Lateral Response of Hollow Circular Caisson Embedded

in Nonlinear Soils
Ramyasri Rachamadugu, Ph.D.1; and Gyan Vikash, Ph.D.2
Downloaded from ascelibrary.org by Indian Institute of Technology, Roorkee on 10/03/22. Copyright ASCE. For personal use only; all rights reserved.

Abstract: This paper presents the numerical analysis of the lateral response of hollow circular caisson embedded in nonlinear soils. The
caisson–soil system is modeled by a three-dimensional nonlinear finite-element method. The caisson is considered as a linear elastic
body, whereas the nonlinear behavior of the adjoining soil is defined by kinematic hardening-based multiyield surface plasticity model.
The interface between the caisson and adjoining soil is modeled by a zero thickness contact element which is defined by the constitutive
relationships capable of describing the relative sliding and the separation at the interface. The present study discusses the lateral response
of the hollow caisson, which consists of the lateral load versus lateral displacement (P–Y) curves, mechanism of deformation of the cais-
son–soil system, and deformation profile of the hollow caisson. It also presents the effect of the D/B ratio, relative stiffness of the cais-
son–soil system, and the effect of the vertical load on the lateral response of the hollow caisson embedded in nonlinear soils. The present
study indicates that the D/B ratio, relative stiffness of the caisson–soil system, and the vertical load have a significant effect on the lateral
response of the caisson–soil system. This study reveals that there occurs elastoplastic deformation during the lateral loading of the caisson-
–soil system. Further, a simplified transitional plasticity-based constitutive law is proposed to model the lateral response of a caisson–soil
system. The proposed model requires only three parameters, which can be calibrated using the lateral load versus lateral displacement
curve obtained from the finite-element analysis or the field test. Furthermore, mathematical relationships between the model parameters
and the material parameters defining the caisson–soil system are established using the least-squares regression analysis. DOI: 10.1061/
(ASCE)GM.1943-5622.0002265. © 2021 American Society of Civil Engineers.
Author keywords: Caisson–soil system; Interface modeling; Soil nonlinearity; Deformation mechanism.

Introduction Gerolymos 2016; Biswas and Choudhury 2020) investigated the

lateral and seismic response of caisson. Gerolymos and Gazetas
Caisson is a large size foundation, which is built deep into the (2006a) proposed a generalized multispring Winkler model to de-
ground to support heavy loads. It is the most common type of foun- termine the lateral response of rigid caisson embedded in linear
dation for major bridge piers and other structures founded on rivers soil. This model considers four different types of linear springs to
with an erodible bed. It is installed to a sufficient depth below the simulate the interaction between the caisson and the soil. These
maximum depth of scouring to support large axial and lateral loads. are lateral translational spring and lateral rotational spring distribu-
The hollow caisson is one of the types of caisson foundations most ted along the length of the caisson and concentrated translational
often used as a foundation for major road and railway bridges built and rotational springs at the base of the caisson. Gerolymos and
on rivers (Ponnuswamy 2008; Mondal et al. 2012; Jain 2016). Gazetas (2006b) further updated this model to incorporate soil
Open caisson, hollow caisson, and well foundation are the terms and interface nonlinearities. Varun et al. (2009) proposed a simple
that are typically used interchangeably to denote the hollow pris- model to obtain the lateral response of the caisson embedded in lin-
matic or cylindrical shaped foundation with a large diameter. The ear elastic soil. Mondal et al. (2012) performed finite-element anal-
top and the bottom ends of such foundations are kept open. A hol- ysis to study the seismic response of a well foundation enclosed by
low caisson is installed to a sufficient depth by the sinking process. cohesionless soil. In this study, the well–soil system was modeled
The top and the bottom ends of the foundation are plugged with as a two-dimensional system wherein the well foundation was con-
concrete, and the cylindrical shell is filled with locally available sidered as a rigid body which was modeled by the rigid beam ele-
sand after the foundation is driven to the desired depth. ment. A review of the literature reveals that the studies done in the
In the recent past, many researchers (Gerolymos and Gazetas past are primarily focused on the lateral and the seismic response of
2006a, b, c; Varun et al. 2009; Kumar and Rao 2012; Mondal the solid caissons. None of the earlier studies explicitly investigate
et al. 2012; Karapiperis and Gerolymos 2014; Zafeirakos and the lateral response of hollow caisson embedded in nonlinear soils.
All the studies done in the past assume that the caisson undergoes
1
Research Scholar, Dept. of Civil Engineering, School of Engineering, rigid body movement under the lateral load. The assumption that
Shiv Nadar Univ., Dadri, Uttar Pradesh 201314, India. Email: rr783@snu the caisson is a rigid body may be valid for solid caisson, but it
.edu.in may not apply to the hollow caisson. The available studies do
2
Associate Professor, Dept. of Civil Engineering, School of Engineer- not explicitly explain the deformation mechanism of a caisson–
ing, Shiv Nadar Univ., Dadri, Uttar Pradesh 201314, India (corresponding soil system subjected to the lateral load. The effect of the ratio of
author). Email: [email protected]
embedment depth and diameter (D/B) of the caisson, vertical
Note. This manuscript was submitted on August 21, 2020; approved
on October 11, 2021; published online on December 1, 2021. Discussion load, soil type, and the relative stiffness of the soil and the caisson
period open until May 1, 2022; separate discussions must be submitted on the lateral response of a hollow caisson are yet to be explored.
for individual papers. This paper is part of the International Journal of Considering these research gaps, a nonlinear finite-element
Geomechanics, © ASCE, ISSN 1532-3641. analysis of a hollow caisson embedded in nonlinear soil subjected

© ASCE 04021275-1 Int. J. Geomech.

Int. J. Geomech., 2022, 22(2): 04021275

 to the lateral load is performed in the present study. For this pur- Tables 1 and 2 are chosen by considering the range of the values
pose, a three-dimensional (3D) finite-element model of the cais- of these parameters given by Yang et al. (2008).
son–soil system is developed. The finite-element model
developed in this study considers the nonlinearities induced due
to the sliding and the separation at the interface between the caisson Numerical Modeling
and the soil, and the nonlinear behavior of the surrounding soil. The
caisson is modeled as a linear elastic body, and the adjoining non- In the present study, OpenSees and GiD software are used to nu-
linear soil is defined by the multiyield surface plasticity model. A merically model the caisson–soil system under the lateral loading
zero thickness contact element is used to model the interface be- condition. A detailed sensitivity analysis is performed to determine
tween the caisson and the soil. The contact element is defined by
the constitutive relationships which are capable of describing the
Downloaded from ascelibrary.org by Indian Institute of Technology, Roorkee on 10/03/22. Copyright ASCE. For personal use only; all rights reserved.

sliding and the separation at the caisson–soil interface. The present

study investigates the deformation mechanism of a laterally loaded
caisson–soil system and the factors affecting the lateral response of
the caisson–soil system by performing a series of finite-element
analyses by varying the D/B ratio, the soil types, and the parameters
defining the mechanical properties of the soils. In addition to this,
the effect of the vertical load applied on the caisson on the lateral
response of the caisson–soil system is also investigated. Based
on the physics of the deformation mechanism of the laterally loaded
caisson–soil system observed in this study, a transitional plasticity-
based constitutive relationship is proposed to model the lateral
response of the hollow caisson embedded in nonlinear soils. The
proposed model requires only three parameters that can be cali-
brated by a single lateral load versus lateral displacement curve.
These model parameters are calibrated with the results obtained
(a)
from the finite-element analysis done in the present study. Further,
relationships between the model parameters and the material pa-
rameters describing the soil and the caisson are established using
the least-squares regression analysis.

