PT Stiffness Matrix PDF

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/315743987
FINITE ELEMENT ANALYSIS OF POST-TENSIONED CONCRETE BOX GIRDERS
Thesis · May 2004

DOI: 10.13140/RG.2.2.23073.81769
CITATIONS READS
0 316
1 author:
Mohanned Al Gharawi
University of Missouri
6 PUBLICATIONS 2 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
CONCRETE BOX GIRDERS View project
A Framework for Automated Lock-In Thermography Data Processing View project
All content following this page was uploaded by Mohanned Al Gharawi on 02 April 2017.
The user has requested enhancement of the downloaded file.

University of Baghdad
College of engineering
Department of civil engineering
FINITE ELEMENT ANALYSIS OF POST-TENSIONED

CONCRETE BOX GIRDERS
A THESIS
SUBMITTED TO THE COLLEGE OF ENGINEERING
OF THE UNIVERSITY OF BAGHDAD IN PARTIAL
FULFILLMENT OF THE REQUIREMENTS FOR
THE DEGREE OF MASTER OF SCIENCE
IN CIVIL ENGINEERING
(STRUCTURES)
BY
MOHANNED IBRAHIM MOHAMMED HUSSAIN
AL-GHARRAWI
APRIL 2004 MUHARRAM 1425

ِ‫ﺑِﺴِﻢ ﱠ‬
‫اﻪﻠﻟ اﻟ ﱠﺮ ْﺣ َﻤ ِﻦ اﻟ ﱠﺮِﺣﻴِﻢ‬ ْ
   

  
 
 
‫ﻴﻢ‬ ِ ‫اﻪﻠﻟ ا ْﻟﻌﻠِﻲﱡ ا ْﻟﻌ‬

‫ﻈ‬
ُ َ َ ُ‫ﺻﺪَ َق ﱠ‬َ
(85:‫)اﻹﺳﺮاء‬
 



 
 

 
 
 
 



 
2004

(SUPERVISOR CERTIFICATE)
I certify that the preparation of this thesis titled “Finite Element Analysis of Post-
Tensioned Concrete Box Girders” accomplished by “Mohanned Ibrahim
Mohammed” was made under my supervision at the Department of Civil
Engineering, University of Baghdad as a partial fulfillment of the requirement for
degree of Master of Science in Civil Engineering (Structure).
Signature:
Name: Prof. Dr. Husain Mohammed Husain
Date:
In view of the available recommendations, I forward this thesis for debate by the
examination committee.
Dr. Abdul Muttalib I. S. Al-Musawi, Head,

Department of Civil Engineering,
(EXAMINATION COMMITTEE CERTIFICATE)
We certify that we have read this thesis titled “FINITE ELEMENT

ANALYSIS OF POST-TENSIONED BOX GIRDER”. Presented by “ Mohanned
Ibrahim Mohammed Hussien”, and as an examining committee, we examined the
student in its contents and in what is connected with it, and that in our opinion it meets
the standards of a thesis for the degree of master of science in Civil Engineering
(Structures).
Signature: Signature:
Name: Prof. Dr. Husain M. Husain Name: Dr. Ihsan A. S. Al-Shaarbaf
(Supervisor) (Member)
Date: / / 2004 Date: / / 2004
Signature: Signature:
Name: Dr. Nazar K. Al-Oukaili Name: Prof. Dr. Khalid S. Mahmoud
(Member) (Chairman)
Date: / / 2004 Date: / / 2004
Approved by the Dean of College of Engineering
Signature:
Name: Prof. Dr. Ali Al-Kiliddar
Dean, College of Engineering
Date: / / 2004
ACKNOWLEDGEMENT
Firstly, all thanks and praise be to GOD who enabled me to complete this
work.
I would like to express my deepest gratitude and gratefulness to
Prof. Dr. Hussain M. Hussain the supervisor for his valuable guidance,
encouragement, constructive suggestions, and his assistance throughout
the final preparation of this work.
Sincere thanks to my colleague the Ph. D. student Wamidh A'Jel
for his help.
Special thanks to my colleagues and friends Mohammed Najim,
Nassir, Ghaidaq, Ahmed Adnan, Ali Ahmed Salman and Sabah for
their encouragements.
Special thanks to my real friends Amjad, Najah, Adil, Dhirgham
and Ahmed Khudher for their helps to complete this thesis.
Cordial thanks are also to my colleagues Miss Ruwayda, Miss
Hind and Mirs. Rana for their encouragements throughout all stage of
this study.
Finally, I would like to express my deepest gratitude to all who
gave helpful hand throughout this study, especially to my family for their
encouragement and assistant during this work.
MOHANNED
2004
I
ABSTRACT
The behavior of prestressed concrete box-girder bridges has been studied

under short term loading. The box-girder with its two cantilever flanges is
simulated by isoparametric 20-noded brick elements with three translation
degrees of freedom at each node. The reinforcing bars are idealized as axial
members embedded within the brick elements. The behavior of concrete in
compression was simulated by an elastic-plastic work hardening model
followed by a perfectly plastic response, which is terminated at the onset of
crushing. In tension, a smeared crack model with fixed orthogonal cracks is
used with the inclusion of models for the retained post-cracking stress and
reduced shear modulus.
Prestressing tendons are modeled by using the equivalent nodal
forces concept. Also, the prestressing tendons are treated by using Lin's
method. The contribution of the prestressing tendon stiffness to the global
stiffness matrix is considered by treating the tendon with axial member
embedded within the brick element. Two types of short term prestress
losses are considered in this study. The bond-slip phenomenon at concrete-
tendon is considered by reducing the tendon axial stiffness.
The nonlinear equations of equilibrium have been solved by using the
incremental-iterative technique operating under load control. The solution
algorithms are the standard and the modified Newton-Raphson methods.
The numerical integration has been conducted by using 27 point Gaussian
rule.
Several examples of prestressed concrete box girders are analyzed
and compared with available experimental and theoretical studies in order
to demonstrate the validity and efficiency of the proposed method. Good
agreements between the results are obtained.
II
CONTENTS
SUBJECT PAGE
Acknowledgement I
Abstract II
Content III
List of tables VI
List of figures VII
Notations. XI
Chapter One: INTRODUCTION

1.1 General 2
1.2 Method of Analysis 3
1.2.1 Folded Plate Method 3
1.2.2 Finite Strip Method 4
1.2.3 Finite Element Method 4
1.3 Objective and Scope 5
1.4 Layout of Thesis 6
Chapter Two: REVIEW OF LITERATURE

2.1 Introduction 7
2.2 Analysis of Box-Girder Bridges 7
2.3 The Finite Element Method 11
Chapter Three: THE FINITE ELEMENT FORMULATION AND

NON-LINEAR SOLUTION TECHNIQUES
3.1 Introduction 21
3.2 Formulation of Element Stiffness Matrix 21
3.3 Material Representation 24
3.3.1 Concrete Idealization 24
3.3.1.1 Shape Function 25
3.3.1.2 Evaluation of Element Stiffness Matrix 26
III
3.3.2 Reinforcing Bar Representation 28
3.3.3 Bond and Bond Slip Representation 31
3.4 Numerical Integration Technique 33
3.5 Nonlinear Solution Techniques 35
3.6 General Nonlinear Solution Procedure 36
3.6.1 Incremental Technique 36
3.6.2 Iterative Technique 37
3.6.3 Combined Incremental-Iterative Technique 39
3.7 Convergence Criteria 41
Chapter Four: MODELING OF MATERIAL PROPERTIES

4.1 Introduction 42
4.2 Modeling of Concrete 42
4.2.1 Stress-Strain Models 43
4.2.2 Behavior of Concrete In Compression 44
4.2.2.1 The Yield Criterion 45
4.2.2.2 The Hardening Rule 46
4.2.2.3 The Flow Rule 48
4.2.2.4 Incremental Stress-Strain Relationship 50
4.2.2.5 The Crushing Condition 51
4.2.3 Behavior of Concrete in Tension 52
4.2.3.1 Model of Crack Representation 52
4.2.3.2 The Cracking Criterion 54
4.2.3.3 Tension Stiffening Model 58
4.2.3.4 Shear Retension Model 59
4.3 Modeling of Reinforcement 60
Chapter Five: FORMULATION OF PRESTRESSING

5.1 Introduction 61
5.2 Short-Term Prestress Losses 62
5.2.1 Frictional Losses 62
IV
5.2.2 Anchoring Losses 63
5.3 Algorithm for Use with the Finite Element Analysis 65
5.3.1 Basic Assumption 65
5.3.2 Geometry of the Tendon and Variation of the Prestressing
Forces 65
5.3.3 Element Local Loads Due to Prestressing 67
5.3.4 Vector of Primary Nodded Forces of the Element 68
5.4 Algorithm for Use the Lin's Method 70
5.5 Computer Program 71
Chapter Six: APPLICATIONS AND RESULTS

6.1 Introduction 72
6.2 Simply Supported Single-Cell Prestressed Concrete Box-Girder
Bridge 72
6.3 Simply Supported Two-Cell Prestressed Concrete Box-Girder
Bridge 91
Bridge with Inclined Webs 113
Chapter Seven: CONCLUSIONS AND RECOMMENDATIONS

7.1 Conclusions 132
7.2 Recommendations 134
REFERENCES 135
APPENDIX A
APPENDIX B
V
LIST OF TABLES
TABLE
TITLE PAGE
NO.
3.1 Shape functions for the 20-node hexahedral 26
3.2 Weight and location of sampling points in the 27-point rule 35
6.1 Material properties of the one-cell box-girder bridge 76
6.2 Variation longitudinal stresses at mid span for Pu/P=0.250 and 0.875
81
6.3 Variation longitudinal stresses at quarterspan for Pu/P=0.250 and 0.875

82
6.4 Material properties of the two-cell box-girder bridge. 94
6.5 Variation of longitudinal stresses at midspan for P/Pu=0.182 and 0.727

99
Variation of longitudinal stresses at quarterspan for P/Pu=0.182 and
6.6
0.727 100
Material properties for prestressed box-girder bridge with inclined
6.7
webs 116
6.8 Variation of deflections for cross-section at midspan (mm) 118
6.9 Variation of longitudinal stresses for cross-section at midspan (MPa)

119
VI
LIST OF FIGURES
FIG.
TITLE PAGE
NO.
1.1 Development of the box-girder cross-section 2
1.2 Three-cell girder being discretized into finite element 5
2.1 Curved box-girder of Aneja and Roll 12
2.2 Skew ended element used by Scordelis 15

Box-girder macro-element formed by an assembly of special-purpose
2.3
elements 16
2.4 Thin-walled beam element with three nodes 17
2.5 Typical shell element used in the analysis 18
2.6 Typical 8-noded, 6 D.O.F. layered isoparametric flat shell element 19
2.7 Degenerated Mindlin-type thick curved shell element 20
3.1 20-Noded isoparametric brick element 25
3.2 Alternative representation of reinforcement 30
3.3 Representative bond stress-slip curves 33

Distribution of the sampling points over the element in the 27-gaussian
3.4
points integration rule 34
3.5 Typical structure response 36
3.6 Basic techniques for solving nonlinear equations 38
3.7 Standard and modified Newton-Raphson methods 40
4.1 Uniaxial stress-strain curve for concrete 45
4.2 Idealization of discrete cracking model 53
4.3 Idealization of smeared cracking model 53
4.4 Failure cracking for triaxially loaded concrete 57
4.5 Post-cracking for concrete in tension 58
4.6 Shear retention model for concrete 60
VII
4.7 Stress-strain relationship of steel bars 60
5.1 Prestress forces losses due to anchor slip 64
5.2 Typical segment of prestressing tendon traversing a brick element 66
5.3 Analysis of curved cable 70
6.1 Structural details of the bridge analyzed for one-cell box-girder 74

Finite element idealization of half bridge model for one-cell box-girder
6.2
before and after loading 75
Simulated Ontario highway bridge design trucks (OHBDC) for one-cell
6.3
box-girder bridge 77
Analytical and experimental load-span deflection curve for one-cell box-
6.4
girder bridge 77
Deflected shape of the bridge at various load levels for one-cell box-girder
6.5
bridge 78
6.6 Longitudinal normal stress at top midspan for one-cell box-girder bridge 78
Longitudinal normal stress at top quarterspan for one-cell box-girder
6.7
bridge 79
Variation of concrete longitudinal normal stresses along the center line of
6.8
the top slab for one-cell box-girder bridge 79
Longitudinal normal stress variation across the section at midspan for
6.9
one-cell box-girder bridge 80
Longitudinal normal stress variation across the section at quarterspan for
6.10
one-cell box-girder bridge 80
6.11 Variation of longitudinal stresses in the top surface of the top flange 83
6.12 Variation of transverse stresses in the top surface of the top flange 84
Variation of longitudinal stresses in the bottom surface of the bottom
6.13
flange 85
6.14 Variation of transverse stresses in the bottom surface of the bottom flange
86
6.15 Variation of longitudinal stresses in the outer surface of the web 87
6.16 Variation of transverse stresses in the outer surface of the web 88
6.17 Variation of longitudinal stresses in the inner surface of the web 89
6.18 Variation of transverse stresses in the inner surface of the web 90
6.19 Structural details of the bridge analyzed for two-cell box-girder 92

Finite element idealization of half bridge model for two-cell box-girder
6.20
before and after loading 93
VIII
Simulated Ontario Highway Bridge Design trucks (OHBDC) for two-cell
6.21
Analytical and Experimental Load-Midspan Deflection Curves for two-
6.22
cell box-girder bridge 95
Deflected shape of the bridge at various load levels for two-cell box-girder
6.23
bridge 96
Longitudinal normal stress on the top slab at midspan for two-cell box-
6.24
girder bridge 96
Longitudinal normal stress on the top slab at quarterspan for two-cell
6.25
Variation of concrete longitudinal normal stress along the Top Slab for
6.26
two-cell box-girder bridge 97
Longitudinal normal stress variation across the section at midspan for
6.27
Longitudinal normal stress variation across the section at quarterspan for
6.28
6.31
flange 103
6.32 Variation of transverse stresses in the bottom surface of the bottom flange 104
6.33 Variation of longitudinal stresses in the outer surface of the right web 105
6.34 Variation of transverse stresses in the outer surface of the right web 106
6.35 Variation of longitudinal stresses in the inner surface of the right web 107
6.36 Variation of transverse stresses in the inner surface of the right web 108
6.37 Variation of longitudinal stresses in the outer surface of the left web 109
6.38 Variation of transverse stresses in the outer surface of the left web 110
6.39 Variation of longitudinal stresses in the inner surface of the left web 111
6.40 Variation of transverse stresses in the inner surface of the left web 112
Structural details and cable profile of prestressing box-girder bridge with
6.41
inclined webs 114
6.42 Finite element mesh for the bridge model before and after loading 115
6.43 Deflection at cross-section for midspan for Jirousek 117
6.44 Longitudinal stresses at cross-section for midspan for Jirousek 117
IX
6.47
flange 122
6.48 Variation of transverse stresses in the bottom surface of the bottom flange
123
6.49 Variation of longitudinal stresses in the outer surface of the left web 124
6.50 Variation of transverse stresses in the outer surface of the left web 125
6.51 Variation of longitudinal stresses in the inner surface of the left web 126
6.52 Variation of transverse stresses in the inner surface of the left web 127
6.53 Variation of longitudinal stresses in the outer surface of the right web 128
6.54 Variation of transverse stresses in the outer surface of the right web 129
6.55 Variation of longitudinal stresses in the inner surface of the right web 130
6.56 Variation of transverse stresses in the inner surface of the right web 131
X
NOTATIONS
The major part of the symbols used in the text is listed below. Others are defined they
first appear.
General Symbols
[A]T,{a}T Transpose of matrix [A] and vector {a}
[A]-1 Inverse of matrix [A]
d,  Differential symbols
, det. Determinate of matrix or absolute value
{} Vector
[] Matrix
Scalar
Aps Area of prestressing steel
As Area of reinforcing steel
Cp Plasticity coefficient
dλ Plastic multiplier
Ec Modulus of elasticity of concrete
Eps Modulus of elasticity of prestressing steel
Es Modulus of elasticity of reinforcing concrete
f Yield function
fc' Uniaxial compressive strength of concrete
ft Uniaxial tensile strength of concrete
fpy Yield strength of prestressing bar
fy Yield strength of reinforcing bar
H' Hardening parameter
I1 First stress invariant
XI
I1' First strain invariant
J Jacobian
J2 Second deviatoric stress invariant
J2' Second deviatoric strain invariant
l, m, n Direction cosine of principal stresses
la The elongation of the after anchor release
q Constant profile curvature
P Force in jack end
Po Force in tendon at distance x
Pl, Pm Concentrated force at ends
Pn Normal component at distributed tensile force
Pt Tangential component at distributed tensile force
T Tensile force in tendon
u, v, w Displacement components
V Volume
W Weight of a sampling point
C, S Direction cosine of in-plane principal stresses
X, Y, Z Global coordinate system
x', y', z' Local coordinate system
Xc Vector of global Cartesian coordinates of any point of cable
 Angle change in prestressing tendon

 1 , 2 Tension stiffening parameters
β Shear retention factor or material constant

s Prestressing tendon anchor slip
γ Shear strain
XII
 1 , 2 , 3 Shear retention parameters
ε Strain
ε cu Concrete ultimate strain
εe Elastic strain
Strain corresponding to peak Uniaxial concrete compressive
εo
stress
 o Total strain corresponding to the parabolic part of uniaxial
compressive strength stress-strain curve
εp Plastic strain
 Natural coordinates for tendon
μ Curvature friction coefficient
 Wobble friction coefficient
λ Reduction factor
υ Poisson's ratio
σ Stress
o Effective stress at onset of plastic deformation
 Effective stress
τ Shear stress
 12 , 13 , 23 Shear stresses in principal planes
 12 ,  13 ,  23 Shear strains in principal planes
ξ,η,ζ Curvilinear coordinate system
Matrices
[A] Displacement gradient matrix of concrete element
[B] Strain-displacement matrix
[D] Material constitutive matrix
[D'] Constitutive matrix for steel bar
[J] Jacobian matrix
XIII
[K] Stiffness matrix of concrete element
[K'] Stiffness matrix of bar element
[Mi] Shape function of the tendon
[N] Shape function of concrete element
[T] Transformation matrix
Vectors
{a} Nodal displacement or flow vector
{B'} Strain-displacement vector of the bar element
{b} Body forces
{f} External nodal forces
{P} Internal load vector
{r} Residual load vector
{t} Surface traction
{u} Displacement vector
{σ} Stress vestor
{ε} Strain vector
XIV
Chapter One Introduction
Chapter One
INTRODUCTION
1.1 General
Box-girder bridges have been used extensively over the past decades, especially
in urban areas where aesthetics is a major principle in construction. The box-
girder cross section evolved structurally from the hollow cell deck bridge or the
T-beam bridge. The first box-girder cross-sections possessed deck slabs that
cantilevered out only slightly from the main box portion Fig. (1.1 a-e)[57]. With
prestressed concrete, the length of the cantilever could be increased. The high
formwork costs caused a reduction in the number of cells Fig. (1.1 f, g)[57]. In
order to reduce the construction loads to the minimum possible or to require
only one longitudinal girder in the working state even with multiple traffic lanes,
the one-cell built-up cross-section constructed in modular fashion emerged as
the last development, Fig. (1.1 h)[57].
The relative economy of the box-girder bridges contributed greatly to its
popularity, as it has its relatively slender and unencumbered appearance. Some
proponents of the box-girder bridge have claimed its smooth soffit to be very
desirable in urban areas for reasons of esthetics. The structural simplicity of the
box-girder bridges, particularly in continuous structures of medium to long
spans, has been well demonstrated. The efficiency of the cross-section for
positive and negative longitudinal bending moments, as well as torsional
moments is apparent even to casual observer. Special advantages for this mode
of concrete bridge construction include the low depth-to-span ratio that can be
economically achieved together with the ease with which variable conditions of
bridge width, superelevation and curvature, both vertical and horizontal can be
1
accommodated. Variable superstructure depth can also be accommodated with

relatively less difficulty[37].
The determination of the stress resultants (due to warping, shear lag, and
cross sectional distortion) and deformations in straight and curved, simple or
continuous box-girders, is complex and requires specified relationships between
geometry, loads and deformation.
For straight box girders, these stresses may be small since design loads
generally produce very little twist. In curved box-girders, however, twisting
moments are produced even for symmetric loading and distortional
deformations may be significant. Various analytical and numerical methods
have been developed for the analysis of box-girder bridges. Simple solutions for
curved box-girder bridges are only available in special cases. The curvilinear
nature of this type of structure along with the complex deformation patterns and
stress fields developed under full and partial truck loading conditions led
designers to adopt methods of analysis limited in accuracy and scope of
application due to their inherent simplifying assumptions [10].
a) c)
b) d)
e)
f)
g)
h)
Fig. (1.1) Development of the box-girder cross-section[57].
2
A box-girder bridge is a particular case of a folded plate structure where

the plate elements are arranged to form a closed section. One of the main
differences between the general shell and the box-girder (folded plate) is that in
the general shell only two elements can meet at the same edge and the change of
slope is smooth, while in the box-girder more than two elements can meet at the
same edge and at different inclinations. This causes a problem of slope
discontinuity at corners[10].
Problem of the slope discontinuity at corners by the use of brick elements
does not exist due to existing of three transition degrees of freedom at each
node.
1.2 Method of Analysis

