el PAOLUCCI AND PECKER
derived which can be used for practical purposes.
KINEMATIC APPROACH
‘The strip foundation in Fig. 1 is resting on a homogene-
ous Mohr-Coulomb soil, with cohesion c and friction an-
fle ¢. The width of the foundation is B. Under static con-
ditions, the system is subjected to a vertical action V
transmitted by the structure, the lateral overburden g,
and the weight of the soil y=, The seismic excitation is
assumed to be applied statically (pseudo-static approach).
The structure transmits a horizontal action H to the foun-
dation and an overturning moment M=V-e, e being the
load eccentricity. In addition, the effect of the horizontal
components of the body force in the soil (f.) and the
lateral overburden (q,), due to the soil inertia, must be
considered.
In the presence of a significant vertical component of
seismic excitation, the weight of the structure and of the
soil should be multiplied by (1~K,), k, being the vertical
seismic coefficient. For simplicity, this effect has been
neglected herein, although it can be easily taken into ac-
count.
The set of loads acting on the foundation system can
therefore be described by a 7-dimension load vector =
{V, H, M, fy Sy Quy Gy}. The kinematic approach of the
yield design theory (Salengon, 1990) provides an upper
‘bound for Q, i.e. determines the region of loads that will
certainly cause the failure of the system.
Several kinematic mechanisms were considered, in or-
der to determine the best upper bound approximation for
the limit load. These mechanisms are the extension to the
Mohr-Coulomb material of those developed by Salencon
and Pecker (1995 a, b) for a Tresca material. We will illus-
trate here the most significant ones.
‘The kinematic mechanism shown in Fig. 2, referred to
as Mec and also described in Dormieux and Pecker
(1995), is of the Prandtl type and is defined by the two ge-
ometric parameters a and 4. AJD and DKL are trian-
gular rigid blocks, whereas DJK is a deformable block
bounded by a logarithmic spiral centered in D and de-
scribed by DM=DJ-e"**, The velocity field Uis defined
in each zone as follows:
AID: U is uniform (/U|=U) and perpendicular to
Ds;
—DIK: in a polar reference system with center D, U is
represented by the following expression:
0
=Ue™,
U(r, 8) wo
thus its value is constant along any line DM, M being a
generic point on the log-spiral DJ, and it is perpendicu-
lar to
=DKL: U is uniform (| U1 =
dicular to DK.
‘The kinematic mechanism shown in Fig. 3, referred to
as Mec2, consists of a rigid block IID, bounded by the
foundation on the upper side and by a logarithmic spiral
U-e8-«- %¢ and perpen-
‘Mobe-Couomb sil
ee
Fig. 1. Shallow foundation on a homogeneous Mobr-Coulomb soll
Fig. 2, Kinematic mechanism Mecl, defined by two geometric
rameters a and
Fig. 3. Kinematic mechanism Mec2,
Parameters 2, Py cand yt
defined by four geometric
centered in @ on the lower side, of a deformable block
DJK defined by a logarithmic spiral centered in D, and of
a triangular deformable block DKL. The geometry of
Mec? is determined completely by four parameters: a, ,
4 and . The friction angle atthe soil-foundation inter-
face is 5. The velocity field U is defined in each zone as
follows:
—IJD: the motion consists of a rigid rotation around 2
with angular velocity «2.
—DIK: in a polar reference with center D, U's given by:
0
10 Oy, =a(aDer™—r) ®SEISMIC BEARING CAPACITY 97
—DKL: denoting by Ma generic point along DK and by
‘M.Mza segment parallel to KL, the velocity field is con-
stant along MM: and forms an angle ¢ with the KL.
direction. The absolute value of the velocity at the
‘generic point Mz inside DKL is given by:
U(M)=e(@De*"- ™*—DM,) @
‘The velocity field across the boundary between zones
DIK and DKL is continuous, as can be seen from (2) and
) by setting r=DM,, 6=x—a—m and recalling that
U(M)=UML).