Numerical Modelling of the Caisson–Soil System

Fig. 1 shows the top view and the sectional view of the hollow cir-
cular caisson considered in the present study. The outer diameter of
the caisson is 12 m, and the inner diameter is 9 m. The hollow por-
tion of the caisson is filled with sand having a modulus of elasticity
equal to 50 MPa. The thickness of the bottom plug, the top plug,
and the top cap of the caisson are 5, 1, and 2 m, respectively.
The bottom plug of the caisson is generally a curved surface,
which is approximated in this study by a straight line. The bottom
and the top plugs are made of M25 grade concrete, whereas the top
cap and the steining are made of M35 grade concrete. The maxi-
mum depth of the scour (H ) is considered equal to 12 m from
the riverbed. The embedment depth (D) of the caisson is taken as (b)
0.5B, 1B, 1.5B, and 2B to study the effect of the D/B ratio on the
lateral response of the caisson. The minimum value of the embed- Fig. 1. Geometrical details of circular caisson considered in the present
ment depth (D) is chosen by considering the criterion given in IRC study: (a) plan view; and (b) sectional view.
78-2000 (IRC 2000) and Ponnuswamy (2008) to decide the mini-
mum value of the grip length. Fig. 1(b) shows the lateral load, P,
and the vertical load, V, applied on the caisson. The lateral load Table 1. Values of the parameters defining the mechanical properties of
the clayey soils
is applied at the top of the caisson, and the vertical load is applied
at the center of the well cap. The present study considers both the Soft Medium Stiff
clayey soil and the sandy soil as the embedment soil wherein the Parameters clay clay clay
caisson is embedded. To investigate the effect of the stiffness of Reference shear modulus (Gr, MPa) 13 60 150
the embedment soil on the lateral response of the caisson, three dif- Reference bulk modulus (Br, MPa) 65 300 750
ferent types of the clayey soil, namely, soft clay, medium clay, and Density (ρ, kg/m3) 1,300 1,500 1,800
stiff clay and three different types of sandy soil, namely, loose sand, Undrained cohesion (c, kPa) 18 37 75
medium sand, and dense sand are considered in this study. The ma- Shear strain at failure (γmax) 0.1 0.1 0.1
terial parameters and their value to define the mechanical behavior Reference mean effective confining 100 100 100
pressure (Pr, kPa)
of the embedment soils considered in the present study are listed in
Number of yield surfaces 20 20 20
Tables 1 and 2. The values of the material parameters shown in

© ASCE 04021275-2 Int. J. Geomech.

Int. J. Geomech., 2022, 22(2): 04021275

 Table 2. Values of the parameters defining the mechanical properties of radially varying elements whose size varies from 1.2 m × 2.6 m ×
the sandy soils 2 m to 36 m × 28 m × 2 m. The first stage of the sensitivity analysis
Type of ρ Φ ΦpT is executed by extending far end boundaries of the soil domain up
soil (kg/m3) ϑ Gr (kPa) (°) γmax D (°) co to 5B, 10B, 15B, and 20B from the outer surface of the foundation.
4 The total depth of the domain of interest is considered herein equal
Loose sand 1,700 0.34 5.5 × 10 29 0.1 0.5 29 0.21
to D + 3B. From the first stage of the sensitivity analysis, it was ob-
Medium 2,000 0.35 1 × 104 37 0.1 0.5 27 0.05
sand served that the results converge after 10B, which are almost the
Dense sand 2,100 0.37 2.5 × 105 40 0.1 0.5 27 0.02 same for 10B, 15B, and 20B. Hence, this study considers 15B as
the optimum soil domain extended on both sides of the foundation.
Note: ϑ = Poisson’s ratio; Φ = friction angle; d = pressure-dependent The second stage of the analysis is performed by considering three
coefficient; ΦpT = phase transformation angle; and co = non-negative
different depths of the element, which are 0.5, 1, and 2 m, respec-
constant defining the rate of shear-induced decrease in volume.
Downloaded from ascelibrary.org by Indian Institute of Technology, Roorkee on 10/03/22. Copyright ASCE. For personal use only; all rights reserved.

tively, and keeping the width of the element the same as those of
the first stage of the sensitivity analysis. This stage of the sensitivity
the optimum size of the domain and the optimum size of the ele- analysis indicates that the results for the element with depths 1 and
ment to discretize the domain of interest. In order to perform the 2 m are nearly the same. Therefore, the optimum depth of the ele-
sensitivity analysis, boundary conditions consistent with the phys- ment is considered equal to 2 m. Fig. 2 shows the geometrical de-
ics of the problem, suitable constitutive relationships to define the tails of the 3D finite-element model of the circular caisson obtained
mechanical behavior of the soil and the caisson, and constitutive from the present sensitivity analysis.
law to define the interface between the caisson and the soil are
needed to define. These are classified herein as geometrical model-
ing, material modeling, and interface modeling, respectively, which Material Modeling
are described in the subsequent sections. The present study defines the surrounding soil using kinematic
hardening multiyield surface elastoplastic constitutive model,
Geometrical Modeling which was proposed by Iwan (1967), Mroz (1967), and Prevost
(1985). This model was further modified by Parra-Colmenares
In the present study, a 3D finite-element model of the caisson–soil (1996) and Yang (2000), and it was added into the OpenSees ma-
system is developed by considering interface and material nonlin- terial model library. The pressure-independent multiyield surface
earities. For geometrical modeling, a caisson–soil system is consid- plasticity model, which considers the von Mises criterion to define
ered as a combination of two units. The first unit is the hollow its yield surface and failure surface, was used for clayey soil;
caisson filled with sandy soil, and the second unit is the surround- whereas the pressure-dependent multiyield surface plasticity
ing soil medium. These two units are connected by zero thickness model, which considers the Druger–Prager criterion to define its
interface elements. Both the optimum size of the soil domain, yield surface and failure surface, was used for sandy soil. A detailed
wherein caisson is embedded, and the optimum size of the element description of these models is given by Parra-Colmenares (1996)
to discretize the domain of interest are unknown in the present and Yang (2000). The parameters defining the material properties
problem, and these unknowns have to be determined by performing and their value for different types of soil considered in this study
the sensitivity analysis. The sensitivity analysis herein is performed are given in Tables 1 and 2, respectively. Caisson and the soil filled
in two stages to determine these unknowns. The first stage of the in the hollow portion of the caisson are considered as a linear elastic
sensitivity analysis is done to determine the optimum size of the do- body, which is defined by the isotropic elastic material model avail-
main of the soil medium. It is done with the presumed mesh size for able in OpenSees. The isotropic elastic material model requires
the caisson as well as for the surrounding soil domain. The second three model parameters, namely, modulus of elasticity (E), Pois-
stage of the sensitivity analysis is done to determine the optimum son’s ratio (ϑ), and density (ρ). Table 3 shows the value of E, ϑ,
size of the element to discretize the domain of interest, which is and ρ considered for the caisson and the filled sand. The caisson
done with the optimum size of the soil domain obtained from the is made of reinforced cement concrete whose mechanical property
first stage of the sensitivity analysis. The second stage of the sensi- is defined by its characteristic strength ( fck). As the material model
tivity analysis is performed by varying the depth of the element and requires E as an input parameter, it is needed to determine the value
keeping the width of the element for both the caisson and the soil of E for given fck. The value of E is determined by the formula
domain the same as those of the first stage of the sensitivity analy- given in IS 456 (IS 2000), which relates E and fck as
sis. For both the stages, boundary conditions were defined by pro- 




