Edwards, D. H., Zdravkovic, L. & Potts, D. M. (2005). Géotechnique 55, No.

10, 755–758

TECHNICAL NOTE

Depth factors for undrained bearing capacity

D. H . E DWA R D S * , L . Z D R AVKOV I C a n d D. M . P OT T S

KEYWORDS: bearing capacity; footings/foundations; numerical footing will hence be stiff and probably rough, and will
modelling and analysis therefore differ from the assumptions made in the analyses.
Prior to Salgado et al. (2004), Hu et al. (1999), using
displacement finite element analysis, and Martin (2001) (see
INTRODUCTION also Martin & Randolph, 2001), using upper- and lower-
In engineering practice, the bearing capacity of foundations (applying the method of stress characteristics) bound ap-
(both drained and undrained) is usually estimated using proaches, addressed the same issue by considering the
well-established bearing capacity formulae (e.g. Terzaghi, vertical bearing capacity of skirted circular foundations. In
1943). These formulae are based on a number of empirical their analyses the D/B ratio and the gradient, k, of the
and semi-empirical correlations that are not uniquely defined variation of undrained strength with depth were varied, as
and have been adapted according to local design practice. well as the roughness of the skirt/soil interface, whereas the
This technical note is concerned with the undrained bear- foundation base was considered as fully rough. Subsequently,
ing capacity of foundations embedded in clay, which is Houlsby & Martin (2003), using the stress field approach,
normally calculated according to the following expression considered conical footings embedded in clay. They also
varied the D/B ratio, the cone angle (the limit being 1808,
qult ¼ sc d c Nc su þ q0 (1) i.e. a flat footing), the gradient k (the limit being 0, i.e.
constant su with depth) and the footing roughness (from
In this expression qult is the ultimate bearing stress on the
perfectly smooth—their parameter Æ ¼ 0—to perfectly
foundation (calculated as the ultimate vertical force, Qult ,
rough: Æ ¼ 1). The boundary conditions above the base of
divided by the footing base area, A); sc is the shape factor
the footing simulated those of a rigid smooth-sided shaft.
(defined as the ratio of the ultimate bearing stress of a
When the results from these studies are compared for a
surface footing of a particular shape to that of a surface strip
circular footing, constant undrained strength with depth and
footing); dc is the depth factor (usually defined as the ratio
a fully rough soil/footing base interface, a significant differ-
of the net ultimate bearing stress, qult
q0 , of a strip footing
ence in the bearing capacity factor (sc dc Nc ) is observed
at depth to that of a strip footing at the ground surface); Nc
(shown later in Fig. 3). At first sight this could be attributed
is the bearing capacity factor for a strip footing (¼ 2 +
;
to the different boundary conditions adopted for simulating
Prandtl, 1920); su is the undrained strength of the soil; and
the footing in the various studies. Consequently, the aim of
q0 is the surcharge at the footing base level. Of the bearing
the study presented in this technical note was to investigate,
capacity factors, only Nc for the strip footing on the ground
using the displacement finite element method, the merits of
surface is exact, and the factors sc and dc are semi-empiri-
the various solutions.
cal.
In one of the latest studies Salgado et al. (2004) investi-
gated this problem using the finite element limit analysis
FINITE ELEMENT ANALYSIS
approach. They performed a series of upper- and lower-
Small-strain finite element analyses of embedded strip and
bound analyses of strip, circular, square and rectangular
circular foundations were performed using the Imperial Col-
footings of width/diameter B, embedded at depth D below
lege Finite Element Program (ICFEP) (Potts & Zdravko-
the ground surface. In their analyses the load on the footing
vic, 1999). The D/B ratio was varied from 0 (footing at the
was modelled by expanding a horizontal cavity (i.e. a slot),
ground surface) to 4. Owing to the symmetry in geometry
at depth D, of the same shape and size as the foundation
and loading conditions, only half of the domain was discre-
base. Elements that were on the floor of the cavity were
tised, using 8-noded quadrilateral elements, as shown in Fig.
loaded vertically to failure, whereas the elements on the
1. To represent an embedded footing, the mesh for a surface
ceiling of the cavity had a constant normal stress applied,
footing was extended vertically, above the boundary BC, and
equal to the self-weight of the soil above. The interface of
the initial stresses updated with respect to the new ground
both surfaces with the surrounding soil was rough, and the
surface. Zero thickness interface elements (Day & Potts,
soil had a constant undrained strength with depth. Whereas
1994) were placed along the boundary ABC to simulate
they obtained a very narrow band of upper- and lower-bound
different interface conditions between the footing and the
solutions for a strip footing, the two solutions diverged at
soil.
greater depths for other foundation shapes (as shown later in
The soil was modelled using the Tresca constitutive
Figs 2 and 3). The boundary conditions adopted simulate an
model, in order to be consistent with existing design expres-
isolated thin footing at depth with soil above it. However, in
sions and the comparative studies described above. A con-
most practical situations a deep footing will either extend to
stant undrained strength, su , with depth equal to 50 kPa, a
the ground surface or support a wall. The sides of the
Young’s modulus, E, of 105 kPa and a Poisson’s ratio, , of
0.499 were assigned to the soil. The interface elements along
Manuscript received 11 April 2005; revised manuscript accepted 5
October 2005. the foundation base (boundary BC) always represented a
Discussion on this paper closes 1 June 2006; for further details see rough interface. They were assigned equal normal and shear
p. ii. stiffness of 105 kN/m3 and an undrained strength of 50 kPa.
* Department of Civil and Environmental Engineering, Imperial To model the rough interface between the side of the footing
College, London, UK. and the soil, interface elements along the boundary AB were

