Salman Dissertation 2018

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 188

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

net/publication/344991560

Parametric Investigation of Integral Abutment Bridges: Behavior and Pile


Buckling Analysis

Thesis · May 2018


DOI: 10.13140/RG.2.2.10938.21444

CITATIONS READS
0 558

1 author:

Nassr Salman
University Of Kufa
7 PUBLICATIONS   11 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Nassr Salman on 30 October 2020.

The user has requested enhancement of the downloaded file.


Parametric Investigation of Integral Abutment Bridges: Behavior and Pile Buckling
Analysis

BY

NASSR NOORI IBRAHIM SALMAN


B.S., University of Kufa, Iraq, 1998
M.S., University of Technology, Iraq, 2001

THESIS
Submitted as partial fulfillment of the requirements
for the degree of Doctor of Philosophy in Civil Engineering
in the Graduate College of the
University of Illinois at Chicago, 2018

Chicago, Illinois

Defense Committee:

Mohsen Issa, Chair and Advisor


Ahmed Shabana, Mechanical and Industrial Engineering
Craig Foster
Krishna Reddy
Eduard Karpov
ACKNOWLEDGEMENTS

I would to thank the Ministry of Higher Education and Scientific Research in Iraq for providing

the financial support for my PhD study in the United States according to a scholarship program.

I would like to express my deepest gratitude to my advisor, professor Mohsen Issa, for his

continuing support and guidance throughout research stages, which without him this work couldn’t

be accomplished.

Special thanks are due to committee members, Alexander Chudnovsky, Craig Foster, Eduard

Karpov, Krishna Reddy and Ahmad Shabana for their valuable suggestions about thesis subject

and teaching throughout PhD program.

I am deeply grateful to my wife, Fatimah, for always being there for me. I thank my brothers

Mohammed and Firas for their support and handling many issues in Iraq while I was away.

I am also very grateful to my friend Timothy Doane from University of California, Davis for

reviewing the manuscript of this thesis and his valuable suggestions. His efforts are greatly

acknowledged.

Thanks to all my colleagues who in one way or another helped me in accomplishing this work.

ii
ABSTRACT

Integral Abutment bridges (IABs) are special types of bridges where expansion joints in the

superstructure are eliminated and the thermally induced lateral demand is transferred to the

supporting substructure system. As such, the substructure system moves back and forth following

seasonal bridge expansion and contraction. These bridges have gained a wide popularity and have

become a preferred choice among Department of Transportations and design offices around the

world because they have numerous advantages over the conventional bridges, including low

construction and maintenance cost, longer serviceability and higher stability, improved riding

quality and better seismic performance. Although of the proven advantages of IABs, there are no

uniform national guidelines for designing or constructing these structures, and each US state has

its own design limitations based on experience and performance of the previously constructed

bridges. Absence of the design guidelines is attributed to their complex behavior which is not fully

understood. The current study is two-fold; the first part involves a calibration process for an

instrumented bridge using a three-dimensional finite element (FE) model. The abutment and pile

displacements were calibrated with their experimental counterparts. Several shrinkage models and

temperature gradient scenarios were examined to predict the most representative parameters in

simulating realistic behavior. Based on the calibration process, a parametric study was conducted

to investigate the effect of bridge length, pile size and orientation, and type and stiffness of the soil

around the pile on the critical bridge responses which include: abutment displacement, pile

displacement, and deck and girder stresses. The second part addresses pile buckling under

combined effect of axial load and lateral cyclic displacement. Eleven detailed nonlinear finite

element models, experimentally validated, were established for steel HP sections with two axis

orientations to estimate the displacement capacities of the piles supporting IABs. A coupon test-

iii
validated cyclic plasticity model is incorporated in the finite element analysis to capture the

hysteresis of the steel behavior under cyclic loading. Displacement capacities are also compared

with an available analytical method. Length limits for IABs were estimated based on displacement

capacities of the HP sections and are compared with the limitations of the US Department of

Transportations’ current practice. Design recommendations for IABs are also presented.

iv
1. TABLE OF CONTENTS

CHAPTER
1. INTRODUCTION ...................................................................................................................... 1
1.1 General ................................................................................................................................... 1
1.2 Problem Statement ................................................................................................................ 3
1.3 Objective of the Study ........................................................................................................... 4
1.4 Structure of the Study ............................................................................................................ 5
2. LITERATURE REVIEW ........................................................................................................... 7
2.1 Introduction ........................................................................................................................... 7
2.2 Field Monitoring..................................................................................................................... 8
2.3 Finite Element Modelling ..................................................................................................... 13
2.4 Soil Modelling Methods ....................................................................................................... 14
2.5 Parametric Studies ............................................................................................................... 17
2.6 Stability Requirements and Fatigue ..................................................................................... 18
2.7 Bridge Geometric Limitations .............................................................................................. 19
3. SOIL-STRUCTURE INTERACTION ..................................................................................... 23
3.1 Introduction ......................................................................................................................... 23
3.2 Soil-Pile Interaction .............................................................................................................. 24
3.2.1 p-y Curves ..................................................................................................................... 25
3.2.1.1 p-y Method for Clay .............................................................................................. 25
3.2.1.1.1 Soft Clay Having Free Water: Static Loading Case ......................................... 25
3.2.1.1.2 Soft Clay Having Free Water: Cyclic Loading Case ......................................... 28
3.2.1.1.3 Stiff Clay Having Free Water: Static Loading Case ......................................... 29
3.2.1.1.4 Stiff Clay in the Presence of Free Water: Cyclic Loading ............................... 33
3.2.1.1.5 Stiff Clay with no Free Water: Static Loading ................................................ 34
3.2.1.1.6 Stiff Clay with no Free Water: Cyclic Loading ................................................ 35
3.2.1.2 p-y Curves for Sand ............................................................................................... 36
3.2.1.3 p-y Curves for Soil Having Cohesion and Angle of Friction................................... 41
3.2.2 Program LPILE ............................................................................................................... 45
3.3 Abutment-Backfill Interaction ............................................................................................. 46

v
4. CALIBRATION OF THE NUMERICAL MODEL................................................................. 52
4.1 Introduction ......................................................................................................................... 52
4.2 Bridge Description................................................................................................................ 53
4.3 Finite Element Modelling ..................................................................................................... 55
4.4 Modelling Soil-Pile Interaction............................................................................................. 56
4.5 Modelling Soil-Abutment Interaction .................................................................................. 58
4.6 Pile Finite Element Model .................................................................................................... 60
4.7 Finite Element Model of the Bridge ..................................................................................... 70
4.8 Calibration Matrix ................................................................................................................ 74
4.9 Results .................................................................................................................................. 77
4.9.1 Log-Spiral Theory .......................................................................................................... 78
4.9.2 Rankine Case ................................................................................................................. 82
4.10 Summary and Conclusions ................................................................................................. 88
5. PARAMETRIC STUDY .......................................................................................................... 91
5.1 Introduction ......................................................................................................................... 91
5.2 Parametric Study Base Case................................................................................................. 92
5.3 Selection of Parameters ....................................................................................................... 95
5.4 Soil Properties Used in the Parametric Study ...................................................................... 96
5.5 Finite Element Model Setup ................................................................................................ 98
5.6 Effect of Bridge Length....................................................................................................... 100
5.7 Effect of Bridge Length on Pile Stresses and Displacements Profiles ................................ 112
5.8 Effect of Soil Type on Pile Stresses and Displacements Profiles........................................ 114
5.9 Effect of Pile Size on Pile Stresses and Displacements Profiles ......................................... 116
5.10 Conclusions ...................................................................................................................... 118
6. PILE BUCKLING ANALYSIS OF HP PILES IN INTEGRAL ABUTMENT BRIDGES .. 120
6.1 Introduction ....................................................................................................................... 120
6.2 Finite Element Model Setup .............................................................................................. 121
6.3 Validation of the FE model................................................................................................. 124
6.4 Pile selection, Boundary Conditions and Loading History ................................................. 133
6.5 Soil-Pile Interaction ............................................................................................................ 137
6.6 Buckling Failure Criterion ................................................................................................... 138

vi
6.7 Results ................................................................................................................................ 139
6.8 Effect of Soil Types Surrounding Piles................................................................................ 146
6.9 Comparison with Abendroth and Greimann (1989) Method ............................................ 148
6.10 Integral Abutment Bridge Length .................................................................................... 152
6.11 Summary and Conclusions ............................................................................................... 154
7. Conclusions and Recommendations ....................................................................................... 155
7.1 Conclusions ........................................................................................................................ 155
7.2 Recommendations for Future Work .................................................................................. 157

vii
LIST OF TABLES

TABLE 2.1 BRIGE DETAILS MONITORED BY Frosch et al. (2004, 2011) ........................... 10
TABLE 2.2 MAXIMUM LENGTH LIMITS FOR STEEL AND CONCRETE IABS IN CLAY
(Dicleli and Albhaisi, 2003) .......................................................................................................... 19
TABLE 2.3 SUMMARY OF IAB RESPONSES (ADAPTED FROM KUNIN AND
ALLAMPALI, 2000) .................................................................................................................... 21
TABLE 2.4 BUCKLING DISPLACEMENT CAPACITIES OF HP SECTIONS (Frosch et al.,
2004) ............................................................................................................................................. 22
TABLE 3.1 TYPICAL VALUES OF 50 FOR NORMALLY CONSOLIDATED CLAY (Peck et
al., 1974) ....................................................................................................................................... 27
TABLE 3.2 TYPICAL VALUES OF 𝑘𝑠 ...................................................................................... 30
TABLE 3.3 TYPICAL VALUES OF 50 FOR OVERCONSOLIDATED CLAY (Skempton,
1951) ............................................................................................................................................. 31
TABLE 3.4 REPRESENTATIVE VALUS FOR 𝑘𝑝𝑦 FOR SAND BELOW WATER TABLE
(SUBMERGED) ........................................................................................................................... 39
TABLE 3.5 REPRESENTATIVE VALUES FOR 𝑘𝑝𝑦 FOR SAND ABOVE WATER TABLE
....................................................................................................................................................... 39
TABLE 3.6 APPROXIMATE WALL MOVEMENT RRQUIRED TO REACH MAXIMUM
AND MINUMUM EARTH PRESSURE (Clough and Duncan, 1991) ....................................... 49
TABLE 4.1 PILE PROPERTIES OF BRIDGE SR-18 ................................................................ 62
TABLE 4.2 SOIL BORING FOR BENT 1 OF SR-18 (Frosch et al., 2011)................................ 62
TABLE 4.3 SOIL BORING DATA FOR BENT 6 OF SR-18 (Frosch et al., 2011) .................. 63
TABLE 4.4 CLASSIFICATION OF SOIL CONSISTENCY (Prakash and Sharma, 1990) ....... 65
TABLE 4.5 SPT-N VALUES VS ANGLE OF INTERNAL FRICTION FOR COHESIONLESS
SOILS (Fang, 1991) ...................................................................................................................... 65
TABLE 4.6 DRY UNIT WEIGHT FOR DIFFERENT SOIL TYPES (Das, 2011)..................... 66
TABLE 4.7 SOIL PARAMETERS USED IN LPILE MODELS ................................................ 67
TABLE 4.8 CALBRATION MATRIX ....................................................................................... 75
TABLE 5.1 PennDOT 28/78 PRESTRESSED GIRDERS DIMENSIONS (PennDOT) (ALL
DIMENSIONS ARE IN INCHES) ............................................................................................... 94
TABLE 5.2 PARAMERTERS USED IN THE STUDY .............................................................. 95
TABLE 5.3 HP SECTIONS PROPERTIES ................................................................................. 96
TABLE 5.4 SOIL PROPERTIES AROUND THE PILES ........................................................... 97
TABLE 5.5 NORMALIZED DISPLACEMENTS VERSUS BRIDGE LENGTHD ................ 105
TABLE 6.1 DISPLACEMENT PROTOCOL USED IN THE TEST BY Frosch et al. (2004) . 127
TABLE 6.2 HP SECTIONS CONSIDERED IN THE PARAMETRIC STUDY ...................... 134
TABLE 6.3 DISPLCEMENT CAPACITIES OF A SET OF HP SECTIONS EMBEDDED IN A
MEDIUM SAND ........................................................................................................................ 139
TABLE 6.4 FE vs Frosch et al. (2004, 2011) BUCKLING CAPACITES ................................. 145
TABLE 6.5 SAND PROPERTIES USED IN THE STUDY ..................................................... 147

viii
TABLE 6.6 CLAY PROPERTIES USED IN THE STUDY ..................................................... 147
TABLE 6.7 BUCKLING DISPLACEMENT COMPARISONS FOR VARIOUS SOIL TYPES
..................................................................................................................................................... 148
TABLE 6.8 COMPARISON BETWEEN ABENDROTH ET AL. (1989) AND FEM ............. 151
TABLE 6.9 MAXIMUM LENGTH OF INTEGRAL ABUTMENT BRIDGES....................... 153

ix
LIST OF FIGURES

Figure 1.1 Typical ordinary and IAB .............................................................................................. 2


Figure 2.1 Abutment displacement before and after traffic opening (Kim and Laman, 2012) ...... 9
Figure 2.2 Ratcheting phenomenon for the IAB (Frosch et al., 2011) ........................................ 12
Figure 3.1 Set of p-y curves along the pile depth (Reese and Imple, 2001) ................................. 25
Figure 3.2 Schematic representation of p-y curves for soft clay with free water under static load
case (Matlock, 1970) ..................................................................................................................... 27
Figure 3.3 P-y curve for soft clay with free water under cyclic loading (Matlock, 1970) ........... 29
Figure 3.4 p-y method for stiff clay under static loading in the presence of free water ............... 30
Figure 3.5 Graphical values of A (subscripts refer to static and cyclic loading). (Reese and Imple,
2001) ............................................................................................................................................. 32
Figure 3.6 p-y method for stiff clay under cyclic loading in the presence of free water (Reese et
al. 1975) ........................................................................................................................................ 34
Figure 3.7 p-y method for stiff clay under static loading with no free water (Welch and Reese,
1972) ............................................................................................................................................. 35
Figure 3.8 p-y curves for stiff clay with no free water under cyclic loading case (Reese and
Imple, 2001) .................................................................................................................................. 36
Figure 3.9 Coefficients of As and Ac ............................................................................................ 38
Figure 3.10 Nondimensional soil resistance coefficients with depth ........................................... 38
Figure 3.11 Computation of p-y curve for sand (Reese et al., 1974) ............................................ 40
Figure 3.12 coefficient Kc ............................................................................................................. 43
Figure 3.13 coefficient K ............................................................................................................. 44
Figure 3.14 proposed p-y curve for c- soil (Reese and Impe, 2001) .......................................... 45
Figure 3.15 Lateral earth pressure cases ....................................................................................... 48
Figure 3.16 Wall movement versus earth pressure for ideal conditions (Fang, 1991) ................. 50
Figure 3.17 Wall movement versus earth pressure for compacted backfill (Fang, 1991) ............ 50
Figure 4.1 SR-18 over Missessenewa River, IN (Lovell, 2010) ................................................... 54
Figure 4.2 Plan and cross section of bridge SR-18 over Mississinewa River .............................. 54
Figure 4.3 Typical details of abutment of SR-18 (Lovell, 2010).................................................. 55
Figure 4.4 Clough and Duncan digitized curves ........................................................................... 59
Figure 4.5 pile distribution of bridge SR-18 (Frosch et al., 2004)................................................ 61
Figure 4.6 spring tributary area for the pile .................................................................................. 64
Figure 4.7 p-y curves for medium-Sand ....................................................................................... 68
Figure 4.8 Displacement along pile depth for different soil types ................................................ 68
Figure 4.9 ABAQUS vs LPILE computed displacements ............................................................ 70
Figure 4.10 Bridge FE model........................................................................................................ 71
Figure 4.11 Shrinkage strain of the bridge deck ........................................................................... 72
Figure 4.12 Force-displacement curves for SR-18 Backfill (Rankine with =35o)...................... 73
Figure 4.13 Flow chart of the FE modelling ................................................................................. 76
Figure 4.14 Location of the convergence meter (Frosch et al., 2010) .......................................... 77
Figure 4.15 Case (1) TD, TG, SD, SG, Log-spiral theory ............................................................ 79

x
Figure 4.16 Case (2) TD, TG, SD, Log-spiral theory ................................................................... 79
Figure 4.17 Case (3) TD, SD plus SG (CEB-90-90), Log-spiral theory ...................................... 80
Figure 4.18 Case (4) TD, SD (CEB-90-90), Log-spiral theory .................................................... 81
Figure 4.19 Case (5) TD plus 50% TG, SD plus SG (CEB-90-90), Log-spiral theory ................ 82
Figure 4.20 Case (6) TD, TG, SD plus SG (CEB-90-90), Rankine theory .................................. 83
Figure 4.21 Case (7) TD plus 50% TG, SD plus SG (CEB-90-90), Rankine theory ................... 84
Figure 4.22 Case (8) TD plus 50% TG, SD (CEB-90-90), Rankine theory ................................. 85
Figure 4.23 Case (9) TD plus 50% TG, SD (CEB-90-90), Rankine theory ................................. 86
Figure 4.24 Case (10) TD plus 50% TG, SD (ACI-209), Rankine theory ................................... 86
Figure 4.25 Calculated vs Measured earth pressure behind abutment.......................................... 88
Figure 5.1 Bridge 211 in central Pennsylvania (Kim and Laman, 2012) ..................................... 92
Figure 5.2 Bridge 211 a. bridge layout b. section through bridge ................................................ 93
Figure 5.3 PennDOT 28/78 prestressed concrete girder section (PennDOT) ............................... 94
Figure 5.4 Load-displacement of the backfill soil ........................................................................ 97
Figure 5.5 Typical bridge FE model used in the parametric study ............................................... 99
Figure 5.6 substructure response considering various bridge span lengths. ............................... 101
Figure 5.7 Abutment displacement vs bridge length for different pile sections and orientations
..................................................................................................................................................... 103
Figure 5.8 Abutment displacement vs bridge length for HP10x57 and weak axis orientation
(bridge expansion) ...................................................................................................................... 103
Figure 5.9 Abutment displacement vs bridge length for HP10x57 and weak axis orientation
(bridge contraction) ..................................................................................................................... 104
Figure 5.10 Normalized bridge displacement versus bridge length (bridge expansion) ............ 105
Figure 5.11 proposed equation compared to finite element model results ................................. 106
Figure 5.12 Max pile displacement vs bridge length for various pile sections and orientations
(bridge expansion) ...................................................................................................................... 107
Figure 5.13 Max pile displacement vs bridge length for HP10x57 and weak axis orientation
(bridge expansion) ...................................................................................................................... 108
Figure 5.14 Max pile axial stress vs bridge length for various pile sections and orientations
(bridge expansion) ...................................................................................................................... 109
Figure 5.15 Max pile axial stress vs bridge length for HP10x57 (bridge expansion) ................ 110
Figure 5.16 Max deck axial stress vs bridge length for HP10x57 and weak axis orientation
(bridge expansion) ...................................................................................................................... 111
Figure 5.17 Maximum girder axial stress vs bridge length for HP10x57 and weak axis orientation
(bridge expansion) ...................................................................................................................... 112
Figure 5.18 Displacement profile for H10x57 and medium sand for various bridge lengths and
pile orientations ........................................................................................................................... 113
Figure 5.19 Von-Mises stress profile for H10x57 and medium sand for various bridge lengths
and pile orientations .................................................................................................................... 114
Figure 5.20 Displacement profile for H10x57 for various soil types and weak axis orientation
(Bridge length = 200 ft) .............................................................................................................. 115
Figure 5.21 Max Von-Mises stress profile for H10x57 for various soil types and weak axis
orientation (Bridge length = 200 ft) ............................................................................................ 115

xi
Figure 5.22 Displacement profile for H10x57 for soft clay and weak axis orientation.............. 117
Figure 5.23 Von-Mises stress profile for H10x57 for soft clay and weak axis orientation ........ 117
Figure 6.1 One-dimensional representation of the nonlinear model (ABAQUS manual) .......... 122
Figure 6.2 Three-dimensional representation of the nonlinear model. (ABAQUS manual) ...... 123
Figure 6.3 Calibration of the of the material cyclic response ..................................................... 124
Figure 6.4 Schematic diagram of the experimental testing (Frosch et al., 2011) ....................... 125
Figure 6.5 Typical Pile model in IAB ......................................................................................... 125
Figure 6.6 Typical FE model of the test setup ............................................................................ 128
Figure 6.7 HP10X42 response: Load-deflection hysteresis at column top ................................. 129
Figure 6.8 HP10X42 response: Lateral load fluctuations along step 2....................................... 130
Figure 6.9 HP10X42 response: Buckling of column in the experimental testing ...................... 131
Figure 6.10 HP10X42 response: FE buckling mode .................................................................. 131
Figure 6.11 HP12X53 response: Load-deflection hysteresis at column top ............................... 132
Figure 6.12 HP12X53 response: Lateral load fluctuations along step 2..................................... 133
Figure 6.13 Typical pile idealization in parametric studies: weak axis ...................................... 135
Figure 6.14 Typical pile idealization in parametric studies: strong axis .................................... 136
Figure 6.15 Load-deflection relationship for HP12x63 weak axis ............................................. 141
Figure 6.16 Lateral load fluctuation with time for HP12x63 weak axis ..................................... 142
Figure 6.17 variation of strain with step time for HP 12x63 at buckling displacement ............. 143
Figure 6.18 Buckling ductility ratios for HP sections in medium sand ...................................... 144
Figure 6.19 Typical cantilever pile model in IABs (Frosch et al., 2004) ................................... 145
Figure 6.20 Lateral displacement profiles of HP sections at buckling ....................................... 150

xii
1. INTRODUCTION

1.1 General

Integral abutment bridges (IABs), or jointless bridges, are single or continuous multiple span

bridges that have no expansion joints in the superstructure or bearing system in the support. The

girders and deck slab are cast integrally with abutments at the end spans. Figure 1.1 shows a typical

ordinary bridge and an IAB with the obvious distinctions in the end connection. The main

difference between conventional bridges and the IAB is that the thermally induced lateral demand

and time-dependent effects (creep and shrinkage) in the conventional bridges are accommodated

by the expansion joints in the superstructure while in case of IABs, this demand is carried by the

abutment-pile system at the bridge ends. The piles in the substructures of IABs are often designed

as a single row of HP steel sections orientated to the weak axis to increase the system flexibility.

This system moves back and forth following superstructure expansion and contraction due to

seasonal and daily temperature variations. This cyclic movement induces stresses in the

superstructure and results in pile hinging at the pile-abutment interface which must be considered

in the pile design. Furthermore, the pile in IABs is vulnerable to local flange buckling over

continuous cycling process, which leads to a substantial decrease in the axial load carrying capacity

of the pile because of the strength degradation. However, IABs provide many advantages over

conventional jointed bridges, including eliminating the initial construction and life cycle

maintenance cost of expansion joints, reducing the deterioration of substructure elements by

preventing leaking water and deicing chemicals from defective joints, in addition to providing

better riding quality. Some studies have demonstrated their high seismic performance as well

(Frosch et al, 2009; Burke, 1993). Defective expansion joints and other underlying affected

1
substructure components cost millions of dollars in maintenance each year (Alampalli and

Yannotti, 1998). The cost of direct and indirect damage of the bridges due to corrosion was

estimated to be $80 billion in the year 1998 (Koch et al. 2001). For that reason, IABs have become

the preferred alternative for many bridge owners and designers. Kunin and Alampalli (2000)

reported in a survey that more than 30 US states and Canadian provinces stated that IABs perform

as well as or even better than conventional bridges. However, there are many new challenges

associated with the new technology. The system has a complex behavior and there are no stationary

provisions for designing or constructing these types of structures. Because of the limited

knowledge about the behavior of IABs, design procedures are either conservative or based on the

performance of previously constructed bridges. Researchers have been working since the early

1980s to develop, enhance, and refine their design guidelines. It is expected that these bridges will

soon replace conventional bridges for short and medium length spans.

