Scholars' Mine

Masters Theses Student Theses and Dissertations

Spring 2010

Nonlinear finite element formulation of the soil structure

interaction through two parameter foundation model
Taraka Ravi Shankar Mullapudi

Follow this and additional works at: https://scholarsmine.mst.edu/masters_theses

Part of the Civil Engineering Commons

Department:

Recommended Citation
Mullapudi, Taraka Ravi Shankar, "Nonlinear finite element formulation of the soil structure interaction
through two parameter foundation model" (2010). Masters Theses. 6938.
https://scholarsmine.mst.edu/masters_theses/6938

This thesis is brought to you by Scholars' Mine, a service of the Missouri S&T Library and Learning Resources. This
work is protected by U. S. Copyright Law. Unauthorized use including reproduction for redistribution requires the
permission of the copyright holder. For more information, please contact [email protected].
NONLINEAR FINITE ELEMENT FORMULATION OF THE SOIL STRUCTURE
INTERACTION THROUGH TWO PARAMETER FOUNDATION MODEL

by

TARAKA RA VI SHANKAR MULLAPUDI

A THESIS

Presented to the Faculty of the Graduate School of the

MISSOURI UNIVERSITY OF SCIENCE AND TECHNOLOGY

In Partial Fulfillment of the Requirements for the Degree

MASTER OF SCIENCE IN CIVIL ENGINEERING

2010

Approved by

Genda Chen, Advisor

Ashraf Ayoub
Roger A. LaBoube
Xiaoping Du
 © 2010
TARAKA RA VI SHANKAR MULLAPUDI
All Rights Reserved
To my professor and guru, Dr. Ashraf Ayoub, for his continued support and inspiration
 IV

PUBLICATION THESIS OPTION

This thesis is composed of two papers and has been prepared in the style as

specified by the first paper Journal of Mechanics Research Communications and the

second paper Journal of Computers and Geotechnics.

 v

ABSTRACT

The response of shallow and raft foundations is having a significant importance

due to its complex behavior because of the semi-infinite soil media. Winkler's model is

the simplest model to deal with the structure and soil. The Winkler model represents the

foundation reaction as proportional to the soil displacement at a particular point, which

results in the elasticity of the soil being the only parameter in consideration. But in reality

the soil cohesiveness is having a significant contribution in soil structure interaction, and

therefore the consideration of coupling effects of Winkler springs need to be accounted.

Most of the existing elements either consider certain parameters of the foundation or

assume an elastic beam and foundation response. In this research a new finite element

formulation was developed in which these limitations were eliminated. This improved

model can be viewed as a soil with a combination of cohesive behavior which transmits

the rotation due to bending in addition to the Winkler effect.

The non linear response of structures resting on this improved foundation model

can be analyzed by assuming that the foundation resists compression and tension. In

reality soil is very weak in tension and its tension capacity needs to be neglected, which

leads to lift-off regions at different locations. This phenomenon becomes much more

complicated by considering the inelastic soil structure behavior, which leads to a highly

nonlinear problem. In order to estimate the necessary nonlinear soil parameters. an

analytical procedure based on the Vlasov model is proposed. The presented solutions and

applications show the superiority of the proposed nonlinear foundation model.

 vi

ACKNOWLEDGMENTS

I would like to express my appreciation and gratitude to the members of my

advisory committee, Dr. Genda Chen, Dr. Ashraf Ayoub, Dr. Roger A. LaBoube and Dr.
Xiaoping Du.
I am especially grateful to Dr. Ashraf Ayoub, for his guidance, advice, and
consistent encouragement in this work. He has been a tremendous source of motivation in
realizing this work. I am grateful, to my advisory committee for their unending support as
I completed this study under stressful constraints. I am grateful, also, to Dr. Abdeldjelil
Belarbi for his guidance and support.
My heartfelt thanks go to Suriya Shanmugam, Shravan Vudumu, Amol Sawant, Kazi
Karim, Mike Murphy, Carlos Ortega, Brian Wood, Mei Wang, my fellow graduate
students and good friends. My heartfelt thanks go to Drs. Young-min You, Sang-Wook
Bae, and Gary Greene for their help and assistance during the research. A special thank
you goes to Vicki Hudgins, graduate studies specialist at Missouri S&T for her
administrative help and careful editing of this thesis.
Financial support to complete this study was provided by the University
Transportation Center at the Missouri University of Science and Technology through its
Graduate Research Assistantship Program. This support is greatly appreciated.
Finally, I am eternally indebted to my wife and family for their love,
encouragement, and assistance, without which this work would not have been possible.
 VII

TABLE OF CONTENTS

Page
PUBLICATION THESIS OPTION ................................................................................... iv
ABSTRACT ........................................................................................................................ v
ACKNOWLEDGMENTS ................................................................................................. vi
LIST OF ILLUSTRATIONS ............................................................................................. ix
PAPER
I. Inelastic analysis of semi-infinite foundation elements ............................................. I
Abstract ..................................................................................................................... I
I. Introduction ........................................................................................................ I
2. Governing Equations of Beams on Vlasov Foundations .................................... 3
3. Hellinger-Reissner Formulation of Beam on Vlasov Foundation ...................... 6
4. Evaluation of Model by Numerical Studies ....................................................... 9
4.1. Inelastic Beam on Tensionless Vlasov Foundation ................................. 9
4.2. Numerical Correlation with Experimental Results ................................ 14
5. Conclusions ...................................................................................................... 15
References................................................................................................................ I 5
2. Nonlinear Finite Element Modeling of Beams on Two-Parameter Foundations ..... 18
Abstract ................................................................................................................... I 8
1. Introduction ...................................................................................................... 18
2. Governing Equations for Beams on Two-Parameter Foundations ................... 21
2.1. Equilibrium ............................................................................................ 21
2.2. Compatibility ......................................................................................... 24
2.3. Material Constitutive Laws .................................................................... 24
3. Displacement Formulation of Beam on Two-Parameter Foundation .............. 24
4. Mixed Formulation of Beam on Two-Parameter Foundation ......................... 26
4.1. Stability of Mixed Formulation ............................................................. 29
5. Evaluation of Model by Numerical Studies ..................................................... 30
5.1. Elastic Beam on Vlasov Foundation ...................................................... 30
5.2. Inelastic Beam on Two-Parameter Foundation ...................................... 34
 viii

5.3. Shear Wall Foundation Structure ........................................................... 35

6. Conclusions ...................................................................................................... 42
References ................................................................................................................ 42
VITA ................................................................................................................................ 46
 IX

LIST OF ILLUSTRATIONS

Figure Page

PAPER I
I. Infinitesimal segment of a beam on Vlasov foundation ............................................... 4
2. (a) Reinforced concrete beam on tensionless foundation; (b) Kent and Park model
for concrete ................................................................................................................. 10
3. Load-displacement response ofRC beam on tensionless Vlasov foundation (disp.
model) ......................................................................................................................... II
4. Load-displacement response ofRC beam on tensionless Vlasov foundation (mixed
model) ......................................................................................................................... 12
5. Curvature distribution of RC beam on tensionless Vlasov foundation (disp. model) 12
6. Curvature distribution of RC beam on tensionless Vlasov foundation (mixed model)
..................................................................................................................................... 12
7. Moment distribution of RC beam on tensionless Vlasov foundation (disp. model) ... 13
8. Moment distribution of RC beam on tensionless Vlasov foundation (mixed model) 13
9. Foundation force distribution ofRC beam on tensionless Vlasov foundation (disp.
model) ......................................................................................................................... 13
I 0. Foundation force distribution ofRC beam on tensionless Vlasov foundation (mixed
model) ......................................................................................................................... 14
II. Footing moment-rotation response ofGajan eta!. specimen SSG04-06 .................... 15

PAPER 2
I. Infinitesimal segment of a beam on two-parameter foundation ................................. 21
2. Beam on elastic foundation ......................................................................................... 22
3. Beam with free ends .................................................................................................... 31
4. Moment-rotation response of free beam ..................................................................... 31
5. Vertical displacement for free beam ........................................................................... 32
6. Bending moment distribution tor free beam ............................................................... 32
7. Tensionless beam with free ends ................................................................................ 32
8. Tensionless foundation moment-rotation for free beam ............................................. 33
9. Tensionless foundation vertical displacement for free beam ...................................... 33
 X

I 0. Tensionless foundation rotation for free beam ........................................................... 33

II. Inelastic beam on tensionless foundation ................................................................... 36
12. Moment rotation response of beam on tensionless foundation (mixed model) .......... 36
13. Moment rotation response of beam on tensionless foundation (displ. model) ........... 3 7
14. Curvature distribution for beam on tensionless foundation (mixed model) ............... 3 7
15. Curvature distribution for beam on tensionless foundation (displ. model) ................ 38
16. Vertical displacement distribution for beam on tensionless foundation (mixed
model) ......................................................................................................................... 38
17. Foundation force distribution for beam on tensionless foundation (mixed model) .... 39
18. Moment distribution for beam on tensionless foundation (mixed model) .................. 39
19. Foundation rotation for beam on tensionless foundation (mixed model) ................... 40
20. Foundation moment distribution for beam on tensionless foundation (mixed model)40
21. Moment-rotation response of Gajan et al. specimen .................................................. 41
22. Foundation force distribution ofGajan et al. specimen .............................................. 41
 1. Inelastic analysis of semi-infinite foundation elements
T. Ravi S Mullapudi a, Ashraf Ayoub a,*

a Department of Civil and Environmental Engineering, Missouri University of Science

and Technology, Rolla, MO 65409, USA

Abstract

The inelastic response of shallow and raft foundations are significantly complex due to
the behavior of the surrounding semi-infinite soil media. The Winkler approach models
the soil as a single layer, and assumes that the foundation reaction at a particular point is
proportional to the soil displacement. In reality, the soil is a semi-infinite medium that
can not be modeled as a single layer. In this paper a new finite element formulation was
developed in which the soil can be viewed as a semi-infinite inelastic element that can
resist bending, in addition to the well-known Winkler effect. A parametric analysis of an
inelastic reinforced concrete foundation element is presented.

