Upper-bound solutions for bearing capacity of

foundations
Abdul-Hamid Soubra

To cite this version:

Abdul-Hamid Soubra. Upper-bound solutions for bearing capacity of foundations. Journal of Geotech-
nical and Geoenvironmental Engineering, American Society of Civil Engineers, 1999, 125 (1), pp.59-68.
�10.1061/(ASCE)1090-0241(1999)125:1(59)�. �hal-01007412�

HAL Id: hal-01007412

https://hal.archives-ouvertes.fr/hal-01007412
Submitted on 16 Jun 2014

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.

Distributed under a Creative Commons Attribution| 4.0 International License

 UPPER-BOUND SOLUTIONS FOR BEARING CAPACITY OF FOUNDATIONS
By Abdul-Hamid Soubra1

ABSTRACT: The static and seismic bearing capacity problem of shallow strip footings is investigated. Two
kinematically admissible failure mechanisms M1 and M2 are considered in the framework of the upper-bound
method of the limit analysis theory. The M1 mechanism is symmetrical, and it permits the calculation of the
bearing capacity in the case of no-seismic loading. It is composed of a triangular active wedge under the footing
and two radial shear zones composed of a sequence of rigid triangles. The M2 mechanism is nonsymmetrical and
is composed of a single radial shear zone. This mechanism permits the calculation of the bearing capacity in the
presence of seismic loading. Quasi-static representation of earthquake effects using the seismic coefficient concept
is adopted. The solutions obtained are rigorous upper-bound ones in the framework of the limit analysis theory.
The numerical results of the static and seismic bearing capacity factors are presented in the form of design charts
for practical use in geotechnical engineering. These results are compared with results of other authors.

INTRODUCTION power dissipated by any kinematically admissible velocity

The bearing capacity of strip footings in no-seismic areas has field can be equated to the power dissipated by the external
been extensively studied by several investigators [Terzaghi loads and so enables a strict upper-bound on the true limit
(1943), Caquot and Kérisel (1953), Meyerhof (1963), Vesic load to be deduced. A kinematically admissible velocity field
(1973), and Chen (1975) among others]. However, very few is one that satisfies compatibility, the flow rule, and the ve-
attempts have been made to study the effect of an earthquake locity boundary conditions. To provide solutions that are use-
on the bearing capacity of foundations. The few studies avail- ful in practice, the upper-bound theorem is often used in tan-
able in the literature describing the seismic effect on the bearing dem with the lower-bound theorem. The latter also assumes a
capacity concern the work of Meyerhof (1951) and Shinohara perfectly plastic soil model with an associated flow rule and
et al. (1960). Both approaches are pseudostatic: horizontal and states that any statically admissible stress field (which satisfies
vertical accelerations are applied to the center of gravity of the equilibrium and the stress boundary conditions and nowhere
structure and the problem is reduced to a static case of bearing violates the yield criterion) will furnish a lower-bound estimate
capacity with inclined eccentric loads. However, in these solu- of the true limit load. By using these two theorems, the exact
tions, the inertia of the soil mass is not included. Sarma and limit load can often be bracketed with an accuracy that is
Iossifelis (1990) and Richards et al. (1993) suggested more rig- sufficient for design purposes.
orous approaches for calculating the seismic bearing capacity In this paper, only the upper-bound theorem of limit analysis
of strip footings in seismic areas by considering the inertia is applied to the static and seismic bearing capacity problem
forces on all parts of the soil-structure system (soil and foun- using kinematically admissible velocity fields. It should be
dation). The theoretical approaches they used are based on the noted here, that the upper-bound theorem gives an unsafe es-
limit-equilibrium method. It is well known that this method timate of the ultimate bearing capacity. The aim of this work
gives an approximate solution of the failure load and that the is to improve the best available upper-bound solutions given
solution cannot be said to be an upper- or a lower-bound one by Chen (1975) in the symmetrical failure mechanism and by
with respect to the exact solution. Recently, Dormieux and Soubra (1997) in the nonsymmetrical mechanism.
Pecker (1995) and Soubra (1997) used the upper-bound method THEORETICAL ANALYSIS OF STATIC AND SEISMIC
of the limit analysis theory and developed upper-bound solu- BEARING CAPACITY PROBLEM
tions of the seismic bearing capacity factors. These solutions
are rigorous upper-bound ones with respect to the exact solu- A soil-foundation system with translational movement is as-
tions for an associated flow rule Coulomb material. sumed. Two distinct translational failure mechanisms, referred
In this paper, both the static and seismic bearing capacity to as the M1 and M2 mechanisms, are utilized in the analysis.
problems are investigated by the upper-bound method of the In the following investigation, the terms ‘‘mechanism’’ and
limit analysis theory using respectively symmetrical and non- ‘‘velocity field’’ will be used interchangeably. Note that the
symmetrical failure mechanisms. These mechanisms allow the velocity fields used are composed of rigid blocks that move
slip surface to develop more freely in comparison with the with constant velocities. Since no general plastic deformation
available mechanisms given by Chen (1975) and Soubra of the soil mass is permitted to occur, the power is dissipated
(1997); hence, they lead to smaller upper-bound solutions of solely at the interfaces between adjacent blocks, which con-
the bearing capacity problem. stitute velocity discontinuities.
The soil is homogeneous and isotropic. It is assumed to be
UPPER AND LOWER BOUND THEOREMS OF LIMIT an associated flow rule Coulomb material obeying Hill’s max-
ANALYSIS imal work principle. The consequence of applying the nor-
The upper-bound theorem, which assumes a perfectly plastic mality condition to a frictional soil with its angle of internal
soil model with an associated flow rule, states that the internal friction equal to ␾ will be a necessary occurrence of a volume
expansion with ⌿ = ␾ during the plastic flow. However, fric-
1
Maı̂tre de Conférences, Ecole Nationale Supérieure des Arts et In- tional soils are found experimentally to dilate at increments
dustries de Strasbourg, 24, Bd de la victoire, 67084 Strasbourg cedex, considerably less than those predicted by the normality con-
France. dition, that is, ⌿ < ␾. Hence, real soils do not obey the as-
sociative flow rule. Furthermore, it is well known that a non-
associative material cannot be stronger than the associative
one. On the other hand, it should be mentioned that in trans-
lational failure mechanisms, the energy balance approach for
evaluation of the limit load is always equivalent to the equi-
librium of forces approach, because the energy balance equa-

