / Computers and Geotechnics 69 (2015) 210–218 215

Fig. 8. HV failure envelopes in terms of h0 = H/Bcu1 and v0 = V/Bcu1.

Fig. 9. HV failure envelopes in terms of h = H/Hult and v = V/Vult.

Fig. 9a–c. For a foundation resting on a soft-over-stiff clay deposit As a special case, the failure envelope for D/B = 2 is close to the
(i.e., cu1 6 cu2), the failure envelope is close to the failure envelope failure envelope of Eq. (3) given by Gourvenec [10], as illustrated in
for cu1 = cu2. The failure envelope for cu1 = cu2 provides a lower limit Fig. 9d.
of the failure envelope for a foundation resting on a two-layered
clay deposit (i.e., cu1 > cu2). By taking advantage of this finding, h ¼ 4v ð1  v Þ
the failure envelope for cu1 = cu2 can provide a conservative approx- ð3Þ
h ¼ 1 for v 6 0:5
imation expression of the envelopes for an obliquely loaded foun-
dation resting on two-layered clay. which is an algebraic fit to the Green solution.
216 P. Rao et al. / Computers and Geotechnics 69 (2015) 210–218
The failure mechanism for a = 15° is presented in Fig. 10 and is
similar to the mechanism used in the method of characteristics by
Hansen [14].
4.3. VM loading plane
4.3.1. Interface with zero tensile capacity
In the case of a footing–soil interface with zero tensile capacity,
separation between the soil and foundation may occur under
moment loading even for low values of vertical load, which causes
a reduced contact area and a reduction in the moment capacity of
the foundation [16]. The failure envelope can be stated in terms of
the normalized loads v = V/Vult and m = M/Mult, where Vult and Mult
are the ultimate vertical and moment capacity, respectively. At the
limit of zero vertical loads, no moment can be sustained. The initial
slope of the failure envelope at the origin on the VM loading plane
is constant for all cases considered here. For D/B = 0.25, the failure
envelopes are symmetric about V = 0.5Vult, where the ultimate
moment is mobilized, as illustrated in Fig. 11a. However, as D/B
increases, the failure envelopes for footings on a stiff-over-soft clay
deposit become asymmetric when D/B 6 1, as shown in
Fig. 11b and c. Fig. 11d shows the failure envelopes for D/B = 2
and clearly shows that the strength ratio cu1/cu2 does not affect
the failure envelopes. In this case, the calculated failure envelopes
Fig. 10. Failure mechanism for an obliquely loaded foundation.
Fig. 11. MV failure envelopes in terms of v = V/Vult and m = M/Mult for a footing–soil interface with zero tensile capacity.
are equivalent to those of the homogeneous soil layer. According to
the effective width rule B0 = B  2e, the failure envelope can be
expressed as the following equation (considering that e = M/V
and that V/Vult = B0 /B)
m ¼ 4v ð1  v Þ ð4Þ
Eq. (4) has also obtained by Houlsby and Puzrin [16] using the scal-
ing concept, which is an extension of the effective width rule.
Fig. 10d shows that the lower bound solution is in good agreement
with Eq. (4).
Fig. 12 shows the failure mechanism of eccentrically loaded
footings for e/B = 1/3, which is similar to the scoop-wedge mecha-
nism, as suggested by Bransby and Randolph [1].
4.3.2. Interface with non-zero tensile capacity
For a footing–soil interface with non-zero tensile capacity, the
maximum moment capacity can be mobilized at zero vertical loads
due to the adhesion along the interface, as shown in Fig. 13a–d.
This difference between the failure envelopes for footing–soil
interfaces with tensile capacity and with non-zero tensile capacity
has also been observed by Gourvenec and Randolph [9]. The failure
envelope in terms of v = V/Vult and m = M/Mult is a function of the
strength ratio cu1/cu2 and D/B. Similarly, the failure envelope for
Fig. 12. Failure mechanism for an eccentrically loaded foundation.
 P. Rao et al. / Computers and Geotechnics 69 (2015) 210–218 217

Fig. 13. MV failure envelopes in terms of v = V/Vult and m = M/Mult for a footing–soil interface with non-zero tensile capacity.

Fig. 14. MV failure envelopes in terms of v0 = V/Bcu1 and m0 = M/B2cu1 for footing–soil interfaces with zero tensile capacity and non-zero tensile capacity.