Key Word'i: Beam on foundation; Vlasov foundation; Semi-infinite foundation; Mixed

finite element; Hellinger-Reissner variational principle.
1. Introduction
The inelastic response of shallow and raft foundations are significantly complex due to
the behavior of the surrounding semi-infinite soil media. Winkler ( 1867) developed a
simple model that accounts for the behavior of both the foundation and soil. The Winkler
model represents the soil beneath the foundation as a system of similar but mutually
independent elastic springs. In this model, it is assumed that the foundation reaction at a
particular point is proportional to the soil displacement. The Winkler model is considered
therefore a single parameter model with the spring's elasticity as its only parameter.
While this model is associated with closely spaced elastic springs, in reality these springs
should be dependant on each other. To address these drawbacks, some modified

• Corresponding author. Tel.: +I 713 743 4285; fax: +I 713 743 4260.
E-mail address: [email protected] (Ashraf Ayoub).
 2

approaches have been proposed such as the model developed by Vlasov ( 1966) , which
belongs to the family of multiple-parameter foundation models.
The stiffness matrix of a beam on multiple-parameter foundation element can be
derived based on different orders of displacement shape functions or by using the exact
displacement function obtained from the solution ofthe differential equations governing
the behavior. Biot (1937) studied the foundation as an elastic continuum and gave an
exact solution to an infinite beam under a concentrated load. Kerr ( 1965) studied the
foundation response using an elastic continuum approach by connecting each two spring
layers with an interconnecting shear layer. Harr et al. ( 1969) analyzed elastic beams on
semi-infinite elastic foundations based on Vlasov general variational method. Yang
(1972) introduced a numerical iterative procedure on the basis of the finite element
method for analyzing plates on elastic foundations. Zhaohua and Cook ( 1983) developed
the finite element formulation of an elastic beam on two-parameter foundation using
both, an exact displacement function and a cubic displacement function for the case of
distributed loads acting along the entire beam length. Chiwanga and Valsangkar ( 1988)
extended the approach for the case of a generalized distributed load. Their work though
was limited to a specific combination of beam and foundation stiffnesses. Razaqpur and
Shah ( 1991) further extended the work by considering beams and foundations with any
stiffness. Shirima and Giger (1990) developed the stiffness matrix and nodal-action
column vectors for a Timoshenko beam on a two-parameter foundation element. Morfidis
and A vramidis (2002) derived the element stiffness matrix based on the exact solution of
the differential equations with the ability to account for shear deformations, semi-rigid
connections, and rigid end offsets. In most practical applications the foundation is
typically assumed to be tensionless not elastic. Kaschiev and Mikhajlov ( 1995) used the
finite element method as a general numerical technique to solve the problem of elastic
beams on tensionless foundations for different loading conditions. Coskun (2003) studied
the tensionless Pasternak formulation which results in lift off regions between the beam
and foundation, and presented the roots of a nonlinear equation to calculate the contact
length of the beam. Celep and Demir (2005) studied the tensionless behavior which
showed that the problem becomes highly non-linear due to the lift off of the beam from
the foundation.
 3

In all previously described work, the beam was assumed elastic, while the foundation
was assumed either elastic or tensionless. In reality, beams are typically made up of
reinforced concrete, which can undergo deformation in the plastic range due to
reinforcement yielding or concrete crushing. The analysis of inelastic beam on foundation
elements is very complex due to the combined effect of beam plasticity as well as the
spatial variation of soil resistance. To solve this problem displacement formulation
requires more elements and also are plagued by instability problems. Ayoub and Filippou
(2000) and Ayoub (2003) proposed a consistent mixed formulation based on a Hellinger-
Reissner variational principle for inelastic analysis of composite structures and inelastic
beams on Winkler's foundations, respectively. The mixed formulation proved to
overcome most of the difficulties associated with the standard displacement approach
derived from a minimum potential energy principle (Zhaohua and Cook, 1983), and to
provide a more efficient numerical platform for analysis of these types of structures. In
this paper, the mixed approach was formulated for the problem of beams resting on semi-
infinite foundations by adopting a Vlasov approach, in which the soil parameters are
determined based on a plane strain approach.
In the next sections, Hellinger-Reissner finite element formulation for beams on
Vlasov foundations are developed. The model is implemented in the finite element
program FEAP, developed by R.L. Taylor, and described in details in Zienkiewicz and
Taylor (1989). Numerical examples that compare the behavior of both models are then
performed, and conclusions based on these results are derived. The governing equations
of beams on Vlasov foundations are presented first.
2. Governing Equations of Beams on Vlasov Foundations
The equilibrium of an element of length dx of a beam element resting on a Vlasov
foundation as shown in Fig. I is given by:
 4

w
I__ LI:::I::"TT~
M I' I' dl' M dM

( )~~-~~ )

Fig. I. Infinitesimal segment of a beam on Vlasov foundation

V:x + (W- p) = 0
(1)
M ,X +V=O

Where V, and M denote the shear force and bending moment, respectively; p is the
total foundation load per unit length acting on the beam, w denotes the distributed load
on the beam, and a comma denotes derivation. According to Vlasov's hypothesis, and
assuming linear soil behavior, the foundation load per unit length is related to the
transverse displacement as follow:
p = k 1v(x)- knr v.u (x) (2)

Where v is the vertical displacement of the beam, k1 is the Winkler's modulus and km
is Vlasov's parameter that depends on both the soil and foundation characteristics. For
inelastic behavior, both parameters are based on nonlinear functions as will be described
later. The foundation force term corresponding to Vlasov's parameter can be viewed as
an additional moment resistance provided by the foundation following elementary thin-
plate theories. Accordingly, the Winkler force t 1 , and Vlasov force t, which can be

viewed as a moment applied to the beam by foundation, are defined respectively as

follows:
t1 = krv
(3)
tm = k.,v,x

From Eqs. (I )-(3):

M .xx -tm,x +tf -w=O (4)
The values of the two foundation parameters k1and km are evaluated as proposed by
Vlasov ( 1966). These equations account for the semi-infinite dimension of the underlying
 5

soil by adopting a plane strain approach. In general the soil beneath the foundation has
different stratums with different thicknesses and different soil properties. As a result, the
normal stress in the soil changes with respect to the depth. Vlasov derived an equation
that represents the different layer properties with a single equivalent layer with elastic
modulus and Poisson ratio of E11 and v0 , respectively. The normal stress is assumed to be

constant within this equivalent layer. According to Vlasov analysis, Zhaohua and Cook
( 1983) evaluated the parameters for foundations of infinite depth as follow:

(5)

where
l ,J''
A= 2D(I-l',;l
f"h
and D= Eh bh' ,
12(1-v;)
(6)

The elastic constants of the equivalent layer Eo and v0 are defined as:

E v (7)
E =-'-and v =-.,-
o 1-vX2 0 1-vX

y is a coefficient that depends on the elastic properties of the foundation and

determines the rate of decrease of displacements over the foundation depth: hand hare
the width and height ofthe beam section, respectively. Eb andEs are the elastic moduli of
the beam and soil, respectively and uh and Us are the Poisson ratios of the beam and soil,
respectively.
The curvature at a section x is related to the transverse displacements by:

"-xx- X= 0 (8)

where x is the section curvature.

The internal moment ofthe beam M(x) is related to the curvature x by a nonlinear
constitutive relation

M(x) = g(x(x)) (9)

In this study the nonlinear relation in Eq. (9) is derived from a discretization of the
cross section of the beam into several fibers with nonlinear uniaxial stress-strain relations
 6

fc1r the constituent materials. For a nonlinear soil rather than the elastic one assumed by
Ylasov. the two f()l(ndation forces t 1 and 1,11 are related to their respective deformations

by two other non Iin ear relations as follows:

(I 0)

In the next sections. the strong form Eqs. (1)-(4) and (8)-(10) are solved for using the
finite element method. Due to the nonlinear nature of Eqs. (9) and (I 0). a Newton-
Raphson iteration strategy is used. The following discussion refers to a single Newton-
Raphson iteration denoted by subscript i.
3. Hellinger-Reissner Formulation of Beam on Vlasov Foundation
In a I Jell inger-Reissner formulation. the differential equations are solved based on
both a displacement and a force field (mixed formulation). For the foundation problem.
Ayoub (2003) proved earlier that this mixed approach is very advantageous from a
numerical standpoint. Accordingly:

v(x) = a(x)v (I I)

In addition M(x) = b(x)M ( 12)

where M (x) is the bending moment. and b(x) is a matrix of""' force interpolation

functions, and M is the vector of element end moments.

The finite element formulation is derived by considering the f(1llowing two variational
equations that correspond to the compatibility Eq. (8) and the equilibrium Eq. ( 4):

c)/ 1, = J(5M' (xJ[v"- x}In= 0 (13)

n

81,. = fc)\• 1 (x)[M" +1 1 -1 11 , , -H']dO=O n = [o. L] ( 14)

!l

Using an incremental Newton-Raphson iterative technique to solve the nonlinear Eq.