1
tion can be interpreted as an expression of the virtual rate of pacity in the case of no-seismic loading. The wedge ABC is
work principle. This observation has often been made (Davis translating vertically as a rigid body with the same initial
1968; Mroz and Drescher 1969; Michalowski 1989; Salençon downward velocity as the footing. The downward movement
1990; De Buhan and Salençon 1993; Drescher and Detournay of the footing and wedge is accommodated by the lateral
1993). The equivalence of the two approaches plays a key role movement of the adjacent soil as indicated by the two radial
in the derivations of the limit loads for nonassociative mate- shear zones. The angles ␪, ␣i, and ␤i (i = 1, . . . , n) are as yet
rials. Recent theoretical considerations made on translational unspecified. Since the movement is symmetrical about the
failure mechanisms (Drescher and Detournay 1993; Michal- footing, it is only necessary to consider the movement on the
owski and Shi 1995, 1996) allow one to conclude that for a right-hand side of Fig. 1.
nonassociative material, the limit load can be obtained by the The radial shear zone BCD is composed of n triangular rigid
use of the flow rule associated with a new yield condition in blocks. As shown in Fig. 2(a), all the triangles move as rigid
which c and ␾ are replaced by c* and ␾* as follows: bodies in directions that make an angle ␾ with the disconti-
nuity lines di (i = 1, . . . , n). The velocity of each triangle is
cos ⌿ sin ␾
tan ␾* = (1) determined by the condition that the relative velocity between
1 ⫺ sin ⌿ sin ␾ the triangles in contact must have the direction that makes an
cos ⌿ cos ␾ angle ␾ to the contact surface. The velocity hodographs are
c* = c (2) shown in Fig. 2(b). The velocities so determined constitute a
1 ⫺ sin ⌿ sin ␾
kinematically admissible velocity field.
Hence, the results presented in the present paper can be used As shown in Fig. 3, the external forces contributing to the
for nonassociative material provided the internal friction angle incremental external work consist of the foundation load, the
␾ and the cohesion c are replaced with ␾* and c* calculated weight of the soil mass, and the surcharge q on the foundation
from (1) and (2), respectively. level. The incremental external work for the different external
forces can be easily obtained; the calculations are presented in
Failure Mechanisms Appendix I.
Energy is dissipated at the discontinuity surfaces di (i = 1,
M1 Mechanism
. . . , n) between the material at rest and the material in motion
The M1 mechanism is shown in Fig. 1. This mechanism is and at the discontinuity surfaces li (i = 1, . . . , n) within the
symmetrical, and it permits the calculation of the bearing ca- radial shear zone. The incremental energy dissipation per unit

FIG. 1. Failure Mechanism M1 for Static Bearing Capacity Analysis

FIG. 2. (a) Velocity Field of M1 Mechanism; (b) Velocity Hodographs

2
 FIG. 3. Free-Body Diagram for M1 Failure Mechanism

length along a velocity discontinuity or a narrow transition

zone can be expressed as
⌬DL = c⌬V cos ␾ (3)
where ⌬V = incremental displacement or velocity that makes
an angle ␾ with the velocity discontinuity according to the
associated flow rule of perfect plasticity; and c = cohesion
parameter. Calculations of the incremental internal energy dis-
sipation along the different velocity discontinuities are given
in Appendix I.
Equating the total rate at which work is done by the force
on the foundation, the soil weight in motion, and the surcharge FIG. 4. Failure Mechanism M2 for Seismic Bearing Capacity
Analysis
loading [(26) in Appendix I] to the total rate of energy dissi-
pation along the lines of velocity discontinuities [(33) in Ap-
pendix I], it is found, after some simplifications, that an upper
bound on the bearing capacity of the soil is
PS B0
qcS = =␥ N␥S (␪, ␣i , ␤i) ⫹ qNq S (␪, ␣i , ␤i) ⫹ cNcS (␪, ␣i , ␤i)
B0 2
(4)
in which the static bearing capacity factors N␥S (␪, ␣i , ␤i),
NqS (␪, ␣i , ␤i), and Nc S (␪, ␣i , ␤i) can be expressed in terms of
the (2n ⫹ 1) as yet unspecified angles (␪, ␣i , ␤i). They are
given as follows:
N␥S = ⫺( f1 ⫹ f2) (5)
Nq S = ⫺f3 (6)
NcS = 2( f4 ⫹ f5 ⫹ f6) (7)
The ultimate static bearing capacity of the foundation is ob-
tained by minimization of qcS [(4)] with regard to the mech-
anism’s parameters. However, in practice, the minimum values
of the three factors N␥S (␪, ␣i , ␤i), NqS (␪, ␣i , ␤i), and NcS (␪, ␣i,
␤i) are determined independently of each other, and therefore
their use errs on the safe side (see Static Bearing Capacity
Factors in the fourth section).
FIG. 5. (a) Velocity Field of M2 Mechanism; (b) Velocity Hodo-
graph
M2 Mechanism
The M2 mechanism is shown in Fig. 4. This mechanism is level and an eccentricity for the vertical foundation load. The
nonsymmetrical, and it permits the calculation of the bearing moment due to the seismic load on the structure is not con-
capacity in the presence of seismic loading. As is well known, sidered. Only the base shear load will be taken into account.
an earthquake has two possible effects on a soil-foundation Except for the triangular area directly below the base of the
system. One is to increase the driving forces, and the other is footing, the M2 nonsymmetrical mechanism is similar to the
to decrease the shearing resistance of the soil. In this paper, right-hand side of the M1 mechanism. Wedge ABC is trans-
only the reduction of the bearing capacity due to the increase lating as a rigid body with a downward velocity V1 inclined
in driving forces is investigated under seismic loading condi- at an angle ␾ to the discontinuity line AC (Fig. 5). The foun-
tions. The shear strength of the soil is assumed to remain un- dation is assumed to move with the same velocity as wedge
affected by the seismic loading. On the other hand, the earth- ABC (i.e., V1). The rest of the mechanism is similar in form
quake acceleration for both the soil and the structure is to the right-hand side of the M1 mechanism.
assumed to be the same: Only the horizontal seismic coeffi- As shown in Fig. 6, the external forces contributing to the
cient Kh is considered, the vertical seismic coefficient often incremental external work consist of the force acting on the
being disregarded. Finally, the earthquake load on the structure footing, the weight of soil in motion, the surcharge loading,
is represented by the base shear load acting at the foundation and the different inertia forces. These inertia forces concern

3
 Fig. 4) and ␣i ⫹ ␤i ⱖ ␤i⫹1 [cf. Fig. 5(b)]} for the M2 mech-
anism. The method of minimization used is the general re-
duced gradient method. Additional information on Solver op-
tions and algorithms can be found in the Microsoft Excel
Solver’s help file and at the website www.frontsys.com.
In the following sections, we present and discuss in succes-
sion (1) the static bearing capacity factors N␥S , NqS , and NcS
given by both the M1 and M2 mechanisms; and (2) the seismic
bearing capacity factors N␥E , NqE , and NcE given by the M2
nonsymmetrical mechanism.