1.2.1 Folded Plate Method
A folded plate is a three-dimensional structure of thin plane plates jointed
together along their longitudinal edges. Normally an analysis of a continuous
box-girder bridge by the folded plate method is based on the elasticity theory
developed by Goldberg and Leve (1957) and developed further by De Fries-
Skene and Scordelis (1964) and by Scordelis (1966)[21].
The method has been developed for a structure which is simply supported
at its two ends. It has end diaphragms which are considered to be infinitely rigid
in their own plane but perfectly flexible in a direction perpendicular to their own
plane.
The box- beam bridge deck may be considered as a particular type of
folded plate structure. The structure consists of a number of rectangular plates
connected at longitudinal joints. Harmonic analysis may be used to represent the
applied loads. The stiffness matrix for each plate is then expressed in terms of
the harmonics of a half-range Fourier series. The direct stiffness matrix method
is used to analyze the complete structure.
3
It should be noted that the application of the method is restricted to the

right cellular bridge decks of uniform cross section and which must be simply
supported at the extreme ends with rigid diaphragms positioned over the end
supports[21].
1.2.2 Finite Strip Method

This method was first put forward by Cheung[16] for rectangular slabs.
Subsequent rapid development has been made in three centers: in Canada by
Cheung[17], in the United States by Scordelis[58], William and Meyer[46], and in
Britain by Loo and Cusens[39].
Analysis by means of finite strips lies somewhere between the rigorous
folded plate theory and the finite element method. The behavior of a structural
shell is approximated by decreasing the amount of calculation required with any
sort of boundary conditions, a selected displacement pattern is varying
periodically in the longitudinal direction in which the boundary conditions are
satisfied at the end of supports, while in the transverse direction a polynomial
function is assumed to represent the behavior of the strip in the total
structure[4][16][17].
This method is simple but expensive. Its application is restricted to simple
boundary conditions and geometry[36]. The spline finite strip method was
devised by Cheung et al (1982)[19] to analyze right box-girder bridges with
various types of end and interior supports. In 1988, the spline finite strip method
was extended by Li et al[36] to the elasto-plastic analysis of circular and
noncircular box-girder bridges.
1.2.3 Finite Element Method:

The finite element method is the most versatile and appropriate numerical
method that can cater to most of the following requirements: detailing,
geometric and material behavior, loading characteristics, and the boundary
4
conditions of the structure and any significant interaction among them. The
finite element method employs an assemblage of discrete one, two and three-
dimensional members to represent the structure. The structure is divided into
elements that are only connected at their nodes, Fig. (1.2), which possess an
appropriate number of degrees of freedom[57].
Recently the development of two and three-dimensional finite elements
capable of simulating the actual nonlinear behavior of prestressed and reinforced
concrete after the formation of cracks has been a considerable progress.
Thus, the finite element method may be seen to be very general in
application and, indeed, for difficult bridge deck problems it is sometimes the
only valid form of analysis[30].
applied loading
top flange
finite element
nodal points
web
fine mesh
bottom flange
coarse mesh
Fig. (1.2)Three-cell girder being discretized into finite element[57].
1.3 Objective and Scope

This work is devoted to study the overall prestressed concrete box-girder bridges
subjected to monotonically increasing load. In order to achieve this main
objective, a computer program is used, which was originally developed by Al-
Shaarbaf[9], but modified to be capable of analyzing prestressed concrete box-
girders, by developing a system to include the effect of prestressing in the
element formulation. The modified computer program has also been used to
study the material nonlinearities due to cracking of concrete, plastic flow or
crushing of concrete, and yielding of reinforcing bars. The maximum tensile
5
strength with stiffening concept is used to describe the post-cracking behavior of

concrete. The aggregate interlocking effect and the dowel action are
incorporated in the finite element model by using an appropriate shear retention
model.
To demonstrate the applicability of the modified computer program,
different kinds of examples have been selected, such as, straight bridges with
single or two cells and with various boundary conditions, and various loading
cases. The behavior of load-deflection curves and comparing these with the
available experimental data are fully examined. Besides, the stress distributions
under various affecting parameters are investigated.
1.4 Layout of Thesis

This thesis is divided into seven chapters. A general introduction describing the
significance of prestressed concrete in construction and the objectives of the
investigation are presented in Chapter One. Chapter Two reviews the research
works on the prestressed and reinforced concrete box-girders. In Chapter
Three, the basic concept of the finite element method and the derivation of the
governing equilibrium equations by using the principle of virtual displacement
are given. The 20-node isoparametric brick elements that have been used to
model the concrete and the embedded reinforcement are described. The
numerical integration scheme used to determine the element stiffness matrix and
the solution of the nonlinear equations of equilibrium are also presented.
Material constitutive models are presented to describe the idealization of
concrete in compression and in tension, and the representations of reinforcement
steel are given in Chapter Four. Chapter Five presents the basic procedure for
the formulation and solution algorithm used to analyze prestressing problem. In
Chapter Six, several numerical examples are presented and verified with
previous experimental studies. Chapter Seven contains several conclusions and
recommendations for future studies on this subject.
6
Chapter Two Review of Literature
Chapter Two
REVIEW OF LITERATURE
2.1 Introduction
During the past three decades, interest in the nonlinear analysis of concrete
structures has increased steadily. The nonlinear finite element method of
structural analysis has been ensured the wide use of both reinforced and
prestressed concrete structures. This is because of the development of a
relatively powerful analysis technique implemented on electronic computers.
In this chapter, a brief review of literature of the main contributions in the
field of the analysis of box-girder bridges is presented. However, a literature
review for more specific area will be included in the following sections related
to this topic. This is believed to give a clearer picture to the several topics which
this study deals with.
2.2 Analysis of Box-Girder Bridges

The earliest published study of box girders belongs to Scordelis A. C., Davis R.
E. and Lo K. S. in 1969[60]. In their study, theoretical solutions were developed
using the folded plate theory, for the elastic analysis of multi-celled box-girder
bridges with or without interior diaphragms or supports. The results of
parametric studies with respect to load distribution on example bridges with
3,4,6 and 8 cells were analyzed. The study showed that for box-girder bridges,
load distribution is influenced by the number of parameters such as the number
and the dimensions of cells, depth to span ratio, width to span ratio, number of
diaphragms and other factors. Also, it was apparent that a box-girder bridge is
inherently stiffer in a transverse direction than the slab-on-beam bridges, which
lack transverse continuity in the bottom flange.
7
In 1970, Meyer C. and Scordelis A. C.[46] presented an alternative general

computer program CURSTR, based on the curved strip method, which can be
used for circularly curved box-girder bridges. Using this approach they also
treated sloping decks and webs and box girders having any desired orthotropic
material properties. But with that analysis, the structure must be simply
supported along the straight radial edges[63].
In 1974, Yoo C. H., Evick D. R. and Heins C. P.[67] used the finite
difference technique to solve Vlasov equations for the analysis of single and
continuous span curved girders. The solution by this means permits the inclusion
of point-wise property variations (Io, Kt, Ix) and the interaction of the vertical
and torsional deformations. The solution of two simultaneous equations gives
deflections and rotations at each point. The forces (bending moment, shear, pure
torsion, warping torsion, and bimoment) at each of these nodes are then
calculated by using differences relationships. The entire difference analysis has
been programmed for use on an UNIVAC 1108 computer, with FORTRAN IV
language. In order to demonstrate the versatility and reliability of the computer
program, three example problems were given, prismatic curved girder, non-
prismatic straight girder and two span non-prismatic curved girders. Good
results were obtained when comparing with available exact solution.
In 1976, Heins C. P. and Oleinik J. C.[30] used the finite difference
procedure to determine the response of a single and a multi-span curved single-
box beam bridge with any number of interior diaphragms. They divided the
analysis of the box-girder into two parts, with the effects of both parts being
summed to give the total normal stress distribution. In the first part the load-
deformation response of the box-girder by a formulation developed by
Vlasov[66] was used to predict the behavior of the rigid section. In the second
part, the effects of the cross sectional deformations were taken into
consideration. These deformations were predicted by an equation given by
Darbrowski[22]. The formulations based on numerical solutions are more
8
amenable to a separate evaluation of the effects of distortion, warping, and shear

lag although the generality of solutions seem to be more limited. The
distortional effects considered by Heins and Oleinik are limited to consideration
of a single angular change parameter rather than the general transverse bending
of the cross section[65].
In 1985, Scordelis A. C., Chen E. C. and Ketchum M. A.[59] demonstrated
the capabilities of three computer programs MUPD4, CURDI4 and CELL4
using folded plate method, finite strip method, and finite element method
respectively. The MUPD4 consists of plate elements extending with the length
of the structure. One element between webs is sufficient for modeling the top
and bottom slabs by the elasticity elements used and thus resulting in 15
elements and 12 joints. The CURDI4 model consists of straight finite strip
elements extend with the length of the structure. Two elements between webs
were used for the top and bottom slabs for the finite strip elements to accurately
predict local bending moments, and each web was subdivided into two elements,
thus resulting in 30 elements and 27 joints. 100 Fourier series terms were used.
The CELL4 model is a finite element mesh with one element over the depth of
each web, two elements between webs for top and bottom slabs and 10 elements
over the length of each span, resulting in a total of 460 elements and 231 nodes.
MUPD4 and CURDI4 give identical results, while CELL4 gives slightly smaller
(stiffer) displacements and moments.
In 1989, Mavaddat S. and Mirza M. S.[44] presented three computer
programs, written in FORTRAN WATIF, which were developed to analyze
straight, monolithically cast, symmetric concrete box beams with one, two or
three cells and side cantilevers over a simple or over two spans with symmetric
mid-span loadings. The analysis, based on Maisel’s formulation, is performed in
three stages. First, the structure is idealized as a beam and the normal and shear
stresses are calculated using the simple bending theory and St-Venant’s theory
of torsion. The secondary stresses arising from torsional and distortional
9
warping and shear lag are calculated in the second and third stages, respectively.
These stresses arising in each stage of analysis are then superposed to determine
the overall response of the box section to the applied loading. The results of only
one example were presented for illustrating the application of the computer
programs. They agree precisely with Maisel’s detailed hand-calculation results.
In 1990, Abdullah M. A. and Abdul-Razzaq A. A.[1] used the same
examples that were analyzed by Scordelis et al[59] but they used the finite strip
method for the analysis of a prestressed concrete box-girder bridge using higher
order bending and in-plane strips. The auxiliary nodal line technique was
adopted for both bending and in-plane strips. A flexibility procedure was
incorporated for the solution of continuous box-girder bridges when redundant
forces are induced by intermediate supports. Extension of the finite strip method
to the analysis of prestressed forces was described. The higher order finite strip
gives good results for the analysis of both simply supported and continuous
prestressed box-girder bridges compared with the elasticity solution.
In 1993, Abdullah M. A. and Abdul-Razzaq A. A.[2] applied a refined
finite strip method for the analysis of continuous box-girder bridges using higher
order bending and inplane strips. The auxiliary nodal line technique has been
adopted for both bending and inplane actions. A flexibility approach was used
for the solution of continuous box-girder bridges. Sixth order bending strip was
combined with third order inplane strip to form a flat shell strips. At strip
boundaries, displacement parameters were similar to those of lower order finite
strip. Static condensation technique was adopted to eliminate the auxiliary nodal
line parameters for both bending and inplane actions at strip level. Therefore, the
solution is marginally higher than the lower finite strip although extra manual
effort is necessary in developing the condensed stiffness and force matrices at
the strip level. Afterwards Abdul-Razzaq A. A. developed and used a higher
conical frustum shell strip to analyze curved box girders[4].
10
Later, a finite strip method for the elastic analysis of simply supported and
continuous prismatic cellular box-girder bridges was presented by Abdul-
Razzak[3]. Higher order finite strip method with auxiliary nodal line technique
has been used for the analysis. A flexibility approach is incorporated.
2.3 The Finite Element Method

The earliest published application of the finite element method to box girders
belongs to Mehortra B. L., Mufti A. A. and Redwood R. G. in 1969[45]. In their
study, they used triangular elements for analyzing thin walled box-girder bridge.
The structure is analyzed by a system of flat triangular plate elements. The
solving program incorporates two different functions, which describe the
element bending behavior. One is a simple noncompatible function, which has
been in plate bending problems, and the other consists of a set of linear
functions. The compatibility between adjacent elements is achieved by the
imposition of linear displacement and normal slope variation on the element
side. The program uses six degrees of freedom at node. Close agreement was
obtained with experimental results of Sawko and Cope[56].
In 1971, Aneja L. K. and Roll E.[13] performed a model study for box-
girder bridge. The investigation was done on a curved box-girder made of
Plexiglass with geometry and cross section, as shown in Fig. (2.1). Flat plate
element was used to analyze this model. The element behavior is represented by
an assumed displacement field with five degrees of freedom. The sixth degree of
freedom (rotation about normal) was neglected[32]. The experimental results
obtained disagree considerably with the finite element analysis made by the
same author (according to Zyle S. F. and Scordelis A. C.[70]).
11
Fig. (2.1) Curved box-girder of Aneja and Roll.
In 1973, Tirkha D. N. and Edwards A. D.[64] presented an automatic

incremental iterative finite element procedure for the analysis of prestressed box
girders subjected to combined bending and torsion in the uncracked and post-
cracking ranges. The analysis included the nonlinear behavior of concrete, the
change of structural topology due to the initiation and propagation of cracks and
the provision for the gradual but progressive breakdown of bond. Rectangular
and triangular membrane elements were used to idealize the concrete. The
triangular elements were used adjacent to the inclined prestressing steel. Since
the flexural rigidity of the elements forming the structure was neglected, large
distortional warping stresses were developed in the post-cracking phase, leading
to unstable transverse cracks. In order to include the distortional rigidity of the
cell, imaginary diaphragms at each mesh line was introduced with stiffness
made equivalent to the distortional rigidity of the cell. The contribution of
prestressing steel to the overall stiffness was neglected as long as tendons were
not grouted. Grouted tendons were assumed to be completely bonded with their
stiffness included in the overall structure stiffness. Fairly good results were
obtained when compared with experimental results.
In 1975, Fam A. and Turkstra C.[24] developed a new finite element
scheme for box-girder analysis. The 12-degrees of freedom nonconforming
12
rectangular plate element have been used for webs and the 12-degrees of
freedom annular plate bending element for flanges. Good agreement had been
obtained with both curved folded plate theory and finite strip method.
The same authors made a model study in 1976[25] to investigate the effect
of different parameters in the linear elastic region. A three-dimensional finite
element program was developed. Plexiglass was chosen as the model material.
Three kinds of elements were used; for flanges the annular element with corner
nodes, cylindrical elements for webs, and rectangular elements for diaphragms.
The results of three-dimensional finite element program developed for this
purpose were found to be in good agreement with the experimental results for
deflections, radial stresses, and tangential stresses.
In 1977, Scordelis A. C. and Larsen P. K[61][62] made a model study on
multi-cell reinforced concrete box-girder. The model construction was similar to
that used for the prototype in the field. In that study theoretical and experimental
results were compared for reactions, deflections, and moments due to point
loads. Three different analytical methods were used. The computer programs
associated with these three analytical solutions are titled SAP, CURDI, and
CELL. The SAP program was used to analyze the bridge as an idealized three-
dimensional frame using one-dimensional members. A total of 24 straight beam
elements were used to model the curved bridge. In each beam member’s section
properties represent the entire four-cell bridge cross section. In CURDI analysis
the bridge structure was modeled as an assemblage of flat and curved plate
elements spanning between the end supports. The solution was based on the
harmonic finite strip analysis. The CELL program is a general finite element
program for cellular bridges, in which the structure is analyzed as a three-
dimensional assemblage of plane plate elements representing top and bottom
flanges and vertical webs. Both CELL and CURDI accounted for both the
membrane and plate bending action in the elements comprising the structure.
From this study it was obvious that SAP program can only give an indication of
13
the longitudinal distribution of total reactions, moments, and the center-line

deflection. It gives no information on the transverse distribution of these
quantities or the internal membrane forces and plate bending moments in each
element, which were obtained from CURDI and CELL programs. The results
from CELL were used in the comparisons between the theory and the
experiment because CELL solution allows more accurate representation of the
bridge structure. The theoretical and experimental results show only minor
difference.
In 1978, Zyle S. F. and Scordelis A. C.[70] used the skew-ended element
and analyzed the same examples of Anja and Roll. In their study the highly
curved box beam is approximated by a series of skew-ended elements. In this
element the cross-sections are not required to be normal to the element axis. The
end of an element can be made skew so as to coincide with the end of adjacent
element meeting at an angle. This makes the element fully three-dimensional
and compatible. Fig. (2.2) illustrates the division of a curved box-girder into
skew-ended finite elements as used. There was a close agreement between the
results obtained by Scordelis method and the experimental results of Anja
model. Scordelis extended his method to take into account prestressing and
time-dependent deformations. The equivalent load method is used to compute
the forces applied by tendon upon the structure.
In 1978, Turkstra C. J. and Fam A. R. M.[65] performed a parametric study
on a curved box-girder. The method utilized annular flange elements, cylindrical
web elements, and rectangular in-plane elements for diaphragms. The degrees of
freedom and displacement functions for all elements were chosen specifically to
model the curved box behavior as in Ref. 25. Numerical results have been
selected to indicate the influence of curvature, web spacing and the number of
intermediate diaphragms. From the preceding study it was concluded that under
certain conditions maximum flange stresses could be significantly greater than
those predicted by the curved beam theory. Also in the curved box-girder, web
14
spacing and curvature affect the moment distribution and peak longitudinal in-
plane stresses even under symmetric loading. Qualitatively, an increase in
curvature or increase in web spacing increases the difference in flexural stiffness
of the inner and outer halves of the cross section. A slight reduction of the
maximum stress occurred with introduction of intermediate diaphragms.
Fig. (2.2) Skew ended element used by Scordelis.
In 1979, Jirousek J., Bouberguig A. and Saggun A.[32] presented an

efficient macro-element formulation for static analysis of curved box-girder
bridges with variable cross section. The macro-elements were having the form
of transversal slices of identical topology. Each slice is formed using two types
15
of special purpose elements; a modified Ahmad’s thick shell element and an

assembly element, while the purpose of the latter element is to enable
assemblage of up to four thick shell elements meeting at different angles,
Fig. (2.3). The effect of prestressing is properly taken into account in the form of
appropriate local forces considered at the element level. Three examples were
analyzed, a straight single cell box-girder bridge, a curved box-girder bridge and
a prestressed concrete, straight box-girder bridge. The analysis gave good results
comparing with the experimental or other theoretical results.
Fig. (2.3) Box-girder macro-element formed by an assembly of

special-purpose elements: superparametric assembly element (1),
modified superparametric thick shell element (2).
In 1982, Ghali A. and Ghonhm G. A.[27] developed a computer program to

determine the response of reinforced and prestressed concrete structure such as
box-girder, that is, any spatial structure that can be idealized as an assemblage of
thin plate elements subjected to bending and membrane forces. Each element
was assumed to be composed of a number of layers and the stiffness was
allowed to vary over the area of the element and from one layer to the other
according to the associated state of stress and loading history. The reinforcement
16
is idealized either by smearing a mesh of bars into a layer, or as a discrete bar

element when heavy reinforcement or prestressing is presented.
In 1984, Zhang S. H. and Lyons L. R.[69] had presented another approach
for the analysis of single and multi-cell box-girder bridges. In this approach the
thin-walled beam theory established by Vlasov has been directly combined with
the finite element by including three extra degrees of freedom over the normal
six degrees of freedom in beam formulation, to take into account the warping
and distortional effects as well as shear, Fig. (2.4). The results produced were
identical to the theoretical differential equation solution given by Heins and
Oleinik, which was regarded as an exact solution.
Fig. (2.4) Thin-walled beam element with three nodes.
In 1989, Razaqpur A. G., Nofal M. and Mirza M. S.[53] utilized the