viding roller support at the far end of the sides of the soil domain E = 5,000 × fck (1)
and fixed support at the bottom end of the soil domain. Caisson
is defined by a linear elastic material model, whereas the surround-
ing soil is modeled using the nonlinear elastoplastic material Interface Modeling
model. Details of the material models are described in the subse- The present study uses a zero thickness contact element to model
quent section. Both the caisson and the surrounding soil are discre- the interface between the adjoining soil and the caisson. It has
tized using eight-noded brick elements; caisson is discretized using two nodes, which are termed master nodes and slave nodes. The
uniform meshing, whereas the soil domain is discretized using non- master node of the contact element is connected with the caisson,
uniform mesh. Nonuniform meshing for the soil domain is done to and the slave node is connected with the adjoining soil. The degree
get an in-depth understanding of the mechanism of deformation of freedom of the nodes of the caisson and the adjoining soil
near the interface between the soil and the caisson. connected with the contact element is kept the same to satisfy the
The caisson with a D/B ratio of 0.5 embedded in soft clay with displacement compatibility condition at these nodes. The elastic–
the lateral load, P, equal to 1 MN is considered for the sensitivity perfectly plastic gap material model and the elastic–perfectly
analysis. The first stage of the sensitivity analysis is performed plastic material model are considered to define the separation and
by discretizing the caisson with the element of size 1.2 m × the sliding mechanism at the caisson–soil interface, respectively.
0.75 m × 2 m. The soil domain is discretized by nonuniform Details of these models can be found in the OpenSees manual

© ASCE 04021275-3 Int. J. Geomech.

Int. J. Geomech., 2022, 22(2): 04021275

Downloaded from ascelibrary.org by Indian Institute of Technology, Roorkee on 10/03/22. Copyright ASCE. For personal use only; all rights reserved.

Fig. 2. Three-dimensional (3D) finite-element model of hollow circular caisson in nonlinear soil.

Table 3. Mechanical properties of the caisson Table 4. Iterations performed in the penalty method
Materials E (GPa) ϑ ρ (kg/m ) 3
Iterations Normal stiffness (KN) Tangential stiffness (KT)
M35 grade concrete 29.5 0.15 2,500 1 Es 0.01Es
M25 grade concrete 25 0.15 2,500 2 10Es 0.1Es
Medium sand 0.05 0.3 1,900 3 100Es Es
4 200Es 2Es
5 100Es 10Es
(McKenna and Fenves 2001). The elastic–perfectly plastic gap ma- 6 100Es 100Es
terial model is defined by only one parameter, KN, which is the nor- Note: Es = modulus of surrounding soil.
mal stiffness of the contact element in the compressive condition.
The elastic–perfectly plastic material model, which is considered
to define the sliding mechanism at the interface, requires two pa- Validation of the Proposed Numerical Approach
rameters, namely, tangential stiffness, KT, and failure stress, Fy.
Thus, three parameters, KN, KT, and Fy, are required to define The finite-element model developed in the present study is validated
completely the separation and sliding mechanism at the interface. with the experimental studies available in the literature. The experi-
Because KN is the normal stiffness of the contact element and KT mental studies were done by Anagnostopoulos and Georgiadis
is the tangential stiffness of the contact element, these are assigned (1993), and Kumar and Rao (2012) are considered for the validation
perpendicular to the plane of interface and tangential to the plane of purpose. Anagnostopoulos and Georgiadis (1993) performed the lab-
the interface, respectively. The penalty method proposed by oratory tests on the aluminum closed-ended piles driven in the soft
Zienkiewicz (1974) was used to determine the value of KN and clay bed subjected to the vertical load as well as the combined loading
KT. According to the penalty method, a little amount of relative (both vertical and horizontal). Kumar and Rao (2012) experimentally
investigated the lateral response of caissons by performing a series of
penetration between the bodies in contact is needed to allow to
experiments on the model caissons driven in marine clay subjected to
avoid numerical instability while determining the optimum value
the lateral load. The detailed description of these experimental studies
of KN. The present study considers 0.2 mm as the tolerance limit
and the material properties can be found in Anagnostopoulos and
for the relative penetration to estimate the optimum value of KN.
Georgiadis (1993) and Kumar and Rao (2012). A numerical simula-
The value of KN and KT considered in the successive iterations is
tion of the experimental studies done by Anagnostopoulos and
given in Table 4. This table depicts that the optimum value of KN
Georgiadis (1993), and Kumar and Rao (2012) are performed in
and KT can be considered equal to 100 times the elastic modulus
the present study to validate the proposed numerical scheme. Fig. 3
of the adjoining soil (Es) as no significant variation in the result
presents the comparison of the lateral response of the caisson ob-
is observed after the fourth iteration.
tained experimentally and that obtained from the numerical approach
The remaining third parameter Fy for clay was determined using
considered in the present study. This figure depicts that there is a good
von Misses failure criteria, which defines Fy as Fy = √3Su; where
agreement between the experimentally observed lateral response of
Su is the undrained strength, which is equal to undrained cohesion
the caissons and that predicted from the present numerical analysis.
(c) for saturated clay. For sandy soil, Fy was determined using the
Mohr–Columb failure criterion, which defines Fy as Fy = μtN;
where μ is the frictional coefficient and tN is the normal component
of the traction at the interface. The value of μ was considered herein Results and Discussion
equal to tan (δ), where δ is the angle of friction between the caisson
and adjoining soil, which is determined as δ = 2/3Φ (where Φ rep- The hollow circular caissons with D/B ratios 0.5, 1, 1.5, and 2 are
resents the angle of friction of soil). considered to understand the effect of the D/B ratio on the lateral

© ASCE 04021275-4 Int. J. Geomech.

Int. J. Geomech., 2022, 22(2): 04021275

Downloaded from ascelibrary.org by Indian Institute of Technology, Roorkee on 10/03/22. Copyright ASCE. For personal use only; all rights reserved.

