You are on page 1of 13

International Journal of Geotechnical Engineering

ISSN: 1938-6362 (Print) 1939-7879 (Online) Journal homepage: http://www.tandfonline.com/loi/yjge20

Seismic bearing capacity of shallow foundations


on partially liquefiable saturated sand

M. F. Mansour, M. A. Abdel-Motaal & A. M. Ali

To cite this article: M. F. Mansour, M. A. Abdel-Motaal & A. M. Ali (2016) Seismic bearing
capacity of shallow foundations on partially liquefiable saturated sand, International Journal of
Geotechnical Engineering, 10:2, 123-134, DOI: 10.1179/1939787915Y.0000000020

To link to this article: http://dx.doi.org/10.1179/1939787915Y.0000000020

Published online: 04 Sep 2015.

Submit your article to this journal

Article views: 29

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=yjge20

Download by: [Dr Mohamed Mansour] Date: 15 May 2016, At: 05:24
Seismic bearing capacity of shallow
foundations on partially liquefiable saturated
sand
M. F. Mansour*1, M. A. Abdel-Motaal1 and A. M. Ali2
The seismic bearing capacity of partially liquefiable saturated sand is investigated by numerical analysis.
An empirical correlation in the literature assumes a punching shear mechanism, which underestimates
the seismic bearing capacity due to omission of the q-term. In this paper, the effects of the footing width,
soil relative density, foundation depth, and maximum seismic acceleration on the seismic bearing capacity
of partially liquefiable saturated sands are investigated. The results show that the foundation depth has
Downloaded by [Dr Mohamed Mansour] at 05:24 15 May 2016

a remarkable effect on the seismic bearing capacity. Hence, the failure mode is not likely punching. The
seismic bearing capacity of partially liquefiable saturated sands is redefined with new seismic bearing
capacity factors, Nq and Nγ. The paper presents also a guideline to estimate the allowable seismic bearing
capacity considering both the ultimate and serviceability limit states.
Keywords: Seismic bearing capacity, Shallow foundations, Partially liquefiable saturated sand, Punching shear mechanism, Finite element analysis, Post-
earthquake settlement

Introduction The Euro Code states that sands with FSliq lower than 1.25 are
susceptible to liquefaction (Eurocode 8 – Part 5, 1998).
The behavior of loose to medium saturated sands subjected The seismic bearing capacity of shallow foundations resting
to seismic loads is quite complex. These soils suffer from the on partially liquefiable sands did not receive the deserved atten-
generation of excess pore pressures, which causes a reduction tion in the literature. Many researchers investigated the problem
in the soil shear strength. Seed and Idriss (1971) defined the of the seismic bearing capacity of liquefiable sand overlain by
factor of safety against liquefaction (FSliq) as the ratio between a non-liquefiable clay crust. Naesgaard et al. (1998) assumed
the cyclic resistance ratio (CRR) and the cyclic stress ratio that punching shear failure will occur in the clay crust along
(CSR). Very loose and loose saturated sands have FSliq lower with a wedge-type failure within the liquefiable sand layer,
than unity; i.e. they lose their shear strength during seismic and developed an equation to calculate the factor of safety
excitation and completely liquefy. On the other side, dense against bearing capacity failure. Cascone and Bouckovalas
and very dense sands experience low excess pore pressures, (1998) investigated the same problem, and proposed a formula
and the shear strength is not affected by the seismic shaking. to compute the degraded internal friction angle of the liquefia-
Day (2010) considered that saturated sands are completely ble sand due to excess pore pressure generation. Bouckovalas,
safe against liquefaction when FSliq is higher than 2.0. In the Valsamis and Andrianopoulos (2005) suggested a modifica-
intermediate zone of partially liquefiable soils; i.e. FSliq lies tion to the expression of the degraded internal friction angle
between 1.0 and 2.0, the excess pore pressures lead to decreas- by Cascone and Bouckovalas (1998) to include the effect of
ing the effective stresses, and, hence, the shear strength and the residual shear strength of the liquefied sand. Karamitros,
seismic bearing capacity are reduced (Day 2010). Dickenson et Bouckovalas, Chaloulos and Andrianopoulos (2013) conducted
al. (2002) indicated that full liquefaction should be considered an effective stress finite difference analysis to study the bear-
to occur when FSliq is lower than 1.1. They indicated also that ing capacity degradation of liquefiable sand overlain by a clay
for FSliq higher than 1.4, the excess pore pressures are very crust. Karamitros et al. (2013) presented a relationship between
low, and, hence, the sand shear strength should not be reduced. the clay crust thickness and the footing width in order to avoid
The Building Seismic Safety Council (BSSC) recommended the hazard of soil liquefaction.
that FSliq of 1.5 is appropriate for building sites (BSSC 1997). Day (2010) derived an expression to calculate the seismic
bearing capacity of shallow foundations resting directly on
partially liquefiable saturated sand. According to Day (2010),
partially liquefiable sand corresponds to FSliq in the range of
1
Ain Shams University, Egypt
1.0–2.0. Day (2010) assumed that the likely mode of failure is
2
Engineering Consultants Group S. A. (ECG), Egypt

*Corresponding author, email Mohamed.Mansour@eng.asu.edu.eg

© 2016 Taylor & Francis


Received 23 June 2015; accepted 15 August 2015
DOI 10.1179/1939787915Y.0000000020 International Journal of Geotechnical Engineering    2016  VOL. 10  NO. 2 123
Mansour et al. Seismic bearing capacity of shallow foundations