D/B = 2 is also equivalent to the case of a homogeneous soil layer,
which is close to the solution given by Gourvenec and Randolph
[9], as shown in Fig. 13d. Fig. 13a and b show that the failure envel-
opes for D/B = 0.25 and 0.5 fall in a very tight band that can also be
approximated according to the solution of Gourvenec and
Randolph [9]. Fig. 13c shows that the size of the failure envelope
for D/B = 1 decreases with increasing strength ratio cu1/cu2 and that
the failure envelope for cu1 = cu2 provides a lower limit with respect
to that of cu1/cu2 > 1. By taking advantage of this feature, the failure
envelope for cu1 = cu2 can provide a conservative approximation
expression to the failure envelope of an eccentrically loaded foun-
dation resting on two-layered clay. According to the work of
Gourvenec and Randolph [9], the failure envelope for cu1 = cu2 can
be described by the following power law
v ¼ ð1  mÞ0:23 ð5Þ
218 P. Rao et al. / Computers and Geotechnics 69 (2015) 210–218
4.3.3. Remarks
Fig. 14 shows the failure envelopes in terms of v 0 = V/Bcu1 and
m0 = M/B2cu1, which were calculated for all combinations of the
parameters: cu1/cu2 = 0.2, 0.5, 1, 2, 3, and 4; and D/B = 0.25, 0.5, 1
and 2. These failure envelopes represent the absolute size of the
failure envelope. When the thickness D of the upper soil layer is
small or comparable to the footing width B (i.e., D 6 B), the expan-
sion of the failure envelopes along both load axes indicates the
increased load-carrying capacity available with the decreased
strength ratio cu1/cu2, as shown in Fig. 14a–c.
According to the above discussion, the complexity and variabil-
ity of the calculated failure envelopes imply that simple equations
such as those previously recommended for homogeneous soil pro-
files (e.g., [1,9]) cannot describe the capacity of shallow founda-
tions for a range of different shear strength ratios and
thicknesses of the upper soil layer. Therefore, the design charts
shown in Figs. 8 and 14 can be used in practice.
5. Conclusions
A series of lower bound limit analyses in conjunction with finite
elements and SOCP are conducted to investigate the capacity of
strip footings on two-layered clays under combined vertical, hori-
zontal and moment loading. The results for the ultimate vertical
bearing capacity are presented for validation, which can be approx-
imated according to Eqs. (2a) and (2b). In addition, the failure
envelopes in the MV and HV loading spaces are investigated, which
has not previously been undertaken. Based on the results of the
numerical study presented in the paper, we reach the following
conclusions:
(1) The ultimate vertical capacity is shown to be proportional to
the strength ratio cu1/cu2 in the log–log scale for cu1/cu2 < (-
cu1/cu2)crit, and reaches to the limit value when cu1/cu2 -
P (cu1/cu2)crit, as shown in Eq. (2). Good agreement has
been obtained with the available theoretical solutions for
the ultimate capacity.
(2) For footings on soft clay overlying stiff clay (i.e., cu1 < cu2), the
failure envelope in terms of dimensionless loads h0 = H/Bcu1
and v0 = V/Bcu1 can be described by the failure envelope for
cu1 = cu2 (see Fig. 8). This treatment leads to predictions that
are on the conservative side. The failure envelope in terms of
v = V/Vult and m = M/Mult (see Fig. 13) is also applicable in the
case of a footing–soil interface with non-zero tensile
capacity.
(3) For footings on stiff clay overlying soft clay (i.e., cu1 > cu2), the
undrained shear strength ratio cu1/cu2 and the thickness D/B
of the upper soil layer can significantly affect the shape of
the failure envelopes. Because the failure envelopes in VH
or HM load space vary in a quite complex manner with cu1/-
cu2 and D/B, no simple expressions were found. For practical
applications, the present lower bound analyses can be used,
which provide a conservative prediction of the bearing
capacity.
Acknowledgements
This work is supported by the National Natural Science
Foundation of China (project number: 51208301) and the
Funding Scheme for Training Young Teachers for Universities in
Shanghai (project number: slg12019). The authors wish to express
their gratitude for the above financial support.
References
[1] Bransby MF, Randolph MF. Combined loading of skirted foundations.
Géotechnique 1998;48(5):637–55.
[2] Brown JD, Meyerhof GG. Experimental study of bearing capacity in layered