Deck Approach Slab Deck Approach Slab

Integral Abutment Integral Abutment


Girder Girder

Expansion Joint
Pile Pile

Ordinary Bridges Integral Abutment bridges

Figure 1.1 Typical ordinary and IAB

2
1.2 Problem Statement

Despite the proven advantages of IABs over conventional jointed bridges which are reflected in

terms of cost, structural performance and long-life serviceability, there are still many controversies

surrounding design criteria of these structures, including the maximum safe length of the bridge,

size and orientation of the piles, the soil-structure interaction, and time-dependent effects. There

is no consensus among states’ departments of transportations over the design guidelines of IABs,

and construction practices vary widely from one state to another. Each US state has its own

geometric limitations built on past practical experience and performance of previously constructed

bridges, rather than implementing a rigorous scientific approach (Greimann et. al., 1986). Even

the AASHTO specifications do not contain guidelines to design or construct IABs. The design of

IABs should ideally consider both flexibility and strength capacity of the substructure so that

neither excessive stresses are induced in the superstructure due to end restraint of the substructure

and adjacent soil, nor failure occurs in the piles due to the lateral cyclic demand combined with

axial loads imposed by the built-in superstructure. This balanced design method requires extensive

and comprehensive investigation to capture the exact behavior of the critical parameters affecting

the bridge response. Furthermore, the response of the piles under prescribed loading conditions

must be carefully addressed to set the safe length of IABs.

3
1.3 Objective of the Study

The objective of the current study is twofold; the first part involves predicting the parameters

affecting bridge behavior. These parameters include the potential loading and boundary conditions

that govern the bridge response. This entails calibrating a three-dimensional finite element based

on numerical models with an experimentally tested bridge. Based on the outcomes of the

calibration process, a parametric study is presented which investigates the effect of various bridge

parameters on bridge critical response. The second part addresses HP-pile buckling by developing

a detailed three-dimensional finite element model to investigate displacement capacities of IAB.

Maximum length of IABs is estimated based on the displacement capacities of HP piles. To that

end, the following tasks have been performed:

1- Reviewing the available analytical procedures to simulate IABs

2- Calibrating the bridge response using three-dimensional finite element models against

experimentally monitored data.

3- Conducting a parametric study to examine the effect of various bridge parameters and

boundary conditions on the critical response.

4- Conducting a pile buckling analysis to predict the displacement capacities for HP sections

in IABs which are then used in estimating maximum length of IAB.

4
1.4 Structure of the Study

This dissertation is structured as follows:

Chapter 1: Introduction:

This chapter provide an overview of integral abutment bridges, their advantages and their

associated problems.

Chapter 2: Literature Review

This chapter surveys the available studies conducted on monitoring, modelling, and assumptions

regarding soil structure interaction and finite element modelling techniques.

Chapter 3: Soil-structure Interaction

This chapter present soil-structure interaction for IABs. This includes:

- Nonlinear soil- pile interaction (p-y method).

- Nonlinear soil-abutment interaction (lateral earth pressure).

Chapter 4: Calibration of the numerical model

This chapter involves three-dimensional finite element modelling of IAB using the commercial

package Abaqus. The calculated results are compared to the monitored data and the most

representative loading and boundary conditions are obtained.

Chapter 5: Parametric Study:

In this chapter, a parametric study is conducted to study the effect of the bridge geometric and

boundary conditions on the design criteria of IABs. These parameters include:

5
- Bridge length.

- Pile type and orientation.

- Soil type and stiffness around the pile.

Chapter 6: Pile buckling analysis of HP piles in Integral Abutment Bridges:

This chapter incorporates three-dimensional finite element models for HP piles embedded

in soil medium and subjected to combined axial load and lateral cyclic displacement. The

study aims at predicting displacement capacities of HP sections to set the maximum

length limits of IABs.

References

Appendices

6
2. LITERATURE REVIEW

2.1 Introduction

Research on integral abutment bridges dates to the past three decades when early attempts were

begun to set rational analysis and design methods for this type of structure (Abendroth and

Greimann, 1989). Since then, successive studies have followed to obtain more insight about their

complex behavior and to establish guidelines for designing and constructing them. Research is

basically focused on four major approaches:

1. Field monitoring of newly constructed or in-service bridges.

2. Finite element simulations for full or partial bridge elements.

3. Parametric investigation.

4. Displacement capacity of the HP piles supporting abutments.

In this chapter, a survey is presented to review the available work conducted considering these

four approaches. A historical sequence will be considered.

7
2.2 Field Monitoring

Field monitoring programs of integral abutment bridges were generally launched in early 1990s.

Studies comprised of instrumenting various locations in bridge superstructure and/or substructure

elements. The bridge monitoring process is intended to help understand the behavior of full-scale

bridges and to calibrate simulated numerical models. Major efforts focused on the soil-pile and

soil-abutment interaction due to cyclic thermal variations. Monitoring periods ranged from short

period programs to comprehensive long-term projects. Girton et al. (1991) monitored two bridges

in Iowa to verify available design procedures for piles. The first one was a 324.5 ft. long,

prestressed girder skewed bridge, having an angle of 45o skew. The second one was a 320 ft. long

composite bridge, with a 30o skew angle. The monitoring period spanned over two years, from

January 1987 to February 1989. Measurements included bridge temperature, bridge displacements,

pile strains, and coefficient of thermal expansion which were based on concrete core samples.

Simplified analytical models for the bridge and the piles were also presented and compared to the

experimental data.

Lawver et al. (2000) conducted a field experimental program for an IAB in Minnesota. The bridge

was 216.6 ft. long and composed of three-span prestressed girders supported by six HP 12x53 piles

on each side, oriented along the weak axis. The bridge was monitored for seven years to measure

the horizontal movement and rotation of the abutment, abutment pile strain, earth pressure behind

abutments, prestressed girders strains, deck strains, and girder displacements. Weather conditions

like air temperature and solar radiations were also recorded. The study incorporated two live load

tests of the bridge to understand the performance of the bridge under live load effects. It was

8
revealed that the bridge builds up an inward abutment movement over time which results in a

continued overall shortening of the bridge.

Abendroth and Greimann (2005) conducted a field instrumentation program for two bridges in the

state of Iowa. The measurements included longitudinal and lateral displacement of the abutment,

abutment rotation, and pile and girder strain, in addition to ambient temperature.

Kim and Laman (2012) monitored four integral abutment bridges in central Pennsylvania. Bridges

selected were short to medium in length. The instrumentation plan started in 2002 and continued

for 7 years. Measurements included abutment displacement, abutment backfill pressure, abutment

rotation, girder rotation, girder bending moment, girder axial force, pile bending moment, pile

axial force, and strain in approach slab. All responses were reported with and without traffic

loading. The study revealed that traffic loading has insignificant effects on all bridge responses

as shown Figure 2.1.

Figure 2.1 Abutment displacement before and after traffic opening (Kim and Laman, 2012)

9
Frosch et al. (2004, 2011) conducted a comprehensive program to study the behavior of IABs in

early 2000s. Study incorporated field monitoring and analytical study for three bridges in Indiana.

These bridges were as follows: bridge I-65 having a total length of (152 ft.), bridge SR-18 having

a total length of (367 ft.), and bridge US-231 with a total length of (221 ft.). Dimensions and details

of the bridges is tabulated in Table 2.1. Bridges were highly instrumented to measure multiple

responses including abutment displacement, abutment rotation, earth pressure behind abutment,

and transverse abutment movement. A unique feature in the instrumentation in this program is that

in bridge SR-18, pile curvature had been measured, which later led to predict lateral deflection

along pile depth. This work has never been reported in any of the monitoring programs ever

performed. Moreover, it was a very significant step towards understanding the behavior of piles in

the IABs, as it provides an insight about how the pile responds to the bridge thermal movement.

For this feature, and because of the availability of bridge plans, this bride (SR18) has been

considered for the calibration process in this dissertation.

TABLE 2.1 BRIGE DETAILS MONITORED BY Frosch et al. (2004, 2011)

Southbound I-65 SR-18 US-232


Total Length (ft.) 152 367 221
Skew Angle (Deg.) 25 8 33.8
Girder Type W36x150 Prestressed Concrete Prestressed Concrete

Pile Type HP12x53 / CFT 14” CFT 14” CFT 14”

Date Instrumented Summer 2000 Summer 2003 Fall 2006

10
Frosch et al. (2011) reported that there is a continuing inward shortening of the IAB, which was

noted at bridge SR-18 as shown in Figure 2.2. This was observed in the monitoring process which

extended for seven years. Lawver et al. (2000) reported the same response in a monitored bridge

in Minnesota. However, none of the investigators fully justified this behavior. Lawver et al. (2000)

attributed it to soil compaction behind the abutment and build-up of debris at the end joints. Frosch

et al. (2011) attributed it to the accumulated concrete shrinkage strain over time. Frosch et al.

(2011) had also conducted field experimental tests on a single-span, quarter-scale IA bridge to

investigate the effect of high skew angle of IAB.

11
Figure 2.2 Ratcheting phenomenon for the IAB (Frosch et al., 2011)

12
2.3 Finite Element Modelling

Researchers have employed several techniques to model the IAB structural elements. In general,

the finite element (FE) method in 2D frame and three-dimensional space frame has been widely

used. A 2D FE model is commonly employed because of its simplicity and computational time

savings. It has been used by Diclelei and Albhaisi (2003); Fennema et al. (2005); Civjan et al

(2007); Pugasap et al. (2009); and Kim and Laman (2010). The 2D methodology involves selecting

a tributary area of the bridge deck and the abutment equal to the spacing between girders (Dicleli

and Albhaisi, 2003). Three-dimensional space frame modelling has also been implemented by

many researchers: Faraji et al. (2001); Pugasap et al. (2009); Frosch et al. (2011), Albhaisi (2012)

and Quinn and Civjan (2016). Both methods are economic in terms of time, however, the three-

dimensional method is more accurate than 2D because it provides information about response in

the transverse direction.

Modelling of IAB parts generally involves simple one-dimensional or two-dimensional finite

elements. Girders, transverse diaphragms, and piles are typically modelled using beam/frame

elements. Deck slab and abutments are modeled using shell elements. Rigid links sometimes are

used between the girders and the slab to maintain full composite action between them. Fennema

et al. (2005) used STAAD pro to create a 2D and three-dimensional finite element frame models

for a three-span IAB in Pennsylvania. In the 2D frame model, the prestressed girders and the deck

slab were transformed into an equivalent beam element based on the properties of the girders and

slab. The abutments and piers are transformed into beam elements as well. Steel HP piles were

lumped into a single pile composed of multiple beam elements. In the three-dimensional frame

model, the girders, intermediate diaphragms, and the piles were modelled as beam elements. The

deck, abutments and approach slab were modelled using shell elements. Faraji et al. (2001); Huang

13
et al. (2005) and Albhaisi (2012) likewise used the same modelling techniques to represent girders,

deck, abutments and piles.

Several researchers have implemented three-dimensional finite element solid elements to model

the IABs. Although this method is rather expensive in terms of the model building and solving

time, it proved to be more accurate than the conventional 2D or three-dimensional space frame FE

models because it tracks the stress variation within elements. Khodair et al. (2004) built a FE model

of the substructure of an IAB in New Jersey using three-dimensional solid elements in the

commercial software ABAQUS. The piles and the surrounding soil were modelled as eight-node

linear brick solid elements (C3D8R). A surface-to-surface contact algorithm is used to simulate

the contact between the soil and the piles. The elastic-perfectly plastic material model is assumed

for the steel. The concrete material is assumed as elasto-plastic model while Morh-Coloumb failure

criterion is used for the soil.

Khasawneh (2014) has also used three-dimensional solid elements to model the IAB. The

superstructure of the bridge is modelled as a 4-node thermally coupled tetrahedral element with

coupled thermal/mechanical elastic material. Soil is modelled as a 4-node tetrahedral element with

elastoplastic material model. A user-defined subroutine is developed for the soil model and

calibrated to laboratory testing.

2.4 Soil Modelling Methods

The soil-structure interaction is the most challenging part in the analysis and design of any

structure adjacent to the soil. Uncertainties arise from the complex nature of soil behavior under

different loading conditions. The soil-structure interaction in IABs is a problem of laterally loaded

piles and abutment-backfill interaction. The p-y curve method is the most widely used procedure

14
to model the interaction between the piles and the surrounding soil. It correlates the lateral pressure

of the soil on the pile and the corresponding pile lateral displacement. The p-y curve method was

first introduced by Terzaghi (1952) to help model the soil-structure interaction as a linear soil

modulus along the pile depth. Successive studies tried to develop empirical formulas based on

calibrating the soil response with full-scale experimental tests. Abutment-backfill interaction is a

lateral earth pressure problem like the retaining wall case but with relative movement of the

abutments relative to the wall. Expansion and contraction of the IAB induce passive and active

earth pressure to the abutments. Modelling of the abutment-backfill interaction involves pressure-

displacement curves like p-y curves of the piles. Chapter 3 will present both soil-pile and soil-

abutment interaction procedures in detail.

Dicleli and Albhaisi (2003, 2004) used elasto-plastic linear soil springs around the piles to build

2D FE models which were employed to estimate the maximum length of IABs in sand and clay.

The slope of the elastic linear part of the spring behavior is represented by initial soil modulus,

while the plastic part is defined as the ultimate soil resistance.

Faraji et al. (2001) used nonlinear soil springs to model the backfill soil behind the abutment of

IAB and around the piles. For the backfill soil, they implemented the National Cooperative

Highways Research Program (NCHRP, 1991) design curves to build the force-deflection response

of the nonlinear springs. The American Petroleum Institute (API, 1993) procedure was used by the

researchers to build the p-y curves for the springs representing the soil around the pile.

Frosch et al. (2004) used linear soil springs in two perpendicular dimensions to model the soil

around the pile. The models were built using SAP2000 to calibrate monitored piles of IABs.

Frosch et al. (2011) represented the backfill soil behind abutments as nonlinear springs utilizing

15
Duncan and Mokwa’s (2001) method for idealizing the spring behavior. Pile springs were also

considered linear.

Modelling by the three-dimensional finite element method has also been employed in the soil-

structure interaction problem. The soil is simulated as two- or three-dimensional finite elements.

Isoparametric quadrilateral elements or triangular elements are used in the case of two-dimensional

plane stress or strain, and hexahedral (often called brick elements) or tetrahedral elements or a

combination of both is used in the case of three-dimensional analysis. The interface between the

soil and the pile is represented by thin or zero thickness interface elements (Desai et al., 1984)

which help in providing friction and slippage between the two surfaces in the case of compression

and deboning in the case of tension. The material behavior of the soil and the interface is the most

challenging and controversial part in the modelling process because no consensus exists about

their most satisfactory representation. Different types of soil have different material behaviors and

they are highly dependent on the structure and composition of the soil, type of loading, and water

condition. Different models have been presented in the literature to represent various types of soil.

Muqtadir and Desai (1986) used a nonlinear three-dimensional finite element model combined

with an elastic-plastic soil behavior in a pile-group analysis problem. The soil and pile were

idealized as 8-noded hexahedral elements and the interface elements between the soil and the piles

are assumed as 8-noded thin-layer element with normal and shear stiffness constitutive model.

Trochanis (1988) used ABAQUS software to model soil and pile as a solid 27-node quadratic

element, and a 9-node thin-layer interface element was used to model the surface between the pile

and soil. A modified Drucker-Prager nonlinear model was used in the model to represent the soil.

Other investigators who also discretized the soil mass as a continuum or solid elements include:

Brown and Shie (1990); Bhowmik (1992); Khohair (2004) and khasawneh (2014).

16
2.5 Parametric Studies

Parametric studies have been conducted by many researchers to investigate the effects of various

geometric and boundary conditions on the behavior of IA bridges. Huang et al. (2004) performed

a parametric study for a concrete IA bridge using three-dimensional finite element model utilizing

ANSYS software. They studied five variables derived from the base case (monitored bridge) which

was considered as well in the parametric study. The five variables were: abutment- pile cap

connection; pile type, size and orientation; girder dimensions; wing wall configuration; and soil

surrounding piles. The study indicated a necessity to balance the stresses between the piles and

superstructure in the case of long spans and stiff soil conditions. It also showed that the hinged

connection at the pile-abutment interface would reduce the stresses in girder by 30% compared to

the fixed connection. Faraji et al. (2001) studied the effect of the level of compaction of the soil

behind the abutment and around the piles. A three-dimensional FE model was built for a short-

span bridge in Massachusetts using software GTSTRUDL. They found that the level of

compaction of the backfill behind the abutment has a significant impact on the bridge response in

general, particularly on the axial forces and moments generated in the deck. Albhaisi et al. (2012)

investigated the effect of substructure stiffness on the abutment and pile displacement and on

girder and deck stresses by investigating four soil types surrounding the piles. Two in-service IABs

were modified to be applicable to the parametric study. The researchers found that the soil stiffness

has an influential part in governing bridge response under thermal loading. Frosch et al. (2011)

conducted a parametric investigation to examine the effect of length and skew of the IAB and their

influence on the pile lateral displacement demand. The investigators also studied the effect of other

parameters on the bridge response including: span length, shrinkage models, stiffness of the soil,

and pile size and orientation. The study showed that the ultimate lateral displacement exerted on

17
piles is 0.24 in. from 200 ft. bridge; 1.28 in. from 600 ft. bridge; and 2.34 in. from 1000 ft. bridge.

The study also revealed that if the shrinkage stain is added to the temperature strain, the lateral

displacement demand will be overpredicted by a constant factor of 0.6.

2.6 Stability Requirements and Fatigue

Integral abutment bridges are subjected to cyclic lateral movement resulting from seasonal and

daily temperature fluctuations. This lateral movement is translated to the supporting piles which

must be flexible enough to sustain such loading. On the other hand, piles should meet stability

requirements for global and local buckling resulting from the weight of the bridge combined with

the lateral cyclic demand from thermal fluctuations. Global instability is not considered in

designing the piles of IABs because piles are embedded in soil which provides a restraining effect

against global buckling. However, the flanges and web of the HP sections are vulnerable to local

buckling due to lateral cyclic expansion and contraction along the bridge. AISC LRFD 2005 set

limits for flange and web width-to-thickness ratios to control local buckling, and compact sections

are classified as the sections that can develop a full plastic moment capacity before buckling is

induced. Piles in IABs are also prone to fatigue due to repetitive movement of the bridge. Dicleli

and Albhaisi (2003, 2004) initiated an approach to estimate the maximum length of IABs based

on low-cycle fatigue of steel piles under thermal effects. They implemented a fatigue damage

model for daily/weakly and annual temperature variations to estimate the limiting strain for fatigue

failure based on a 75-year bridge life. The strain is then used to estimate the maximum lateral

displacement that the piles can sustain and is used to estimate the maximum length of the IABs

assuming the bridge expansion follows a basic thermal linear expansion equation. Table 2.2

presents the maximum length limits of IABs found for piles embedded in clay.

18
TABLE 2.2 MAXIMUM LENGTH LIMITS FOR STEEL AND CONCRETE IABS IN CLAY
(Dicleli and Albhaisi, 2003)

Concrete bridge maximum length (m) Steel bridge maximum length (m)
Pile Size Moderate Cold Moderate climate Cold
climate climate climate
HP310×125 320 265 220 145
HP310×110 300 250 205 135
HP250×85 240 195 160 110
HP200×63 180 150 125 80

2.7 Bridge Geometric Limitations

The geometric limitations of IABs have been the subject of research and controversy among

departments of transportation for decades. The problem encompasses many parameters like

maximum abutment height, pile types, pile size and orientation, allowable skew angle of the

bridge, degree of compaction of the backfill soil and the approach slab-abutment connection.

Greimann, et al. (1984) reported in a survey that construction details for IABs vary widely from

state to state. Vermont Agency of Transportation (Vtrans) proposed a simplified design method

for the IABs, based on experience of both structural and geotechnical engineers. The agency

suggested the following criteria that must be met for the method to be applicable:

- The bridge skew angle should be less than 20 degrees

- The bridge should have straight beams, whether it is a straight or curved bridge. The beams

should be parallel to each other.

- The piles supporting the abutments should be H-section of grade (50) steel.

- The flange of the H-piles should be greater than or equal to 10 inches.

19
- Abutments and piers should be parallel to each other.

- Abutment height should not exceed 13 ft. to reduce earth pressure against it.

- The maximum total bridge length from centerlines of abutments must be as follows:

o 395 ft. for steel bridges.

o 695 ft. for concrete bridges.

- Individual span lengths should not exceed 145 ft.

- Wing walls should be built monolithically with abutments with a length of 10 ft. or less.

The Vtrans guidelines stated that beyond the above requirements, a more detailed analysis is

required.

Kunin and Allampalli (2000) conducted a survey to obtain states’ departments of transportation

practices in design and construction of IABs. The study incorporated 39 states and agencies in the

United States and Canada. The researchers highlighted many facts about geometric limitations and

construction practices implemented. The maximum length was found ranged from 25-140 m for

steel bridges, 18 – 244 m for precast concrete bridges, and 48.8-244 m for cast-in-place concrete

girder bridges. The maximum allowable skew angle for the IABs was 45 degrees but many state

departments do not impose a limit on skew angle. Abutment height limits range from 0.9-4.3 m.

Table 2.3 summarizes this survey.

20
TABLE 2.3 SUMMARY OF IAB RESPONSES (ADAPTED FROM KUNIN AND
ALLAMPALI, 2000)

Longest built
Total built First built Last built
Steel Girder CIP Concrete CIP Concrete
Girder Girder
318.4 maximum 358.4 maximum 290.4 maximum
Over 9773 1905 1996
24.4 minimum 15.9 minimum 29.3 minimum

Frosch et al., 2004 conducted a series of experimental tests on steel HP-piles to predict the

maximum lateral displacement that the steel and CFT piles could sustain without buckling. Six

different HP-sections and three concrete-filled steel tube piles (CFT) were tested for combined

low-cycle large amplitude lateral displacement and maximum axial load permitted by AASHTO

standard specifications. The axial stress limit by AASHTO specifications is 25% of steel yield

strength in case of steel sections while it is 25% of steel yield strength + 40% of concrete

compressive strength for concrete-filled steel tube. Displacement capacities of HP sections

predicted by the researchers are reported in Table 2.4. These limiting displacements were used to

estimate the maximum length of IABs based on a linear thermal expansion equation. The

maximum length of IABs is estimated to be 500 ft.

21
TABLE 2.4 BUCKLING DISPLACEMENT CAPACITIES OF HP SECTIONS (Frosch et al.,
2004)

Specimen Orientation Axial Load (kips) Buckling capacity (in)


HP8x36 Weak 95.4 2.0
HP8x36 Strong 95.4 1.5
HP8x36 45o 95.4 1.5
HP8x36 Weak 191 2.25
HP10x42 Weak 112 1.25
HP12x53 Weak 140 1.25
CFT8 - 129 1.5
CFT8 - 173 1.5
CFT10 - 204 1.75

Azizinamini et al. (2015) discussed the effect of the flexibility in the pile-head connection. They

showed that a hinge connection can improve the displacement capacity of the piles by four times

more than the fixed connection. They demonstrated that through the stiffness and bending moment

values developed in the pile, which were less in case of a pinned connection. They also indicated

that strong axis orientation of the pile would increase the displacement capacity by 70% compared

to weak axis orientation. In their study, they found that the maximum principle plastic strain for a

modelled prestressed pile in ABAQUS is much higher in the case of a fixed connection than in the

case of a pin connection for the same lateral displacement at the pile head.

22
3. SOIL-STRUCTURE INTERACTION

3.1 Introduction

The soil-structure interaction is the most challenging part in the analysis and design of any

structure in contact with soil because of the uncertainties related to soil state response under

various loading. The complexity of the soil-structure interaction emerges from the nature of soil

as a nonhomogeneous material, and its behavior and properties differ from one spot to another

even on the same area. However, it is important to identify the right soil type and properties which

closely reflect the real behavior in the field. The IABs, unlike other bridge types, are in a semi-

static state and have a strong and pertinent relation to the supporting soil. IABs are subjected to

daily and seasonal movements due to thermal expansion and contraction, and these movements

are converted into lateral forces on the supporting substructure elements. Basically, the soil-

structure interaction in IABs is of two categories: pile-soil interaction and abutment-soil

interaction. This chapter will focus on the theoretical approach for modelling both types of

interactions.