755
756 EDWARDS, ZDRAVKOVIC AND POTTS
B/2 Detail All analyses were performed by applying uniform vertical
D
displacements and zero horizontal displacements to the
A
nodes along the boundary ABC until failure was reached (as
shown later in Fig. 4). The ultimate load on the footing was
C B
then calculated as the sum of the vertical reaction forces
5B along the base BC. This load is equal to the net bearing
force, whose value is equivalent to sc dc Nc su A. It does not
Detail of mesh around include the resistance from any shear stresses mobilised
footing along the side of the footing.

7·5B

Fig. 1. Finite element mesh RESULTS

Initially, the bearing capacity factors from finite element
analyses for strip (Nc ) and circular (sc Nc ) surface footings
9 were compared with the known solutions to establish the
accuracy of the finite element analyses. An Nc factor of 5.18
was obtained for the strip footing, which is 0.8% in error
compared with the 2 +
solution of Prandtl (1920). In a
similar way, the sc Nc factor of 6.09 for the circular footing
8 is 0.6% in error compared with the 6.05 solution of Eason
& Shield (1960). The accuracy of the finite element analysis
results is therefore excellent, being within 1% of the closed-
form solutions. The results from all finite element analyses
are given in Table 1.
d c Nc

7
Figure 2 shows the results from the strip footing analyses,
as compared with the upper- and lower-bound calculations
of Salgado et al. (2004). The diagram shows the variation of
dc Nc (as the shape factor is equal to 1) with embedment
Lower bound, Salgado et al. (2004)
6 ratio D/B. The rough interface conditions at the base of the
Upper bound, Salgado et al.(2004)
FE, smooth-sided footing footing and the uniform undrained strength in the soil are
FE, rough-sided footing the same for both studies. The results from the finite
Skempton (1951), equation (2)
Fig. 2 from Skempton (1951)
element analyses of the smooth-sided footings fall within the
very narrow range of the upper- and lower-bound solutions,
5 thus confirming their accuracy and demonstrating that they
0 1 2 3 4 essentially simulate a smooth-sided footing, even though
D/B they model soil above the footing. For rough-sided founda-
tions the finite element results, however, have greater ulti-
Fig. 2. Strip footing mate capacity than that obtained by the upper-bound
approach. This is due to the vertical shear stress mobilised
along the side of the footing, which provides greater resis-
14
tance to the upward movement of the soil around the corner
of the footing and hence greater capacity. However, at
shallower depths (< 0.25D/B), the smooth- and rough-sided
12 solutions are equal, owing to shallow tension cracks that
form along the side of the rough-sided foundation. Also
shown in the figure are Skempton’s (1951) empirical solu-
scdcNc

10 tions for the bearing capacity factors of an embedded strip

footing. Both the actual solutions (reproduced from Fig. 2 in
Lower bound, Salgado et al. (2004)
Upper bound, Salgado et al. (2004)
8 FE, smooth-sided footing
FE, rough-sided footing
14
Houslby & Martin (2003)
Lower bound, Martin (2001)
Upper bound, Martin (2001) 12
Skempton (1951),equation (3)
6 Fig. 2 from Skempton (1951)
0 1 2 3 4 10
D/B
ä/B 5 0·3
8
scdcNc

Fig. 3. Circular footing

6
E/su 5 500, smooth-sided
4 E/su 5 2000, smooth-sided
assigned the same properties as along the base. In the case
E/su 5 2000, rough-sided
of a smooth interface, the interface elements along AB were 2
assigned the same normal stiffness, but negligible shear
stiffness (0.001 kN/m3 ) and undrained strength (1 kPa), such 0
that there was essentially no shear stress mobilised along the 0 0·5 1·0 1·5 2·0 2·5 3·0 3·5
side of the footing. In all cases the interface elements were ä/B
not allowed to sustain any tensile normal stress (i.e. a
tension crack could open). Fig. 4. Load–displacement curves
 DEPTH FACTORS FOR UNDRAINED BEARING CAPACITY 757
Table 1. Ratio of net bearing capacity to undrained strength, (qult 2 q0 )/su , for strip
and circular foundations in clay

