Geo-Congress 2023 GSP 341 614

Seismic Bearing Capacity of an Embedded Strip Footing on Slope Using Modified Pseudo-
Dynamic Method

K. Halder, Ph.D.1; and D. Chakraborty, Ph.D.2

1
Research Fellow, Dept. of Civil Engineering, Univ. of Nottingham, UK.
Email: [email protected]
2
Assistant Professor, Dept. of Civil Engineering, Indian Institute of Technology Kharagpur,
Kharagpur, India. Email: [email protected]

ABSTRACT

By combining the modified pseudo-dynamic method with the lower bound finite element
limit analysis technique, the bearing capacity of an embedded strip footing placed on a soil slope
and under seismic loading is estimated. A series of parametric studies by varying coefficients of
seismic acceleration, friction angle of soil, inclination of the slope, embedded depth of footing,
and soil damping ratio is carried out. Obtained results are shown as the variation between the
bearing capacity factor and considered parameters. It is found that with the increasing footing
embedment, seismic bearing capacity increases considerably. The range of increment in the
seismic bearing capacity of an embedded footing varies from seven to ten times that of no
embedment. However, irrespective of footing embedment depth, the bearing capacity of footing
reduces with increasing values of seismic acceleration coefficients. When both the horizontal and
vertical seismic acceleration coefficients are considered together, seismic bearing capacity
reduces more. However, it is also found that the footing bearing capacity is higher for soil with a
higher damping ratio than that of the footing on a soil with a lower damping ratio. With
increasing friction angle of soil, footing bearing capacity increases, but reduces with increasing
slope inclination. State of stresses throughout the problem domain are plotted in the two-
dimensional axes to understand the failure mechanism of slope under pseudo-dynamic loading.
Results obtained from the study will be useful to the design engineers.

INTRODUCTION

Many times, foundations for various engineering structures like bridge abutments, hanging
ropeway, buildings, and transmission towers are constructed on the sloping ground, particularly
in the hilly areas. The instability and vulnerability of these structures increase when subjected to
seismic loading. In the past, various researchers (Kumar and Rao 2003; Askari and Farzaneh
2003; Chakraborty and Kumar 2015; Casablanca et al. 2021; Izadi et al. 2021; Yang et al. 2021)
studied the effect of seismicity on the bearing capacity of shallow foundation on slopes for
several combinations of slope geometry, foundation embedment, soil property, and seismic
loading. In all the above-mentioned studies, seismic loading is considered through pseudo-static
analysis, where the peak ground acceleration is represented by a constant horizontal and vertical
body forces throughout the soil mass. The pseudo-static analysis fails to reciprocate the effects of
phase and frequency difference of seismic loading with time. By including the effects of phase
and frequency difference of seismic loading with time, Steedman and Zeng (1990) first
formulated the pseudo-dynamic method. Later, many researchers (Ghosh 2008; Ghosh and
Choudhury 2011) estimated footing bearing capacity using the pseudo-dynamic method.

© ASCE
Geo-Congress 2023 GSP 341 615

Bellezza (2014) and Pain et al. (2015) modified the pseudo-dynamic method by incorporating (1)
ground amplification, (2) no stress at boundary, (3) soil damping, and (4) frequency. Nadgouda
and Choudhury (2021) utilized the modified pseudo-dynamic method and assumed a failure
surface to estimate the footing bearing capacity under seismic loading. However, to the best
knowledge of the authors, no such studies are carried out to estimate the footing bearing capacity
under pseudo-dynamic loading.
Therefore, the present study aims to compute the bearing capacity of a footing embedded on
a cohesionless soil slope under pseudo-dynamic loading using the lower bound finite element
limit analysis (LB-FELA) framework. It is to be noted that the LB-FELA provides a safe
estimate of ultimate load and there is no requirement of assuming the failure surface of the slope
before the analysis. Following, Halder et al. (2018), the footing is embedded at the slope edge.
Different parametric studies are carried out by varying coefficients of seismic acceleration (kh,
kv), friction angle of soil (ϕ), inclination of the slope (β), embedded depth of footing (H), and soil
damping ratio (ξ). Bearing capacity factor associated with soil unit weight (Nγ) are presented as
design charts for different mentioned parameters. The failure mechanism corresponding to the
critical or resonant time period is obtained by plotting the stress states within the slope.