23
3.2 Soil-Pile Interaction

Soil-pile interaction studies have evolved in the middle of the last century in response to the need

for understanding the behavior of offshore platforms under wind, wave and other lateral and axial

loads (Reese and Impe, 2011; Desai and Zaman, 2014). The most common and widely used method

for modelling soil-pile response is the p-y method. This method represents the soil pressure at any

point along pile depth versus the corresponding lateral displacement in a form of nonlinear springs.

A set of p-y curves or springs for each pile is required to represent the effect of the soil on the pile

under various loadings, Fig (3.1). These curves are functions of soil type and properties, water

level, and loading pattern. Most departments of transportation and engineering bureaus depend on

the p-y method due to main two reasons: First, it is simple, wherein the equations are empirical

and based only on soil properties and constants and does not require complex soil models like the

three-dimensional FE method. Second: it is highly accurate and thought to best represent the real

soil behavior, because it is calibrated using full-scale tested models. For these two reasons, and

because of the lack of validation for other modelling techniques, it was selected as the focus of

the current study.

Reese and Imple (2001) presented a detailed description on how to compute the p-y curves for

various soil types and loading conditions.

24
Figure 3.1 Set of p-y curves along the pile depth (Reese and Imple, 2001)

3.2.1 p-y Curves

In general, three soil types are commonly considered in soil mechanics books: clay, sand, and silt.

In the following sections, methods for computing p-y curves for these three soil types will be

presented with various loading conditions.

3.2.1.1 p-y Method for Clay

3.2.1.1.1 Soft Clay Having Free Water: Static Loading Case

Matlock (1970) suggested an equation for computing the p-y curves for soft clay as follows:

25
1
p  y 3
 0.5   3.1
pult  y50 

Where:

  ' J  
 3  z  z  cu b 
pult  min  cu b   3.2
 9cu b 
 

average effective unit weight from ground surface to the point of spring under consideration.

z: depth from the ground surface to the point of spring under consideration.

cu: shear strength of the soil at depth of z.

J: A factor equals to 0.5 in case of soft clay and 0.25 in case of medium clay.

b: Pile width.

y50  2.5 50b

50: The strain that corresponds to 50% the maximum principal stress difference. Table 3.1 shows

typical values for 50. Fig. 3.2 depicts the typical p-y curve for soft clay. The value of the pressure

remains constant after y = 8y50.

26
TABLE 3.1 TYPICAL VALUES OF 50 FOR NORMALLY CONSOLIDATED CLAY (Peck et
al., 1974)

Consistency of Clay Average values for shear strength (kPa) 50

Soft <48 0.02

Medium 48-96 0.01

Stiff 96-192 0.005

Figure 3.2 Schematic representation of p-y curves for soft clay with free water under static load

case (Matlock, 1970)

27
3.2.1.1.2 Soft Clay Having Free Water: Cyclic Loading Case

The computation of the p-y curve in cyclic loading case is similar to the static loading case for

values of p less than 0.72pult . A transition depth 𝑧 is computed by solving the two equations in

(3.2) simultaneously. In the case where the unit weight and shear strength are constant, zr can be

calculated from the following equation:

6cu b
zr  3.3
  ' b  Jcu 

If the spring has a depth equal to or greater than z r, p will have a constant value of 0.72𝑝 for all

y values beyond 3𝑦 . For depths smaller than zr, the value of p decreases from 0.72𝑝 at y=3𝑦

to the value in eq. (3.4) at y=15𝑦 . Fig (3.3) shows the procedure graphically.

p  z 
 0.72   3.4
pu  zr 

28
Figure 3.3 P-y curve for soft clay with free water under cyclic loading (Matlock, 1970)

3.2.1.1.3 Stiff Clay Having Free Water: Static Loading Case

Reese et al. (1975) presented a method for predicting the p-y curve for stiff clay. The method

involves computing the ultimate value of the soil resistance (force per unit length) as the minimum

of the following two equations. Figure 3.4 explains the procedure:

pct  2ca b   ' bz  2.83ca z 3.5

pcd  11cu b 3.6

Where:

𝑐 ∶ average undrained shear strength over the depth z.

𝑐𝑢: undrained shear strength at depth z.

': soil submerged unit weight.

29
b: pile diameter.

Figure 3.4 p-y method for stiff clay under static loading in the presence of free water

The p-y curve is composed of several portions. The first straight line portion is computed from the

following equation:

p   ks z  y 3.7

Where ks obtained from Table 3.2.

TABLE 3.2 TYPICAL VALUES OF 𝑘

undrained shear strength of the soil (kPa)

50-100 100-200 300-400

kpys (static) MN/m3 135 270 540

kpyc (cyclic) MN/m3 55 110 540

30
The second portion of the p-y curve (first parabolic portion) is computed from the following

equation (the portion between eq. 3.7 and the point of y=A sy50:

0.5
 y 
p  0.5 pc   3.8
 y50 

Where:

Pc is computed as the smaller from eq. (3.5) or eq. (3.6)

y50=50b, where 50 is obtained from laboratory tests or from Table 3.3.

TABLE 3.3 TYPICAL VALUES OF 50 FOR OVERCONSOLIDATED CLAY (Skempton,


1951)

Consistency of Clay Average values for shear strength (kPa) 50

Soft 50-100 0.007

Medium 100-200 0.005

Stiff 300-400 0.004

The second parabolic portion of the p-y curve is computed from to the following equation which

defines the part of p-y curve bounded by y=𝐴 𝑦 to y=6𝐴 𝑦 :

0.5 1.25
 y   y  As y50 
p  0.5 pc    0.055 pc   3.9
 y50   As y50 

Where As could be found from the Figure 3.5.

31
Figure 3.5 Graphical values of A (subscripts refer to static and cyclic loading). (Reese and Imple,
2001)

The next straight line in the p-y curve is constructed according the following equation:

0.0625
p  0.5 pc (6 As )0.5  0.411 pc  pc  y  6 As y50  3.10
y50

Which defines the portion of the portion of the p-y curve between y=6A sy50 and y=18Asy50.

The final straight line can be found from the following two equations:

p  0.5 pc  6 As   0.411 pc  0.75 pc As


0.5
3.11


p  pc 1.225 As  0.75 As  0.411  3.12

32
And defines the p-y curve from point of y=18𝐴 𝑦 thereafter.

3.2.1.1.4 Stiff Clay in the Presence of Free Water: Cyclic Loading

For cyclic loading, the procedure is like that of static loading except that the portion of the

parabolic curve (Eq. 3.8) is defined as follows:

 y  0.45 y p
2.5

p  Ac pc 1   3.13
 0.45 y p 
 

Where 𝐴 can be found from Figure 3.5. Eq. 3.13 represents the portion of the curve bounded the

intersection of Eq. 3.7 and y=0.6𝑦 , where 𝑦 is defined as follows:

y p  4.1Ac y50 3.14

The straight line of the curve defined by the limits y=0.6𝑦 and y=1.8𝑦 is computed from the

following equation:

0.085
p  0.936 Ac pc  pc  y  0.6 y p  3.15
y50

The last part straight line from point of y= 1.8𝑦 thereafter is computed from the following

equation:

0.102
p  0.936 Ac pc  pc y p 3.16
y50

Figure 3.6 illustrates the procedure for the cyclic loading.

33
Figure 3.6 p-y method for stiff clay under cyclic loading in the presence of free water (Reese et

al. 1975)

3.2.1.1.5 Stiff Clay with no Free Water: Static Loading

Welch and Reese (1972) developed a method for formulating p-y curves for stiff clay with no free

water based on field tests in Houston, TX. The method is the same for soft clay with free water

except the equation describing the p-y curve (equation 3.1) is defined as follows:

0.25
p  y 
 0.5   3.17
pu  y50 

For y values beyond 16𝑦 , p is equal to pult. Figure 3.7 illustrates the procedure.

34
Figure 3.7 p-y method for stiff clay under static loading with no free water (Welch and Reese,

1972)

3.2.1.1.6 Stiff Clay with no Free Water: Cyclic Loading

The procedure is like the case of static load in Sec. 3.3.1.5, but a new value of y for each cycle is

determined from the following equation:

yc  ys  y50 C log N 3.18

Where:

4
 p 
C  9.6   3.19
 pult 

𝑦 : Deflection under N-cycles of load

𝑦 : Deflection under static loading

N: Number of cycle to which curve is desired.

35
Figure 3.8 illustrates the method of computing p-y curves for stiff clay under cyclic loading with

no free water.

Figure 3.8 p-y curves for stiff clay with no free water under cyclic loading case (Reese and Imple,

2001)

3.2.1.2 p-y Curves for Sand

Reese et al. (1974) performed a series of tests at Mustang Island to develop p-y curves for sand.

The procedure incorporates three segments: one parabolic part and two straight line segments. The

method starts by computing pult and pm from the following equations:

pult  As ps 3.20

pult  Ac ps 3.21

The corrspoding y intercept (x-axis herein) is computed as follows :

36
3b
yu  3.22
80

pm  Bs ps 3.23

pm  Bc ps 3.24

And the corrspoding y intercept:

b
ym  3.25
60

The parameters As , Ac , B s and B c can be found from Fig 3.9 and Fig 3.10. The ultimate soil

pressure ps per unit length of the pile is computed from the following two equations:

 k z tan  sin  tan  


pst   z  o   b  z tan  tan    ko z tan   tan  sin   tan    kab 
 tan      cos  tan      

3.26

psd  kab z  tan8   1  kob z tan  tan 4  3.27

The above two equations are solved simultaneously to estimate the depth yt. Above this depth

Eq. 3.24 is used and below it Eq. 3.25 is implemented.

The first segment of the p-y curve is constructed from the following equation:

p   k py z  y 3.28

Where 𝑘 can be obtained from Table 3.4 or Table 3.5

37
Figure 3.9 Coefficients of As and Ac

Figure 3.10 Nondimensional soil resistance coefficients with depth

38
TABLE 3.4 REPRESENTATIVE VALUS FOR 𝑘 FOR SAND BELOW WATER TABLE
(SUBMERGED)
Soil relative density Loose Medium Dense

kpy (MN/m3) 5.4 16.3 34

TABLE 3.5 REPRESENTATIVE VALUES FOR 𝑘 FOR SAND ABOVE WATER TABLE

Soil relative density Loose Medium Dense

kpy (MN/m3) 6.8 24.4 61

The parabolic part of the curve between point k and m is computed from the following equation:

p  Cy1/n 3.29

Where:

pm
n 3.30
mym

pm
C 3.31
ym1/ n

The slope of the line between point m and u is computed from the following equation:

pu  pm
m 3.32
yu  ym

39
Point k (the point of intersection of the first segment and the middle one) is computed based on

the following formula:

n
 C  n 1
yk   3.33
 k py z 
 

Figure 3.11 illustrates the procedure for calculating p-y curve for sand

Figure 3.11 Computation of p-y curve for sand (Reese et al., 1974)

40
3.2.1.3 p-y Curves for Soil Having Cohesion and Angle of Friction

There are no explicit recommendations for creating p-y curves for soils having both cohesion and

angle of internal friction. However, a procedure suggested by Reese and Impe (2001), is presented

herein for the sake of completeness. It assumes the stress-strain curve for the c- soil closely

matches that for the cohesionless soil. The procedure is as follows:

1- Establish yu as compute Pult from the following two equations for static and cyclic loading,

respectively:

pult  As pult  pultc 3.34

pult  Ac pult  pultc 3.35

And the corresponding y intercept:

3b
yu  3.36
80

2- Compute pm from the following two equations:

pm  Bs ps 3.37

pm  Bc ps 3.38

And the corrspoding y intercept:

b
ym  3.39
60

The parameters As , Ac , B s and B c can be found from Fig 3.9 and Fig 3.10 as in the case of

cohesionless soil. The ultimate soil pressure ps is computed from equations 3.26 and 3.27.

41
The parameter A can be obtained from Fig. 3.9 and the friction part of Eq. 3.33 can be found

from the following two equations (smaller governs):

 K tan  sin  tan  


pult   z  o   b  z tan  tan    3.40
 tan      cos  tan      

pult  K a b z  tan8   1  Kob z tan  tan 4  3.41

The cohesion part of the equation pultc is computed from the following two equations:

 ' J 
pultc   3  z  z c b 3.42
 c b 

pultc  9 c b 3.43

The first segment of the curve is constructed according the following equation:

p   k py z  y 3.44

Where 𝑘 is obtained from Eq. 3.45 and z is the depth at which the p-y curve is desired.

k py  kc  k 3.45

The coefficients kc and k can be estimated using Fig. 3.12 and 3.13.

The parabolic part of the curve between point k and m is computed from the following equation:

p  Cy1/n 3.46

Where:

pm
n 3.47
mym

42
pm
C 3.48
ym1/ n

The slope of the line between point m and u is computed from the following equation:

pu  pm
m 3.49
yu  ym

Figure 3.12 coefficient Kc

43
Figure 3.13 coefficient K

Point k (the point of intersection of the first segment and the middle on) is computed based on the

following formula:

n / n 1
C
yk    3.50
 kz 

Figure 3.14 illustrates the procedure for calculating p-y curve for cohesionless soil.

44
Figure 3.14 proposed p-y curve for c- soil (Reese and Impe, 2001)

3.2.2 Program LPILE

LPILE is a Window-based commercial software developed by professor Lymon Reese at the

University of Texas at Austin to solve the problem of a single pile subjected to combined

axial/bending moment and lateral load/displacement. Currently, LPILE is the most widely used

software among departments of transportation in the US to solve the problem of laterally loaded

piles. The program considers the theories presented in the previous sections in addition to other

recommendations for computing p-y curves by other organizations, such as API (1993). It

considers various pile sections, soil types and loading scenarios. It also quantifies lateral pile

deflection, bending moment, shear force, and soil pressure distribution along pile depth. This

program will be used for computing p-y curve of the soil spring for piles in the following chapters.

The nonlinear force-displacement data are obtained by converting the pressure of the p-y data into

45
lumped forces concentrated at nodes of the pile. The forces can be calculated by multiplying the

value of the pressure at a specific depth below ground surface by the average distance between

springs on the top and bottom of that depth for the center springs and half the distance in the edge

springs. A series of force-displacement data can be generated for the springs at various depths

along the pile, which are inputted into the ABAQUS program to represent the soil effect on the

IAB.

3.3 Abutment-Backfill Interaction

Integral Abutments experience lateral movement following bridge expansion and contraction. This

movement exerts passive pressure on the adjacent soil in the case of bridge expansion and active

pressure in the case of contraction. The horizontal soil pressure on the abutment is a function of

soil properties, friction between the soil and interfacing structure, shape of the structure, and depth

of the point below ground level.

The magnitude of lateral earth pressure on an abutment can be calculated from the following

equation:

p=zk 3.51

Where:

p: horizontal soil pressure,

: effective unit weight of the backfill,

z: depth of the point below ground level and

k: coefficient of lateral earth pressure.

46
The of lateral earth pressure coefficient (k) represents the ratio of the horizontal effective stress

to the vertical effective stress. The value of earth pressure coefficient depends of the direction of

movement of the wall/abutment relative to the soil. Three scenarios have been reported in soil

mechanics references:

- At-rest case

The soil in this case is in a state of equilibrium, in which the soil does not experience

compression or expansion. The earth pressure coefficient is referred to as at-rest pressure

coefficient (ko)

- Active case

If the confinement has been released due to movement of the retaining structure away from

the soil, an active case ensues, and the corresponding coefficient is referred to as active

earth pressure coefficient (ka). This case represents the lower bound value (minimum earth

pressure coefficient).

- Passive case

When the retaining structure or abutment moves towards the soil compressing its particles,

passive earth pressure arises, and the coefficient is referred to as the passive earth pressure

coefficient (kp). The passive case represents the upper bound case. (maximum earth

pressure coefficient). Figure 3.15 depicts the three cases of lateral earth pressure

mobilization.

Movement of the structure relative to the soil invokes shear stresses due to friction in the

interface, resulting in an inclined reaction at an angle () with the normal to the interface.

The value of () is a function of the soil properties and the material of the interfacing wall.

47
At-rest condition Active condition Passive condition

Figure 3.15 Lateral earth pressure cases

In IABs, passive earth pressure induces as the structure expands due to temperature rise in hot

season, applying compressive forces to the abutments. When the abutments move away from the

backfill during cold seasons, active earth pressure develops gradually until it reaches a constant

value. In general, the active earth pressure is relatively small compared to passive pressure and it

is typically neglected in the analysis of IABs (Albhaisi et al., 2012).

There are three main theories for computing the passive earth pressure: Coulomb theory (1776),

Rankine theory (1857), and log spiral theory (Terzaghi, 1943). Rankine theory underestimates the

value of earth pressure, while Coulomb theory overestimates the value of earth pressure for values

of (/) > 0.4 (Cole and Rollins 2006). Log spiral theory provides a more accurate prediction of

passive earth pressure than both Rankine and Coulomb theories for values of (/) > 0.4, where 

is the angle of internal friction of the soil (Duncan and Mokwa, 2001)

48
The main three earth pressure theories can only predict the maximum value of earth pressure;

however, they do not explain how that pressure develops with the displacement up to the maximum

value. There are a very limited number of studies that address the development of passive earth

pressure with the displacement of the wall. Clough and Duncan (1991) conducted a series of field

tests and developed finite element models to estimate the amount of displacement to achieve

maximum passive earth pressure. They also developed graphs for calculating earth pressure

coefficients versus normalized maximum wall displacement behind a retaining wall for loose and

dense sand, for both compacted and non-compacted backfill (Figure 3.16 and 3.17). These authors’

work has been adopted by the NCHRP (1991) manual for describing changes in the passive earth

pressure coefficient with wall top displacement. However, no clues exist on how to compute the

variation in earth pressure coefficient with displacement for various soil types or properties. Table

3.6 shows the normalized displacement of the wall with respect to the wall height for various soil

types.

TABLE 3.6 APPROXIMATE WALL MOVEMENT RRQUIRED TO REACH MAXIMUM


AND MINUMUM EARTH PRESSURE (Clough and Duncan, 1991)
Values of /H

Type of Backfill Soil Active Passive

Dense Sand 0.001 0.01

Medium-dense Sand 0.002 0.02

Loose Sand 0.004 0.04

Compacted Silt 0.002 0.02

Compacted lean clay 0.01 0.05

Compacted fat clay 0.01 0.05

49
Figure 3.16 Wall movement versus earth pressure for ideal conditions (Fang, 1991)

Figure 3.17 Wall movement versus earth pressure for compacted backfill (Fang, 1991)

50
In the current study, graphs developed by Clough and Duncan (1991) were utilized to compute the

force-displacement relationship for the abutment springs. Clough and Duncan had only depended

on Rankine earth pressure theory and three friction angles for soil in their graphs. As such, this

method is expanded herein to incorporate more friction angles and earth pressure theories. This

has been performed by digitizing the original curves and proportioning to the desired angle of

friction of the soil or to the log-spiral earth pressure theory as will be described in subsequent

chapters.

51
4. CALIBRATION OF THE NUMERICAL MODEL

4.1 Introduction

To examine the validity of the numerical model, a verification procedure was conducted by

creating a finite element model using ABAQUS with the same geometry and loading conditions

as the experimentally tested bridge. Bridge components were simulated using 3D solid elements

to best reflect the real behavior of a monitored bridge response under measured field temperatures.

Nonlinear p-y curves for the soil-pile and soil-abutment interactions were employed for the

calibration process, combined with several shrinkage models for the concrete deck and girders. A

separate pile model is created and calibrated with experimental data based on the lateral

displacement profile. Full bridge model is later established utilizing the outcomes of the pile

modelling and a calibration procedure is conducted based on the lateral abutment displacement.

The earth pressure response is also presented and compared.

52
4.2 Bridge Description

The integral abutment bridge located on SR-18 over the Mississenewa river in Indiana, USA was

used for calibration (Figure 4.1). This bridge has been previously instrumented and monitored for

seven years (Frosch et al., 2006; 2011). It was selected for the current study because of the long

monitoring period and availability of design plans. Moreover, its length is typical for bridge spans

over highways in the United States. The bridge has a total length of approximately 112 m (367 ft)

divided into five spans and has an 8° skew angle as shown in Figure 4.2. The bridge has CIP 200

mm (8 in) thick deck, supported by five 1.524 m (60 in) prestressed bulb T-beams (INDOT Design

Manual, 2010), spaced at 3.1 m (10 ft-2 in). The bridge is integrally built with two abutments

having 2.74 m (9 ft) height and 1 m (39 in) thickness and supported by 10 concrete filled tube piles

CFT 355 x 8 mm (CFT14 in x 0.312 in) in a single row. Typical section in the abutment is shown

in Figure 4.3. The average length of the piles is 6.3 m (20.8 ft) and 8.2 m (27 ft) under bent 1 and

bent 2, respectively. Instrumentation data of the bridge includes pile strains along pile depth,

abutment displacement, soil pressure behind abutment in addition to temperature variation along

monitoring period. Pile and abutment displacements were selected as an index for comparison with

their experimental counterparts.

53
Figure 4.1 SR-18 over Missessenewa River, IN (Lovell, 2010)

18.9 m 24.7 m 24.7 m 24.7 m 18.9 m


Bent 1 Bent 6
A

14.6 m

1.1 m 3.1 m 3.1 m 3.1 m 3.1 m 1.1 m

Section A-A

Figure 4.2 Plan and cross section of bridge SR-18 over Mississinewa River

54
Figure 4.3 Typical details of abutment of SR-18 (Lovell, 2010)

4.3 Finite Element Modelling

ABAQUS/CAE v14 was used to generate two comprehensive 3D FE models. The first model is a

separate pile of a previously monitored IAB and the second model is a full bridge model using the

same bridge from which the first model is derived. The pile model was built to calibrate the pile

displacement against the experimental data, eventually predicting the soil type that best matches

the field soil-pile experimental response. A full bridge model is based on pile model outcomes to

investigate the abutment hysteresis behavior.

55
Modelling of IAB behavior requires valid assumptions for the boundary conditions and loading

scenarios. Soil-structure interactions are the most challenging part of the modelling process. In

general, two principal procedures are available to model the soil-structure interaction: finite

element discretization and Winkler springs. Soil modelling as finite elements is a time-consuming

process, especially in large or 3D models, and it has multiple drawbacks including uncertainties

regarding elastoplastic soil parameters prediction and the inability to track soil behavior during

cyclic loading. The Winkler springs method, on the other hand, provides a very economical

alternative in terms of computational time for modelling soil effects on adjacent structures. More

importantly, the springs method has a wide range of calibration procedures against full-scale field

data, and as such it gains better reliability in modelling. Therefore, soil springs were adopted

throughout this study to model the soil-pile and soil-abutment interactions.

4.4 Modelling Soil-Pile Interaction

The most common and widely used method for modelling the soil-pile response in laterally loaded

piles is the p-y curve method. This method, as its name implies, represents the soil pressure per

unit length of the pile at any point along pile depth versus the corresponding lateral displacement

of that point. Nonlinear springs are attached to the nodes of the FE model at the locations of the

soil-pile interface. A set of p-y curves for the springs along the pile is required to represent the

effect of the soil on the pile under lateral loading. The p-y method is accredited by most

departments of transportation and engineering bureaus across the United States as a method for

analysis of laterally loaded piles (Kunin and Alampalli, 2000) due to main two reasons: First, it is

simple to implement; the equations are empirical and based only on soil properties and pile section

without complex procedures. Second, it is highly accurate and thought to best represent the real

56
soil response because it is calibrated against full-scale field tests (Matlock H., 1970; Reese et al.,

1975). For these two reasons, and because of the lack of validation procedures for other modelling

techniques, like 3D FE, the p-y curve method was incorporated into the current study. LPILE

v2015 (Ensoft, Inc) software, which adopts the p-y curve methodology, was used for quantifying

the nonlinear p-y curves for the soil around the piles. Pressure-displacement (p-y) data of the

springs is extracted from LPILE according the soil type, then converted into force-displacement

data by multiplying the pressure by the corresponding tributary length for each spring. The force-

displacement data for each spring along the pile depth are assigned to the FE model with nonlinear

elastic behavior. Each spring is modelled as a connector element in Abaqus with a corresponding

connector section for that depth. Two orthogonal springs are attached at each node in the pile to

account for the effect of movement in both perpendicular directions. Connector elements in

Abaqus have combined tension/compression behavior and are not able to model tension-only or

compression only behavior. In the case of laterally loaded piles, soil springs should compress

against the pile while the pile moves towards the soil; conversely, in the case of pile movement

away from soil, springs must release the pile to avoid applying tensile force at the pile, assuming

negligible soil tension. To overcome this inevitable precondition in Abaqus, the supposed

compression effect on the other side of the pile in the case of countermovement is treated as reverse

tension on the same spring. As such, one spring in each direction is sufficient to represent the soil

effect on the pile in one direction. Due to the high number of springs required to represent the soil

effect on the pile, the process of spring generation was automated in the FE model. A Python

scripting property available in Abaqus was adopted herein for this purpose. To simplify the

procedure of scripting and make it more efficient, the code was not written from scratch, but

instead extracted from the original companion script file in Abaqus (journal file) and

57
encapsulated within adequate loops to perform the intended operation of spring planting to the FE

model.