punching shear. Hence, the q-term of the bearing capacity equa- sand layer, which overlies a very dense sand layer having a
tion is omitted, as shown in equation (1) thickness of 15.0 m. The soil profile is illustrated in Fig. 2.
1 The groundwater table is taken at the foundation level. A rela-
qult = B𝛾 N (1 − ru ) (1) tively big thickness of the medium dense sand layer is adopted
2 sub 𝛾
to simulate an extended deposit beneath the foundation level.
where B is the footing width, γsub is the buoyant unit weight Hence, the seismic bearing capacity and the post-earthquake
of the soil beneath the footing, Nγ is the bearing capacity factor, settlement are conservatively estimated.
which was suggested by Day (2010) to be taken equal to the
static value, and ru is the pore pressure ratio (the ratio between
the excess pore pressure and the effective stress). Day (2010) Parametric study
recommended determining the pore pressure ratio from the Four strip footings widths are investigated: 1.0, 3.0, 5.0, and
charts developed by Marcuson and Hynes (1990). 7.0 m. The relative density (Dr) is varied according to NSPT of 10,
In this paper, the validity of Day's assumptions is investigated 15, 20, 25, and 30. The corresponding relative density values
by numerical analysis. The seismic bearing capacity of partially are 35.0, 42.5, 50.0, 57.5, and 65.0%. The effect of the foun-
liquefiable saturated sands without an overlying clay crust is dation depth (Df) is investigated by comparing the behavior of
investigated. The most adverse consequence of partially lique- surface footings (Df = 0.0 m) with that of footings founded at
fiable sands is the value of the permanent settlement after the 1.5 m below the ground surface. Therefore, the primary study
end of an earthquake episode. Several researchers indicated involves 40 analyses. In each analysis, the stress–settlement
that post-earthquake deformations are actually driven by static curve is constructed by varying the applied pressure at the
forces rather than dynamic forces (Finn, Yogendrakumar, Lo and foundation level and calculating the permanent post-earthquake
Yoshida 1988; Law et al. 2005; Kourkoulis, Anastasopoulos, settlement after the end of the seismic shaking.
Downloaded by [Dr Mohamed Mansour] at 05:24 15 May 2016

Gelagoti and Gazetas 2010; US Bureau of Reclamation, 2012).


USBR (2012) pointed out that many observations of embank-
ment instability from seismic loading indicated that most of Methodology
the deformation occurs after the end of the seismic episode. The numerical analyses are conducted in three steps using the
Therefore, the permanent deformation of saturated sands should finite element program GeoStudio 2007. The three steps are
be calculated by redistributing the unbalanced stresses during explained as follows:
the earthquake using a static numerical analysis. 1. The initial effective static stresses due to the soil own
An extensive parametric study is conducted where the ulti- weight and the footing pressure are established by con-
mate seismic bearing capacity of partially liquefiable sand is ducting a two-dimensional plane strain finite element
determined from the stress–settlement curve. Different footings analysis. The lower boundary is fixed in the horizontal
widths, foundation depths, soil density conditions, and input and vertical directions. The side boundaries are restrained
motion intensities are investigated. The conducted analyses are in the horizontal direction, as shown in Fig. 2. The ele-
two-dimensional plane strain analyses. Hence, strip footings ment size around the footing is taken 0.5 m, and the mesh
only are investigated. A non-linear dynamic constitutive model gets coarser towards the lower and side boundaries. For
is adopted to characterize the soil response. The output of the surface footings, i.e. zero foundation depth, the footing
seismic analysis is introduced into a static load-deformation is simulated as a beam element. For 1.5 m-deep footings,
analysis to redistribute the seismic stresses and calculate the the footing is simulated as a solid element.
permanent deformations. 2. A seismic finite element analysis is conducted based on
the output of the static analysis. The input earthquake
motion is applied at the lower boundary of the model
Soil profile (horizontal and vertical fixation). The side boundaries are
The soil under study is partially liquefiable sand, i.e. with a restrained in the vertical direction only to ensure that the
factor of safety against liquefaction ranging from 1.0 to 2.0 earthquake waves are not radiated back into the investi-
according to Day (2010). Preliminary calculations are per- gated domain. Preliminary sensitivity analyses have indi-
formed to determine the range of the standard penetration test cated that a boundary distance of six times the footing
corrected blows count (NSPT) that corresponds to the range of width ensures a minimal effect of the model boundaries
factor of safety of 1.0–2.0. The factor of safety against lique- on the results (see Fig. 2). The finite element mesh has
faction is calculated using the simplified procedure by Seed and the same configuration as the static mesh except for the
Idriss (1971). Three peak ground accelerations of 0.10, 0.15, side boundary conditions.
and 0.20 g are applied at the bottom of a 15 m-deep saturated 3. The seismic shear stresses and excess pore pressures are
sand deposit. At each peak acceleration value, the upper and calculated from the seismic analysis. The results are used
lower bounds of NSPT, which correspond to factors of safety of as input for a static finite element analysis to calculate the
1.95 and 1.05, respectively, are back calculated. The results permanent post-earthquake deformation resulting from
are shown in Fig. 1. The figure shows that partially liquefiable the exceedance of the shear strength by the seismic shear
sands correspond to NSPT range of 10–30, which is the range stresses.
of medium dense sands. Therefore, NSPT of 10, 15, 20, 25, and A stress–settlement relationship is plotted by varying the
30 are chosen for the parametric study. applied stress and calculating the post-earthquake settlement.
The soil profile beneath the footing consists of 10.0 m-thick Typically, four to eight points are used to plot each stress–set-
medium dense sand layer followed by a 15.0 m-thick dense tlement curve.

124 International Journal of Geotechnical Engineering  2016  VOL. 10   NO. 2


Mansour et al. Seismic bearing capacity of shallow foundations

1 Ranges of NSPT values corresponding to peak accelerations of 0.10, 0.15, and 0.20 g
Downloaded by [Dr Mohamed Mansour] at 05:24 15 May 2016

2 Geometrical configuration of the numerical model showing the finite element mesh

Input motion that the peak acceleration response (amax) at the foundation level
equals 0.15 g. The artificial earthquake duration is taken 20 s.
The input motion used to perform the seismic analysis is an Additional numerical simulations are conducted to investigate
artificial earthquake acceleration–time history based on the the effect of changing the earthquake intensity on the ultimate
standard response spectrum of the Uniform Building Code seismic bearing capacity. Mokhtar, Abdel-Motaal and Wahidy
(Abdel-Motaal 1999), as shown in Fig. 3. The acceleration– (2014) concluded that the lateral seismic displacement of piles
time history is applied at the lower boundary of the model. in liquefiable soils is slightly affected when the earthquake
Preliminary analyses are conducted to determine the peak duration increases from 20 to 40 s. Hence, the earthquake dura-
ground acceleration at the lower boundary of the model such tion is taken 20 s in the conducted analyses.