( 13 ):

5 r.\/ = ('> r-l

.\1 eM 5 r-l
+ _!!___ \1
dM' = o (IS)
 7

The left hand side of Eq. ( 15) is:

t5I;, = ft5M' (.yJ[ v',.,- x' ]dn = 0 n = [o. L] ( 16)

n

Substituting the predefined force interpolation functions into Eq. ( 16 ):

t5I;, =t5M' fb'(xJ[v~,,-x']dn=O O=[O,L) ( 17)

n

The second term in Eq. ( 15) is equal to:

. , 6 1;7 1dM'
f-
cM
= [ (5M 1 ( Jb(x)' cr~,(x
n
;) ) ~v
;) 1/-l dQ-
cv eM
Jb(x) 1
n
.,
~x(x)
eM (x)
1/-l dQ) l
dM'

= [6M 1 ( Jb(x)'a,,(x) cJv 1'- 1

dQ- Jb(x) 1 (.f- 1 )'- 1 (x)b(x)dQ)ldM'
n cJM n

n =[o.L] (18)
Eq. ( 18) could be written as:

~c51,'-1 1 dM' =c5M' <Jb(x)' a (x)dQ)dv' -JM' <Jb(x)' (k'- 1 r- (x)b(x)dQ)dM'

1
aM · ll
··'·' ll

O=[O.L) (19)

The first term in Eq. (15) is equal to:

c51;7 1 =8M' [- Jb' (X)X'- 1(.Y)d0+ Jb'v'~,1 (x)dQ]

ll ll
O=[O.L) (20)

=i>'M' [- Jb' (X)X'-'(x)d0+ Jb'a.n(x)d0\·'- 1 ]

- ll !l

Substituting Eqs. ( 19) and (20) into Eq. ( 15) and using the arbitrariness of 8M yields:

(21)

where T = Jb 1 (x)a,,(x)dQ O=[O.L] (22)

(l
 8

F'- 1 = fb 1 (x)f'- 1 (x)b(x)dQ, f(x) = f:'- 1(x) = k- 1 (x), Q = (0, L) (23)

i!

v;.- 1 = fb 1 (x)x'- 1(x)dQ-Tv'- 1 Q = (0, L) (24)

n

F is the element flexibility matrix, and vr is the element residual deformation vector.

Similarly. using an incremental Newton-Raphson technique to solve the nonlinear Eq.

( 14):

(25)

The second term of the right hand side of Eq. (25) is equal to:

!_SI'- dv' =liiv'(Ja(x) 1 b(xl'''\1 (x)r·- 1 d0.+ fa(.x)' cl 1 (x)l'- + fa(x) 1 <7l,,.(x)
1
1
av'(x)~·- 1 d0.)1d,'
()\' I' U 1\ (]\• U iJ\'' f) I av'(x) rl\' J

~ r'iv 1 lJ aL>d,, b(x

n
)dn] riM (x)' + l J a(x ) 1 k'1 ,_ 1 ( x)a( x)dU + J a (x ) 1,
n u
,g;,,'- 1 ( x )a ( x) , d!l )] dv'

O=[O,L) (26)

The first term of the right hand side of Eq. (25) is equal to:

/il:- 1 = dv 1 [ fa(x) 1,, b(x)M'- 1 (x)dQ + Ja(x)1 t~- 1 (x)dQ + fa(x) 1,(,~ 1 (x)dQ +BTl
o o n

n=[OJ] (27)

Substituting Eqs. (26) and (27) in Eq. (25), and from the arbitrariness of Sv. we get:

T / (IM' + K'-I 1( / v + K'- 1 /v'

I
Ill (
= P-T 1 M'- 1 -M'-I 1 -MH Ill (28)

Where K7 = fa
1 1 (x)k~- 1 (x)a(x)dx is the Winkler foundation element stiffness matrix.
()

k 1 (x) = g>(x) is the foundation force tangent stiffness term.

I

K;,~ 1 = Ja 1, (x)k,',~ 1 (x)a, (x)d• is the Vlasov foundation element stitTness matrix.
II
 9

k",(x) = ,~;"(x) is the foundation moment tangent stiffness term, M'1- 1 = Ja 1 (x)t;- 1(x)d\
()

is the Winkler foundation element resisting load vector, M;,~ 1 = Ja~,(x)t;,~ 1 (x)cl\ is the

Vlasov foundation element resisting load vector, T is as defined before in Eq. (22), and P
is the vector of applied external loads accounting for the boundary term BT.
Writing Eqs. (21) and (28) in matrix form:

[-;: (29)

It is important to note that at convergence, the residual deformation vector v,. reduces

to zero inside each clement satisfying compatibility. As discussed by Ayoub (200 I), two
algorithms for the mixed formulation exist. In the first algorithm, the system of equations
in Eq. (29) is solved for globally with the displacements and moments as degrees of
freedom. This algorithm, however, results in oscillations of results. In the second
algorithm, the moment degrees of freedom are condensed out from the first of Eqs. (29)
at the element level resulting in a generalized displacement stiffness matrix. Accordingly:

An internal element iteration is required in order to zero the residual deformation

vector v r in every element. In addition, in accordance with the Babuska-Brezzi (B-B)
stability condition ( 1973, 1974), the order of the displacement interpolation functions
needs to be larger by two than that of the force interpolation functions. The algorithm
used lt)llows the same procedure as the one discussed in Ayoub (200 I). Numerical
examples to evaluate the perltmnance ofthe model are presented next.
4. Evaluation of Model by Numerical Studies
4.1. Inelastic Beam on Tensionless V/asov Foundation
The numerical example represents reinforced concrete beams resting on tensionless
foundations, which arc quite common in structural design. These primary and secondary
beams are useful to connect different columns such that the load will be transferred to the
 10

soil more evenly. The reinforced concrete beam, shown in Fig. 2(a), has a length of I 0 m,
and a rectangular cross-section with a width and depth of 400 mm and 500 mm.
respectively. The beam is having 4 No.I 0 (32.26 mm diameter) longitudinal bars at the
top and bottom. These bars are uniformly spaced along the width with a concrete cover of
50mm. No.3 (9.53 mm diameter) steel stirrups are provided with a spacing of200 mm.
The concrete compressive strength is assumed to be 42.0 MPa (6 ksi). Each beam section
is discretized into 16 fibers with uniaxial stress-strain relations for the constituent
materials, as discussed in Eq. (9). The concrete stress-strain behavior followed the Kent
and Park model ( 1971) (Fig. 2(b )), and assuming its Young's modulus to be Eh=25113
MPa and its Poisson ratio to be vb =0.2. The steel's stress-strain curve is assumed to be

elastic perfectly plastic with a yield stress of 413MPa (60 ksi) and a Young's modulus of
200,000 MPa (29,000 ksi). The underlying soil is Ottawa Sand with properties as given
by Park and Desai (2006) as follows: elastic modulus Es= 193 MPa, Poisson ratio,., c~o.4,

and r =1.5. The soil behavior is assumed tensionless and elasto-plastic in compression.

From Eqs. (5)-(7) the values of the foundation elastic parameters were found to be: k1
=113.2 MPa and k111 =1 0,076,000 N. A central vertical load is applied proportionally
under displacement control. A mesh consisting of62 displacement-based elements with
fifth order polynomials was shown to represent the converged solution.

(b) f
f''
I'
. A : •
.,!1111J1)lll

• • • I
I'" ]oo),,,,,,ludn••ll•,,

(II) ' :"11(111)111

'" ,,, ],.,,,.,,,,],,.,] "·"

I
~ lz

... r
IUm
l'J.l (I \1Pa . A
... f f I,ooo2-
z, I ooo2'' '
ll ~I,,

I' f f 1 Zt• -0 002)

-~~~~- -------·------
\' " llllfl:' 1:-:0 1:

Fig. 2. (a) Reinforced concrete beam on tensionless foundation: (b) Kent and Park
model for concrete.

Figs. 3- 10 show the behavior ofboth the displacement and mixed models using 8
elements with different order of interpolation functions. Figs. 3 and 4 show the load -
midspan displacement behavior of the Vlasov beam using the displacement and mixed
 II

formulations, respectively. For the displacement formulation, two shape functions were
used. namely cubic hermitian polynomials and fifth order polynomials. For the mixed
model, two force interpolation functions were used, namely linear bending moment
functions along with the hermitian displacement functions: and cubic moment functions
along with the fifth order displacement functions. In the figures points A. A· and A"
represent yield points for the converged. higher order and lower order solutions.
respectively. Similarly, points 8, B' and B" represent the corresponding points for the
ultimate state. The plots reveal that the mixed model can capture the yielding point more
accurately, while having a higher convergence rate than the displacement model. In fact.
tor the displacement model, little difference could be observed between using the low
order hermitian functions and the fifth order polynomials. while 14 mixed elements with
cubic moments were sunicient to reach convergence. The superiority ofthe mixed model
could be explained by considering the distribution ofthe local parameters along the
length ofthe beam, namely the curvature, bending moment, and f(mndation forces. These
parameters are shown in Figs. 5- I0 for both the displacement and mixed models.

3000
- ·- · • Otsp. model (cub1c dlsp.)
2500 · · · · Dsp. model (5th order dsp.)
-Converged solut1on

2000
'j
"'
.; 1500
§
-'
1000

4 6 8
~lid<ipan displacemrnt (n•n)

Fig. 3. Load-displacement response of RC beam on tensionless Vlasov t()LIIldation

(disp. model)
 12

3000

2500

2000
r
-=- 1500
""'
0

1000

500

6
Midspan displan~mfnt (nun)

Fig. 4. Load-displacement response of RC beam on tensionless Vlasov foundation

(mixed model)

25000
- - ·. Dsp_ model (cut:Jc d1sp.)
70000

15000
- .. --- Dsp. model (5th ordP.r disp.)