nonlinear program NONLACS to analyze single-cell and two-cell prestressed
concrete box-girder bridges. In this study, the four-node quadrilateral membrane
element and the rectangular plate bending element are used to model the box-
girder. The two elements are combined to produce an isotropic facet shell
element in which the membrane and bending actions are coupled, Fig. (2.5). The
membrane element has three degrees of freedom at each node. The reinforcing
and prestressing steel could be idealized as a steel layer or as a one-dimensional
bar. The latter element is suitable for the large reinforcing or prestressing bars
and tendons, while the former element is appropriate for distributed steel in
slabs and walls. The NONLACS program used an incremental-iterative
17
approach, based on the tangent stiffness method, to solve the nonlinear problem
in a series of incrementally linear analysis. For comparison, the load-deflection
curves show that the predicted and measured strength of the bridges are in
excellent agreement for the single-cell box while for the two-cell box, the
difference being less than 4.3 %. However, the overall analytical response is
slightly stiffer than the experimental response.
Fig. (2.5) Typical shell element used in the analysis.
In 1990, Ghalib A. M. A.[28] analyzed three-dimensional, multi-planar

concrete structures such as box girders which had been treated under short term
loading. These multi-planar systems were represented by an assemblage of a set
of isoparametric, layered, Q8 flat shell elements with a total of six degrees of
18
freedom per node, as shown in Fig. (2.6). Reinforcement was represented by

either a smeared, fully bonded steel layer of orthotropic properties to represent
equally spaced reinforcement arrangements, or as a discrete, generally curved
steel strand element to model special heavy reinforcement concentrations.
Prestressing tendons are considered as an integral part of the structure.
Generally, curved bonded/unbonded prestressing strand element embedded in
the Q8 flat shell element has been developed. Several short-term prestress losses
are included in this study. Several numerical examples were worked out and
compared with existing experimental studies in order o demonstrate the validity
and efficiency of the proposed method. Good agreements between the results
were obtained.
Y, v, Py
θ y , My
Z, w, Pz
σy
θz , M z τxy
σx
τxy midsurface
X, u, Px
shell layers θ x , Mx
Fig. (2.6) Typical 8-noded, 6 D.O.F. layered isoparametric flat shell element.
In 2002, Al-Temimi J. M. E.[10] adopted Ahmad’s degenerated thick shell

element and modified it for the analysis of reinforced and prestressed concrete
box-girder bridges, as shown in Fig. (2.7). The box with its two cantilever
flanges was simulated by an assemblage of a set of shell elements, with five
degrees of freedom at each of the nine nodes of the element, three displacements
and two local rotations at each node. The problem of slope discontinuity at
corners and where more than two elements are meeting at angle has been solved
19
by adopting the technique of transformation of the two local rotations to three

global rotations at the nodes of the junctions. The equivalent nodal forces
concept was used to include the effect of prestressing. The effect of intermediate
diaphragms on transverse distribution of stress and deflection, as well as the end
diaphragms has been simulated by using the same thick flat shell elements.
Several numerical examples were analyzed and compared with available
experimental and theoretical studies in order to demonstrate the validity and
efficiency of the proposed method in the analysis of straight and curved single
and multi-cell bridges. Good agreement between the results was obtained.
ξ
ξ
top surface
η
ζ midsurface
(b) bottom surface

(a) normal
ζ
Fig. (2.7) Degenerated Mindlin-type thick curved shell element.
In the present study, the 20-noded isoparametric brick element has been
used to model the concrete, each node has three transition degrees of freedom.
The reinforcing bars have been idealized as axial members within the concrete
elements. Bond slip between the steel and concrete is considered. The equivalent
load method is used to compute the force applied by the tendon upon structure.
The program is extended to include the prestressing effects. The program is
applied to several prestressed box-girder bridge.
20
Chapter Three Formulation of Finite Element & Non-Linear Solution Technique
Chapter Three
FINITE ELEMENT FORMULATION
3.1 Introduction
The finite element method is a popular numerical technique that has been widely
used by engineers and scientists, besides it is a very powerful and modern
computational tool. This technique has the potential to play an increasingly
important role in all areas of reinforced and prestressed concrete research,
design, and analysis.
The finite element method consists of replacing the actual structural
system being analyzed by an assemblage of finite elements. These elements may
be one, two or three-dimensional, each of them is bounded by a defined number
of nodes. These nodes represent the points of connection between adjoining
elements.
3.2 Formulation of Element Stiffness Matrix

For a general three-dimensional body, subjected to a system of external forces
which consists of surface traction and body forces, the individual element
stiffness matrix can be determined easily by using the principle of virtual
displacements.
The shape functions N  are used to express the displacements at any
point within the element, ue , in terms of the nodal displacements, ae , as:
ue  N ae …(3.1)
By differentiation of the displacements, the corresponding strains  e ,
can be written as:

 e   Aue …(3.2)
21
where,  A , is the matrix of differential operators. Substituting of Eq. (3.1) into

(3.2) yields:
 e  B ae …(3.3)
where, B  , is the strain-displacement matrix given by:

B    A .N  …(3.4)
The stresses  e can be determined from the corresponding strains, using the
general stress-strain relationship as:

 e  D  . e …(3.5)
where  e is the stress vector, given by:
 e  [ x  y  z  xy  yz  zx ] T
…(3.6)
D is the constitutive matrix and  e , is the strain vector given by:
 u    
 x   x 0 0
 v    
 x    0 0
    y   y 
  y   w    u
  z   z   0
 0  
z   
 e       u v      v  …(3.7)
 xy      
0 w
 yz   y x   y x  
   v w    
 zx     0 
  z  y   z y 
  u  w   
0
 z x   z x 
Substitution of Eq. (3.3) into Eq. (3.5) gives the stress-displacement
relationship:
 e  DB ae …(3.8)
The principle of virtual displacements of a deformable body is used to
establish the governing equations of static equilibrium. It states that,” if a
general structure in equilibrium is subjected to a system of small virtual
22
displacements with a compatible state of deformation, the virtual work due to

the external action,  W ext , is equal to virtual strain energy due to the internal
stress,  W int [9][23] “, thus:
 W int   W ext  0 …(3.9)
A system is considered of a volume V and a surface area s subjected to
body force bi and surface traction t i , then the external work done in moving
these forces through the virtual displacement u , is as:
 W ext   uT bdV   uT t ds …(3.10)

V s
The internal virtual work is given by:
 W int    T  dV …(3.11)

V
Substitution of Eq. (3.5) into Eq. (3.11) yields:
 W int    T D  dV …(3.12)

V
By making use of Eqs. (3.10) and (3.12), Eq. (3.9) may be expressed as:
   D dV   u bdV   u t ds  o

T T T
…(3.13)
V V s
The above expression represents the equation of static equilibrium for a

general body. Therefore, by making use of Eqs. (3.1) and (3.3), the discretized
form of Eq. (3.13) can be written as:
T  
a    B  D B  dVeae    N  be dVe   N  t e ds e   0
T T T
 n Ve n Ve se 
…(3.14)
where n is the total number of the elements of the discrete system.
Since the vector of the virtual nodal displacements, a is arbitrary, the
T
following set of algebraic equations may be obtained:

 f   K a …(3.15)
23
where K  is the stiffness matrix of the element assemblage and it is given by:
K    k     B T DB dVe …(3.16)

n n Ve
where k  is the element stiffness matrix. While,  f , is the element assemblage

of external nodal forces vector given by:
 f     N T be dVe    N T t e dse …(3.17)

n Ve n se
3.3 Material Representation

In the present work, concrete is simulated by hexahedral brick elements and the
reinforcing steel bars are modeled as one-dimensional elements embedded into
the three-dimensional concrete elements.
3.3.1 Concrete Idealization

The quadratic 20-nodes brick element shown in Fig. (3.1) is adopted in the
present study. This element has been successfully used in three-dimensional
non-linear reinforced and prestressed concrete analysis[9][12]. The mid-side nodes
must be positioned the same way as the sides are divided into two equal
segments in local coordinate system.
Fig. (3.1) 20-Noded isoparametric brick element[9].
24
3.3.1.1 Shape Function

The shape functions for the 20-noded brick element are introduced to express
the coordinates and displacements within the element in terms of the nodal
coordinates and displacements (isoparametric element). The origin of the local
coordinates system is placed at the center of the brick element, each of the local
coordinate lines ranges from (-1) to (+1) and given in terms of  , and  as
shown in Fig. (3.1). The displacements of the brick element are given by:
20
u , ,    Ni  , ,  . u i
i 1
20
v  , ,    Ni  , ,  . v i …(3.18)
i 1
20
w  , ,    Ni  , ,  . w i
i 1
where Ni  , ,  is the shape function at the i-th node and ui , v i and w i are
the corresponding nodal displacements. The shape functions for the 20-noded
brick element are given in Table (3.1).
Table (3.1) Shape functions for the 20-node hexahedral element.
Location ξi ηi ζi NI=(ξi, ηI, ζi)
0.125(1+ξξ i)(1+ηηi)
Corners ±1 ±1 ±1
(1+ζζi)( ξξ i+ ηηi+ ζζi-2)
Mid-Side
0 ±1 ±1 0.25(1-ξ2)(1+ηηi)(1+ζζi)
Nodes
Mid-Side
±1 0 ±1 0.25(1+ξξ i)(1-η2)(1+ζζi)
Nodes
Mid-Side
±1 ±1 0 0.25(1+ξξ i)(1+ηηI)(1-ζ2)
Nodes
ξi, ηi, ζi denote the coordinates of i-th node.
25
3.3.1.2 Evaluation of Element Stiffness Matrix

The shape function Ni , is a function of both local and global coordinates, while
the strains in Eq. (3.7) are functions of the global coordinates. Therefore, a
relationship between the derivatives in the two coordinate systems must be
defined. The total derivatives of a typical shape function Ni are given as:
 N i   x y z   N i   N i 
        
        x
   x 
 N i   x y z   N i   N 
     J  i  …(3.19)
          y   y 

 N i   x y z  N i 
  N i 
         z   z 
  
where, J  is the Jacobian matrix.

The global coordinates in terms of the local coordinates are:
20 20 20
x  N i xi , y   N i yi , z   N i zi …(3.20)
i 1 i 1 i 1
where x i , y i and z i are the nodal coordinates.

Using of these relations, the Jacobian matrix J  in Eq. (3.19) becomes:
 20 N i 20
N 20N i 
 xi   i y i   z i 
i  1   i1 i 1 
 20 20 20 N 
J     N i x i 
N i
yi  i
z i …(3.21)
 i  1  i1   i  1  
 20 N 20
N i 20 N 
 i
xi   y i   z i 
i
 i  1  i1 i 1 
The global derivative is found by inverting J  , as:
 N i   N i 
    
  x   
 N i   1  N i 
   J    …(3.22)
  y    
 N i   N i 
 z    
 
26
For a three-dimensional element of volume V , the differential volume

dV e , can be written as:
dVe  dx .dy .dz …(3.23)
Also it can be written for local coordinates:
dVe  J d .d .d …(3.24)
where J is the determinant of the Jacobian matrix.

By substituting Eq. (3.24) into (3.16) and setting the limits of integration
in the coordinates to be varied from (-1) to (+1), the element stiffness matrix in
the local coordinates may be written in the form:
1 1  1
K e     B T DB  J .d .d .d
 1 1 1
…(3.25)
Numerical integration (Guass quadrature) is used to evaluate the element
stiffness matrix K e given by Eq. (3.25).
3.3.2 Reinforcing Bar Representation

Representation of reinforcing steel in a finite element model exists generally in
three types, these are as follow:
1- Distributed Representation:
In this type of representation the reinforcement is assumed to be distributed over
the element in any specified direction, Fig. (3.2.a). A composite concrete-
reinforcement constitutive relation is used in this case. Perfect bond has been
assumed between the concrete and reinforcement when deriving the constitutive
relations.
2- Discrete Representation:
In this type, one-dimensional discrete element has been used to idealize the
reinforcement, Fig. (3.2.b). Axial force members and bar links may be used and
assumed to be pin-connected with two degrees of freedom at each connection.
27
Elements with three degree of freedom at each nodal point such as beam
elements subjected to axial, shear and bending forces can also be used. The
major advantages of this type of representation are its simplicity and ability to
represent the bond slip and dowel action phenomena. A major disadvantage of
this type is the extremely expensive computation time since a large number of
elements are implemented.
3- Embedded Representation:
This approach is often used in connection with high-order isoparametric
elements. In this approach perfect bond (or specified bond-slip relation) is
assumed to occur between the reinforcing bars and the surrounding concrete.
This assumption of compatibility of displacements and strains between the steel
and concrete allows the reinforcing bars to be treated as integral parts of the
element[68]. The stiffness of steel bars is added to that of the concrete to obtain
the global stiffness matrix of the element. Reinforcing bars are assumed to be
capable of transmitting axial forces only. It is further assumed that such bars are
restricted to be parallel to the local coordinate lines  , and  of the brick
element.
In the current study, the embedded representation is adopted. The
derivation of the stiffness matrix for a bar parallel to the local coordinate axis 
(for example) is represented in this section. A similar derivation can be used for
bars parallel to  and  axes.
A bar is considered lying parallel to the local coordinate axis  with
   c and    c , Fig. (3.2.c). The shape functions of the brick element can be
used to represent the displacements of the bar[9]:
20
u   Ni (  ).ui
i 1
20
v   Ni (  ).vi …(3.26)
i 1
20
w   Ni (  ).wi
i 1
28
While the strain-displacement relationship is given by:

 C 1 C2 C 3   N i x    u i 
20
 1 C  
   2  2 C4 C 5   N i y    v i  …(3.27)
i 1 h
 C 3 C5 C 6   N i z    w i 
where,
h  C 12  C 42  C 62 …(3.28)
and
x 2 x  y x z
C1  ( ) ,C 2  ( ), C 3  ( )
     
…(3.29)
y 2 x  z z 2
C4  ( ) ,C 5  ( ) ,C6  ( )
    
In Eq. (3.16), dV can be written as: dV  As .dx  As .h .d . Finally the
stiffness matrix of the embedded bar K e can be expressed as:
1
K e  As  B T D B  .h. d …(3.30)
1
where, As is the cross sectional area of the bar, and D is the constitutive
matrix that represents the modulus of elasticity of the steel bar. Similar
derivation can be used for bars parallel to  and  axes.
29
a-Distributed representation
Axial Element Flexural Element

b-Discrete representation
c-Embedded representation
Fig. (3.2) Alternative representation of reinforcement.
30
3.3.3 Bond and Bond Slip Representation

The interaction between concrete and steel is one of the most important factors
behind the wide spread use of reinforced concrete as a structural material. Such
interaction, or stress transfer, between concrete and steel is called bond. It is
made of the following component[41]:
1) Chemical bond.
2) Friction.
3) Mechanical interaction between concrete and steel.
Bond of plain bars depends mainly on the first two factors. Deformed
bars, however, depend on the first two factors for bond only at early loading
stages beyond which the mechanical interaction becomes predominant.
As a result of the transfer of stress by the mechanical interaction, relative
movement or bond slip takes place.
In the finite element modeling of reinforced concrete structures, the bond
between reinforcement and concrete is mostly idealized by discrete springs
connecting the bars at different points along its length to the concrete. Each
spring represents the bond resistance along its tributary length of the bar. The
horizontal spring represents the bond slip relationship, while the vertical spring
has a very high stiffness value to restrain the vertical movement of the
reinforcement relative to the concrete. So the need for pairs of extra nodes along
each bar and on the adjacent concrete greatly increases the cost of the
analysis[11].
For the embedded bar model, perfect bond between steel and concrete was
assumed in previous research, but in the recent studies, Allwood et al[11],
Mahmood[43], and Kindeel[34], bond slip between steel and concrete is
considered.
Since the problem of bond and bond slip involves many variables, so the
best that can be done now is to try fitting experimental data.
31
The present bond and bond slip formulation is based on the experimental
results of Nilson[48]. In the study of bond slip, Nilson found that the bond slip
relationship depends on the compressive strength of concrete as well as the
distance from loaded face (or face of crack), C. The results of Nilson are
illustrated in Fig. (3.3).
In this study, the bond-slip curve with C=152 mm (6 in.) is used. This
curve represents the upper bound of the peak bond stress of experimental results
of Nilson. Two polynomials are used to describe this curve, one for ascending
portion, and the other for the descending part, as proposed in Refs.[7] and [8]:
u
 0.083 ( 7.5  3  25  2  27.5  ) for 0    1 …(3.31)
f c
u
 0.083 ( 2.5  3  15  2  22.5  ) for   1 …(3.32)
f c
where u = bond stress, MPa.

 , normalized slip.
p
 = slip, mm, and  p = slip at peak bond stress, mm.
To obtain the bond stiffness Kb, Eq. (3.31) and Eq. (3.32) are
differentiated with respect to the slip  :
f c
Kb 
u

 0.083
p
 
22.5 2  50  27.5 for 0    1 …(3.33)
f c
Kb 
u

 0.083
p
 
7.5 2  30  22.5 for   1 …(3.34)
Since the stiffness of steel bar is calculated separately, as in Eq. (3.30), by

assuming fixed bond, and then added to the concrete element stiffness, then, to
account for bond slip, the steel axial stiffness is reduced by the bond slip
stiffness[34].
32
Slip (mm)
C  6'' (15.3 cm)

Bond Stress Ratio
C=4'' (10.2 cm)
C=3'' (7.6 cm)
C=2'' (5.1 cm)
Slip  10-3( inches)
Fig. (3.3) Representative bond stress-slip curves.[48].
3.4 Numerical Integration Technique

The evaluation of the element stiffness matrix involves some difficult
integration. Explicit integration might be difficult or even impossible, for such
functions. Usually the Gauss-Legendre scheme is used to perform the
integration required to set up the element stiffness matrix. This method has been
found to be both accurate and convenient for finite element work[68].
By using this technique, Eq. (3.25) can be rewritten in the following form:
 1 1 1
 
ni nj nk
I    f  , , d .d .d     w i . w j . w k . f  i , j , k …(3.35)
 1 1 1 i 1 j1 k 1
where ni , n j and nk are the numbers of Gaussian points in  , and 
direction respectively, w i , w j and w k are the weight factors of the i-th, j-th and
33
k-th integration points and 

f  i , j ,  k  represent the results of the
multiplication B DB J . In general, the number of integration points is

T
taken equal in the three directions.