 International Journal of Geotechnical Engineering  2016  VOL. 10   NO. 2 125


Mansour et al. Seismic bearing capacity of shallow foundations

definitions of the excess pore pressure variation and the vol-


umetric changes that occur due to the seismic shaking. These
parameters are discussed in the following sections.

Maximum shear modulus (Gmax )


The empirical formula proposed by Seed and Idriss (1970) is
used, in SI units, to calculate the maximum shear modulus, as
shown in equation (2)
( )0.5
Gmax = 218.79 ∗ K2,max ∗ 𝜎m� (2)
where Gmax is the maximum shear modulus (kPa), K2,max is a unit-
less factor that depends on the value of the relative
) density, and
3 Artificial earthquake acceleration–time history with
𝜎m′ is the mean principal stress 𝜎v� + 2 ∗ K0 𝜎v� ∕3 (kPa) (kPa).
(
duration of 20 s and peak ground acceleration of 0.15 g
Seed and Idriss (1970) provided guide values for the factor
(after Abdel-Motaal 1999)
K2,max, based on the relative density. Seed, Wong, Idriss and
Tokimatsu (1984) reported that the value of K2,max for gravels
ranges from 80 to 180. Table 2 shows the K2,max values used
Table 1 Strength and deformation parameters in the static
analyses
in this study. The values of Gmax are defined in the numerical
model as a function of the effective vertical stress.
Downloaded by [Dr Mohamed Mansour] at 05:24 15 May 2016

Internal friction Elastic deformation


Soil layer angle,φ (degrees) modulus, E (MPa)
Medium 32, 33, 34, 35, and 40.0 Damping and maximum damping ratios
dense sand 36
Dense sand 37 75.0 The damping ratio is calculated from the correlation developed
Very dense 40 150.0 by Ishibashi and Zhang (1993). The correlation accounts for
sand the effects of the mean principal stress, the plasticity index (for
cohesive soils), and the modulus reduction factor (G/Gmax), as
shown in equation (3)
Constitutive models and material
[ )2 ]
1 + exp(−0.0415PI1.3 ) G G
(
𝜉 = 0.333 0.586 − 1.547 +1 (3)
2 Gmax Gmax
properties
This section presents the constitutive models adopted in the where
static and seismic analyses, and the values of the material prop- G ( )m(𝛾,PI)−m0
erties of the different soil types. = K(𝛾, PI) 𝜎m� (3.a)
Gmax

Static conditions { [ ( )0.4 ]}


0.000102 + n(PI)
Mohr–Coulomb failure criterion is used to define the constitu- K(𝛾, PI) = 0.5 1 + tanh In
𝛾 (3.b)
tive behavior under static conditions; i.e. the constitutive law
is elastic-perfectly plastic. The investigated materials have no
cohesion. Poisson's ratio (ν) is taken 0.33 for all layers. Table
1 summarizes the strength and deformation parameters used { [ ( )0.4 ]}
0.000556
in the static analyses. m(𝛾, PI) − mo = 0.272 1 − tanh In
𝛾
exp(−0.0145PI1.3 ) (3.c)

Seismic conditions 0.0 for PI = 0


3.37 ∗ 10−6 PI1.404 for 0 < PI ≤ 15 (3.d)
The non-linear dynamic constitutive model is selected to sim- n(PI) =
7.0 ∗ 10−7 PI1.976 for 15 < PI ≤ 70
ulate the soil behavior in the seismic analysis. The non-linear 2.7 ∗ 10−5 PI1.115 for PI >70
dynamic model computes the excess pore pressures at each time
step during the earthquake shaking, and, hence, can simulate
Table 2 Values of K2,max
the generation of the excess pore pressure and the subsequent
decrease in the effective stress during the earthquake. Soil type NSPT (unitless) K2max (unitless)
The stress–strain relationship in the non-linear dynamic
Medium dense sand 10 31
model can be approximated by a hyperbolic curve (Kramer 15 40
1996; Kramer and Elgamal 2001). The curve is defined by two 20 50
parameters; the slope at zero strain (the maximum shear mod- 25 60
30 69
ulus, Gmax) and the asymptotes at large strains (the maximum Dense sand 30–50 70
shear strength, τ). The non-linear dynamic model requires also Very dense sand >50 80

126 International Journal of Geotechnical Engineering  2016  VOL. 10   NO. 2


Mansour et al. Seismic bearing capacity of shallow foundations

Excess Pore Pressure (Δu) function stress paths of triaxial specimens subjected to undrained mono-
tonic loading, and truncated at the steady-state strength (Css).
Several methods are available in the literature to define the The collapse surface is defined by two parameters; the collapse
excess pore pressure function. The methods are based on stress surface angle and the steady-state strength (Css). Its slope is
path models, volumetric strain models and empirical models taken two-thirds the critical state or the steady-state line slope
(Ishihara, Tatsuoka and Yasuda 1975; Martin, Finn and Seed (Kramer 1996). The steady-state strength (Css) is computed by
1975; Seed, Idriss and Arango 1983; Wu 1996; Wu 2001). The the equation developed by Stark and Mesri (1992), based on
volumetric strain model by Martin et al. (1975) is adopted in the initial effective stress and the corrected SPT blows number
this study. Martin et al. (1975) defined the excess pore pressure equivalent to clean sand condition (N160-cs), as shown in equation
Δu, as shown in equation (4) (7)
Δu = Er ⋅ Δ𝜀vd (4) Sr
= 0.0055 N1 60−cs (7)
( )

𝜎vo
where Δu is the excess pore pressure, Er is the rebound or the where Sr is the residual strength, and 𝜎vo

is the effective vertical
recoverable modulus, and Δεvd is the incremental plastic volumet- stress.
ric strain that would occur under drained cyclic loading conditions. Stark and Mesri (1992) computed the “Residual strength” (Sr)
The approach by Martin et al. (1975) relates the volumetric not the “Steady-state strength” (Css). The “Residual strength”
strain resulting from a simple shear test on a soil specimen is defined as the shear strength that can be mobilized by an ele-
under drained cyclic loading conditions and the excess pore pres- ment of soil when strained monotonically to very large strains
sure generated under undrained cyclic loading conditions. The (Kramer and Elgamal 2001). In this study, the steady-state
adopted dynamic constitutive model requires definitions of the strength (Css) is considered equal to the residual strength (Sr).
recoverable modulus (Er) and the incremental plastic volumetric
Downloaded by [Dr Mohamed Mansour] at 05:24 15 May 2016

strain (Δεvd). Hence, the excess pore pressure (Δu) is calculated.