-- - Con-.erged solutiOn r1
10000

5000 ;~,\
0 ---~~ ~~-----
-5000
10

l.ength alonl.! ht.·:tm (m)

Fig. 5. Curvature distribution of RC beam on tensionless Vlasov foundation (disp.

model)

25000

20000 M1xt>d 1110del (cubic 1liOtTlenl)

- - (:onwrqr·d .solut1on

15000

~ 10000

;
5000
'
-=

-5000
10

I .t·ngth o1long IM·am (m)

Fig. 6. Curvature distribution of RC beam on tensionless Vlasov foundation (mixed

model)
 13

900
- ·--- Dlsp. model {cubic d1sp.)
'2 BOO
·--- ·- Dsp. model (5th order disp.) . 1 ~­
; 700
~-Converged solut1on
-::.. 600
~ 500
§ 400
=
~J;
300
:.; 200
~ 100
=
10

l.t•ngth along lx·am (m)

Fig. 7. Moment distribution ofRC beam on tensionless Vlasov foundation (disp.

model)

900
500 ---·-Mixed modelllmear nmnlfln!)
~ 100 M1xed model iCUbiC moment)
~ 500 - Con1.erged solut1on r 1A',
- SQQ ,: ) \I
o ; I \
E
E4oo
300
·'/
.
1' ·.~ I

~- 200 r· / \\~_
~100
~ 0 ----·--~=~'--(,(
1 ~
\,;.._-~----~
"' -100 '
-200
0 10

I~nJ!_dt aim~ bram(m)

Fig. 8. Moment distribution ofRC beam on tensionless Vlasov foundation (mixed

model)

~\ /-
=
0 -800
'\\. /
;·
0
'\_j
.
] -1000

-1200
---
- Dsp model 1cub1c d1sp)
Dsp model (Sih O!der d1<;p)
Con~rged solut1on

-1400
0 10

l£~th alollJ! lx-am (Ill)

Fig. 9. Foundation force distribution of RC beam on tensionless Vlasov foundation

(disp. model)
 14

-200
5
; -400
-="
- -600
~
·~ -800
-o
§ -1000
~
-1200 ~.: ~---~ ~:;:~ ~~~= ~~~~~~ ~~~~~g
--Converged solutiOfl

-1400
0 4 10
l.t>nglh itlone beam (m)

Fig. I0. Foundation force distribution of RC beam on tension less Vlasov foundation
(mixed model)

Figs. 5 and 7 show the curvature and bending moment distributions for the
displacement model using both the hermitian cubic polynomials and lifth order
polynomials. Due to the high value of the bending moment at midspan. the curvature is
localized in this region as shown in Fig. 5. The displacement model can not capture this
phenomenon accurately. even if using the higher order formulation. since it is based on
polynomial displacement shape functions. Fig. 9 shows the foundation force distribution
ofthe Vlasov beam with the displacement model.
The same plots were repeated for the mixed model. Figs. 6 and 8 show the curvature
and bending moment distributions. respectively. Since the mixed model is based on
approximating the smooth bending moment through the force interpolation functions. it
can accurately represent the curvature localization in the plastic zone. as shown in Fig. 6.
In tact. the higher order mixed formulation with cubic moment distributions produce
results almost identical to the converged solution. The t()undation vertical l(wcc is shm\ n
in Fig. I 0. The deflected shape. as well as the lift off at the beam ends\\ here the
foundation force vanishes. was captured rather well.
4.2. Numerical Correlation with Experimental Re.mlts
Numerical analysis using the mixed model was conducted for the Aluminum shear
wall foundation structure SSG04-06 tested by Gajan et al. (2006) under an increasing
wall lateral load. The footing is 2.8 m x 0.65 m. and has a Young·s modulus of70.000
MPa. The underlying soil is Nevada sand with modulus of elasticity 45 MPa. Poisson
ratio 0.4. The yield ofthe soil is assumed to be at 35% of its bearing capacity. From Fqs.
 15

(5)- (7) the values ofthe foundation parameters were found to be: y =1.5. k1 =13.5 MPa
and km= 12. 170,000 N. Fig. II shows the monotonic envelope of the moment-rotation plot
at the middle of the foundation for both the Vlasov and well-known Winkler models, as
well as the experimental results. From the figure. it is observed that the Vlasov model is
able to predict the behavior reasonably well, while the Winkler model under-predicts the
foundation moment capacity.

--VI.'"'\ \1o,kl
\\, 111~k1 r>..1odc:l
• 1\)'elllllO.::Illc'IIIC:]\l)'

......... .......
... ............ -
,.
..•
... .......
~

...

Fig. II. Footing moment-rotation response of Gajan et al. specimen SSG04-06

5. Conclusions
This paper presents a new inelastic element for the analysis of semi-infinite foundation
problems. The element is derived from a two-field mixed formulation, where f'lxces and
deformations are approximated with independent interpolation functions. The nonlinear
response ofthe foundation is analyzed following a Vlasov approach to represent the
semi-infinite soil medium. Numerical examples to evaluate the performance of the model
were conducted. The studies revealed the superiority of the proposed mixed model in
evaluating the inelastic complex behavior of these types of structures.

References
Ayoub, A.S .. 200 I. A Two-Field Mixed Variational Principle for Partially Connected
Composite Beams. Finite Elements in Analysis and Design. 37, 929-959.

Ayoub. A., 2003. Mixed formulation of nonlinear beam on t()undation elements.

Computers and Structures. 81 , 411--421.
 16

Ayoub, A.S., Filippou, F.C., 2000. Mixed Formulation ofNonlinear Steel-Concrete

Composite Beam Element. Journal of Structural Engineering, ASCE. 126(3) , 3 71-
381.

Babuska, I. 1973. The Finite Element Method with Lagrange Multipliers. Num. Math.,
20, 179-192.

Brezzi, F. 1974. On the Existence, Uniqueness and Approximation of Saddle Point

Problems Arising from Lagrangian Multipliers. RAJ RO, 8-R2, 129-151.

Biot, M.A., 1937. Bending ofan infinite beam on an elastic foundation. J. Appl. Mech.
A4 J-7.

Celep, Z., Demir, F., 2005. Circular rigid beam on a tensionless two-parameter elastic
Foundation. Zamm-Zeitschrift fUr Angewandte Mathematik und Mechanik. 85(6),
431-439.

Chiwanga, M.,Valsangkar, A.J., 1988. Generalized beam element on two-parameter

elastic foundation. J. Struct. Engng. 114(6), 1414-27.

Coskun. 1., 2003. The response of a finite beam on a tensionless Pasternak foundation
subjected to a harmonic load. European Journal of Mechanics- A/Solids. 22( I), 151-
161.

Gajan, S., Phalen, J., Kutter. Band Kutter, B., 2006. Soil-foundation-structure
interaction: shallow foundations. Center for Geotechnical Modeling, University of
California, Davis, CA

Harr. E.M., J.L.Davidson and S.V.Ramaswamy .. 1969. Euler Beams on Two Parameter
Foundation Model. .Journal of the Soil Mechanics and Foundations Division. ASCL
95(4) .933-948.

Kaschiev, M.S.,Mikhajlov. K.. 1995. A Beam Resting on a Tensionless Winkler

Foundation. Computers and Structures. 55(2). 261-264.

Kent, D.C., Park, R.. 1971. Flexural Members with Confined Concrete . .1. Struct. Engrg ..
ASCE. 97(ST7).1969-1990.

Kerr. A.D., 1965. A study of a new foundation model. Acta Mechanica. 1(2) .135-147.
 17

Morfidis, K., Avramidis, I.E., 2002. Formulation of a generalized beam element on a

two-parameter elastic foundation with semi-rigid connections and rigid offsets.
Comp. and Struct. 80, 1919-1934.

Park, I. J., Desai, C.S., 2006. Cyclic Behavior and Liquefaction of Sand Using Disturbed
State Concept, Journal of Geotechnical & Geoenvironmental Engineering.
126(2),834-846.

Razaqpur, G.A .. Shah, K.R., 1991. Exact analysis of beams on two-parameter elastic
foundation. Int J Solids Struct. 27, 435-454.

Shirima, L.M.,Giger, M. W., 1990. Timoshenko beam element resting on two-parameter

elastic foundation. J Engng. Mech. 118(2), 280-95.

Vlasov ., Leontiev, U.N .. 1966. Beams, plates. and shells on elastic foundation.
Jerusalem: Israel Program tor Scientific Translations [translated from Russian!.

Winkler, E., 1867. Theory of Elasticity and Strength. Dominicus Prague. Czechoslovakia.

Yang, T.Y ., 1972. A finite element analysis of plates on a two parameter foundation
model. Computers and Structures. 2, 593-614.

Zhaohua, F., Cook. R. D., 1983. Beam elements on two parameter elastic foundations.
Journal of Engineering Mechanics, ASCE. I09( 6). 1390-1402.

Zienkiewicz. 0. C.,Taylor. R. L.. 1989. The Finite Element Method Volume I. Basic
Formulation and Linear Problems. Fourth Edition. McGraw Hill. London.
 18

2. Nonlinear Finite Element Modeling of Beams on

Two-Parameter Foundations
T. Ravi S Mullapudi a and Ashraf Ayoub a,*

a Department of( 'il'il and Environmental Engineering, Missouri University o(.)'cience

and Techno/oA'Y· Rolla, MO 65.:/09, USA

Abstract
This paper presents an inelastic element for the analysis of beams resting on two-
parameter foundations. The element is derived from a two field mixed formulation with
independent approximation of forces and displacements. The values for the two
parameters of the foundation are derived through an iterative technique that is based on
an assumption of plane strain for the soil medium. This iterative behavior is repeated at
each time step of the nonlinear solution algorithm. The nonlinear response of structures
resting on this improved two-parameter foundation model is analyzed following both a
Vlasov and a Pasternak approach. Numerical examples that clarify the advantage of the
newly developed model are conducted. These studies confirmed the importance of
accounting for the foundation second parameter. and the etriciency and accuracy ofthe
proposed model.

