You are on page 1of 24

10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022].

See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Received: 10 June 2022 Revised: 24 November 2022 Accepted: 26 November 2022

DOI: 10.1002/eqe.3794

RESEARCH ARTICLE

A new nonlinear spring-dashpot model of CPRF of NPP


structure based on coupled BEM-FEM approach

Mohd Firoj1 Bal Krishna Maheshwari2

1 Department of Earthquake Engineering,


Indian Institute of Technology Roorkee, Abstract
Roorkee, Uttarakhand, India In this paper, a BEM-FEM coupled model of combined piled raft foundation
2 Department of Earthquake Engineering, (CPRF) is proposed for both the frequency domain and the time domain dynamic
Indian Institute of Technology Roorkee,
Roorkee, Uttarakhand, India analyses of a NPP structure in the linear, equivalent linear and nonlinear soil
domain. The piles are modeled using the finite element method as a beam ele-
Correspondence
ment, while the raft and internal domain of soil are modeled using the eight
Bal Krishna Maheshwari, Shamsher
Prakash Chair Professor, Department of nodded brick element. It is assumed that the piles are rigidly connected with
Earthquake Engineering, Indian Institute the raft. The nonlinearity of soil is modeled using the hyperbolic stress–strain
of Technology Roorkee, Uttarakhand
relation and Modified Drucker–Prager Model. The boundary of the layered soil
247667, India.
Email: bk.maheshwari@eq.iitr.ac.in domain is modeled using the boundary element method. For simplification,
it is assumed that the raft and piles are rigidly connected with the soil that
means there are no slip and separation conditions at the interface of pile-soil
and raft-soil. In order to validate the proposed approach, numerical results, that
is, impedances in the horizontal, vertical and rocking directions, are compared
with the published literature. Further, dynamic analyses of a lumped structure
founded on CPRF is carried out in the time domain and frequency domain using
these impedances and results are compared with the FEM. It is found that the
spring-dashpot models give the reasonably good agreement with finite element
method for linear, equivalent linear and nonlinear models for both frequency
domain and time domain analyses. Moreover, practical application of proposed
model is shown for a very large problem of raft-pile-soil-structure interaction
such as a NPP structure on the layered soil. It was found that the nonlinearity
significantly increases the response of the NPP structure. The proposed model
has many advantage over the finite element model, for example, first it is very
simple and second it reduced the computation time significantly with the good
accuracy.

KEYWORDS
boundary element method, combined piled raft foundation, finite element method, hyperbolic
stress–strain model, modified Drucker–Prager model, NPP, Spring–Dashpot model

Earthquake Engng Struct Dyn. 2022;1–24. wileyonlinelibrary.com/journal/eqe © 2022 John Wiley & Sons Ltd. 1
10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
2 FIROJ and MAHESHWARI

HIGHLIGHTS
∙ The dynamic impedances are evaluated for combined piled raft foundation (CPRF).
∙ The effect of the soil nonlinearity on the dynamic impedance of CPRF is examined considering equivalent
linear, hyperbolic stress–strain and modified Drucker–Prager Models.
∙ An equivalent spring-dashpot model is proposed and validated against finite-element model for linear,
equivalent linear and nonlinear cases.
∙ To capture the material nonlinearity of the soil, algorithms for computation has been developed.
∙ Using the proposed spring-dashpot model, a practical application to evaluate seismic response of a NPP
structure founded on the CPRF is demonstrated considering the nonlinearity of the soil.

1 INTRODUCTION

In order to carry out the dynamic analysis of the very large problem such as a NPP on the layered soil, the direct method
of analysis for such a complex system become computationally time consuming. To reduce the computation time period,
substructure method is used in which the impedance is applied at the base of the foundation. In this context, researchers1–3
developed the impedance functions for the pile group foundation system under the harmonic excitation which can be
further used in the dynamic analysis of the superstructure analysis. However, in these studies, nonlinearity of soil was
not considered. The numerical analyses of single pile and pile group were performed by various researchers4–7 using the
coupled boundary element method (BEM) and finite element method (FEM) for the soil-pile system. However, in these
studies, the modelling of soil is done using the BEM which is not capable to model the nonlinear soil behavior.
Cairo et al.8,9 proposed a simplified method for the dynamic analysis of pile group under the vertical loading using
closed form stiffness matrices derived by Kausel and Roësset.10 However, that study was restricted to the pile group
foundation under the vertical harmonic excitation only. Maheshwari and Emani11 carried out the 3D nonlinear FEM cou-
pled with consistent infinitesimal finite-element cell method (CIFECM) for the seismic analysis of pile groups. Syed and
Maheshwari12 performed the nonlinear time domain analysis for the single pile with the coupled FEM-SBFEM and calcu-
lated the dynamic impedance and kinematic interaction factors. Wang and Ai13 carried out the vertical vibration analysis
of pile group in the poroelastic layered soil media using the FEM technique. In order to overcome the limitation of the
nonlinear behavior of soil, various researchers14–17 used the direct method of analysis using finite element method (FEM)
to analyze the high rise buildings, bridges, tunnels along with the soil in the time domain formulation. However, these
studies are very complex and computationally inefficient which require highly computational efficient workstations. To
avoid the wave reflection from the boundary nodes, various researchers used the nonreflecting boundary in the modelling
of SSI problems, for example, frequency-dependent Kelvin elements18–21 for direct method of SSI and FEM-SBFEM12 for
the substructure method.
Recently, instead of pile foundation, the combined piled raft foundation (CPRF) is successfully used under the tall build-
ing structures.22–24 Recently many researchers25–27 used the finite element modelling technique to compute the behavior
of combined piled-raft foundation. Bhaduri and Choudhury28 carried out the steady state response of CPRF under the
dynamic loading. These studies are for either static or dynamic analysis of the CPRF foundation. However, no study
is reported for the seismic analysis of CPRF considering nonlinearity. Further, no substructure method is available to
evaluate the impedance of the CPRF for the nonlinear dynamic analysis of superstructure in the time domain formulation.
The NPPs are constructed on different types of foundations, depending on the soil classification and geological con-
ditions. In India, many NPPs are forthcoming on soft soil such as alluvium and requires intensive dynamic analysis.
In the western region, where the basaltic formation exists, the NPP structures are founded on rock. However, in the
Indo-Gangetic plain where alluvium extends to depth in excess of 200 m, nuclear structures are required to be especially
designed to take into account the poor ground condition. Therefore, CPRF for the NPPs is a better choice for soft soil as
reported by Firoj and Maheshwari.29 The importance of CPRF become more significant for a heavy structure like NPP.
Saxena et al.30 and Saxena and Paul31 studied the effect of embedment, slip and separation on the seismic behavior of
NPP. Bhaumik and Raychowdhury32 and Kumar et al.33 carried out the seismic performance analysis of NPP structure
using the Beam on Nonlinear-Winkler-Foundation (BNWF) approach for the soil foundation interface. Abell et al.34 used
the domain reduction method (DRM) to couple the seismic (near field) imitations with local soil–structure–interaction
10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
FIROJ and MAHESHWARI 3

response. Borbón et al.35 studied the effect of soil stiffness, SSI and infill soil stiffness on the seismic response of NPP.
Patil et al.36 studied the various aspects of CPRF of NPP structure such as soil medium, raft response, pile response and
NPP performance under the static vertical loading. Veeraraghavan et al.37 developed the open source software for the
nonlinear seismic response analysis and risk assessment analysis of the NPP structure. In most recent study, Mahesh-
wari and Firoj,38 proposed a methodology to estimate the impedances for the embedded raft foundation. The impedance
used in that study was reproduced form the Veletsos and Verbič.39 Further, nonlinearity was not considered. There, it was
assumed that spring-dashpot value is fixed corresponding to the predominant frequency of excitation. In another study by
Firoj and Maheshwari,29 the response of the NPP structure is examined using the direct method approach which requires
relatively more computation time period. Keeping in view these reported studies, it can be concluded that there is no
well-established nonlinear substructure method available to calculate the seismic response of the NPP structure founded
on CPRF with layered soil system.
Even today, for the nonlinear region of soil, there is no well-established method available to evaluate the dynamic
impedances of CPRF and to use these impedances in time history analysis of the superstructure. This is in contrast to the
dynamic analysis of superstructure, for which the nonlinear properties are available for the finite element modelling.40
Further, most of the studies are performed for the pile groups. Moreover, the nonlinearity is considered only for the super-
structure not for the whole system. Therefore, in this paper linear, equivalent linear (EL) and nonlinear spring-dashpot
model of NPP structure founded on CPRF is developed using the coupled FEM-BEM approach. Moreover, a nonlinear
spring-dashpot approach considering substructure method is presented for the whole system in the time domain analysis.
Further, the aim of this study is to reduce the computation time of large soil–structure interaction problems by modelling
using the proposed spring-dashpot model. Following are the major novel contributions of this study:

∙ The dynamic impedances for the CPRF are evaluated using the coupled BEM-FEM approach.
∙ The effects of the soil nonlinearity on the dynamic impedance of CPRF are considered using three different constitutive
relationships for the soil, that is, EL, Hyperbolic Stress-Strain Model (HSSM) and Modified Drucker–Prager Model
(MDPM).
∙ An equivalent spring-dashpot model which replaced soil and pile group (used in CPRF) is proposed and results are
compared with the finite element.
∙ To capture the material nonlinearity of the soil, a MATLAB code and a FORTRAN code in ABAQUS are developed. A
good agreement is found.
∙ A practical application to evaluate seismic response of the NPP structure founded on the CPRF is demonstrated using
the proposed spring-dashpot model considering the nonlinearity of the soil.

2 RAFT, PILES AND SOIL: FEM MODELLING AND FORMULATION

Three noded embedded beam element is used for the finite element modelling of piles with each node having three trans-
lations and three rotational degrees of freedom. Three-node beam element has a great advantage over two-node elements,
which provides solving large displacement geometry in nonlinear static and dynamic analysis.41 Although three-node
elements increased computational efforts at element level, the computers’ treatment polynomial function efficiency is
very high, and thus overall, efforts of computation are not increased. Raft and internal domain of soil are modelled using
8 noded brick elements. For simplicity, it is assumed that the raft-pile, raft-soil and pile-soil are rigidly connected with
each other, which provide a perfect bond between nodes in the model. The rigid connection has two sides, dependent and
independent. The independent side governs the movement of a node on the dependent side. The dependent side has six
degrees of freedom that can control and fix selectively. The rotational degree of freedom is must to determine the rocking
impedance for CPRF. Therefore, rotational DOFs are considered. The rotation DOFs of pile is connected with the brick
element by allowing the rotation at the common node. Wu and Finn42 used similar technique. The behavior of the raft,
piles and soil subjected to dynamic load can be defined by using the equation of motion40 :

[𝑀] {𝑥̈ (𝑡)} + [𝐶] {𝑥̇ (𝑡)} + [𝐾] {𝑥 (𝑡)} = {𝑓 (𝑡)} (1)

̈
where, [𝑀], [𝐶] and [𝐾] are the mass, damping and stiffness matrices, respectively for the raft, piles and soil. 𝑥(𝑡), ̇
𝑥(𝑡)
and 𝑥(𝑡) are the nodal acceleration, velocity and displacement, respectively. 𝑓(𝑡) is the nodal force vector over the raft and
variable t denotes the time.
10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
4 FIROJ and MAHESHWARI

For the frequency domain analysis, it is supposed that the top of raft is subjected to force then in that case harmonic
forcing function

𝑓 (𝑡) = 𝐹𝑒𝑖𝜔𝑡 (2a)

The solution of Equation (1) considering phase-lag (α) will be:

𝑥 (𝑡) = 𝑥𝑜 𝑒(𝑖𝜔𝑡+𝛼) = 𝑥𝑜 𝑒𝑖𝜔𝑡 𝑒𝛼 (2b)

where, 𝑥𝑜 is the amplitude of nodal displacement and rotation vector for the raft and piles. F is the amplitude of
nodal forces. 𝜔 is angular frequency of the excitation. Substituting Equation (2b) and its first and second derivatives in
Equation (1):
( )
[𝐾] − 𝜔2 [𝑀] + 𝑖𝜔 [𝐶] 𝑒𝛼 𝑥𝑜 = 𝐹 (3)

Once the elemental stiffness, damping and mass matrix is ready then it is assembled in the global degree of freedom for
the whole system, one can write the following equation:

̄ 𝑥𝑜 = 𝐹
[𝐾] (4a)

where, ̄ = {[𝐾] − 𝜔2 [𝑀] + 𝑖𝜔[𝐶]}𝑒𝛼


[𝐾] (4b)

where, [𝐾]̄ is composite stiffness matrix of the global system, in which the real and imaginary parts correspond to the
dynamic stiffness and damping of the system, respectively.
For the linear analysis, Newmark method (Chopra40 ) is used as described by Firoj and Maheshwari.29 For the nonlinear
time domain analysis, [𝐾] and [𝐶] matrices are updated using the modified Newton–Raphson procedure. The convergence
is achieved by the unbalanced force and displacement criteria with a very small tolerance value of 10–5 .

3 MODELLING OF UNBOUNDED SOIL DOMAIN USING BEM

The boundary element method (BEM) has emerged over the last 20 years as an alternative numerical technique because
it requires only surface discretization. This feature makes it particularly efficient in dealing with infinite and semi-infinite
problems, which commonly occur in the geotechnical earthquake engineering field.43 The BEM has significant advan-
tages over the FEM as the former requires only the boundary data as inputs. Thus, infinite domain can be modelled.
Further, the meshing effort is limited, and the system matrices are smaller. In the present study, FEM is used to deal
with interior domain which may be heterogeneous and nonlinear. Thus, advantages of both BEM and FEM are exploited
in the present analysis. The coupled BEM-FEM reduced the domain size of the finite element mesh significantly. The
boundary of the outer node of the truncated soil domain is demonstrated using the boundary BEM. The boundary sur-
face is modelled using the four-node quadrilateral elements. The boundary integral equation in the general form can be
written as43 :

𝐻𝑢 = 𝐺𝜎 (5)

where, 𝐻. and 𝐺 are the system


.
matrices of the order of node number over the boundary element and can be expressed
as 𝐻 = ∫ 𝜎𝑑Γ and 𝐺 = ∫ 𝑢𝑑Γ, respectively. 𝑢 and 𝜎 are the displacement and stress vectors, respectively. Γ is boundary
𝚪 𝚪
length. The details of the BEM modelling can be found in the literature.43–45 The unbounded soil media consists of free
surface nodes (f) and interaction nodes (i) at the boundary connected with finite element nodes as shown in Figure 1. The
10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
FIROJ and MAHESHWARI 5

F I G U R E 1 Coupled BEM-FEM model of 2 × 2 pile group embedded in a half-space (𝐷 = diameter of pile; 𝐿 = length of pile;
tr = thickness of raft; s = spacing of pile; Fx and Fz are forces in x and z directions, respectively; Mθ = moment about y axis)

equation for far field domain can be extended as:


[ ] { } [ ] { }
𝐻𝑓𝑓 𝐻𝑓𝑖 𝑢𝑓 𝐺𝑓𝑓 𝐺𝑓𝑖 𝜎𝑓
= (6)
𝐻𝑖𝑓 𝐻𝑖𝑖 𝑢𝑖 𝐺𝑖𝑓 𝐺𝑖𝑖 𝜎𝑖

In this study, it is assumed that the free surface is unloaded. Therefore, the traction vector (𝜎𝑓 ) at the free surface
will be zero. Solving Equation (6) for the traction vector at the interface, which leads to Equation (7) in condensed form
as:
( )−1 ( )
−1 −1
𝐺𝑖𝑖 − 𝐻𝑖𝑓 𝐻𝑓𝑓 𝐺𝑓𝑖 𝐻𝑖𝑖 − 𝐻𝑖𝑓 𝐻𝑓𝑓 𝐻𝑓𝑖 𝑢𝑖 = 𝜎𝑖 (7)

̃ in both side of Equation (7), which contains the shape and area or length (ds)
Multiplying the distribution matrix (𝐴)
of element and is defined as:
.
𝐴̃ = ∫ 𝑁𝑁 𝑇 𝑑𝑠 (8)
𝑠

where, N is the shape function of the of the boundary element. Now Equation (7) can be written as:

𝑆𝐵∞ 𝑢 = 𝐹 (9)

where, 𝑆𝐵∞ , u and F are the dynamic stiffness matrix, displacement and force vectors of infinite far field soil domain, respec-
tively. Equation (7) is used only for the far field nodes and it is added with near field domain of FEM using substructure
method. 𝑆𝐵∞ and F of the far field soil domain are defined as:

( )−1 ( )
𝑆𝐵∞ = 𝐴̃ 𝐺𝑖𝑖 − 𝐻𝑖𝑓 𝐻𝑓𝑓
−1
𝐺𝑓𝑖 −1
𝐻𝑖𝑖 − 𝐻𝑖𝑓 𝐻𝑓𝑓 𝐻𝑓𝑖 (10)

̃ 𝑖
𝐹 = 𝐴𝜎 (11)

4 NONLINEAR MODELLING OF SOIL

During the seismic loading, the behavior of soil is highly nonlinear; therefore, material nonlinearity at high strain is very
significant and need to be considered for Raft-Pile-Soil-Structure Interaction (RPSSI) analysis. In this study, nonlinear
10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
6 FIROJ and MAHESHWARI

FIGURE 2 (a) Hyperbolic stress–strain relationship, (b) successive increment technique for analysis

behavior of the soil is considered using three constitutive behavior, that is, EL approach, hyperbolic stress–stress rela-
tionship and Modified Drucker–Prager constitutive model. The EL method and its procedure is described in Maheshwari
and Firoj.38 The hyperbolic stress–strain model and Modified Drucker–Prager model are described in the following
subsections.

4.1 Hyperbolic Stress-Strain Model (HSSM) Relationship

Duncan and Chang46 by the finding of Kondner47 shows that the nonlinear behavior of sandy and clayey soil can be
approximated by the hyperbolic stress–strain relationship. The hyperbolic equation proposed by Kondner47 was


𝜎1 − 𝜎3 = (12)
𝑎+𝑏 ∈

where, 𝜎1 and 𝜎3 are the major and minor principal stresses, respectively; ∈ is the axial strain; a and b are the constant
which can be determined using the experimental data. In Equation (12), a is the reciprocal of initial tangent modulus (𝐸𝑖 )
and b is the ratio of failure ratio and difference of ultimate stress at which the soil fails as shown in Figure 2a. Thus,

1
𝑎 = (13a)
𝐸𝑖

𝑅𝑓
𝑏 = (13b)
(𝜎1 − 𝜎3 )𝑓

Therefore, Equation (12) can be written as:


𝜎1 − 𝜎3 = ∈𝑅𝑓
(13c)
1
+
𝐸𝑖 (𝜎1 −𝜎3 )𝑓

where, 𝑅𝑓 is the failure ratio which is always less than unity. For different types of soil, it varies from 0.75 to 1. In the
present analysis, 𝑅𝑓 is considered 0.91. Janbu’s48 experimental studies recommended that the initial tangent modulus of
the soil is dependent on the confining pressure and it can be written as
( )𝑛
𝜎3
𝐸𝑖 = 𝐾𝑝𝑎 (14)
𝑝𝑎
10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
FIROJ and MAHESHWARI 7

where, 𝑝𝑎 is atmospheric pressure; K is modulus number and n is the exponent defining the variation of 𝐸𝑖 . In the present
analysis, K = 2000 and n = 0.54 for the sand are considered. As this model is applicable for both the loading and unloading
conditions, its failure criteria can be defined using the Mohr–Coulomb model as:

2𝑐 cos 𝜙 + 2𝜎3 sin 𝜙


(𝜎1 − 𝜎3 )𝑓 = (15)
1 − sin 𝜙

where, c and 𝜙 are the cohesion and angle of internal friction of soil, respectively.
The nonlinearity of the soil in the FEM analysis is approximated by assigning the different modulus of elasticity to each
of element, which depends on the stress–strain relationship. It is necessary to perform the repeated analyses to assure
that the modulus and stress values correspond to each of the element in the CPRF system. For assigning the modulus
value to each of the element, successive increment technique is used as shown in Figure 2b. At the beginning of each new
increment of loading, the modulus value is changed according to the stress condition in each element. For this, tangent
modulus (𝐸𝑡 ) is used that can be described as:

[ ]2 ( )𝑛
𝑅𝑓 (1 − sin 𝜙) (𝜎1 − 𝜎3 ) 𝜎3
𝐸𝑡 = 1 − 𝐾𝑝𝑎 (16a)
2𝑐 cos 𝜙 + 2𝜎3 sin 𝜙 𝑝𝑎

which can be simplified using Equations (14) and (15) as follows

[ ]2
𝑅𝑓 (𝜎1 − 𝜎3 )
𝐸𝑡 = 1 − 𝐸𝑖 (16b)
(𝜎1 − 𝜎3 )𝑓

A MATLAB code is prepared to implement the nonlinear behavior of soil into the BEM-FEM modelling of CPRF system.
Moreover, this constitutive soil model was not available in the ABAQUS.49 Therefore, a FORTRAN subroutine is also coded
as a User-defined Material (UMAT) and it is implemented into the ABAQUS49 material library.

4.2 Modified Drucker–Prager Model (MDPM)

For modelling material nonlinearity (plasticity) of soil, an advanced constitutive model, that is, Modified Drucker-Prager
model (MDPM), also known as Drucker–Prager with Cap, model is used. The primary importance of this model is to
consider the hysteretic response of soil using the strain hardening of cap. The detailed formulation of this model is reported
in Firoj and Maheshwari.29

5 VALIDATION OF COUPLED BEM-FEM MODEL

In the literature, the impedance functions are available only for the pile group for the linear soil behavior. Therefore, first
coupled BEM-FEM is verified for the pile group foundation in linear soil. Further, authors also carried out the validation
of CPRF under the vertical loading in which the nonlinearity of soil is considered. There is no study in which impedances
are calculated for the CPRF foundation in the nonlinear soil behavior. However, sufficient validation was carried out in
previous study (Firoj and Maheshwari29 ) for the modelling of CPRF in nonlinear soil. It was observed that the model can
predict the nonlinear behavior of soil. The results were in good agreement with the centrifuge and numerical modelling
of the raft, pile group and CPRF under the static loading.
In addition, in the present study, the coupled BEM-FEM model of pile groups is validated in terms of dynamic
impedances considering the behavior of soil as a linear with the published literature. The model of pile group with rigid
massless cap is analyzed for the horizontal and vertical vibrations of the pile groups. Here, two sets of pile group are used
one is 2 × 2 and another is 4 × 4 as presented by Miura et al.7 The material and geometrical properties used for the valida-
tion is the same as those used by Miura et al.7 and given in Table 1. The dynamic impedances for the harmonic excitation
10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
8 FIROJ and MAHESHWARI

TA B L E 1 Simulation details in model


Modulus of
Elasticity Poisson’s Density, Damping L/D ratio s/D ratio
(GPa) ratio, ν ρ (kg/m3 ) Ratio, ζ for pile of pile
Soil 0.25 0.4 1750 5% – –
Pile 25 0.25 2500 0 20 5

F I G U R E 3 Horizontal impedances of 2 × 2 and 4 × 4 pile groups. Comparison with Miura’s solution (a) stiffness coefficient, (b)
damping coefficient (Ep /Es = 100; s/D = 5)

were calculated using the relationship as:


[ ]
𝐾𝑖𝑗 𝐺 = 𝐾𝑖𝑗 𝑘𝑖𝑗 + 𝑖𝑎𝑜 𝑐𝑖𝑗 (17)

where, 𝐾𝑖𝑗 𝐺 is the dynamic stiffness matrix of the pile group which relates the force or moment at the top of the pile cap
to the resulting displacement or rotation at the same point. 𝐾𝑖𝑗 is static stiffness. 𝑘𝑖𝑗 and 𝑐𝑖𝑗 are the frequency dependent
stiffness and damping coefficients of pile groups respectively. 𝑎𝑜 is the dimensionless frequency which is defined as, 𝑎𝑜 =
𝜔𝐷∕𝑉𝑠 , where, 𝑉𝑠 is shear wave velocity of soil, 𝜔 is angular frequency of excitation and D is the diameter of the pile
which is taken as 0.5 m. The impedances were normalized by N (number of piles) times of static stiffness of single pile.
The horizontal and vertical impedance of the pile groups using the proposed coupled BEM-FEM technique are shown
in Figure 3 and Figure 4, respectively. It can be observed that the findings are in very good promise for both sets of pile
groups for the linear soil behavior. There is minor deviation in the vertical impedance due to the different methodology
used by Miura et al.7 The present study used the approximate technique of coupled BEM-FEM while Miura et al.7 used
the exact solution using the Green function for the calculation of dynamic impedance of pile groups.
In the other set of results, the proposed model is validated with the closed form solution proposed by Kaynia.50 Hori-
zontal and vertical impedances (stiffness and damping coefficient) of 2 × 2 pile group are calculated in the homogeneous
elastic half space as shown in Figure 5 and Figure 6, respectively. Three pile spacing were considered, that is, s/D = 2, 5,
10.
The properties used are: damping coefficient of soil β = 0.05; ratio of pile to soil stiffness Ep /Es = 1000; ratio of soil to
pile densities ρs /ρp = 0.72; Poisson’s ratios νs = 0.4 (for the soil) and νp = 0.25; length to diameter ratio L/D = 15; diameter
of pile D = 0.6 m.
It can be perceived that the estimated impedances are in close agreement with those reported by Kaynia.50 All results
of Kaynia50 were reproduced approximately. The results from the present study to that reported in literature suggest that
the proposed coupled BEM-FEM technique can be further used in the nonlinear dynamic analysis of CPRF.
Figure 7 shows the comparison of the proposed coupled BEM-FEM model with FEM model of 2 × 2 pile group under
the vertical mode of vibration. The material properties and pile configuration used in the FEM model of Dezi et al.51 is
10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
FIROJ and MAHESHWARI 9

F I G U R E 4 Vertical impedances of 2 × 2 and 4 × 4 pile groups. Comparison with Miura’s solution (a) stiffness coefficient, (b) damping
coefficient (Ep /Es = 100; s/D = 5)

F I G U R E 5 Horizontal impedances of 2 × 2 pile groups. Comparison with Kaynia’s closed form solution (a) stiffness coefficient, (b)
damping coefficient (Ep /Es = 1000)

the same as that of Kaynia.50 The same has been used in the present study. However, the modelling technique of soil is
different in three cases. The size of domain used in the present study is the same as that reported by Dezi et al.,51 where
Winkler’s springs are used to model the boundary. It can be observed that the results from the present study is in good
agreement with Kaynia’s closed form solution as compared to the results from FEM model of Dezi et al.51 Thus, it can
be concluded that the proposed coupled BEM-FEM technique is in a better position to calculate the impedances of piles
group. Therefore, this is further used in evaluation of the dynamic impedances of CPRF.

6 EFFECT OF SOIL PLASTICITY

The effect of nonlinearity on the raft-pile-soil interface is calculated using the developed coupled BEM-FEM model of
2 × 2 size piled raft foundation. The nonlinearity of soil is considered using the EL, HSSM and MDPM. The G/Gmax and
damping ratio curves used in the EL approach are shown in Figure 8. Table 2 summarize the material properties of the
soil, pile and raft. The effect of nonlinearity was investigated on the horizontal, vertical and rocking impedances of CPRF
along with the increasing frequency of excitation.
10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
10 FIROJ and MAHESHWARI

F I G U R E 6 Vertical impedances of 2 × 2 pile groups. Comparison with Kaynia’s closed form solution (a) stiffness coefficient, (b)
damping coefficient (Ep /Es = 1000)

F I G U R E 7 Vertical impedances of 2 × 2 pile groups. Comparison with FEM model (a) stiffness coefficient, (b) damping coefficient
(Ep /Es = 1000; s/D = 5)

TA B L E 2 Material properties of soil, pile and raft


Parameter Soil Pile Raft
3
Unit weight (kg/m ) 1750 2500 2500
Young’s modulus (MPa) 100 50,000 50,000
Poisson’s ratio 0.4 0.25 0.25
Friction angle 43o – -
Cohesion (kPa) 18 – –

The dynamic impedances of the piled raft foundation are required for the time harmonic or time domain analysis of the
superstructure supported on the piled raft foundation. These impedances are highly effected by the nonlinearity of soil
during the seismic loading. Figure 9a,b shows the horizontal impedances in which both stiffness and damping coefficients
decreases significantly from linear to EL case. However, there is small difference in the results of EL, HSSM and MDPM.
This may be because EL approach captures most of the nonlinearity. The values of impedances further reduced using
10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
FIROJ and MAHESHWARI 11

F I G U R E 8 Modulus reduction and


damping curve (After, Seed and Idrees52 )

TA B L E 3 Geometrical and material properties of raft-pile-soil (data from Horikoshi et al.53 and Eslami et al.54 )
Parameter Soil Pile Raft
Size (m) 28 × 28 × 16 0.5(D) × 9(L) 4×4×1
Unit Weight (kN/m3 ) 16.3 24 27
Young’s modulus (MPa) 40 41,700 70,000
Poisson’s ratio 0.3 0.25 0.35
Cohesion (kPa) 10 – –
Friction Angle 31o – -

HSSM and MDPM. MDPM reduced the stiffness and damping coefficients little more as compared to HSSM because of
the hardening and softening considered by the MDPM.
Figures 9c,d and 9e,f show the vertical and rocking impedances, respectively, in which the stiffness and damping coef-
ficients are decreasing with nonlinearity in the similar fashion as for the horizontal case. From Figure 9c, it can be
comprehended that nonlinear effect on the stiffness is more prominent in the frequency range of ao = 0.5–0.9. However,
difference in linear and EL impedances are for wider range of frequency (ao < 0.9). While damping coefficient is signifi-
cantly effected in the frequency range of ao < 0.6 (Figure 9d). This kind of variation in the impedance of CPRF is attributed
to the variation in fundamental frequency of the raft-pile-soil arrangement due to the nonlinearity of soil. Thus, it can be
concluded that nonlinearity using the three models (EL, HSSM, and MDPM) has significantly reduced the stiffness and
damping of the CPRF for the all modes of vibration, that is, horizontal, vertical and rocking with marginal differences
among them. The effect of nonlinearity is more clearly predicted by the time domain analysis of raft-pile-soil-structure
system presented in the next section.

7 COMPARISON OF SPRING-DASHPOT MODEL WITH FEM MODEL

In the present study, two techniques are used to evaluate the response of a lumped structure founded on CPRF subjected
to seismic loading and results are compared. In the first technique, finite element (FE) modelling is employed to model
the soil. In the second technique, the soil is replaced by the spring-dashpot models with the frequency dependent stiffness
and damping coefficients. For both the techniques, analyses are carried out assuming soil behavior Linear, EL, HSSM and
Modified Drucker-Prager Model (MDPM).
FE model of lumped mass rested on the CPRF is created in the ABAQUS.49 Figure 10a shows the generalized form of
raft-pile-soil-structure system. In this model; raft, pile and soil are modelled using the 3D finite element formulation as
direct method of analysis. The geometrical and material properties used for raft-pile-soil are listed in Table 3. The soil was
modelled using the 8 noded solid brick elements while 20 noded quadratic brick elements for the raft and 3 noded beam
element for the piles were used. At the outer nodes on the side boundaries, viscous dashpots are attached in vertical and
horizontal directions for absorbing the incoming waves. All the base nodes are fixed and earthquake motion was applied
in the X direction.
10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
12 FIROJ and MAHESHWARI

F I G U R E 9 Effect of nonlinearity (a) horizontal stiffness coefficient, (b) horizontal damping coefficient, (c) vertical stiffness coefficient,
(d) vertical damping coefficient, (e) rocking stiffness coefficient, (f) rocking damping coefficient of 2 × 2 CPRF (s/D = 5, Ep /Es = 500,
tr /D = 1.5, L/D = 20, ζs = 0.05)

