Professional Documents
Culture Documents
and shape and depth factors could not. Most efforts since then have been devoted to determining
methods to estimate the values of these factors [7, 8, 30, 22, 19, 21, 29].
Key Words and Phrases. Footings, cone penetration test, constitutive modeling, foundation design, sand, Schmert-
manns method, settlement.
2003 ASCE DOI: 10.1061/(ASCE)1532-3641(2002)2:1(1)
ISSN 1532-3641
Downloaded 17 Apr 2009 to 128.211.178.132. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
2 Junhwan Lee and Rodrigo Salgado
Settlements are considered tolerable if they do not impair the functionality or serviceability
of foundations or the supported superstructures under the design loads. Most methods used in
practice to estimate the settlements of footings in sand are based on the linear elastic approach.
Results of this approach are strongly dependent on the reasonable selection of a representative
set of elastic parameters. This is a difcult judgment call that can be made easier if a number of
load test results or other forms of well-documented preexisting information are available. In fact,
well-designed foundations induce stress-strain states in the soil that are neither in the linear elastic
range nor in the range usually associated with perfect plasticity. This requires the consideration
of the nonlinear stress-strain relationship of soils for accurate estimation of settlements under
working loads. Estimated settlement should be no greater than the maximum tolerable settlement
chosen on the basis of type and details of the superstructure, as well as its purpose and architectural
nishings [46, 2].
Footing settlement in sand deposits is often estimated using the results of in situ tests, mainly
the standard penetration test (SPT) and the cone penetration test (CPT). In this approach, the soil
stiffness is estimated from measured penetration resistance in terms of either the SPT blow count
N or the cone resistance q
c
. While the SPT is still widely used, the CPT has become popular for
a number of reasons, including the much lower level of uncertainty associated with q
c
than with
N. There have been several methods proposed for the use of CPT results in footing settlement
calculations [26, 40, 41, 3]. Most of these methods estimate representative soil compressibility
or elastic moduli from cone resistance q
c
. Schmertmanns method is one of the most popular
methods, in part due to its simplicity. It is based on linear elasticity and the observation that
settlement results from strains that initially increase with depth (measured from the base of the
footing), but then peak and drop toward zero. The soil elasticity modulus at different depths for
use in calculations is determined by multiplying q
c
by an empirical factor.
In this article, vertically loaded footings in sand, including both isolated and strip footings,
are analyzed for various soil conditions using the nite element method. The analyses in this
study take into account the nonlinearity of sands, and the effects of footing size and relative
density. The load-settlement responses obtained from these analyses are compared with those
from Schmertmanns method and from eld load tests. Based on these results, a new approach
for estimating the settlement of footings in sand is presented. It is based on the elastic framework
of Schmertmann [40], but allows more realistic accounting of factors such as the settlement level,
footing size and relative density on the estimation of soil stiffness from CPT cone resistance.
II. Review of methods of settlement estimation for footings bearing on sand
A. SPT-based methods
There are several methods available for the calculation of footing settlements using SPT
results. Most of these methods are based on elasticity, and thus focus on determination of soil
compressibility, with consideration of footing size. Meyerhof [26] suggested the following rela-
tionship for the settlement of spread footings on sand:
s =
0.203L
R
N
45
_
q
b
p
A
_
for B 1.2L
R
(1)
s =
0.305L
R
N
45
_
q
b
p
A
__
B
B +L
R
/3.28
_
2
for B > 1.2L
R
(2)
where s =footing settlement; q
b
=unit load at base of footing; N
45
=SPT blow count corrected
for an energy ratio of 45%, following Skempton [42]; B =footing width; L
R
=reference length
= 1 m = 3.28 ft = 39.37 in; p
A
= reference pressure = 100 kPa 1 tsf. In (1) and (2), the SPT
Downloaded 17 Apr 2009 to 128.211.178.132. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
Estimation of Footing Settlement in Sand 3
N
45
values are not corrected for water table, vertical effective stress, or other factors.