(a) (a)

(b) (b)

(c) (c)

Fig. 3. Validation of the proposed numerical approach with (a) Kumar Fig. 4. Lateral response of the hollow caisson embedded in (a) soft
and Rao (2012) and Anagnostopoulos and Georgiadis (1993); (b) ver- clay; (b) medium clay; and (c) stiff clay.
tical response; and (c) lateral response.

observations can also be made from Figs. 5(a–c), which present

response of the caisson–soil system. The present study considers the lateral load versus lateral displacement curve of the caisson em-
both clayey soils and sandy soils as the soil bed wherein the caisson bedded in sandy soils. Fig. 6 shows the general trend of the lateral
is embedded. The lateral load, P, is applied at the top of the caisson load versus lateral displacement curve of the hollow circular cais-
incrementally with an increment of 1 MN till the lateral displace- son driven in nonlinear soils. It is clear from this figure that the ini-
ment of the caisson at the ground level reaches 0.15B. The magni- tial portion and the later portion of the load–displacement curve are
tude of the ultimate lateral displacement is chosen equal to 0.15B somewhat linear, as indicated by the range OA and BC, respec-
by considering the suggestion given by Broms (1964). In the pre- tively. A zone of nonlinearity (AB) can be observed between
sent numerical analysis, it is observed that the complete mechanism these two linear portions where a smooth transition of the curve
of deformation of the laterally loaded caisson–soil system can be from the initial linear portion to the later linear portion occurs.
obtained by allowing the ground level lateral displacement equal Fig. 6 depicts that point A is the point where the deviation of the
to 0.15B. curve from the linear part of the curve begins, and point B is the
point from which the intermediate nonlinear part of the curve be-
comes asymptotic to the later linear part of the curve. The subse-
Lateral Load–Displacement Curves
quent section describes the deformation mechanism of the
Fig. 4 depicts the lateral load versus lateral displacement curve of caisson–soil system in these three zones, namely, OA, AB, and
the hollow circular caissons embedded in clayey soils. It can be ob- BC. It is important to note from Figs. 4 and 5 that the initial linear
served from Figs. 4(a–c) that the lateral load–lateral displacement portion of the load–displacement curves for a given soil type is
curve is nonlinear for all the cases of the caissons irrespective of nearly the same for all the D/B ratios considered herein. This obser-
the stiffness of the soil wherein it is embedded. Similar vation indicates that the D/B ratio has an insignificant effect on the

© ASCE 04021275-5 Int. J. Geomech.

Int. J. Geomech., 2022, 22(2): 04021275

Downloaded from ascelibrary.org by Indian Institute of Technology, Roorkee on 10/03/22. Copyright ASCE. For personal use only; all rights reserved.

(a)
Fig. 7. Effect of the stiffness of the adjoining soil on the lateral re-
sponse of the hollow caisson embedded in clayey soils and sandy soils.

in the D/B ratio. Fig. 7 depicts the effect of the stiffness of the ad-
joining soil on the load–displacement curve of the hollow caisson
enclosed with the clayey soils and the sandy soils, respectively.
In these figures, the lateral load at the point of intersection of the
extended part of the initial and the later linear portions of the
curve (PD, as shown in Fig. 6) is considered to observe the effect
of the stiffness of the adjoining soil on the lateral response of the
caisson. It can be observed from Fig. 7 that, for all the values of
(b) the D/B ratio, the magnitude of PD increases as the stiffness of
the adjoining soil increases. This observation indicates that the
stiffness of the caisson–soil system increases as the stiffness of
the adjoining soil increases.

Mechanism of Deformation of the Caisson–Soil System

Fig. 6 illustrates that there exist three zones, namely, initial linear
zone (OA), intermediate transitional zone (AB), and later linear
zone (BC), in the lateral load versus lateral displacement curve of
the hollow caisson embedded in the nonlinear soils. To observe
the deformation mechanism of the caisson–soil system during the
lateral loading, three points on the load–displacement curve are
considered. These points are point 1, which is within zone OA;
(c) point 2, which is within zone AB; and point 3, which is within
zone BC. Fig. 8 depicts the contour of the deformation field at
Fig. 5. Lateral response of the hollow caisson embedded in (a) loose
points 1, 2, and 3 of the caissons with a D/B ratio of 2 embedded
sand; (b) medium sand; and (c) dense sand.
in medium clay. This figure depicts that at point 1 whole soil do-
main is influenced during the lateral loading. The zone of influence
seems to be decreased as the load–displacement curve reaches point
2. At point 3, the zone of influence appears to be get localized near
the interface between the caisson and the adjoining soil. Similar ob-
servations have been observed for all the cases considered in the
present study. This observation indicates that the deformation
near the interface between the caisson and the adjoining soil be-
comes predominant because the load–displacement curve moves
away from the initial linear part of the curve. Fig. 9 shows a closer
view of the deformation field of the caisson–soil system. This fig-
ure indicates that the separation between the caisson and the soil at
the interface does not occur at point 1. Whereas, separation on the
left side of the caisson is quite noticeable at point 2. At point 3, sep-
aration on the left side as well as at the base of the caisson can be
Fig. 6. General trend of the lateral response of the caisson–soil system. observed from Fig. 9. Separation at the sides of the caisson (Δh)
measured at the ground surface and the separation at the base of
the foundation (Δv) (as shown in Fig. 9) are considered to further
initial stiffness of the caisson–soil system subjected to the lateral analyze the mechanism of deformation of the caisson–soil system
load. However, Figs. 4 and 5 indicate that the influence of the during the lateral loading. Figs. 10 and 11 present the evolution
D/B ratio becomes more prominent at the later part of the curve of Δh and Δv along the load–displacement curve of the caissons em-
as the stiffness of the caisson–soil system increases with an increase bedded in the medium clay and the medium sand, respectively.

© ASCE 04021275-6 Int. J. Geomech.

Int. J. Geomech., 2022, 22(2): 04021275

Downloaded from ascelibrary.org by Indian Institute of Technology, Roorkee on 10/03/22. Copyright ASCE. For personal use only; all rights reserved.

Fig. 8. Contour plot of the deformation field at points 1, 2, and 3 of the caissons with a D/B ratio of 2 embedded in medium clay.

Fig. 9. Closer view of the deformation field of the caisson–soil system. Fig. 10. Variation of Δh and Δv with the lateral load of the caisson
embedded in the medium clay.

These figures illustrate that both Δh and Δv are very small, which
can be reasonably considered to be negligible, in the initial linear irrecoverable deformation because of the relative sliding and the
part of the load–displacement curve. The magnitude of Δh increases separation at the interface begins after this point.
along the intermediate nonlinear part of the curve and it becomes
significant in the later linear part of the load–displacement curve.
Deformation Profile of the Caisson Subjected
From Figs. 10 and 11, it can be readily observed that the magnitude
to the Lateral Load
of the separation at the base, Δv, shows an increasing trend with the
lateral load along the P–Y curve beyond the initial linear part of the Figs. 12 and 13 depict the deformation profile of the caissons
curve; however, its magnitude is very less in comparison to Δh. It is driven in the clayey soils and the sandy soils, respectively. These
important to note that the similar trend of the evolution of Δh and figures show the deformation profile of the caissons corresponding
Δv has been observed for the caissons embedded in soft clay, stiff to the ground level lateral displacement equal to 0.05 m. These
clay, loose sand, and dense sand also. It can be therefore inferred plots have been made between the nondimensional axes Y/Ymax
from these observations that the lateral response of the caisson– and z/D, where Y is the lateral displacement, Ymax is the lateral dis-
soil system is linear and elastic in the initial part of the curve and placement at the ground level, z is the spatial location along with
it is plastic along the P–Y curve beyond the initial linear part of the depth of the foundation, and D is the embedment depth.
the curve. This study reveals that the mechanism of failure of the From Fig. 12, it can be observed that the caissons driven in the
caisson–soil system is predominately governed by the relative slid- soft clay rotate as a rigid body irrespective of the D/B ratio.
ing and the separation at the interface between the caisson and the However, it is evident from this figure that there occurs bending de-
adjoining soil. The experimental observations made by Kumar and formation in the caissons driven in the medium clay and the stiff
Rao (2012) reinforce this finding as they observed that the separa- clay. It can also be observed from Fig. 12 that the curvature of
tion between the caisson and the soil at the ground surface increases the bending deformation of the caisson increases as the D/B ratio
as the lateral load increases, which becomes significant at the ulti- of the caisson increases. Similar observations can also be made
mate value of the lateral load. Based on these observations, it can be for the caissons driven in the sandy soils, which can be observed
suggested that the caisson–soil system should be designed for the from Fig. 13. Further, the effect of the stiffness of the adjoining
allowable lateral load less than the lateral load at point A as the soil on the deformation profile of the caisson can also be observed

© ASCE 04021275-7 Int. J. Geomech.