PROBLEM STATEMENT

By combining the modified pseudo-dynamic and LB-FELA methods, the seismic bearing
capacity factor (Nγ) corresponding to the unit weight of soil is obtained for a shallow embedded
strip foundation located at a cohesionless slope edge. The two-dimensional plane strain problem
domain used in the present study is shown in Figure 1(a). Here, footing width is B, and slope
angle is denoted by β. A non-inclined and non-eccentric load of Qv is acting on the strip footing.
Footing embedment from the ground is H. Footing side and bottom interfaces with soil are
considered rough. Response of soil under loading is modelled by using the Mohr-Coulomb
constitutive relation. Dimensions of the problem domain in the horizontal and vertical directions
are fixed after carrying out several trials so that (1) the value Nγ does not alter with changing
problem dimensions, and (2) stress at failure does not stretch to the boundary. Depending upon
the above-mentioned requirements, the problem domain is expanded up to 21.12B in the
horizontal direction; whereas 14.29B to 16.29B in the vertical direction.
Shear and normal stresses are zero along slope face (JA) and horizontal ground surface (FG).
Fully rough interfaces are considered between (1) footing bottom and soil (KI), (2) footing sides
and soil (KG, IJ) and assumed to follow the condition:  xy  (c −  y tan  ) . A representative
finite element mesh used is shown in Figure 1(b).

METHODOLOGY

The present study follows the plane strain LB-FELA formulation of Sloan (1988). An
admissible stress field is first set up by implementing element equilibrium, stress boundary, and
yield criteria over the elements, boundary nodes, and whole domain, respectively.
In order to incorporate the effect of pseudo-dynamic loading, element equilibrium conditions
for static loading as detailed in Sloan (1988) are modified. Modified element equilibrium
conditions to incorporate seismic loading are expressed in Eq. 1.

𝜕𝜎𝑥 𝜕𝜏𝑥𝑦
+ = 𝑘ℎ𝑚𝑜𝑑 𝛾 (1a)
𝜕𝑥 𝜕𝑦

© ASCE
Geo-Congress 2023 GSP 341 616

𝜕𝜏𝑥𝑦 𝜕𝜎𝑦
+ = (1 − 𝑘𝑣𝑚𝑜𝑑 )𝛾 (1b)
𝜕𝑥 𝜕𝑦

Soil unit weight is expressed as γ. Here, khmod and kvmod are modified seismic acceleration
coefficients in the horizontal and vertical directions, which are dependent on (1) the total height
of soil domain (Ht), (2) angular frequency (ω), (3) shear (Vs), and primary (Vp) wave velocities,
(4) damping ratio of soil (ξ). Detailed formulation to obtain khmod and kvmod from these parameters
is provided in Pain et al. (2015) and Halder and Chakraborty (2022).

X
(a)
Ht = 14.29B to 16.29B

(b)

Figure 1. (a) Boundary conditions applied over the problem domain; (b) discretized
domain of slope with β = 20°, ϕ = 30°, H/B = 2.

An additional equality constraint Qh = khmodQv is employed to consider the inertial force that
occurs in the superstructure due to seismic loading. Shear stresses along with the soil-footing
base interface (KI) and normal stresses along with the soil-footing wall interfaces (KG, IJ) are
integrated to obtain Qh. After the formulation of the admissible stress field, maximization of
collapse load using nonlinear programming (Makrodimopoulos and Martin, 2006) is carried out.

© ASCE
Geo-Congress 2023 GSP 341 617

The final form of the optimization is expressed as follows in Eq. 2. A LB-FELA code is written
in MATLAB. The Second-order conic optimization is performed using MOSEK.

Maximize
𝑇
{𝑔𝑜𝑏𝑗 } {𝜎 𝑔 } (2a)

Subjected to

[𝐴 𝑔 ]{𝜎 𝑔 } = {𝑏 𝑔 } (2b)

In the above equation, [Ag] and {bg} are global matrices and vectors of the coefficients
obtained from the implementation of different constraints. {σg} is the global vector of all
unknown nodal stresses associated with different constraints.