Martin et al. (1975) calculated the rebound modulus (Er) from the Analysis results
results of unloading tests in the oedometer apparatus. Hence, the
This section presents the results of the numerical analysis
recoverable or rebound modulus is taken equal to the unloading–
conducted to calculate the ultimate seismic bearing capacity
reloading modulus (Eur). The value of Eur for sandy soils is taken
of partially liquefiable sands. The investigated parameters are
three times the secant modulus at 50% of the failure stress (E50).
the footing width (B), the relative density (Dr) represented by
According to Schanz, Vermeer and Bonnier (1999), the value of
the corrected SPT blows count (NSPT), the foundation depth
E50 can be determined from equation (5), as follows
( � )m (Df), and the peak acceleration response at the foundation level
𝜎3 (amax).
E50 = Eref ⋅ (5)
pref A stress–settlement curve is constructed for each case by
applying different contact stresses, running static and seis-
where 𝜎3′ is the effective confining pressure, which equals the
mic finite element analyses, and calculating the permanent
at-rest earth pressure coefficient (Ko) times the effective vertical
post-earthquake deformation. The ultimate stress is defined
stress (𝜎v′ ), pref is the reference pressure (100 kPa), m is a stress
as the value above which the settlement tends to be infinite.
exponent and equals 0.5 for cohesionless soils (Janbu 1963),
Graphically, two tangents are drawn from the first and last
and Eref is the modulus corresponding to the reference pressure.
points on the curve. The intersection of the two tangents defines
The value of Eref is taken equal to NSPT in units of megapascals
the ultimate seismic bearing capacity, as illustrated in Fig. 5.
(Schanz and Vermeer 1998).
Martin et al. (1975) developed a four constants equation to
calculate the plastic volumetric strain increment (Δεvd), as Ultimate Seismic bearing capacity
shown in equation (6)
C3 𝜀2vd Table 3 summarizes the values of the ultimate seismic bearing
Δ𝜀vd = C1 𝛾 − C2 𝜀vd + (6)
( )
𝛾 + C4 𝜀vd capacity obtained from the numerical model (called hereinafter

The volume change constants, C1, C2, C3, and C4, are related to
the soil density. The plastic volumetric strain increment (Δεvd)
is a function of the total accumulated volumetric strain (εvd)
and the amplitude of the current shear strain (γ). Bhatia (1982)
conducted extensive laboratory tests to determine the volume
change constants for sand with relative densities of 45, 54,
60, and 68%. Interpolation is used to estimate the values of
the volume change constants that correspond to the relative
densities investigated in this study (35.0, 42.5, 50.0, 57.5, and
65.0%), based on Bhatia (1982) results. The volumetric strain
functions for the different values of NSPT are shown in Fig. 4.

Steady-state strength (Css)


4 Volumetric strain functions at a shear strain amplitude
Sladen, D'Hollander and Krahn (1985) defined the collapse of 0.2% considering different NSPT values (by
surface as the line passing through the peaks of the effective interpolation of Bhatia 1982 results)

 International Journal of Geotechnical Engineering  2016  VOL. 10   NO. 2 127


Mansour et al. Seismic bearing capacity of shallow foundations

that extends to a depth of B below the foundation level. Table


4 shows the values of the pore pressure ratio (ru). The case
of Df = 0.0 m is used to calculate the ultimate seismic bearing
capacity based on Day (2010) equation, because the equation
assumes a nil effect of the foundation depth on the ultimate
seismic bearing capacity (punching shear failure assumption).
Examining the results shown in Figs. 7–11, the following
can be established •:
1. The numerical seismic bearing capacity increases in
general with the footing width, not according to a per-
fect linear relationship like the ultimate static bearing
capacity. This is attributed to the effect of the excess
pore pressure, which does not increase linearly with the
footing width.
5 Stress–settlement curves for NSPT = 20 and B = 3.0 m 2. The numerical seismic bearing capacity for a foundation
depth of 1.5 m is always larger than the corresponding
value for a surface footing. This indicates that the founda-
the numerical seismic bearing capacity). Samples of the stress– tion depth affects the seismic bearing capacity of medium
settlement curves are shown in Figs. 5 and 6 for the cases of dense saturated sands. The effect of the foundation depth
NSPT = 20 and footing widths of 3.0 and 5.0 m. Each figure shows is investigated in more detail in the following section.
two curves; one for the case of a foundation depth (Df) of 0.0 m, 3. In comparison with the numerical results, the empiri-
Downloaded by [Dr Mohamed Mansour] at 05:24 15 May 2016

and the other for the case of Df = 1.5 m. cal equation by Day (2010) overestimates the ultimate
seismic bearing capacity of surface footings. The over-
estimation increases with both the footing width and the
Effect of the footing width (B) soil relative density (or NSPT). For footings founded at a
Figures 7–11 show the effect of the footing width on the ulti- depth of 1.5 m, the empirical equation underestimates
mate static and seismic bearing capacity. The numerical seismic the ultimate seismic bearing capacity for NSPT = 10, 15,
bearing capacity is compared with the ultimate static bearing and 20. For NSPT = 25 and 30, the empirical equation
capacity and the ultimate seismic bearing capacity calculated underestimates the ultimate seismic bearing capacity
assuming punching shear failure (Day 2010). for footing widths smaller than 6.0 and 5.0 m, respec-
The ultimate static bearing capacity (qult-net) is calculated tively, whereas it overestimates the ultimate seismic
from the general bearing capacity equation, which is expressed bearing capacity for greater footing widths.
as follows for a strip footing resting on a cohesionless soil 4. The overestimation of the ultimate seismic bearing
capacity of surface footings by Day's equation suggests
1 that the seismic bearing capacity factor, Nγ, should not
qult-net = q.Nq + B𝛾2 N𝛾 − q (8)
2 be taken equal to the static value, as suggested by Day
where q is the effective overburden stress at the foundation (2010). This observation is confirmed by comparing
level. The bearing capacity factors, Nq and Nγ, are calculated the slopes of the curves of the static and numerical
from the following equations seismic bearing capacity. The slopes of the numerical
seismic bearing capacity curves are 0.48–0.54 times the
corresponding static bearing capacity slopes. The slope
( )
𝜑�
Nq = e𝜋 tan 𝜑 tan2 45 +