4.5 Modelling Soil-Abutment Interaction

Integral abutments experience lateral movement following bridge expansion and contraction. This

movement exerts passive pressure on the adjacent soil in the case of bridge expansion and active

pressure in the case of contraction. In general, the active earth pressure is relatively small compared

to the passive pressure and it is typically neglected in the analysis of IABs (Albhaisi et al., 2012).

The magnitude of lateral earth pressure on an abutment can be calculated from the following

equation:

p=zk 4.1

Where p: horizontal soil pressure, : effective unit weight of the backfill, z: depth of the point

below ground surface and k: earth pressure coefficient.

There are three main theories for explaining lateral earth pressure: Rankine, Coulomb, and

logarithmic spiral. Two earth pressure theories, Rankine and log spiral theory, were examined in

the current study to investigate their potential to represent the real field behavior. Coulomb theory

was excluded because it overestimates the value of lateral earth pressure (Cole and Rollins, 2006).

The main three earth pressure theories can only predict the maximum value of earth pressure; they

do not explain how pressure develops with displacement up to the maximum value. Very few

studies address the development of passive earth pressure with the displacement of the wall.

Clough and Duncan (1991) developed graphs for calculating earth pressure coefficients versus

58
normalized maximum wall displacement behind a retaining wall. This work was also adopted by

the NCHRP (1991) manual for prescribing the change of the passive earth pressure coefficient

with the wall top displacement. However, no indications exist on how to compute earth pressure

coefficient variation with displacement for various soil types or properties. As such, a modified

form of the Clough and Duncan method was considered to compute the force-displacement

relationship for the abutment springs. Figure 4.4 shows a digitized form of the graphs, wherein the

Rankine theory is used to compute the lateral earth pressure coefficient (k P) for a soil having

coefficient of internal friction ( =37o). Curves for other parameters (=35 and log-spiral) were

estimated by proportioning the values to the original curves.

8
Passive Earth Pressure Coef. (Kp)

2 Rankine f=35o(predicted)
Rankine f=37o(digitized)
Log-Spiral f=35o(predicted)
0
0.00 0.01 0.02 0.03 0.04 0.05
D/H

Figure 4.4 Clough and Duncan digitized curves

59
The passive resistance for a specific point in the force-displacement curve is computed by

multiplying the vertical stress  at a specified depth z by the lateral earth pressure coefficient (Kp)

obtained from the digitized curves in Figure 4.4 and then by the tributary area for each spring (the

horizontal distance between two springs times the vertical distance between two springs). The

corresponding lateral displacement is collected from the abscissa values multiplied by the

abutment height to generate a set of springs data. A Python code was also used herein to add soil

springs to the FE model through connector elements, as in the case of the pile springs. Appendix

A includes typical Python code for adding springs to the abutment.

4.6 Pile Finite Element Model

One of the two centrally located piles under bent 1 of the bridge on SR-18 (pile No.6 in Figure

4.5) has been modelled using ABAQUS/CAE. The pile is 6.8 m (22.25 ft) in length measured from

the ground level. The dimensions and material properties of the pile are listed in Table 4.1. 3D

solid elements (C3D8R) were used to model both the concrete core and the steel tube. The element

is an 8-node linear brick, reduced integration, hourglass control and capable of representing 3D

state of stress (ABAQUS/CAE manual). Hexahedral elements were used to model the pile shell

and wedge elements for modelling the concrete core. Concrete and steel were modeled as linear

elastic materials as the displacement range for the pile is not excessive under thermal bridge

movement and is considered acceptable for calibration process. Full composite action was assumed

between the concrete core and the steel pipe through connecting both contacting surfaces by TIE

constraint.

60
Figure 4.5 pile distribution of bridge SR-18 (Frosch et al., 2004)

The soil around the piles is composed mostly of silty loam according to the site investigations as

can be seen from Table 4.2 and 4.3. The water table lies 10.2 m (33.5 ft) below ground level. This

depth is below the pile tip; therefore, a dry soil is assumed in calculations. N in tables refers to the

number of blows in Standard Penetration Test (SPT).

61
TABLE 4.1 PILE PROPERTIES OF BRIDGE SR-18

Pile Property (CFT 14)


Outer diameter 14 in (360 mm)
Nominal wall thickness 0.312 in (7.9 mm)
Design wall thickness 0.291 in (7.4 mm)
Modulus of elasticity of steel 29000 ksi (200 GPa)
Poisson’s ratio of steel 0.3
Yield strength of steel 35 ksi (241 MPa)
compressive strength of concrete 4 ksi (27.6 Mpa)
Modulus of elasticity of concrete 3605 ksi (24.8 GPa)
Poisson’s ratio of concrete 0.2

TABLE 4.2 SOIL BORING FOR BENT 1 OF SR-18 (Frosch et al., 2011)
Depth below
N Soil Composition
Ground Level (ft)
0 - 0.5 - Asphalt
0.5 - 1.4 - Concrete
1.4 - 2.9 17 Silty loam, Slightly Moist, Stiff, Tan
2.9 - 5.0 23 Silty clay loam, stiff, slightly moist, gray
Silty loam, some sand, gravel, stiff, slightly moist,
5.0 - 10.0 20
Brown
10.0 - 15.0 21 Silty loam, stiff, slightly moist, Tan
Silty clay loam, stiff, slightly moist, grayish brown,
15.0 - 20.0 16
medium
20.0 - 25.0 15 Silty loam, medium stiff, slightly moist, gray
25.0 - 30.0 6 Silty loam, soft, moist, gray
30.0 - 35.0 47 Silty loam, soft, gray
35.0 - 45.0 78 Silty loam, hard, dry, gray

62
TABLE 4.3 SOIL BORING DATA FOR BENT 6 OF SR-18 (Frosch et al., 2011)

Depth blow
N Soil Composition
Ground Level (ft)
0.0 – 0.6 - Asphalt
0.6 – 1.2 - Concrete
1.2 – 2.7 19 Silty loam, stiff, slightly moist, brown
2.7 – 8.0 28 Sandy loam, medium dense, slightly moist, gray
8.0 – 10.0 16 Silty loam, medium stiff, slightly moist, brown
10.0 – 15.0 8 Sandy loam, loose, slightly moist, brown
15.0 – 20.0 12 Sand, loose, slightly moist, tan
20.0 – 25.0 19 Sand+ some gravel, medium dense, moist, brown
25.0 – 30.0 63 Silty loam, hard, slightly moist, gray
30.0 -35.0 112 Silty clay loam + gravel, hard, dry, gray

The soil was modelled using nonlinear springs following the guidelines in chapter 3. The springs

were created through connector elements using Axial and Elasticity options to generate the force-

deflection data of the nonlinear p-y relationship. Springs were distributed evenly at one foot apart

along the pile depth. Forces in the springs were computed by multiplying the soil pressure by the

distance between springs. The first spring at the pile-abutment connection (depth=0) and the last

spring (depth =L) has a tributary length equal to half the distance of the intermediate springs

because the intermediate springs receive half the spacing between nodes from both sides while the

edge spring has spacing from only one side as shown in Figure 4.6.

63
Edge Springs

0.50 ft
Intermediate Springs

1.00 ft

Figure 4.6 spring tributary area for the pile

Computation of p-y data requires predicting soil parameters such as soil shear strength, effective

unit weight, and angle of internal friction. The soil boring data available for this project include

only soil type and SPT-N values and there are no laboratory tests such as triaxial or direct shear

available to predict soil strength parameters. Therefore, processing is required to extract the

required p-y curves parameters from the available SPT data and soil type.

To obtain the soil parameters, the SPT-N values were used as a benchmark for determining the

soil parameters based on typical values found in geotechnical references. A weighted average

procedure is suggested for the SPT number of the blows across the soil profile for bent 1 and it is

estimated to be 18. Prakash and Sharma (1990) values shown in Table 4.4 were used to compute

the undrained shear strength of the soil (Cu) based on the estimated average value of SPT-N for

the soil at this site. The soil consistency, according to typical reference values, is very stiff, which

is in the range (120 kN/m2) corresponding to the estimated average SPT-N value.

64
TABLE 4.4 CLASSIFICATION OF SOIL CONSISTENCY (Prakash and Sharma, 1990)

Consistency Undrained Shear Strength, kPa SPT-N Values


(lb./ft2)
Very soft <12 (250.63) <2
Soft 12-25 (250.6-522) 2-4
Firm 25-50 (522-1044) 4-8
stiff 50-100 (1044-2088) 8-15
Very stiff 100-200 (2088-4177) 15-30
Hard >200 (4177) >30

Fang (1991) values are used herein for correlating SPT-N values and the angle of internal friction

for cohesionless soils, based on two investigators: Peck (1974) and Meyerhof (1956). The average

SPT-N value for the soil indicates that the soil is medium graded, the value of the angle of internal

friction () could be interpolated from the table to be 32o based on Peck and 37o based on

Meyerhof, therefore, an average value of 34° is considered.

TABLE 4.5 SPT-N VALUES VS ANGLE OF INTERNAL FRICTION FOR COHESIONLESS


SOILS (Fang, 1991)
Angle of internal friction, 

SPT-N Relative Density Peck et al. (1974) Meyerhof (1956)


Very loose sand <4 < 0.2 < 29 < 30
Loose sand 4-10 0.2-0.4 29-30 30-35
Medium sand 10-30 0.4-0.6 30-36 35-40
Dense sand 30-50 0.6-0.8 36-41 40-45
Very dense sand >50 > 0.8 > 41 > 45

65
The water table is well below the tip of the pile, as such, a dry unit weight is considered for the

calculations of p-y curves. Table 4.6 by Das (2011) includes the dry unit weight for different soil

types, it is used herein to estimate the dry unit weight of the soil.

TABLE 4.6 DRY UNIT WEIGHT FOR DIFFERENT SOIL TYPES (Das, 2011)

Dry unit weight, d


Soil Type Void ratio, e Ib/ft3 kN/m3
Loose uniform sand 0.8 92 14.5
Dense uniform sand 0.45 115 18
Loose angular-grained silty sand 0.65 102 16
Dense angular-grained silty sand 0.4 121 19
Stiff clay 0.6 108 17
Soft clay 0.9-1.4 73-93 11.5-14.5
Loess 0.9 86 13.5
Soft organic clay 2.5-3.2 38-51 6-8
Glacial till 0.3 134 21

Soil borings indicate multiple soil types contributing to the mixture of the soil around the pile. As

such, three soil types were selected for the current study: silt, very stiff clay, and medium sand, to

examine the potential representation of those types compared to experimental behavior. Table 4.7

includes parameters of the three soil types used in LPILE. The strains that correspond to one-half

of the principle stress difference 50 and the soil modulus Kpy in Table 4.7 are based on laboratory

tests for soil and were taken and from Reese and Van Impe (2011) according to the soil type used.

66
TABLE 4.7 SOIL PARAMETERS USED IN LPILE MODELS

Undrained Shear Strength Kpyd,


Soil Type b dc, kN/m3 50d
Cu,a lb/ft2(kN/m2) lb/in3(kN/m3)
Silt 2500 (120) 34° 110 (17) 0.005 1090 (295,430)
Very stiff clay 2500 (120) - 110 (17) 0.005 1000 (271,000)
Medium Sand - 34° 100 (16) - 90 (24,430)
a
Interpolated value based on Prakash and Sharma (1990)
b
Based on Fang (1991)
c
Suggested values for dry unit weight by Das (2010)
d
According to LPILE manual, 2015.

Figure 4.7 depicts the p-y curves for medium sand computed using LPILE; p-y curves for other

soil types (silt and very stiff clay) are in Appendix A. Curves are plotted every 0.6 m (2 ft) apart

to remove congestion and for more clarity; the actual curves used in the FE model were spaced 0.3

m (1 ft) apart.

To investigate the type of soil at the site, the three assumed soil types were analyzed with LIPLE

and compared to the experimental response. Loading of the model includes applying an axial load

of 355.85 kN (80 kips) according to the design calculations, combined with a lateral displacement

of 9.7 mm (0.38 inches) which represents the reading of the convergence meter at a temperature

differential of 15.5 °C (60° F) (Frosch et al., 2006).

Figure 4.8 shows the lateral pile displacements along its depth for the three soil types. The

computed displacements were compared to their experimental counterpart (Frosch et al., 2006).

67
y (mm)
0.0 3.0 6.1 9.1 12.2 15.2
6000 1050.7
z = 1 ft
z = 3 ft
5000 z = 5 ft
z = 7 ft 840.6
z = 9 ft
4000 z = 11 ft
z = 13 ft
630.4

p (kN/m)
z = 15 ft
p (lb/in) z = 17 ft
3000 z = 19 ft
z = 21 ft 420.3
2000

210.1
1000

0 0.0
0.0 0.1 0.2 0.3 0.4 0.5
y (in)

Figure 4.7 p-y curves for medium-Sand

Displacement (mm)
-2.54 0.00 2.54 5.08 7.62 10.16
0 0.00

-5 -1.52 Depth (m)


Depth (ft)

-10 -3.05

-15 -4.57
Experimental (Frosch et al.,2006)
LPILE (Silt)
-20 LPILE (Very Stiff Clay) -6.10
LPILE (Medium Sand)
-0.1 0.0 0.1 0.2 0.3 0.4
Displacement (in)

Figure 4.8 Displacement along pile depth for different soil types

68
It is obvious from the plot that the sandy soil curve is the closest one to the experimental response,

although the difference among the soil types is minimal. The shape of the curves is also well

matched to the experimental response. Therefore, the medium sand is considered as the

representative site soil and is used in the entire SR-18 bridge model because of its proximity to the

experimental behavior.

To validate the FE model and examine the performance of the nonlinear springs in ABAQUS

compared to LPILE, a pile model with medium sand is established in ABAQUS using the same

geometry and loading conditions of the LPILE model. The top of the pile is assumed to move

freely in the horizontal direction and is restrained against rotations in all directions. These

boundary conditions are assumed to simulate the pile head encased in a concrete abutment and

permitted to translate laterally. The axial load and lateral displacement were applied to a central

reference point tied to the pile head surface to avoid local stress concentrations and maintain rigid

body movement.

Nonlinear soil springs were attached to the pile along its depth at a spacing of 0.3 m (1 ft) in two

orthogonal directions according to the procedure presented above. The p-y curves were obtained

from LPILE and the soil springs were attached to the model using a connector element with

nonlinear elastic behavior. The displacements calculated by ABAQUS are compared with those

by LPILE as shown in Figure 4.9. It can be observed from the figure that the displacements of both

software outputs precisely coincide with each other, which is logical as both programs share the

same p-y curves. This is further proof of the efficiency of connector elements in representing soil

behavior around the pile. It is noteworthy that LPILE depends on the finite difference method in

solving the differential equation of the soil-pile interaction, which is less accurate than the finite

element method but still gives identical results in small displacement theory.

69
Displacement (mm)
0.0 2.5 5.1 7.6 10.2
0 0.00
LPILE (Medium Sand)
ABAQUS (Medium Sand)
-5 -1.52

Depth (m)
Depth (ft) -10 -3.05

-15 -4.57

-20 -6.10

0.0 0.1 0.2 0.3 0.4


Displacement (in)

Figure 4.9 ABAQUS vs LPILE computed displacements

4.7 Finite Element Model of the Bridge

A 3D finite element model was developed for the entire bridge using software ABAQUS/CAE

Standard. All bridge components were constructed using solid elements (C3D8R) as in the case of

the pile model. Supports on all piers were modelled as rollers except at the bridge centerline, where

the supports were restrained (prevented from movement) in both horizontal and vertical directions.

The connection between the bridge deck and the girders was assumed to be fully composite and a

tie constraint was enforced between them. Connections between girders and abutments, piles and

abutments, and the concrete pile core and pile shell were also all assumed to be fully composite

and a tie constraint used between them. Figure 4.10 shows the finite element discretization of the

bridge model with all details simulated as the actual geometry of the bridge.

70
Figure 4.10 Bridge FE model

Material properties of the bridge deck, girders and abutments are taken according to the project

information (Frosch et al., 2011) as follows: The modulus of elasticity of the deck and abutments

is 24.8 GPa (3605 ksi) and for the prestressed girders is 30.4 GPa (4415 ksi). A linear elastic

behavior was assumed for all bridge materials. The three most commonly used shrinkage models

in practice were selected to calibrate the finite element model with the experimental data. The

models are ACI-209, CEB-MC-90-99, and CEB-MC-90-90. The GL-2000 model was excluded

because it yielded higher results for shrinkage in preliminary calculations. A MATLAB code was

developed to compute the shrinkage strains according the guidelines of ACI 209.2R-08. Shrinkage

strains were computed for both the deck and girders and were included in the FE model through

equivalent temperatures. Conversion of shrinkage strain into temperature was performed by

dividing the shrinkage strain by the coefficient of thermal expansion of the concrete, where the

thermal expansion of concrete was taken as 9.9 x10-6/°C. Figure 4.11 shows the shrinkage models

for the deck of the bridge.

71
400

300
me

200

ACI-209 Model
100 CEB-MC-90-99 Model
CEB-MC-90-90 Model
GL2000 Model
0
0 500 1000 1500 2000
Time (days)

Figure 4.11 Shrinkage strain of the bridge deck

The abutment-backfill springs were computed considering two earth pressure theories: Rankine

and log-spiral theory. The backfill soil was assumed to be dense sand with a unit weight of 20

kN/m3 (130 lb/ft3) and angle of internal friction of (35°) as per Frosch et al. (2011). Wall friction

between the abutment and the backfill was assumed to be two thirds the angle of internal friction.

Based on these soil properties, the coefficient of passive earth pressure in the Rankine case is

3.69 and, in the log-spiral case is 7.78. Force-displacement curves for the Rankine earth pressure

theory with =35o is shown in Figure 4.12.

72
Displacement (mm)
0.0 3.0 6.1 9.1 12.2 15.2
5000 22.2
z= 1ft
z= 2 ft
z= 3 ft
4000 z= 4 ft
17.8
z= 5 ft
z= 6 ft
3000 z= 7 ft 13.3

Force (kN)
Force (lb)

z= 8 ft
z= 9 ft
2000 8.9

1000 4.4

0 0.0
0.00 0.01 0.02 0.03 0.04 0.05
Displacement (ft)

Figure 4.12 Force-displacement curves for SR-18 Backfill (Rankine with =35o)

The dead load of the bridge was included in the model for all its components through the “Gravity”

option in ABAQUS using the respective densities of each material. Live load was not included in

the model, because it has been shown by Kim and Laman (2012) that traffic loading has an

insignificant effect on the bridge response. The temperature values were taken from the field

measurements of the bridge. An average temperature differential of 70°F was taken and used in

the model as “predefined fields”.

Two bridge models were established to track bridge expansion and contraction phases separately,

due to the unavailability of compression-only-springs within ABAQUS to model the soil behavior

behind the abutment as stated earlier. Unlike the pile case, the abutment has soil on only one side

73
and in the contraction phase, the springs should be nulled as the abutment moves away from the

soil, assuming negligible active pressure. As such, one model is required for the expansion phase

and another for the contraction phase, and the overall behavior is obtained by adding up the

contraction and expansion displacements at the peak points (equivalent to the average maximum

and minimum temperatures of 90°F and 20°F, respectively) to track the response along

instrumentation time.

4.8 Calibration Matrix

To calibrate the FE model with the experimental data, the key parameters governing bridge

response were varied to help capture the experimental response. Those parameters include:

temperature variation across superstructure depth, time dependent effects, and boundary

conditions. Multiple finite element models were built for the bridge to account for various cases.

The abutment displacement and earth pressure were selected as the critical responses for

calibration. Ten cases were investigated with different combinations. Temperature variations

across the superstructure profile were assumed for three scenarios: deck only temperature (TD),

deck and girder temperature (TDG) and deck temperature plus 50% girder temperature. The third

scenario implies full measured field temperature given to the deck and half of the temperature

given to the girders. This distribution is considered to simulate the potential temperature gradient

across the bridge superstructure, whereas the two other options (TD or TDG) suggest full

temperature potential given to the deck or girders.

Time-dependent effects were incorporated through shrinkage strains; creep and prestressed losses

were ignored in the current study because their effects are minimal. Shrinkage strains were applied

74
according to two cases: deck only shrinkage (SD), and deck and girder shrinkage combined

(SDG), with three shrinkage models as previously stated. Two theories for passive earth pressure

were considered: Rankine and log-spiral theory. This calibration improvement progress was

evaluated for each case and was adjusted for subsequent cases by maintaining the successful

assumptions and excluding cases leading to anomalous results, until the response was fully

optimized compared to experimental behavior. The calibration matrix with all potential

combinations is shown in Table 4.8. Figure 4.13 shows a schematic diagram for the sequence of

operations of the FE modelling of the whole bridge model.

TABLE 4.8 CALBRATION MATRIX

Shrinkage Models
Earth Pressure
Temp CEB-90-99 CEB-90-90 ACI-209
Theory
Log-
Case No. Deck Girder Deck Girder Deck Girder Deck Girder Rankine
Spiral
1 x x x x - - - - - x
2 x x x - - - - - x
3 x - x x - - - - - x
4 x - x - - - - - x
5 x 50% x x - - - - - x
6 x x x x - - - - x -
7 x 50% x x - - - - x -
8 x 50% x - - - - - x -
9 x 50% - - x - - - x -
10 x 50% - - x - - - x -

75
MATLAB Code Get Backfill Soil Data Get Pile Soil Data
(Shrinkage Models)

Digitizing and computing the p- LPILE (p-y curves for piles)


y curves for the abutment

Nonlinear Force-Displacement Data Nonlinear Force-Displacement


for abutment springs data for pile springs

Python Codes to
attach springs to the
ABAQUS model
Temperature Data

ABAQUS Final Model

Figure 4.13 Flow chart of the FE modelling

76
4.9 Results

The location of the convergence meter where the lateral displacement of the abutment is collected

during the experimental program (Frosch et al, 2011) was is selected to compare the monitored

versus the calculated displacements. The convergence meter was mounted 0.38 m (15 in) from

the bottom of the abutment (Figure 4.14). This same point in the FE model was used for

calibration of computed results. Results were adequately categorized according to earth pressure

theories; other cases are discussed inclusively to cover the calibration matrix cases.

Figure 4.14 Location of the convergence meter (Frosch et al., 2010)

77
4.9.1 Log-Spiral Theory

Five cases were evaluated considering the log-spiral lateral earth pressure theory for the soil behind

the abutment, including different combinations of temperature and shrinkage scenarios, as listed

in Table 4.8. Figure 4.15 shows the abutment displacement response along seven years, with

temperature and shrinkage given for both deck and girders. In this figure (case 1), two scenarios

are compared to the measured behavior, one including a shrinkage model (CEB-90-99) and the

other without shrinkage. The general trend of the abutment displacement tends to move upward

hysterically with time. This “ratcheting” is attributed to the difference in abutment displacement

values in the contraction and expansion phases. In the case of bridge contraction, there are no soil

springs resisting abutment while it moves inward. The reverse can be noted in the case of

expansion, in which the soil springs push against the abutment movement and hence reduce the

values of displacement. The figure indicates that the amplitude of the displacement is higher than

the experimental amplitude both with and without shrinkage. The curve is higher even before

superimposing shrinkage strains over the abutment displacement.