COMPARISON

As there is no available study related to the estimation of footing bearing capacity of an

embedded strip footing under pseudo-dynamic loading, the present results are validated with one
of the reported solutions using pseudo-static analysis. Casablanca et al. (2021) combined the
method of stress characteristics and pseudo-static analysis to estimate the footing bearing
capacity on the slope. Figure 2 shows the variation between the results obtained from the present
study and that of Casablanca et al. (2021) for a slope combination of β = 20°, ϕ = 30°, kv/kh = 0,
0.5 with varying kh. The same LB-FELA code is used by modifying for the pseudo-static
analysis with khmod = kh and kvmod = kv. It is evident from Figure 2 that the present results are
lower than that obtained by Casablanca et al. (2021). It is attributed to the fact that Casablanca et
al. (2021) assumed the one-sided failure geometry and used the method of stress characteristics.
But the failure surface propagates on both sides of the footing, which is also obtained from the
present LB-FELA. However, the trend of the results is found to match for both the present study
and Casablanca et al. (2021).

RESULTS

Different parametric studies are carried out by varying (1) footing embedment depth (H/B =
0, 1, 2), seismic acceleration coefficients in the (2) horizontal directions (kh = 0, 0.05, 0.10, 0.15,
0.20, 0.25, 0.30), and (3) vertical directions (kv/kh = 0, 0.5, 1), (4) slope inclinations (β = 20°,
25°), (5) friction angle of soil (ϕ = 30°, 35°, 40°), (6) soil damping ratio (ξ = 5%, 10%, 20%).
Required values of angular frequency (ω), shear (Vs), and primary (Vp) wave velocities are
considered from Chakraborty and Choudhury (2014). Variation between Nγ and mentioned
parameters are plotted.
Figure 3(a-c) shows that the value of Nγ increases with the increasing footing embedment
depth for any values of inclinations of the slope, friction angle of soil, seismic acceleration
coefficient, and soil damping ratio. A significant amount of additional passive resistance from
both sides of embedment is offered by the soils that increases the footing bearing capacity. For
an instance, the Nγ value increases from 4.54 to 43.59 with increasing values of H/B from 0 to 2
with a slope of β = 20°, ϕ = 30°, kh = 0.05, kv/kh = 0.5, and ξ = 10%. With the increasing
horizontal seismic loading, the footing bearing capacity reduces continuously before reaching

© ASCE
Geo-Congress 2023 GSP 341 618

zero under a limiting kh value while keeping other parameters as same. The Nγ value reduces
between 49.26 and 31.85 with an increasing kh value between 0.05 and 0.15 with kh = 0.20. The
Nγ magnitude becomes zero for a slope having β = 20°, ϕ = 30°, H/B = 2, kv/kh = 0.5. The footing
bearing capacity under pseudo-dynamic loading increases if the soil damping ratio increases. An
increasing value of soil damping ratio means increasing soil absorptance capacity under seismic
load. For instance, with a slope combination of β = 20°, ϕ = 40°, H/B = 2, kh = 0.15, kv/kh = 0.5,
value of bearing capacity factor increases from 72.58 to 118.79 with the increasing value of soil
damping ratio from 5% to 20%. Figure 3(c-e) shows that the Nγ value increases with the
increasing ϕ value irrespective of the magnitude of seismic loading. The Nγ magnitude increases
from 45.83 to 158.53 with increasing ϕ value between 30° and 40° for a slope having β = 20°,
H/B = 2, kh = 0.05, kv/kh = 0.5, and ξ = 10%. It can be found from Figure 3(f) that with increasing
slope angle, bearing capacity reduces. Figure 3(g) shows that when both vertical and horizontal
seismic loading acts together, the magnitude of seismic bearing capacity becomes lowest. For
example, with β = 20°, ϕ = 30°, H/B = 2, ξ = 10%, the values of Nγ are 23.22 and 42.90
corresponding to kh = 0.10, kv = 0.10 and kh = 0.10, kv = 0.0, respectively.

6
Casablanca et al. (2021), kv/kh = 0
kv/kh = 0.5
5 Present study, kv/kh = 0
= 0.5
4
Nγ

2
β = 20 ,  = 30
1
0.05 0.1 0.15
kh

Figure 2. Validation of present results with Casablanca et al. (2021).