FIG. 6. Free-Body Diagram for M2 Failure Mechanism Static Bearing Capacity Factors

the base shear load, the inertia forces of the soil in motion, First, the results given by the M1 symmetrical mechanism
and the surcharge loading. Energy is dissipated along the lines will be presented and compared to those given by other ex-
li (i = 1, . . . , n ⫺ 1) and di (i = 1, . . . , n). Calculations of isting solutions. Second, the results of the M2 nonsymmetrical
the incremental external work and the internal energy dissi- mechanism for Kh = 0 will be presented and compared to those
pation along the different velocity discontinuities are given in given by the M1 symmetrical mechanism. This permits us to
Appendix II. estimate the difference between results when considering a
Equating the total external rate of work [(44) in Appendix nonsymmetrical mechanism for a centrally loaded footing.
II] to the total internal rate of energy dissipation [(49) in Ap- Table 1 presents the N␥S factor obtained from the M1 mech-
pendix II], it is found that the value of the upper bound on anism for ␾ = 45⬚ and for various values of n (the number of
the bearing capacity is the triangular rigid blocks). It can be observed that the upper-
bound solution can be improved by increasing the number of
PE B0 rigid blocks. The reduction in the N␥S value decreases with the
qcE = =␥ N␥E (␣i , ␤i) ⫹ qNq E (␣i , ␤i) ⫹ cNcE (␣i , ␤i) (8) n-increase and attains 0.2% for n = 14. It should be mentioned
B0 2
that the same trend is also observed for the NqS and NcS factors.
in which the seismic bearing capacity factors N␥E (␣i , ␤i), Fig. 7 shows the critical slip surface obtained from the nu-
NqE (␣i , ␤i), and NcE (␣i , ␤i) can be expressed in terms of the merical minimization of the N␥S factor for ␾ = 45⬚ and for n
(2n) as yet unspecified angles (␣i , ␤i). They are given as fol- = 12. It can be observed that the critical failure mechanism
lows: obtained by the computer program is composed of two radial
shear zones sandwiched between an active triangular wedge
1
N␥E = ⫺ (g1 ⫹ Kh g2) (9) under the footing and a Rankine passive wedge. It should be
sin(␤1 ⫺ ␾) ⫹ Kh cos(␤1 ⫺ ␾) noted that the radial shear zones are not bounded by log-spiral
1 slip surfaces as is the case of the Prandtl mechanism. Finally,
Nq E = ⫺ (g3 ⫹ Kh g4) (10) note that all subsequent calculations are made for n = 14.
sin(␤1 ⫺ ␾) ⫹ Kh cos(␤1 ⫺ ␾)
Table 2 presents the N␥S , NqS , and Nc S factors obtained from
1 the computer program for ␾ ranging from 0 to 50⬚.
NcE = (g5 ⫹ g6) (11) To check the effect of the superposition method, one cal-
sin(␤1 ⫺ ␾) ⫹ Kh cos(␤1 ⫺ ␾)
culates the ultimate load Pdirect obtained by direct numerical
From these equations, it is clear that only the N␥E factor in- minimization of PS [(4)] and compares it to the one obtained
cludes the soil inertia. The NqE factor includes the inertia of by the superposition method Psuperposition using the N␥S , NqS , and
the foundation load and the surcharge loading; however, the NcS factors. For ␾ = 30⬚, c = 10 kPa, q = 10 kPa, B0 = 1 m,
NcE factor only includes the inertia of the foundation load and and ␥ = 18 kN/m3, one obtains Pdirect = 726.13 kN/m and
thus corresponds to the case of a footing subject to an inclined
load. TABLE 1. N␥S Value versus Number of Rigid Blocks n for ␾ =
As in the static case, the minimum value of qc E gives the 45ⴗ from M1 Symmetrical Mechanism
ultimate seismic bearing capacity of the foundation. However,
Reduction
in practice, the minimum values of the three factors N␥E (␣i , n N␥S (%)
␤i), NqE (␣i , ␤i), and NcE (␣i , ␤i) are determined independently (1) (2) (3)
of each other, and therefore their use errs on the safe side.
2 741.93 —
3 447.94 39.6
NUMERICAL RESULTS 4 384.28 14.2
5 359.50 6.4
The most critical bearing capacity factors can be obtained 6 347.19 3.4
by minimization of these factors [(5) – (7) and (9) – (11)] with 7 340.16 2.0
regard to the mechanism’s parameters. The minimization pro- 8 335.76 1.3
cedure can be performed using the optimization tool available 9 332.82 0.9
in most spreadsheet software packages. In this paper, one uses 10 330.77 0.6
the Solver optimization tool of Microsoft Excel. Two computer 11 329.27 0.5
12 328.14 0.3
programs using the Visual Basic programming language that 13 327.27 0.3
resides in Microsoft Excel have been written to define the 14 326.59 0.2
static and seismic bearing capacity factors as function of the
mechanism’s parameters [(5) – (7) and (9) – (11)]. Initial values
need to be assigned to the different angular parameters. The
solver tool is then invoked to ‘‘minimize’’ the bearing capacity
factor ‘‘by changing’’ the angular parameters, ‘‘subject to’’ the
constraints {␪ ⫹ 兺ni=1 ␣i = ␲ (cf. Fig. 1) and ␣i ⫹ ␤i ⱖ ␤i⫹1
[cf. Fig. 2(b)]} for the M1 mechanism and {兺ni=1 ␣i = ␲ (cf. FIG. 7. Critical Slip Surface for ␾ = 45ⴗ and n = 12