Figure 4.16 depicts case (2) in the calibration matrix. In this case, girder shrinkage is omitted from

the previous case, while keeping the same other parameters to investigate girder shrinkage. The

same response of Figure 4.15 is still dominant, but the displacement curves with shrinkage were

shifted slightly downward after removing girder shrinkage. The amplitude of the displacement

remained the same.

78
2.0 50.80
Measured
FEM without shrinkage
1.6 40.64
Abutment Displacement (in)

Abutment Displacement (mm)


FEM with CEB-90-99 model

Contraction
1.2 30.48

0.8 20.32

0.4 10.16
Expansion

0.0 0.00

-0.4 -10.16

2003 2004 2005 2006 2007 2008 2009 2010 2011


Time

Figure 4.15 Case (1) TD, TG, SD, SG, Log-spiral theory

2.0 50.80
Measured
FEM without shrinkage
1.6 40.64
Abutment Displacement (in)

Abutment Displacement (mm)


FEM with CEB-90-99 model
Contraction

1.2 30.48

0.8 20.32

0.4 10.16
Expansion

0.0 0.00

-0.4 -10.16

2003 2004 2005 2006 2007 2008 2009 2010 2011


Time

Figure 4.16 Case (2) TD, TG, SD, Log-spiral theory

79
Figure 4.17 and Figure 4.18 show the displacement of the abutment for cases (3) and (4),

respectively. The temperature is given to the deck only, without girder temperature, to study the

effect of temperature across bridge superstructure on the behavior. It is obvious in these cases that

the displacement amplitude is significantly reduced compared to the experimental curve. The

computed results still exceed the measured data, indicating higher computed displacements in the

FE model. This clearly indicates that this assumption for the temperature load is ineffective, and

further adjustment is justified regarding the distribution of the temperature across the height of the

superstructure.

2.0 50.80
Measured
FEM without shrinkage
1.6 40.64
Abutment Displacement (in)

Abutment Displacement (mm)


FEM with CEB-90-99 model
Contraction

1.2 30.48

0.8 20.32

0.4 10.16
Expansion

0.0 0.00

-0.4 -10.16

2003 2004 2005 2006 2007 2008 2009 2010 2011


Time

Figure 4.17 Case (3) TD, SD plus SG (CEB-90-90), Log-spiral theory


80
2.0 50.80
Measured
FEM without shrinkage
1.6 40.64
Abutment Displacement (in)

Abutment Displacement (mm)


FEM with CEB-90-99 model

Contraction
1.2 30.48

0.8 20.32

0.4 10.16
Expansion

0.0 0.00

-0.4 -10.16

2003 2004 2005 2006 2007 2008 2009 2010 2011


Time

Figure 4.18 Case (4) TD, SD (CEB-90-90), Log-spiral theory

Figure 4.19 represents case (5), in which a full temperature change is assumed for the bridge deck

and half of the temperature change for the girders. This attempt aims at examining further potential

distribution of the temperature across the bridge superstructure. In this case, it can be noted that

the computed displacement amplitude closely matches the experimental values, indicating a more

optimized distribution for temperature across the superstructure. However, even though

displacement response coincides with experimental response in the early stages of monitoring,

there is an obvious shift upwards in the later monitoring years. Moreover, the shrinkage addition

seems to overestimate the overall displacement.

81
2.0 50.80
Measured
FEM without shrinkage
1.6 40.64
Abutment Displacement (in)

Abutment Displacement (mm)


FEM with CEB-90-99 model

Contraction
1.2 30.48

0.8 20.32

0.4 10.16
Expansion

0.0 0.00

-0.4 -10.16

2003 2004 2005 2006 2007 2008 2009 2010 2011


Time

Figure 4.19 Case (5) TD plus 50% TG, SD plus SG (CEB-90-90), Log-spiral theory

4.9.2 Rankine Case

As a further step towards refining the FE response versus the experimental response, five more

cases were also investigated using Rankine earth pressure theory to capture the behavior of SR-

18. The cases were indicated in the calibration matrix with almost the same configuration as the

log-spiral theory cases, but with greater focus on the promising temperature scenario across the

superstructure section, as was concluded from the previous section. Figure 4.20 (case 6) shows the

displacement response with temperature given for both the bridge deck and the girders. The

amplitude of the displacement remains higher than the experimental displacement. The shrinkage,

however, shifts the curve upward in this case. Figure 4.21, which represents case (7), also indicates

that the temperature gradient is more representative of the experimental data in terms of amplitude

82
height. The displacements tend to match experimental data in the early stages but deviate at later

stages. In general, this case and case (5) explain the distribution of temperature over the

superstructure height by giving full temperature change for the deck and half to the girders; such

an approach yielded good results in comparison to experimental results. As such, this distribution

was maintained in the subsequent cases while continuing to change other parameters, to obtain a

more refined representation of the experimental behavior.

2.0 50.80
Measured
FEM without shrinkage
1.6 40.64
Abutment Displacement (in)

Abutment Displacement (mm)


FEM with CEB-90-99 model
Contraction

1.2 30.48

0.8 20.32

0.4 10.16
Expansion

0.0 0.00

-0.4 -10.16

2003 2004 2005 2006 2007 2008 2009 2010 2011


Time

Figure 4.20 Case (6) TD, TG, SD plus SG (CEB-90-90), Rankine theory

83
2.0 50.80
Measured
FEM without shrinkage
1.6
Abutment Displacement (in)
40.64

Abutment Displacement (mm)


FEM with CEB-90-99 model

Contraction
1.2 30.48

0.8 20.32

0.4 10.16
Expansion

0.0 0.00

-0.4 -10.16

2003 2004 2005 2006 2007 2008 2009 2010 2011


Time

Figure 4.21 Case (7) TD plus 50% TG, SD plus SG (CEB-90-90), Rankine theory

Figure 4.22 (case 8) represents a case under the same conditions of the previous case (case 7) but

here shrinkage is only given to the deck. This case seemed to better match the experimental

response than all the previous trials. It is obvious that girder shrinkage has a negligible influence

over the bridge response. It could be thought of as if the girders have consumed all the shrinkage

stain during the curing period before coming to be mounted at the bridge location. The deck

shrinkage, however, proved to be the major determinant the displacement response.

Figure 4.23 (case 9) shows the response of the abutment using the CEB-90-90 model as a shrinkage

substitute of the previous case. In this case, the behavior of the abutment also matches experimental

behavior reasonably well. Shrinkage strains herein satisfactorily pulled the computed curve to the

84
desired bridge monitored counterpart. The same results were almost obtained for the ACI-209

shrinkage model as shown in Figure 4.24 (case 10), in which the displacement curve coincides

fairly well with experimental behavior with a slight deviation in the response at the beginning of

the bridge monitoring.

2.0 2.0
Measured
FEM without shrinkage
1.6 1.6
Abutment Displacement (in)

Abutment Displacement (mm)


FEM with CEB-90-99 model
Contraction

1.2 1.2

0.8 0.8

0.4 0.4
Expansion

0.0 0.0

-0.4 -0.4

2003 2004 2005 2006 2007 2008 2009 2010 2011


Time

Figure 4.22 Case (8) TD plus 50% TG, SD (CEB-90-90), Rankine theory

85
2.0 50.80
Measured
FEM without shrinkage
1.6 40.64
Abutment Displacement (in)

Abutment Displacement (mm)


FEM with CEB-90-90 model

Contraction
1.2 30.48

0.8 20.32

0.4 10.16
Expansion

0.0 0.00

-0.4 -10.16

2003 2004 2005 2006 2007 2008 2009 2010 2011


Time

Figure 4.23 Case (9) TD plus 50% TG, SD (CEB-90-90), Rankine theory

2.0 50.80
Measured
FEM without shrinkage
1.6 40.64
Abutment Displacement (in)

Abutment Displacement (mm)


FEM with ACI-209 model
Contraction

1.2 30.48

0.8 20.32

0.4 10.16
Expansion

0.0 0.00

-0.4 -10.16

2003 2004 2005 2006 2007 2008 2009 2010 2011


Time

Figure 4.24 Case (10) TD plus 50% TG, SD (ACI-209), Rankine theory

86
It can be concluded that the CEB-90-90 and ACI-209 models best represent shrinkage effects

on concrete IABs. The Rankine earth pressure theory was found to be a reasonable model

representing the influence of the soil behind abutments, more so than the log-spiral theory.

Furthermore, temperature distribution has proved to be more effective in capturing the real

behavior than assuming a uniform distribution across the superstructure section. The distribution

considered in the current study is very simple and easily applicable in analysis and design, which

helps at avoiding sloping gradients that complicate the implementation of finite element models.

The optimal temperature distribution was easily established by assuming full temperature change

for the bridge deck and half the temperature change for the girders.

As further verification of the abutment response, the computed lateral earth pressure is compared

with its measured experimental counterpart. Measured earth pressure was collected from the

pressure cells mounted at the back of the abutment at the location of the longitudinal displacement

measurement point (15 in from the base of the abutment). Computation of the earth pressure at

the point of interest (earth pressure location) is conducted by forcing the mesh size to coincide

with that point. The pressure is given as the force in the spring divided by the tributary area for

that spring. The force in the spring is computed from the displacement of that point (node)

projected at the force-displacement curve, which is extracted from LPILE.

Figure 4.25 shows the earth pressure behind the abutment for the two theories, Rankine and log-

spiral. As can be seen here, the Rankine theory yielded results closer to the measured earth pressure

compared to the log-spiral theory. There is a slight shift in the computed pressure even in Rankine

theory, which might be attributed to the gap which generates between the abutment wall and the

backfill over the contraction phase during thermal hysteretic cycling. In general, it can be
87
concluded that Rankine theory is the best representative model for the backfill soil. This further

validates what has been concluded previously regarding Rankine theory. The log-spiral theory

yielded higher pressure values and therefore prevented the abutment from necessary displacement,

permitting the bridge response in general to move more inward (shorten), and elevating the

displacement curve above the experimental behavior.

Computed (Rankine)
4000 Computed (Log-Spiral) 191.5
Measured

Earth Pressure (kN/m2)


Earth Pressure (psf)

3000 143.6

2000 95.8

1000 47.9

0 0.0

2003 2004 2005 2006 2007 2008 2009 2010 2011

Time

Figure 4.25 Calculated vs Measured earth pressure behind abutment

4.10 Summary and Conclusions

The FE modelling of a prestressed concrete IAB showed very good agreement with field

monitoring behavior over an instrumentation period of seven years. All bridge components were

modeled using 3D solid finite elements and nonlinear soil springs for the soil surrounding the

88
piles and behind the abutment. Soil springs were generated using p-y curves which are computed

by software LPILE. Abutment springs were incorporated into the model through digitized Clough

and Duncan curves to represent soil passive earth pressure against bridge expansion. A Python

code was developed to attach soil springs to the FE model. Three shrinkage models were used to

include shrinkage strain effects in the bridge model. The shrinkage strains were converted into

temperature change equivalents and applied to the deck and girders of the bridge. Abutment

displacements were calibrated against experimental counterparts to come up with the best loading

and boundary conditions affecting the behavior. Soil pressure behind abutments is also examined

for additional validation. The following conclusions can be drawn from this study:

1. The CEB-90-90 shrinkage model is the most representative model for concrete behavior

over the monitoring time.

2. The deck contributes substantially to shrinkage effects on the concrete in IAB, however,

the prestressed girders have little or no shrinkage influence on the bridge response.

3. The Rankine earth pressure theory outperforms other theories in capturing the abutment-

backfill interaction behavior.

4. The distribution of the temperature along the depth of the superstructure is important in

simulating the displacement amplitude of the abutment; the best simple distribution over

the entire depth was found by giving full temperature change to the deck and half of the

temperature change to the girders. Other distributions either over- or underestimate the

displacement amplitude.

5. Abutment displacement ratcheting (deck shortening) over time occurs due to the soil

restraining effect to the abutment during the expansion phase of the bridge. The

displacement induced in the passive case (bridge expansion) is less than the displacement

89
in the active case (bridge contraction); this disparity in displacements leads to an

accumulative difference in the net response which is responsible for the overall inward

shortening of the bridge.

The study contributes towards understanding the behavior of IABs, and also provides a sound

evaluation of modelling techniques in parametric studies and in designing various IAB

components. The study is limited to prestressed concrete IABs. Other bridges like steel stringer

IABs or concrete IABs require additional investigation towards this end.

90
5. PARAMETRIC STUDY

5.1 Introduction

This chapter presents a parametric study for a concrete integral abutment bridge supported by steel

HP piles to investigate the effect of bridge length and pile size, orientation, soil type and stiffness

on limiting maximum values for IAB design parameters: pile displacement, abutment

displacement, pile stresses, deck and girder stresses. A three-dimensional nonlinear finite element

models have been built using software ABAQUS utilizing outcomes of the calibration procedure

presented in the previous chapter. The backfill soil behind the abutments and the soil around the

piles are modelled using nonlinear springs. Results of the parametric study are presented and

discussed, and conclusions are drawn.

91
5.2 Parametric Study Base Case

The base case which has been chosen for the current parametric study is bridge No. 211 located in

central Pennsylvania (Kim and Laman, 2012) and shown in Figure 5.1. The bridge is a single span

IAB with a total length of 34.7m (114 ft) and 13.8m (45 ft) in width. The deck slab is CIP concrete

and supported by four PennDOT 28/78 prestressed concrete girders. Figure 5.2 shows the layout

and section of the bridge. Girder dimensions are shown in Figure 5.3 and Table 5.1. Abutments

are 4.3 m (14’1”) in height and 0.914 m (36 in) in thickness and supported by 11 HP12x74 steel

piles spaced at 1.24 m (49 in) oriented on the weak axis bending. This bridge was chosen because

of the high single-span length and the available plans and details.

Figure 5.1 Bridge 211 in central Pennsylvania (Kim and Laman, 2012)

92
a. Bridge 211 layout

b. Section through bridge 211

Figure 5.2 Bridge 211 a. bridge layout b. section through bridge

93
Figure 5.3 PennDOT 28/78 prestressed concrete girder section (PennDOT)

TABLE 5.1 PennDOT 28/78 PRESTRESSED GIRDERS DIMENSIONS (PennDOT) (ALL


DIMENSIONS ARE IN INCHES)

Beam Designation W2 W1 D W3 A (in2) I (in4)


28/78 42 28 78 8 1133 898984

94
5.3 Selection of Parameters

Four parameters were selected for the current study: bridge length, pile section, pile orientation,

and soil type around the pile as shown in Table 5.2. Single span length of all bridges was taken as

100 ft. to the base case in the previous section. Bridges with one, two, three, and four spans were

considered to study the effect of length on the key design parameters. Three pile sections with two

orientations, weak and strong, were also considered. Two soil types around the pile were

investigated: sand and clay, with two stiffness values each. The soil behind the abutment was kept

constant throughout the study and set to be medium sand. Bridge width, thickness, and abutment

height were also kept constant throughout the study and were equal to 13.7 m (45 ft.), 200 mm (8

in), and 4.27 m (14 ft.), respectively. Table 5.3 shows the properties of HP sections used in the

parametric study.

TABLE 5.2 PARAMERTERS USED IN THE STUDY

Parameter Value
Bridge length 100, 200, 300,400
HP10x57
Pile section HP12x74
HP14x102
Pile orientation Weak, Strong
Clay (stiff, soft)
Soil type
Sand (dense, medium)

95
TABLE 5.3 HP SECTIONS PROPERTIES

Depth Flange Width Flange Web Thickness Area


Pile (in) Thickness
bf (in) tw (in) (in2)
tf (in)
HP10x57 10.0 10.2 0.565 0.565 16.7

HP12x74 12.1 12.2 0.61 0.605 21.8

HP14x102 14.0 14.8 0.705 0.705 30.1

5.4 Soil Properties Used in the Parametric Study

A medium sand was assumed as the backfill material behind abutments with an angle of internal

friction of 35° and a unit weight of 100 lb./ft3. Backfill springs were computed based on Clough

and Duncan (1991) curves and were digitized to the desired angle of internal friction as in the

previous chapter. Figure 5.4 shows the nonlinear force-displacement curves for the backfill soil

that were used in the FE models throughout the parametric study.

Four soil types around the piles were considered: medium sand, dense sand, soft clay and stiff clay.

Their properties were estimated based on typical values in references and the Lpile manual and are

shown in Table 5.4. Lpile software was used to compute the pressure-displacement relationships

and then converted into force-displacement data by multiplying by the tributary length (distance

between springs) and inputted into ABAQUS software by using a connector element (Refer to

Chapter 4 for the procedure of calculating abutment and pile springs).

96
TABLE 5.4 SOIL PROPERTIES AROUND THE PILES

Soil Type C (lb/ft2)   (lb/ft3) 50 k (lb/in3)


Medium Sand - 35 100 - 90

Dense Sand - 40 115 - 225

Soft Clay 250 - 80 0.02 30

Stiff Clay 2000 - 108 0.005 500


Passive Resistance, (lbs)

Figure 5.4 Load-displacement of the backfill soil

The range of temperatures that was considered for the models is based on AASHTO 2012

specifications. The minimum and the maximum design temperature for concrete bridges are 10°F

(-12°C) and 80°F (27°C), respectively, for moderate climate. For cold climates, the minimum and

the maximum design temperature for concrete bridges are 0°F (-18°C) and 80°F (27°C),

97
respectively. To study the maximum effects of the temperature variations on the IA bridges, cold

climate temperatures were chosen in this parametric study. The mean temperature was taken as the

construction temperature in summer which will be 80°F (27°C) for cold climate. As the

temperature drops to zero during winter, the temperature increment T will be 80°F (27°C). As

such, the temperature change will be 80°F and -80°F for the expansion and contraction phases,

respectively. The temperature distribution along the superstructure was taken according to chapter

4 calibration process, by giving full temperature change for the deck and half temperature change

for the girders.

5.5 Finite Element Model Setup

All bridge components were modelled using an 8-node linear reduced integration with hour glass

control solid elements (C3D8R) for both concrete and the steel components. The same finite

element assumptions that were considered previously in the calibrations process were also

implemented in the current analysis. Linear elastic behavior was assumed for both materials as the

displacement range is not too high under thermal bridge movement.

Full composite action was assumed between girders and the deck and between the pile head and

the abutment bottom by enforcing a (TIE constraint) between them. Lateral displacement along

the bridge length was accounted for by providing girder supports in multi-span bridges having a

free translation degree of freedom along bridge length while restraining all other degrees of

freedom to impose maximum effect on the pile-abutment system. Piles were assumed fully fixed

at the bottom by restraining all degrees of freedom. Soil springs were attached to the abutment

every 0.3 m (1 ft) apart horizontally and vertically. In case of the piles, soil springs were attached

98
evenly at 1 ft apart along its depth. Guidelines for attaching the soil springs to the FE models are

presented in Chapter 3. Python code was used to attach soil springs to the abutments and the piles

to save modelling time. T A typical FE model of the bridge is shown in Figure 5.5. Dead load of

the bridge was included in analysis through the “GRAVITY LOAD” option using the density of

the bridge component materials. Temperature load was applied to the deck and the girders through

the “PREDEFINED FIELDS” option. Full temperature difference (80°F) was applied to the deck

and half of the temperature to the girders to account for the temperature gradient across the

superstructure utilizing the outcomes of chapter 3. Bridge deck and abutments were meshed with

a 12-in element size while other bridge components (girders and piles) were meshed with 6-in

element size to capture response accurately in small locations. The analysis was performed using

the “STATIC” option with geometric nonlinearity.

Figure 5.5 Typical bridge FE model used in the parametric study

99
5.6 Effect of Bridge Length

Figure 5.6 shows substructure displacement profile for four bridge span lengths: 100, 200, 300,

and 400 ft. HP10x57 with weak axis bending was selected to investigate the response. Two phases

of bridge movement were studied: expansion and contraction. It is obvious from the shape of the

figure that the displacements are almost identical for the expansion and contraction phases for all

bridge lengths. The figure also indicates that the depth of the equivalent cantilever of the pile

increases as the bridge span increases; however, the difference in this depth for various lengths is

insignificant. This figure also validates the use of only one bridge phase (expansion or contraction)

to investigate all responses without repetition. The expansion phase was chosen in the current

study for comparing various bridge responses. The abutment displacement to pile head

displacement ratio decreases as the bridge length increases. This ratio is a function of soil type

behind abutment, abutment height, soil pile, size and orientation of the pile.

100
Contraction Expansion
0

Distance from Abutment top (in)


Abutment

-100

-168

-200
Pile
Bridge Length =100 ft (Exp)
Bridge Length =100 ft (Cont)
Bridge Length =200 ft (Exp)
-300 Bridge Length =200 ft (Cont)
Bridge Length =300 ft (Exp)
Bridge Length =300 ft (Cont)
Bridge Length =400 ft (Exp)
Bridge Length =400 ft (Cont)
-400
-0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8
Lateral Displacement (in)

Figure 5.6 substructure response considering various bridge span lengths.

To study the effect of bridge length on design criteria, three pile sizes with two orientations, weak

and strong, were examined to investigate the effect of bridge length on critical parameters, which

include maximum values of abutment displacement, pile displacement, pile stresses, deck stresses,

and girder stresses. Figure 5.7 shows the relationship between bridge length and abutment

displacement for three pile sizes: HP10x57, HP12x74, and HP14x102, with two axis orientations.

One soil (dense sand) is chosen around the pile to focus on pile stiffness. The figure indicates that

pile size and orientation have an insignificant effect on the abutment displacement. This is

attributed to the small stiffness of the pile compared to overall bridge stiffness. To investigate the

effect of soil stiffness around the pile on abutment displacement, one pile (HP10x57) was selected

with two axis orientations to study the response. Four soil types were studied: dense sand, medium

sand, stiff clay and soft clay. Figure 5.8 and 5.9 show the effect of soil types around the pile on

101
abutment displacement considering four bridge lengths for expansion and contraction phases,

respectively. The figures indicate that neither the soil type nor stiffness have a pronounced effect

on the abutment response for the same reason given above.

It is obvious from the figures that the change in the abutment displacement follows a linear trend

for both of expansion and contraction, which validates the employment of a linear thermal

expansion equation in estimating lateral thermal bridge displacement. Linear thermal expansion of

the abutment is computed from the following equation and is presented in Figures 5.8 and 5.9:

2=  L T/2 5.1

Where 2: is the displacement of the bridge abutment.

L: total length of the bridge.

coefficient of thermal expansion which for normal weight concrete equals 6.0 × 10 –6/°F (10 ×
10–6/°C)
A temperature change (T) was taken as 80°F as in the FE model. It is obvious that the

displacement computed by the equation is higher than the bridge displacement computed by the

finite element model, which is attributed to the restraining effect of the abutment and soil to bridge

expansion in the finite element model while in the thermal expansion equation, the length is free

to expand. The difference between the equation and the FE model displacements is very small at

short bridge lengths, although the difference significantly grows larger as the bridge length

increases.

102
1.0
H10x57 Weak axis
0.9 H12x74 Weak axis
H14x102 Weak axis

Abutment Displacement (in)


0.8 H10x57 Strong axis
H12x74 Strong axis
0.7 H14x102 Strong axis

0.6

0.5

0.4

0.3

Dense Sand
0.2
100 150 200 250 300 350 400
Bridge Length (ft)

Figure 5.7 Abutment displacement vs bridge length for different pile sections and orientations

Dense sand weak axis


1.2 Medium Sand weak axis
Soft Clay weak axis
Stiff Clay weak axis
Abutment Displacement (in)

Dense sand strong axis


1.0
Medium Sand strong axis
Soft Clay strong axis
Stiff Clay strong axis
0.8 Thermal Expansion equation

0.6

0.4

0.2
100 150 200 250 300 350 400
Bridge Length (ft)

Figure 5.8 Abutment displacement vs bridge length for HP10x57 and weak axis orientation
(bridge expansion)

103
-0.3

-0.4
Abutment Displacement (in)

-0.5

-0.6

-0.7

-0.8

-0.9
Dense Sand
-1.0 Medium Sand
Stiff Clay
-1.1 Soft Clay
Thermal Expansion Equation
HP10x57
-1.2
100 150 200 250 300 350 400
Bridge Length (ft)

Figure 5.9 Abutment displacement vs bridge length for HP10x57 and weak axis orientation
(bridge contraction)

Table 5.5 presents normalized abutment displacements (1/2), where 1 is the abutment

displacement computed by the finite element model. To correlate the displacements calculated by

both methods, normalized displacements are plotted against bridge lengths in Figures 5.10 for the

expansion phase. Curve fitting was used to find a power-based equation and a good correlation is

achieved compared to the observed response (R2=0.97). This equation was used to develop an

empirical equation to quantify the abutment displacement based on a simple thermal expansion

equation.