D/B Strip Circular

Smooth-sided Rough-sided Smooth-sided Rough-sided

0 5.18 5.18 6.09 6.09

0.25 5.77 5.77 7.42 7.45
0.5 6.21 6.30 8.15 8.52
0.75 6.46 6.62 8.74 9.15
1 6.66 6.91 9.21 9.73
1.5 6.98 7.41 10.02 10.57
2 7.24 7.78 10.70 11.23
2.5 7.42 8.08 11.28 11.82
3 7.58 8.20 11.80 12.23
3.5 7.76 8.25 12.27 12.70
4 7.92 8.28 12.70 13.10

Skempton, 1951) and his simplified design equation (given rough-sided footing at the same embedment depth and E/su
below) are presented ¼ 2000. The figure shows that in all three analyses the
d c Nc ¼ 5ð1 þ 0:2D=BÞ < 7:5 (2) footing reaches a clear limit load, which was the case for all
the analyses presented in the present study. The two analyses
Figure 3 summarises the results from the circular footing with the same soil stiffness and different side roughness
analyses and compares them with those of Salgado et al. have a different limit load (as opposed to Hu et al., 1999),
(2004), Houlsby & Martin (2003) and Martin (2001) for the whereas the analyses with different soil stiffness but the
same footing and soil conditions (i.e. rough soil/base and same side roughness reach the same limit load, but clearly
smooth soil/skirt interfaces). The diagram shows the varia- after different displacements. For the analysis with E/su ¼
tion of sc dc Nc with embedment ratio D/B. Again, the finite 500 the mobilised bearing capacity factor equals 10 at a
element results for the smooth-sided footing plot within the displacement /B ¼ 0.3 and is therefore in agreement with
upper- and lower-bound solutions of Salgado et al. (2004), Hu et al. (1999). However, this is not the ultimate bearing
although for D/B . 2.0 the lower-bound solutions deviate capacity.
significantly from both the finite element and upper-bound The finite element results are also used to derive depth
results. As observed for the strip footing case, the solutions factors. This is achieved by dividing the bearing capacities
for the rough-sided circular footing plot above these upper- obtained for the footings at depth by that obtained for the
bound solutions for D/B > 0.25. In contrast, the stress field surface footing. This has been done for both strip and
solutions of Houlsby & Martin (2003) predict much smaller circular footings, and the results are presented in Fig. 5.
bearing capacity factors than the present finite element Clearly the shape of the footing has a large influence, as the
analyses, and even smaller than the lower-bound solutions of depth factors are consistently larger for circular than for
Salgado et al. (2004). However, the lower-bound solutions of strip footings. It should be noted that by plotting the data in
Martin (2001) are almost identical to those of Salgado et al. this manner it is implied that the shape factor sc is
(2004) for D/B < 1.27, whereas the upper-bound solutions unaffected by depth. An alternative interpretation used by
are much higher. Similar to the strip footing results, Fig. 3 Salgado et al. (2004) was to assume that the depth factor
also shows the empirical solutions of Skempton (1951) for derived for the strip footing applies to all footing shapes,
the bearing capacity factors of an embedded circular footing, and then to use this to determine how the shape factor sc
together with his simplified design equation varied with depth. This interpretation could equally well be
applied to the finite element results in the present study.
sc d c Nc ¼ 6ð1 þ 0:2D=BÞ < 9:0 (3)
Both options are equally valid and have their own strengths
The results from the finite element study of Hu et al. and weaknesses.
(1999) are, for clarity, not presented in Fig. 3, but they Salgado et al. (2004) suggested the following relation
provide yet another different set of results. The solutions between dc and D/B for a strip footing
from this study, for both smooth- and rough-sided embedded
circular foundation with a rough base, are almost identical, rffiffiffiffi
D
with the bearing capacity factor increasing from about 6 at d c ¼ 1 þ 0:27 (4)
D/B ¼ 0 to about 10 at D/B ¼ 1, and remaining at this B
magnitude up until D/B ¼ 5. However, it is recognised in
the study that for D/B . 1 it was not possible to reach the This relationship is compared with the finite element results
limit load, and the presented bearing capacity factors are in Fig. 5. Also shown is the curve derived from Fig. 2 of
those mobilised at a foundation displacement, , equal to Skempton (1951) by dividing the bearing capacity factor of
0.3B. In addition, the E/su ratio in the Hu et al. (1999) study embedded strip footing by its value for the footing on the
was equal to 500, which is smaller than the value of 2000 ground surface. This ratio was also calculated for the
adopted for the present finite element study. Consequently, circular footing, but as the two curves are almost identical,
the analysis for D/B ¼ 4 (the highest embedment considered only that for the strip footing is shown in Fig. 5. Both
in the present study) and a smooth-sided footing was curves are in good agreement with the finite element results
repeated with E/su ¼ 500 (i.e. E ¼ 25 MPa and su ¼ for smooth-sided strip footing and provide conservative
50 kPa). The resulting load–displacement curve is compared estimate of dc for the other cases considered. In addition,
with that from the analysis with E/su ¼ 2000 in Fig. 4. Also the suggested design equation of Skempton (1951) is also
shown in this figure is the load–displacement curve for the shown:
758 EDWARDS, ZDRAVKOVIC AND POTTS
2·50 in good agreement with the finite element results from
Strip, smooth-sided
Strip, rough-sided the present study for a smooth-sided strip footing.
Circle, smooth-sided
Circle, rough-sided
However, for all other cases they are conservative,
2·25 especially for circular footings.
Salgado et al.(2004), equation (4)
Fig. 2 from Skempton (1951)
Skempton (1951), equation (5)
2·00
NOTATION
Depth factor, dc

