Академический Документы
Профессиональный Документы
Культура Документы
by
SUDARSHAN C KASERA
A THESIS
UNIVERSITY OF CINCINNATI
Cincinnati, OH
2014
Committee Members:
Gian Andrea Rassati, Ph.D.
Ala Tabiei, Ph.D.
Abstract
iv
3.3.2.1 Concrete material model ....................................................................................... 39
3.3.2.2 Girder concrete...................................................................................................... 43
3.3.2.3 Deck concrete........................................................................................................ 44
3.4 Assembly ............................................................................................................................ 44
3.5 Defining steps and loadings ................................................................................................ 48
3.5.1 Steps ............................................................................................................................. 49
3.5.2 Dead load ..................................................................................................................... 50
3.5.3 Creep and shrinkage ..................................................................................................... 50
3.5.3.1 ACI – 209 Report .................................................................................................. 51
3.5.3.1.1 Correction factor for creep and shrinkage coefficient for girder and deck .... 53
3.5.3.2 User subroutine to calculate creep and shrinkage strain ....................................... 56
3.5.4 Live load ...................................................................................................................... 57
3.6 Field output requests ........................................................................................................... 57
3.7 Submission of job for analysis ............................................................................................ 57
3.8 Post processing in ABAQUS using the visualization module ............................................ 58
Appendix B. Volume to surface ratio for the girder and the deck .............................................. 100
Appendix D. Script file for submission of job to the supercomputer ......................................... 104
v
Appendix E. MS Excel Macro to create plots............................................................................. 107
vi
List of Figures
Figure 1.1 Percentage of bridges built annually with three major construction materials ............. 1
Figure 1.2 Number of new and replaced bridges (both Federal and Non-Federal aid) using
different materials from 2003 to 2010 .................................................................................... 2
Figure 1.3 Building material for new and replaced bridges in percentages between 2003 and
2010......................................................................................................................................... 2
Figure 1.4 Prestressed girder in the construction yard .................................................................... 4
Figure 1.5 Prestressed girders erected on the supports (piers) at the construction site .................. 5
Figure 1.6 Complete bridge structure with slab, diaphragm and negative reinforcement .............. 5
Figure 2.1 Positive moment in the connection due to creep in the girder ...................................... 9
Figure 2.2 Negative moment in the connection due to the differential shrinkage in girder and slab
................................................................................................................................................. 9
Figure 2.3 RESTRAINT Model (Miller et al. 2004) .................................................................... 16
Figure 2.4 Comparison of RESTRAINT with PCA study (Miller et al. 2004) ............................ 16
Figure 2.5 Connection capacity (stub girder) specimen (Miller et al. 2004) ................................ 17
Figure 2.6 Variation of strain at bottom of diaphragm with time (Miller et al. 2004) ................. 20
Figure 2.7 Comparison of ABAQUS and FHWA data (Stephen 2006) ....................................... 24
Figure 2.8 Initial temperature rise in bridge decks (Curtis 2007) ................................................. 26
Figure 2.9 Cross section of single cell precast pre-stressed box girder, Unit = mm (Debbarma and
Saha 2011) ............................................................................................................................ 27
Figure 2.10 Prediction of shrinkage strain using model codes (Debbarma and Saha 2011) ........ 28
Figure 2.11 Prediction of creep compliance using model codes (Debbarma and Saha 2011) ...... 28
Figure 2.12 Development of strain vs. time in concrete box girder section (Debbarma and Saha
2011) ..................................................................................................................................... 29
Figure 2.13 Deflection profile of prestressed concrete box girder (Debbarma and Saha 2011) .. 29
Figure 2.14 Pattern of deflection of box girder G-13 at different temperature (Debbarma and
Saha 2011) ............................................................................................................................ 30
Figure 2.15 Compressive strain vs temperature in soffit slab of box girder bridge at the mid-span
(Debbarma and Saha 2011) ................................................................................................... 30
vii
Figure 2.16 Tensile strain vs temperature in deck slab of box girder bridge at mid-span
(Debbarma and Saha 2011) ................................................................................................... 31
Figure 2.17 Temperature readings in deck and top-and bottom flanges (Okeil et al. 2013) ........ 32
Figure 2.18 Rotation of girder ends (Okeil et al. 2013) ................................................................ 32
Figure 2.19 Strains in hairpin bars at both sides of continuity diaphragm (Okeil et al. 2013) ..... 33
Figure 3.1 Model database in Abaqus........................................................................................... 34