Key Words: Two-parameter foundation; Winkler foundation; Pasternak foundation:

Vlasov foundation: Mixed finite element.
1. Introduction
The inelastic response of shallow and raft foundations is significantly complex due to
the behavior of the surrounding semi-infinite soil media. Winkler's model [I] is the
simplest element that account tor the behavior of both the foundation and soil. The
Winkler approach models the soil as a single layer, and assumes that the foundation
reaction at a particular point is proportional to the soil displacement. The Winkler model
is considered therefore a single-parameter model with the spring's elasticity as its only
parameter. While this model is associated with closely spaced independent elastic

• Corresponding author. Tel.: +I 713 743 4285.

r,·-mail address: asayoub(i1!uh.edu (Ashraf Ayoub).
 19

springs, in reality these springs should be dependant on each other. To address these
drawbacks, several modified approaches have been proposed such as the ones developed
by Filonenko-Borodich [2], Hetenyi [3], Pasternak [4], and Vlasov and Leontiev (5].
These models belong to the family of multiple-parameter foundation models because. in
addition to the first parameter spring's elasticity, they assumed a second parameter that
accounts tor the effect of the adjacent soil medium. In this paper, the Pasternak (4]. and
Ylasov and Leontiev [5] models were adopted. In the Pasternak model, the shear
interaction between the Winkler spring elements is considered. The shear layers consist
of incompressible vertical elements which deform only in transverse shear. In the Ylasov
and Leontiev model, the second parameter is considered by extending the Pasternak
spring elements with a consideration ofthe effect ofthe soil on both sides of the beam.
The stiffness matrix of an elastic beam on multiple-parameter foundation element can
be derived based on different orders of displacement shape functions or by using the
exact displacement function obtained from the solution of the ditTerential equations
governing the behavior. Biot [61 studied the foundation as an clastic continuum and
derived an exact solution to an infinite beam under a concentrated load. Kerr f71 studied
the foundation response using an elastic continuum approach by connecting each two
spring layers with an interconnecting shear layer. Reissner [8] formulated the problem of
an elastic plate on an elastic foundation by assuming a transition condition at the interior
of the foundation layer along the cylindrical surface. Harr et al. [9] analyzed beams on
elastic foundations based on Vlasov general variational method in which the elastic
foundation is represented by a single layer. Yang [I 0] introduced a numerical iterative
procedure on the basis of the finite element method for analyzing plates on clastic
foundations. Zhaohua and Cook [II] developed the finite element formulation of an
elastic beam on two-parameter foundation using both. an exact displacement function.
and a cubic displacement function tor the case of distributed loads acting along the entire
beam length. Chiwanga and Yalsangkar [ 12] extended the approach for the case of a
generalized distributed load. Shirma and Giger [15] developed the stiffness matrix and
nodal-action column vectors tor a Timoshenko beam on two-parameter foundation
element. Razaqpur and Shah f 14] derived the stiffness matrix and nodal load vector of an
element representing a beam on two-parameter elastic foundation using polynomial
 20

displacement shape functions. Vallabhan and Das [15] developed a unique iterative
technique to determine the values of the Vlasov parameters used in [9]. De Rosa [ 16]
studied the free vibration ofTimoshenko beams on two-parameter foundations by
considering the second parameter either as the total rotation of the beam or the bending
rotation only. Hou and Tseng [ 17] used cubic polynomial expressions tor the total
deflection and bending slope of the beam, and derived the mass and stiffness matrices of
the element from energy expressions. Patel et al. [ 18] further modified the formulation of
beam on two-parameter elements by considering geometric nonlinearity with Von
Karman's strain-displacement relations. Onu [19] presented a formulation with an
explicit free of meshing stiffness matrix with the consideration of shear deformation
effects. Morfidis and A vramidis [20] derived the element stiffness matrix based on the
exact solution of the differential equations with the ability to account tor shear
deformations, semi-rigid connections, and rigid offsets. Coskun [21] studied the
tensionless Pasternak formulation which results in lift-off regions between the beam and
foundation, and presented the roots of a nonlinear equation to calculate the contact length
of the beam. In most practical applications though, the foundation is typically assumed to
be tensionless not elastic. Celcp and Demir [22] studied the tensionless behavior of
beams resting on Winkler foundations, which showed that the problem becomes highly
nonlinear due to the lift-offofthe beam from the foundation. Kaschiev and Mikh~jlov

[23] used the finite element method as a general numerical technique to solve the
problem of elastic beams on tensionless foundations for different loading conditions.
Ayoub and Filippou [24] and Ayoub [25] proposed a consistent mixed formulation f(lr
inelastic analysis of composite structures. The mixed formulation proved to overcome
most of the ditlicultics associated with the standard displacement approach and to
provide a more etlicient numerical platf(mn for analysis of these types of structures.
Ayoub [26] confirmed the advantages of the mixed formulation over displacement-based
formulations for beams on Winkler foundations.
The objective of this paper is to adopt the numerically efficient mixed formulation tor
developing a new element for inelastic analysis of beams resting on two-parameter
foundations. In the next sections, both a displacement and a mixed finite clement
formulation tor beams on two-parameter foundations are developed. The models arc
 21

implemented in the finite element program FEAP, developed by R.L. Taylor, and
described in details in Zienkiewicz and Taylor [27]. Numerical examples that compare
the behavior of both models are then performed, and conclusions based on these results
are derived. The governing equations of beams on two-parameter foundations are
presented first.
2. Governing Equations for Beams on Two-Parameter Foundations
2. I. Equilibrium
The equilibrium of an element of length d"t of a beam element resting on a two-
parameter foundation, as shown in Fig. (I), is given by:

v.+(w-t.)c=(}
.r I (I)

M ,X +V=O (2)

11'

r .L-:-Iu~:r=r~=:-~
\I I' 1· dl' If d.\1

( l [l ~~~~--11 )
~- ..

Fig. I. Infinitesimal segment of a beam on two-parameter 1t1tll1dation

Where V and M denote the shear force and bending moment. respectively. 11 is the

foundation force per unit length, w denotes the distributed load on the beam. and a
comma denotes a derivative. According to the two-parameter foundation hypothesis. and
assuming linear soil behavior, the foundation force per unit length is related to the
transverse displacement as follow:

II = k I v(x)- kill vr, (X) (3)

 22

Where kt is the Winkler's modulus and km is a second parameter that depends on both the
soil and foundation characteristics. For inelastic behavior, both parameters will be based
on nonlinear functions as will be described later. The foundation force term
corresponding to the second parameter can be viewed as an additional moment resistance
provided by the foundation following elementary thin-plate theories. Accordingly, the
foundation moment per unit length, assuming linear behavior, is defined as follows:

1111 = k 111 Vx (4)

From equations (I) to (3 ):

M .xx -I m.x +I f - HJ
,., = () (5)

The values of the two foundation parameters ktand km are typically evaluated based
on two-parameter equations [5]. For an approximate analysis, Vlasov and Leonticv [5]
assumed the transverse displacement v(x,y) as a function of a vertical surface
displacement v(x) and a shape function h(y). These equations were derived for a beam
of finite width resting on an elastic foundation layer in a plane strain condition. as shown
in Fig. 2:

Beam
Oj
....
X L
....
~
lv
'
y
Foundation lv ~ u II

Fig. 2. Beam on elastic foundation

v(x,y) = v(x)h(y) (6)

Where h(O) =I. h(H) = 0: H being the depth ofthe soil layer. and
 23

sinh r(l- ~)
h(y) = ---'-------'- (7)
sinh r

Based on these assumptions, the parameters krand k111 are evaluated as:

k = (I- v,) E,h ysinh ycosh y + y~

(8)
.t (1+v,)(l-2vJH 2sinh 2 y

k = E,hH sinh ycosh y- y

(9)
Ill 2(1+l/.,.) 2 ysm
. h'-y

The r parameter is a coefficient that determines the rate of decrease of the

displacements over the depth of the foundation and is evaluated from the equation given
below [5]:

(I 0)

Where E,, v, are elastic modulus and Poisson ratio of the soil, respectively and L is

the soil length. For the Vlasov model, an additional soil length. typically taken as twice
the beam length on both of its sides is assumed. This length was proven to be sutlicicnt to
capture the semi-infinite soil effect [20]. In the Pasternak model, this additional soil
length is ignored.
To determine the parameter r an iterative method developed by Vallabhan and Das
[ 15] is adopted as follows: First assume a value of r and calculate the k 1 and kill

parameters form Eqs. (8) and (9). respectively. With these parameters computed. the new
surface displacement v(x) is evaluated. The parameter r is then recalculated from Eq.
(I 0), and compared to the previously evaluated r value. The process is repeated at each

time step ofthe nonlinear solution algorithm until convergence is achieved within an
acceptable tolerance.
 24

2.2. Compatihili(v
The curvature at a section x is related to the transverse displacements by:

v"- x =0 (I I)

where v is the vertical displacement ofthe beam, and x is the curvature.

2.3. Material Constitutive Laws
The internal moment of the beam M{.t) is related to the curvature X by a nonlinear
constitutive relation

M(x) = g(x(x)) ( 12)

In this study the nonlinear relation in (12) is derived from a tiber discretization ofthe
cross section of the beam with nonlinear uniaxial stress-strain relations for the constituent
materials. The two foundation forces 11 and 1111 are related to their respective

deformations by two other nonlinear relations as follows:

1t = a..t (v)
h
& t 111 ,"r (vI)
= c....,m ( 13)

In the next sections, the strong form equations (I) to (5) and (II) to ( 13) arc solved for
using the tinite element method. Due to the nonlinear nature of equations ( 12) and ( 13 ). a
Newton-Raphson iteration strategy is used. The following discussion refers to a single
Newton-Raphson iteration denoted by subscript i.

3. Displacement Formulation of Beam on Two-Parameter Foundation

In a displacement formulation. the differential equations arc solved based on a
displacement field. Accordingly:

v(x) = a(x)V ( 14)

where v(x) is the vertical displacement. and a(x) is a matrix of 17" shape functions. 17"

depends on the order of displacement shape functions. and V is the vector of clement end
displacements.

The finite element formulation is considered by deriving the weighted integral of the
equilibrium equation:
 25

( 15)

where , denotes derivation.