clays. In: Proceedings of 7th international conference on soil mechanics and
foundation engineering, Mexico, vol. 2; 1969. p. 45–51.
[3] Button SJ. The bearing capacity of footings on a two-layer cohesive subsoil. In:
Proceedings of 3rd international conference on soil mechanics and foundation
engineering, Zürich, vol. 1; 1953. p. 332–5.
[4] Burd HJ, Frydman S. Bearing capacity of plane strain footings on layered soils.
Can Geotech J 1997;34:241–53.
[5] Chen WF. Limit analysis and soil plasticity. Amsterdam: Elsevier; 1975.
[6] Ciria H, Peraire J, Bonet J. Mesh adaptive computation of upper and lower
bounds in limit analysis. Int J Numer Meth Eng 2008;75:899–944.
[7] Florkiewicz A. Upper bound to bearing capacity of layered soils. Can Geotech J
1989;26:730–6.
[8] Georgiadis M, Michalopoulos AP. Bearing capacity of gravity bases on layered
soil. J Geotech Eng 1985;111(6):712–29.
[9] Gourvenec S, Randolph M. Effect of strength non-homogeneity on the shape of
failure envelopes for combined loading of strip and circular foundations on
clay. Géotechnique 2003;53(6):575–86.
[10] Gourvenec S. Failure envelopes for offshore shallow foundations under general
loading. Géotechnique 2007;57(9):715–28.
[11] Gourvenec S. Effect of embedment on the undrained capacity of shallow
foundations under general loading. Géotechnique 2008;58(3):177–85.
[12] Green AP. The plastic yielding of metal junctions due to combined shear and
pressure. J Mech Phys Solids 1954;2(3):197–211.
[13] Griffiths DV. Computation of bearing capacity on layered soil. In: Proceedings
of 4th international conference of numerical methods in geomechanics,
Edmonton, Alberta, Canada, May. Balkema, Rotterdam, the Netherlands, vol.
1; 1982. p. 163–70.
[14] Hansen JB. A revised and extended formula for bearing capacity. Danish
Geotechnical Institute, Bulletin No. 11; 1970. p. 38-46.
[15] HKS. ABAQUS users’ manual, version 6.4’’. Hibbit, Karlsson and Sorensen Inc;
2004.
[16] Houlsby GT, Puzrin AM. The bearing capacity of a strip footing on clay under
combined loading. Proc R Soc Lond Ser A 1999;455:893–916.
[17] Lyamin AV, Sloan SW. Lower-bound limit analysis using nonlinear
programming. Int J Numer Methods Eng 2002;55:573–611.
[18] Mandel J, Salençon J. Force portante d’un sol sur une assise rigide (étude
théorique) (Bearing capacity of soil on rigid layer (theoretical study)).
Géotechnique 1972;22(1):79–93.
[19] Merifield RS, Sloan SW, Yu HS. Rigorous plasticity solutions for the bearing
capacity of two-layered clays. Géotechnique 1999;49(4):471–90.
[20] Meyerhof GG, Hanna AM. Ultimate bearing capacity of foundations on layered
soils under inclined load. Can Geotech J 1978;15(4):565–72.
[21] Michalowski RL, Shi L. Bearing capacity of footings over two-layer foundation
soils. J Geotech Eng ASCE 1995;121(5):421–8.
[22] Michalowski RL. Collapse loads over two-layer clay foundation soils. Soils
Found 2002;42(1):1–7.
[23] MOSEK ApS, 2013. The MOSEK Optimization Tools Manual Version 7.0
(Revision 81) <http://www.mosek.com>.
[24] Reddy AS, Srinivasan RJ. Bearing capacity of footings on layered clays. J Soil
Mech Found Div ASCE 1967;93(2):83–99.
[25] Sloan SW. Lower-bound limit analysis using finite elements and linear
programming. Int J Numer Anal Methods Geomech 1988;12(1):61–77.
[26] Sloan SW, Kleeman PW. Upper-bound limit analysis using discontinuous
velocity fields. Comput Methods Appl Mech Eng 1995;127:293–314.
[27] Tang C, Phoon KK, Toh KC. Effect of footing width on Nc and failure envelope of
eccentrically and obliquely loaded strip footings on sand. Can Geotech J 2014.
http://dx.doi.org/10.1139/cgj-2013-0378.
[28] Ukritchon B, Whittle AJ, Sloan SW. Undrained limit analyses for combined
loading of strip footings on clay. J Geotech Geoenviron Eng
1998;124(3):265–76.
[29] Youssef Abdel Massih DS, El-Hachem E, Soubra A-H, 2005. Bearing capacity of
eccentrically and/or obliquely loaded strip footing over two-layer foundation
soil by a kinematical approach. In: VIII International conference on
computational plasticity (COMPLAS VIII), Barcelona.
[30] Yu L, Liu J, Kong XJ, Hu YX. Three-dimensional large deformation FE analysis of
square footings in two-layered clays. J Geotech Geoenviron Eng
2011;137(1):52–8.
[31] Zhan YG. Undrained bearing capacity behavior of strip footings on two-layer
clay soil under combined loading. Electr J Geotech Eng (EJGE) 2011. <www.
ejge.com/2011/Ppr11.017/Ppr11.017r.pdf>.