In a similar manner the steel bar element stiffness matrix can be written
as:
1 ni
K e   f (  ) . d   wi . f (  i ) …(3.36)
1 i 1
In the present study, the 27 (  3  3  3 ) points of Gauss-quadrature

integration rule are employed. The relative positions of the sampling points for
this rule over the volume of the quadratic brick element are shown in Fig. (3.4).
The corresponding weights and locations are defined in Table (3.2).
Fig (3.4) Distribution of the sampling points over the element in

the 27-gaussian points integration rule.
34
Table (3.2) Weight and Location of Sampling Points in the 27-Point Rule.
Sampling Location Coordinates
Weight
Integration Rule Points ξ η ξ
1,3,5,7,19,21,23,
The 27-Point
±0.7746 ±0.7746 ±0.7746 0.171468

25
2,4,6,8,10,12,14,
±0.7746 ±0.7746 ±0.7746 0.274348
16,20,22,24,26
9,11,13,15,17,27 ±0.7746 ±0.7746 ±0.7746 0.438957
18 ±0.7746 ±0.7746 ±0.7746 0.702332
3.5 Nonlinear Solution Techniques

One of the main objectives of the finite element analysis of reinforced and
prestressed concrete structures is to determine the response of the structure
under loading. Fig. (3.5) shows a typical load-deformation curve for a
monotonically loaded member, the response up to point A is linear, beyond this
point a nonlinear load-deformation response occurs. Such a response is due to
either material or geometrical nonlinearity or a combination of both.
The material nonlinear behavior includes sudden changes in the element
stiffness due to cracking and crushing of concrete and yielding of steel
reinforcement, while the geometrical nonlinearity is related to the large
deformations. In the present study, material nonlinearity due to material
nonlinear behavior is considered.
35
Ultimate Load
Load
A
Deformation
Fig. (3.5) Typical structure response.
3.6 General Nonlinear Solution Procedure

Applying the finite element method for any problem leads in general to a
set of algebraic equations (linear or nonlinear) in the following form:
[ K ] a  P …(3.37)
In a simple linear elastic problem the solution for these equations can be
obtained directly. However, the direct solution cannot be achieved when
nonlinearity is present in the stiffness matrix [ K ] , which will depend on the
displacement level. Therefore, the solution cannot be exactly obtained before the
determination of the unknown nodal displacements a. The solution of
nonlinear problems is usually attempted by one of the following techniques:
3.6.1 Incremental Technique

In this technique the load is applied on the structure as a series of equal or
unequal increments and if such increments are sufficiently small, then the
behavior within each increment can be assumed to be linear. Eq. (3.37) can be
rewritten in the following incremental form:
K  . a   P  0 …(3.38)
36
where K  is the stiffness matrix at the beginning of the increment,  a and
 P  represent increments of displacements and loads respectively.
The size of error resulting from this piecewise linearization process
mainly depends on the size of the load increment. The accumulation of this error
over several increments results in an incorrect solution. The incremental process
is repeated until the failure load has been reached. Fig. (3.6.a) shows how such a
solution can deviate from the true one.
3.6.2 Iterative Technique

Two different techniques are usually used with the purely iterative technique.
These are the direct iterative technique and Newton-Raphson technique.
The process of calculation by these techniques is schematically shown in
Fig. (3.6b and c).
I- Direct Iterative Technique
In this technique full load is applied only once at the first iteration and a new
stiffness matrix has to be formed for each iteration, i. e., a full system of Eq.
(3.37) has to be solved in each iteration to evaluated the displacements vector
a. Initially, some values of a  a 0  is assumed, then an improved
approximation is obtained as:
a1  K  1P  …(3.39)

Repetition of the process can be written as:
an   K n 1  1P …(3.40)

and this is terminated when the “error”, i. e., e  an  an  1 becomes sufficiently
small or within an acceptable value, Fig. (3.6.b).
II- Newton-Raphson Technique
In this technique also the total load is applied only once at the first iteration.
After that, the portion of the total loading that is not balanced is calculated and
37
used in the next step to compute an additional increment of displacements

 a i  using the relationship:
 ai   KT ai   1r ai  …(3.41)
where K T a i   is the tangent stiffness matrix at iteration i, and ri  is the
unbalanced nodal forces vector at iteration i given by:
ri   P   f i   p    B T  i dV …(3.42)

vol .
where P  is the external applied load vector,  fi  is the internal nodal load
vector at iteration i, and  i  are the current stresses at iteration i. The
procedure is repeated until the increments of displacements or the unbalanced
forces become sufficiently close to a preselected criterion, Fig. (3.6.c).
The iterative type of technique is not suitable for tracing the nonlinear
equilibrium path because it fails to produce information about intermediate stage
of loading.
f f
3
2
a a
(a) (b)
f
1 2 3 …. n
a
(c)
Fig. (3.6) Basic techniques for solving nonlinear equations.
a- Incremental technique. b- Direct iteration.
c- Newton-Raphson method (conventional and modified).
38
3.6.3 Combined Incremental-Iterative Technique

This technique utilizes a combination of the incremental and iterative
techniques, Fig. (3.7). The load is applied incrementally, and within each
increment of loading iterative cycles are performed in order to obtain
a converged solution for that increment.
The incremental-iterative technique has been widely used in the nonlinear
analysis of reinforced and prestressed concrete structures. This is due to its
capability to trace the response history of the structure and provide information
about cracking of concrete that occurs at different load levels, yield of
reinforcement, crushing of concrete and the load-deflection curve. For these
reasons it is adopted in the present study. Two different methods are usually
used in connection with the incremental-iterative technique. These are the
standard and modified Newton-Raphson methods.
I- Standard Newton-Raphson Method
In the standard, or full, Newton-Raphson method, Fig. (3.7.a) the external load
is incrementally applied. A series of iterations are performed within each
increment of load in order to bring the amount of unbalanced forces to an
acceptable limit. The tangential stiffness matrix is updated and a new system of
equations is solved for each iteration. This is an expensive computational
procedure particularly if the size of the load increments is small. To overcome
this, the algorithm can be modified by only updating the stiffness matrix
occasionally[68].
II- Modified Newton-Raphson Method
The most commonly used modified forms of the standard Newton-Raphson
method are those in which the stiffness matrix is updated only once for each
load increment. Fig. (3.7.b) shows that the stiffness matrix is updated at the
beginning of the first iteration of each load increment, (the KT1 method).
Fig. (3.7.c) shows that the stiffness matrix is updated at the beginning of the
39
second iteration, (the KT2 method), so that the nonlinear effects are more
accurately represented in the stiffness matrix[68].
These methods are generally more powerful than the standard
(or conventional) Newton-Raphson method since they involve fewer stiffness
matrix reformulation and inversion. However, convergence is slower and a large
number of iterations are required to achieve a converged solution. This is
particularly true for an increment of loading at which a sudden softening may
occur due to cracking, yielding or substantial nonlinear behavior of concrete in
compression. In order to make the modified methods more effective at loading
stages in which slow convergence occurs, stiffness matrix may be updated more
than once within such an increment. The developed program incorporates a
modified Newton-Raphson method (KT2a), in which the stiffness matrix is
updated at the 2nd, 12th, 22th, …etc. iterations of each increment of loading.
(a) Standard N.R. method (b) Modified N.R. method
(c) Modified N.R. method
Fig. (3.7) Standard and modified Newton-Raphson methods.
40
3.7 Convergence Criteria

In order to decide when the iterative process should be terminated, a
convergence criterion is required. The convergence criteria generally can be
classified into the following:
a- Displacements convergence criterion.
b- Forces convergence criterion
c- Work done convergence criterion.
In the present work, convergence was achieved by the force convergence
criterion. This type of criterion is favored because it monitors directly the
redistribution of the out-of-balances forces. The violation of the equilibrium is
estimated by the magnitude of the residual unbalanced nodal forces. The
convergence can be considered to be achieved when:
r a T

.r a 
 toler …(3.43)
P P T
where toler is the specified convergence tolerance.
41
Chapter Four Modeling of Material Properties
Chapter Four
MODELING OF MATERIAL PROPERTIES
4.1 Introduction
The most powerful and general solution tool for the nonlinear analysis of
reinforced concrete and prestressed structures is undoubtedly the finite element
method. It is a valuable solution tool, which can be used to predict the load-
deflection response, failure load, stresses and strains at different stages of
loading and crack patterns. Thus, by using the finite element modeling of
reinforced and prestressed concrete members, effects of different parameters on
the structural behavior can be studied. Accurate modeling for reinforced
concrete material is often one of the major factors in enhancing the analysis
results. This is because reinforced concrete is a complicated composite material
and its behavior cannot be accurately predicted without considering the
constitutive relations of its constituents (concrete and reinforcement).
4.2 Modeling of Concrete

Concrete is in effect a heterogeneous nonlinear material and has completely
different properties in tension as compared to compression. It is well known that
nonlinear response of concrete is caused by two major material properties,
cracking in tension and plasticity in compression due to interface bond failure
between aggregate and mortar and cracking of mortar itself. Time-dependent
effects such as creep, shrinkage and temperature change also contribute to the
nonlinear response. These time-dependent effects are not considered in this
study.
42
4.2.1 Stress-Strain Models

Several approaches for defining the complicated stress-strain behavior of
concrete under various stress states can be divided generally into groups:
I- Elasticity theory.
II- Plasticity theory.
I- Elastic Models
Elastic behavior of a material is assured when after unloading, no residual
strains are retained, and the material returns to its original shape. The stress-
strain relations of elastic models may be either linear or nonlinear elastic
behavior.
I-1 Linear Elastic Models
In the linear elastic models, stress-strain relations for uncracked and cracked
concrete are based on the theory of linear elasticity. The linear elastic models
can be quite accurate for structures where the tensile cracking is the major cause
of the nonlinear behavior. However, these models fail to identify inelastic
deformation. This phenomenon becomes apparent when the material experience
unloading. This can be improved by introducing the nonlinear elasticity models.
I-2 Nonlinear Elastic Models
Nonlinear elasticity models are based on two different approaches. These are the
total “hyperelastic” and incremental “hypoelastic” stress-strain formulation. In
the total stress-strain models, the current state of stress is assumed to be
uniquely expressed as a function of the curved state of strain. This type of
formulation is reversible and path-independent which is not true for concrete in
general. When concrete exhibits unloading these models fail to predict the
inelastic deformation. The incremental stress-strain models are used to describe
the behavior of materials in which the state of stress depends on the current state
of strain and on the stress path followed to reach this state. This type of
formulation is incrementally reversible and path dependent and therefore, it
43
gives a good representation of concrete behavior as compared with the total

stress formulation.
II- Plasticity Models
The plasticity-based models have been used extensively in recent years to
describe the nonlinear behavior of concrete in compression. It is known that
under triaxial compression, concrete can flow as a ductile material on the yield
surface before reaching its crushing strain. To account for this plastic flow
ability of concrete before crushing, various plasticity models have been
introduced. In these models, concrete can be modeled as an elastic-perfect
plastic material or as a strain hardening material. The prediction of the overall
behavior using plasticity-based models gives good agreement with the
experimental results. This type of modeling is adopted in the present work.
4.2.2 Behavior of Concrete in Compression

The numerical models used to describe the behavior of concrete in any finite
element analysis of reinforced and prestressed concrete structures must be able
to trace the overall behavior of the structure with sufficient accuracy. At the
same time they must be kept as simple as possible.
In the present study, the behavior of concrete in compression is simulated
with an elastic-plastic work hardening model followed by perfect plastic
response, which is terminated at the onset of crushing, Fig. (4.1). The plasticity
model can be illustrated in terms of the following constructions:
-The yield criterion.
-The hardening rule.
-The flow rule.
-The crushing condition.
44
C p  1.0
f c
E
C p f c 1
o c
 cu
 o
Fig. (4.1) Uniaxial stress-strain curve for concrete [9].
4.2.2.1 The Yield Criterion

The strength of concrete under multiaxial stresses is a function of the state of
stresses and cannot be predicted by limitations of simple tensile, compressive
and shearing stresses independently of each other. Therefore, strength of
concrete can be properly determined only by considering the interaction of the
various components of the state of stresses. Under triaxial state of stresses, the
yielding criterion for concrete is generally assumed to be dependent of three
stress invariants. However, a yield criterion dependent only on two stress
invariants has been proved to be adequate for most practical situations[14][15][26].
In the present work, a yield criterion based on two stress invariants has
been adopted, and this is given by:
F    f  I 1 , J 2    I 1  3  J 2   o …(4.1)
where  and  are material parameters, I1 is the first stress invariant given by:
I1   x   y   z …(4.2)
and:
1
  
J 2     x2   2y   z2   x  y   y  z   z  x    xy
2
  2yz   zx
2
…(4.3)
3
and  o is the equivalent effective stress at the onset of plastic deformation and
this can be determined from uniaxial compression test as:
45
 o  C p f c …(4.4)
where C P is a plasticity coefficient used to mark the initiation of plastic

deformation ( 0.0  C P  1.0 ) (end of elastic and beginning of plastic state).
The material parameters in this criterion have been obtained by fitting
biaxial test results and using the following states of stresses:
-For the uniaxial compression state of stress:
 x   o and  y   z  0 …(4.5)
-For the equal biaxial compression state of stresses:

 x   y   o and  z  0 …(4.6)
If the results obtained by Kupfer et al[35] for a failure envelope of plain

concrete are employed for initial yielding, the value of the constant  becomes
equal to 1.16 for equal principal compressive stresses (at crushing failure).
Manipulation of Eqs. (4.1) into (4.6) gives:
  0.355 o , and   1.35468
Letting C   /( 2 o )  0.17734 …(4.7)
Therefore Eq. (4.1) can be rewritten as:
F    2C o I 1  3  J 2   o …(4.8)
Eq. (4.8) can be simplified to:
F    C .I 1  C .I 1  2  3  J 2 o …(4.9)
4.2.2.2 The Hardening Rule

The hardening rule defines the motion of the subsequent loading surfaces during
plastic deformation. A relationship between the accumulated plastic strain and
the effective stress is required to control the position of the current loading
surface. A number of hardening rules has been proposed to describe the growth
of subsequent loading surface for work-hardening materials[14][15].
46
In the current study an isotropic hardening rule is adopted. The rule

assumes that the yield surface expands uniformly without distortion as plastic
flow occurs[14][15]. Therefore, from Eq. (4.9), the subsequent loading functions
may be expressed as:
F    C .I 1  C .I 1 2  3 J 2  …(4.10)
where  represents the stress level at which further plastic deformation will
occur and it is termed as the effective stress or the equivalent uniaxial stress at
that level.
In the present model, a parabolic stress-strain curve is used for the
equivalent uniaxial stress-strain relationship beyond the limit of elasticity,
C p f c  . This parabolic curve represents the work hardening stage of behavior up
to when the peak compressive stress is reached, where then a perfectly plastic
response is assumed to occur. Returning to Fig. (4.1), the equivalent uniaxial
stress-strain curve is shown at the various stages of behavior.
The equivalent stress-strain relationships at various stages are:
1- Elastic stage:
  E . c for   C p . f c …(4.11)
where,  c is the effective total strain.

2- Work-hardening stage:
2
 C p . f c  E  C p . f c 
  C p . f c  E  c    c   for C p . f c    f c
 E  2 
0  E 
…(4.12)
3- Perfect plastic stage:
f c
  f c for  c   0 or  c  ( 2  C p ) …(4.13)
E
where,  0 is the total strain corresponding to the parabolic part of the curve
given by:
47
f c f
 o   o  C p . or  0  2( 1  C p ) c …(4.14)
E E
Based on experimental data, the value of the plasticity coefficient ( C p )
was set to equal to 0.3 for normal strength concrete and 0.5 for high strength
concrete.
In order to derive the relationship between the effective stress and the
effective plastic strain, the total strain  c , is decomposed into its elastic and
plastic components as:
c  e   p …(4.15)
where,  e is the elastic strain component given by:

e  …(4.16)
E
By substituting Eqs. (4.15) and (4.16) into Eq. (4.12), the effective stress-plastic
strain relation can be expressed as:
  C p f c  E p  2 E 2  0  p …(4.17)
The slope of the effective stress-plastic strain curve, the hardening

coefficient, is used in the formulation of the elastic-plastic incremental stress-
strain relationship. Therefore, by differentiation of Eq. (4.17) with respect to the
effective plastic strain, the hardening coefficient, H  , can be expressed as:
d   0 
H   E  1.0  …(4.18)
d p  2 p 
4.2.2.3 The Flow Rule

The necessary connection between the loading function, f , and the stress-strain
relation for a work-hardening material can be made by means of a flow rule.
When the current yield surface, f , is reached, the material is in a state of plastic
flow upon further loading. In the plasticity theory a flow rule must be defined so
48
that the plastic strain increment can be determined for a given stress increment.
The associated flow rule has been widely used for concrete models mainly
because of its simple formulation. This approach is adopted in the current work.
The plastic strain increment can be expressed as[15][50]:
f (  )
d  p   d …(4.19)

where, d  0 is a scalar hardening parameter which can vary throughout the
f (  )
straining process. The gradient of the yield loading surface , defines the

direction of the plastic-strain increment vector d p while the length is
determined by the hardening or loading parameter d , i.e., the plastic flow

develops along the normal to the loading surface. The yield function derivatives
with respect to the stress components define the flow vector { a } [15][50] as:
T
 
a   f , f , f , f , f , f  …(4.20)
  x  y  z  xy  yz  zx 
These derivations can be evaluated from Eq. (4.9) with the use of Eqs. (4.2) and
(4.3) as:
f
 x

 C  2 ( C 2   ) x  ( 2 C 2   ) (  y   z ) / Q
f
 y

 C  2 ( C 2   ) y  ( 2 C 2   ) (  x   z ) / Q
f
 z

 C  2 ( C 2   ) z  ( 2 C 2   ) (  x   y ) / Q …(4.21)
f   xy
6
 xy Q
f   yz
6
 yz Q
f   zx
6
 zx Q
49
where:
Q  2 [( C 2   )(  x2   y2   z2 )  ( 2 C 2   )(  x  y   y  z   z  x ) 
3  (  xy
2
  yz
2
  zx
2
)] 0.5 …(4.22)
C and  are the material constants previously defined.
4.2.2.4 Incremental Stress-Strain Relationship

The total (elastic plastic) incremental strain vector is given by:
d    d  e   d  p  …(4.23)
By making use of Eq. (4.19), the complete incremental relationship[35] between

stress and strain for elasto-plastic deformation can be written as:
df (  )
d    d  e   d …(4.24)
d( )
The elastic strain increment is related to the stress increment by the elastic
constitutive relationship, which is given by:
d    D  d  e  …(4.25)
where D  , is the elastic constitutive matrix given by:
1     0 0 0 
  1   0 0 0 
 
   1  0 0 0 
 1  2 
D   E
 0 0 0 0 0  …(4.26)
( 1   ) ( 1  2 )  2 
1  2
 0 0 0 0 0 
 2 
 1  2 
 0 0 0 0 0
2 
Substituting of Eq. (4.25) into Eq. (4.24) yields:
d    D  d  d a
1
…(4.27)
50
Pre-multiplying both sides of Eq. (4.27) by, a D , and eliminating, ad  ,
T
the plastic multiplier, d , can be expressed as:

 aT . D  
d    . d   …(4.28)
 
 H  a . D .a
 T
By substituting of Eq. (4.28) into Eq. (4.24) and multiplying both sides by D  ,
the complete elastic-plastic incremental stress-strain relation becomes:
 
d    De  p d   …(4.29)
where:
De  p 
  D aaT D  
 D    …(4.30)
 H  a D a
 T
and where the second term in the bracket represents the stiffness degradation
due to plastic deformation.
4.2.2.5 The Crushing Condition

In the adopted model, the isotropic hardening of the subsequent loading surface
is terminated when the effective stress reaches the peak compressive stress.
Beyond that, a perfectly plastic response is assumed to occur. The perfectly
plastic flow continues until the ultimate deformation capacity of concrete is
reached and the material eventually exhibits a crushing failure. The stresses at
crushing point drop abruptly to zero.
The crushing criterion is obtained by simply converting the yield criterion
in Eq. (4.9), which is written in terms of stresses, directly into strains[15], thus:
C . I 1  ( C . I 1 ) 2  3  . J 2   cu …(4.31)
where I 1 and J 2 are the strain invariants,  cu is the ultimate crushing strain of
concrete, extrapolated from uniaxial test.
I 1   x   y   z …(4.32)
51
 1
   
J 2     x2   y2   z2   x  y   y  z   z  x  (  xy ) 2
 3
 (  yz ) 2  (  zx ) 2 …(4.33)
4.2.3 Behavior of Concrete in Tension

In the present study, the behavior of concrete in tension is modeled as a linear
elastic brittle material and the maximum tensile stress criterion is employed. If
the maximum principal stress exceeds a limiting tensile strength, f t , crack is
formed in a plane perpendicular to the offending stress. Thereafter, the concrete
behavior is no longer isotropic; it becomes an orthotropic material. A smeared
cracks model with fixed orthogonal crack is adopted to represent the fracture of
concrete. The model is described in terms of the cracking criterion, post-
cracking formulation and shear retention model.
4.2.3.1 Model of Crack Representation

In the finite element analysis of concrete members, two main approaches are
adopted for cracks representation:
- Discrete cracking model.
- Smeared cracking model.
In the discrete cracking model, crack can appear only at the element
boundaries[49]. This is normally done by disconnecting displacements at nodal
points of adjoining elements in the finite element mesh, Fig. (4.2). One obvious
difficulty in such approach is that the location and orientation of the crack are
not known in advance. Thus, geometrical restrictions imposed by the preselected
finite element mesh can be hardly avoided. This leads to some extent to the
adjustment of the mesh by increasing the number of element nodes. Such
techniques are complex and time-consuming.
52
a) One-dimensional cracking b) Two-dimensional cracking
Fig. (4.2) Idealization of discrete cracking model[15].
In the smeared cracking model, the concrete is assumed to remain a

continuum, that is, the cracks are smeared out in a continuous fashion[52] rather
then representing a single discrete crack. The present approach has the effect of
representing an infinite number of parallel fissures across the cracked concrete
element, Fig. (4.3). It is assumed that the concrete becomes orthotropic after the
occurrence of first cracking. Such formulations easily allow for gradual release
of stress due to aggregate interlocking which can be accounted for by retaining a
positive shear modulus. For most structural engineering applications the
smeared cracking model is usually used.
Y
Y'
X'
Fig. (4.3) Idealization of smeared cracking model[15].
53
4.2.3.2 The Cracking Criterion