104
TABLE 5.5 NORMALIZED DISPLACEMENTS VERSUS BRIDGE LENGTHD

Bridge Displacement in Displacement Normalized Displacements


Length (ft) Expansion (FE) 1 (Equation) (1/2)
100 0.258 0.288 0.895833
200 0.415 0.576 0.720486
300 0.5823 0.864 0.6739583
400 0.74 1.152 0.642361

0.95

0.9

0.85

0.8
1/2

0.75 y = 2.6713x-0.241

0.7

0.65

0.6
0 50 100 150 200 250 300 350 400 450
Bridge Length (ft)

Figure 5.10 Normalized bridge displacement versus bridge length (bridge expansion)

The following simple equation is suggested to compute abutment displacement from the total

bridge length. Figure 5.11 plots the proposed equation against the finite element results.:

  1.3  T L0.76 5.2

105
1.2 Finite Element Model
Thermal Expansion equation
1.1 Proposed Equation

Abutment Displacement (in)


1.0

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2
100 150 200 250 300 350 400
Bridge Length (ft)

Figure 5.11 proposed equation compared to finite element model results

Figure 5.12 depicts pile displacement versus four bridge lengths considering three pile sizes and

two orientations. The figure shows that neither the pile section nor the pile orientation has a

significant impact on the pile top displacement. The displacements are almost the same for all pile

sections and orientations.

Figure 5.13 shows the bridge length versus the maximum pile displacement using only one pile

section (HP10x57) for the investigation. The figure indicates that the pile displacement increases

as the bridge length increases, which is reasonable given that the lateral displacement is

proportional to the bridge length. The figure also reveals that the soil stiffness has more impact

than the pile orientation on the pile displacement. This can be observed through the fact that the

upper and lower values for the pile displacements are achieved by the soft and stiff clay,

106
respectively. Also, the difference between the weak and strong axis orientation for the dense sand

is very small, which means that the pile orientation has only a superficial influence on the pile

displacement, whereas the soil stiffness plays a vital role in controlling the pile displacement. In

general, the weak axis orientation exhibited higher pile displacement than strong axis. The

difference in the pile displacement is unnoticeable within small bridge lengths; however, as the

bridge length increases, the differences increase as well.

0.35 H10x57 Weak axis


H12x74 Weak axis
0.30 H14x102 Weak axis
H10x57 Strong axis
Max Pile Displacement (in)

0.25 H12x74 Strong axis


H14x102 Strong axis
0.20

0.15

0.10

0.05

0.00

-0.05
100 150 200 250 300 350 400
Bridge Length (ft)

Figure 5.12 Max pile displacement vs bridge length for various pile sections and orientations
(bridge expansion)

107
Dense Sand Weak
0.35
Medium Sand Weak
Soft Clay Weak
0.30 Stiff Clay Weak
Dense Sand Strong
Max Pile Displacement (in)
0.25

0.20

0.15

0.10

0.05

0.00
HP10x57
-0.05
100 150 200 250 300 350 400
Bridge Length (ft)

Figure 5.13 Max pile displacement vs bridge length for HP10x57 and weak axis orientation
(bridge expansion)

In Figure 5.14, a comparison is made to investigate the effect of pile size and orientation on the

pile stresses for different bridge lengths. The Von-Mises stresses are used for the comparison as

the case involves a three-dimensional state of stress. The figure indicates that the stresses increase

as bridge length increases. It is obvious that for the same section, the weak axis suffers higher

stresses than the strong axis. Moreover, piles with small sections sustained higher stresses than

larger sections. The difference between the weak and strong axis orientation is very small.

To obtain more insight about the development of pile stress for different soil types, one pile section

(HP10x57) was selected to investigate the relationship between the bridge length and the pile axial

stresses, considering four soil types and two pile orientations as shown in Figure 5.15. As the

bridge length increases, the stresses (Von-Mises) increase correspondingly. The stiff clay with

108
weak axis orientation exhibited higher stresses, followed by the dense and medium sands with

weak axis orientation. The minimum stresses were observed with the soft clay having strong axis

orientation. It is obvious that pile axial stresses are proportional to the soil stiffness and inversely

proportional to the pile stiffness. The reason behind increased pile stresses as soil stiffness

increases is likely the increased restraining effect of the soil to the confined pile which accelerates

local flange buckling and a consequent spike in pile stresses.

30000
H10x57 Weak axis
H12x74 Weak axis
25000 H14x102 Weak axis
H10x57 Strong axis
Max Pile Axial Stress (psi)

H12x74 Strong axis


20000 H14x102 Strong axis

15000

10000

5000

HP10x57
0
100 150 200 250 300 350 400
Bridge Length (ft)

Figure 5.14 Max pile axial stress vs bridge length for various pile sections and orientations
(bridge expansion)

109
35000 Dense Sand Weak
Medium Sand Weak
Stiff Clay Weak
30000
Soft Clay Weak
Dense Sand Strong
Max pile stress (psi)

25000

20000

15000

10000

100 150 200 250 300 350 400


Bridge Length (ft)

Figure 5.15 Max pile axial stress vs bridge length for HP10x57 (bridge expansion)

Figure 5.16 depicts the relationship between the bridge length and the maximum deck stresses.

The figure indicates that as the bridge length increases, the stresses slightly increase. The stresses

maintained an almost constant value of 1200 psi, which indicates that soil stiffness has only a small

influence on the deck stresses. However, the figure reveals that stiff clay induced higher deck

stresses than soft clay, which means that the reduction of soil stiffness around the pile can

contribute slightly to mitigating deck stresses.

110
1400

Max Deck Stresses (psi)

1200

Dense Sand Weak


Medium Sand Weak
Stiff Clay Weak
Soft Clay Weak
Dense Sand Strong HP10x57
1000
100 150 200 250 300 350 400
Bridge Length (ft)

Figure 5.16 Max deck axial stress vs bridge length for HP10x57 and weak axis orientation
(bridge expansion)

Figure 5.17 depicts the relationship between the bridge length and the maximum girder stresses.

The same trend is evident here for girder stresses as was seen for deck stresses. Soil with higher

stiffness resulted in higher stresses whereas the lower soil stiffness caused minimum stresses. The

stiffness of the soil plays a vital role in mitigating the girder stresses. Furthermore, predrilled holes

appear to be a very effective solution for flexible design of IABs. Controlling deck and girder axial

stresses is important in preventing the incidence of cracks.

111
3000

2500

Max Girder Stresses (psi)


2000

1500

1000
Dense Sand Weak
Medium Sand Weak
500 Stiff Clay Weak
Soft Clay Weak
HP10x57 Dense Sand Strong
0
100 150 200 250 300 350 400
Bridge Length (ft)

Figure 5.17 Maximum girder axial stress vs bridge length for HP10x57 and weak axis orientation
(bridge expansion)

5.7 Effect of Bridge Length on Pile Stresses and Displacements Profiles

To investigate the effect of bridge length on pile stresses and displacement profiles through pile

depth, one pile section (HP10x57) with two orientations was selected with four bridge lengths:

100, 200, 300, and 400 ft. Figure 5.18 shows the effect of bridge length on the displacement profile

of HP10x57 for both weak and strong axis orientations. Medium sand was selected as the soil

around the pile for this comparison. The figure indicates that pile with weak axis orientation has a

slightly higher displacement than the strong axis orientation. The effect increases as the bridge

length increases. The depth of the equivalent cantilever is almost constant for different bridge

lengths and pile orientations.

112
Figure 5.19 presents the pile stress profile for the same pile section and orientations and four bridge

lengths. In general, the stresses increase with bridge length for both orientations. The weak axis

exhibits higher stresses than strong axis orientation along the pile depth, which is attributed to the

low stiffness of the weak axis as compared to the strong axis. The maximum stress occurs at the

second segment of the pile (bounded by the first inflection point and the second infection point).

The maximum stress does not arise in the pile-abutment interface because of the rotation of the

abutment which provide some release of fixity and reduced stresses at the pile head.

0
Distance from pile top (ft)

-5

-10
Weak Medium Sand 100
Weak Medium Sand 200
-15 Weak Medium Sand 400
Strong Medium Sand 100
Strong Medium Sand 200
Strong Medium Sand 400
-20
-0.05 0.00 0.05 0.10 0.15 0.20 0.25 0.30
Displacement (in)

Figure 5.18 Displacement profile for H10x57 and medium sand for various bridge lengths and
pile orientations

113
0

Distance from Pile top (ft)


-5

-10

Weak Medium Sand 100


Weak Medium Sand 200
-15 Weak Medium Sand 400
Srong Medium Sand 100
Strong Medium Sand 200
Strong Medium Sand 400
-20
0 2000 4000 6000 8000 10000
Von-Mises Stress (psi)

Figure 5.19 Von-Mises stress profile for H10x57 and medium sand for various bridge lengths
and pile orientations

5.8 Effect of Soil Type on Pile Stresses and Displacements Profiles

Figure 5.20 shows the effect of soil type on the displacement profile of HP10x57 for both weak

and strong axis orientations. A bridge length of 200 ft was used herein for the comparison. Four

soil types are examined: dense sand, medium sand, soft clay and stiff clay. The shape of the

displacement profile is almost identical for all soil types with a slight difference between them

(0.01 in). The soft clay resulted in the highest displacement at the pile head relative to the other

soil types.

114
0

Distance from pile top (ft)


-5

Weak Dense Sand


-10 Weak Medium Sand
Weak Soft Clay
Weak Stiff Clay
-15 Strong Dense Sand
Strong Medium Sand
Strong Soft Clay
Strong Stiff Clay
-20
-0.02 0.00 0.02 0.04 0.06 0.08 0.10
Displacement (in)

Figure 5.20 Displacement profile for H10x57 for various soil types and weak axis orientation
(Bridge length = 200 ft)

0
Distance from pile top (ft)

-5

-10
Weak Dense Sand
Weak Medium Sand
Weak Soft Clay
Weak Stiff Clay
-15 Strong Dense Sand
Strong Medium Sand
Strong Soft Clay
Strong Stiff Clay
-20
0 1000 2000 3000 4000 5000 6000
Von-Mises Stress (psi)

Figure 5.21 Max Von-Mises stress profile for H10x57 for various soil types and weak axis
orientation (Bridge length = 200 ft)

115
Figure 5.21 shows the effect of soil type on the stress profile of HP10x57 for both weak and strong

axis orientations and for a bridge length of 200 ft. The figure shows that the pile with weak axis in

a stiff clay soil has the lowest stresses at the pile head and maximum stresses in the first segment

of the pile. Higher stresses are induced in pile heads embedded in soft clay. In general, maximum

stresses are induced in the first segment of the pile with the weak axis having the higher stresses.

5.9 Effect of Pile Size on Pile Stresses and Displacements Profiles

Three pile sections were selected to investigate the effect of pile size on displacement and stress

profiles. Axis orientation and soil type were kept constant and selected to be weak axis and soft

clay. Bridge length was also kept constant to concentrate on the effect of pile size.

Figure 5.22 indicates that the pile size has a negligible effect on the displacement profile, since the

pile head displacement is almost the same for all pile sizes. However, a very small difference is

evident between pile sizes for the below-ground displacements.

Figure 5.23 shows the stress distribution along pile depth for the three pile sizes. The smaller pile

size caused higher stresses and the large section induced the lower stresses. The shape of the stress

distribution is identical for all three pile sizes.

116
0

Distance from pile top (ft)


-5

-10

-15
H57WSC400
H74WSC400
H102WSC400
-20
-0.05 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
Displacement (in)

Figure 5.22 Displacement profile for H10x57 for soft clay and weak axis orientation

0
Distance from Pile top (ft)

-5

-10

-15
H57WSC400
H74WSC400
H102WSC400
-20
0 1000 2000 3000 4000 5000 6000 7000
Von-Mises Stress (ksi)

Figure 5.23 Von-Mises stress profile for H10x57 for soft clay and weak axis orientation

117
5.10 Conclusions

The following conclusions are drawn from the following chapter:

1. The change in the abutment displacement follows a linear trend, which validates the

employment of a linear thermal expansion equation in estimating lateral thermal bridge

displacement.

2. The difference between the thermal expansion equation and the FE model displacements

is very small with short bridge lengths; however, the difference grows significantly larger

as bridge length increases.

3. Neither the pile size or pile orientation nor the soil type or soil stiffness have a significant

impact on the abutment displacement.

4. The soil stiffness has more impact than the pile orientation on the pile displacement.

5. The pile stresses are proportional to the soil stiffness and inversely proportional to the pile

stiffness.

6. Weak axis orientation leads to higher stresses than strong axis orientation along the pile

depth, which is attributed to the small stiffness of the weak axis as compared to the strong

axis. The maximum stress occurs at the second segment of the pile (bounded by the first

inflection point and the second infection point). The maximum stress does not appear in

the pile-abutment interface because of the rotation of the abutment which provide some

release of fixity and reduced stresses at the pile head.

7. The soil stiffness has a small influence on the deck stresses. However, the reduction of soil

stiffness around the pile can contribute somewhat to mitigating deck stresses.

118
8. Reducing the stiffness of the soil around the pile can contribute to mitigating stresses on

the bridge girders.

119
6. PILE BUCKLING ANALYSIS OF HP PILES IN INTEGRAL ABUTMENT BRIDGES

6.1 Introduction

This chapter presents a numerical study calibrated to experimental tests to investigate the lateral

displacement capacity of HP piles in integral abutment bridge subjected to combined axial load

and cyclic displacement amplitude. Eleven steel HP sections with two axis orientations (weak and

strong) commonly used in IAB construction practice were investigated. Several soil types

surrounding the piles were examined to investigate the effect of soil stiffness on the displacement

capacity. Based computed pile displacement capacities, maximum length of integral abutment

bridges has been estimated and compared with current practice of US departments of

transportation.

120
6.2 Finite Element Model Setup

A nonlinear finite element (FE) models are established for the piles of IABs utilizing the

commercial software ABAQUS CAE v.14. Quadratic four-node doubly curved shell elements

(S4R) with reduced integration are used throughout this study to discretize the pile models. This

element was used herein to save in the computational running time of the analysis, as it is a very

intensive problem. Also, it has been widely implemented in the buckling analysis of columns under

cyclic loading (Fogarty and Al-Tawil, 2016; Elkadi and Lignos, 2015). A size of (25x25) mm is

used to mesh the models based on a sensitivity study considering a reasonable computational time.

This mesh size had also been validated by other researchers for similar modelling problems (Elkadi

and Lignos, 2015).

A constitutive nonlinear model based on Lemaitre and Chaboche (1990) is used to simulate the

cyclic plasticity of the steel material. The model adopts Von Mises yield criterion and associated

flow rule and assumes combined isotropic/kinematic hardening behavior. Two parts characterizes

this model: The isotropic part and the kinematic part. The isotropic part defines the change in the

yield surface size  and the equivalent plastic strain  , as in the following equation:
0 p

p
 0   |0 Q (1 eb  ) 6.1

where  |0 represents the yield stress when the equivalent plastic strain is zero, Q is the maximum

change in the yield surface size and b is the rate of change of the yield surface size. The kinematic

part of the model defines the change of backstress  , and is given by:

1
  C  p (   )     p 6.2
 0

121
Where C: is the initial kinematic hardening modulus; is the rate at which the backstress

changes with increasing plastic strain  . The maximum change in the backstress in limited by
pl

the ratio of C/. Figure 6.1 and 6.2 show the relation between the backstress and the isotropic

hardening for uniaxial and multiaxial loading, respectively (ABAQUS manual).

Figure 6.1 One-dimensional representation of the nonlinear model (ABAQUS manual)

122
Figure 6.2 Three-dimensional representation of the nonlinear model. (ABAQUS manual)

To this end, an auxiliary FE model is developed to calibrate the hardening parameters of the steel

material under cyclic displacement. A steel grade of ASTM A36 is used throughout this study

because of its wide popularity in construction of most IABs in North America (Dicleli and

Albhaisi, 2003). The stress-strain response of ASTM A36 steel grade used in the FE models is

calibrated against the cyclic coupon tests conducted by (Kaufman et al., 2001) and a reasonable

match is achieved as shown in Figure 6.1. The calibrated cyclic parameters were found to be as

follows: the initial kinematic hardening modulus is 500, the rate at which the backstress changes

is 10, the maximum change in the yield surface size is 8, the rate of change of the yield surface

size is 2. These parameters have been reported on the same range by other researchers (Elkady and

Lignos, 2015; Wigle and Fahnestock, 2010)

123
100 689.5
Experimental
80 551.6
FE Model
60 413.7

40 275.8

Stress (ksi)

Stress (Mpa)
20 137.9

0 0.0

-20 -137.9

-40 -275.8

-60 -413.7

-80 -551.6

-100 -689.5
-0.04 -0.02 0.00 0.02 0.04

Strain

Figure 6.3 Calibration of the of the material cyclic response


(Experimental data from Kaufman et al. 2001)

6.3 Validation of the FE model

Experimental testing conducted by Frosch et al. (2004) on steel HP stub columns were used to

validate the FE modeling. Figure 6.2 shows schematic diagram of the experimental testing.

Two specimens HP10x42 and HP12x53 tested about their weak axis were modelled using the

commercial software ABAQUS CAE. Pile model in IAB is typically divided into three segments

as shown in Figure 6.3. The top segment is bounded by the pile-abutment interface and the first

inflection point; the middle segment is bounded by the first inflection point and the second

inflection point; the third segment is the bottom segment and lie between the second inflection

point and the pile tip. The distance from the pile-abutment interface to the point where lateral

displacement is very small (the depth beyond which has a negligible effect on pile response) is

often called equivalent cantilever.

124
Figure 6.4 Schematic diagram of the experimental testing (Frosch et al., 2011)

Figure 6.5 Typical Pile model in IAB

125
Columns that were modelled herein represent the top segment having a length of 1.524 m (5 ft)

which stands for the distance from the pile abutment interface to the inflection point where the

bending moment vanishes. This length is estimated based on parametric studies assuming different

soil types and pile sections. The loading of the models includes application of a combined axial

load and lateral cyclic displacement history consistent with the experimental testing procedure.

The axial load used was (0.25fyAg); equivalent to the maximum allowable axial stress as per

AASHTO (2002) multiplied by the gross area of the cross section (Ag) where fy the specified yield

strength of steel. Material properties were based on calibrated cyclic coupon tests for Grade A36

steel, which incorporates a range of yield strengths ranging from 40-50 ksi. The yield strengths

were 40 ksi and 49 ksi for specimens HP10x42 and HP12x53, respectively. The cyclic plasticity

parameters presented in the previous section were used for both steel yield strengths, yet each

model was assigned its specified yield strength. Accordingly, the axial loads applied were 124 kips

and 190 kips, for specimens HP10x42 and HP12x53, respectively.

The lateral displacement history is imposed to the column through symmetric revered cyclic

displacement protocol as shown in Table 6.1. The loading scheme started at small displacement

with few cycles to ensure specimen stability. Later, higher displacements are applied combined

with 50 cycles, representing the potential bridge life of 50 years.

126
TABLE 6.1 DISPLACEMENT PROTOCOL USED IN THE TEST BY Frosch et al. (2004)

Cycles
Displacement Range (mm)
HP10X42 HP12X53
0.25 5 5
0.5 10 10
0.75 25 25
1.00 50 50
1.25 50 50
1.5 50 100
1.75 50 70
2.00 50 -
2.25 50 -

The column base is assumed fully fixed to simulate the concrete base which was securely attached

to the floor in the experiments while the other end is assumed free to rotate and displace in the

direction of the cyclic loading, as supposedly encountered during the testing. Figure 6.4 depicts

typical FE model of the specimens. Loading sequence of the model included two steps, the first

step involves axial load applied to the column to provide initial stability. The axial load is then

propagated to the next step to act simultaneously with the displacement protocol. Both of loading

steps were applied to a reference point which was tied to the pile head to provide a rigid body and

distribute loading evenly without potential stress concentration.

127
Figure 6.6 Typical FE model of the test setup

Figure 6.5 shows the lateral load-displacement hysteresis of column HP10x42. The figure indicates

that the FE model almost matches the experimental response for the elastic and yield phases.

However, the lateral load in the experimental results maintained its value at post-buckling

displacements and exhibits a slight shifting past the FE load values, where the lateral load in the

FE model underwent a substantial decrease at a displacement range of (1.25 inch) as can be seen

clearly in Figure 6.6. This is attributed to the deterioration and spalling of the concrete adjacent to

the interface of column and the concrete base which resulted in additional steel material embedded

in the concrete base to emerge and resist the applied loading. The local buckling commenced at a
128
displacement range of 1.25 in according to the experiments; this is the same displacement at which

the lateral load had experienced a sudden drop in the lateral load fluctuations with time in FE

computations as shown in Figure 6.5.

Lateral Displacement (mm)


-50.8 0.0 50.8
30 133.44
FE Model
Experimental (Frosch et al., 2004)
20 88.96
Lateral Load (Kips)

Lateral Load (kN)


10 44.48

0 0.00

-10 -44.48
Buckling Limit

-20 -88.96

1.25581
-30 -133.44
-3 -2 -1 0 1 2 3

Lateral Displacement (in)

Figure 6.7 HP10X42 response: Load-deflection hysteresis at column top

129
30 31.75 mm Amplitude 133.44

20 88.96

Lateral Load (kips)

Lateral Load (kN)


10 44.48

0 0.00

-10 -44.48

-20 -88.96

1.31623
-30 -133.44
1.0 1.2 1.4 1.6 1.8 2.0

Step Time

Figure 6.8 HP10X42 response: Lateral load fluctuations along step 2

The shape of the buckling of the FE model is also fairly matches the experimental behavior as

shown in Figures 6.7 and 6.8. The flanges were wrinkled the opposite way of the experimental

testing due the possible geometric imperfections which is not considered in the FE model. Due to

synchronized occurrence of the local buckling in the experimental testing and the drop of the lateral

load in the FE model, the calibration process is deemed adequate and representative for buckling

prediction of HP sections under prescribed loading scheme.

130
Figure 6.9 HP10X42 response: Buckling of column in the experimental testing

Figure 6.10 HP10X42 response: FE buckling mode

131
Figure 6.9 shows the lateral load-displacement hysteresis for specimen HP12x53. The lateral load

in the FE model experienced a drop at a displacement range of 1 in as can be seen from Figure

6.10. However, the experimental testing indicates that all flanges buckled at a displacement range

of (1.25 inch). This could also be attributed to the deterioration of concrete at the column base as

in the previous case. The figure indicates that the overall response of the FE model is at good

agreement with the experimental results and could be considered, combined with the outcomes of

the other specimen (HP10x42) calibration, an adequate representation to the experimental behavior

at stages prior to fracture. The model is not intended to simulate fracture and is deemed adequate

to model the local buckling before to crack initiation.