(8 – a) of the ultimate bearing capacity versus footing width
2
relationship reflects the combined effects of the buoyant
unit weight of the soil beneath the footing, the bearing
N𝛾 = Nq − 1 ⋅ tan (1.4𝜑� ) capacity factor Nγ, and the pore pressure ratio, ru.
(8 – b)
( )
5. The numerically calculated pore pressure ratios in Table
In order to calculate the ultimate seismic bearing capacity 4 are smaller than those determined from Marcuson
from Day (2010) equation, the pore pressure ratio (ru) should and Hynes (1990) charts. For example, for the cases of
be determined. The pore pressure ratio is calculated from the NSPT = 10 and 15, the factor of safety against liquefaction
output of the seismic analysis as the average value in the zone calculated by the simplified procedure of Seed and Idriss

Table 3 Numerical seismic bearing capacity, qult(s), (kPa)

B = 1.0 m B = 3.0 m B = 5.0 m B = 7.0 m

NSPT Df = 0.0 m Df = 1.5 m Df = 0.0 m Df = 1.5 m Df = 0.0 m Df = 1.5 m Df = 0.0 m Df = 1.5 m


10 58 230 150 350 230 470 320 630
15 78 290 180 440 290 570 390 750
20 85 505 210 600 335 710 470 780
25 110 550 270 640 370 770 500 880
30 120 600 300 700 460 840 670 1000

128 International Journal of Geotechnical Engineering  2016  VOL. 10   NO. 2


Mansour et al. Seismic bearing capacity of shallow foundations

9 Effect of footing width on the ultimate static and


6 Stress–settlement curves for NSPT = 20 and B = 5.0 m seismic bearing capacity (NSPT = 20)
Downloaded by [Dr Mohamed Mansour] at 05:24 15 May 2016

7 Effect of footing width on the ultimate static and 10 Effect of footing width on the ultimate static and
seismic bearing capacity (NSPT = 10) seismic bearing capacity (NSPT = 25)

8 Effect of footing width on the ultimate static and


11 Effect of footing width on the ultimate static and
seismic bearing capacity (NSPT = 15)
seismic bearing capacity (NSPT = 30)

(1971) for a peak ground acceleration of 0.15 g is < 1.0 The results shown in Figs. 7–11 indicate that the empirical
(refer to Fig. 1). This corresponds to a pore pressure ratio equation suggested by Day (2010) neglects the effect of the
of 1.0, i.e. zero seismic bearing capacity. For the cases foundation depth, which leads to a remarkable underestimation
of NSPT = 20, 25, and 30, the values of the factor of safety of the ultimate seismic bearing capacity of partially liquefiable
against liquefaction based on the simplified procedure sands. Regarding the γ-term, the equation overestimates the
equal 1.1, 1.4, and 2.3, respectively. The corresponding value of the bearing capacity factor Nγ, which overestimates
ru values equal 0.6, 0.2, and 0.05, according to Marcuson the bearing capacity, i.e. on the unsafe side. Moreover, Day
and Hynes (1990) charts. The numerical analysis yields (2010) suggested the use of Marcuson and Hynes (1990) charts
smaller values of the pore pressure ratio, except for the to determine the pore pressure ratio. The numerical results show
case of NSPT = 30 where the empirical and numerical val- that Marcuson and Hynes (1990) charts overestimate the value
ues are in practical match. of the pore pressure ratio.

 International Journal of Geotechnical Engineering  2016  VOL. 10   NO. 2 129


Mansour et al. Seismic bearing capacity of shallow foundations

Table 4 Pore pressure ratio (ru) deduced from the seismic numerical analysis

Df = 0.0 m Df = 1.5 m

NSPT B = 1.0 m B = 3.0 m B = 5.0 m B = 7.0 m B = 1.0 m B = 3.0 m B = 5.0 m B = 7.0 m


10 0.18 0.21 0.25 0.28 0.34 0.31 0.30 0.34
15 0.12 0.17 0.16 0.24 0.19 0.23 0.24 0.27
20 0.07 0.10 0.10 0.14 0.11 0.10 0.11 0.16
25 0.05 0.05 0.07 0.11 0.08 0.08 0.09 0.13
30 0.04 0.04 0.05 0.05 0.05 0.07 0.06 0.06

Effect of the foundation depth (Df)


The numerical analysis results indicate that the ultimate seis-
mic bearing capacity of footings founded at a depth of 1.5 m is
greater than the ultimate seismic bearing capacity of surface
footings. Additional numerical analyses are conducted for the
cases of NSPT = 20 and NSPT = 25 with a foundation depth of 2.5 m,
as shown in Figs. 12 and 13, respectively. These figures show
that the numerical seismic bearing capacity versus foundation
depth relationship can be approximated by a linear trend line.
Downloaded by [Dr Mohamed Mansour] at 05:24 15 May 2016