4
TABLE 2. N␥S , Nq S , and Nc S Values from M1 Symmetrical Mech- The values given by Caquot and Kérisel and the expression
anism suggested by Vesic are being increasingly used. Table 3
␾ N␥S Nq S Nc S and Fig. 8 show the comparison with the aforementioned au-
(1) (2) (3) (4) thors. The maximal difference between the present solution
0 — 1.00 5.15
and that of Caquot and Kérisel is smaller than 10% for ␾ ⱕ
1 — 1.09 5.38 45⬚.
2 — 1.20 5.64 On the other hand, rigorous upper-bound solutions for an
3 — 1.31 5.91 associated flow rule Coulomb material are proposed in the
4 — 1.43 6.19 literature. Chen (1975) considered three symmetrical failure
5 — 1.57 6.50 mechanisms referred to as Prandtl1, Prandtl2, and Prandtl3 and
6 — 1.72 6.82
7 — 1.88 7.17
gave rigorous upper-bound solutions in the framework of the
8 — 2.06 7.54 limit analysis theory. Prandtl1 is composed of a triangular ac-
9 — 2.26 7.93 tive wedge under the footing, two radial log-spiral shear zones
10 — 2.47 8.36 and two triangular passive wedges. Prandtl2 differs from
11 — 2.71 8.81 Prandtl1 only in that an additional rigid body zone has been
12 — 2.98 9.30 introduced. Finally, Prandtl3 resembles closely the Prandtl1
13 — 3.27 9.82
14 1.62 3.59 10.39
mechanism; however, each shear zone is now bounded by a
15 1.95 3.95 10.99 circular arc. The upper-bound solutions given by the present
16 2.32 4.34 11.65 M1 mechanism and those given by the three aforementioned
17 2.75 4.78 12.36 mechanisms proposed by Chen are presented in Table 4. It is
18 3.25 5.27 13.13 clear that the present upper-bound solutions are better than
19 3.82 5.81 13.96
20 4.49 6.41 14.86
21 5.26 7.08 15.85 TABLE 3. Comparison of Present N␥S Factor with that of Other
22 6.15 7.84 16.92 Authors
23 7.19 8.68 18.09
24 8.40 9.62 19.37 Present Caquot and Meyerhof Vesic
25 9.81 10.69 20.77 ␾ solution (M1) Kérisel (1953) (1963) (1973)
26 11.46 11.88 22.32 (1) (2) (3) (4) (5)
27 13.39 13.23 24.01 20 4.49 4.97 2.87 5.39
28 15.67 14.76 25.88 25 9.81 10.4 6.77 10.88
29 18.35 16.49 27.95 30 27.51 21.8 15.67 22.4
30 21.51 18.46 30.24 35 49.0 48.0 37.15 48.03
31 25.26 20.70 32.79 40 119.84 113.0 93.69 109.41
32 29.71 23.26 35.62 45 326.59 297.0 262.74 271.76
33 35.02 26.19 38.79
34 41.37 29.56 42.34
35 49.00 33.44 46.33
36 58.21 37.93 50.82
37 69.35 43.13 55.91
38 82.91 49.19 61.68
39 99.48 56.27 68.25
40 119.84 64.58 75.77
41 144.99 74.36 84.40
42 176.23 85.95 94.35
43 215.27 99.73 105.87
44 264.39 116.20 119.29
45 326.59 135.99 134.99
46 405.97 159.91 153.46
47 508.04 189.00 175.31
48 640.42 224.59 201.32
49 813.64 268.44 232.49
50 1,042.48 322.88 270.09

Psuperposition = 680.58 kN/m, which indicates that the superpo-

sition effect errs on the safe side.

Comparison of Results with Existing Solutions

FIG. 8. Comparison of Present N␥S Factor with Results of
N␥S Factor. As is well known, there are a great many Other Authors
solutions for N␥S in the literature based on different methods
and the differences among them are sometimes substantial.
TABLE 4. Comparison of Present N␥S Factor with Other Upper-
Because of the great sensibility of the N␥S factor to the friction Bound Solutions
angle, particularly for ␾ > 30⬚, the tendency today, in practice,
is to use the values given by Caquot and Kérisel (1953), Mey- Chen (1975)
Present
erhof (1963) [cf. (12)], and Vesic (1973) [cf. (13)] ␾ solution (M1) Prandtl1 Prandtl2 Prandtl3
N␥S (Meyerhof ) = (Nq S ⫺ 1)tan 1.4␾ (12) (1) (2) (3) (4) (5)
15 1.9 2.7 2.3 2.1
N␥S (Vesic) = 2(Nq S ⫹ 1)tan ␾ (13) 20 4.5 5.9 5.2 4.6
25 9.8 12.4 11.4 10.9
where NqS is given as follows:

冉 冊
30 21.5 26.7 25.0 31.5
␲ ␾ 35 49.0 60.2 57.0 138.0
Nq S = exp(␲ tan ␾)tan2 ⫹ (14) 40 119.8 147.0 141.0 1,803.0
4 2

5
those of Chen (1975); the improvement attains 15% for ␾ =
40⬚.
NqS and Nc S Factors. Based on the upper-bound method
of the limit analysis theory, the bearing capacity factors ob-
tained from Prandtl mechanism are given by (14) for NqS and
by (15) for Nc S [cf. Chen (1975)]

NcS = (Nq S ⫺ 1)cot ␾ (15)

It should be mentioned here that Chen (1975) has also

shown that the bearing capacity factors NcS and NqS as
given by (14) and (15) are also lower bounds and hence are
the exact solutions in the framework of the limit analysis the-
ory.
The comparison of the present factors with those given by
(14) and (15) has shown that the present solutions are very
close to the exact solutions; the error does not exceed 1.2%.

FIG. 9. Comparison of Bearing Capacity Factors given by M1

TABLE 5. N␥S , Nq S , Nc S Values from M2 Nonsymmetrical Mech-
and M2 Mechanisms
anism