Lateral Displacement (mm)


-50.8 0.0 50.8
30 133.44
FE Model
Experimental (Frosch et al., 2004)
20 88.96
Lateral Load (Kips)

Lateral Load (kN)


10 44.48

0 0.00

-10 -44.48

-20 -88.96
Buckling Limit

1.00958
-30 -133.44
-3 -2 -1 0 1 2 3

Lateral Displacement (in)

Figure 6.11 HP12X53 response: Load-deflection hysteresis at column top

132
25 mm in Amplitude
30 10
25
20
8
15
Lateral Load (kips)

Lateral Load (kN)


10
5 6

0
-5 4
-10
-15
2
-20
-25 1.1409
-30 0
1.0 1.2 1.4 1.6 1.8 2.0

Step Time

Figure 6.12 HP12X53 response: Lateral load fluctuations along step 2

6.4 Pile selection, Boundary Conditions and Loading History

A total of 13 models were created by the FE software ABAQUS CAE. Nine sections oriented to

their weak bending axis and five oriented to their strong bending axis are selected to study the

effect of pile orientation as shown in Table 6.2. Five compact sections and four non-compact

sections were chosen to study the effect of flange-to-thickness effects on the displacement

capacity. These sections are chosen because of their wide popularity in IABs construction as per

Vermont design guidelines for IABs (2008). They have also high ultimate to yield strength ratio

and are believed to withstand severe cyclic loading and ensure full section plastic moment before

local buckling starts. Pile length is chosen to be 30 ft which is deemed enough to secure a fixed

133
end support during modelling. This depth is reached based on solving several LPILE models and

estimating the average depth beyond which no lateral displacement might develop.

TABLE 6.2 HP SECTIONS CONSIDERED IN THE PARAMETRIC STUDY

E =10.23* Area Axial Load


Pile b f / 2t f h/tw  p  0 .38
Fy (kips)
(in2)
HP8x36 9.16 14.2 Compact 10.6 106

HP10x42 12.0 18.9 Non-Compact 12.4 124

HP10x57 9.03 13.9 Compact 16.7 167

HP12x63 11.8 18.9 Non-Compact 18.4 184

HP12x74 10 16.1 Compact 21.8 218

HP12x84 8.97 14.2 Compact 24.6 246

HP14x89 11.9 18.5 Non-Compact 26.1 261

HP14x102 10.5 16.2 Non-Compact 30.1 301

HP14x117 9.25 14.2 Compact 34.4 344

*Based on E=29000 ksi and Fy=40 ksi

Soil effect on the pile is incorporated by attaching horizontal nonlinear springs equally spaced at

0.3 m (1 ft) intervals along the pile depth in the direction of the cyclic loading in case of weak axis

bending and in two perpendicular directions in case of strong axis bending to provide support for

lateral torsional buckling potential in the latter case. Fixed boundary conditions are imposed at the

pile end while the pile top is assumed free to displace in the direction of lateral displacement and

134
in the vertical direction. In the transverse direction, the displacement is restrained assuming

negligible lateral abutment movement. All three rotations were also restrained considering large

abutment stiffness as compared to pile section stiffness. A sustained axial compressive load is

applied at the pile head equal to (0.25f yAg), which stands for the maximum permissible axial load

on the pile per AASHTO standard specifications (2010). The Axial load values are tabulated in

Table 6.2. The yield stress of the steel is taken as 40 ksi for all sections. Figure 6.11 and Figure

6.12 show typical pile models for weak and strong axis bending respectively.

Figure 6.13 Typical pile idealization in parametric studies: weak axis

135
Figure 6.14 Typical pile idealization in parametric studies: strong axis

In this parametric study, an iterative procedure is adopted to predict the displacement at which the

local buckling of the HP section initiates. The displacement capacity of the section is identified as

the displacement at which the lateral load undergoes a substantial drop during 75 cycles, which

represents the expected lifespan of the bridge as per AASHTO LRFD (2012). It involves imposing

a separate constant value for displacement amplitude and observing the respective lateral load

fluctuation with time. Displacements amplitude starts at (0.25 inch) at (0.25 inch) increment until

buckling. For example, a displacement of 0.25 in is applied for 75 cycles while observing the

corresponding lateral load fluctuations, if a drop occurred, the (0.25 inch) will be the displacement

capacity, otherwise displacement will be increased to (0.5 inch) and the process is repeated all over

again by cycling 75 times and so on until the intended displacement is predicted. This procedure

is adopted to study each displacement effect individually and avoiding accumulation of the plastic

136
deformations resulting from successive overlapping of displacement amplitudes to one another in

case of using increasing displacement scheme, as the procedure seen in the experimental testing

by Frosch et al. (2004).

6.5 Soil-Pile Interaction

The soil effect on the pile is incorporated by attaching a set of nonlinear springs to the pile along

its entire depth. P-y curves method is employed to compute the force-deflection relationships for

the soil springs. LPILE (Ensoft Inc., 2015) software is utilized to compute pressure-displacement

relationships for points inside soil medium along pile depth and they were converted into force-

deflection relationships which are connected to the FE nodes at the soil-pile interface. A set of p-

y curves for each spring is required to simulate the soil effect on the pile under lateral loading. P-

y data is computed according the soil type then they are converted into force-displacement set by

multiplying the pressure by the respective tributary length of each spring. Each spring along the

depth of the pile is modelled as a connector element available in ABAQUS with elastic behavior

and corresponding connector section for that depth. Connector elements in Abaqus operate as

nonlinear springs with combined tension/compression behavior. However, these connector

elements are unable to model tension-only or compression only mechanism. In laterally loaded

piles, soil springs are compressed against the pile when the pile moves towards the soil and

supposedly these springs detached from the pile as it moves away from the soil to prevent soil

tension. To overcome this inevitable feature of connector elements in Abaqus, only one spring is

used to represent soil effects for both sides of the soil assuming a compression spring in case of

pile movement towards soil and tension spring in case of the soil moving away from soil to

compensate the compression spring on the other side of the pile, so it works as reversed tension

on the same spring.

137
In this study, a medium sand is used as the soil around the pile. The medium sand is selected

because piles in IAB is constructed with predrilled holes at the top portion of the pile filled with

loose sand which over cycling process, the loose sand condenses into a stiffness close to medium

sand (Dicleli and Albhaisi, 2003). The following properties of the medium sand are used in

calculating the p-y curves: angle of internal friction is 35, unit weight of the soil  is 100 and soil

modulus kpy is 90 (Lpile manual).

6.6 Buckling Failure Criterion

Buckling of the piles under combined axial compressive and lateral cyclic loading is identified by

a substantial decrease in the lateral load values, often characterized by a sudden drop of the lateral

load oscillations with time. Fogarty and Al-Tawil (2016) showed that lateral load reduction is the

dominant criterion among other failure criterion such as bending moment or axial compressive

load reduction. Reduction of the lateral load is also observed by Frosch et al. (2004) as local

buckling commences in the experimental testing. Therefore, the reduction in the lateral load

capacity during cycling is considered as the governing failure criterion in identifying the local

buckling initiation. Many researchers showed that 10-15% reduction in the lateral load can be

considered as the failure point of the section under cyclic load. In the current study, most of the

section underwent a local flange buckling at or less 10% reduction ratio.

138
6.7 Results

Table 6.3 reports buckling capacities, average axial strain at buckling, number of cycles for

buckling, and displacement capacities for the HP sections considered in the current study. The

displacement capacities are the buckling capacity less 0.25 inch, which in this case corresponds to

the previous cycling displacement at which the lateral load sustained its value along 75 cycles. The

axial strain at buckling represents the maximum axial strain induced in the flange tip at the pile-

abutment interface (pile head).

TABLE 6.3 DISPLCEMENT CAPACITIES OF A SET OF HP SECTIONS EMBEDDED IN A


MEDIUM SAND

No. of
Pile Axis b plastic
@ b cycles to b/plastic /y c (in)
Designation orientation (in) (in)
buckling
HP8x36 weak 2.25 0.01628 40 0.6 3.75 11.8 2
HP10x42 weak 1.5 0.01073 31 0.6 2.5 7.8 1.25
HP10x57 weak 2.5 0.01671 49 0.69 3.62 12.1 2.25
HP12x63 weak 1.75 0.01295 33 0.67 2.6 9.4 1.5
HP12x74 weak 2.25 0.01485 50 0.72 3.13 10.8 2
HP12x84 weak 2.5 0.0115 52 0.76 3.29 8.3 2.25
HP14x89 weak 1.75 0.01193 46 0.73 2.39 8.7 1.5
HP14x102 weak 2.25 0.01423 49 0.78 2.88 10.3 2
HP14x117 weak 3 0.01814 43 0.83 3.6 13.2 2.75
HP8x36 strong 1 0.0012 31 0.68 1.47 0.86 0.75
HP10x42 strong 0.75 0.00103 22 0.68 1.1 0.75 0.5
HP10x57 strong 1 0.00092 57 0.78 1.28 0.67 0.75
HP12x74 strong 1 0.00085 50 0.82 1.22 0.62 0.75
HP14x89 strong 1 0.00103 28 0.85 1.18 0.75 0.75

139
Figure 6.13 and Figure 6.14 show the load-displacement hysteresis behavior for HP 12x63 section

oriented to the weak axis and the corresponding lateral load fluctuations with step time. Graphs

for other HP sections can be found in Appendix B. A graphical procedure is employed to extract

the strain at buckling utilizing Origin Lab software to accurately predict the maximum and

minimum values for strain amplitude and an average absolute value is considered.

Figure 6.15 shows illustrates the graphical procedure to determine the strain at buckling

displacement. Flange tip at the pile top end or point of flange wrinkling are the potential locations

for maximum strain induced within the section, those locations were used to collect buckling

strains. Figure 6.16 depicts buckling ductility ratios (strain at buckling to yield strain) for all

sections in this study. It is obvious that buckling ductility in weak axis bending is far exceeding

the one in strong axis. Moreover, for the same orientation (weak axis) there are significant

variations exists among ductility ratios, and they are independent of slenderness ratios or flexural

rigidities. Number of cycles to buckling is predicted in the same way by constructing a vertical

line at the point of lateral load drop on the abscissa (time step) and the number of cycles is

determined by proportioning its value from the total number of cycles (Refer to Figure 6.14).

Failure mode of the sections was local flange only or flange and web buckling at the pile head.

Displacements at buckling ranged from 1.5 inches to 3 inches for weak axis bending; the compact

sections achieved higher capacities than noncompact sections. The strong axis bending capacities

were in the range of 1 inch without significant effect of the compactness criteria on the buckling

capacity. Experimental testing conducted by Frosch (2004, 2011) has demonstrated same

outcomes regarding weak axis-oriented sections which surpassed those oriented to their strong axis

bending. For instance, section H8x36 in weak axis bending achieved 2.25-inch buckling

displacement whereas the same section buckled at 1.5-inch displacement in case of strong axis

140
bending. Moreover, the soil restraining effect that has been considered in the computational models

has an adverse effect on the displacement capacity of the sections if compared to barely tested

columns without soil.

Lateral Displacement (mm)


-50.8 0.0 50.8
50 222.40

40 177.92

30 133.44
Lateral Load (Kips)

20 88.96

Lateral Load (kN)


10 44.48

0 0.00

-10 -44.48

-20 -88.96

-30 -133.44

-40 HP12x63 Weak Axis -177.92


Displacement Amplitude =1.75 inch
-50 -222.40
-2 -1 0 1 2
Lateral Displacement (in)

Figure 6.15 Load-deflection relationship for HP12x63 weak axis

141
1.44
250
33 Cycles
200
183
150 166.5
Lateral Load (kN)

100

50

-50

-100

-150

-200 HP 12x63 Weak Axis


44 mm (1.75 in) Amplitude
-250
1.0 1.2 1.4 1.6 1.8 2.0
Step Time

Figure 6.16 Lateral load fluctuation with time for HP12x63 weak axis

142
0.02

0.01114

0.00
Lograithmic Strain, y

-0.01475
-0.02

-0.04

-0.06

HP 12x63 Weak Axis


1.75 inch Amplitude
-0.08
1.0 1.2 1.4 1.6 1.8 2.0
Step Time

Figure 6.17 variation of strain with step time for HP 12x63 at buckling displacement

Experimental displacements predicted by Frosch et al. (2004) represent the first segment in the

pile equivalent cantilever model as presented in a previous section. The second segment (bounded

by the infection point and the point of fixity in the equivalent cantilever) has a lateral displacement

almost equivalent to the first segment at local buckling level as depicted in Figure 6.17. These two

displacements were added up to obtain the total displacement at the pile head as shown with the

FE results in Table 6.4. Comparing the FE results with experimental total displacement refer that

the experimental displacement exceeds the FE value for the same section and orientation. The

reduction in the FE results is due to inclusion of soil effect which curbed the pile flexibility and

triggered the local buckling sooner. The strain computed by Frosch et al. (2004) is in the order of

0.03 at buckling and the average corresponding ductility was 20, which are also higher than the

computed counterparts in FE results by the order of at least 2. This clearly indicates that the soil

143
effect reduces the displacement capacity of the pile section to the half or more and a predrilled

hole filled with sand or flexible material within the equivalent cantilever length of the pile is

strongly recommended.

14

12

10

8
/y

0
HP8x36W

HP10x42W

HP10x57W

HP12x63W

HP12x74W

HP12x84W

HP14x89W

HP14x102W

HP14x117W

HP8x36S

HP10x42S

HP10x57S

HP12x74S

HP14x89S

Figure 6.18 Buckling ductility ratios for HP sections in medium sand

144
Figure 6.19 Typical cantilever pile model in IABs (Frosch et al., 2004)

TABLE 6.4 FE vs Frosch et al. (2004, 2011) BUCKLING CAPACITES

Buckling Experimental
Section Orientation
Displacement (Frosch, 2004, 2011)

HP8x36 weak 2.25 4


HP10x42 weak 1.5 2.5
HP14x89 weak 1.75 3.5
HP8x36 strong 1 3

145
Displacement at full plastic moment is predicted following a trial-and-error procedure utilizing

program Lpile. The lateral displacement is varied in the program until the corresponding moment

reached the plastic moment capacity (Mp). It is obvious that the buckling displacement is 2.5-3.5

times the plastic moment displacement in case of weak axis bending and on the order of 1.5 the

plastic moment displacement in strong axis bending as can be seen in Table 6.3.

Buckling strain in the weak axis is between 7 and 13 times yield strain, an average value of 10

times yield strain is considered for buckling. The strong axis ductility is less than 1, this indicates

that piles buckles before arriving the yield strain which is attributed to out of plane bending or

lateral torsional buckling.

6.8 Effect of Soil Types Surrounding Piles

Six soil types with different stiffness properties were selected to investigate the effect of soil type

on the displacement capacity of the piles. The sandy soil types are: loose sand, medium sand, and

dense sand. The clayey soil types are: soft clay, medium clay and stiff clay. The sand and clay

properties are respectively presented in Table 6.5 and 6.6. P-y curves are constructed for all soil

types using LPILE software and nonlinear springs are generated depending on the guidelines

presented earlier chapter. Two pile section were selected for the investigation process; HP10X57

and HP12X84, oriented to their weak axis. These piles are compact sections for both A36 and

A575 steel grades (Azizinamini et al., 2015)

146
TABLE 6.5 SAND PROPERTIES USED IN THE STUDY

Soil Type Angle of Friction,  Unit Weight,  lb/ft3 Kpy, lb./in3

Loose Sand 30 90 25

Medium Sand 35 100 90

Dense Sand 40 115 225

TABLE 6.6 CLAY PROPERTIES USED IN THE STUDY

Undrained Shear Unit Weight,


Soil Type 50 Kpy, lb./in3
Strength*, Cu, (lb./ft2)  lb./ft3

Soft clay 500 73 0.02 30

Medium clay 750 85 0.01 100

Stiff clay 1500 108 0.005 500

Table 6.7 show the displacement capacities for the cases presented above. It is clear from the

values that the displacement capacity of the section is reversely proportional to the soil stiffness

for both of sand and clay. This is a further confirmation that the soil stiffness has an adverse effect

of the flexibility of the pile and displacement capacity of the pile. Also, the clayey soil outperforms

the sandy soil in allowing the HP piles additional displacement capacities.

147
TABLE 6.7 BUCKLING DISPLACEMENT COMPARISONS FOR VARIOUS SOIL TYPES

Buckling Displacement (inch)


Soil Type HP10X57 HP12X84
Loose sand 2.75 3
Medium Sand 2.5 2.5
Dense Sand 1.75 2
Soft Clay 4.5 5
Medium Clay 3.25 4
Stiff Clay 2 3

6.9 Comparison with Abendroth and Greimann (1989) Method

To get more insight about the predicted FE results, a comparison with analytical procedure

suggested by Abendroth and Greimann (1989) is presented herein. This method assumed an

inelastic model to quantify the displacement capacity of steel piles in IABs. The method involves

moment-rotation capacity of the steel section according the AISC standard specifications (1984)

which is three times the plastic rotational capacity. Based on structural mechanic principles, the

researchers arrive at the following equation to estimate the displacement capacity of HP piles in

IABs:

   p 0 .6  2.25 C i  6.3

Where:

M pL2
p  6.4
D EI

148
19 b f Fy
Ci   6.5
6 60t f

E: modulus of elasticity of steel. Taken as 29000 ksi in the current study.

I: moment of inertia of the cross section.

D: slope-deflection factor = 6 for fixed-end cantilever.

Mp: Plastic moment capacity of the section =F y Z.

L: The depth of the equivalent cantilever

bf: Flange width.

tf: Flange thickness.

Equivalent cantilever length is predicted using LPILE software for the set of HP section under

consideration as can be seen from the displacement profile of the HP section in Figure 6.18, which

has been established using LPILE. It is obvious that piles having larger sections exhibits deeper

point of fixity or equivalent cantilever.

149
Lateral Displacement (mm)
-12.7 0.0 12.7 25.4 38.1 50.8 63.5 76.2
0 0.00

-5 -1.52

-10 -3.05

Depth (m)
Depth (ft)

-15 HP8x36 -4.57


HP10x42
HP10x57
-20 HP12x63 -6.10
HP12x74
HP12x84
-25 HP14x89 -7.62
HP14x102
HP14x117
-30 -9.14
-0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0
Lateral Displacement (in)

Figure 6.20 Lateral displacement profiles of HP sections at buckling

Table 6.8 shows the displacement capacities of HP sections, computed by Abendroth et al. (1989)

and FE methods. It is evident that FE exhibits slightly smaller displacement capacities as

compared to the analytical method for weak axis bending. The difference is attributed to the

approximation associated with inelastic rotation assumption in the analytical method. For strong

axis bending, the analytical method maintained the same displacement capacities as the weak axis,

whereas the FE results had significantly decreased. This is attributed to the failure of including

lateral torsional buckling effect in the analytical model in case of strong axis bending.

150
TABLE 6.8 COMPARISON BETWEEN ABENDROTH ET AL. (1989) AND FEM

Section L  c b
Orientation Ci Z (in3) I (in4) Mp p
Designation (in) Abendroth, 1989) FEM
HP8x36 weak 108 1 15.2 40.3 608 1.011 2.88 2.25
HP10x42 weak 97.2 0.63 21.8 71.7 872 0.66 1.33 1.5
HP10x57 weak 126 1 30.3 101 1212 1.095 3.12 2.5
HP12x63 weak 136.8 0.69 38.7 153 1548 1.09 2.34 1.75
HP12x74 weak 140.4 1 46.6 186 1864 1.14 3.24 2.25
HP12x84 weak 144 1 53.2 213 2128 1.19 3.39 2.5
HP14x89 weak 158.4 0.65 67.7 326 2708 1.198 2.46 1.75
HP14x102 weak 162 0.95 78.8 380 3152 1.25 3.44 2.25
HP14x117 weak 165.6 1 91.4 443 3656 1.3 3.71 3
HP8x36 strong 129.6 1 33.6 119 1344 1.09 3.11 0.75
HP10x42 strong 118.8 0.63 48.3 210 1932 0.75 1.51 0.75
HP10x57 strong 151.2 1 66.5 294 2660 1.189 3.39 0.75
HP12x74 strong 172.8 1 105 569 4200 1.27 3.6 0.75
HP14x89 strong 187.2 0.65 146 904 5840 1.3 2.68 0.75

151
6.10 Integral Abutment Bridge Length

The maximum length of an integral abutment bridge can be computed according on the

displacement capacity of HP sections. A simple thermal expression equation is utilized herein to

quantify the maximum length. The equation involves lateral displacement of the abutment at the

deck level. The abutments in IABs are often designed with a relatively small height (stub) to

alleviate their high induced bending stresses, this implies the abutment rather displace horizontally

with little or no tilting. Abutment displacement at the deck level is conservatively assumed

equivalent to the pile head displacement (displacement capacity).

The maximum length of IAB can be estimated from the following equation derived from the basic

thermal expansion of a bar:

2c
Lmax  6.6
  T

Where:  is the AASHTO temperature factor,  is the coefficient of thermal expansion which is

assumed as 6 × 10–6/°F (10 × 10–6/°C) and 11.7 × 10–6/°C (6.5 × 10–6/°F) for concrete and for steel,

respectively. AASHTO LRFD (2012) bridge design specifications were considered for

temperature ranges. Steel bridges temperature taken as 60°F and 75°F for moderate and cold

climates, respectively, while concrete bridge temperature is taken as 35°F and 40°F for moderate

and cold climates, respectively,

Table 6.9 presents maximum length limits for steel and concrete bridges, for both of cold and

moderate climates in the United Sates. It is obvious that maximum length ranges from 108 – 298

m for steel bridges and 220-554 m for concrete bridges depending the weather and pile section.

These results exceed the maximum IAB limitations of most department of transportation in

152
United States. Also, it slightly exceeds the maximum length of integral abutment bridge ever

built, which located at Verona, Italy with a length of 400 m. However, most of design bureaus

and construction companies avoid higher length due to lack of rational approaches to predict the

length limits. More lengths can be achieved by allowing additional flexibility to the pile head by

providing hinge details to the pile-abutment connectivity or reduce soil stiffness around the pile.

TABLE 6.9 MAXIMUM LENGTH OF INTEGRAL ABUTMENT BRIDGES


Steel bridge Concrete bridge
Moderate Cold Moderate Cold
Section Displacement Capacity
Climate Climate Climate Climate
HP8x36 2 530 424 985 862
HP10x42 1.25 319 255 593 519
HP10x57 2.25 538 430 999 874
HP12x63 1.5 353 282 656 574
HP12x74 2 452 361 839 734
HP12x84 2.25 502 401 932 816
HP14x89 1.5 326 261 607 531
HP14x102 2 426 340 791 692
HP14x117 2.75 567 454 1054 922

153
6.11 Summary and Conclusions

Displacement capacities of a series of HP sections are predicted using the finite element package

ABAQUS. Two pile orientations are studied to investigate whichever can provide better

performance in employment in the IAB construction. Displacement capacities for the pile

embedded in medium sand is reported to be used by designers as a reference. Six soil types are

also studied investigated to examine effect of the soil stiffness on the displacement capacities. The

following conclusions can be drawn:

1. Weak axis pile orientation of HP sections provides displacement capacities 2-3 times

higher than strong axis orientation. The small stiffness associated with weak axis

orientation provides more flexibility to the section during cyclic loading.

2. Buckling displacement is on an average of 3 and 1.5 times the plastic moment capacity of

HP section for weak and strong axis bending, respectively.

3. Buckling of HP sections under combined axial and lateral cyclic displacement occurs at a

strain in the order of 10 times yield strain in average for weak axis orientation. For strong

axis orientation, the section buckles before or at yield strain due to out-of-plane bending.

4. The stiffness of the soil is reversely proportional to the displacement capacity of the pile.

For example, loose sand provides higher displacement capacities than dense sand. The idea

of providing predrilled holes at the top part of the pile is greatly encouraged in IAB

construction. In general, clay outperforms sand in granting pile better flexibility.

5. Depending on the size of the HP section used as a pile in construction, maximum length of
steel integral abutment bridges is ranging between 250 m and 500 m for cold and moderate

climates, respectively. In concrete integral abutment bridges, the maximum length could

be doubled, in a range of 500 m to 1000 m, for moderate and cold climates, respectively.

154
7. Conclusions and Recommendations

7.1 Conclusions

The following conclusions are drawn based on the study in this dissertation:

1. The CEB-90-90 shrinkage model is the most representative model for the concrete behavior

over the monitoring time.

2. The deck represents a substantial contribution to the shrinkage effects on the concrete in IAB;

however, prestressed girders have little or no shrinkage influence on the bridge response.

3. The Rankine earth pressure theory outperforms other theories in capturing the abutment-

backfill interaction behavior.