Peck and Bazaraa [28] proposed the following relationship, a modication of (2) for esti-
mating the settlements of footings on sand:
s = C
w
C
d
_
0.051L
R
N
B
__
q
b
p
A
__
2B
B +L
R
/3.28
_
2
(3)
N
B
=
_
4N
45
1 +4
v
/p
A
_
v
p
A
for
v
0.75p
A
(4)
N
B
=
_
4N
45
3.25 +4
v
/p
A
_
v
p
A
for
v
> 0.75p
A
(5)
where C
w
=groundwater correction factor; C
d
=depth correction factor; q
b
=unit load at base
of footing;
v
=vertical effective stress; N
45
=measured SPT N values corrected for a 45%SPT
energy ratio; N
B
= stress-normalized SPT N value; B = footing width; L
R
= reference length
= 1 m = 3.28 ft = 39.37 in; p
A
= reference pressure = 100 kPa 1 tsf. Equation (3) is based
on the assumption that settlements predicted by the Terzaghi and Peck [45] correlation produce
excessively conservative results (i. e., excessively large settlements).
Another method for estimating settlements of footings in sand or gravel was proposed by
Burland and Burbidge [9]:
s = f
s
f
I
f
t
I
c
_
q
b
2
3
v,p
__
100
p
A
__
B
L
R
_
0.7
(6)
where s = footing settlement; f
s
= shape factor; f
I
= depth factor for the sand or gravel layer;
f
t
= time factor; q
b
= unit load at footing base;
v,p
= maximum previous vertical stress; B =
footing width; I
c
= compressibility index; L
R
= reference length = 1 m = 3.28 ft = 39.37 in;
p
A
= reference pressure = 100 kPa 1 tsf. Burland and Burbidge [9] presented values of the
compressibility index I
c
as a function of SPT blow count N. The use of
v,p
in (6) allows for the
effect of overconsolidation at the footing base due to the excavation.
B. CPT-based methods
One of the most common methods for the calculation of footing settlement using CPT
results is Schmertmanns method [40, 41]. In this method, the soil prole underneath the footing
is divided into several sublayers. For each sublayer, the soil stiffness is determined based on
the cone resistance q
c
. As shown in Figure 1, the inuence zone for settlement computations
extends down to 2B for square footings and 4B for strip footings. The extent of the inuence
zone and the values of the inuence factor are based on deformation proles beneath footings
obtained from analysis and experiements [40, 41]. Physically, stiffness increases with depth, and
the stresses induced by the applied load decrease with depth; so the contribution of deeper layers
to settlement should be less than that of shallower layers. This is observed to be true, except for
the soil immediately below the footing base. The calculation of footing settlement in sands by
Schmertmanns method is done using the following equations:
s = C
1
C
2
_
q
b
d
_
_
I
z
z
i
E
i
_
(7)
C
1
= 1 0.5
_
d
q
b
d
_
(8)
C
2
= 1 +0.2 log
_
t
0.1 t
R
_
(9)
Downloaded 17 Apr 2009 to 128.211.178.132. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
4 Junhwan Lee and Rodrigo Salgado
FIGURE 1 Inuence factor I
z
vs. depth [41].
where s = settlement caused by applied load; C
1
and C
2
= depth and time factors; q
b
= unit
load on footing base;
d
= vertical effective stress at footing base level; I
z
= depth inuence
factor; z
i
= thickness of each sublayer; E
i
= representative elastic modulus of each sublayer;
t = time; t
R
= reference time = 1 year = 365 days.
In Schmertmanns method, the elastic modulus E
i
of each individual sublayer is obtained
from the representative cone resistance q
ci
for that layer. The correlations between the elastic
modulus E
i
and the cone resistance q
ci
that are most often used are those of Schmertmann et
al. [41] and Robertson and Campanella [33], which may be summarized as
E
i
=
_
_
_
2.5q
ci
for young normally consolidated silica sand
3.5q
ci
for aged normally consolidated silica sand
6.0q
ci
for overconsolidated silica sand .