Int. J. Geomech., 2022, 22(2): 04021275

Downloaded from ascelibrary.org by Indian Institute of Technology, Roorkee on 10/03/22. Copyright ASCE. For personal use only; all rights reserved.

Fig. 11. Variation of Δh and Δv with the lateral load of the caisson Fig. 14. Lateral load–lateral displacement at the base curve of the cais-
embedded in the medium sand. sons for different values of D/B ratio enclosed by medium clay and
medium sand.

Location of the Point of Rotation

This section analyzes the location of the point of rotation of the lat-
erally loaded caisson, and the effect of the D/B ratio, and the relative
stiffness of the caisson–soil system on the location of the point of
rotation. For this purpose, variation of the lateral movement of the
base of the caisson, Yb, with the lateral load, P, with due consider-
ation of the effect of the D/B ratio and the relative stiffness of the
caisson–soil system on the P–Yb curve has been analyzed. Fig. 14
presents the P–Yb curve of the caissons driven in the medium clay
and the medium sand, respectively. These figures illustrate that Yb
is positive throughout the lateral loading and its magnitude in-
creases as the magnitude of the lateral load increases. It is to be
Fig. 12. Deformation profile of the hollow caisson driven in clayey noted that a similar observation is made for the caissons driven in
soils. the soft clay, stiff clay, loose sand, and dense sand also. This indi-
cates that the base of the caisson moves in the lateral direction to-
ward the right side during the lateral loading. Thereby, the
location of the point of rotation of the caisson must fall below the
base of the caisson. It is generally assumed that the point of rotation
of the caisson either locates at the base of the caisson or above the
base of the caisson; in contrast to this, the present study reveals
that the location of the point of rotation of the laterally loaded cais-
son falls below the base of the caisson. It is evident from Fig. 14 that
there is a significant effect of the D/B ratio on the P–Yb curve as the
magnitude of Yb for a given lateral load decreases with an increase in
the value of the D/B ratio. Fig. 15 illustrates the effect of the stiffness
of the adjoining soil on the P–Yb curve of the caisson–soil system
with a D/B ratio of 2 driven in the nonlinear soils considered herein.
It can be observed from this figure that the magnitude of Yb for a
given lateral load decreases as the stiffness of the adjoining soil in-
creases. Variation in the magnitude of Yb with the lateral load as ob-
Fig. 13. Deformation profile of the hollow caisson driven in sandy served in Figs. 14 and 15 indicates that the location of the point of
soils. rotation of the caisson varies with the magnitude of the lateral load.
It is generally assumed that the location of the point of rotation of the
from Figs. 12 and 13. These figures illustrate that the curvature of caisson is fixed during the lateral loading irrespective of the amount
the deformed shape of the caisson increases with an increase in the of the lateral load, D/B ratio, and the relative stiffness of the caisson–
stiffness of the adjoining soil. Generally, the caisson subjected to soil system. However, the present study indicates that the point of
the lateral load is considered to behave as a rigid body. However, rotation of a laterally loaded caisson varies during the lateral loading
the observations made in this study indicate that the laterally loaded and its location depends on the magnitude of the lateral load, D/B
caisson does not necessarily always behaves as a rigid body; ratio, and the relative stiffness of the caisson–soil system.
whether the caisson behaves as a rigid body or flexible body is pre-
dominately depends on its D/B ratio and the relative stiffness of the
Effect of the Vertical Load on the Lateral Response
caisson–soil system. Thus, modeling the laterally loaded caisson as
of the Caisson
a rigid body, which has been done in the past (IRC 45-1972, IRC
1972; IRC 78-1983, IRC 1983; Mondal et al. 2012), is not appro- A numerical analysis of the effect of the vertical load on the lateral
priate. It is, therefore, preferable to model the caisson by an element response of the caisson is presented in this section. The response of
that can define its flexible response. the caisson under the combined loading is investigated in the two

© ASCE 04021275-8 Int. J. Geomech.

Int. J. Geomech., 2022, 22(2): 04021275

Downloaded from ascelibrary.org by Indian Institute of Technology, Roorkee on 10/03/22. Copyright ASCE. For personal use only; all rights reserved.

Fig. 15. Lateral load–lateral displacement at the base of the hollow Fig. 17. Effect of the vertical load on the lateral response of the caisson
caisson with a D/B ratio of 2 embedded in clayey soils and sandy soils. with a D/B ratio of 2 enclosed by medium clay and medium sand.

response of the caisson embedded in clayey soils is insignificant,

whereas the lateral stiffness of the caisson embedded in sandy
soil is decreased due to the presence of the vertical load.

Model to Predict P–Y Curves

A typical plot showing the general trend of the P–Y curve of a cais-
son embedded in nonlinear soil was exhibited in Fig. 6. This figure
elucidated that there exist three zones in the P–Y curve, namely, ini-
tial linear zone, transitional zone, and later linear zone. It was ob-
served that the elastic deformation was predominant in the initial
linear zone as an insignificant amount of relative sliding and sepa-
ration at the interface was observed in this zone. The contribution
Fig. 16. Vertical response of the caisson with a D/B ratio of 2 enclosed of the relative sliding and separation to the total deformation was
by medium clay and medium sand. observed to increase as the curve moved toward the transitional
zone; the relative sliding and separation at the interface eventually
became predominant in the later portion of the P–Y curve. Consid-
stages. In the first stage, the vertical response of the caisson is an- ering this phenomenon, a transitional plasticity-based constitutive
alyzed, which yields a vertical load (V ) versus vertical displace- model is proposed herein to model the lateral response of a hollow
ment curve. The allowable vertical load (Vapp), which is to be caisson embedded in nonlinear soil. The proposed model considers
applied on the caisson, is then estimated from the vertical response that both elastic and plastic deformations develop together during
of the caisson obtained from the first stage. In the second stage, a the lateral loading. It is hypothesized that the plastic deformation
lateral load in addition to Vapp is applied incrementally at the top evolved because of the relative sliding and the separation at the in-
of the caisson to determine its response under the combined load- terface, although to an extent only, has possibility from the begin-
ing. Fig. 16 presents the vertical response of the caisson with a ning of the lateral loading. Thus, incremental variation in the lateral
D/B ratio of 2 driven in the medium clay and the medium sand, re- displacement (dY) due to given lateral load increment (dP) can be
spectively. This figure illustrates that the vertical load versus verti- expressed as a summation of the elastic component of deformation
cal displacement curve is linear in the beginning and then it (dY e) and the plastic component of deformation (dY P), which is
becomes nonlinear. The vertical load after that the curve becomes written as follows:
nonlinear is considered as Vapp. Fig. 17 presents the effect of the
vertical load on the lateral response of the caisson with a D/B dY = dY e + dY p (2)
ratio of 2 driven in the medium clay and medium sand, respec-
Considering linearity of the initial portion of the P–Y curve
tively. The response of the caisson under the combined loading (dY e) for given lateral load increment (dP) is defined as
as shown in Fig. 17 is done by first applying Vapp on the caisson
and letting the caisson displace under the vertical load Vapp. There- 1
dY e = dP (3)
after, the lateral response of the caisson is obtained by considering Ki
the vertical displacement at Vapp as the initial state for it. Fig. 17
depicts that the vertical load does not have a significant effect on where Ki = slope of the tangent drawn on the initial portion of the
the lateral response of the caisson embedded in the medium clayey P–Y curve, as shown in Fig. 18. Further, considering the transi-
soil. This figure shows that there is an almost insignificant effect of tional behavior of the P–Y curve from the initial linear portion to
the vertical load on the initial portion of the curve of the combined the later linear portion of the curve, dY P for given lateral load incre-
loading of the caisson driven in the medium sandy soil; however, ment (dP) is defined as
  