␾ N␥S Nq S Nc S
(1) (2) (3) (4) Comparison of Results with Solutions of M2 Mechanism
0 — 1.00 5.15
1 0.03 1.09 5.38
The N␥S , NqS , and NcS factors given by the M2 nonsym-
2 0.07 1.20 5.64 metrical mechanism for Kh = 0 are presented in Table 5 and
3 0.11 1.31 5.91 compared to those given by the M1 mechanism in Fig. 9.
4 0.17 1.43 6.19 While the NqS and Nc S factors are practically identical in both
5 0.25 1.57 6.50 mechanisms, the M2 mechanism gives greater upper-bound
6 0.33 1.72 6.82 solutions than the M1 mechanism for the N␥S factor. Notice,
7 0.43 1.88 7.17
8 0.55 2.06 7.54
however, that the maximal difference does not exceed 4% for
9 0.69 2.26 7.93 ␾ ⱖ 20⬚.
10 0.85 2.47 8.36
11 1.03 2.71 8.81
12 1.24 2.98 9.30 Seismic Bearing Capacity Factors
13 1.49 3.27 9.82
14 1.77 3.59 10.39 Earthquakes have the unfavorable effect of decreasing the
15 2.10 3.95 11.00 bearing capacity of foundations. To investigate how the bear-
16 2.47 4.34 11.65 ing capacity factors N␥E , NqE , and NcE are affected, extensive
17 2.91 4.78 12.36
numerical results based on the M2 failure mechanism are pre-
18 3.41 5.26 13.13
19 3.99 5.81 13.96 sented in Tables 6 – 8. All results are given for n = 14, which
20 4.67 6.41 14.87 means that the minimization procedure is made with regard to
21 5.45 7.08 15.85 28 angular parameters.
22 6.35 7.84 16.92 Fig. 10 shows the critical slip surfaces obtained from the
23 7.40 8.68 18.09 numerical minimization of the N␥E factor for ␾ = 30⬚ and for
24 8.63 9.62 19.38
three values of Kh (Kh = 0, 0.15, and 0.3). It can be observed
25 10.06 10.69 20.78
26 11.73 11.88 22.32 that the critical slip surface becomes shallower as the accel-
27 13.68 13.23 24.02 eration intensity increases.
28 15.98 14.76 25.89 Charts relating bearing capacity factors N␥E , NqE , and NcE to
29 18.69 16.49 27.96 various governing parameters are presented in Figs. 11 – 13.
30 21.88 18.46 30.25
31 25.67 20.70 32.80
32 30.16 23.25 35.63 TABLE 6. Seismic Bearing Capacity Factor N␥E
33 35.52 26.18 38.80
34 41.93 29.55 42.36 ␾
35 49.62 33.43 46.35
Kh 15 20 25 30 35 40 45
36 58.90 37.91 50.84
(1) (2) (3) (4) (5) (6) (7) (8)
37 70.13 43.11 55.93
38 83.78 49.17 61.71 0 2.10 4.67 10.06 21.88 49.62 120.96 328.88
39 100.47 56.24 68.28 0.05 1.51 3.57 7.91 17.43 39.69 96.48 259.93
40 120.96 64.55 75.80 0.1 1.01 2.61 6.04 13.59 31.23 75.92 203.11
41 146.27 74.33 84.44 0.15 0.58 1.80 4.45 10.35 24.14 58.94 156.98
42 177.70 85.91 94.40 0.2 0.26 1.13 3.14 7.67 18.32 45.12 120.05
43 216.97 99.68 105.94 0.25 0.04 0.62 2.09 5.51 13.61 34.05 90.85
44 266.35 116.13 119.37 0.3 — 0.26 1.28 3.80 9.89 25.31 68.05
45 328.88 135.91 135.09 0.35 — — 0.69 2.49 6.99 18.51 50.44
46 408.65 159.81 153.58 0.4 — — 0.28 1.51 4.77 13.29 36.99
47 511.22 188.86 175.45 0.45 — — 0.04 0.81 3.12 9.34 26.82
48 644.19 224.42 201.50 0.5 — — — 0.35 1.92 6.40 19.19
49 818.14 268.22 232.70 0.55 — — — 0.07 1.08 4.24 13.54
50 1,047.90 322.59 270.36 0.6 — — — — 0.51 2.69 9.38

6
 TABLE 7. Seismic Bearing Capacity Factor Nq E

␾
Kh 15 20 25 30 35 40 45
(1) (2) (3) (4) (5) (6) (7) (8)
0 3.95 6.41 10.69 18.46 33.43 64.55 135.91
0.05 3.52 5.72 9.51 16.35 29.44 56.39 117.46
0.1 3.07 5.02 8.35 14.34 25.70 48.89 100.85
0.15 2.59 4.32 7.24 12.44 22.25 42.08 86.05
0.2 2.07 3.62 6.17 10.67 19.08 35.96 73.00
0.25 1.46 2.94 5.17 9.04 16.22 30.52 61.60
0.3 — 2.25 4.22 7.54 13.65 25.73 51.71
0.35 — 1.46 3.33 6.19 11.37 21.53 43.21
0.4 — — 2.47 4.97 9.36 17.89 35.93
0.45 — — 1.56 3.86 7.59 14.74 29.75
0.5 — — — 2.85 6.04 12.04 24.51
0.55 — — — 1.86 4.69 9.72 20.09
0.6 — — — — 3.49 7.75 16.36

TABLE 8. Seismic Bearing Capacity Factor Nc E FIG. 11. Design Chart for N␥E

␾
Kh 15 20 25 30 35 40 45
(1) (2) (3) (4) (5) (6) (7) (8)
0 11.00 14.87 20.78 30.25 46.35 75.80 135.09
0.05 10.26 13.79 19.14 27.64 41.95 67.84 119.27
0.1 9.50 12.69 17.50 25.09 37.74 60.38 104.74
0.15 8.72 11.60 15.91 22.64 33.76 53.46 91.56
0.2 7.96 10.54 14.37 20.32 30.06 47.12 79.70
0.25 7.21 9.51 12.91 18.14 26.63 41.36 69.14
0.3 6.48 8.53 11.53 16.12 23.50 36.18 59.81
0.35 5.79 7.61 10.25 14.26 20.67 31.56 51.62
0.4 5.14 6.75 9.07 12.58 18.12 27.47 44.49
0.45 4.54 5.96 8.00 11.05 15.85 23.86 38.29
0.5 3.98 5.23 7.02 9.68 13.83 20.69 32.94
0.55 3.47 4.58 6.14 8.46 12.04 17.93 28.33
0.6 3.01 3.98 5.36 7.37 10.48 15.53 24.37

FIG. 12. Design Chart for Nq E

FIG. 13. Design Chart for Nc E

Comparison of Results with Existing Solutions

To see the validity of the present upper-bound solution, the
seismic bearing capacity factors are calculated and compared
with other solutions. The differences between them are dis-
cussed.
Soubra (1997) considered two nonsymmetrical failure
mechanisms and gave rigorous upper-bound solutions in the
FIG. 10. Critical Slip Surfaces for ␾ = 30ⴗ and Kh = 0, 0.15, and 0.3 framework of the limit analysis theory. One mechanism is