A foundation base area

1·75 B foundation width/diameter
D foundation embedment depth
dc depth factor
E Young’s modulus of soil
1·50 k gradient of undrained strength variation with depth
Nc bearing capacity factor
Qult ultimate vertical force on the foundations
1·25 qo overburden stress
qult ultimate bearing stress
sc shape factor
su undrained strength of soil
1·00
 foundation settlement
0 1 2 3 4  Poisson ratio
D/B

Fig. 5. Depth factors from finite element analyses

REFERENCES
Day, R. A. & Potts, D. M. (1994). Zero thickness interface
d c ¼ 1 þ 0:2ð D=BÞ < 1:5 (5) elements; numerical stability and application. Int. J. Numer.
Anal. Methods Geomech. 18, No. 10, 689–708.
Eason, G. & Shield. R. T. (1960). The plastic indentation of a
CONCLUSIONS semi-infinite solid by a perfectly rough circular punch. J. Appl.
The results from the finite element study of the undrained Math. Phys. 11, No. 1, 33–43.
Houlsby, G. T. & Martin, C. M. (2003). Undrained bearing capacity
bearing capacity of strip and circular footings embedded in factors for conical footings on clay. Géotechnique 53, No. 5,
clay have 513–520.
(a) confirmed the accuracy of the upper- and lower-bound Hu, Y., Randolph, M. F. & Watson, P. G. (1999). Bearing response
solutions of Salgado et al. (2004) for embedded strip of skirted foundation on nonhomogeneous soil. ASCE J. Geo-
tech. Geoenviron. Engng 125, No. 11, 924–935.
footings Martin, C. M. (2001). Vertical bearing capacity of skirted circular
(b) indicated that the lower-bound solutions of Salgado foundations on Tresca soil. Proc. 15th Int. Conf. Soil Mech.
et al. (2004) and Martin (2001) or Martin & Randolph Found. Engng, Istanbul, 743–746.
(2001) for circular footings embedded at D/B . 1 Martin, C. M. & Randolph, M. F. (2001). Application of the lower
require further optimisation and upper bound theorems of plasticity to collapse of circular
(c) indicated that the stress field solutions of Houlsby & foundations. Proc. 10th Int. Conf. IACMAG, Tucson, 1417–1428.
Martin (2003) for embedded circular footings produce Potts, D. M. & Zdravkovic, L. (1999). Finite element analysis in
low values of bearing capacity geotechnical engineering: Theory. London: Thomas Telford.
(d ) demonstrated that the depth factors are not unique for Prandtl L. (1920). Uber die Harte Plastischer Korper. Nachr. Ges.
Wiss. Goettingen Math. Phys. Kl., 74–85.
any foundation shape, but depend on the roughness of
Salgado, R., Lyamin, A. V., Sloan, S. W. & Yu, H. S. (2004). Two-
the vertical sides of the footing and three-dimensional bearing capacity of foundations in clay.
(e) demonstrated that depth factors for circular footings are Géotechnique 54, No. 5, 297–306.
considerably larger than those for a strip footing Skempton, A. W. (1951). The bearing capacity of clays. Proc.
( f ) demonstrated that the solutions given by Skempton Building Research Cong. London, 1, 180–189.
(1951), which are still used in engineering practice, are Terzaghi, K. (1943). Theoretical soil mechanics. New York: Wiley.