98 PAOLUCCI AND PECKER
two methods in the case of centered and eccentric (e/
B=0.25) inclined loads. The results are shown in Figs.
(a) and 5(b) respectively. Qualitative good agreement of
the predictions, both in terms of shape and dimensions
of the velocity fields, can be noted, In addition, under ec-
centric load, the uplift of the foundation is predicted by
both methods.
Applying the kinematic approach to the previous foun-
dation system, the interaction curve illustrated in Fig. 6
was obtained, representing the combinations of horizon-
tal (H) and vertical (V) loads which cause the failure of
the system. For low values of vertical load, the interac-
tion curve tends to coincide with the line H=Vand,
indicating a sliding failure mechanism. The results
provided by static finite element analyses for several cases
of inclined and eccentric loads are illustrated by dots in,
the same figure. Generally, the limit loads obtained by
finite elements are ~ 10-15% larger than those provided
by the kinematic approach, due to the greater rigidity of
the finite element mesh.
It is well known that for cohesionless soils the hypothe-
sis of an associative flow rule, implicitly required to give
a sense to the kinematic approach (in terms of upper
‘bound solution), gives rise to unrealistically large volu-
metric strains and to the overestimation of the ultimate
load (see e.g. Chen and Liu, 1990). Compared to the ex-
treme case of @ non-associative flow rule with dilation
prevented, our finite element analyses show that the as-
sociative assumption leads to an overestimation of the ul-
timate load by ~40% (Fig. 6), with volumetric strains at
failure approximately two times greater than those ob-
tained in the non-associative case. Obviously, the latter
are of a purely elastic nature.
COMPARISON WITH RECENT EXPERIMENTAL
DATA
Several experimental researches (e.g. Ingra and
Beacher, 1983; Zadroga, 1994) have shown that the N,
values calculated by the classical solutions of the theory
of plasticity are generally much lower (up to ~ 100%)
than those obtained from laboratory models. Due to the
‘uncertainty upon the accurate evaluation of N,, it is
meaningful to present the results in a format normalized
o)
WAN
ANAS
SASS
Fig.
oad, Same as a) but foram eccentric load
H (MP0)
V (MPa)
Fig. 6, Solid line interaction curve derived by the kinematic approach on a Mohr-Coulomb soil with e=80 kPa, 630°, 615°
‘and eccentricity ratio e/ B=0 lett), e/ B= 0.25 (right). Small squares indicate the ultimate bearing capacity ob
Bosdm, no lateral overburde
a) Velocity fields predicted by stale finite element analysis (up) and assumed in the present kinematic approach (down). Case of centred
2.0
18)
tel ¢/8=0.25
14!
° 4 8 12
v (MPa)
ISKN/m?,
{ained by finite element simulations with associated (Glled) and non-assocated flow rues (empty)SEISMIC BEARING CAPACITY 9
by the ultimate bearing capacity (Vau=qu- B) under verti-
cal centered load. This permits the study of the seismic
effects on the bearing capacity merely in terms of reduc-
tion factors with respect to the static case, neglecting the
problem of the accurate evaluation of the NY, factor (or
Vos)
This approach was followed by Butterfield and
Gottardi (1994), who proposed, based upon accurate
laboratory tests, an analytical expression of the three-
dimensional interaction curve for shallow foundations
on sands, subjected to inclined and eccentric loads. As
confirmed also by the experimental work of Nova and
Montrasio (1991), the failure locus is approximately in-
dependent of the soil density
Effect of Load Eccentricity
‘According to Butterfield and Gottardi, the bearing
capacity reduction due to load eccentricity, for the case
of dry sands, is determined by:
e
035-B
V/ Vea. Equation (7a) is illustrated in Fig. 7,
together with the curve provided by the present method,
and the curve of equation
(7a)
co)
which matches the results of the kinematic approach.
The agreement with the experimental curve is good, at
least for eccentricity ratios e/B<0.3.
Effect of Load Inclination
‘The interaction curves in the hv plane (=H Vass)s
as determined by Butterfield and Gottardi (1994) and by
the present method, are compared in Fig. 8, together
with the experimental points derived from the works of
Nova and Montrasio (1991) and Gottardi and Butterfield
(1993). These points lie approximately between the two
curves. The results of the kinematic approach, for a co-
12
Kinematic approech
1.0]
Equation (7a)
Equation (7b)
Vox
o 1 2 3 4 5
eccentricity ratio «/8
ig. 7. Bearing capaclty reduction due to load eccentricity on co-
Iesionlesssolls without lateral overburden
eet cee
‘> Goltordi end Butterfield (7995)
Nowa end Montraso (1981)
Butterfield ond Gottardi (1994)
cquation (8)
Fig. 8. Comparison of the h—pinteraction curves obtained by labora-
‘tory tests and by the present method
hesionless material, are approximated rather accurately
by the equation:
h=0.850(1 V0) ®
It should be noted that (8), apart from the 0.85 factor,
follows directly a well-known formula for the bearing
capacity reduction factor due to load inclination, in the
case of a cohesionless material (see e.g. Vesic, 1975),
namely:
0
EFFECT OF SOIL INERTIA ON BEARING
CAPACITY
‘As already mentioned in the introduction, the previous
studies of the seismic effects on the bearing capacity of
shallow foundations did not clearly identify the effect of
soil inertia, acting alone. The method discussed herein
‘was applied to study the reduction of the bearing capaci-
ty factors due to the application of a horizontal volume
force f= ks fy f, being the weight of the soil and ky a seis-
mic coefficient. The results are presented in terms of the
ratios 2.=Ne/ Ny Z=Nee/ Non 2)=Nye/ Nps Between the
seismic (Ne Ngo, Nye) and the static (Ne, Non Nj») bear-
ing capacity factors (Fig. 9). Whereas z. was found to be
independent of the friction angle, z, and z, can be approx-
imated by the following equation:
1 ky/tan oy" (ao)
‘These results were also confirmed by static finite ele-
ment analyses and clearly indicate that, at least for a
range of reasonable values of the pseudo-static seismic
coefficient (k,<0.3), the bearing capacity reduction due
to the effects of the inertia of the soil is small and does
not exceed ~ 15-20%. It should also be noted that the
kinematic approach predicts the failure of the founda-
tion for k,=tan@, which corresponds to the general
fluidization level according to Richards etal. (1990).