The slope is nearly the same for both footing widths, and is
about 0.38–0.50 the slope of the ultimate static bearing capacity
curve.
12 Effect of foundation depth on the ultimate static and
seismic bearing capacity (NSPT = 20)
Seismic bearing capacity factors
The results shown in the previous two sections indicate that
the ultimate seismic bearing capacity should include a q-term
in addition to the γ-term. Hence, the ultimate seismic bearing
capacity equation can be written in the following form:
1
qult = q.Nq + B ⋅ 𝛾sub ⋅ N𝛾 ⋅ 1 − ru (9)
( )
2
This section analyzes the study results in order to estimate
representative values of the seismic bearing capacity factors,
Nq and Nγ. Considering that the seismic bearing capacity versus
foundation depth relationship can be approximated by a linear
trend line (see Figs. 12 and 13), the results shown in Table
3 can be used to determine the value of the seismic bearing
capacity factor, Nq. The slope of the seismic bearing capacity 13 Effect of foundation depth on the ultimate static and
versus foundation depth relationship for a strip footing equals seismic bearing capacity (NSPT = 25)
γ1*Nq. The unit weight above the foundation level (γ1) is taken
18.0 kN m− 3 in the conducted analyses. Therefore, the value of
the seismic bearing capacity factor Nq for the studied medium lower than the static value. This indicates a reduced contribu-
dense saturated sand can be calculated as follows tion of the q-term to the ultimate seismic bearing capacity, but
q − Df = 1.5 − qult − Df = 0.0 not a zero effect, as assumed by Day (2010).
Nq = ult The calculation of the seismic bearing capacity factor, Nq,
𝛾1 ⋅ D f and the pore pressure ratio, ru, for each of the investigated cases
(10)
qult − Df = 1.5 − qult − Df = 0.0 enable calculating the seismic bearing capacity factor, Nγ. The
= calculated Nγ values are plotted against the internal friction
18 ∗ 1.5
Equation (10) assumes that the pore pressure ratio is the same angle in Fig. 15. As for the Nq results, the seismic bearing capac-
for foundation depths of 0.0 and 1.5 m. Referring to Table 4, the ity factor, Nγ, is smaller than the corresponding static value.
differences in the pore pressure ratio values between the Df = 0.0 The predictive capability of the deduced equation and the
and 1.5 m cases are not expected to have a big impact on the seismic bearing capacity factors is checked by calculating the
calculated values of Nq from equation (10). Figure 14 shows a ultimate seismic bearing capacity from equation (9), based on
plot of the minimum, maximum, and average seismic bearing the average Nq and Nγ values from Figs. 14 and 15, and the ru
capacity factor, Nq, versus the internal friction angle φ. The values from Table 4. The predicted values are compared with
figure shows also the static bearing capacity factor Nq for com- the values deduced from the stress–settlement curves (Table
parison. The seismic bearing capacity factor Nq is considerably 3). The error is within ± 20%.

130 International Journal of Geotechnical Engineering  2016  VOL. 10   NO. 2


Mansour et al. Seismic bearing capacity of shallow foundations

Allowable seismic bearing capacity where;


qult(a) is the ultimate seismic bearing capacity at the desired
The allowable seismic bearing capacity is the important output peak acceleration response (kPa).
that concerns the design engineer rather than the ultimate bear- qult(0.15 g) is the ultimate seismic bearing capacity at a peak
ing capacity. Under static conditions, the minimum allowable acceleration response of 0.15 g (kPa).
factor of safety ranges from 2.5 to 3.0, and could be greater a is the peak acceleration response at the foundation level
to satisfy the allowable settlement tolerance (serviceability (g).
requirements). Under seismic conditions, the calculation of m is the slope of the linear relationship.
the post-earthquake permanent deformation cannot be usually The value of the slope m varies with NSPT and the footing
quantified with high certainty. This leads to conservative esti- width, as shown in Fig. 23.
mates of the factor of safety against seismic bearing capacity
failure. Day (2010) recommended a factor of safety of 5.0
against seismic bearing capacity failure. Discussion
The numerical analyses conducted in this paper yield stress– The study focuses on saturated sands with FSliq in the range of
settlement curves for the investigated cases. The calculated 1.0–2.0. Hence, the investigated sands may not be completely
numerical seismic bearing capacity values correspond to con- liquefiable but experience excess pore pressures during seismic
siderably high values of settlement. Obviously, these values excitation that may lead to a drop in the effective stresses and,
cannot be accepted from a serviceability point of view. hence, the seismic bearing capacity. The literature is not con-
Using the suggested factor of safety by Day (2010) as a sistent with respect to the definition of FSliq above which the
guide, the allowable seismic bearing capacity versus settle- sand is said to be completely safe against liquefaction, as sum-
ment relationships are developed by considering safety factors marized in the Introduction section. This study covers a range of
Downloaded by [Dr Mohamed Mansour] at 05:24 15 May 2016

starting from 3.0 and up to 12.0. The results are shown in Figs. FSliq that encompasses the limits defined by different codes and
16–20 for the case of a foundation depth of 1.5m, since the case investigators. The results show that both excess pore pressures
of a surface footing is investigated only to check the assump- and post-earthquake deformations control the value of the safe
tions of the empirical equation by Day (2010). The curves can FSliq above which the sand is safe against liquefaction.
be used to determine the stress corresponding to the specified
allowable post-earthquake settlement by the design code.
It has to be noted that the above results are applica-
ble for the case of a peak acceleration response at the
foundation level of 0.15 g, and a medium dense sand layer
thickness of 10 m. The medium dense sand layer thickness
has a significant effect on the settlement value. Additional
analyses are conducted for a 5 m-wide footing resting on
varying thicknesses of the medium dense sand layer, i.e.
10.0, 7.0, and 4.0 m. The value of NSPT of the medium dense
sand is taken 20. The applied stress equals 100 kPa. The
results are shown in Fig. 21. The settlement is significantly
reduced when the medium dense sand layer thickness is
decreased.

Effect of the peak acceleration response (amax) 14 Bearing capacity factor, Nq, of medium dense sand
for static and seismic conditions
All the previous analyses are conducted by applying an artificial
earthquake acceleration–time history at the lower boundary,
such that the peak acceleration response at the foundation level
(amax) equals 0.15 g. Additional analyses are conducted in order
to investigate the effect of the peak acceleration response (amax)
on the value of the ultimate seismic bearing capacity. Footing
widths of 3.0 and 5.0 m founded at 1.5 m, and NSPT of 15, 20,
and 25 are selected for the additional study. Peak acceleration
responses of 0.3 and 0.4 g are investigated. Figure 22 shows
the effect of the peak acceleration response on the ultimate
seismic bearing capacity for the studied cases. As expected,
the ultimate seismic bearing capacity decreases as the peak
acceleration response increases. A linear trend equation may
mathematically express the degradation of the ultimate seis-
mic bearing capacity. Therefore, the ultimate seismic bearing
capacity at peak accelerations >0.15 g can be estimated from
the following equation
15 Bearing capacity factor, Nγ, of medium dense sand
qult(a) = qult(0.15g) − m ∗ (a − 0.15) (11) for static and seismic conditions