7
TABLE 9. Comparison of Present Seismic Bearing Capacity
Factors with Upper-Bound Solutions Given by Soubra (1997) for
␾ = 40ⴗ

N␥E Nq E Nc E
Present Soubra Present Soubra Present Soubra
Kh solution (1997) solution (1997) solution (1997)
(1) (2) (3) (4) (5) (6) (7)
0 121.0 140.5 64.6 64.2 75.8 75.3
0.1 75.9 88.4 48.9 48.7 60.4 60.1
0.2 45.1 53.0 36.0 35.9 47.1 46.9
0.3 25.3 30.1 25.7 25.7 36.2 36.1
0.4 13.3 16.0 17.9 17.9 27.5 27.4
0.5 6.4 7.8 12.0 12.0 20.7 20.7
0.6 2.7 3.3 7.8 7.7 15.5 15.5

FIG. 16. Comparison of Present Nc E Factor with that of Rich-

ards et al. (1993) for ␾ = 30ⴗ

tween the present results and those given by Richards et al.

(1993). From Fig. 14, it can be observed that the solutions
given by Richards et al. (1993) slightly overestimate the N␥E
factor with regard to the present upper-bound solutions. Con-
cerning the NqE factor (cf. Fig. 15), the maximal difference
with Richards et al. (1993) does not exceed 14%. However,
for the NcE factor (cf. Fig. 16), the difference is equal to 11%
for ␾ = 30⬚ and Kh = 0 and attains 40% for Kh = 0.3. This
difference may be explained by the fact that Richards et al.
(1993) have used (15) to calculate the seismic factor NcE with-
out any real justification as they mentioned in their paper.
FIG. 14. Comparison of Present N␥E Factor with that of Rich-
ards et al. (1993) for ␾ = 30ⴗ
CONCLUSIONS

Two failure mechanisms have been considered for the anal-

ysis of the static and seismic bearing capacity factors using
the upper-bound method of the limit analysis theory. The so-
lutions presented are rigorous upper-bound ones in the frame-
work of the limit analysis theory. The numerical results ob-
tained lead to the following conclusions.
For the static case, both the M1 symmetrical and the M2
nonsymmetrical mechanisms give the exact solution of the
static NqS and Nc S factors. For the N␥S factor, the M2 mecha-
nism gives greater upper-bound solutions than the M1 mech-
anism. Notice, however, that the maximal difference does not
exceed 4% for ␾ ⱖ 20⬚. The present upper-bound solutions
are better than those of Chen (1975) since one obtains smaller
upper-bound solutions; the improvement attains 15% for ␾ =
40⬚. On the other hand, the comparison between the present
solutions and the currently accepted values of Caquot and Kér-
FIG. 15. Comparison of Present Nq E Factor with that of Rich- isel has shown that the maximal difference between the results
ards et al. (1993) for ␾ = 30ⴗ is smaller than 10% for ␾ ⱕ 45⬚.
For the seismic case, the present upper-bound solutions
composed of a triangular active wedge under the footing, one given by the M2 nonsymmetrical mechanism are better than
radial log-spiral shear zone, and one triangular passive wedge. those of Soubra (1997) for the N␥E factor, the improvement
The other closely resembles the previous mechanism; how- exceeds 15% when ␾ = 40⬚ and Kh = 0.3. However, the NqE
ever, the shear zone is now bounded by a circular arc. In the and NcE factors are practically identical to those given by
spirit of the upper-bound approach, the lesser of these two Soubra (1997). On the other hand, the comparison with the
solutions was given in the form of design charts [cf. Soubra solutions given by Richards et al. (1993) using the limit-equi-
(1997)]. The upper-bound solutions given by the present M2 librium method has shown that the present upper-bound so-
mechanism and those given by Soubra (1997) are presented lutions are in agreement with regard to the results of Richards
in Table 9. It is clear that the present upper-bound solutions et al. for the N␥E and NqE factors. However, for the NcE factor,
are better than those of Soubra (1997) for the N␥E factor; the the difference between the present solution and that of Rich-
improvement exceeds 15% for Kh = 0.3. However, the NqE and ards et al. attains 40% for ␾ = 30⬚ and Kh = 0.3. The present
NcE factors are practically identical. numerical results are presented in the form of design charts
On the other hand, Figs. 14 – 16 show the comparison be- for practical use in geotechnical engineering.

8
APPENDIX I. M1 Mechanism Incremental Internal Energy Dissipation
In this appendix, we present the different expressions for 1. Along BC
the incremental external work of mechanism M1, together with
the internal energy dissipation from the same mechanism. ⌬DBC = cB0 f4 (␣i , ␤i , ␪)V0 (27)
where
Geometry
cos ␾ cos(␤1 ⫺ ␪ ⫺ ␾)
For the triangular block i, the lengths li and di , and the f4 = (28)
2 cos ␪ sin(␤1 ⫺ 2␾)
surface Si are given as follows:

写
i⫺1 2. Along lines di (i = 1, . . . , n)
B0 sin ␤j
li = (16)
2 cos ␪ j=1 sin(␣j ⫹ ␤j) ⌬Ddi (i=1,...,n) = cB0 f5 (␣i , ␤i , ␪)V0 (29)

写
i⫺1
B0 sin ␣i sin ␤j where
di = (17)
2 cos ␪ sin(␣i ⫹ ␤i) j=1 sin(␣j ⫹ ␤j) cos(␪ ⫺ ␾)cos ␾

写
f5 =
2 cos ␪ sin(␤1 ⫺ 2␾)
i⫺1
B 2
0 sin ␣i sin ␤i sin ␤j
2

冘冋 写 册
Si = (18)
2 4 cos2␪ sin(␣i ⫹ ␤i) j=1 sin2(␣j ⫹ ␤j) n
sin ␣i
i⫺1
sin ␤j sin(␣j ⫹ ␤j ⫺ 2␾)
⭈
i=1 sin(␣i ⫹ ␤i) j=1 sin(␣j ⫹ ␤j)sin(␤j⫹1 ⫺ 2␾) (30)
Incremental External Work
3. Along the radial lines li (i = 2, . . . , n)
The different elements of the incremental external work for
the M1 mechanism can be calculated as follows. ⌬Dli (i=2,...,n) = cB0 f6 (␣i , ␤i , ␪)V0 (31)
1. Incremental external work due to self-weight of triangle
ABC1 where