Integrating by parts twice the first term and once the third term, and ignoring the
distributed load term w:

J5v(x)~"Mdr+ J5v_1,(x)t dt+ J5v 1(x)t 1dt+BT=O 111 ( 16)

l l l

where the Boundary term BT equals to

BT=5vM,_' 0l -5v,M
"
1'- -5vt, (I
11
,\
_ I'
(I

'

The consistent linearization of the nonlinear force-deformation relation for the

beam and foundation yield:

M l =k'-1A)
u.l.r.r +
Ml-1 I

kl-1 A 1-1
I
1111 = 111
I
u.Vx +1 111 ( 17)

Where k, k111 , krare the derivatives of the nonlinear functions g, gill, and g1 •

Substituting ( 17) into ( 16):

1 I I

f)r,, (xl[ r
I 11'l1', + 1
,\r 1 }tr+ f)r, 1 (.\') [ k;.,-l I'll·,'+ t;., I ]dr+ fsr 1 (X) [ k; I Ar + ( I }tr = lrl (IS)

"
Substituting the predefined displacement shape functions a(x) into ( 12). we get:

r
liv' fa' .. (x) k'-'a H dx + fa : (x )k;;' a·' dx + fa' (X )k;-· a d•j ,~\' 1
l n o 1' ( 19)

= liv' rP- fa', (x) ~r-' dx- fa: (x~;,;' dx- f•' (x )1', ' d•j
l (l (I 0
 26

From the arbitrariness of t5v, we get:

(20)

where

/.

K'- = fa:,,(x)k'- 1 (x)a,,(x)dr, is the beam element stiffness matrix,

1

where k(x) is the beam section stiffness term

I.

K'1- 1 = far (x)k;- 1 (x)a(x)dt, is the Winkler foundation element stiffness matrix,
()

where k 1 (x) is the foundation force stiffness term

I.

K;,~ 1 = fa1,(x)k,;~ 1 (x)a,(x)dt, is the two-parameter foundation element stiffness matrix.

0

where k (X) is the foundation moment stiffness term

111

/.

M'- 1 = fa 1,,M'- 1 (x)dr, is the beam element resisting load vector

0

/.

M',- 1 = far (x)t;~ 1 (x)dr, is the Winkler foundation element resisting load vector
0

/.

M;~ 1 = fa~,(x)t;,~ 1 (x)dx, is the two-parameter foundation element resisting load vector
0

and P is the vector of applied external loads.

4. Mixed Formulation of Beam on Two-Parameter Foundation
In a two-field mixed formulation, the differential equations are solved based on both a
displacement and a force field. For the foundation problem. it was proven earlier that this
mixed approach is very advantageous from a numerical standpoint [26]. Accordingly. and
similar to (14):

v(x) = a(x)V (21)

In addition M(x) = b(x)M (22)

 27

Where M(x) is the bending moment, and b(x) is a matrix of n force interpolation 111

functions, and M is the vector of element end moments.

The finite element formulation is considered by deriving the weighted integral forms
of both the compatibility and equilibrium equations:

J8 M 1 (x) [ v,, -X }Lr = 0 (23)

!!

Jov (x)[ M_,, -tm.x +1

1
1 -w }ix = 0 (24)
f!

The incremental section constitutive law is inverted an substituted in (23 ).

Accordingly:

(25)

I.

foM (X) [ v,,; (x)- f'- fuU'- z'- }Lr = 0

1 1 1 (26)
(I

where f'- 1 is the section flexibility term at the previous Newton-Raphson iteration.

Substituting the predefined displacement shape functions and force interpolation

functions into the weak form (26), we get:

<1M' {l ~ b'(x) a,,(x) d¥ v' l -l ~ l ~

b' (xl.r' (x) b(x) d¥ liM' - b' (x) x ,_, (x)<h t= 0

(27)

from the arbitrariness of c5M, we get:

Substituting v' by v'- 1 + ~v', (28) becomes:

T ~v'- F'- 1 ~M' -v r ,_ 1 = 0 (29)

 28

where

T= Jb (x)a"(x)dx.
{)
1

(I

(30)

where F is the element flexibility matrix, and v,. is the element residual deformation

vector.
Integrating by parts twice the first term and once the second term in (24), we get:

I. I. I

f5v 1"(x) M' (x)d'<+ f5v/(x) t;Jx)dx + f5v 1(x) t~ (x)dx =Boundary Tenns (BT)
0 0 0

(31)

The incremental force-deformation relations of the foundation take the form

t't = k'-.t 1 fl v ' + t'-.t 1 and t'm = k'-m 1 fl ~,' + t'-

m
1
(32)

Substituting (32) into (31) results in

I I I

ft5v_,/(x) M' (x)dx + ft5v/(x) [k;,~ 1 ~v' +1~~ 1 }tx + ft5v (x) [ k~- 1 ~v' + ~~- 1 ]d\ = BT
1 (33)
{) 0

Substituting the predefined displacement shape functions and force interpolation

functions into the weak form (33), we get:

5v 1 { ' f•',,<x) b(x)c/x1M' + f• ,' (x) [ k;,;' t.v' +(,;'Jet<+ f•' (x) [ k;-' ill'' + t;-' }~<fl = BT
l () () (I

(34)

From the arbitrariness of Sv and replacing M' = M'- 1 + ~M'. (34) is rewritten as:

(35)

where K',- 1 K'- 1 , M',- 1 , M'- 1 are as defined in (14), Tis as defined before in (30). and P
'I 111 . 111

is the vector of applied external loads.

 29

Writing equations (29) and (35) in matrix form:

[
-f
T/
1 1

Kl-l
I
T ][~M~l
111 V
[V
+ Kl-l ~ I = p- T 1 MI-l~ MI.-I
1
_,

I +
MI-l
m
]
(36)

It is important to note that at convergence, the residual deformation vector v, reduces

to zero inside each element satisfying compatibility. As discussed by Ayoub and Filippou
124], two algorithms for the mixed formulation exist. In the first algorithm, the system of
equations in (36) is solved for globally with the displacements and moments as degrees of
freedom. This algorithm, however, results in oscillations of results. In the second
algorithm. the moment degrees of freedom are condensed out from the first of equations
(36) at the element level resulting in a generalized displacement stiffness matrix.
Accordingly:

(37)

An internal element iteration is required in order to zero the residual deformation

vector v r in every element. The algorithm used follows the same procedure as the one
discussed in Ayoub [25].
4.1. Stability of Mixed Formulation
The order and continuity of stress and displacement interpolation functions are very
important parameters in a mixed formulation. For stability ofthe formulation the rank of

matrix T in the expression T 1 ( F 1

_, r' Tin Eq. (3 7) should not be larger than the rank of
the flexibility matrix F for the limit case where the foundation stiffness matrix is zero.
For this to be the case the number of unknowns n;, in vector v after excluding their rigid

body modes should be less or equal to the number of unknowns n, in vector M:

(38)

While condition (38) is necessary for stability of the problem. there is no accuracy gain
by increasing the order of the force field beyond that of the deformation tield that
respects the strain- displacement compatibility condition. The equality condition of(38).
i.e. n, = n;, is therefore the most efficient choice fi·om a computational standpoint. As a
 30

result. the Babuska-Brezzi (8-B) stability conditions [28-29] for the beam on two-
parameter f()lmdation element states that the order of the displacement interpolation
functions needs to be larger by two than that of the force interpolation functions.
5. Evaluation of Model by Numerical Studies
5. I. Elastic Beam on V/a.mv Foundation
The proposed two-parameter model with Pasternak and Vlasov effect (effect ofthe
soil on either side of the beam is considered) was evaluated by analyzing a beam of tlnite
length resting on an elastic foundation, which was tlrst studied by Shirma and Ginger
1131.
The beam length is 5m. width b=0.4m and depth h=I.Om. The beam is made out of

timber with an elastic modulus Eh= I0.500 MPa and Poisson ratio vb = 0.25 . The elastic

foundation is sandy clay with an elastic modulus Es=45.4 MPa. Poisson

ratio vs = 0.25 and y =l.O. From equations (8) to (I 0). the values of the f()lmdation

parameters are: k1 = 3.081 MPa and k111 =12,449.000 N.

The beam has free ends and is subjected to a concentrated moment of 50 kNm applied
at the center as shown in Fig. (3 ). The beam is discretized into 4 mixed elements with
cubic moment interpolation functions. Five integration points were assumed for each
finite clement. Since the problem is elastic, the load is applied at mid span under load
control. To evaluate the effect ofthe semi-infinite foundation, the beam is analyzed using
a Winkler. Pasternak. and Vlasov formulation. Fig. (4) shows the midspan moment
rotation behavior of all models. From the figure. at the highest moment value of 50 kN .m.
the Winkler rotation was found to be 2.8 times larger than the Vlasov rotation. Fig. (5)
shows the deflected shape of the models. Because ofthe added rotational resistance due
to the Vlasov parameter. the end deflection of the Winkler foundation is about three times
that of the Vlasov foundation. Furthermore. due to the effect of the soil on both sides of
the beam in the Vlasov modeL its end deflection is 2.1 times less than that of the
Pasternak model. The bending moment is slightly underestimated if ignoring the Ylasov
or Pasternak effects. as shown in Fig. (6).
The same beam with free ends is analyzed assuming the j()undation to be tensionless.
as shown in Fig. (7). An axial force P that equals I 00 kN is applied under load control.
 31

while a moment M is applied incrementally under displacement control. Fig. (8) shows
the midspan moment rotation behavior of the beam for the Vlasov and Winkler
foundations, respectively using a mixed model with cubic moment interpolation
functions. The Vlasov foundation moment resistance is found to be about 2.8 times larger
than that of the Winkler foundation moment resistance at a rotation of 0.0 I rad. Fig. (9)
shows the foundation vertical displacement of both models at the ultimate load. which
reveals that the Winkler model is having slightly larger deformations than the Vlasov
model. From Fig. (I 0), which shows the foundation rotation for both models at the
ultimate load, the Winkler model has a rotation II% higher than that of the Vlasov
model. The preceding discussion confirms the need to account for the semi-infinite soil
effects in analyzing beam on foundation problems.