 International Journal of Geotechnical Engineering  2016  VOL. 10   NO. 2 131


Mansour et al. Seismic bearing capacity of shallow foundations

16 Stress–settlement curve for different footing widths 18 Stress–settlement curve for different footing widths
(NSPT = 10, Df = 1.5 m) (NSPT = 20, Df = 1.5 m)
Downloaded by [Dr Mohamed Mansour] at 05:24 15 May 2016

17 Stress–settlement curve for different footing widths 19 Stress–settlement curve for different footing widths
(NSPT = 15, Df = 1.5 m) (NSPT = 25, Df = 1.5 m)

The terms related to liquefaction susceptibility need to be


clearer in the literature. The threshold above which complete
liquefaction occurs is different from the threshold that causes
the excess pore pressures to reduce the seismic bearing capacity.
Karamitros et al. (2013), for example, investigated the seismic
bearing capacity of saturated sands, overlain by clay crust, with
relative densities of 40, 50, and 60%, and they described the
sand as “liquefiable.” The results by Karamitros et al. (2013)
indicate that this sand is not completely liquefiable but rather a
partially liquefiable sand that experiences excess seismic pore
pressure leading to bearing capacity decrease. The Euro Code
definition does not clearly indicate whether sands with FSliq
higher than 1.25 are completely safe against liquefaction such 20 Stress–settlement curve for different footing widths
that the seismic excess pore pressures are to be ignored or not. (NSPT = 30, Df = 1.5 m)
The study results show that each of the footing width, the
soil relative density, and the foundation depth affects the ulti-
mate seismic bearing capacity. Hence, the q-term should not seismic bearing capacity of shallow foundations on partially
be ignored as Day (2010) assumed. Both the q-term and γ-term liquefiable saturated sands. The design procedure is summa-
contribute to the ultimate seismic bearing capacity, but with rized in the following steps:
reduced values of the seismic bearing capacity factors, Nq and a. Use equation (9), the average values of the seismic bear-
Nγ. The ultimate seismic bearing capacity of partially liquefi- ing capacity factors, Nq and Nγ, from Figs. 14 and 15,
able sands can be written in the form shown in equation (9). respectively, and the pore pressure ratio values in Table
The study presents also charts to enable calculating the 4 to calculate the ultimate seismic bearing capacity.
stress corresponding to the allowable post-earthquake settle- b. According to the expected peak acceleration response
ment. Therefore, the conducted study provides a simple frame- at the foundation level, calculate the corrected ultimate
work to enable the design engineer to determine the allowable seismic bearing capacity from equation (11).

132 International Journal of Geotechnical Engineering  2016  VOL. 10   NO. 2


Mansour et al. Seismic bearing capacity of shallow foundations

The deduced bearing capacity equation overcomes the lim-


itations of the equation suggested by Day (2010), which can be
summarized in the following points:
– The equation assumes a punching shear failure, and,
hence, the q-term is ignored.
– The bearing capacity factor, Nγ, is taken equal to the static
value.
– The pore pressure ratio is greater than the values predicted
by numerical analysis.

Conclusion
21 Effect of medium dense sand thickness on the
The conclusions of the conducted study can be summarized in
footing settlement (q = 100 kPa, NSPT = 20, B = 5 m, and
the following points:
Df = 1.5 m)
– The ultimate seismic bearing capacity of partially lique-
fiable sands should be calculated by superposition of both the
q-term and γ-term.
Charts are developed to calculate the seismic bearing capacity
factors, Nq and Nγ, based on the internal friction angle value in
the medium dense range.
– The estimated values of the ultimate seismic bearing
Downloaded by [Dr Mohamed Mansour] at 05:24 15 May 2016

capacity correspond to a peak acceleration response of 0.15 g


at the foundation level. An equation is developed to estimate
the corresponding ultimate seismic bearing capacity at higher
peak acceleration responses.
– Additional correction may be required to account for the
partially liquefiable sand layer thickness. The values proposed
in this study correspond to a thickness of 10.0 m below the
foundation level.
22 Effect of the peak acceleration response at the – simple design framework is developed to enable the
foundation level on the ultimate seismic bearing design engineer to determine the allowable seismic bearing
capacity for the cases of NSPT = 15, 20, 25; B = 3.0, capacity considering both the ultimate and serviceability limit
5.0 m; and Df = 1.5 m states.

References
Abdel-Motaal, M. A. 1999. Soil effect on the dynamic behavior of framed
structures. Ph.D. Thesis, Ain Shams University, Cairo, Egypt..
Bhatia, S. K. 1982. The verification of relationships for effective stress method
to evaluate liquefaction potential of saturated sands. Ph.D. Thesis, University
of British Columbia, Vancouver, BC, Canada..
Bouckovalas, G. D., Valsamis, A. I. and Andrianopoulos, K. I. 2005. Pseudo
static vs. performance based seismic bearing capacity of footings on
liquefiable soil. In: Proceedings of the TC4 Satellite Conference, 16th
ICSMGE. Osaka. pp. 22–29..
Building Seismic Safety Council (BSSC). 1997. NEHRP recommended
provisions for seismic regulations for new buildings and other structures,
Washington, DC, Commentary Prepared for the Federal Emergency
Management Agency (FEMA).
Cascone, E. and Bouckovalas, G. D. 1998. Seismic bearing capacity of
footings on saturated sand with a clay cap. In: Proceedings of the 11th
23 Variation of the slope m with the footing width and NSPT
European Conference in Earthquake Engineering, Paris. Rotterdam: A.A.
Balkema..
Day, R. W. 2010. Foundation engineering handbook – design and construction
c. Calculate an allowable seismic bearing capacity by divid- with the 2009 international building code, 2nd ed, McGraw Hill.
ing the ultimate value obtained in (b) by a factor of safety Dickenson, S., McCullough, N. J., Barkau, N. J. and Wavra, B. J. 2002.
Assessment and mitigation of liquefaction hazards to bridge approach
of 5.0, as suggested by Day (2010). embankments in Oregon, Washington, DC, Federal Highway Administration.
d. According to the specified tolerable post-earthquake Eurocode 8 – Part 5. 1998. Design provisions for earthquake resistance of
settlement by the design code, use the charts in Figs. structures. Part 5: foundations, retaining structures and geotechnical aspects.
16–20 to obtain another value of the allowable stress that British Standards Institution..
corresponds to the allowable settlement. Finn, W. D. L., Yogendrakumar, M., Lo, R. C. and Yoshida, N. 1988. Direct
computation of permanent seismic deformations. In: Proceedings of the
e. The design allowable seismic bearing capacity is the 9th World Conference on Earthquake Engineering, Tokyo-Kyoto, Japan,
smaller value obtained from (c) and (d). VIII. pp. 219–224..