In the present model, the initiation of cracking is controlled by the maximum
tensile stress criterion. For a previously uncracked sampling tensile point, the
principal stresses and their directions are calculated. If the major principal
stress,  1 , exceeds a limiting tensile stress value, a crack is assumed to form.
The limiting tensile stress required to define the onset of cracking can be
calculated for states of triaxial tensile stress and for combinations of tensile and
compressive principal stresses as follow[9]:
I) For the triaxial tension zone (  1   2   3  0 ):
 i   cri  f t where i=1,2,3 …(4.34)
II) For the case of tension-tension-compression zone (  1   2  0 , 3  0 ):
0.75  3
 i   cri  f t [ 1.0  ] where i=1,2 …(4.35)
f c
III) For the case of tension-compression-compression zone
(  1  0 , 2   3  0: )
0.75  2 0.75  3
 i   cri  f t [ 1.0  ] .[ 1.0  ] …(4.36)
f c f c
where,  cr is the cracking stress and f t , f c are the tensile and compressive
concrete strengths (positive value).
The last two Eqs. (4.35) and (4.36) show that the compression in one
direction favors cracking in the other (i.e. reduces the tensile capacity of the
material). After cracking, the elastic modulus in the direction of maximum
tensile stress,  1 , is reduced. Also, because of the lack of interaction between
orthogonal planes caused by cracking, Poisson’s ratio is set to zero. Therefore,
the incremental stress-strain relationship in the local material axes may be
expressed as:
54
  1   E1 0 0 0 0 0    1 
    0
 2  E /( 1   2 )  E /( 1   2 ) 0 0 0     2 
   3   0  E /( 1   2 ) E /( 1   2 ) 0 0 0     3 …(4.37)
 
   
   12   0 0 0 1 0 0     12 
   23   0 0 0 0 G 0     23 
    
   31   0 0 0 0 0  1G     31 
Eq. (4.37) may be written in the form:

  Dcr . …(4.38)
Then, the stress increment in the global axes (x, y, z) may be obtained by the
coordinate transformation matrix such that:
   T T .Dcr .T .  …(4.39)

where T , is the transformation matrix expressed in of direction cosine as[20]:
 l12 m12 n12 l1 m1 m1 n1 n1 l1 

 2 
 l2 m22 n22 l2 m2 m2 n2 n2 l2 
 l32 m32 n32 n3 l3 
T   l3 m3 m3 n3
 …(4.40)
 2l1l2 2m1 m2 2n1 n2 ( l1 m2  l2 m1 ) m1 n2  m2 n1 n1 l2  n2 l1 
2l l 2m m 2n n ( l m  l m ) m n  m n n l  n l 
 23 2 3 2 3 2 3 3 2 2 3 3 2 2 3 3

2l3 l1 2m3 m1 2n3 n1 ( l3 m1  l1 m3 ) m3 n1  m1 n3 n3 l1  n1 l3 
where l i , m i and n i (I=1,2,3) represent the direction cosines the local
coordinates x', y' and z' with respect to the global coordinates x , y and
z direction, respectively.
For the tension-tension-compression and the triaxial tension state of
stress, the cracking criterion may be validated by the major principal stress,  1 ,
and the second principal stress,  2 , simultaneously. Thus, two sets of
orthogonal failure planes develop. These planes are perpendicular to the
principal axes 1 and 2, respectively. In this case, Poisson’s ratio is set to zero in
all directions and the constitutive matrix in the local material axes become a
diagonal matrix:
55
E1 0 0 0 0 0 
0 E2 0 0 0 0 

0 0 E 0 0 0 
Dcr     …(4.41)
0 0 0 1 G 0 0 
0 0 0 0 2 G 0 
 
0 0 0 0 0  G
In the fixed crack model, the rotation of the principal stresses after
formation of the first crack is ignored. However, if the second crack initiates at a
subsequent stage of loading, rotation of the principal stresses within the planes
perpendicular to the direction of  1 might be taken into account. Magnitudes
and direction of the new second and minor principal stresses,  2 and  3 , may
be calculated from the original principal stresses  2 and  3 , and the shear
stress  23 accumulated during loading stage beyond the formation of the first
crack. Thus by using Mohr circle[15]:
2 3 (  2   3 )2
 i     23
2
i=2,3 …(4.42)
2 2
For tension-tension-compression state, the onset of the second crack is
expected in terms of the new principal stresses as:
 0.75  3 
 2   cr  f t 1.0   …(4.43)
 f c 
Failure planes corresponding to the second crack are assumed to be

perpendicular to the direction of the second principal stress  2 , Fig. (4.4). The
angle between the original and the new in-plane sets of principal directions,  p ,
can be calculated as:

1 2  23
p  tan  1 ( ) …(4.44)
2  2  3
After formation of the second crack, the local in-plane constitutive matrix
with respect to the new principal directions ( 2  ,3  ) is expected as:
56
E2 0 0 
Dcr    0 E 0 

…(4.45)
 0 0  2 G 
This matrix may be transformed to the original directions (2,3) using the
in-plane coordinate transformation matrix. Therefore, the in-plane incremental
stress-strain relationship in the original principal directions can be written as:
  2  E2 0 0    2 
 
  3   T p   T  0
 E  
 

0  T p   3  …(4.46)
    0  2 G    
 23  0  23 
 
where, T p is the in-plane transformation matrix given by:
 C2 S2 CS 
T 
p

  S2 C2 CS 

…(4.47)
  2CS 2CS C 2  S 2 

where C  cos  p , and S  sin p .
In the present model, a maximum number of three sets of cracks are

allowed to form at each sampling point, Fig. (4.4c). A third crack may occur for
triaxial tension states, when the minor principal stress,  3 (or  3 ) exceeds the
uniaxial tensile strength, f t .
p 2 ´ p 2
´ 2
p 2
2
´
2 2 2
2
2
2
1 ´
3 1
´ 1 1 ´
3
p
3
3 3
p p
3
3
3
3 3
3
3 c)
b) Doublecracks
b) Double cracks c) Triple cracks
Triple cracks
a)a)Single
Singlecracks
cracks
Fig. (4.4) Failure cracking for triaxially loaded concrete[9].
57
4.2.3.3 Tension Stiffening Model

Cracking of concrete is further complicated by the presence of reinforcement,
and the bond stress developed between the steel and the surrounding concrete.
The ability of concrete between adjacent primary cracks to carry tension is
known as the tension-stiffening phenomenon.
Tension stiffening effect is found to be quite significant under surface
load condition by increasing the overall stiffness of the system in the post-
cracking range.
In the finite element modeling of reinforced concrete two approaches have
been suggested to account for this effects[51]. The first approach is characterized
by increasing the stiffness of steel bars. The second approach is characterized by
a gradual decrease of the tensile stress in the cracked concrete over a specified
strain range. A lot of research[5][6][9][12] used the second approach to account for
the tension-stiffening effect in conventionally reinforced concrete members,
Fig. (4.5).
 cr
 2 . cr (  1   a  cr )
 a   2 . cr
( 1  1 )
E
1 ε
 cr n  1 . cr
Fig. (4.5) Post-cracking for concrete in tension[9].
In the present work, the second approach is adopted to represent the

tension-stiffening effect. The formulation of this approach is described as
follows:
58
I) For  cr   n   1  cr
n
( 1  )
 cr
   2  cr …(4.48)
 1  1.0 
II) For  n   1  cr
 n  0.0 …(4.49)
where  n and  n are the stress and strain normal to the crack plane.
 cr is the cracking strain associated with the cracking stress  cr .
 1 and  2 are the tension-stiffening parameters, (  1 is the rate of post-cracking

stress decay as the strain increases, and  2 is the sudden loss in stress at instant
of cracking ).
4.2.3.4 Shear Retention Model

The shear stiffness at a cracked sampling point becomes progressively smaller
as the crack widens. So the shear modulus of elasticity is reduced to  G . Before
cracking the factor  is set equal to 1.0. When the cracks propagate, the shear
reduction factor (  ) is assumed to descrease linearly, Fig. (4.6)[51].
When the crack is sufficiently opened, a constant value is assigned to  ,
to account for the dowel action. The following relations are used to account for
the shear retention effect.
I)   1.0 for  n   cr (no cracking) …(4.50)
( 2   3 ) 
II)   (  1  n )   3 for  cr   n   1  cr …(4.51)
(  1  1.0 )  cr
III)    3 for  n    cr (with cracking) …(4.52)
where,  1 ,  2 and  3 are the shear retention parameters.

 1 , Represents the rate of decay of shear stiffness as the crack widens.
 2 , Represents the sudden loss in the shear stiffness at the onset of cracking.
59
 3 , Represents the residual shear stiffness due to the dowel action.


1.0 1 2   
   1  n    3
 1  1.0   cr 
2
3
ε
 cr  1  cr
Fig. (4.6) Shear retention model for concrete[9].
4.3 Modeling of Reinforcement

Modeling of reinforcing and prestressing steel in connection with the finite
element analysis of reinforced and prestressed concrete members is much
simpler than the modeling of concrete. The stress-strain behavior of reinforcing
and prestressing steel is identical in tension and compression. The reinforced
and prestressed steel bars are long and relatively slender, and therefore, they can
be assumed to transmit axial force only.
In the current work, an elastic linear work hardening model is adopted to
simulate the uniaxial stress-strain behavior of reinforcing and prestressing steel
bars, Fig. (4.7).
It is worth to noting that the prestressing force is included in this study
and will be extensively discussed in Chapter Five.
σ
y
Es Es s
 su
Fig. (4.7) Stress-strain relationship of reinforcing and prestressing steel bars.
60
Chapter Five Formulation of Prestressing
Chapter Five
FORMULATION OF PRESTRESSING
5.1 Introduction
Prestressed concrete has become one of the major structural building materials
and it is widely used all over the world. In practice, concrete box-girder bridges
are most often prestressed and therefore the method of their analysis would be
incomplete without taking this important factor properly into account. Since this
study is devoted for the analysis of prestressed concrete structures, this chapter
considers in detail the derivation of an efficient prestress model capable of being
incorporated in a finite element solution[28].
There are different methods to take into analysis the prestressing tendon.
The simplest one is the method used by Lin and applied by Loo and Cusens[39].
It is also applied by Abdulla and Abdul-Razzak[1]. In there method the tendon is
assumed to be frictionless and acts in the neutral plane of the strip. The tendon
may be replaced by two types of in-plane forces: two end anchorage forces and a
uniform pressure along the span of the bridge structure.
In the second method used by Ghalib[28], the prestressing tendons are
considered as an integral part of the structure. A very sophisticated prestressing
tendon model, capable of modeling effectively the interaction between concrete
and the housed tendon throughout various prestressing processes is developed.
For bonded tendons, a generally curved tendon element embedded in a higher-
order isoparametric flat shell element is developed. For unbonded tendons, since
interaction between the prestressing tendon and the structure occurs mainly at
anchorages, the average extension method is considered very efficient in
modeling the stiffness contribution of such element.
61
Another method was used by Kang et al[33], they analyzed the prestressing
tendon by using approximately a series of prestressing steel segments each of
which is straight, spans a frame element, and has a given profile, initial
tensioning force, and a constant cross-sectional area along its length. The stress
relaxation is computed utilizing the formula developed by Magura et al[42].
In the present study, the method used by Zyle and Scordelis[70] and
applied by Jirousek and Bouberguig[32] is considered. In this approach, the
equivalent load method is used to compute the force applied by the tendon upon
the structure where the prestressing tendon takes the form of a particular loading
case and as such it may be taken into consideration in the form of appropriate
local loads at the level of each element. Practically, this means that the effect of
prestressing manifests itself in the evaluation of the vectors of primary nodal
forces only[32]. The development of the necessary relations for practical
computation is briefly outlined hereafter.
Same approach was implemented by Roca and Mari[54][55] in 1993 with
including bonded and unbonded post-tensioned tendon with general formulation
for material and geometric nonlinearity.
5.2 Short-Term Prestress Losses

5.2.1 Frictional Losses
For post-tensioned members, the tendons are usually anchored at one end and
stretched by jacks at the other end. As the steel slides through the duct, frictional
resistance is developed, with the result that the tension at the anchored end is
less than the tension at the jack. The total friction loss is the sum of the wobble
friction due to unintentional misalignment, and the curvature friction due to the
intentional curvature of the tendon. The following well-known equation is used
to calculate the prestress loss at any point in the tendon at distance x from the
anchorage end[38]:
62
P  Po . e  (    x ) …(5.1)
where, p o : force in jack end (x=0).
p : force in tendon at distance x.
 : curvature friction coefficient.
 : angle change in prestressing tendon over distance x.
 : wobble friction coefficient.
For parabolic profiles of constant curvature, Eq. (5.1) can be written as
follows:
p  p o . e  qx …(5.2)
where q a
q : constant profile curvature (   ax ).
The values of  and  are listed in Appendix A[49].
5.2.2 Anchoring Losses

Prestress loss due to slip-in of the tendon when the prestress jack end is released
is present in post-tensioned as well as pretensioned construction. Although it has
a negligible effect on long tendons, it may become very significant for short
tendons. However, anchor slip loss is mostly confined to a region close to the
jacking anchorage. Distribution along the tendon is prevented by reverse friction
as the tendon slips inward, and the steel stress throughout much of the tendon
length may be unaffected by anchorage slip, Fig. (5.1).
Haung[29] proposed a method for solving this problem. The force p 2 and
length l a over which anchor slip takes place are unknown. Considering the fact
that the area under the curve represents the elongation of the tendon, the
following equation containing the unknown length l a can be obtained:
2  s As E s
la  …(5.3)
p o ( 1  e  2 qla )
63
where,  s : prestressing tendon anchor slip.

As : prestressing tendon area.
E s : prestressing tendon modulus of elasticity.
A detailed formulation of the above expression can be found in Appendix
(B).
Using Eq. (5.3), length l a can be evaluated by using ordinary nonlinear
solvers such as the iterative Newton-Raphson algorithm.
As shown in Fig. (5.1) the force in the tendon is then calculated as
follows:
p  p o . e  q ( 2 la  x ) x  la …(5.4)
p  p o . e  qx x  la …(5.5)
Po
(a) Prestressing tendon layout

Prestressing force (P)
Po
P2
P1
La
Distance (x)
(b) Prestress force variation along the tendon both
before and after anchor release
Fig. (5.1) Prestress forces losses due to anchor slip[28].
64
5.3 Algorithm for Use with the Finite Element Analysis

5.3.1 Basic Assumptions
Few assumptions have to be made in order to obtain a workable mathematical
model[32].
1. The weakening of the concrete section by the holes provided for the
prestressing tendons may be neglected.
2. The tension in the prestressing tendons is not affected by the elastic
deformation of the structure.
5.3.2 Geometry of the Tendon and Variation of the Prestressing Forces

A particular brick element is considered where it is transversed by a prestressing
tendon as shown in Fig. (5.2). The geometric definition of the tendon segment
corresponding to a particular brick element will be supposed of the following
form[32]:
 xc   x ci 
  m
 
X c   yc    M i (  ) y ci  …(5.6)
z  i 1 z 
 c  ci 
In this equation, X c stands for the vector of global Cartesian coordinates
associated with a general point C situated on the axis of the tendon,  is a non-
dimensional parameter varying from –1 to +1 between the points of the segment,
and x ci , y ci , z ci (i=1,2…m) represent given Cartesian coordinates of m
particular points C 1 , C 2 ,...C m distributed as uniformly as possible on the axis of
the tendon. The base function M i (  ) associated with a particular node C i , by
taking a value of unity in C i and zero at all other nodes C k  i , is represented by
the Lagrange polynomial[32]:
(    1 )...(    i  1 )(    i  1 )...(    m )
Mi (  )  …(5.7)
(  i   1 )...(  i   i  1 )(  i   i  1 )...(  i   m )
65
where:
2 i1
 1  1 ,  2  1  , ... i  1  2 , ... m  1 …(5.8)
m1 m1
In practice, the quadratic ( m  3 ) or cubic ( m  4 ) approximations of

the tendon geometry will, in general, give satisfactory results[32].
Fig. (5.2) Typical segment of prestressing tendon traversing a brick element[32].
The variation of tension T  T (  ) in the tendon will most adequately be

defined in the form consistent with the deflection of the tendon geometry,
namely[32]:
m
T  M i (  ).Ti …(5.9)
i 1
where Ti (i=1,2,…m) are given magnitudes of the tension at nodal points. They
will be specified on the basis of the customary prediction of loses of the
prestressing, such as those mentioned in section (5.2).
66
5.3.3 Element Local Loads Due to Prestressing

The local action of a prestressing tendon on a particular brick element may be
represented by a distributed line load acting on the element along the
corresponding segment C 1 C m of the tendon axis and, if revelant, by a
concentrated anchoring force applied to the element at the point ( C 1 C m )
where the extremity of the tendon has been anchored in the concrete, as shown
in Fig. (5.2).
The distributed line load has two components, the tangential component:
dT
Pt   …(5.10)
ds
and the normal component:

T
Pn  …(5.11)
R
acting on the element (Fig. (5.2)) in the direction of the main normal n of the
element axis (R is the radius of the principal curvature). Observing that the
global Cartesian components of the unit vectors t and n may be defined:
Vt dx
t with Vt  …(5.12)
Vt d
and:
Vn 1 d 2 xc a dx c
n with V n  (  ) …(5.13)
Vn Vt
2
d 2 Vt
2
d 
d 2 x c dx c
where: a  .
d 2 d
It is possible to combine the global components of the tangential and
normal loads into a unique global Cartesian local vector:
67
 Px 
 
p   Py   Pt t  Pn n …(5.14)
P 
 z
where:
dT 1 dT T
Pt    and Pn   V n T …(5.15)
ds V t d R
If one of the points C 1 or C m of the tendon segment coincides with the

end of the tendon, the end anchoring force P1 T 1 or Pm  Tm (Fig. (5.3)) must
be applied to the element as a concentrated local load. This load, tangential to
the tendon axis, will most conveniently be specified by giving the vector its
Cartesian global components:
 Px 1 
 
P1   Py 1   ( T .t )  1 …(5.16)
P 
 z1 
 Pxm 
 
Pm   Pym   ( T .t )  1 …(5.17)
P 
 zm 
5.3.4 Vector of Primary Nodal Forces of the Element

The displacement definition of the brick element can be written as:
ue  N  .ae …(5.18)
and it is assumed that the element is transversed by only one prestressing
tendon. Then, using the principle of virtual work, it can be shown that the local
loads in Eqs. (5.14), (5.16) and (5.17) are balanced by the primary nodal forces
of the element loads to[49]:
68
Fs     N T (  c , c , c ) P (  ) ds  ( N T (  c 1 , c 1 , c 1 ) P1 or
l
N (  cm , cm , cm ) Pm )
T
Fs     N T (  c , c , c ) P (  ) Vt (  ) d 
l …(5.19)
( N T (  c 1 , c 1 , c 1 ) P1 or N T (  cm , cm , cm ) Pm )
The large brackets, containing the end anchoring forces, have been used
to indicate that the use of these terms is limited to cases where the point C 1 or
C m coincides with the end of the tendon. If more than one tendon transverse the
element considered, the effect of each one is computed separately, using Eq.
(5.19), the results being summed.
The integration in Eq. (5.19) is carried out numerically by using the
Gaussian quadrature rule. A three-point Gaussian rule is adopted in this study.
Since Gaussian points have to be distributed along the curved tendon
length, one is confronted with the problem of determining the particular values
 c , c and  c of the curvilinear coordinate  , and  at these points which have
given values of the parameter  . Thus an inverse mapping procedure from the
global coordinates system to natural coordinates system is required in order to
locate the Gaussian points in the natural coordinate system. The problem is non-
linear and leads to the following system of three non-linear equations[32]:
x(  c , c , c )  x c (  ) …(5.20)
where the left-hand term is represented by the coordinate definition Eq. (3.18)
for the brick element, whilst the second term is computed from the tendon
definition Eq. (5.6). Since the relationship (3.18) in Chapter three between the
Cartesian and curvilinear coordinates is one-to-one, the system (5.20) may easily
be numerically solved for any given value of the parameter  by the use of the
Newton-Raphson method.
69
5.4 Algorithm for Use the Lin's Method

Another method is used in the present study. This method is used to analyze the
parabolic tendon in prestressed concrete box-girder bridges. The tendon is
assumed to be frictionless and acts in the neutral plane of the brick element. The
parabolic tendon may be replaced by two types of in-plane force: two end
anchorage forces and a uniform pressure along the span of the bridge structure.
Fig. (5.3) illustrates a curved post-tensioned tendon in a brick element.
(y3,z3)
-P (y2,z2)
(y1,z1)
θ Y
y
Fig. (5.3) Analysis of curved cable[1].
This treatment of curved tendon follows the procedure used by Lin and
applied by Loo and Cusens[39][40].
The span of the cable is assumed to be parabolic and the total change of
the slope is calculated as:
  tan 1 ( B  2C z 3 )  tan 1 ( B  2C z 1 ) …(5.21)
where:
y1  y 2 ) y  y3
( )( 2 )
z1  z2 z2  z3
C …(5.22)
( z1  z3 )
70
y2  y3
B  C ( z2  z3 ) …(5.23)
z2  z3
in which (y1,z1), (y2,z2) and (y3,z3) are the coordinates for any three points in the
parabolic curve spanning a brick element. The uniform pressure along the
tendon duct may be replaced by a uniformly distributed in-plane load P along a
line parallel to the z-axis, as shown in Fig. (5.3), and:
T
P …(5.24)
( z3  z1 )
where T is the prestressing force in the tendon. The load is assumed to act as
line y , where:
y  ( y1  y 2  y 3 ) / 3 …(5.25)
In the present study, the uniform pressure is distributed equally upon the
nodes for the elements spanning with parabolic tendon.
5.5 Computer Program

In the present study, the computer program P3DNFEA (Three-Dimensional
Non-linear Finite Element Analysis), has been adopted. The program was
originally developed by A-Shaarbaf[9]. The main objective of the program is to
analysis prestressed concrete box-girder bridges. Modifications and newly added
subroutines were necessary to incorporate the effect of prestressing forces. The
program is coded in FORTRAN 77 language.
71
Chapter Six Applications and Results
Chapter Six
APPLICATIONS AND RESULTS
6.1 Introduction
The aim of this chapter is to verify the present nonlinear finite element model
developed to investigate the behavior and ultimate load capacity of prestressed
concrete box girders. A computer program has been developed to carry out the
analysis. It is also intended to check the accuracy and validity of the computer
program which deals with box-girder bridges subjected to nonproportional loads
and initial prestressing forces. Several examples are considered in this chapter.
The results of the analysis of these examples are discussed and the load-
deflection behaviour obtained by the finite element analysis is compared with
those obtained from the available experimental investigation.