4. The distribution of the temperature along the depth of the superstructure is important in

simulating the displacement amplitude of the abutment, and the best simple distribution over

the entire depth was found by giving full temperature change for the deck and half of the

temperature change to the girders. Other distributions either over- or underestimate the

displacement amplitude.

5. Abutment displacement ratcheting (deck shortening) over time occurs due to the restraining

effect of the soil towards the abutment during the expansion phase of the bridge. The

displacement induced in passive cases (bridge expansion) is less than the displacement in

active cases (bridge contraction); this disparity in displacements leads to an accumulative

difference in the net response which is responsible for the overall inward shortening of the

bridge.

6. The change in the abutment displacement follows a linear trend, which validates the

employment of a linear thermal expansion equation in estimating lateral thermal bridge

displacement.

155
7. The difference between the equation and the FE model displacements is very small with short

bridge lengths but grows significantly larger as the bridge length increases.

8. Neither the pile size or pile orientation nor soil type or soil stiffness have a significant impact

on the abutment displacement.

9. The soil stiffness has more impact than the pile orientation on the pile displacement.

10. The pile stresses are proportional to the soil stiffness and inversely proportional to the pile

stiffness.

11. Weak axis orientation lead to higher stresses than strong axis orientation along the pile depth,

which is attributed to the small stiffness of the weak axis as compared to strong axis. The

maximum stress occurs at the second segment of the pile (bounded by the first inflection point

and the second infection point). The maximum stress is not induced in the pile-abutment

interface because of the rotation of the abutment, which provide some release of fixity and

reduced stresses at the pile head.

12. The soil stiffness has a small influence on the deck stresses. However, the reduction of soil

stiffness around the pile can contribute to mitigating part of the deck stresses.

13. Reducing the stiffness of the soil around the pile can contribute in mitigating the stresses on

bridge girders.

14. Weak axis pile orientation of HP sections provides displacement capacities 2-3 times higher

than strong axis orientation. The small stiffness associated with weak axis orientation provides

more flexibility to the section during cyclic loading.

15. Buckling displacement is an average of 3 and 1.5 times the plastic moment capacity of HP

section for weak and strong axis bending, respectively.

156
16. Buckling of HP sections under combined axial and lateral cyclic displacement occurs at a

strain on the order of 10 times the average yield strain for weak axis orientation. For strong

axis orientation, the section buckles before or at yield strain due to out-of-plane bending.

17. The stiffness of the soil is inversely proportional to the displacement capacity of the pile. For

example, loose sand provides higher displacement capacities than dense sand. The idea of

providing predrilled holes at the top part of the pile is greatly encouraged in IAB construction.

In general, clay outperforms sand in granting a pile better flexibility.

18. Depending on the size of the HP section used as a pile in construction, the maximum length

of steel IABs is 250 m and 500 m for cold and moderate climates, respectively. In concrete

IABs, the maximum length can be doubled to 500 m to 1000 m for moderate and cold climates,

respectively.

7.2 Recommendations for Future Work

1. Expanding the calibration process to incorporate more bridge types and pile sections will

permit more insight about the behavior of integral abutment bridges. Also, more responses can

be studied, including bridge transverse displacement, abutment rotation, pile curvature, and

girder stresses and strains.

2. Parametric studies can be expanded to include the effect of backfill soil stiffness on bridge

various responses. Bridge skew effects on the bridge response can also be highlighted by

considering a three-dimensional finite element model.

3. A pile buckling analysis may be conducted by assuming solid modelling instead of p-y curves

for the soil surrounding piles. In this method, the effect of local buckling on displacement

157
capacities will be obvious. This method is very expensive in terms of computational time for

the solution, but it is very promising in terms of the outcomes.

158
CITED REFERNCES

1. American Association of State Highway and Transportation Officials (AASHTO):

AASHTO LRFD Bridge Design Specification, 5th Edition, Washington, D.C., 2010.

2. AASHTO Standard Specifications for Highway Bridges, 17th Edition. Washington, DC,

USA, 2002.

3. Abendroth, R.E., Greimann, L.F. and Ebner, P.B.: Abutment Pile Design for Jointless

Bridges. Journal of Structure Engineering, Vol. 115, No. 11, 2914-2929, 1989.

4. Abendroth, R.E. and Greimann, L.F.: Field Testing of Integral Abutment Bridge. Final

Report, Iowa Department of Transportation Project HR-399. 2005.

5. ACI 209: Guide for Modeling and Calculating Shrinkage and Creep in Hardened Concrete.

American Concrete Institute (ACI) Committee 209, Farmington Hills, Michigan. 2008.

6. Alampalli, S. and Yannotti, A.: In-Service Performance of Integral Bridges and Jointless

Deck. Transportation Research Record: Journal of the Transportation Research Board. Vol.

1624, pp. 1-7. 1998.

7. American Petroleum Institute (API): Recommended practice for planning, designing, and

constructing fixed offshore platforms-working stress design. 20th ed., API RP2A-WSD,

Washington, D.C. 1993

8. Azizinamini, A, Yakel, A., Sherafati, A., Taghinezhad, R., and Gull, J.H.: Flexible Pile

Head in Jointless Bridges: Design Provisions for H-Piles in Cohesive Soils” Journal of

Bridge Engineering, Vol. 21, Issue 3. 2016.

9. Bertero, V., and Popov, E.: Effect of Large Alternating Strains of Steel Beams” J. Struct.

Eng., 91(1), 1-12. 1965.

159
10. Bhowmik, S.K.: Three-dimensional non-linear finite element analysis of laterally loaded

piles in clay. PhD dissertation, University of Illinois at Urbana-Champaign, Urbana, IL.

1992.

11. Brown, D.A. and Shie, C.: Three-dimensional finite element model of laterally loaded

piles. Computers and Geotechnics, Vol 10, Issue 1, 1990, Pages 59-79, 1990.

12. Burke Jr., M.P.: The design of integral concrete bridges. Concrete International, Vol. 15,

June, pp. 37-42, 1993.

13. Civjan, S.A., Kalayci, E., Quinn, B.H., Brena, S.F., and Allen, C.A.: Observed integral

abutment bridge substructure response. Engineering Structures, 56 (2013) 1177–1191,

2013.

14. Das, B.M.: Principles of Geotechnical Engineering. Cengage Learning Engineering; 5 th

Ed., 2011.

15. Dassault Systems: Abaqus CAE user’s manual. 6.14, 2014.

16. Desai, C., and Zaman, M.: Advanced Geotechnical Engineering: Soil-Structure Interaction

using Computer and Material Models. CRC press, 2013.

17. Dicleli, M., and Albhaisi, S.M.: Maximum length of integral bridges supported on steel H-

piles driven in sand. Engineering Structures, 25(2003), 1491–1504, 2003.

18. Dicleli, M., and Albhaisi, S.M.: Effect of cyclic thermal loading on the performance of

steel H-piles in integral bridges with stub-abutments. J. Constr. Steel Res., 60(2), 161–182,

2004.

19. Duncan, J.M. and Mokwa, R.L.: Passive Earth Pressure: Theories and Tests. Journal of

Geotechnical and Geoenvironmental Engineering, Vol. 127, No.3, 2001.

160
20. Elkady, A., and Lignos, D.: Analytical investigation of the cyclic behavior and plastic

hinge formation in deep wide-flange steel beam columns. Bull. Earthquake Eng., 13, 1097–

1118, 2015.

21. Ensoft, Inc.: LPILE v2015 for Windows-A Program for the Analysis of Piles and Drilled

Shafts Under Lateral Loads. Available from: <http://www.ensoft.com>.

22. Fang, H.: Foundation Engineering Handbook. 2nd Edition, Edited by H. Y., Fang, Van

Nostrand Reinhold, New York, USA, 1991.

23. Faraji, S., Ting, J.M., Crovo, D.S., and Ernst, H.: Nonlinear Analysis of Integral Bridges:

Finite Element Model. Journal of Geotechnical and Geoenvironmental Engineering, Vol.

127, No.5, 454-461, 2001.

24. Fennema, J.F., Laman, J.A., and Linzell, D.J.: Predicted and Measured Response of an

Integral Abutment Bridge. Journal of Bridge Engineering, Vol. 10 Issue 6, 2005.

25. Fogarty, J., and El-Tawil, S.: Collapse Resistance of Steel Columns under Combined Axial

and Lateral Loading. Journal of Structure Engineering, Vol. 142, No. 1., 2016.

26. Frosch, R.J., Chovichien, V., Durbin, K.O., and Fedroff, D.: Jointless and Smoother

Bridges: Behavior and Design of Piles. Joint Transportation Research Program,

FHWA/IN/JTRP-2004/24, 488 pp., 2004.

27. Frosch, R.J., Kreger, M.E., and Talbott, A.M.: Earthquake Resistance of Integral Abutment

Bridges. Publication FHWA/IN/JTRP-2008/11. Joint Transportation Research Program,

Indiana Department of Transportation and Purdue University, West Lafayette, Indiana,

2009.

161
28. Frosch, R.J., and Lovell, M.D.: Long-Term Behavior of Integral Abutment Bridges.

Publication FHWA/ IN/JTRP-2011/16. Joint Transportation Research Program, Indiana

Department of Transportation and Purdue University, West Lafayette, Indiana, 2011.

29. Girton, D.D., Hawkinson, T.R., and Greimann, L.F.: Validation of Design

Recommendations for Integral Abutment Bridges. Journal of Structural Engineering,

ASCE, Vol. 117, No. 7., 1991.

30. Greimann, L.F., Yang, P.S., and Wolde-Tinsae, A.M.: Design of Piles for integral

Abutment Bridges. Final Report, Department of Civil Engineering, Engineering Research

Institute, Iowa State University, Ames, Iowa, August 1984, 260 pp.

31. Greimann, L.F., Yang, P.S., and Wolde-Tinsae, A.M.: Nonlinear Analysis of Integral

Abutment Bridges. Journal of Structural Engineering, ASCE, Vol. 112, No. 10, 1986, pp.

2263-2280, 1986.

32. Huang, J., Shield, C.K., and French, C.E.W.: Parametric Study of Concrete Integral

Abutment Bridges. Journal of Bridge Eng., 13(5), 511-526, 2008.

33. Kaufman, E., Metrovich, B., and Pense, A.: Characterization of cyclic inelastic strain

behavior on properties of A572 Gr. 50 and A913 rolled sections. ATLSS Rep. No. 01-13,

National Center for Engineering Research on Advanced Technology for Large Structural

Systems, Lehigh Univ., Bethlehem, PA., 2001.

34. Khasawneh, Y.A.: Soil-structure interaction of integral abutment bridges. Ph.D. thesis,

Purdue University, West Lafayette, IN., 2014.

35. Khodair, Y.A., and Hassiotis, S.: Analysis of Soil-Pile Interaction in Integral Abutment.

Journal of Computers and Geotechnics, 32, 201-209, 2005.

162
36. Kim, W., and Laman, J.: Seven-Year Field Monitoring of Four Integral Abutment

Bridges. Journal of Performance of Constructed Facilities, Vol. 26, No. 1, 2012.

37. Koh, S., and Stephens, R.: Mean Stress Effects on Low Cycle Fatigue for a High Strength

Steel. Fatigue Fract. Engng. Mater. Struct., 14(4):413–428, 1991.

38. Kunin, J., and Alampalli, S.: Integral Abutment Bridges: Current Practice in United States

and Canada. Journal of Performance of Constructed Facilities, ASCE, Volume 14, Issue 3,

2000.

39. Lawver, A., French, C., and Shield, C.: Field Performance of Integral Abutment Bridge.

Transportation Research Record: Journal of the Transportation Research Board, Vol.

1740, pp. 108-117, 2000.

40. Lemaitre, J., and Chaboche, J.L.: Mechanics of solid materials. Cambridge University

press, 1990.

41. Matlock, H.: Correlations for Design of Laterally Loaded Piles in Soft Clay. Offshore

Technology Conference, OTC 1204, 1970.

42. Meyerhof, G.G.: Penetration Tests and Bearing Capacity of Cohesionless Soils. Journal of

the Soil Mechanics Division, ASCE, Vol. 82, SM1, pp. 1-12, 1956.

43. Muqtadir, A., and Desai, C.S.: Three‐dimensional analysis of a pile‐group foundation.

International Journal for Numerical and Analytical methods in Geomechanics, Vol 10 Issue

1, 1986.

44. National Cooperative Highway Research Program (NCHRP): Manuals for the design of

bridge foundations. Barker, R.M., Duncan, J.M., Rojiani, K.B., Ooi, P.S.K., Tan, C.K. and

Kim, S.G., Rep. 343, Transportation Research Board, Washington, D.C., 1991.

163
45. Prakash, S., and Sharma, H.D.: Pile Foundations in Engineering Practice. John Wiley &

Sons, 1990.

46. Pugasap, K., Kim, W., and Laman, J.A.: Long-Term Response Prediction of Integral

Abutment Bridges. Journal of Bridge Eng., 14(2), 129-139, 2009.

47. Quinn, B.H., and Civjan S.A.: Parametric Study on Effects of Pile Orientation in Integral

Abutment Bridges. J. Bridge Eng., 22(4), 2017.

48. Rankine, W.J.M.: On the Stability of Loose Earth. Philosophical Trans., Royal Soc.,

London, 1857.

49. Reese, L.C., Cox, W.R., and Koop, F.D.: Analysis of Lateral Loaded Piles in Sand.

Offshore Technical Conference, Dallas, Texas, Paper No. OTC 2080, 1974.

50. Reese, L.C., and Welch, R.C.: Lateral Loading of Deep Foundations in Stiff Clay. Journal

of the Geotechnical Engineering Division, ASCE, Vol. 101, No. GT7, pp. 633-649, 1975.

51. Reese, L.C., and Van Impe, W. F.: Single Piles and Pile Groups Under Lateral Loading.

A. A. Balkema, Rotterdam, 2001.

52. Rollins, K.M., and Cole, R.T.: Cyclic Lateral Load Behavior of Pile Cap and Backfill.

Journal of Geotechnical and Geoenvironmental Engineering, Vol. 132, No. 9, 2006.

53. Skempton, A.W.: The Bearing Capacity of Clays. Proc. Building Research Congress, pp.

180-189, 1951.

54. Terzaghi, K.: Evaluation of Coefficients of Subgrade Reaction. Geotechnique, Vol 5, 297-

326, 1955.

55. Trochanis, A., Bielak, J., and Christiano, P.: Three-Dimensional Nonlinear Study of Piles.

Journal of Geotechnical Engineering, ASCE, 117(3), 429-447, 1991.

164
56. Vermont Agency of Transportation (VTrans): Integral Abutment Bridge Design

Guidelines. 2nd Ed., Montpelier, Vermont, 2008.

57. Welch, R.C., and Reese, L.C.: Laterally loaded behavior of drilled shafts. Research Report

3-5-65-89. Center for Highway Research. University of Texas, Austin, 1972.

58. Wigle, V.R., and Fahnestock, L.A.: Buckling-restrained braced frame connection

performance. Journal of Constructional Steel Research, Vol. 66 (1), 65-74, 2010.

165
APPENDIX A

Figure A.1 Pile Finite Element Model

Figure A.2 Lateral Displacements in the Pile Model

166
Figure A.3 Bridge Finite Element Model

167
Python code to generate abutment springs (Node numbers are changeable according the
problem)

1 from part import *


2 from material import *
3 from section import *
4 from assembly import *
5 from step import *
6 from interaction import *
7 from load import *
8 from mesh import *
9 from optimization import *
10 from job import *
11 from sketch import *
12 from visualization import *
13 from connectorBehavior import *
14 L=0
15 v=[693,699,709,717,1067,1115,1752,1753,1754,1755,1756,1757,1758,
16 1759,1760,1761,1762,1763,1764,1765,1766,1767,1768,1769,1770,
17 1771,1772,1773,1775,1777,1778,1779,1780,1781,1782,1783,1784,
18 1785,1786,1787,1788,1789,1790,1791,1792,1793]
19 for i in range(1751,1797): # add 46 to the last loop value
20 mdb.models['Model-1'].rootAssembly.WirePolyLine(mergeType=IMPRINT,
meshable=
21 False, points=((
22 mdb.models['Model-1'].rootAssembly.instances['Abutment-1'].vertices[v[L]],
23 None), ))
24 mdb.models['Model-1'].rootAssembly.features.changeKey(fromName='Wire-
'+str(i),
25 toName='Wire-'+str(i))
26 mdb.models['Model-1'].rootAssembly.Set(edges=
27 mdb.models['Model-1'].rootAssembly.edges.getSequenceFromMask(('[#1 ]', ),
)
28 , name='Wire-'+str(i)+'-Set-1')
29 mdb.models['Model-1'].rootAssembly.SectionAssignment(region=
30 mdb.models['Model-1'].rootAssembly.sets['Wire-'+str(i)+'-Set-1'],
sectionName=
31 'Backfill_14')
32 L=L+1

168
Python code to generate pile springs (Node numbers are changeable according the problem)

1 from part import *


2 from material import *
3 from section import *
4 from assembly import *
5 from step import *
6 from interaction import *
7 from load import *
8 from mesh import *
9 from optimization import *
10 from job import *
11 from sketch import *
12 from visualization import *
13 from connectorBehavior import *
14 # springs
15 k=0
16 L=0
17 j=1850 #add 22 in the next pile
18
v=[8,2,17,23,37,41,53,61,65,77,81,93,101,105,117,121,133,141,145,157,161,173,
181,185,197,
201,210,218]
19 for i in range(1853,1881): # add 22 to the last loop value
20 if k > 22: # springs after depth 22 will be the same 22 ft
21 k=22
22 mdb.models['Model-1'].rootAssembly.WirePolyLine(mergeType=IMPRINT,
meshable=
23 False, points=((None,
24 mdb.models['Model-1'].rootAssembly.instances['Pile Sell
R-1-lin-10-1'].vertices[v[L]]),
25 ))
26 mdb.models['Model-1'].rootAssembly.features.changeKey(fromName='Wire-
'+str(i),
27 toName='Wire-'+str(i))
28 mdb.models['Model-1'].rootAssembly.Set(edges=
29 mdb.models['Model-1'].rootAssembly.edges.getSequenceFromMask(('[#1 ]', ),
)
30 , name='Wire-'+str(i)+'-Set-1')
31 mdb.models['Model-1'].rootAssembly.SectionAssignment(region=
32 mdb.models['Model-1'].rootAssembly.sets['Wire-'+str(i)+'-Set-1'],
sectionName=
33 'Pile_spring_'+str(k))
34 mdb.models['Model-1'].rootAssembly.sectionAssignments[j].getSet()
35 mdb.models['Model-1'].rootAssembly.ConnectorOrientation(localCsys1=
36 mdb.models['Model-1'].rootAssembly.datums[19734], region=
37 mdb.models['Model-1'].rootAssembly.allSets['Wire-'+str(i)+'-Set-1'])
38 k=k+1
39 L=L+1
40 j=j+1

169
APPENDIX B

Lateral Displacement (mm)


-50.8 0.0 50.8
30 133.44

20 88.96
Lateral Load (Kips)

Lateral Load (kN)


10 44.48

0 0.00

-10 -44.48

-20 -88.96
HP8x36 Weak Axis
Displacement Amplitude = 57 mm (2.25 in)
-30 -133.44
-2 -1 0 1 2
Lateral Displacement (in)

Figure B.1a. Load-deflection relationship for HP8x36 weak axis


1.54
150
40 cycles

100 101
89.9
Lateral Load (kN)

50

-50

-100
HP 8x36 Weak Axis
57 mm (2.25 in) Amplitude)
-150
1.0 1.2 1.4 1.6 1.8 2.0
Step Time

Figure B.1b. Lateral load fluctuation with time for HP8x36 weak axis

170
Lateral Displacement (mm)
-50.8 0.0 50.8
50 222.40

40 177.92

30 133.44

Lateral Load (Kips)


20 88.96

Lateral Load (kN)


10 44.48

0 0.00

-10 -44.48

-20 -88.96

-30 -133.44

-40 HP12x63 Weak Axis -177.92


Displacement Amplitude =1.75 inch
-50 -222.40
-2 -1 0 1 2
Lateral Displacement (in)

Figure B.2a. Load-deflection relationship for HP12x63 weak axis

1.44
250
33 Cycles
200
183
150 166.5
Lateral Load (kN)

100

50

-50

-100

-150

-200 HP 12x63 Weak Axis


44 mm (1.75 in) Amplitude
-250
1.0 1.2 1.4 1.6 1.8 2.0
Step Time

Figure B.2b. Lateral load fluctuation with time for HP12x63 weak axis

171
Lateral Displacement (mm)
-50.8 0.0 50.8
60 266.88
50 222.40
40 177.92
30 133.44

Lateral Load (Kips)

Lateral Load (kN)


20 88.96
10 44.48
0 0.00
-10 -44.48
-20 -88.96
-30 -133.44
-40 -177.92
-50 HP12x84 Weak Axis -222.40
Displacement Amplitude =63.5 mm (2.5 in)
-60 -266.88
-3 -2 -1 0 1 2 3
Lateral Displacement (in)

(a) (b)
Figure B.3a. Load-deflection relationship for HP12x84 weak axis

1.69
300 52 Cycles
242.4
200 227.5
Lateral Load (kN)

100

-100

-200

HP 12x84 Weak Axis


-300 63.5 mm (2.5 in) Amplitude
1.0 1.2 1.4 1.6 1.8 2.0

Step Time

Figure B.3b. Lateral load fluctuation with time for HP12x84 weak axis

172
Lateral Displacement (mm)
-50.8 0.0 50.8
100 444.80

80 355.84

60 266.88

Lateral Load (Kips)


40 177.92

Lateral Load (kN)


20 88.96

0 0.00

-20 -88.96

-40 -177.92

-60 -266.88

-80 HP14x117 Weak Axis -355.84


Displacement Amplitude = 76 mm (3 in)
-100 -444.80
-3 -2 -1 0 1 2 3
Lateral Displacement (in)

(a) (b)
Figure B.4a. Load-deflection relationship for HP14x117 weak axis

1.58
43 cycles
400 380.8
346
Lateral Load (kN)

200

-200

-400 HP 14x117 Weak Axis


76 mm (3 in) Amplitude
1.0 1.2 1.4 1.6 1.8 2.0
Step Time

Figure B.4b. Lateral load fluctuation with time for HP14x117 weak axis

173
Lateral Displacement (mm)
-50.8 0.0 50.8
50 222.40

40 177.92

30 133.44

Lateral Load (Kips)


20 88.96

Lateral Load (kN)


10 44.48

0 0.00

-10 -44.48

-20 -88.96

-30 -133.44

-40 HP8x36 Strong Axis -177.92


Displacement Amplitude = 25.4 mm (1 inch)
-50 -222.40
-1 0 1
Lateral Displacement (in)

Figure B.5a. Load-deflection relationship for HP8x36 strong axis

300
1.43
32 cycles

200 204.2
195.6
Lateral Load (kN)

100

-100

-200
HP 8x36 Strong Axis
25.4 mm (1 inch) Amplitude
-300
1.0 1.2 1.4 1.6 1.8 2.0
Step Time

Figure B.5b. Lateral load fluctuation with time for HP 8x36 strong axis

174
View publication stats

VITA

NAME: Nassr Noori Ibrahim Salman

EDUCATION: Bachelor of Science, Civil Engineering, University of Kufa, Najaf,


Iraq, 1998

Master of Science, Civil Engineering, University of Technology,


Baghdad, Iraq, 2001

Doctor of Philosophy, Civil Engineering, University of Illinois at


Chicago, Chicago, Illinois, 2018

TEACHING: Department of Civil Engineering, University of Kufa, Najaf, Iraq,


2006-2013

PROFESSIONAL ASCE, ACI


MEMBERSHIP
PHD PUBLICATIONS Salman, N., and Issa M. A. (2018). “Prediction of the Measured
Response of an Integral Abutment Bridge Using a Finite-Element
Analysis.” Manuscript submitted for publication.

Salman, N., and Issa M. A. (2018). “Displacement Capacities of HP


Piles in Integral Abutment Bridges.” Manuscript submitted for
publication.

175

You might also like