 International Journal of Geotechnical Engineering  2016  VOL. 10   NO. 2 133


Mansour et al. Seismic bearing capacity of shallow foundations

Ishibashi, I. and Zhang, X. 1993. Unified dynamic shear moduli and damping Naesgaard, E., Byrne, P. M. and Ven Huizen, G. 1998. Behaviour of light
ratios of sand and clay, Soils and Foundations, 33, (1), 182–191. structures founded on soil ‘crust’ over liquefied ground, in Geotechnical
Ishihara, K., Tatsuoka, F. and Yasuda, S. 1975. Undrained deformation and earthquake engineering and soil dynamics III, geotechnical special
liquefaction of sand under cyclic stresses, Soils and Foundations, 15, (1), publication no. 75, (eds. Dakoulas P., et al.). 422–433, Seattle, USA, ASCE.
29–44. Schanz, T. and Vermeer, P. A. 1998. Pre-failure deformation behaviour
Janbu, N. 1963. Soil compressibility as determined by oedometer and triaxial of geomaterials, Institution of Civil Engineers, 383–387,
tests. In: Proceedings of 3rd European Conference on Soil Mechanics and http://dx.doi.org/10.1680/pdbog.26421.
Foundation Engineering. Wiesbaden. pp. 19–25. Schanz, T., Vermeer, P. A. and Bonnier, P. G. 1999. Formulation and verification
Karamitros, D. K., Bouckovalas, G. D., Chaloulos, Y. K. and Andrianopoulos, of the Hardening-Soil model, in Beyond 2000 in computational geotechnica,
K. I. 2013. Numerical analysis of liquefaction-induced bearing capacity (ed. Brinkgreve R. B. J.., 281–290, Rotterdam, Netherlands, Balkema.
degradation of shallow foundations on a two-layered soil profile, Soil Seed, H. B. and Idriss, I. M. 1970. Soil moduli and damping factors for dynamic
Dynamics and Earthquake Engineering, 44, 90–101. response analyses, Berkeley, CA, USA, Earthquake Engineering Research
Kourkoulis, R., Anastasopoulos, I., Gelagoti, F. and Gazetas, G. 2010. Center, University of California [Report EERC 70-10].
Interaction of foundation-structure systems with seismically precarious Seed, H. B. and Idriss, I. M. 1971. Simplified procedure for evaluating soil
slopes: numerical analysis with strain softening constitutive model, Soil liquefaction potential, Journal of Geotechnical Engineering Division, ASCE,
Dynamics and Earthquake Engineering, 30, 1430–1445. 97, (9), 1249–1273.
Kramer, S. L. and Elgamal, A. W. 2001. Modeling soil liquefaction hazards Seed, H. B., Idriss, I. M. and Arango, I. 1983. Evaluation of liquefaction
for performance-based earthquake engineering. Pacific Earthquake potential using field performance data, Journal of Geotechnical Engineering,
Engineering Research Center (PEER) Report. Available at: < http://peer. ASCE, 109, (3), 458–482.
berkeley.edu/publications/peer_reports/reports_2001/0113.pdf> [Accessed Seed, H. B., Wong, R. T., Idriss, I. M. and Tokimatsu, K. 1984. Moduli and
June 2015]. damping factors for dynamic analyses of cohesionless soils, Journal of
Kramer, S. L. and Hall, W. J. eds 1996. Geotechnical earthquake engineering, Geotechnical Engineering, ASCE, 112, (11), 1016–1032.
in Prentice-Hall International Series in Civil Engineering and Engineering Sladen, J. A., D'Hollander, R. D. and Krahn, J. 1985. The liquefaction of sands, a
Mechanics, collapse surface approach, Canadian Geotechnical Journal, 22, (4), 564–578.
Law, K. T., Refahi, K., Ko, P., Lam, T. and Hassan, P. 2005. Seismic deformation Stark, T. D. and Mesri, G. 1992. Undrained shear strength of sands for stability
Downloaded by [Dr Mohamed Mansour] at 05:24 15 May 2016

of Waba Dam. In: Conference Proceeding: Canadian Dam Association, analysis, Journal of Geotechnical Engineering, ASCE, 118, (11), 1727–1747.
Calgary, Alberta, Canada.. US Bureau of Reclamation. 2012. Seismic risks for embankments. Available
Marcuson, W. F. and Hynes, M. E. 1990. Stability of Slopes and Embankments at: < http://www.usbr.gov/ssle/damsafety/Risk/BestPractices/27-
during Earthquakes. In: Proceedings of the Geotechnical Seminar Sponsored SeismicEmbankment20121207.pdf> [Accessed June 2015].
by ASCE and the Pennsylvania Department of Transportation, Hershey, PA.. Wu, G. 1996. Volume change and residual pore water pressure of saturated
Martin, G. R., Finn, W. D. L. and Seed, H. B. 1975. Fundamentals of liquefaction granular soils to blast loads, Canada, Research Report Submitted to
under cyclic loading, Journal of the Geotechnical Engineering Division, NSERC. Available at: < http://www.wutecgeo.com/downloads/Wu_NSERC-
101, (GT5), 423–438. BLAST(2).pdf> [Accessed June 2015].
Mokhtar, A. A., Abdel-Motaal, M. A. and Wahidy, M. M. 2014. Lateral Wu, G. 2001. Earthquake-induced deformation analyses of the Upper San
displacement and pile instability due to soil liquefaction using numerical Fernando Dam under the 1971 San Fernando Earthquake, Canadian
model, Ain Shams Engineering Journal, 5, (4), 1019–1032. Geotechnical Journal, 38, (1), 1–15.

134 International Journal of Geotechnical Engineering  2016  VOL. 10   NO. 2

You might also like