Bridge
One-seventh scale direct model of a single-cell prestressed box-girder bridge,
simply supported at its ends was tested experimentally at McGill University by
Razaqpur et al[53]. They also analyzed the same bridge by using the nonlinear
finite element technique. The nonlinear program NONLACS (NONlinear
Analysis of Concrete and Steel structure) was utilized for the analysis. They
used a typical shell element.
The geometry and finite element mesh are shown in Figs. (6.1) and
(6.2 a). The applied loading for the bridge was considered to be of a scaled
modeled of the Ontario Highway Bridge Design truck (OHBDC) as shown in
Fig. (6.3). The positioning of the trucks on the bridge model is shown in Fig.
(6.2). The material properties of the concrete, reinforcing and prestressing steel
are listed in Table (6.1).
72
Since the one-cell box-girder was symmetrically loaded with respect to its
longitudinal axis, only one-half of the box-girder is modeled. The one-half
structure has been modeled by 252 brick elements with a total number of 1815
nodal points, as shown in Fig. (6.2 a).
Fig. (6.4) represents the load-midspan deflection curve of the prestressed
box-girder bridge. The obtained results show close agreement in comparison
with the experimental results. The box-girder bridge failed due to crushing in
concrete at failure load level of 56.425 kN. Comparison between the present
study 56.425 kN (4.41 trucks) and experimental value 55.0 kN (4.33 trucks)
shows good agreement with an error just exceeding 2.53%. This is in the same
comparison with the NONLACS study.
Fig. (6.5) shows the deflected shape of the bridge at various load levels.
The level of the load is indicated by the ratio P/Pu, where P is the load at which
the deflection is evaluated and Pu is the ultimate load for the bridge. These
deflections were measured along the longitudinal centerline of the bridge. The
obtained results are in good agreement with the experimental and NONLACS
studies.
The concrete longitudinal normal stresses at various locations in the
bridge are illustrated in Figs. (6.6), (6.7) and (6.8). Fig. (6.6) shows the
longitudinal normal stress in the centerline of the top slab at midspan versus the
level of load. In this figure, the initial longitudinal stress at zero loads in the
present study is less than the initial longitudinal stress in the experimental and
NONLACS results. But after the loading reaching to 40 kN, the longitudinal
stress were near to the experimental results. Generally, the obtained results were
nearly close to the experimental results. While, Fig. (6.7) shows the longitudinal
normal stress on the centerline of the top slab at quarterspan versus the level
load. The initial longitudinal stress at zero loads also gives a value less than the
initial stress in the experimental and NONLACS studies.
73
The variation of concrete longitudinal normal stresses along the centerline

of the top slab is shown in Fig. (6.8). From these figures, the obtained results are
in good agreement with the experimental and NONLACS results along the left
part of the bridge. But these are unfairly close along the right part of the bridge
when comparing with the experimental and NONLACS results.
Figs (6.9) and (6.10) show the variation of longitudinal normal stress at
the cross-sections at midspan and quarterspan for the bridge at load ratio P/Pu
equal to 0.250 and 0.875. These results are listed in tables (6.2) and (6.3).
The deflected shape of the one-half of the one-cell box-girder bridge is
shown in Fig. (6.2 b). The displacements of the whole structure are exaggerated
10 times for clearness.
The stress distribution for the whole structure is shown in Figs. (6.11) to
(6.18).
C
L
282
25
35
a- cross section 56
12 56
12
17 17
b- profile of side view
C
L
Midspan Section End Section
225
129
73
41
29
17
17
70 3 X 130 8 X 145 70
1690
All dimensions in mm
Fig. (6.1) Structural details of the one-cell box-girder bridge
74
a- Finite element idealization with 252 brick elements b- Deflected shape due to prestressing and
for one-half for one-cell box-girder external loading
Fig. (6.2) Finite element idealization for half bridge model of one-cell box-
girder before and after loading
75
Table (6.1) Material properties of the one-cell box-girder bridge.
Concrete
Elastic modulus, Ec (MPa) 22552

Compressive strength, fc` (MPa) 30
Tensile strength, ft (MPa) 1.75*
Poisson’s ratio, υ 0.15*
Compressive strain at fc` 0.0018
Ultimate compressive strain 0.0045
Cracking tensile strain 0.0002*
Reinforcing steel
Elastic modulus, Es (MPa) 200000

Yield stress, fy (MPa) 480.0
Diameter (mm) 1.59
Bar area, (mm2) 2.00
Ultimate strain 0.018
Yield strain 0.0018
Poisson’s ratio 0.3*
Prestressing steel

Diameter (mm) 5.00
Yield strain 0.0035
Initial prestressing force, Po (kN) 14.406*
*assumed
Po= 0.7 Aps fy
76
1 2 3 4 No. of axles
2.78 2.78 3.99 3.16 Axle load (kN)

1.39 1.39 1.995 1.58 Wheel load (kN)
169 845 1014

2028
85
85
TRAVEL
424
85
85
85
All dimension in mm
35
Fig. (6.3) Simulated Ontario Highway Bridge Design trucks (OHBDC)
for one-cell box-girder bridge
60.00
50.00
40.00 with losses

load (kN)
without losses
30.00
20.00
Experimental Ref.(53)
NONLACS Ref.(53)
Present study
10.00
0.00
0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00

deflection (mm)
Fig. (6.4) Analytical and experimental load-midspan deflection
curve for one-cell box-girder bridge
77
distance from left upport (m)

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50
0.00
P/Pu=0.250
2.00
P/Pu=0.500
4.00
deflection (mm)
6.00 P/Pu=0.750
8.00
10.00
12.00
P/Pu=0.875
NONLACS Ref.(53)
14.00
Present study with losses
16.00 Present study without losses
Fig. (6.5) Deflected shape of the bridge at various load levels

60.00
50.00
without losses
40.00
with losses
load (kN)
30.00
20.00 Experimental Rfe.(53)

NONLACS Rfe.(53)
Present study
10.00
0.00
0.00 5.00 10.00 15.00 20.00 25.00

stress (MPa)
Fig. (6.6) Longitudinal normal stress on the top slab at midspan

78
60.00
50.00
without losses
with losses
40.00
load (kN)
30.00
20.00
NONLACS Ref.(53)
10.00 Present study
0.00
0.00 2.00 4.00 6.00 8.00 10.00 12.00

stress (MPa)
Fig. (6.7) Longitudinal normal stress on the top slab at quarterspan
Distance from left support (m)

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50
0.00
2.00
4.00 P/Pu=0.250
6.00
P/Pu=0.500
Stress (MPa)
8.00
10.00
P/Pu=0.750
12.00
14.00
16.00
NONLACS Ref.(53)
18.00 Present study with losses

P/Pu=0.875
Present study without losses
20.00
Fig. (6.8) Variation of concrete longitudinal normal stress along the

centerline of the top slab for one-cell box-girder bridge
79
Experimental
NONLACS
Present study
P/Pu=0.250 P/Pu=0.875
Fig. (6.9) Longitudinal normal stress variation across the section at

midspan for one-cell box-girder bridge
Experimental
NONLACS
Present study
P/Pu=0.250 P/Pu=0.875
Fig. (6.10) Longitudinal normal stress variation across the section of slab
at quarter span for one-cell box-girder bridge
80
Table (6.2) Variation of longitudinal stresses at midspan for Pu/P=0.250 and 0.875.
P/Pu=0.250 P/Pu=0.875
Points Present Present
Exp. NONLACS Exp. NONLACS
study study
A -1.9645 -3.2274 -3.341 -3.5065 -13.1415 -17.209

B - - -3.727 - - -19.723
C - -3.648 -4.287 - -15.3409 -21.537
D - - -3.73 - - -20.028
E -3.929 -3.508 -3.453 -19.286 -19.286 -17.908
C’ - -3.508 -3.026 - -8.7663 -4.406
F - - -3.626 - - -0.959
G -2.806 - -3.618 0.0 0.0 0.959
H - - -3.295 - - 0.521
I - -2.8064 -2.997 - 0.0 0.403
J - - -3.028 - - 0.159
K -0.7016 -2.1048 -3.178 1.7533 0.8766 -0.004
A B C D E
’
C
F
I J K
81
Table (6.3) Variation of longitudinal stresses at quarter span for Pu/P=0.250 and
P/Pu=0.250 P/Pu=0.875
study study
A -1.455 -1.531 -1.373 -2.0358 -6.786 -5.674

B - - -1.293 - - -5.523
C - -1.531 -1.328 - -6.786 -5.547
D - - -1.352 - - -5.85
E -2.143 -1.531 -1.334 -6.786 -6.786 -5.603
C’ - -1.76 -2.725 - -5.70 -5.532
F - - -2.725 - - -3.347
G -0.765 - -3.553 0.0 0.0 -1.616
H - -1.684 -4.046 - - -0.12
I - -1.531 -4.48 - 0.0 0.244
J - - -3.448 - - 0.5
K -0.574 -1.531 -3.445 1.62864 0.0 1.53
0.875.
A B C D E
’
C
F
I J K
82
2
0
-2
-4
-6
-8
-10
-12
-14
-16
-18
-20
-22
-24
-26
Fig. (6.11) Variation of longitudinal stresses in the top surface of the top flange
83
1
0.5
0
-0.5
-1
-1.5
-2
-2.5
-3
-3.5
-4
Fig. (6.12) Variation of transverse stresses in the top surface of the top flange
84
6.5
6
5.5
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
-0.5
Fig. (6.13) Variation of longitudinal stresses in the bottom surface of the bottom
flange
85
1.5
1
0.5
0
-0.5
-1
-1.5
-2
-2.5
-3
-3.5
-4
Fig. (6.14) Variation of transverse stresses in the bottom surface of the bottom
flange
86
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
-1
-2
Fig. (6.15) Variation of longitudinal stresses in the outer surface of the web
87
2.2
2
1.8
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
-1
-1.2
-1.4
Fig. (6.16) Variation of transverse stresses in the outer surface of the web
88
12
11
10
9
8
7
6
5
4
3
2
1
0
-1
-2
Fig. (6.17) Variation of longitudinal stresses in the inner surface of the web
89
1.20
1.00
0.80
0.60
0.40
0.20
0.00
-0.20
-0.40
-0.60
-0.80
-1.00
-1.20
Fig. (6.18) Variation of transverse stresses in the inner surface of the web
90
6.3 Simply Supported Two-Cell Prestressed Concrete Box-Girder Bridge

The same authors[53] of the previous example tested and analyzed another
example. It was a two-cell box-girder, simply supported at its two ends.
The geometry and the finite element mesh are shown in Figs. (6.19) and
(6.20 a). The applied loading for the bridge is shown in Fig. (6.21). The
positioning of the trucks on the bridge model is shown in Fig. (6.20). The
material properties of concrete, and reinforcing and prestressing steel are listed
in Table (6.4).
One- half of the bridge is modeled due to the symmetry of loading with
respect to the longitudinal axis. The two-cell box-girder was modeled with 308
brick elements with a total number of 2266 nodal points, as shown in Fig. (6.20).
In Fig. (6.22), the load-deflection curve of the bridge is shown. Good agreement
with experimental and NONLACS results is satisfied throughout most loading
levels. Failure is initiated due to crushing in concrete and yielding at load equal
to 153.89 kN (11.78 trucks). Comparison between the present study value and
the experimentally reported ultimate load level of 139.0 kN (10.64 trucks),
indicates an error not exceeding 9.68%. But, it is within an error of 5.78% when
comparing with the ultimate load of NONLACS 145.0 kN (11.10 trucks). The
deflected shape due to external loading is shown in Fig. (6.23), it is measured at
various levels of loading. The deflected shapes were measured along the
longitudinal centerline of the bridge. Excellent agreement exists with the
experimental results at various ratios of P/Pu, except the curve at the ratio
P/Pu=0.909. Fig. (6.24) represents the development of longitudinal normal
stress on the top slab at midspan versus the loading. It can be noted that the rate
of development of stress is almost linear. The linear curve was due to the
behavior of the structure. It can be noted that when the applied load is equal to
80.0 kN, a possible crack takes place. This means that no more change occurs in
the properties of concrete and steel (i. e. no further cracking or no yielding of
steel is occurs). Therefore, the crushing in concrete is sudden. Generally, the
91
obtained results are in good agreement with respect to the experimental and
NONLACS results until the load is nearly equal to 135.0 kN. A good agreement
with respect to the experimental results at the top slab at quarter span of the
bridge is also shown in Fig. (6.25). It can be noted that, the obtained results in
the present study are more close to the experimental results than the NONLACS
results. The variation of concrete longitudinal normal stresses along the
centerline of the top slab is shown in Fig. (6.26). The comparison is fairly close
with respect to the experimental results. Figs. (6.27) and (6.28) show the
variation of longitudinal normal stress at the cross-section at midspan and
quarter span for the bridge at a ratio P/Pu equal to 0.182 and 0.727. These
results are listed in tables (6.5) and (6.6).
The deflected shape due to the loading is shown in Fig. (6.20 b). All
displacements in the structure are multiplied by 10 for clearness. The
distribution of the longitudinal and transverse normal stresses for the bridge is
shown in Figs. (6.29) to (6.40).
L
C 560
30
a- cross section 50
135
35 19 30
19
Midspan Section End Span
b-- profile of side view L

C
357
135
35
19
19
3 X 115 2 X 152 2 X 254 406 203
1766
All dimensions in mm
Fig. (6.19) Structural details of the two-cell box-girder bridge

92
a- Finite element mesh for the structure b- Deflected shape for the two-cell box-girder
with 308 elements due to prestressed and external loads
Fig. (6.20) Finite element idealization of half bridge model of two-cell

box-girder before and after loading
93
Table (6.4) Material properties of the two-cell box-girder bridge.
Concrete

Compressive strength, fc` (MPa) 37
Poisson’s ratio, υ 0.18*
Compressive strain at fc` 0.0018
Ultimate compressive strain 0.0045
Reinforcing steel

Diameter (mm) 4
Yield strain 0.0018
Prestressing steel

Diameter (mm) 5.00
Yield strain 0.0035
Initial prestressing force, Po (kN) 21.266*
*assumed
Po=0.7 Aps fy
94
1 2 3 4 No. of axles
2.86 2.86 4.08 3.26 Axle load (kN)

1.43 1.43 2.04 1.63 Wheel load (kN)
171 857 1029

2057
86
86
TRAVEL
257
429
86
86
86
36
All dimension in mm
Fig. (6.21) Simulated Ontario Highway Bridge Design trucks (OHBDC)
for two-cell box-girder bridge
180.00
without losses
160.00
140.00
120.00
with losses
load (kN)
100.00
80.00
60.00 Experimental Ref.(53)

NONLACS Ref.(53)
40.00
Present study
20.00
0.00
0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00

deflection (mm)
Fig. (6.22) Analytical and experimental load-midspan
deflection curves for two-cell box-girder bridge
95
distance from left support (m)

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00
0.00
P/Pu=0.182
2.00 P/Pu=0.364
4.00 P/Pu=0.546
6.00
P/Pu=0.727
8.00
deflection (mm)
10.00
12.00 P/Pu=0.909
14.00
16.00
18.00 NONLACS Ref.(53)

20.00
Present study without losses
22.00
Fig. (6.23) Deflected shape of the bridge at various

load levels for two-cell box-girder bridge
180.00
with losses
160.00
without losses
140.00
120.00
load (kN)
100.00
80.00
60.00
NONLACS Ref.(53)
40.00 Present study
20.00
0.00
-5.00 0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00
stress (MPa)
Fig. (6.24) Longitudinal normal stress on the top slab at

midspan for two-cell box-girder bridge
96
180.00
with losses
160.00
without losses
140.00
120.00
load (kN)
100.00
80.00
60.00 Experimental Ref.(53)

NONLACS Ref.(53)
40.00 Present study
20.00
0.00
0.00 3.00 6.00 9.00 12.00 15.00 18.00 21.00 24.00

stress (MPa)
Fig. (6.25) Longitudinal normal stress on the top slab at quarter span
for two-cell box-girder bridge
distance from left support (m)

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00
0.00
2.00
P/Pu=0.182
4.00
stress (MPa)
P/Pu=0.364
6.00
P/Pu=0.546
8.00
10.00
NONLACS Ref.(53)
P/Pu=0.727 Present study without losses
12.00
Fig. (6.26) Variation of concrete longitudinal normal stresses along

the top slab for two-cell box-girder bridge
97
Experimental
NONLACS
Present study
P/Pu=0.727 P/Pu=0.182

midspan for two-cell box-girder bridge
Experimental
NONLACS
Present study
P/Pu=0.727 P/Pu=0.182

quarter span for two-cell box-girder bridge
98
Table (6.5) Variation of longitudinal stresses at midspan for P/Pu=0.182 and 0.727.
P/Pu=0.182 P/Pu=0.7275
study study
A -3.245 -5.409 -1.849 -7.768 -8.4152 -11.23

B - - -1.989 - - -12.147
C -5.409 - 2.159 -8.4152 - -13.88
D - - -1.982 - - -12.254
E -6.49 -5.950 -1.854 -9.710 -9.71 -11.425
F - - -1.992 - - -12.49
G -5.409 - -2.17 -8.4152 - -14.269
H - - -1.20 - - -12.325
I -2.8125 -5.409 -1.869 -9.0625 -9.0625 -11.645
C' - -6.490 -1.979 - -5.567 -2.787
J - - -2.135 - - -1.158
L -3.894 - -2.446 0.0 0.0 -0.596
N - -2.163 -2.665 - 2.589 1.277
G' - -6.49 -2.121 - -4.855 2.92
K - - -2.292 - - 1.323
M -4.111 - -2.456 0.0 0.0 -0.621
R' - -2.163 -2.679 - 2.589 -1.217
N' - -1.082 -2.649 - 3.884 -2.272
O - - -2.683 - - -2.284
P -1.947 - -2.747 3.237 - -2.342
Q - - -2.695 - - -2.204
R - 1.082 -2.665 - 2.589 -2.236
A B C D E F G H I
C' G'
J K
L M
N R'
N' O P Q R
99
Table (6.6) Variation of longitudinal stresses at quarter span for P/Pu=0.182 and 0.727.
P/Pu=0.182 P/Pu=0.7275
study study
A - -2.344 -1.469 - -5.729 -6.443