It should be noted that the introduction of the
parameter 2 allows the taking into account of the possi-
ble uplift of the foundation in the presence of strong load
‘eccentricities. Generally, in the case of zero or low eccen-
tricity, the limit values provided by Mec! and Mec? tend
to coincide. The kinematic mechanisms proposed in the
literature (Sarma and lossifelis, 1990; Budhu and Al-
Karni, 1993) are also of the Prandtl type, but do not take
into account the effects of eccentricity. As a rule, Mec2
was found to give the best upper bound results; thus, the
solutions that will be shown in the following have been
obtained using this mechanism.
‘The application of the kinematic approach of the yield
design theory leads to the necessary condition of stabil-
ity:
P=(Q, ¥, U)sP™(¥, U) “
where P* is the power of the external forces and P™ the
maximum power of the resisting forces; Wis a vector con-
taining the geometric parameters of the kinematic
mechanism. The analytical expressions of P™(Q, ¥, U)
and P™(, U) for MecI and Mec2 are given in Appen-
dix L
‘The upper bound of the external load Q was found by
the following procedure:
—the only load variable is the horizontal action H; the
other 6 load components which define @ are assumed
to be given;
—applying the necessary condition (4), the following ine-
can be derived (see (A.4) and (A.9) in Appen-
HOV, €, fof Gn In ) ()
—a minimization algorithm of the simplex type was ap-
plied to find the minimum value of and the corre-
sponding geometric parameters (i.e. « and u for Mec!;
a, u, 2 and f for Mec2).
VALIDATION OF THE METHOD.
Bearing Capacity Factors
‘A first validation of the results of the kinematic ap-
proach was provided by the static approach of yield de-
sign theory. The construction of statically admissible
stress fields gives an indication of the quality of the
kinematic solution. This approach, which has not been
reported herein for simplicity, was followed by Salengon
and Josseron (1994) who have proven that in the case of a
centered, inclined load the kinematic mechanism Mect
provides an upper bound which practically coincide with
the lower bound of the static approach.
‘The results of the present kinematic approach were
then compared with the classic solutions for the ultimate
bearing capacity (q,) of a shallow foundation under verti-
cal loads. As is well known, this is usually calculated
from the superposition formula:
1
=e-NA0)+ Gy Nd@)+5 YBN) o
where:
Ny=e"*-tan? (45°+9/2) 6)
N=(N,-l cot d (6)
‘The theoretical values of N;(¢) may be calculated by clas-
sical methods of the theory of plasticity (Lundgren and
Mortensen, 1953). Whereas the N, and N. values ob-
tained by the present approach coincide with those given
by (), the Ny values exceed the theoretical ones, up to
~70-80% (Fig. 4). As a reference, the N, values
prescribed by Eurocode 7 (1994) are also shown in Fig. 4.
It should be noted however, that the present N, values
practically coincide with those found by other authors
(Sarma and Tossifelis, 1990; Budhu and Al-Karni, 1993),
‘who used essentially the same Prandtl-type mechanism.
Comparison with Static Finite Element Analyses
‘The results of the present method were compared with
those obtained by numerical analyses carried out using
the nonlinear finite element code DYNAFLOW (Prevost,
1993). A rigid strip foundation, 4 m wide, was consid-
ered, founded on a homogeneous Drucker-Prager soil
with unit weight y=15 KN/m’, The parameters of the
model were chosen so that the Drucker-Prager failure
criterion coincides in plane strain conditions with the
Mohr-Coulomb one (Chen and Liu, 1990), with cohesion
c=80 kPa and friction angle ¢=30°. The soil-founda-
tion interface was modeled by frictional joint elements
with angle J= 15°. The flow rule is associated, to allow a
direct comparison with the results of the kinematic ap-
roach.
‘The velocity fields at failure were calculated using the
30?
101
£7: 2(N—1Ntang
Plasticity ‘Theory
Kinematié approach
108
2 28 30 35 40 45
# (deg)
Fig. 4. Comparison of diferent evaluations of the 1, factor