␥B20 cos(␪ ⫺ ␾)cos ␾

⌬WABC1 = [ f1(␣i , ␤i , ␪)]V0 (19) f6 =
2 cos ␪ sin(␤1 ⫺ 2␾)
2

冘冋 写
n i⫺1
where sin(␤i⫺1 ⫺ ␤i ⫹ ␣i⫺1) sin ␤j
⭈
sin(␤i ⫺ 2␾) sin(␣j ⫹ ␤j)
tan ␪ i=2 j=1

写 册
f1 = (20) i⫺2
2 sin(␣j ⫹ ␤j ⫺ 2␾)
⭈
sin(␤j⫹1 ⫺ 2␾) (32)
2. Incremental external work due to self-weights of the re- j=1

maining 2n triangular blocks The total incremental energy dissipation is twice the summa-

冘 tion of these three parts, that is, (27), (29), and (31)

冘
2n
␥B 2

[ f2 (␣i , ␤i , ␪)]V0
0
[⌬W ] j = (21)
j=1 2 [⌬D] = 2(⌬DBC ⫹ ⌬Ddi (i=1,...,n) ⫹ ⌬Dli (i=2,...,n) ) (33)
where
APPENDIX II. M2 Mechanism
cos(␪ ⫺ ␾) In this appendix, we present the different expressions for
f2 =
2 cos2␪ sin(␤1 ⫺ 2␾) the incremental external work of mechanism M2, together with

冘冋 冉 冘 冊
n i⫺1 the internal energy dissipation for the same mechanism.
sin ␣i sin ␤i
⭈ sin ␤i ⫺ ␪ ⫺ ␣j ⫺ ␾
sin(␣i ⫹ ␤i)
i=1 j=1
Geometry

写 册
i⫺1
sin2␤j sin(␣j ⫹ ␤j ⫺ 2␾) For the triangular block i, the lengths li and di , and the
⭈
j=1 sin2(␣j ⫹ ␤j)sin(␤j⫹1 ⫺ 2␾) (22) surface Si are given as follows:

写
i
3. Incremental external work due to the foundation load sin ␤1 sin ␤j
li = B0 (34)
⌬WPS = PSV0 (23) sin(␣1 ⫹ ␤1) j=2 sin(␣j ⫹ ␤j)

写
i
4. Incremental external work due to the surcharge loading sin ␤1 sin ␣i sin ␤j
di = B0 (35)
sin(␣1 ⫹ ␤1) sin ␤i j=2 sin(␣j ⫹ ␤j)
⌬Wq = qB0 f3 (␣i , ␤i , ␪)V0
写
(24) i
B02 sin2␤1 sin ␣i sin(␣i ⫹ ␤i) sin2␤j
where Si = (36)
2 sin (␣1 ⫹ ␤1)
2
sin ␤i sin (␣j ⫹ ␤j)
2

冉 冘 冊
j=2
n ⫺1
cos(␪ ⫺ ␾) sin ␤n
f3 = sin ␤n ⫺ ␪ ⫺ ␣j ⫺ ␾ Incremental External Work
cos ␪ sin(␤1 ⫺ 2␾) sin(␣n ⫹ ␤n) j=1

写 The different elements of the incremental external work for

n ⫺1
sin ␤j sin(␣j ⫹ ␤j ⫺ 2␾)
⭈ the M2 mechanism can be calculated as follows.
j=1 sin(␣j ⫹ ␤j)sin(␤j⫹1 ⫺ 2␾) (25) 1. Incremental external work due to self-weights and inertia
The total incremental external work is the summation of forces of the n triangular rigid blocks
these four contributions, that is, (19), (21), (23), and (24) ␥B20
⌬Wsoil = [g1(␣i , ␤i) ⫹ Kh g2 (␣i , ␤i)]V1
冘 冘
2n
(37)
2
[⌬W ]ext = ⌬WABC1 ⫹ [⌬W ] j ⫹ ⌬WPS ⫹ ⌬Wq (26)
j=1 where

9
 冘冋 冉 冘冊
n i⫺1
sin2␤1 sin ␣i sin(␣i ⫹ ␤i) Chen, W. F. (1975). Limit analysis and soil plasticity. Elsevier Scientific
g1 = sin ␤i ⫺ ␾ ⫺ ␣j Publishing Company, London, 637.
sin2(␣1 ⫹ ␤1) i=1 sin ␤i j=1 Davis, E. H. (1968). ‘‘Theories of plasticity and the failure of soil

写 写 册
i i⫺1 masses.’’ Soil mechanics: Selected topics, I. K. Lee, ed., Butterworth’s,
sin ␤j
2
sin(␣j ⫹ ␤j ⫺ 2␾) London, 341 – 380.
⭈
j=2 sin2(␣j ⫹ ␤j) j=1 sin(␤j⫹1 ⫺ 2␾) (38) De Buhan, P., and Salençon, J. (1993). ‘‘A comprehensive stability anal-

冘冋 冉 冘冊
n i⫺1 ysis of soil nailed structures.’’ Eur. J. Mech. Ser. A/Solids, Paris, 12(3),
sin2␤1 sin ␣i sin(␣i ⫹ ␤i) 325 – 345.
g2 = cos ␤i ⫺ ␾ ⫺ ␣j Dormieux, L., and Pecker, A. (1995). ‘‘Seismic bearing capacity of foun-
sin (␣1 ⫹ ␤1)
2
i=1 sin ␤i j=1
dations on cohesionless soil.’’ J. Geotech. Engrg., ASCE, 121(3),

写 写 册
i i⫺1 300 – 303.
sin2␤j sin(␣j ⫹ ␤j ⫺ 2␾)
⭈ Drescher, A., and Detournay, E. (1993). ‘‘Limit load in translational fail-
j=2 sin2(␣j ⫹ ␤j) j=1 sin(␤j⫹1 ⫺ 2␾) (39) ure mechanisms for associative and non-associative materials.’’ Géo-
technique, The Institution of Civil Engineers, London, 43(3), 443 – 456.
2. Incremental external work due to the foundation load and Meyerhof, G. G. (1951). ‘‘The ultimate bearing capacity of foundations.’’
the corresponding inertia force Géotechnique, The Institution of Civil Engineers, London, 2, 301 – 332.
Meyerhof, G. G. (1963). ‘‘Some recent research on the bearing capacity
⌬WPE = PE [sin(␤1 ⫺ ␾) ⫹ Kh cos(␤1 ⫺ ␾)]V1 (40) of foundations.’’ Can. Geotech. J., Ottawa, 1(1), 16 – 26.
Michalowski, R. L. (1989). ‘‘Three-dimensional analysis of locally loaded
3. Incremental external work due to the surcharge loading slopes.’’ Géotechnique, The Institution of Civil Engineers, London,
and the corresponding inertia force 39(1), 27 – 38.
Michalowski, R. L., and Shi, L. (1995). ‘‘Bearing capacity of footings
⌬Wq = qB0 [g3 (␣i , ␤i) ⫹ Kh g4 (␣i , ␤i)]V1 (41) over two-layer foundation soils.’’ J. Geotech. Engrg., ASCE, 121(5),
where 421 – 428.