B - - -1.54 - - -6.607
C -2.344 - -1.594 -7.563 - -6.802
D - - -1.537 - - -6.618
E -1.875 -2.344 -1.487 -5.271 -5.156 -6.521
F - - -1.551 - - -6.675
G -1.641 - -1.612 -4.583 - -6.846
H - - -1.556 - - -6.668
I -1.875 -1.875 -1.504 -6.875 -4.01 -6.573
C' - -2.156 -1.766 - -5.156 -4.97
J - - -2.161 - - -1.682
K -1.219 - -2.521 0.0 0.0 0.872
L - -1.406 -3.194 - 2.979 0.992
G' - -1.875 -2.083 - -5.156 -2.367
Q - - -2.498 - - 0.275
R -0.9375 - -2.533 0.0 0.0 0.878
P' - -1.172 -3.561 - 5.042 1.409
L' - -0.9375 -3.206 - 2.292 1.5
M - - -3.195 - - 1.476
N -1.406 - -3.23 3.438 3.438 1.461
O - - -3.205 - - 1.476
P - -1.219 -3.592 - 2.929 0.896
A B C D E F G H I
C' G'
J Q
K R
L P'
L' M N O P
100
0
-2
-4
-6
-8
-10
-12
-14
-16
-18
-20
-22
-24
101
1.5
0.5
-0.5
-1
-1.5
-2
-2.5
-3
102
1.5
1
0.5
0
-0.5
-1
-1.5
-2
-2.5
-3
-3.5
-4
flange
103
0.5
0
-0.5
-1
-1.5
-2
-2.5
-3
-3.5
-4
-4.5
-5
-5.5
flange
104
1
0
-1
-2
-3
-4
-5
-6
-7
-8
-9
-10
Fig. (6.33) Variation of longitudinal stresses in the outer surface of the right web
105
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
-0.1
-0.2
-0.3
-0.4
-0.5
-0.6
-0.7
-0.8
-0.9
-1
-1.1
Fig. (6.34) Variation of transverse stresses in the outer surface of the right web
106
0.5
-0.5
-1.5
-2.5
-3.5
-4.5
-5.5
-6.5
-7.5
-8.5
-9.5
Fig. (6.35) Variation of longitudinal stresses in the inner surface of the right web
107
0.6
0.5
0.4
0.3
0.2
0.1
0
-0.1
-0.2
-0.3
-0.4
-0.5
Fig. (6.36) Variation of transverse stresses in the inner surface of the right web
108
0
-1
-2
-3
-4
-5
-6
-7
-8
-9
-10
-11
-12
Fig. (6.37) Variation of longitudinal stresses in the outer surface of the left web
109
0.8
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
-1
-1.2
-1.4
-1.6
Fig. (6.38) Variation of transverse stresses in the outer surface of the left web
110
0
-1
-2
-3
-4
-5
-6
-7
-8
-9
-10
-11
-12
Fig. (6.39) Variation of longitudinal stresses in the inner surface of the left web
111
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
-1
-1.2
-1.4
-1.6
Fig. (6.40) Variation of transverse stresses in the inner surface of the left web
112
6.4 Simply Supported Single-Cell Prestressed Concrete Box-Girder with

Inclined Web
A longitudinally prestressed single-cell box-girder, simply supported at both
ends was analyzed by Jirousek et al[32]. They used three macro-elements to
model one half-span of the bridge. Each macro-element was formed by two
identical assembly elements and four modified thick shell elements. The
equivalent nodal force method was adopted to represent the prestressing forces.
Using Ahmed degenerated shell element, Al-Temimi analyzed the same
example. In that study, one half of the bridge is modeled with 24 elements. The
equivalent nodal force method was also adopted in that study.
The box-girder is longitudinally prestressed by parabolic tendons located
within the webs. The profile of the tendons and geometry of the bridge model
are shown in Fig. (6.41). The material properties of the bridge model are listed
in Table (6.7). Each web of the bridge is provided with one-parabolic cable as
shown in Fig. (6.41). The intensity of cable tension was assumed constant.
Due to symmetry of loading and geometry, one half of the bridge was
modeled with 176 brick elements and 1312 nodal points as shown in Fig. (6.42).
In this example, the procedure used by Lin[38] and applied by Loo and Cusens[39]
is used to represent the prestressing forces at the nodes, as explained in Section
(5.4). This method is also used by Abdul-Razaq[1].
Fig. (6.43) shows the vertical deflections for the cross-section at midspan
due to prestressing forces only. Good agreement is obtained by comparing with
Jirousek et al[32] and Al-Temimi[10] solutions. In Fig. (6.44), the distribution of
longitudinal stresses at cross-section at midspan is shown. The obtained results
are fairly close to Jirousek et al[32] solution. The results of Figs. (6.43) and (6.44)
are listed in Tables (6.8) and (6.9).
The deflected shape for one half-span of the bridge is shown in Fig.
(6.42 b). These displacements are multiplied by 30 for clearness.
113
The distribution of longitudinal and transverse stresses for the whole

structure is shown in Figs. (6.45) to (6.56).
25.00 cm
56.00 cm
28500 kN 150.00 cm
25.00 cm
10.00 cm
625.00 cm 625.00 cm 625.00 cm
a- profile
190.00 cm
35.00 cm 375.00 cm
190.00 cm
b- plan
190.00 cm 185.00 cm
35.00 cm
22.50 cm
150.00 cm
32.00 cm
c- cross-section 25.00 cm
200.00 cm
Fig. (6.41) Structural details and cable profile of prestressing for box-girder
bridge with inclined webs
114
a- Finite element mesh for half bridge with 176 brick

elements
b- Deflected shape due to prestressing loading only

magnified by 30 times
Fig. (6.42) Finite element mesh for the bridge model before and after loading
115
Table (6.7) Material properties for the prestressed box-girder bridge with inclined
webs.
Concrete

Compressive strength, fc` (MPa) 33.64*
Poisson’s ratio, υ 0.15
Compressive strain at fc` 0.0018*
Ultimate compressive strain 0.0045*
Initial prestressing force, Po (kN) 28500
*assumed
*E c  5000 f c
*f t  0 .54 f c
116
Experimental
NONLACS
Present study
Fig. (6.43) Deflection at cross-section for midspan for Jirousek
Fig. (6.44) Longitudinal stresses at cross-section for midspan for Jirousek
117
Table (6.8) Variation of deflections for cross-section at midspan (mm).
Points Jirousek et al Al-Temimi present study
A 25.2 24.87 25.92

B 25.1 26.0 25.95
C 25.0 24.8 25.99
D 25.0 26.02 26.07
E 25.0 24.77 26.16
F 25.0 26.02 26.27
G 25.30 24.8 26.41
H 25.1 26.0 26.54
I 25.2 24.87 26.66
J 25.0 25.71 25.98
K 25.0 25.71 26.42
L 25.0 25.62 25.98
M 25.0 25.62 26.06
N 25.0 25.62 26.14
O 25.0 25.62 26.24
P 25.0 25.62 26.36
118
Table (6.9) Variation of longitudinal stresses for cross-section at midspan (MPa).
Points Jirousek et al present study
A 0.283 0.3049
B .126 0.3203
C .078 0.2998
D 0.138 0.3060
E 0.189 0.3497
F 0.138 0.3519
G 0.078 0.303
H 0.126 0.3213
I 0.283 0.3059
J -0.752 -1.008
K -0.752 -1.015
L -2.851 -2.836
M -2.851 -2.841
N -3.601 -5.242
O -3.601 -5.238
P -4.171 -5.242
Q -4.314 -5.238
R -4.2 -5.38
S -4.314 -5.251
T -4.171 -5.242
A B C D E F G H I
J K
L M
N L
P Q R S T
119
-0.35
-0.45
-0.55
-0.65
-0.75
-0.85
-0.95
-1.05
-1.15
-1.25
-1.35
120
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
-0.1
-0.2
-0.3
-0.4
-0.5
-0.6
-0.7
121
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
-1
flange
122
3.5
2.5
1.5
0.5
-0.5
-1
flange
123
19
17
15
13
11
9
7
5
3
1
-1
Fig. (6.49) Variation of longitudinal stresses in the outer surface of the left web
124
2.4
2
1.6
1.2
0.8
0.4
0
-0.4
-0.8
-1.2
-1.6
Fig. (6.50) Variation of transverse stresses in the outer surface of the left web
125
19
17
15
13
11
9
7
5
3
1
-1
Fig. (6.51) Variation of longitudinal stresses in the inner surface of the left web
126
2.2
2
1.8
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
Fig. (6.52) Variation of transverse stresses in the inner surface of the left web
127
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
Fig. (6.53) Variation of longitudinal stresses in the outer surface of the right web
128
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
-0.1
-0.2
-0.3
-0.4
-0.5
-0.6
-0.7
-0.8
Fig. (6.54) Variation of transverse stresses in the outer surface of the right web
129
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
-1
-2
Fig. (6.55) Variation of longitudinal stresses in the inner surface of the right web
130
1.1
0.9
0.7
0.5
0.3
0.1
-0.1
-0.3
-0.5
-0.7
-0.9
Fig. (6.56) Variation of transverse stresses in the inner surface of the right web
131
Chapter Seven Conclusions and Recommendations
Chapter Seven
CONCLUSIONS AND RECOMMENDATIONS
7.1 Conclusions
The present study has been used concrete brick element with embedded steel bar
element for the analysis of prestressed concrete box-girder bridges. The
prestressing forces are dealt by using the equivalent nodal force method and by
Lin's method. The losses in prestressing forces had been considered in the
analysis. Several examples are analyzed by the present method and the results
are compared with the available experimental and other analytical results. From
these results, the following conclusions are drawn.
1. The nonlinear finite element method presented in this study has shown to
be capable of predicting the structural response of prestressed concrete
box-girder bridges. The isoparametric brick element proved to be suitable
for predicting the state of ultimate load, deflection and stress with good
accuracy. The difference for the single-cell box-girder with respect to
experimental results of Ref.(53), in ultimate load is 2.59%, and in stress at
quarterspan is 2.27%, and it is equal to 12.69% for the stress at midspan.
In the second example, the two-cell box-girder bridge, the difference is to
be less than 9.15% comparing with experimental results in Ref.(53) for
ultimate load, but it is less than 5.23% comparing with theoretical results
for the same authors. The tolerance is equal to 2.06% for the deflection at
midspan, and it is equal to 8.57% for the stress at quarterspan, and it is
less than 9.97% for the stress at midspan. For the prestressed concrete
box-girder with inclined webs, the third example, the maximum
difference with respect to the theoretical results in Ref.(32), in deflection
132
is equal to 8.0% at top flange, and in stress is to be less than 14.0% at the
bottom flange.
2. The losses in prestressing forces used in the present study improved the
obtained results. For the first example, it was found that the accuracy
increased with 2.83% for the ultimate load, and it is increased with 4.67%
for the deflection at midspan. Besides, the accuracy is increased with
0.75% for the stress at quarterspan, and it is increased with 0.26% for the
stress at midspan. In the second example, the tolerance increased with
4.51% for the ultimate load, and it is increased with 2.6% for the
deflection at midspan. While, the accuracy is increased with 11.8% for the
stress at quarterspan, and it is increased with 15.0% for the stress at
midspan. For the third example, the tolerance increased with 12.0% for
the deflection at the top flange.
3. The concept of equivalent nodal forces used in the present study is
capable to simulate the loads exerted by the prestressing tendon upon the
structure with fair accuracy. Also, the Lin's method is proved to be
suitable to simulate the forces due to the parabolic tendon upon the
structure.
4. The contribution of the prestressing tendon stiffness to the element
stiffness is contributed, and it is affected on the analysis.
5. Boundary conditions caused most of the differences in the results. The
boundary conditions should be taken as exactly as possible it is in the
experimental work.
133
7.2 Recommendations
The following suggestion for the possible extension and improvement of the
present study can be made:
1. The dependent effects are needed to be incorporated in this study. These
effects include concrete shrinkage, creep and aging of concrete as well as
relaxation in prestressing steel.
2. Dynamic response of prestressed concrete box-girder bridges due to blast
load and earthquakes may be studies.
3. Modified the available program to take thermal loading into account.
4. Modified the available program by treating the tendon as an integral part
of the structure.
134
References
REFERENCES
1) Abdulla M.A. and Abdul-Razzak A.A., “Finite Strip Analysis of

Prestressed Box-Girder”, Computers and Structures, Vol.36, No.5, pp.817-
822, 1990.
2) Abdulla M.A. and Abdul-Razzak A.A., “Solution of Continuous Box-
Girder Bridges Using Higher Order Finite Strip Method”, Engineering
and Technology, Baghdad, Vol.12, pp. 50-68, 1993.
3) Abdul-Razzak A.A., “Distribution of Moments and Shear Forces in Box-
Girder Bridges Using Higher Order Finite Strip”, Al-Rafidain
Engineering, Mosul, Vol.1, No.2, pp.32-45.
4) Abdul-Razzak A.A., “Finite Strip Analysis of Curved Box-Girders”, Civil
Engineering Department, University of Mosul, Mosul, Iraq. (cited in Ref. 8)
5) Al-Bahadly H.A.K., “Nonlinear Analysis of Elastic Plastic Steel and
Reinforced Concrete Beam in Torsion Using Three-Dimensional Finite
Element Model”, M.Sc. Thesis, University of Technology, 1995.
6) Allose L.E.H., “Three-Dimensional Nonlinear Finite Element Analysis of
Steel Fiber Reinforced Concrete Beam in Torsion”, M.Sc., Thesis,
University of Technology, 1995.
7) Al-Mahaidi R.S.H., “Non-Linear Finite Element Analysis of Reinforced
Concrete Deep Members”, Report No.79.1, Department of structural
Engineering, Cornell University, Ithaca, New York, January 1979. (cited in
Ref. 8)
8) Al-Sabah A.S.I., “Non-Linear Time-Dependent Finite Element Analysis
of Plane Reinforced Concrete Members”, M.Sc. Thesis, University of
Baghdad, February 1983.
135
References
9) Al-Shaarbaf I., “A Nonlinear Three-Dimensional Finite Element Analysis

Program for Steel and Reinforced Concrete Beam in Torsion”, Ph.D.
Thesis, University of Bradford, 1990.
10) Al-Temimi J.E.M., “Analysis of Reinforced and Prestressed Concrete
Box-Girder Bridges by Ahmad Degenerated Shell Element”, M.Sc.
Thesis, University of Al-Mustansiria, 2002.
11) Alwood R.J. and Bajarwan A.A., “Modeling Non-Linear Bond-Slip
Behavior for the Finite Element Analysis of Reinforced Concrete
Structures”, ACI Structural Journal, Vol.96, No.5, pp.538-544, Sep.-Oct.
1996. (cited in Ref.65)
12) Al-Zahawi S.K., “Nonlinear Finite Element Analysis of Reinforced
Concrete Voided Slab Strips”, M.Sc. Thesis, University of Technology,
1999.
13) Aneja L.K. and Roll E., “Model Analysis of Curved Box-Beam Highway
Bridge”, Proceedings of the American Society of Civil Engineering ASCE,
Vol.97. No.ST12, December 1971.
14) Cervenka V., “Constitutive Model for Cracked Reinforced Concrete”, ACI
Journal, Vol.82, No.6, pp.877-882, November-December 1985.
15) Chen W.F., “Plasticity in Reinforced Concrete”, McGraw-Hill, 1982.
16) Cheung Y.K., “Analysis of the Box-Girder Bridges by Finite Strip
Method”, ACI Publication, Vol.1, SP-26, 1971.
17) Cheung Y.K., “Finite Strip Method in Structural Analysis”, Pergamon
Press, 1976. (cited in Ref.44)
18) Cheung Y.K., “The Finite Strip Method in Analysis of Elastic Plates with
Two Opposite Simply-Supported Ends”, Proceedings, Institution of Civil
Engineers, U.K., Part 1, December 1968.
19) Cheung Y.K. and Fam S.C., “Static Analysis of Right Box-Girder Bridges
by Spline Strip Method”, Proceedings, Institution of Civil Engineers, U.K.,
Part2, pp.311-323, 1983.
136
References
20) Cook R.D., “Concepts and Applications of Finite Element Analysis”,

Second Edition, John Wiley and Sons, New York, 1981.
21) Cusens A.R. and Pama R.P., “Bridge Deck Analysis”, John Wiley and
Sons, 1975.
22) Dabrowski R., “Curved Thin-Walled Girders, Theory and Analysis”,
Springer, 1968. (cited by Heins and Sheu[28])
23) Dawe D. J., “Matrix and Finite Element Displacement Analysis of
Structures”, Clarendon Press, Oxford, 1984.
24) Fam A. and Turkstra C., “A Finite Element Scheme for Box Bridges
Analysis”, Computers and Structures, Vol.5, pp.117-136, 1975.
25) Fam A. and Turkstra C., “Model Study of Horizontally Curved Box-
Girders”, Journal of the Structural Division, Proceedings of the American
Society of Civil Engineers ASCE, Vol.102, No.ST5, pp.1097-1108, May
1976.
26) Gerstle K.H., “Behavior of Concrete under Multiaxial Stress State”,
Journal of the Engineering Mechanic Division, ASCE, Vol.106, No-EM6,
pp.1383-1403, December 1980.
27) Ghali A. and Ghonhm G.A., “Nonlinear Analysis of Concrete Structures”,
Canadian Journal of Civil Engineering, Vol.9, 1982.
28) Ghalib A.M.A., “Inelastic Finite Element Analysis of Prestressed
Concrete Multi-Planar Systems”, M.Sc. Thesis, Department of Civil
Engineering, University of Baghdad, 1990.
29) Haung T., “Anchorage Take-Up Loss in Post-Tensioned Members”, PCI
Journal, Vol.14, August 1969.
30) Heins C.P. and Olienik J.C., “Curved Box Beam Bridge Analysis”,
Computers and Structures, Vol.6, pp.65-73, 1976.
31) Heins C.P. and Sheu F.H., “Design/Analysis of Curved Box-Girder
Bridge”, Computers and Structures, Vol.4, pp.675-698, 1974.
137
References
32) Jirousek J. and Bouberguig A., “A Macro-Element Analysis of Prestressed

Curved Box-Girder Bridges”, Computers and Structures, Vol.10, pp.467-
482, 1979.
33) Kang Y. J. and Scordelis A.C., “Nonlinear Analysis of Prestressed
Concrete Frames”, Journal of the Structural Division, Proceedings of the
American Society of Civil Engineers ASCE, Vol.106, No.ST2, pp.445-462,
February 1980.
34) Kindeel Z.M.R., “Nonlinear Finite Element Analysis of Reinforced
Concrete Beam-Column Joints”, M.Sc. Thesis, University of Al-
Mustansiria, July 2003.
35) Kupfer H., Hilsdorf H.K. and Raush H., “Behavior of Concrete under
Biaxial Stresses”, ACI Journal, Vol.66, No.8, pp.656-666, August 1969.
36) Li W.Y., Tham L.G. and Cheung Y.K., “Curved Box-Girder Bridges”,
Journal of Structural Engineering ASCE, Vol.114, No.6, June 1988.
37) Libby J.R. and Perkins N.D., “Highway Bridge Superstructure, Modern
Prestressed Concrete, Design Principles and Construction Methods”, San
Diego, California 1976.
38) Lin T.Y. and Burns N.A., “Design of Prestressed Concrete Structures”,
Third Edition, John Wiley and Sons, 1982.
39) Loo Y.C. and Cusens A.R., “A Finite Strip Program for the Analysis of
Box-Girder Bridges”, User Manual, CIRIA Report 127/7, Civil
Engineering Department, University of Dundee, 1972. (cited in Ref. 1)
40) Loo Y.C. and Cusens A.R., “The Finite Strip Method on Bridge
Engineering”, Eyre Spottiswoode, London 1978. (cited in Ref. 1)
41) Lutz L.A. and Gergely P., “Mechanical of Bond and Bond Slip of
Deformed Bars in Concrete”, ACI Journal, Vol.64, No.11, pp.711-721.
42) Magura D.D., Sozen M.A. and Siess C.P., “A Study of Stress Relaxation in
Prestressing Reinforcing”, PCI Journal, Vol.2, April 1964.
138
References
43) Mahmood A.A., “Nonlinear Finite Element Analysis of Reinforced