2500 mm

- 2."00 llllll

Fig. 3. Beam with free ends

-\),,,.") .. ,... .~,,,.,,,),,,,)\),,,j.),,,,,.,,,,., ...,,,

- - - l'."t.-rn.t~ I "''"'\.'1'''" \I ,,,·d \ 1·~1, I ,, nl••· "'' "''' 111 ·
•• • · • \\Ill~ lei ) ••UIIol.Lih•ll \)L\,d \I•~), I r,uln, 1ll••L11>!11

llooi.Lllo>lliL,I<IL

Fig. 4. Moment-rotation response of free beam

 32

- \ l.o~· '' I •llllld.ott• 111 \li\<'d \h•d..:l 1 cllhtc III(IJ\JC!II,

---! ,J,I<'I!I.J~ I •>lillol.ill•'IL \h\<.\J \lo>Jd l<.lli'l< llJ<Illl<..:lll 1

1\ .. ,kkT lo>Liild.oll• io\1;\c'd\J.,,!.Jo,LJbi,!l\PIIK'lll>

_,

L.:ngth alnng hcamtnll

Fig. 5. Vertical displacement for free beam

.II

~(I

~~~
"E
;i:
-:! 111
c
§ -
.
II
i!!
~

-I !I

J - ..~0

- ~u

--W

Length along beam tm)

Fig. 6. Bending moment distribution for free beam

.... ~~()(\ 111111 ~)00 111111

.
Fig. 7. Tensionless beam with free ends
 33

600
- - Vla~ov foundataon m•xed model (cub•c moment)
500 · · W1nkler foundation mxed model (cubic moment)
;
-: 400

ic 300

~
200

=
~ 100

0
0 0.002 0.004 0.006 0.008 0.0\

Rotation {md)

Fig. 8. Tensionless foundation moment-rotation for free beam

30
- - Vlasov foundataon m1xed model {cubic moment)
~ 20 ------·Winkler foundation mixed model (cubic moment)

~ 10
E
.1!
c.
.i!/
~ -10

'f
-~ -20

-30
0

l.e~th aim~ beam(m)

Fig. 9. Tensionless foundation vertical displacement for free beam

.,! 0.0098
0.0096
0.01

c 0.0084
-~ 0.0092
--VIasov foundal1on rnixed model
e 0.000
(GUbic moment)

..
-~
""!§
0.0088
0.0086
--Winkler foundation mixed model
(cubic moment)
0.0084
...
~
0.0082
0.008

IA·ngth along beam (m)

Fig. I 0. Tensionless foundation rotation for free beam

 34

5.2. Inela.\·tic Beam on Two-Parameter Foundation

The second numerical example represents an inelastic beam resting on a tensionless
foundation. The main objective of this example is to compare the behavior of the
Winkler one-parameter model. to the Pasternak and Vlasov two-parameter models. The
beam is shown in Fig. (II). and has a length of I 0 m. and a square cross-section with I 00
mm dimension. For the Pasternak model, an additional soil length that equals twice the
beam length on both of its sides was used. The adjacent soil effect on the beam depends
on the soil modulus and depth of the soil layer. For the Ylasov model, an additional soil
length of twice the foundation length (20m) was added on both sides of the beam. The
beam uniaxial stress-strain relation is elasto-plastic with Young's modulus E = 200 GPa.
yield strength of207 MPa, and a hardening slope that equals 1.4%. The beam section is
subdivided into 16 fibers. The underlying soil is I 0 m Nevada Sand with properties as
given by Pradhan and Desai [30] as follows: elastic modulus Es=40.85 MPa. and Poisson
ratio v, =0.316 .The soi I parameters are being calculated with the analytical method

proposed by Vallabhan and Das [15] and described in equations (6) to (I 0).
The loading condition consists of a transverse force and a moment acting at midspan.
which is typical of foundation structures. The transverse force equals 70 kN. and is
applied under load control. while the moment is applied incrementally under
displacement control. The converged midspan moment rotation behavior of the beam is
shown in Fig ( 12) tor the mixed model with 32 elements. In the figure. points A. A· and
A., represent yield points for the Ylasov. Pasternak and Winkler models. respectively.
Similarly. points B. B' and B" represent the corresponding points at the ultimate state.
The plot reveals that the stiffness is highest tor the Vlasov model and is lowest for the
Winkler model. The Vlasov model has also a higher yield and ultimate moment
capacities, while the Winkler model has the lowest. The same plot is repeated in Fig. ( 13)
using the displacement-based model. The distribution of the local parameters along the
length ofthe beam. namely the bending moment. curvature. vertical displacement.
rotation, and foundation forces are shown in Figs. (I 4-20) at the ultimate load stage.
These figures reveal that the displacement model did not achieve convergence. even with
32 elements.
 35

Fig. ( 14) shows the curvature distributions ofthe Winkler, Pasternak, and Vlasov
models using the mixed formulation, while Fig. ( 15) shows the same distributions using
the displacement formulation. The plots revealed that the mixed model was able to
capture the curvature localization near the midspan and accurately predict the maximum
curvature. The displacement modeL however, failed to capture this behavior as it
underestimated the maximum beam curvature value by a factor of2.3. This is in part due
to the assumed displacement shape functions, which can not represent the steep curvature
distribution accurately.
As observed in Figs. ( 16) and ( 17) the lift-off at the beam ends where the foundation
force vanishes is severe for the Winkler model, and is much less for the Pasternak and
Vlasov models. In addition, Fig. ( 17) reveals that the lift-off region is slightly higher tor
the Pasternak than for the Vlasov model, and that the foundation force is much smaller
tor the Pasternak than tor the Vlasov model. Furthermore, From Fig. ( 16 ). due to
consideration ofthe surrounding soil effect, the displacement at midspan is also much
lower f(x the Vlasov and Pasternak models than tor the Winkler model. The beam
bending moment values along the length are higher tor the Vlasov than for the Pasternak
and Winkler models. as shown in Fig. ( 18), due to the additional moment resistance
provided by the semi-infinite soil ellects. The foundation rotation along the beam length
is higher tor the Winkler model than tor the Pasternak and Vlasov models as shovm in
Fig. ( 19). The foundation moment resistance is zero for the Winkler model: while it is
higher for the Vlasov model than tor the Pasternak modeL as shown in Fig. (20).
5.3. Shear Wall Foundation Structure
Numerical analysis using the mixed model was conducted tor the Aluminum shear
wall f(Jundation structure SSG04-06 tested by Ciajan et al. [31 J under an increasing lateral
load. The footing is 2.8 m x 0.65 m, and has a Young's modulus of70.000 MPa. The
underlying soil is Nevada sand with modulus of elasticity 45 MPa and Poisson ratio 0.4.
The yield Ioree of the soil is 1238 kN/m. which is assumed to be at 35% of its bearing
capacity. Fig. (21) shows the monotonic envelope of the moment-rotation plot at the
middle ofthe foundation for both the Vlasov and Winkler models. as well as the
experimental results. From the figure. it is observed that the Ylasov model is able to
predict the behavior reasonably welL while the Winkler model under-predicts the
 36

moment capacity of the foundation. Fig. (22) shows the foundation force distribution at
the ultimate load. The maximum foundation force equals 1504 kN/m. which exceeds the
soil yield force, indicating that the soil has undergone inelastic deformations.

'··!AS 1111111
rr (i\/1'<1)

107 I. (k,\' Ill)

,., !mel)

l'(lll)

](I()

5000 111111

-
Fig. II. Inelastic beam on tensionless foundation
)()()() \1\\l\
...

120
--- Vlaso,·
· · · · · l'asternak
-Winkler
100 ll
_..-- -eto
ll'
-,.--:-
i 80
i
-:.
=
s" 60
=
s
.
=
c:.
:E
:; 40 ·I"

20

0
0 0.002 11.004 11.006 11.008 0.01 0.012 0.1114 11,111(> 11.1111! 11.112
\lids11an rotation lrad)

Fig. 12. Moment rotation response ofbeam on tensionless foundation (mixed

model)
 37

120

- - - \'hlSO\
----- Pasternak
100 -\\inkier
1\
>~a
ll'
.,.-_.,-/'·""

80
~
z
"'
~
;
/.
-
60 /.-:·--:·./,.
=
= /-:-·
;~~~'
~

~ w
:E
:::: ~0 ,. /"

/"
,-. /"
,r- A"
/
,.. /"

20 "
,.. /"
/ '"
/
/
/
/
0
0 0.1102 11.004 0.006 11.008 0.01 0.1112 1).(114 11.1116 ll.IIIS 11.112
\lid,p:m rol:ttiun (nut)

Fig. 13. Moment rotation response of beam on tensionless foundation (dis pl. model)

0.3
--- \'lasov
----- Pastt.·rnak
IUS
-Winkler

0.2

s
:c II.IS
E
~
.2
::::::" 0.1
:..,

II.IIS

II I
I
,,

-II. OS
0 6 HI
Length along beam tml

Fig. 14. Curvature distribution for beam on tensionless foundation (mixed model)
 38

0.11

--- \'laSO\
Pasternak I
II
0.1 -Winkler II
I
I

0.08

0.06

!
.=
"
>
(J.IJ-1

=
'- 0.02

I
I
i
I ,->
-0.02 ~/-

-0.0~

Lt•ngth 11long ht·11m (m) "'

Fig. 15. Curvature distribution for beam on tensionless foundation (displ. model)

711

--- \'lascn
60

/
· Pasterna"-
-Winkler
50

.to
E"
5 30
= 20
~
,;:
~ 10
'S
"§
0
-~
> -10
/
/
/
-211

-311

-.tO
II Ill