(10)
A different approach for footing settlement estimation using CPT results can be found in
Berardi et al. [3]. In this approach, footing settlements in sands are calculated from the following
equation from elasticity theory:
s = I
s
q
b
B
E
(11)
where s = footing settlement; I
s
= inuence factor depending on the shape and rigidity of the
foundation; q
b
= unit load on footing base; B = footing width; E
J
2o
_
J
2 max
J
2o
_
g
_
_
I
1
I
1o
_
n
g
(12)
where G = secant shear modulus; G
o
= initial shear modulus; J
2
= second invariant of the
deviatoric stress tensor;
J
2
,
J
2o
and
J
2 max
represent the current, initial, and maximum
shear stress in three dimensions; I
1
and I
1o
are the rst invariants of the stress tensor at the current
and initial states; f , g, and n
g
=material parameters. Equation (12) represents the degradation of
shear modulus from its initial maximum value G
o
as a function of the shear (
J
2
) and conning
stress (I
1
) levels. Lee and Salgado [23, 25] also proposed values of f and g as a function of the
relative density D
R
.
Sand usually behaves as a linear elastic material with shear modulus G
o
for shear strains
up to approximately 10
5
, after which the stress-strain relationship is strongly nonlinear. In this
study, the following empirical equation, based on the work of Hardin and Black [17], was used
to estimate the initial maximum shear modulus of sand:
G
o
= C
g
_
e
g
e
o
_
1 +e
o
p
1n
g
A
_
m
_
n
g
(13)
where C
g
, n
g
, and e
g
= intrinsic material variables; e
o
= initial void ratio; p
A
= reference
pressure = 100 kPa; and
m
= initial mean effective stress in the same unit as p
A
. It should be
noticed that the parameter n
g
is the same as appears in (12): in both equations, n
g
represents the
dependence of the shear modulus on connement.
The elastic stress-strain relationship may be expressed by two constants; the bulk modulus
K and the shear modulus G are often used. The full description of nonlinear elastic response
requires proper representation of the variation of K as well as G with changes in stress state. The
bulk modulus depends mainly on the amount of conning stress [27]. Based on the discussion
of the K-G model by Naylor et al. [27],the tangent bulk modulus K
t
can be represented by the
following equation:
K
t
= D
s
_
m
_
n
k
(p
A
)
(1n
k
)
(14)
where p
A
= reference pressure = 100 kPa;
m
= mean effective stress in the same units as
p
A
; D
s
= material constant that can be calculated from the initial values of bulk modulus and
conning stress; and n
k
can be taken as 0.5.
It has been known that hyperbolic soil models with varying G and K cannot describe ad-
equately the soil behavior near failure [10]. In order to describe failure and post-failure soil
Downloaded 17 Apr 2009 to 128.211.178.132. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
6 Junhwan Lee and Rodrigo Salgado
response, the Drucker-Prager failure criterion was adopted. The DruckerPrager failure criterion
is given by:
F =
_
J
2
(I
1
+) = 0 (15)
where
=
2 sin
p
3
_
3 sin
p
_ (16)
=
6c cos
p
3
_
3 sin
p
_ (17)
where c =cohesion and
p
=peak friction angle. The cohesive intercept c is zero in sands. The
peak friction angle in sands can be expressed in terms of the friction angle at the critical state and
the dilatancy angle [4, 39]. The critical-state friction angle for a given soil is independent of stress
state and density. The dilatancy angle is a function of density and connement, increasing with
increases in density and decreasing with increases in connement. Consequently, the envelope of
the failure surface is nonlinear. Bolton [4] proposed the following equation to estimate the peak
friction angle in sand:
p
=
c
+0.8
p
(18)
where
p
= peak friction angle,
c
= critical state friction angle and
p
= peak dilatancy angle
given by
p
=
_
6.25I
R
for plane strain conditions
3.75I
R
for triaxial conditions
. (19)
The dilatancy index I
R
that appears in (19) is given by
I
R
= I
D
_
Q+ln
_
p
A
100p
P
__
1 (20)
where I
D
=relative density (as a number between 0 and 1); p
A
=reference pressure = 100 kPa;
p
P
= mean effective stress at peak strength in the same units as p
A
; and Q = intrinsic soil
variable, approximately equal to 10 for silica sands. Equations (18) through (20) were used to
dene the nonlinear DruckerPrager failure surface.