beyond the initial portion of the curve, the stiffness of the cais- 1 1 δ γ
son–soil system in the lateral direction decreases on the application dY =P
− 1− dP (4)
Ku Ki δmax
of the vertical load. A similar trend is also observed for the caissons
with all the D/B ratios and the soil types considered herein. It is in- where Ku = slope of the tangent drawn on the later linear portion of
teresting to note that the influence of the vertical load on the lateral the curve; δmax = distance measured from the beginning of the

© ASCE 04021275-9 Int. J. Geomech.

Int. J. Geomech., 2022, 22(2): 04021275

Downloaded from ascelibrary.org by Indian Institute of Technology, Roorkee on 10/03/22. Copyright ASCE. For personal use only; all rights reserved.

(a)

Fig. 18. Schematic representation of the model parameters and the

state variables of the proposed model.

curve to the point where the extended later linear part of the curve
meets of the P-axis; δ = distance measured from the current state
of the load (P) to its corresponding load on the extended later linear
part of the curve; and γ = transitional parameter introduced to control
the transition of the curve from the initial linear part to the later linear
part of the curve. Ku, δmax, and δ are shown explicitly in Fig. 18.
Using Eqs. (3) and (4), dY for given lateral load increment dP
can be written as
(b)
   
1 1 1 δ γ
dY = + − 1− dP (5) Fig. 19. Variation of (a) the tangential stiffness with the lateral load;
Ki Ku Ki δmax and (b) lateral load–lateral displacement curve for different values of γ.
Analysis of Eq. (5) elucidates that the plastic component of the
deformation, dY P, is almost negligible in the initial zone where the
Table 5. Values of the model parameters (Ki, Ku, and γ) calibrated from the
lateral load begins to apply as δ tends to δmax in the initial zone of finite-element analysis result of the caissons embedded in clayey soils
the curve. Therefore, in the initial zone of the curve,
Soil type D/B Ki (MN/m) Ku (MN/m) γ
lim dY P ≅ 0 (6a)
δ
δmax Soft clay 0.5 476.81 31.75 13
1 550 70 10
which results in 1.5 559.54 90 8.75
2 600 170 7.75
1
lim dY ≅ dP (6b) Medium clay 0.5 2,526.01 100 9
δ
δmax Ki
1 2,692 225 7
The contribution of the plastic component of the deformation, 1.5 2,743.39 400 5.9
dY P, to the total deformation, dY, is maximum in the later linear 2 2,898.18 520 5.4
part of the P–Y curve as δ approaches zero in this zone. Therefore, Stiff clay 0.5 4,892.07 320 7
in the later linear part of the curve, 1 5,567.25 750 5.3
  1.5 5,896.19 1,200 4.7
1 1
lim dY P ≅ − dP (6c) 2 5,917.09 1,800 4
δ
0 Ku Ki
As a result of Eq. (6c), dY in the later linear part can be ex-
Table 6. Values of the model parameters (Ki, Ku, and γ) calibrated from the
pressed as
finite-element analysis result of the caissons embedded in sandy soils
1 γ
lim dY ≅ dP (6d) Soil type D/B Ki (MN/m) Ku (MN/m)
δ
0 Ku
Loose sand 0.5 1,800 40 9
Using Eq. (5), overall tangential stiffness of the caisson–soil 1 2,000 75 7
system, K, can be written as 1.5 2,100 190 6
2 1,800 40 5
1
K =    (7) Medium sand 0.5 3,550.53 150 8
1 1 1 δ γ
+ − 1− 1 3,600 420 5.9
Ki Ku Ki δmax 1.5 3,750 500 5.3
2 4,140.37 850 4.75
Fig. 19(a) shows the variation of K with the lateral load for dif-
ferent values of γ. The plot shown in Fig. 19(a) was drawn by as- Dense sand 0.5 7,984.37 600 6
suming the value of Ki, Ku, and PU equal to 2,000 kN/m, 200 kN/ 1 8,360.34 1,100 4.8
1.5 8,565.76 1,500 4
m, and 250 kN, respectively, where PU is the intercept of the later
2 8,844.45 2,000 3.6
linear part of the curve on the P-axis. This figure illustrates that the

© ASCE 04021275-10 Int. J. Geomech.

Int. J. Geomech., 2022, 22(2): 04021275

Downloaded from ascelibrary.org by Indian Institute of Technology, Roorkee on 10/03/22. Copyright ASCE. For personal use only; all rights reserved.

(a)

(b)

Fig. 20. Comparison of the load–displacement curves obtained from the finite-element analysis to those obtained from the proposed model for the
caisson embedded in (a) clayey soils; and (b) sandy soils.

tangential stiffness of a caisson–soil system, K, is bounded between Calibration of Model Parameters and Model Prediction
Ki and Ku. It can be observed from Fig. 19(a) that K varies smoothly
The proposed model requires three model parameters, namely, Ki,
from Ki to Ku for the lower value of γ; for the higher value of γ, K is
Ku, and γ, to model the nonlinear trend of the P–Y curve. Ki can be
nearly equal to Ki till the point of intersection of the extended por- determined as the slope of the tangent drawn on the initial part of
tion of the initial and the later linear parts of the curve thereafter it the curve; Ku can be determined as the slope of the tangent
becomes almost equal to Ku. The sudden variation in the magnitude drawn on the later part of the curve, and γ can be estimated by try-
of K for the higher value of γ indicates that the proposed model with ing several values of γ and finding its value which best fits the P–Y
a higher value of γ yields a bilinear P–Y curve. The P–Y curves cor- curve. Besides these model parameters, two state variables, δmax
responding to Ki, Ku, PU, and γ previously considered are shown and δ, are required to predict the P–Y curve using the proposed
in Fig. 19(b), which shows that the P–Y curve is a smoothly varying model. Determination of δmax and δ requires mathematical equation
nonlinear curve for the lower value of γ, whereas it is the bilinear of the later linear part of the curve, which can be written as
curve for the higher value of γ. Thus, it can be inferred that the tran-
sitional parameter, γ, introduced in the proposed model has the po- P u = P U + Ku Y (8)
tential to model smooth transition of the curve from the initial linear
part to the later linear part of the curve as well as to model the bi- PU is the intercept of the straight line defined by Eq. (8) on the
linear trend of the P–Y curve. P-axis that yields δmax. δ is defined as the difference of the current

© ASCE 04021275-11 Int. J. Geomech.