Lt.•nJ!th alonJ! ht'am (m)

Fig. 16. Vertical displacement distribution for beam on tensionless foundation

(mixed model)
 39

II

-~

-Ill
E
; ,1
.l

-=t: -I='
i
i
... .)
l
..:
.I

·~ -211
-g"
=
:£
-2='

- - - \'laso\
-311 · · · · · Pasternak
-Winkler

-3S
II 6 8 10
Length along beam (m)

Fig. 17. Foundation force distribution tor beam on tensionless foundation (mixed
model)

80
--- \'lasn'
· Pasternak
60 -Winkler

i
40 I
I

z=
:.
.=
c

0
20

= I
"c
'5
I
I
~ I
"' -20
I
I
.; .;
r

I ./
I ·/
I.'!
i/
'"

-60
II 10

Length alonJ! hl'am (m)

Fig. 18. Moment distribution tor beam on tensionless foundation (mixed model)
 40

0.!12S
--- \'laso\'
0.02 ----- l'asternak
\
--Winkler \
\
O.OIS \
'-,\
~ \
f: 11.111
"\
'
·!" n.nns
..
~
0

;c:"
0 n , __.......
..."
~_

............. ~

..."= -ll.OOS
0 -·...:.:_,:.
----
.;..·)

-ll.lll

-O.OIS

-0.02
0 2 6 8 HI
Length along beam (m)

Fig. 19. Foundation rotation for beam on tensionless foundation (mixed model)

70
--- \'laso\
(,() ----- l'asternak
--Winkler
sn h.
I\.

e
e 40 '
I'
J \
\'
\
z. I
J
\
\
-=- JO
=
...
e 20 :·'
J '·
\
\,.
=
e I '

""
;c: 10 ./ J '-
.... ,
........
..,_
..."' I
(
...... -:-,
,__
..."="
I
n I

,I
(
-10 i
.)

·'
-20

-30
n 2 8 10
Length along beam (m)

Fig. 20. Foundation moment distribution tor beam on tensionless foundation (mixed
model)
 41

XIIII

Ylasm Model
" Winkler Model
Icxp.:rimcnt em clop

"11111

coo • I

. /_'

II i
II 0"01 002 0"0.1 () 0~

Rotation (rad)

Fig. 21. Moment-rotation response ofGajan et al. specimen

(I

- Ylasm Model
--- W inkier Model

-400

-1200

-1600
() 05 1.5 2 2.5 3
Length along beam (m)

Fig. 22. Foundation force distribution ofGajan et al. specimen

 42

6. Conclusions
The paper presents a new inelastic element for the analysis of two-parameter beam on
foundation problems. The element is derived from a two-field mixed formulation. where
forces and deformations are approximated with independent interpolation functions. An
iterative rational procedure to estimate the values of the two parameters of the foundation
based on an assumption of plane strain for the soil medium was presented. This iterative
behavior is conducted at each loading step of the nonlinear solution algorithm. The
nonlinear response of structures resting on this newly developed two-parameter
foundation model is analyzed following both a Vlasov and a Pasternak approach.
Numerical examples to compare the behavior of the one-parameter and two-parameter
models were conducted. The studies confirmed the importance of including the second
parameter in estimating the foundation behavior, and revealed that accounting tor the
effect of the soil on both sides ofthe beam by adopting a Vlasov approach can
substantially affect the nonlinear response. The studies also confirmed the superiority of
the proposed mixed model in evaluating the inelastic complex behavior of these types of
structures.

References
[I] Winkler, E. ''Theory of Elasticity and Strength,'' Dominicus Prague. Czechoslovakia:
1867.

[2] Filonenko-Borodich. M.M. Some approximate theories of elastic foundation.

Uchenyie Zapiski MoskovskogoGosudarstuennogo Universiteta Mekhanika. Moscow.
1940; 46:3-18 (in Russian).

[3] Hetenyi, M. Beams on elastic foundation. The University of Michigan Press. Ann
Arbor: 1946.

[4] Pasternak, P.L. On a new method of analysis of an elastic foundation by means of two
foundation constants. Gosudarstvennoe lzdatelstvo Literaturi po Stroitelstuve i
Arkhitekture, Moscow: 1954; (in Russian).

[5] Vlasov and Leontiev, U.N. Beams. plates. and shells on elastic foundation.
Jerusalem: Israel Program for Scientific Translations: 1966 [translated from Russian J.
 43

J6J Biot. M.A. Bending of an infinite beam on an elastic foundation, J. Appl. Mech.
1937: A4:1-7.

171 Kerr. A.D .. A study of a new foundation model, Acta Mechanica 1965; 1/2:135-147.

181 Rcissncr. E. Note on the formulation ofthe problem of the plate on an elastic
fi.)undation. Acta Mechanica 1967: 4:88-91.

191 llarr. E.M .. Davidson. J.L., and S.V.Ramaswamy. Euler Beams on Two Parameter
Foundation ModeL Journal of the Soil Mechanics and Foundations Division :
proceedings ofthc American Society ofCivil Engineer 1969; 95(4): 933-948.

[I OJ Yang. T.Y. A finite element analysis ofplates on a two parameter foundation modeL
Computers and Structures 1972; 2: 593-614.

Jill Zhaohua. F.. and Cook. R.D. Beam elements on two parameter elastic foundations.
Journal of Engineering Mechanics, ASCE 1983: 109(6): 1390-1402.

f 12] Chiwanga M. and Valsangkar AJ. Generalized beam element on two-parameter

elastic foundation. J. Struct. Engng 1988: 114(6): 1414-27.

[ 13] Shirima LM. Giger MW. Timoshenko beam element resting on two-parameter
clastic f(Hmdation". J Engng Mech ( 1990 ); 118(2):280-95.

[ 14] Razaqpur. G.A. and Shah, K.R. Exact analysis of beams on two-parameter elastic
foundation. lnt J Solids Struct. 1991: 27: 435-454.

[15] Vallabhan. C.V.G., and Das, Y.C. Modified vlasov model for beams on elastic
foundations, Journal of Geotechnical Eng 1991: 117 (6): 956-966.

[ 16] De Rosa, M.A. Free vibrations ofTimoshenko beams on two-parameter elastic

t<.)undation, Computers and Structures 1995; 57( I): 151-156.

[ 17] Hou Y. C., Tseng, C. H. and Ling, S. F. A new high-order non-uniform Timoshenko
beam finite element on variable two-parameter foundations tor vibration analysis. Journal
of Sound and Vibration 1996: 191 (I), 91-1 06.

[ 18] Patel, B.P., M. Ganapathi and M. Touratier. Nonlinear free flexural vibrations /post-
buckling analysis of laminated orthotropic beams/columns on a two parameter elastic
foundation, Composite & Structures , 1999; 46: 189-196.
 4~

[ 19] Onu, G. Shear effect in beam finite element on two-parameter clastic f(Jundation, J
Struct Eng. 2000: 126: 1104-1107.

[20] Morfidis, K., and Avramidis, I.E. Formulation of a generalized beam element on a
two-parameter elastic foundation with semi-rigid connections and rigid offsets,
Computers and Structures 2002; 80: 1919-1934.

[21] Coskun, I. The response of a finite beam on a tensionless Pasternak foundation

subjected to a harmonic load, European Journal of Mechanics- A/Solids 2003: 22( I):
151-161

[22] Celep Z, and Demir F. Circular rigid beam on a tensionless two-parameter elastic
Foundation, Zamm-Zeitschrift tUr Angewandte Mathematik und Mechanik 2005: 85(6):
431-439.

[23] Kaschiev, M.S .. and Mikhajlov, K. A Beam Resting on a Tensionless Winkler

Foundation. Computers and Structures 1995: 55(2): 261-264.

[24] Ayoub, A.S., and Filippou, F.C. Mixed Formulation of Nonlinear Steel-Concrete
Composite Beam Element. Journal ofStructural Engineering, ASCE 2000: 126(3): 371-
381.

[25 J Ayoub, A.S. A Two-Field Mixed Variational Principle for Partially Connected
Composite Beams, Finite Elements in Analysis and Design, 2001: 37: 929-959.

[26] Ayoub, A. Mixed formulation of nonlinear beam on foundation elements. Computer~

and Structures. 2003; 81: 41 1-421

[27] Zienkiewicz, 0. C.. and Taylor. R. L. The Finite Element Method Volume I. Basic
Formulation and Linear Problems, Fourth Edition. McGraw Hill. London: 1989.

[28] Babuska, I. The Finite Element Method with Lagrange Multipliers. Num. Math.
1973: 20, 179-192.

[29] Brezzi, F. On the Existence. Uniqueness and Approximation of Saddle Point

Problems Arising from Lagrangian Multipliers. RAIRO 1974: 8-R2. 129-151.
 45

[30] Pradhan, S.K., and Desai, Chandra S. DSC model for soil and interface including
liquefaction and prediction of centrifuge tests, Journal of Geotechnical &
Geoenvironmental Engineering 2006: 132(2): 214-222.

[31] Gajan, S., Phalen, J., Kutter, Band Kutter, B. Soil-foundation-structure interaction:
shallow foundations 2006. Center for Geotechnical Modeling. University ofCalif()fnia.
Davis. CA.
 4<

VITA

Taraka Ravi Shankar Mullapudi was born on June 17. 1978 in Tallapudi. India.
He received his Bachelor of Technology in Civil Engineering from National Institute of
Technology. Warangal, India in the summer of2002. He enrolled at Missouri University
of Science and Technology to pursue his Master of Science in Civil Engineering. His
graduate studies focused on soil-structure interaction. He was fortunate enough to receive
a Graduate Research Assistantship from the Department of Civil Engineering. Aller
fulfilling the requirements. Mr. Mullapudi received his MS in Civil Engineering in Ma)
2010.