B. Numerical modeling of footing load tests
Five footing load tests were performed at the Texas A&M University Riverside campus
for the spread footing prediction symposium organized for the Settlement 94 ASCE Special
Conference [5]. Five square footings were tested: two 3 3 m footings (south and north sides),
one 2.5 2.5 m footing, one 1.5 1.5 m footing, and one 1 1 m footing. The test site consists
predominantly of sand down to a depth of 11 m. Beneath this sand layer, there is a very stiff clay
deposit extending down to a depth of approximately 33 m. The water table was observed at a
depth of 4.9 m from the ground surface. Figure 2 shows the subsoil prole for the test site. Grain
size analysis showed the amount of nes content to vary with depth, from 2 to 8% to 5 to 30%
nes contents down to depths of 3 and 9 m, respectively.
All ve footing load tests were modeled numerically using the nite element method. These
analyses aimed at validating in a general way our numerical analysis by comparison with existing
Downloaded 17 Apr 2009 to 128.211.178.132. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
Estimation of Footing Settlement in Sand 7
FIGURE 2 Sub-soil prole at the Texas A&M footing load test site [16].
settlement observations. The commercial nite element program ABAQUS was used to model
the footings, with a subroutine specically written for the nonlinear elastic-plastic stress-strain
model previously described. In the nite element analyses, the soil layers below the foundation
level were modeled based on the soil prole (shown in Figure 2) determined from extensive site
characterization done before the conference.
The bottom boundaries of the nite element models were located at a depth of 17 m. The
lateral boundaries were located at a distance of 14 m (for the 1-m and 1.5-m footings), and 18 m
(for the 2.5-m and 3-m footings) from the axis of the footings. The footings were modeled as
circular footings with diameters equal to 1.13 m, 1.69 m, 2.80 m, and 3.40 m, with areas equivalent
to those of the 1-m, 1.5-m, 2.5-m, and 3-m square footings, respectively. Differences in stresses
at the same depths due to the use of circular rather than square footings were less than 2%,
based on linear elastic calculations. Eight-noded axisymmetric elements were used in the nite
element meshes to model both the soil and the footings. In addition, interface elements were used
between the footing base and the soil with a Coulomb friction coefcient of 0.7, corresponding
to an interface friction angle of about 35
. Figure 4
shows the nite element meshes for the 1-m, 2-m, and 3-m footings, respectively.
Figure 5 shows results from the nite element analyses in terms of unit load at the footing
base versus normalized settlement. The settlement was normalized with respect to the footing
diameter B. For the normalized load-settlement curves of footings on sand, Briaud and Jeanjean [6]
suggested the use of a unique load-settlement curve, without consideration of the effects of footing
size. The results obtained in this study, however, indicate that the effect of footing size on the
Downloaded 17 Apr 2009 to 128.211.178.132. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
Estimation of Footing Settlement in Sand 11
FIGURE 3 Measured and predicted load-settlement response for (e) 3-m footing (north side).
TABLE 2
Basic Properties of Ticino Sand (after Ghionna et al. [15])
D
10
D
50
max
min
(mm) (mm) G
a
s
U
b
c
e
max
e
min
(KN/m
3
) (KN/m
3
) C
g
n
g
e
g
0.36 0.54 2.623 1.5 34.8