Int. J. Geomech., 2022, 22(2): 04021275

 Table 7. Values of the relative stiffness of the caisson–soil systems (Ec/Es)
considered in the present study
Soil Soft Medium Stiff Loose Medium Dense
type clay clay clay sand sand sand
Ec
875.16 189.62 75.85 200.68 109.56 43.18
Es

curve obtained using the proposed model with that obtained by

Karapiperis and Gerolymos (2014). This figure indicates that the
P–Y curve obtained from the numerical analysis is predicted well
Downloaded from ascelibrary.org by Indian Institute of Technology, Roorkee on 10/03/22. Copyright ASCE. For personal use only; all rights reserved.

using the proposed model because the result predicted from the pro-
posed model differs maximum by 2% from the results stated by
Karapiperis and Gerolymos (2014).
Fig. 21. Comparison of the load–displacement curve obtained from the Further, an attempt has been made to obtain the relationship of
proposed model with that stated by Karapiperis and Gerolymos (2014). the model parameters (Ki, Ku, and γ) with the D/B ratio and the
material parameters describing the soil and the caisson. Fig. 22(a)
presents the variation of Ki with the relative stiffness of the caisson-
–soil system (Ec/Es) for the D/B ratio equals 2. Ec is the modulus of
elasticity of concrete M35 grade of which caisson is made of, and
Es is the modulus of elasticity of the adjoining soil. Table 7 presents
the value of Ec/Es corresponding to different adjoining soils consid-
ered in this study. Fig. 22(b) depicts the variation of Ki with the D/B
ratio for a given value of Ec/Es. From Figs. 22(a and b), it can be
observed that the value of Ki decreases with an increase in the
value Ec/Es, whereas the value of Ki increases with the increase
in the value of the D/B ratio. Therefore, considering the trend of
variation of Ki with Ec/Es and D/B ratio, the following mathemati-
cal expression is considered to relate Ki with Ec/Es and D/B ratio:
 b  −c
(a) D Ec
Ki = a (9)
B Es
There constants, namely, a, b and c, are introduced in the pre-
ceding equation, which has been determined by least-squares re-
gression analysis. To maintain consistency in the units, a
dimensionless parameter, which is the ratio of Ki to EcB, is defined
which is as follows:
 0.1  −0.8
Ki D Ec
= 0.46 (10)
EC B B Es
Following the method considered to obtain the relationship of Ki
with D/B ratio and Ec/Es, the relationship for Ku in terms of the D/B
(b) ratio, undrained cohesion (c) and coefficient of friction (μ) are es-
tablished. The expression for Ku for the caisson embedded in
Fig. 22. Variation of the estimated values of Ki: (a) with the relative clayey soil and sandy soil are given in the following equations:
stiffness of the caisson–soil system (Ec/Es) with a D/B ratio of 2; and  1.3
Ku D
(b) with the D/B ratios for the caisson embedded in medium clay. = 4,000 (c)1.6 (11a)
B B
load and its corresponding load on the straight line defined by
 0.79
Eq. (8) at the given value of the lateral displacement, Y. Therefore, Ku D
δ can be determined as, δ = PU−P. Tables 5 and 6 present the value = 0.0097 (μ)6.41 (11b)
Ec B B
of the model parameters (Ki, Ku, and γ) for the caissons embedded
in the clayey soils and the sandy soils, respectively. Fig. 20 com- Similarly, the relationship of Υ with D/B and Ec/Es is obtained,
pares P–Y curves predicted using the proposed model with those which as follows:
obtained from finite-element analysis. It can be observed from  −0.37  0.25
D Ec
these figures that the proposed model predicts well the P–Y re- γ = 1.83 (12)
sponse of caissons embedded in nonlinear soils. The proposed B Es
model is further evaluated with the lateral response of the caisson
with a D/B ratio of 1 embedded in the cohesive soil obtained
from the numerical analysis performed by Karapiperis and Summary and Conclusions
Gerolymos (2014). Details of the geometry of the caisson and the
mechanical properties of the caisson and the soil can be found in The present study enunciates numerical investigation of the lateral
Karapiperis and Gerolymos (2014). Fig. 21 compares the P–Y response of a circular-shaped hollow caisson embedded in

© ASCE 04021275-12 Int. J. Geomech.

Int. J. Geomech., 2022, 22(2): 04021275

 nonlinear soils with due consideration of interface nonlinearity. A Ec = modulus of elasticity of caisson;
detailed parametric study was performed through 3D nonlinear Es = modulus of elasticity of surrounding soil;
finite-element analysis. Parameters included were the D/B ratio, fck = characteristic strength of concrete;
the relative stiffness of the caisson–soil system, and the soil type. Gr = reference modulus of shear;
Emphasis was given on observing the trend of the lateral load ver- H = maximum scour depth;
sus lateral displacement curves, failure mechanism, deformation KN = normal stiffness of contact element;
profile of caisson, location of the point of rotation, and influence KT = tangential stiffness of contact element;
of the vertical load on the lateral response of caisson embedded Ki = slope of the tangent drawn on the initial portion of the
in nonlinear soils. The results obtained from the present analysis lateral load–lateral deflection curve;
elucidated that there exist three zones in the P–Y curve, namely, ini- Ku = slope of the tangent drawn on the later linear portion of
tial linear zone, transitional zone, and later linear zone. The obser- the lateral load–lateral deflection curve;
Downloaded from ascelibrary.org by Indian Institute of Technology, Roorkee on 10/03/22. Copyright ASCE. For personal use only; all rights reserved.

vations made in this study indicate that the mechanism of failure of P= lateral load;
a caisson–soil system is primarily controlled by the relative sliding PD = point of intersection of the extended part of initial and
and the separation which occurs at the caisson–soil interface. It is later linear portions of the curve;
observed that the initial stiffness of the caisson–soil system does PU = intercept of the later linear part of the curve on the load
not depend on the D/B ratio. The effect of the D/B ratio is observed axis;
to be more prominent in the transitional zone of the curve and the Pr = reference mean effective confining pressure;
later linear part of the curve. The lateral capacity of the caisson–soil Su = undrained strength;
system beyond the initial linear part increases with an increase in tN = normal component of the traction at the interface;
the value of the D/B ratio. The present analysis evinced that the spa- V = vertical load;
tial location of the point of rotation of a caisson is not fixed as it is Vapp = allowable vertical load;
generally considered rather it depends on the D/B ratio, soil stiff- Y= ground level deflection of foundation;
ness, and the amount of the lateral load applied on the caisson. YP = plastic component of deformation;
The results obtained from the present study indicated that the influ- Yb = movement of the base of the caisson in the lateral
ence of vertical load on the lateral response of caisson foundation is direction;
insignificant for the caisson embedded in clayey soils. However, Ye = elastic component of deformation;
the vertical load has a significant effect on the caisson embedded z= spatial location along with the depth of the foundation;
in sandy soils as the lateral load capacity of the caisson–soil system γ= transitional parameter;
decreases on the application of the vertical load. γmax = shear strain at failure;
Based on the observations made in the present study, a transi- Δh = separation at the sides of the caisson measured at the
tional plasticity-based constitutive model was proposed to predict ground level;
the P–Y curve of a laterally loaded caisson embedded in nonlinear Δv = separation at the base of the foundation;
soil. The proposed model considers that both elastic and plastic de- δ= distance measured from the current state of the load to its
formations develop together from the beginning of the lateral load- corresponding load on the extended portion of the later
ing. Three parameters are required to define the P–Y curve using the linear part of the curve;
proposed model. These parameters were calibrated using the results δ= angle of friction between the caisson and adjoining soil;
obtained from the present finite-element analysis. Further, relation- δmax = distance measured from the beginning of the curve to the
ships between model parameters and material parameters describing intercept of the later linear part of the curve meets the load
the soil and the caisson were established using least-squares regres- axis;
sion analysis. It was observed that the load–displacement (P–Y) ρ= density;
curves obtained from the proposed model match very well with Φ= friction angle of soil; and
the load–displacement curves obtained from the finite-element anal- ϑ= Poisson’s ratio.
ysis. The difference in results was less than 18% in all the cases.
References