FAILURE PATTERNS

The slope failure mechanism under the strip and pseudo-dynamic loading are obtained by
plotting stress states in the coordinate axes. The stress state at each node is quantified as a
dimensionless ratio a/d, where, a = ( x −  y ) + (2 xy ) and d = [−( x +  y ) sin  ] . Failure at
2 2 2

any node is indicated by a/d =1; whereas, a/d < 1 indicates no failure. Figure 4 shows different
failure patterns for various combinations of slope and footing geometry and seismic loading.
When a surface footing is subjected to only static loading, a non-yielding zone occurs beneath
the footing, whereas a plastic zone develops on the footing sides (Figure 4a). With pseudo-
dynamic loading, the plastic zone propagates more towards the slope face side (Figure 4b).
Figure 4(c-d) shows that when footing embedment is present, the plastic zone propagates below
and around the footing, i.e., passive resistance of soil improves. As a result of that footing
bearing capacity is also enhanced. With increasing footing embedment depth, the area of the
plastic zone expands as shown in Figure 4(b).

© ASCE
Geo-Congress 2023 GSP 341 619

10 40
β = 20 , ϕ = 30 , H/B = 0 ξ = 20% β = 20 , ϕ = 30 , H/B = 1 ξ = 20%
= 10% = 10%
= 5% 30 = 5%

5 20

Nγ
Nγ

10

0 0
0.00 0.05 0.10 0.15 0.20 0.25 0.00 0.05 0.10 0.15 0.20 0.25
kh kh
(a) (b)
60 120
β = 20 , ϕ = 30 , H/B = 2 ξ = 20% β = 20 , ϕ = 35 , H/B = 2 ξ = 20%
50 = 10% 100 = 10%
= 5% = 5%
40 80

30 60

Nγ
Nγ

20 40

10 20

0 0
0 0.05 0.1 0.15 0.2 0.25 0.00 0.10 0.20 0.30 0.40
kh kh
(c) (d)
250 160
β = 20 , ϕ = 40 , H/B = 2 ξ = 20% β = 25 , ϕ = 40 , H/B = 2 ξ = 20%
200 = 10% = 10%
= 5% 120 = 5%
150
80
Nγ

Nγ

100

50 40

0 0
0.00 0.10 0.20 0.30 0.40 0 0.1 0.2 0.3 0.4
kh kh
(e) (f)
60
β = 20 , ϕ = 30 , H/B = 2, kv/kh = 0.0
ξ = 10% = 0.5
= 1.0
40
Nγ

20

0
0.05 0.10 0.15 0.20 0.25
(g) kh

Figure 3. Variation between (a) Nγ and kh, ξ for β = 20°,  = 30°, H/B = 0; (b) Nγ and kh, ξ
for β = 20°,  = 30°, H/B = 1; (c) Nγ and kh, ξ for β = 20°,  = 30°, H/B = 2; (d) Nγ and kh, ξ
for β = 20°,  = 35°, H/B = 2; (e) Nγ and kh, ξ for β = 20°,  = 40°, H/B = 2; (f) Nγ and kh, ξ
for β = 25°,  = 40°, H/B = 2; (g) Nγ and kh, kh/kv for β = 20°,  = 30°, H/B = 2, ξ = 10%.

© ASCE
Geo-Congress 2023 GSP 341 620

Figure 4. Failure patterns obtained for a slope combination of β = 20°,  = 30° with (a) no
seismic loading (kh = kv = 0) and H/B = 0; (b) kh = 0.1, kv/kh = 0.5, ξ = 10%, H/B = 0; (c) kh =
0.1, kv/kh = 0.5, ξ = 10%, H/B = 1; (d) kh = 0.1, kv/kh = 0.5, ξ = 10%, H/B = 2.

CONCLUSION

The lower bound value of seismic bearing capacity is obtained for an embedded footing
located at a cohesionless soil slope edge. Seismic loading is incorporated by utilizing the

© ASCE
Geo-Congress 2023 GSP 341 621

modified pseudo-dynamic method. With the increasing footing embedment depth, passive
resistance offered by the soils on the sides of the footing increases that ultimately increases the
seismic bearing capacity of the footing placed on the cohesionless soil slope. However, the
seismic bearing capacity of footing reduces considerably with the increasing seismic loading.
This rate of reduction in the magnitude of the seismic bearing capacity of the footing reduces
when it is placed on a soil slope with a higher damping ratio. Design charts presented in this
study illustrate that the seismic bearing capacity increases with increasing soil friction angle and
reduces with increasing slope angle irrespective of the magnitude of seismic loading. From the
failure plots, it is observed that when the footing is embedded, the plastic zone propagates below
and around the footing which helps in increasing the bearing capacity of the foundation. The
plastic zone propagates more towards the direction of the seismic loading which is in the slope
face direction.