In spring-dashpot model, soil and pile below the raft is replaced by the series of spring-dashpot (Voigt model) as shown in
Figure 10b and c. The springs were applied in all three translations and three rotational directions. The spring and dashpot
for soil and pile below the structure is calculated using the coupled BEM-FEM approach. The numerical calculations for
the spring-dashpot model were employed in the MATLAB. The procedure for the EL approach is the same as described in
the Maheshwari and Firoj.38 For the nonlinear analysis, spring stiffness is varied according to load-deflection curve which
is derived from the coupled BEM-FEM model √ of CPRF. The load deflection curves are independent of the frequency of
excitation. The damping coefficient (c = ζ 𝑘𝑚) used in the study is varied according to the stiffness calculated from
the load-deflection curves, where, ζ = hysteretic damping, k = stiffness of soil and m = mass of structure. Figure 11a
and Figure 11b show the load-deflection curve under the horizontal and vertical loading, respectively, at the top of raft
foundation for the linear and nonlinear soil behavior. The load-deflection varies linearly for linear soil model. While
10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
FIROJ and MAHESHWARI 13

FIGURE 10 (a) BEM-FEM model of CPRF, (b) spring-dashpot model (x–z plan) and (c) spring-dashpot model (y–z plan)

FIGURE 11 Displacement at the top of raft under (a) horizontal load, (b) vertical load

TA B L E 4 Properties of earthquake motion data


Sl. Earthquake PGA Predominant
No. event Magnitude (g) frequency (Hz)
1. Kobe (1995) 6.9 0.347 0.58
2. Bhuj (2001) 7.7 0.320 1.20
3. Northridge 6.7 0.568 1.22
(1994)
4. Chamoli 6.6 0.300 1.66
(1999)
5. Uttarkashi 7.0 0.300 2.90
(1991)

it varies nonlinearly for the MDPM and HSSM soil models in which the stiffness is reduced by the significant amount
increasing the displacements significantly as compared to linear model. Moreover, MDPM soil model gives slightly greater
displacement as compared to HSSM at the same load due to its consideration of softening and hardening behavior of soil.

7.1 In time domain formulation

In order to check the applicability of the proposed spring-dashpot model, the seismic response analysis is carried out under
five earthquake motions with different duration and frequency. Table 4 shows the characteristics of the input motions of
10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
14 FIROJ and MAHESHWARI

F I G U R E 1 2 (A) Displacement time history and (B) spectral acceleration at the top of lumped mass (a) linear (b) equivalent linear (c)
HSSM (d) MDPM for Kobe earthquake (1995)

F I G U R E 1 3 (A) Displacement time history and (B) Spectral displacement at the top of lumped mass (a) linear (b) equivalent linear (c)
HSSM (d) MDPM for Bhuj earthquake (2001)

Kobe (1995), Bhuj (2001), Northridge (1994), Chamoli (1999) and Uttarkashi (1991) earthquakes. The maximum PGA of the
input motion varies from 0.300 to 0.568 g and predominant frequency is in the range of 0.58–2.90 Hz. These motions are
used as a bedrock motion for finite element analysis. The input motion used for the spring-dashpot model is determined
by the free field analysis of layered soil mass at the top layer of soil (just below the base of the raft).
The seismic response of the lumped structure founded on the CPRF is calculated in terms of the horizontal displace-
ment, and response spectrum at the top of the structure. The seismic response of lumped structure is calculated assuming
behavior of soil as linear, EL, nonlinear (HSSM and MDPM). The results are compared with those obtained using finite
element model. Figures 12A–16A show the displacement time histories and spectral accelerations for the linear, EL, HSSM
10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
FIROJ and MAHESHWARI 15

F I G U R E 1 4 (A) Displacement time history and (B) Spectral acceleration at the top of lumped mass (a) linear (b) equivalent linear (c)
HSSM (d) MDPM for Northridge earthquake (1994)

F I G U R E 1 5 (A) Displacement time history and (B) Spectral acceleration at the top of lumped mass (a) linear (b) equivalent linear (c)
HSSM (d) MDPM for Chamoli earthquake (1999)

and MDPM cases, under the Kobe (1995), Bhuj (2001), Northridge (1994), Chamoli (1999) and Uttarkashi (1991) earthquake,
respectively. It can be observed that for all five earthquakes, the results of spring-dashpot model are in good agreement
with that using the finite element model for linear, EL and HSSM and MDPM cases except some marginal deviation. The
permanent deformation is not significantly observed in the Bhuj (2001) and Northridge (1994) earthquakes using spring-
dashpot and FEM model for the nonlinear case. Therefore, proposed model can calculate the residual displacement of the
structure efficiently.
Table 5 summarizes the maximum displacement values for the spring-dashpot and FEM model under different earth-
quake motions. The maximum displacement in the spring-dashpot model is increased by 2%–7.6%, 7.4%–10.9%, 7.8%–10.8%
10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
16 FIROJ and MAHESHWARI

F I G U R E 1 6 (A) Displacement time history and (B) Spectral acceleration at the top of lumped mass (a) linear (b) equivalent linear (c)
HSSM (d) MDPM for Uttarkashi earthquake (1991)

and 10.2%–11.1% in the linear, EL, MDPM and HSSM case, respectively, with respect to the finite element model. The
change in the response is expected due to the different models used, still it is less than 11%. Therefore, even for nonlinear
case, the proposed spring-dashpot model can be used in the seismic analyses of the structures. From the time histories
(Figures 12A–16A), it can be observed that EL and nonlinearity significantly increased the displacement of the lumped
structure as compared to linear case. However, EL method also reduced the stiffness in the soil at higher strain but gives
zero deformation of structure at the end of time history. However, in the nonlinear case, both MDPM and HSSM predicts
the permanent deformation of the structure at the end of time history due to the high strain rate in each element.
Figures 12B–16B show response spectrum at the top of the lumped mass for linear, EL, HSSM and MDPM, under the
Kobe (1995), Bhuj (2001), Northridge (1994), Chamoli (1999) and Uttarkashi (1991) earthquake, respectively. The results
of spring-dashpot model are in the same trend as of finite element model. However, some deviation is found between
peak values using the finite element and spring-dashpot model. The peak spectral accelerations observed in the spring-
dashpot and finite element model are shown in Table 5 under the various earthquake motions. In the spring dashpot
model, the peak spectral acceleration of the lumped structure is increased in the range of 3.7%–5.4%, 2.4%–9.2%, 3.5%–
11.1% and 6.6%–11% in the linear, EL, HSSM and MDPM case, respectively, with respect to finite element model. It
can also be noted that peak spectral acceleration of the NPP structure is significantly increased with the inelasticity of
the soil.
Thus, in general, the results of spring-dashpot model of lumped structure founded on CPRF are in good agreement with
those obtained using FEM model. The overall variation in the results are in the range of 1 to 11% which is in acceptable
limit. Table 6 shows the comparison of computation time required in evaluating the seismic response of the lumped
structure on the CPRF, using the FEM and the spring-dashpot model. It can be observed that with a little compromise in
accuracy (shown in Table 5) there is significant reduction in the computation time. The computation time using spring-
dashpot model is reduced in the range of 37% to 48% depending on the soil model used. Therefore, the proposed model
can evaluate the seismic response of an NPP structure founded on CPRF with good accuracy and less computation time.