Concrete Structures Under Long-Term Loading”, M. Sc. Thesis,
University of Baghdad, Iraq 2001. (cited in Ref.65)
44) Mavaddat S. and Mirza M.S., “Computer Analysis of Thin-Walled
Concrete Box Beams”, Canadian Journal of Civil Engineering, Vol.16,
p.902-909, 1989.
45) Mehrotra B.L., Mufti A.A. and Redwood R.G., “Analysis of Three-
Dimensional Thin-Walled Structures”, Journal of the Structural Division,
Proceedings of the American Society of Civil Engineers ASCE, No. ST12,
pp. 2863-2872, 1969.
46) Meyer C. and Scordelis A.C., “Analysis of Curved Folded Plates
Structures”, Journal of the Structural Division, Proceedings of the
American Society of Civil Engineers, Vol.97, No.ST10, pp.2459-2480,
October 1971.
47) Ngo D. and Scordelis A.C., “Finite Element Analysis of Reinforced
Concrete Beams”, ACI Journal, Vol. 64, No.3, pp.152-163 March 1967.
48) Nilson A.H., “Internal Measurement of Bond Slip”, ACI Journal, No.69,
pp.439-441, July 1972.
49) Nilson A.H. and Darwin D., “Design of Concrete Structure”, Twelfth
Edition, 1997.
50) Owen D. R. and Hinton E., “Finite Element in Plasticity, Theory and
Practice”, Pineridge Press, Swansea, U. K. 1980.
51) Pagnani T., Slater J., Aneur-Mosor R. and Boyukourk O., “A
Nonlinear Three-Dimensional Analysis of Reinforced Concrete Beam on
Boundary Surface Model”, Computers and Structures, Vol.43, No.1, pp.1-
12, 1992.
52) Rashid Y.R., “Analysis of Prestressed Concrete Pressure Vessels”,
Nuclear Engineering and Design, Vol.7, No.4, April 1968.
139
References
53) Razaqpur A.G., Nofal M. and Mirza M.S., “Nonlinear Analysis of

Prestressed Concrete Box-Girder Bridges under Flexure”, Canadian
Journal of Civil Engineering, Vol.16, pp.845-853, 1989.
54) Roca P. and Mari A.R., “Numerical Treatment of Prestressing Tendons in
the Nonlinear Analysis of Prestressed Concrete Structures”, Computers
and Structures, Vol. 46, No. 5, pp.905-916, 1993
55) Roca P. and Mari A.R., “Geometric and Material Analysis of Prestressed
Concrete General Shell Structures”, Computers and Structures, Vol.46,
No. 5, pp.917-929, 1993.
56) Sawko F. and Cope R.J., “Linear Analysis of Complete Structures by
Computers”, Proceeding, Institution of Civil Engineers, U.K., Vol.40, June
1968. (cited by Mehrotra et al[34])
57) Schlaich J. and Seheel H., “Concrete Box-Girder Bridges”, International
Association for Bridge and Structural Engineers, Swiss, 1982.
58) Scordelis A.C., “Analysis of Continuous Box-Girder Bridges”, Report
SESM 67-25, University of California, Berkeley, November 1967. (cited in
Ref. 21)
59) Scordelis A.C., Chan E.C., Ketchum M.A. and Walt P., “Computer
Programs for Prestressed Concrete Box-Girder Bridges”, Structural
Engineering and Structural Mechanics, Department of Civil Engineering,
Report No.UCB/SEM-85/03, University of California, Berkeley, 1985.
60) Scordelis A.C., Davis R.E. and Lo K.S., “Load Distribution in Concrete
Box-Girder Bridges”, ACI Publications, SP23-8, pp.117-136, 1969.
61) Scordelis A.C. and Larsen P.K., “Structural Response of Curved
Reinforced Concrete Box-Girder Bridge”, Journal of the Structural
Division, Proceedings of the American Society of Civil Engineers ASCE,
No.ST8, pp.1507-1524, August 1977.
62) Scordelis A.C., Larsen P.K. and Elfgren L.G., “Ultimate Strength of
Curved Reinforced Concrete Box-Girder Bridges”, Journal of the
140
References
Structural Division, Proceedings of the American Society of Civil

Engineers ASCE, Vol. ST8, pp.1525-1542, August 1877.
63) Scordelis A.C., Meyer C. and Seni A., ”Analysis of Horizontally Curved
Box-Girder Bridges”, Journal of the Structural Division, Proceedings of the
American Society of Civil Engineers ASCE, No.ST8, pp.1878-1880,
August 1972.
64) Trikha D.N. and Edwards A.D., “Analysis of Concrete Box-Girders before
and After Cracking”, Proceedings of the Institution of Civil Engineers,
U.K., Part 2, Paper No.7571, 1973.
65) Turkstra C.J. and Fam A.R.M., “Behavior Study of Curved Box Bridges”,
Journal of the Structural Division, Proceedings of the American Society of
Civil Engineers ASCE, No.ST3, pp.453-462, March 1978.
66) Valsov V.Z., “Thin Walled Elastic Beam”, Second Edition, NSF,
Washington, D.C., 1971. (cited by Heins and Sheu[28])
67) Yoo C., Evick D.R. and Heins C.P., “Non-Prismatic Curved Girder
Analysis”, Computers and Structures, Vol. 4, pp.675-698, 1974.
68) Zeinkiweicz O.C., “The Finite Element Method”, Third Edition, Mc Graw,
Hill, London, 1977.
69) Zhang S.H. and Lyons L.D.R., “A Thin-Walled Box Beam Finite Element
for Curved Bridge Analysis”, Computers and Structures, Vol.18, No.6,
pp.1035-1046, 1984.
70) Zyle V.S.F. and Scordelis A.C., ”Analysis of Segmentally Erected
Prestressed Concrete Box-Girder Bridges”, Structural Engineering and
Structural Mechanics, Department of Civil Engineering, Report
No.UC/SESM-78/02, University of California, Berkeley, January 1978.
141
Appendix A
Appendix A
The values of  and  are very appreciably, depending on construction

methods and materials used. The values listed in below table may be used as a
guide in the ACI Code Commentary[70].
Table A-1 Friction coefficients for post-tensioned tendons.
Type of Tendon Wobble coeff.  , per ft Curvature coeff. 
Grouted tendons in metal sheathing

Wire tendons 0.0010-0.0015 0.15-0.25
High strength bars 0.0001-0.0006 0.08-0.30
Seven-wire strand 0.0005-0.0020 0.15-0.25
Unbonded Tendons
Mastic-coated wire tendons 0.0010-0.0020 0.05-0.15

Mastic-coated seven-wire strand 0.0010-0.0020 0.05-0.15
Pregreased wire tendons 0.0003-0.0020 0.05-0.15
Pregreased seven-wire strand 0.0003-0.0020 0.05-0.15
A
Appendix B
Appendix B
Referring to Eq. (5.3):

P  po . e  (     x ) …(A.1)
The curvature of a parametric curve “a” is constant, thus  can be written
as follow:
  ax
then   x  qx
where q  a   …(A.2)
Eq. (A.1) then becomes:
P  Po e  qx …(A.3)
From Fig. (5.2) it can be stated that:
P1  Po . e  qla
P2  P1 . e  qla
or P2  Po . e  2 qla …(A.4)
The average force loss P over length l a is approximately:
P  ( Po  P2 ) / 2
thus: P  Po ( 1  e  2 qla ) / 2 …(A.5)
From the elastic properties of the tendon, P can also be expressed as:
P  (  s A s E s ) / l a …(A.6)
where:
 s : anchor slip.
As : tendon cross section area.
E s : tendon elastic modulus.
Substituting Eq. (A.5) into Eq. (A.6), l a can be expressed as:
2  s As E s
la  …(A.7)
Po ( 1  e  2 qla )
‫اﻟ ﻼﺻﺔ‬
‫ـ ﻋﺔ ﻣـ ﺧرﺳـﺎﻧﺔ ﻣ ـ ﻘﺔ‬ ‫ـ ر ذات اﻟ ﻘـﺎ ﻊ اﻟ ـ دوﻗ ﺔ واﻟ‬ ‫ﺗ ﻓـﻲ ﻫـذﻩ اﻟدراﺳـﺔ ﺗ ﻠ ـﻞ اﻟ‬
‫ــر اﻟ ـ از‬ ‫اﻟ ﻬــد ﺗ ــت ﺗــﺄﺛ ر اﻷﺣ ــﺎل ﻗ ـ رة اﻷﻣــد‪ .‬ﺗ ـ ﺗ ــﻞ اﻟ ﻘــﺎ ﻊ اﻟ ـ دوﻗ ﺔ ﺎﺳ ـ دام اﻟﻌ‬
‫ﻋﻘدة ذات درﺟﺎت اﻟ ﻼﻗﺔ اﻟ ﻼﺛﺔ )ﺛـﻼث ازاﺣـﺎت(‪ .‬أﻣـﺎ دراﺳـﺔ ﺣدﯾـد اﻟ ـﻠ ﺢ ﻓﻘـد‬ ‫اﻟ ﺎﺑ ﻗﻲ ذو اﻟﻌ ر‬
‫ر اﻟ رﺳﺎﻧﺔ‪ .‬ﺳﻠ ك اﻟ رﺳﺎﻧﺔ ﺗ ـت‬ ‫ر ﻓﻲ ﻋ‬ ‫ر أﺣﺎد اﻟ ﻌد )ﻣ ر ( ﻣ‬ ‫ﺗ ﺗ ﻠﻪ ﺎﺳ دام ﻋ‬
‫ذج ﻣرن‪-‬ﻟدن ﯾ ﻌﻪ ﺟزء ﻟدن ﺗﺎم اﻟذ ﯾ ﻬﻲ ﻋ د ﺑدا ﺔ ﺗﻬ ـ اﻟ رﺳـﺎﻧﺔ‪ .‬أﻣـﺎ ﺗ ـت‬ ‫اﻻﻧ ﻐﺎط ﻣ ﻞ ﺑ‬
‫اﻟ ــد ﻓﻘــد اﺳ ـ دم ﻧ ـ ذج ﺗ ز ــﻊ اﻟ ــﻘ ق اﻟﻌ د ــﺔ ﻋﻠــﻰ ﻣ ــﺎور اﻹﺟﻬــﺎد اﻟرﺋ ـ ﺔ ﻣــﻊ اﻷﺧــذ ﺑ ــر‬
‫اﻟ ﻘ ق واﻟ ﻐ ر ﻓﻲ ﻣﻌﺎﻣﻞ اﻟﻘص‪.‬‬ ‫اﻻﻋ ﺎر ﻣ ﺎﻫ ﺔ اﻟ رﺳﺎﻧﺔ اﻟ ﺟ دة ﺑ‬
‫ﻠ ﺔ‬ ‫ﻞ ﻗ ة اﻟ د اﻟ‬ ‫اﻟﻌﻘد ﺔ اﻟ ﺎﻓئﺔ ﻟ‬ ‫ﺎﻋ ﺎد ر ﻘﺔ اﻟﻘ‬ ‫ﺗ إدﺧﺎل ﺗﺄﺛ ر اﻟ ﻬد اﻟ‬
‫ــﻞ اﻟ ﻬــد‬ ‫ــر‪ .‬ﺗـ اﺳـ دام ر ﻘــﺔ ﺛﺎﻧ ــﺔ ﻟ‬ ‫ﻣ ــﻠ ﺔ ﻓــﻲ اﻟﻌﻘــد ﻟ ــﻞ ﻋ‬ ‫ﻣـ ﻗ ــﻞ ﺣ ــﻞ اﻟ ــد إﻟــﻰ ﻗـ‬
‫‪(Global‬‬ ‫ــﻔ ﻓﺔ اﻟ ﻠ ــﺔ‬ ‫ﻫــﻲ ر ﻘــﺔ ﻟـ )‪ . (Lin‬ﺗـ إدﺧــﺎل ﻣ ــﻔ ﻓﺔ ﺣ ــﻞ اﻟ ــد إﻟــﻰ اﻟ‬ ‫ـ‬ ‫اﻟ‬
‫ــﺎ درﺳــت ﺣﺎﻟــﺔ‬ ‫ــر اﻟ رﺳــﺎﻧﺔ‪.‬‬ ‫ــر أﺣــﺎد اﻟ ﻌــد )ﻣ ـ ر ( ﻣ ـ ر ﻓــﻲ ﻋ‬ ‫)‪ stiffness‬ﺎﻋ ــﺎرﻩ ﻋ‬
‫ـ ‪ .‬وﻗـد اﺧـذ ﻓـﻲ اﻟ ﻠ ـﻞ ـﺎﻫرة اﻧـزﻻق‬ ‫اﺳ ﺎق اﻟ ﻬد ﻋ د وﺟ د ﺧ ﺎﺋر ﻗ ـ رة اﻟ ـد ﻓـﻲ اﻟ ﻬـد اﻟ‬
‫اﻟ رﺳﺎﻧﺔ وﺣ ﻞ اﻟ د‪.‬‬ ‫اﻟر ط ﻋ د ﺳ ﺢ اﻟ ﻼﻣس ﺑ‬
‫ﺗ ﺣﻞ ﻣﻌﺎدﻻت اﻟ ازن اﻟﻼﺧ ﺔ ﺎﺳـ دام ر ﻘـﺔ ﺗزاﯾد ـﺔ‪-‬ﺗﻌدد ـﺔ ﺗﻌ ـﻞ ﺗ ـت أﺣ ـﺎل ﻣ ـ رة‬
‫)‪ (Newton-Raphson‬اﻟﻌﺎﻣﺔ واﻟ ﻌدﻟﺔ‪ .‬أﺟر ت اﻟ ـﺎﻣﻼت‬ ‫واﺳ دﻣت ﻓﻲ اﻟ ﻞ ر ﻘﺔ ﻧ ﺗ ‪-‬راﻓ‬
‫اﻟﻌدد ﺔ ﺎﺳ دام ﻗ اﻋد اﻟ ﺎﻣﻞ ذات ‪ 27‬ﻧﻘ ﺔ ﺗ ﺎﻣﻞ‪.‬‬
‫ــﻬﺎ أو ﺗ ﻠ ﻠﻬــﺎ ﻣ ـ ﻘﺎً ﺎﺳـ دام‬ ‫ــﺎرة واﻟ ــﻲ ﺗـ ﻓ‬ ‫ﺗ ــت دراﺳــﺔ وﺗ ﻠ ــﻞ ﻋــدد ﻣـ اﻷﻣ ﻠــﺔ اﻟ‬
‫رق ﻋ ﻠ ﺔ وﻧ ر ﺔ ﻣﻌ دة أﺧر ‪ .‬أ ﻬـرت ﻧ ـﺎﺋﺞ اﻟ ﻘﺎرﻧـﺔ ـﺎن ر ﻘـﺔ اﻟﻌ ﺎﺻـر اﻟ ـددة أﻋ ـت ﻧ ـﺎﺋﺞ‬
‫ﺟ دة ﻣﻘﺎرﻧﺔ ﻣﻊ اﻟ ﺎﺋﺞ اﻟﻌ ﻠ ﺔ واﻟ ر ﺔ‪.‬‬

‫ﺟﺎﻣﻌﺔ ﺑﻐﺪﺍﺩ‬
‫ﻛﻠﻴﺔ ﺍﻟﻬﻨﺪﺳﺔ‬
‫ﻗﺴﻢ ﺍﻟﻬﻨﺪﺳﺔ ﺍﻟﻤﺪﻧﻴﺔ‬
‫دوﻗ ﺔ‬ ‫ﻘﺔ اﻹﺟﻬﺎد ذات اﻟ ﻘﺎ ﻊ اﻟ‬ ‫اﻟ ﻠ ﻞ اﻹﻧ ﺎﺋﻲ ﻟﻠﻌ ارض اﻟ رﺳﺎﻧ ﺔ اﻟ‬
‫ﺎﺳ دام ر ﻘﺔ اﻟﻌ ﺎﺻر اﻟ ددة‬
‫إ روﺣﺔ ﻣﻘدﻣﺔ‬
‫إﻟﻰ ﻠ ﺔ اﻟﻬ دﺳﺔ ﻓﻲ ﺟﺎﻣﻌﺔ ﻐداد‬
‫ر ﻋﻠ م ﻓﻲ‬ ‫وﻫﻲ ﺟزء ﻣ ﻣ ﻠ ﺎت ﻧ ﻞ درﺟﺔ اﻟ ﺎﺟ‬
‫اﻟﻬ دﺳﺔ اﻟ دﻧ ﺔ )إﻧ ﺎءات(‬
‫ﻣ ﻗﻞ‬
‫اﻟﻐراو‬ ‫ﻣﻬ د إﺑراﻫ ﻣ د ﺣ‬
‫) ﺎﻟ ر س ‪(2000‬‬
‫ﺻﻔر ‪1425‬ﻫـ‬ ‫ﻧ ﺎن ‪2004‬م‬
‫‪View publication stats‬‬

PT Stiffness Matrix PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

PT Stiffness Matrix PDF

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

FINITE ELEMENT ANALYSIS OF POST-TENSIONED CONCRETE BOX GIRDERS

Thesis · May 2004

CONCRETE BOX GIRDERS View project

A Framework for Automated Lock-In Thermography Data Processing View project

The user has requested enhancement of the downloaded file.

FINITE ELEMENT ANALYSIS OF POST-TENSIONED

APRIL 2004 MUHARRAM 1425

   

‫ﻴﻢ‬ ِ ‫اﻪﻠﻟ ا ْﻟﻌﻠِﻲﱡ ا ْﻟﻌ‬

Dr. Abdul Muttalib I. S. Al-Musawi, Head,

We certify that we have read this thesis titled “FINITE ELEMENT

Approved by the Dean of College of Engineering

The behavior of prestressed concrete box-girder bridges has been studied

List of figures VII

Chapter One: INTRODUCTION

Chapter Two: REVIEW OF LITERATURE

Chapter Three: THE FINITE ELEMENT FORMULATION AND

Chapter Four: MODELING OF MATERIAL PROPERTIES

Chapter Five: FORMULATION OF PRESTRESSING

Chapter Six: APPLICATIONS AND RESULTS

Chapter Seven: CONCLUSIONS AND RECOMMENDATIONS

3.1 Shape functions for the 20-node hexahedral 26

3.2 Weight and location of sampling points in the 27-point rule 35

6.1 Material properties of the one-cell box-girder bridge 76

6.3 Variation longitudinal stresses at quarterspan for Pu/P=0.250 and 0.875

6.5 Variation of longitudinal stresses at midspan for P/Pu=0.182 and 0.727

6.9 Variation of longitudinal stresses for cross-section at midspan (MPa)

1.2 Three-cell girder being discretized into finite element 5

2.1 Curved box-girder of Aneja and Roll 12

2.2 Skew ended element used by Scordelis 15

2.5 Typical shell element used in the analysis 18

2.6 Typical 8-noded, 6 D.O.F. layered isoparametric flat shell element 19

2.7 Degenerated Mindlin-type thick curved shell element 20

3.1 20-Noded isoparametric brick element 25

3.2 Alternative representation of reinforcement 30

3.3 Representative bond stress-slip curves 33

3.6 Basic techniques for solving nonlinear equations 38

3.7 Standard and modified Newton-Raphson methods 40

4.1 Uniaxial stress-strain curve for concrete 45

4.2 Idealization of discrete cracking model 53

4.3 Idealization of smeared cracking model 53

4.4 Failure cracking for triaxially loaded concrete 57

4.5 Post-cracking for concrete in tension 58

4.6 Shear retention model for concrete 60

5.1 Prestress forces losses due to anchor slip 64

5.2 Typical segment of prestressing tendon traversing a brick element 66

5.3 Analysis of curved cable 70

6.1 Structural details of the bridge analyzed for one-cell box-girder 74

6.16 Variation of transverse stresses in the outer surface of the web 88

6.17 Variation of longitudinal stresses in the inner surface of the web 89

6.18 Variation of transverse stresses in the inner surface of the web 90

6.19 Structural details of the bridge analyzed for two-cell box-girder 92

6.43 Deflection at cross-section for midspan for Jirousek 117

6.44 Longitudinal stresses at cross-section for midspan for Jirousek 117

[A]T,{a}T Transpose of matrix [A] and vector {a}

[A]-1 Inverse of matrix [A]

Aps Area of prestressing steel

As Area of reinforcing steel

Ec Modulus of elasticity of concrete

Eps Modulus of elasticity of prestressing steel

Es Modulus of elasticity of reinforcing concrete

fc' Uniaxial compressive strength of concrete