冉 冘冊
Michalowski, R. L., and Shi, L. (1996). ‘‘Closure on ‘Bearing capacity
n ⫺1
sin ␤1 of footings over two-layer foundation soils.’ ’’ J. Geotech. Engrg.,
g3 = sin ␤n ⫺ ␾ ⫺ ␣j 122(8), 701 – 703.
sin(␣1 ⫹ ␤1) j=1 Mroz, Z., and Drescher, A. (1969). ‘‘Limit plasticity approach to some

写 写
n n ⫺1 cases of flow of bulk solids.’’ J. Engrg. Ind. Trans., ASCE, 91, 357 –
sin ␤j sin(␣j ⫹ ␤j ⫺ 2␾)
⭈ 364.
j=2 sin(␣j ⫹ ␤j) j=1 sin(␤j⫹1 ⫺ 2␾) (42) Richards, R., Elms, D. G., and Budhu, M. (1993). ‘‘Seismic bearing ca-

冉 冘冊
n ⫺1 pacity and settlement of foundations.’’ J. Geotech. Engrg., ASCE,
sin ␤1 119(4), 662 – 674.
g4 = cos ␤n ⫺ ␾ ⫺ ␣j
sin(␣1 ⫹ ␤1) j=1
Salençon, J. (1990). ‘‘An introduction to the yield design theory and its
application to soil mechanics.’’ Eur. J. Mech. Ser. A/Solids, Paris, 9(5),

写 写
n n ⫺1
sin ␤j sin(␣j ⫹ ␤j ⫺ 2␾) 477 – 500.
⭈ Sarma, S. K., and Iossifelis, I. S. (1990). ‘‘Seismic bearing capacity fac-
j=2 sin(␣j ⫹ ␤j) j=1 sin(␤j⫹1 ⫺ 2␾) (43) tors of shallow strip footings.’’ Géotechnique, The Institution of Civil
Engineers, London, 40(2), 265 – 273.
The total incremental external work is the summation of these Shinohara, T., Tateishi, T., and Kubo, K. (1960). ‘‘Bearing capacity of

冘
contributions, that is, (37), (40), and (41) sandy soil for eccentric and inclined load and lateral resistance of single
piles embedded in sandy soil.’’ Proc., 2nd World Conf. on Earthquake
[⌬W ]ext = ⌬Wsoil ⫹ ⌬WPE ⫹ ⌬Wq (44) Engrg., Tokyo, Gabujutsu bunken sukyu-rai, Vol. 1, 265 – 280.
Soubra, A.-H. (1997). ‘‘Seismic bearing capacity of shallow strip footings
Incremental Internal Energy Dissipation in seismic conditions.’’ Proc., Instn. Civ. Engrs., Geotech. Engrg., Lon-
don, 125(4), 230 – 241.
1. Along lines di (i = 1, . . . , n) Terzaghi, K. (1943). Theoretical soil mechanics. Wiley, New York, 510.
Vesic, A. S. (1973). ‘‘Analysis of ultimate loads of shallow foundations.’’
⌬Ddi (i=1,...,n) = cB0 g5 (␣i , ␤i)V1 (45) Proc., ASCE, 99(1), 45 – 73.
where
APPENDIX IV. NOTATION

冘冋 写
n i
sin ␤1 cos ␾ sin ␣i sin ␤j The following symbols are used in this paper:
g5 =
sin(␣1 ⫹ ␤1) i=1 sin ␤i j=2 sin(␣j ⫹ ␤j)

写 册
i⫺1
B0 = width of footing;
sin(␣j ⫹ ␤j ⫺ 2␾) c = cohesion;
⭈ c* = residual cohesion due to nonassociativeness;
j=1 sin(␤j⫹1 ⫺ 2␾) (46)
di , li = discontinuity lines;
2. Along the radial lines li (i = 1, . . . , n ⫺ 1) Kh = horizontal seismic coefficient;
N␥E , Nq E , Nc E = seismic bearing capacity factors;
⌬Dli (i=1,...,n ⫺1) = cB0 g6 (␣i , ␤i)V1 (47) N␥S , Nq S , Nc S = static bearing capacity factors;
where n = number of rigid blocks in failure mechanisms;

冘冋 写
n ⫺1 i
PE = seismic ultimate load;
sin ␤1 cos ␾ sin(␤i ⫺ ␤i⫹1 ⫹ ␣i) sin ␤j PS = static ultimate load;
g6 =
sin(␣1 ⫹ ␤1) i=1 sin(␤i⫹1 ⫺ 2␾) j=2 sin(␣j ⫹ ␤j) q = surcharge loading;

写 册
i⫺1
qcE = seismic bearing capacity of footing;
sin(␣j ⫹ ␤j ⫺ 2␾) qcS = static bearing capacity of footing;
⭈ Si = area of block i;
j=1 sin(␤j⫹1 ⫺ 2␾) (48)
V0 = initial downward velocity of footing for M1
The total incremental energy dissipation is the summation of mechanism;

冘
these two parts, that is, (45) and (47) V1, V2 , . . . , Vn = velocities of blocks 1, 2, . . . , n;
␥ = unit weight of soil;
[⌬D] = ⌬Ddi (i=1,...,n) ⫹ ⌬Dli (i=1,...,n ⫺1) (49) ⌬V = velocity along velocity discontinuity;
␪, ␣i , ␤i = angular parameters of failure mechanisms;
APPENDIX III. REFERENCES ␾ = angle of internal friction of soil;
Caquot, A., and Kérisel, J. (1953). ‘‘Sur le terme de surface dans le calcul ␾* = residual friction angle due to nonassociative-
des fondations en milieu pulvérulent.’’ Proc., 3rd Int. Conf. on Soil ness; and
Mech. and Found. Engrg., ICOSOMES, Zurich, Vol. I, 336 – 337. ⌿ = dilatancy angle.

10