7.2 In frequency domain formulation

In order to check the applicability of the proposed spring-dashpot model with the FEM model, in the frequency domain,
a harmonic excitation is applied at the base. In this context, effect of excitation frequency is studied on the response of
the lumped structure resting on the CPRF for the FEM model and spring-dashpot model and results are compared. The
10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
FIROJ and MAHESHWARI 17

TA B L E 5 Maximum horizontal displacement and spectral acceleration under various earthquake motion
Maximum displacement (m) Peak spectral acc. (g)
Spring- Spring-
Soil dashpot dashpot
Earthquakes behaviour FEM model % change FEM model % change
Kobe (1995) Linear 0.098 0.100 2.0 2.24 2.36 5.4
Eq. linear 0.114 0.125 9.6 2.29 2.46 7.4
Nonlinear 0.107 0.118 10.2 2.42 2.69 11.1
(HSSM)
Nonlinear 0.117 0.130 11.1 2.56 2.77 8.2
(MDPM)
Bhuj (2001) Linear 0.276 0.297 7.6 2.40 2.52 5.0
Eq. linear 0.358 0.389 8.6 3.03 3.31 9.2
Nonlinear 0.365 0.402 10.1 3.76 4.14 10.1
(HSSM)
Nonlinear 0.379 0.420 10.8 3.95 4.24 7.3
(MDPM)
Northridge Linear 0.114 0.122 7.0 3.47 3.60 3.7
(1994)
Eq. linear 0.135 0.145 7.4 4.68 5.07 8.3
Nonlinear 0.140 0.151 7.8 4.80 5.06 5.4
(HSSM)
Nonlinear 0.143 0.158 10.5 4.81 5.16 7.3
(MDPM)
Chamoli (1999) Linear 0.094 0.098 4.3 2.09 2.20 5.3
Eq. linear 0.115 0.127 10.4 2.53 2.59 2.4
Nonlinear 0.120 0.133 10.8 2.57 2.66 3.5
(HSSM)
Nonlinear 0.136 0.151 11.0 2.59 2.76 6.6
(MDPM)
Uttarkashi Linear 0.049 0.051 4.1 3.00 3.12 4.0
(1991)
Eq. linear 0.055 0.061 10.9 3.80 4.11 8.2
Nonlinear 0.077 0.083 7.8 3.86 4.29 11.1
(HSSM)
Nonlinear 0.078 0.086 10.2 4.00 4.44 11.0
(MDPM)

TA B L E 6 Comparison of computation time using FEM and spring-dashpot model for Kobe earthquake
Computation TIME (min)
Spring-dashpot model
For evaluating For evaluating
springs and response of the
Methodology FEM dashpots system Total % reduction
Linear 126 60 10 70 44.45
Eq. linear 141 73 11 84 48.22
Nonlinear 150 80 13 93 38.00
(HSSM)
Nonlinear 158 85 14 99 37.34
(MDPM)
10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
18 FIROJ and MAHESHWARI

FIGURE 17 Horizontal response at the top of structure with excitation frequency (a) linear (b) equivalent linear (c) HSSM (d) MDPM

FIGURE 18 Vertical response at the top of structure with excitation frequency (a) linear (b) equivalent linear (c) HSSM (d) MDPM

results are simulated for varying dimensionless frequency of excitation (a0 ) by applying the sine wave of amplitude 0.3 g.
Where a0 is defined as 𝜔𝐷∕𝑉𝑠 .
The response of the lumped structure is calculated in the horizontal, vertical and rocking directions. Figure 17a, b, c and
d shows the horizontal response of the structure under the harmonic excitation for the linear, EL, HSSM and MDPM. The
horizontal response of the system is increasing up to the dimensionless frequency 0.2 and then decreasing up to 0.7. It can
be observed that spring-dashpot model are in good agreement with finite element model of CPRF. The maximum horizon-
tal displacement in the spring-dashpot model is increased by 3.67%, 3.84%, 4.08% and 3.57% in case linear, EL, HSSM and
MDPM, respectively, compared to the FEM model. Furthermore, horizontal displacement significantly increases with
the frequency by considering the nonlinearity of the soil. Similar, trend in results is observed for the vertical response
of the lumped structure on the CPRF (Figure 18). The maximum vertical displacement in the spring-dashpot model is
increased by 4.32%, 3.86%, 4.22% and 3.60% in case of linear, EL, HSSM and MDPM, respectively compared to the FEM
model. Therefore, proposed model can calculate the vertical response of the lumped structure.
10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
FIROJ and MAHESHWARI 19

FIGURE 19 Rotational response at the top of structure with excitation frequency (a) linear (b) equivalent linear (c) HSSM (d) MDPM

F I G U R E 2 0 Effects of soil inelasticity on the


amplification ratio of lumped structure

Figure 19 shows the rotational displacement of the lumped structure for the linear, EL, HSSM and MDPM. The rotational
displacement increases up to the dimensionless frequency 0.2 and then decreases up to 0.7, similar to horizontal and
vertical response. It is observed that rotational displacement in the spring-dashpot model is over predicted by 3.25%, 3.39%,
3.21% and 4.64% in case of linear, EL, HSSM and MDPM, respectively, as compared to the FEM model. It is also noted
that there is marginal difference of 18%–32% in the linear and nonlinear response of the lumped structure for horizontal,
vertical and rotational displacement. Hence, it can be said that proposed model can calculate the nonlinear response of the
structure founded on the CPRF in the frequency domain. Thus, the proposed model indicated relatively better agreement
with FEM model in the frequency domain where deviation in the results is smaller as compared to that in the time domain.
This may be due the impedances used in the presented study, which is developed in the frequency domain. This limitation
may be reduced by developing the impedance in the time domain formulation.
Figure 20 shows the effect of nonlinearity on the lumped mass founded on CPRF using the spring-dashpot approach. It is
examined that amplification factor of the lumped structure is significantly increased by incorporating the soil nonlinearity.
The peak values of amplification are increased by 12.35%, 17.63% and 22.10% in case of EL, HSSM, and MDPM, respectively,
as compared to the linear case. It is also predicted that fundamental frequency is also increased by 7.60%, 9.86%, and 10.57%
in case of EL, HSSM, and MDPM, respectively, as compared to the linear case, which by virtue of decrease in the stiffness
and reduced in damping of system in case of nonlinear soil models.
10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
20 FIROJ and MAHESHWARI

F I G U R E 2 1 Schematic view of raft, pile,


soil and lumped structure

TA B L E 7 Index properties and shear wave velocity


Shear Wave
Cohesion Density Velocity Gmax
Depth (m) SPT (N) (kPa) ϕ (kN/m3 ) Type of Soil PI (m/s) (MPa)
3 7 20 30 18.2 Clayey Silty 2 176 56.60
Sand
6 23 18 40 18.3 4 264 127.79
9 39 22 32 18.2 12 316 182.00
12 45 20 38 18.3 13 332 201.71
15 22 22 38 18.5 5 260 125.34
18 34 23 40 18.6 9 302 169.43
21 31 28 36 18.9 8 292 161.69
24 27 35 40 18.0 7 279 140.18
27 43 53 25 18.2 14 327 194.50
30 17 23 48 17.9 3 238 101.77
33 70 54 20 17.9 15 386 266.44

8 PRACTICAL APPLICATION OF THE PROPOSED MODEL

The nuclear structure considered in this section is a lumped model of a typical Indian nuclear reactor building as reported
in Paul and Saxena.55 The raft foundation is circular having 45 m diameter and 4 m thickness with 2 m embedment in
the soil as shown in Figure 21. In addition to the raft, a pile group of circular pattern is considered in the present study to
represent the CPRF. The soil below the raft is heterogeneous having 3 m depth of each layer. Total depth of soil domain
considered is 33 m and width is 200 m.
The shear wave velocity (Vs ) of the soil layer is calculated using the correlation given by Imai56 as reported by Hasancebi
and Ulusay.57

𝑉𝑠 = 91𝑁 0.340 (23)

The calculated maximum shear modulus and other material properties are shown in Table 7. The site was reported
clayey silt with SPT value varying from 7 to 70. At the bottom layer, SPT value is more than 50, which is considered as
refusal. Therefore, fixed base condition is used in the coupled BEM-FEM model of CPRF foundation. In addition, the
spring-dashpot model is considered for the dynamic analysis of the NPP structure. For this, the same procedure is used
as in the preceding section. The G/Gmax and damping curves used in the EL analysis are selected as per the PI value as
shown in Figure 22.
The effect of nonlinearity is examined at the top of the NPP structure. The nonlinearity is considered using the eq. linear,
MDPM and HSSM. The results are compared with the elastic behavior of soil. For this, horizontal harmonic acceleration
is applied at the base of soil at different frequencies, and calculated motion at the top of foundation is used as an input
motion for the spring-dashpot model of NPP. The effect of nonlinearity is presented in Figure 23 in terms of amplification
10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
FIROJ and MAHESHWARI 21

FIGURE 22 G/Gmax and damping curves of soil (After, Vucetic and Dobry58 )