Data Availability Statement Anagnostopoulos, C., and M. Georgiadis. 1993. “Interaction of axial and
lateral pile responses.” J. Geotech. Eng. 119 (4): 793–798. https://doi
Some or all data, models, or code generated or used during the .org/10.1061/(ASCE)0733-9410(1993)119:4(793).
study are available in a repository online in accordance with funder Biswas, S., and D. Choudhury. 2020. “Behavior of caisson foundations under
data retention policies. All the data and some of the codes are avail- lateral loading in layered cohesive Soil.” In Geo-Congress 2020:
Foundations, soil improvement, and erosion, edited by J. P. Hambleton,
able in https://drive.google.com/drive/folders/1NJdK5qtr_826wF1
R. Makhnenko, and A. S. Budge, 23–32. Reston, VA: ASCE.
mXEApe8dxLsr6WsdE?usp=sharing. Broms, B. B. 1964. “Lateral resistance of piles in cohesive soils.” J. Soil
Mech. Found. Div. 90 (2): 27–63. https://doi.org/10.1061/JSFEAQ
Notation .0000611.
Gerolymos, N., and G. Gazetas. 2006a. “Winkler model for the lateral re-
The following symbols are used in this paper: sponse of rigid caisson foundations in linear soil.” Soil Dyn. Earthquake
B = diameter of the hollow caisson; Eng. 26 (5): 347–361. https://doi.org/10.1016/j.soildyn.2005.12.003.
Br = reference bulk modulus; Gerolymos, N., and G. Gazetas. 2006b. “Development of Winkler model
for static and dynamic response of caisson foundations with soil and in-
C = undrained cohesion;
terface nonlinearities.” Soil Dyn. Earthquake Eng. 26 (5): 363–376.
co = a non-negative constant defining the rate of shear-induced https://doi.org/10.1016/j.soildyn.2005.12.002.
decrease in volume; Gerolymos, N., and G. Gazetas. 2006c. “Static and dynamic response of
D = grip length; massive caisson foundations with soil and interface nonlinearities—
d = pressure-dependent coefficient; Validation and results.” Soil Dyn. Earthquake Eng. 26 (5): 377–394.
E = Young’s modulus; https://doi.org/10.1016/j.soildyn.2005.12.001.

© ASCE 04021275-13 Int. J. Geomech.

Int. J. Geomech., 2022, 22(2): 04021275

 IRC (International Rescue Committee). 1972. Recommendations for esti- cohesionless Soil.” Earthquake Spectra 28 (3): 1117–1145. https://doi
mating the resistance of soil below the maximum scour level in the .org/10.1193/1.4000074.
design of caissons of bridges. IRC 45-1972. New Delhi: IRC. Mroz, Z. 1967. “On the description of anisotropic workhardening.”
IRC (International Rescue Committee). 1983. Standard specifications and J. Mech. Phys. Solids 15 (3): 163–175. https://doi.org/10.1016/0022
code of practice for road bridges, section: VII. Foundations and sub- -5096(67)90030-0.
structure. IRC 78-1983. New Delhi: IRC. Parra-Colmenares, E. J. 1996. “Numerical modeling of liquefaction
IRC (International Rescue Committee). 2000. Standard specifications and and lateral ground deformation including cyclic mobility and dilation
code of practice for road bridges, section: VII. Foundations and sub- response in soil systems.” Doctoral dissertation, Dept. of Civil and
structure. IRC 78-2000. New Delhi: IRC. Environmental Engineering, Rensselaer Polytechnic Institute.
IS (Indian Standard). 2000. Plain and reinforced concrete. Code of Practice Ponnuswamy, S. 2008. Bridge engineering. New York: McGraw-Hill.
(4th revision). IS:456-2000. New Delhi: IS. Prevost, J. H. 1985. “A simple plasticity theory for frictional cohesionless
Iwan, W. D. 1967. “On a class of models for the yielding behavior of soils.” Int. J. Soil Dyn. Earthquake Eng. 4 (1): 9–17. https://doi.org/10
Downloaded from ascelibrary.org by Indian Institute of Technology, Roorkee on 10/03/22. Copyright ASCE. For personal use only; all rights reserved.

continuous and composite systems.” J. Appl. Mech. 34 (3): 612–617. .1016/0261-7277(85)90030-0.

https://doi.org/10.1115/1.3607751. Varun, D. Assimaki, and G. Gazetas. 2009. “A simplified model for the lat-
Jain, S. K. 2016. “Earthquake safety in India: Achievements, challenges eral response of large diameter caisson foundations—Linear elastic for-
and opportunities.” Bull. Earthquake Eng. 14 (5): 1337–1436. mulation.” Soil Dyn. Earthquake Eng. 29 (2): 268–291. https://doi.org
https://doi.org/10.1007/s10518-016-9870-2. /10.1016/j.soildyn.2008.02.001.
Karapiperis, K., and N. Gerolymos. 2014. “Combined loading of caisson Yang, Z. 2000. Numerical modeling of earthquake site response including
foundations in cohesive soil: Finite element versus Winkler modeling.” dilation and liquefaction. New York: Columbia Univ.
Comput. Geotech. 56: 100–120. https://doi.org/10.1016/j.compgeo Yang, Z., J. Lu, and A. Elgamal. 2008. Opensees soil models and solid-
.2013.11.006. fluid fully coupled elements: User’s manual. San Diego: Dept. of
Kumar, N. D., and S. N. Rao. 2012. “Lateral load: Deflection response of an Structural Engineering, Univ. of California.
embedded caisson in marine clay.” Mar. Georesour. Geotechnol. Zafeirakos, A., and N. Gerolymos. 2016. “Bearing strength surface for
30 (1): 1–31. https://doi.org/10.1080/1064119X.2011.562444. bridge caisson foundations in frictional soil under combined loading.”
McKenna, F., and G. Fenves. 2001. The OpenSees command language man- Acta Geotech. 11 (5): 1189–1208. https://doi.org/10.1007/s11440-015
ual: Version 1.2. Berkeley: Pacific Earthquake Engineering Center, Univ. -0431-7.
of Calif. Accessed March 16, 2020. https://opensees.berkeley.edu Zienkiewicz, O. C. 1974. “Constrained variational principles and penalty
/OpenSees/OpenSeesCommands.pdf. function methods in finite element analysis.” In Conf. on the
Mondal, G., A. Prashant, and S. K. Jain. 2012. “Significance of interface Numerical Solution of Differential Equations, 207–214. Berlin,
nonlinearity on the seismic response of a well-pier system in Heidelberg: Springer.

© ASCE 04021275-14 Int. J. Geomech.

Int. J. Geomech., 2022, 22(2): 04021275