REMARK

The effect of soil dilatancy is not considered in the present study. However, the effect of soil
dilatancy can be considered in the Mohr-Coulomb constitutive model and lower bound finite
element limit analysis technique by following the approach of Drescher and Detouney (1993)
(refer Halder et al. 2018).

REFERENCES

Askari, F., and Farzaneh, O. (2003). “Upper-bound solution for seismic bearing capacity of
shallow foundations near slopes.” Geotechnique, 53(8), 697-702.
Bellezza, I. (2014). “A new pseudo-dynamic approach for seismic active soil thrust.”
Geotechnical and Geological Engineering, 32(2), 561-576.
Casablanca, O., Biondi, G., Cascone, E., and Filippo, G. D. (2021). “Static and seismic bearing
capacity of shallow strip foundations on slopes.” Geotechnique, 1-15.
Chakraborty, D., and Choudhury, D. (2014). “Sliding stability of non-vertical waterfront
retaining wall supporting inclined backfill subjected to pseudo-dynamic earthquake forces.”
Applied Ocean Research, 47, 174-182.
Chakraborty, D., and Kumar, J. (2015). “Seismic bearing capacity of shallow embedded
foundations on a sloping ground surface.” International Journal of Geomechanics, 15(1),
04014035.
Drescher, A., and Detournay, E. (1993). “Limit load in translational failure mechanisms for
associative and non-associative materials.” Geotechnique, 43(3), 443–456.
Ghosh, P. (2008). “Upper bound solutions of bearing capacity of strip footing by pseudo-
dynamic approach.” Acta Geotechnica, 3(2), 115-123.
Ghosh, P., and Choudhury, D. (2011). “Seismic bearing capacity factors for shallow strip
footings by pseudo-dynamic approach.” Disaster Advances, 4(3), 34-42.
Halder, K., Chakraborty, D., and Dash, S. K. (2018). “Seismic bearing capacity of a strip footing
situated on soil slope using a non-associated flow rule in lower bound limit analysis.” In
Geotechnical Earthquake Engineering and Soil Dynamics V: Numerical Modeling and Soil
Structure Interaction Reston, VA: American Society of Civil Engineers, 454-463.
Halder, K., and Chakraborty, D. (2022). “Estimation of seismic active earth pressure on
reinforced retaining wall using lower bound limit analysis and modified pseudo-dynamic
method.” Geotextiles and Geomembranes. (Accepted In Press).

© ASCE
Geo-Congress 2023 GSP 341 622

Izadi, A., Foroutan Kalourazi, A., and Jamshidi Chenari, R. (2021). “Effect of roughness on
seismic bearing capacity of shallow foundations near slopes using the lower bound finite
element method.” International Journal of Geomechanics, 21(3), 06020043.
Kumar, J., and Mohan Rao, V. B. K. (2003). “Seismic bearing capacity of foundations on
slopes.” Geotechnique, 53(3), 347-361.
Makrodimopoulos, A., and Martin, C. (2006). “Lower bound limit analysis of cohesive‐frictional
materials using second‐order cone programming.” International Journal for Numerical
Methods in Engineering, 66(4), 604-634.
Nadgouda, K., and Choudhury, D. (2021). “Seismic bearing capacity factor Nγe for dry sand
beneath strip footing using modified pseudo-dynamic method with composite failure
surface.” International Journal of Geotechnical Engineering, 15(2), 171-180.
Pain, A., Choudhury, D., and Bhattacharyya, S. K. (2015). “Seismic stability of retaining wall–
soil sliding interaction using modified pseudo-dynamic method.” Geotechnique Letters, 5(1),
56-61.
Sloan, S. W. (1988). “Lower bound limit analysis using finite elements and linear
programming.” International Journal for Numerical and Analytical Methods in
Geomechanics, 12(1), 61-77.
Steedman, R. S., and Zeng, X. (1990). “The influence of phase on the calculation of pseudo-
static earth pressure on a retaining wall.” Geotechnique, 40(1), 103-112.
Yang, S., Leshchinsky, B., Cui, K., Zhang, F., and Gao, Y. (2021). “Influence of failure
mechanism on seismic bearing capacity factors for shallow foundations near slopes.”
Geotechnique, 71(7), 594-607.

© ASCE