F I G U R E 2 3 Effects of nonlinearity on the


magnitude of response of NPP structure

FIGURE 24 Displacement time history at top of NPP structure for Kobe earthquake input motion

factor which is ratio of response at the top of NPP structure to input bedrock motion. It can be observed that the effect of
nonlinearity is more prominent near the natural frequency of the system. However, at the higher frequency there is not
much effect of nonlinearity. Maheshwari and Sarkar20 had also shown the similar observation.
In order to examine the influence of soil nonlinearity in time domain, a real time earthquake motion (Figure 16) is
applied at the base of soil. The NPP structural response is calculated and displacement is compared for different kinds
of soil nonlinearity as shown in Figure 24. The maximum displacement at the top of the structure in elastic, EL, HSSM
and MDPM soil conditions are 0.143, 0.178, 0.186 and 0.193 m, respectively. Maximum displacement is increasing in the
similar trend as that for amplification (Figure 23) for different types of nonlinearity. Moreover, it can also be noted that
more advanced constitutive soil model (HSSM and MDPM) gives the more displacement as compared to linear and EL
soil behavior. MDPM and HSSM leads to permanent distortion of the structure by virtue of reduced stiffness of soil.
10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
22 FIROJ and MAHESHWARI

9 SUMMARY AND CONCLUSIONS

In this study, a 3D coupled BEM-FEM model of CPRF in the inelastic region of soil has been presented. The raft, piles and
internal soil domain is modelled using the FEM in the inelastic range while the truncated soil mass using the BEM in the
linear elastic range. The nonlinearity of soil was taken into account using the EL, HSSM and MDPM. The effect of soil
plasticity is evaluated on the dynamic impedances of 2 × 2 piled raft foundation.
Further, based on coupled BEM-FEM approach, a spring-dashpot model representing 3D finite element model is
proposed. A comparison of results obtained from the proposed model with those obtained by the finite element
model shows that proposed model can calculate the seismic response (displacement and spectral acceleration) with
reasonable accuracy in both frequency and time domains. The proposed model has many advantage over the finite
element model, for example, it is very simple and second it reduced the computation time significantly with good
accuracy.
The following conclusions can be drawn from the results presented in this paper:
The proposed method of coupled BEM-FEM model shows good agreement with the well-established results reported
in the literature. The proposed method can be used for the different pile configuration and varying pile spacing.
In the dynamic analysis of piled raft foundation, nonlinearity plays a vital role and dynamic impedances significantly
decreased from linear to equivalent-linear case. However, there is small difference in the results of EL, HSSM and MDPM.
This is because of EL approach captured the most of the nonlinearity. Reduction in the stiffness and damping coefficients
using MDPM is greater than that using HSSM. This may be due the hardening and softening behavior considered by the
MDPM model.
An equivalent spring-dashpot model for soil-pile substructure is proposed and the results are compared with the FEM
model. The difference in the maximum displacement using the spring-dashpot model and the finite element model is only
in the range of 2%–11%, even considering soil nonlinearity, for different earthquake motions. Thus, the spring-dashpot is
reasonably accurate. Further, there is significant percentage decrease in the computation time, which is in the range pf
37% to 48%. Thus, the proposed spring-dashpot model can be efficiently used for the time domain dynamic analysis of the
NPP structure founded on CPRF with good accuracy.
The proposed spring-dashpot model can calculate the dynamic response of the lumped structure for linear, EL, HSSM
and MDPM cases on the CPRF in the frequency domain analysis with good accuracy. The maximum difference in the
spring-dashpot and FEM results is in the range of 3% to 5%.
A practical application of the proposed equivalent spring-dashpot model for a NPP structure is illustrated. The maxi-
mum displacement at the top of the considered NPP structure in elastic, EL, HSSM and MDPM soil conditions are 0.143,
0.178, 0.186 and 0.193 m, respectively. Thus, it can be concluded that the proposed model is capable of capturing the effect
of soil nonlinearity.
The amplification ratio (normalized value of the response at the top of NPP structure) is significantly affected by
nonlinearity of soil near the natural period of the structure. However, at the higher period, there is not much effect of
nonlinearity.
The present study draws the useful conclusion with reference to impedance calculation of CPRF for linear, eq. lin-
ear, HSSM and MDPM soil models using the coupled BEM-FEM approach. It has been found that material nonlinearity
has significant effects on the dynamic impedances. These effects shall be definitely incorporated in the analysis and
design of NPP structures founded on CPRF. Here, a spring-dashpot model of CPRF of NPP structure is proposed and
validated for evaluating the nonlinear seismic response of NPP structure with good accuracy and less computation
time.

AC K N OW L E D G E M E N T S
The research presented in this paper is supported by a fellowship to the first author by the Ministry of Education,
Government of India. This support is gratefully acknowledged.

D A T A AVA I L A B I L I T Y S T A T E M E N T
Some or all data, models, or code that support the findings of this study are available from the corresponding author upon
reasonable request.

ORCID
Bal Krishna Maheshwari https://orcid.org/0000-0002-7306-4932
10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
FIROJ and MAHESHWARI 23

REFERENCES
1. Taherzadeh R, Clouteau D, Cottereau R. Simple formulas for the dynamic stiffness of pile groups. Earthquake Engng Struct Dyn.
2009;38(15):1665-1685.
2. Wang J, Zhou D, Liu W. Horizontal impedance of pile groups considering shear behavior of multilayered soils. Soils Found. 2014;54(5):927-
937.
3. Cemalovic M, Husebø JB, Kaynia AM. Simplified computational methods for estimating dynamic impedance of batter pile groups in
homogeneous soil. Earthquake Engng Struct Dyn. 2021;50(14):3894-3915.
4. Mamoon SM, Kaynia AM, Banerjee PK. Frequency domain dynamic analysis of piles and pile groups. J Eng Mech. 1990;116(10):2237-2257.
5. Gazetas G, Fan K, Kaynia A, Kausel E. Dynamic interaction factors for floating pile groups. J Geotech Eng. 1991;117(10):1531-1548.
6. Kaynia AM, Novak M. Response of pile foundations to Rayleigh waves and obliquely incident body waves. Earthquake Engng Struct Dyn.
1992;21(4):303-318.
7. Miura K, Kaynia AM, Masuda K, Kitamura E, Seto Y. Dynamic behaviour of pile foundations in homogeneous and non-homogeneous
media. Earthquake Engng Struct Dyn. 1994;23(2):183-192.
8. Cairo R, Conte E, Dente G. Analysis of pile groups under vertical harmonic vibration. Comput Geotech. 2005;32(7):545-554.
9. Cairo R, Conte E, Dente G. Interaction factors for the analysis of pile groups in layered soils. J Geotech Geoenviron Eng. 2005;131(4):525-528.
10. Kausel E, Roësset JM. Stiffness matrices for layered soils. Bull Seismol Soc Am. 1981;71(6):1743-1761.
11. Maheshwari BK, Emani PK. Three-dimensional nonlinear seismic analysis of pile groups using FE-CIFECM coupling in a hybrid domain
and HISS plasticity model. Int J Geomech. 2015;15(3):04014055.
12. Syed NM, Maheshwari BK. Non-linear SSI analysis in time domain using coupled FEM–SBFEM for a soil–pile system. Géotechnique.
2017;67(7):572-580.
13. Wang LH, Ai ZY. Vertical vibration analysis of a pile group in multilayered poroelastic soils with compressible constituents. Int J Geomech.
2021;21(1):04020232.
14. Chaudhary MTA. Influence of Pier stiffness degradation on soil-structure interaction in base-isolated bridges. J Bridge Eng. 2004;9(3):287-
296.
15. Maheshwari BK, Syed NM. Verification of implementation of HiSS soil model in the coupled FEM–SBFEM SSI analysis. Int J Geomech.
2016;16(1):04015034.
16. Bolisetti C, Whittaker AS, Coleman JL. Linear and nonlinear soil-structure interaction analysis of buildings and safety-related nuclear
structures. Soil Dyn Earthquake Eng. 2018;107:218-233.
17. Bhaduri A, Choudhury D. Serviceability-based finite-element approach on analyzing combined pile–raft foundation. Int J Geomech.
2020;20(2):04019178.
18. Maheshwari BK, Truman KZ, El Naggar MH, Gould PL. Three-dimensional nonlinear analysis for seismic soil–pile-structure interaction.
Soil Dyn Earthquake Eng. 2004;24(4):343-356.
19. Maheshwari BK, Truman KZ, Gould PL, El Naggar MH. Three-dimensional nonlinear seismic analysis of single piles using finite element
model: effects of plasticity of soil. Int J Geomech. 2005;5(1):35-44.
20. Maheshwari BK, Sarkar R. Seismic behavior of soil-pile-structure interaction in liquefiable soils: parametric study. Int J Geomech.
2011;11(4):335-347.
21. Sarkar R, Maheshwari BK. Effect of soil nonlinearity and liquefaction on dynamic stiffness of pile groups. Int J Geotech Eng. 2012;6(3):319-
330.
22. Katzenbach R, Moormann C, Recommendations for the design and construction of piled rafts. In International Conference on Soil
Mechanics and Geotechnical Engineering. 2001:927-930.
23. Katzenbach R, Bachmann G, Boled-Mekasha G, Ramm H. Combined pile raft foundations (CPRF): an appropriate solution for the
foundations of high-rise buildings. Slovak J Civ Eng (SJCE). 2005:19-29.
24. Yamashita K, Hamada J, Onimaru S, Higashino M. Seismic behavior of piled raft with ground improvement supporting a base-isolated
building on soft ground in Tokyo. Soils Found. 2012;52(5):1000-1015.
25. Nguyen DDC, Jo SB, Kim DS. Design method of piled-raft foundations under vertical load considering interaction effects. Comput Geotech.
2013;47:16-27.
26. Salciarini D, Ronchi F, Cattoni E, Tamagnini C. Thermomechanical effects induced by energy piles operation in a small piled raft. Int J
Geomech. 2015;15(2):04014042.
27. Lee J, Park D, Park D, Park K. Estimation of load-sharing ratios for piled rafts in sands that includes interaction effects. Comput Geotech.
2015;63:306-314.
28. Bhaduri A, Choudhury D. Steady-state response of flexible combined pile-raft foundation under dynamic loading. Soil Dyn Earthquake
Eng. 2021;145:106664.
29. Firoj M, Maheshwari BK. Effect of CPRF on nonlinear seismic response of an NPP structure considering raft-pile-soil-structure-interaction.
Soil Dyn Earthquake Eng. 2022;158:107295.
30. Saxena N, Paul DK, Kumar R. Effects of slip and separation on seismic SSI response of nuclear reactor building. Nucl Eng Des. 2011;241(1):12-
17.
31. Saxena N, Paul DK. Effects of embedment including slip and separation on seismic SSI response of a nuclear reactor building. Nucl Eng
Des. 2012;247:23-33.
10969845, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eqe.3794 by Indian Institute Of Tech - Roorkee, Wiley Online Library on [12/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
24 FIROJ and MAHESHWARI

32. Bhaumik L, Raychowdhury P. Seismic response analysis of a nuclear reactor structure considering nonlinear soil-structure interaction.
Nucl Eng Des. 2013;265:1078-1090.
33. Kumar S, Raychowdhury P, Gundlapalli P. Response analysis of a nuclear containment structure with nonlinear soil–structure interaction
under bi-directional ground motion. Int J Adv Struct Eng (IJASE). 2015;7(2):211-221.
34. Abell JA, Orbović N, McCallen DB, Jeremic B. Earthquake soil-structure interaction of nuclear power plants, differences in response to
3-D, 3× 1-D, and 1-D excitations. Earthquake Engng Struct Dyn. 2018;47(6):1478-1495.
35. Borbón DF, Domizio M, Ambrosini D, Curadelli O. Influence of various parameters in the seismic soil-structure interaction response of a
nuclear power plant. Eng Struct. 2020;217:110820.
36. Patil G, Choudhury D, Mondal A. Three-dimensional soil–foundation–superstructure interaction analysis of nuclear building supported
by combined piled–raft system. Int J Geomech. 2021;21(4):04021029.
37. Veeraraghavan S, Bolisetti C, Slaughter A, et al. MASTODON: an open-source software for seismic analysis and risk assessment of critical
infrastructure. Nucl Technol. 2021;207(7):1073-1095.
38. Maheshwari BK, Firoj M, Equivalent linear spring-dashpot model for embedded foundations of NPP. In 17th World Conference on
Earthquake Engineering, 17WCEE Sendai, Japan. 4c-0014, 2020.
39. Veletsos AS, Verbič B. Vibration of viscoelastic foundations. Earthquake Engng Struct Dyn. 1973;2(1):87-102.
40. Chopra AK. Dynamics of Structures: Theory and Applications to Earthquake Engineering. 4th ed. Prentice Hall; 2011.
41. Jian L, Shenjie Z, Meiling D, Yuqin Y. Three-node Euler-Bernoulli beam element based on positional FEM. Procedia Eng. 2012;29:3703-3707.
42. Wu G, Finn WL. Dynamic elastic analysis of pile foundations using finite element method in the frequency domain. Can Geotech J.
1997;34(1):34-43.
43. Genes MC, Kocak S. Dynamic soil–structure interaction analysis of layered unbounded media via a coupled finite element/boundary
element/scaled boundary finite element model. Int J Numer Methods Eng. 2005;62(6):798-823.
44. Masoumi HR, François S, Degrande G. A non-linear coupled finite element–boundary element model for the prediction of vibrations due
to vibratory and impact pile driving. Int J Numer Anal Methods Geomech. 2009;33(2):245-274.
45. Padrón LA, Aznárez JJ, Maeso O. 3-D boundary element–finite element method for the dynamic analysis of piled buildings. Eng Anal
Boundary Elem. 2011;35(3):465-477.
46. Duncan JM, Chang CY. Nonlinear analysis of stress and strain in soils. J Soil Mech Found Div. 1970;96(5):1629-1653.
47. Kondner RL. Hyperbolic stress-strain response: cohesive soils. J Soil Mech Found Div. 1963;89(1):115-143.
48. Janbu N, Soil compressibility as determined by odometer and triaxial tests. In Proceedings of European Conference of SMFE 1963;1:19-25.
49. ABAQUS. Abaqus Analysis User’s Manual. Dassault Systèmes Simulia Corp.; 2019.
50. Kaynia AM. Dynamic Stiffness and Seismic Response of Pile Groups. Massachusetts Institute of Technology; 1982. Doctoral dissertation.
51. Dezi F, Carbonari S, Leoni G. A model for the 3D kinematic interaction analysis of pile groups in layered soils. Earthquake Engng Struct
Dyn. 2009;38(11):1281-1305.
52. Seed HB, Idriss IM. Soil Moduli and Damping Factors for Dynamic Response Analyses. Earthquake Engineering Research Center, University
of California; 1970. Report No. EERC 70-10.
53. Horikoshi K, Matsumoto T, Hashizume Y, Watanabe T, Fukuyama H. Performance of piled raft foundations subjected to static horizontal
loads. Int J Phys Model Geotech. 2003;3(2):37-50.
54. Eslami MM, Aminikhah A, Ahmadi MM. A comparative study on pile group and piled raft foundations (PRF) behavior under seismic
loading. Comput Methods Civ Eng. 2011;2(2):185-199.
55. Paul DK, Saxena N. Dynamic Uplift Analysis of Reactor Building. Department of Earthquake Engineering, IIT Roorkee; 2005. Project Report.
56. Imai T, P-and S-wave velocities of the ground in Japan. Proceeding of IX International Conference on Soil Mechanics Foundation and
Engineering, 1977;2: 127-132
57. Hasancebi N, Ulusay R. Empirical correlations between shear wave velocity and penetration resistance for ground shaking assessments.
Bull Eng Geol Environ. 2007;66(2):203-213.
58. Vucetic M, Dobry R. Effect of soil plasticity on cyclic response. J Geotech Eng. 1991;117(1):89-107.

How to cite this article: Firoj M, Maheshwari BK. A new nonlinear spring-dashpot model of CPRF of NPP
structure based on coupled BEM-FEM approach. Earthquake Engng Struct Dyn. 2022;1-24.
https://doi.org/10.1002/eqe.3794

You might also like