Figure 3.2 Type of elements used in the model ............................................................................ 35
Figure 3.3 Two dimensional sketch of girder and deck ................................................................ 36
Figure 3.4 Three dimensional extruded view of girder and deck part .......................................... 37
Figure 3.5 Girder and deck part shown as different sections ........................................................ 37
Figure 3.6 Typical steel part ......................................................................................................... 38
Figure 3.7 Response of concrete to uniaxial loading in tension (Hibbitt et al. 2011) .................. 41
Figure 3.8 Response of concrete to uniaxial loading in compression (Hibbitt et al. 2011) .......... 42
Figure 3.9 Girder and deck mesh in cross section ........................................................................ 45
Figure 3.10 Three-dimensional view of girder and deck mesh .................................................... 45
Figure 3.11 Wireframe showing girder with strands and reinforcement ...................................... 46
Figure 3.12 Wireframe showing slab with strands and reinforcement ......................................... 46
Figure 3.13 Wireframe of the complete model ............................................................................. 47
Figure 3.14 Deck reinforcement with part zoomed ...................................................................... 47
Figure 3.15 Strands ....................................................................................................................... 48
Figure 3.16 Creep coefficient vs. time for the girder and the deck slab ....................................... 55
Figure 3.17 Shrinakge strain vs. time for the girder and the deck slab ......................................... 55
Figure 4.1 Internal and external loads on a body (Hibbitt et al. 2011) ......................................... 61
Figure 4.2 Newton-Raphson approach to solve a nonlinear problem (Simulia, Dassault
convergence class) ................................................................................................................ 63
Figure 4.3 Flowchart showing solution procedure (Simulia, Dassault convergence class).......... 65
Figure 4.4 Screenshot for changing default solution control ........................................................ 66
Figure 4.5 Calibration of Viscosity regularization parameter ...................................................... 70
Figure 5.1 Stress distribution in the girder after release of prestress (camber exaggerated). ....... 71
Figure 5.2 Zoomed view of stress distribution near the midspan ................................................. 72
Figure 5.3 Stress in the girder and the slab at the end of 200 days for 90 -day old girder ........... 72
viii
Figure 5.4 Stress distribution at the end section up to 90 days of creep and shrinkage in the girder
............................................................................................................................................... 73
Figure 5.5 Stress distribution at the end section with slab dead load and creep & shrinkage in slab
(Girder age is 90 days at the time of casting of the slab) ...................................................... 74
Figure 5.6 Stress distribution at the midspan section up to 90 days of creep and shrinkage in the
girder ..................................................................................................................................... 75
Figure 5.7 Stress distribution at the midspan section with slab dead load and creep & shrinkage
in the slab (Girder age is 90 days at the time of the casting of the slab) .............................. 76
Figure 5.8 Displacement at girder top flange at midspan due to creep and shrinkage ................. 78
Figure 5.9 Stress at bottom of slab at midspan for different age of girders .................................. 79
Figure 5.10 Stress at top flange of girder for different age of girders .......................................... 79
Figure 5.11 Stress in the top flange of girder due to variation in creep and shrinkage in the slab
(for 10-day old girder at the time of casting of the slab) ...................................................... 80
Figure 5.12 Stress in the top flange of girder due to variation in creep and shrinkage in the slab
(for a 30-day old girder at the time of casting of the slab).................................................... 81
Figure 5.13 Stress in the top flange of girder due to variation in creep and shrinkage in the slab
(for a 60-day old girder at the time of casting of the slab).................................................... 81
Figure 5.14 Stress in the top flange of girder due to variation in creep and shrinkage in the slab
(for a 90-day old girder at the time of casting of the slab).................................................... 82
Figure 5.15 Total displacement in the top flange of girder at mid-span due to variation in creep
and shrinkage in the slab (for a 10-day old girder at the time of casting of the slab) ........... 82
Figure 5.16 Total displacement in the top flange of girder at mid-span due to variation in creep
and shrinkage in the slab (for a 30-day old girder at the time of casting of the slab) ........... 83
Figure 5.17 Total displacement in the top flange of girder at mid-span due to variation in creep
and shrinkage in the slab (for a 60-day old girder at the time of casting of the slab) ........... 83
Figure 5.18 Total displacement in the top flange of girder at mid-span due to variation in creep
and shrinkage in the slab (for a 90-day old girder at the time of casting of the slab) ........... 84
Figure 5.19 Comparison of stresses for cracked and uncracked slab condition at bottom of slab at
midspan (for a 90-day old girder at the time of casting of the slab) ..................................... 85
Figure 5.20 Comparison of stresses for cracked and uncracked slab condition at top of girder at
midspan (for a 90-day old girder at the time of casting of the slab) ..................................... 86
ix
Figure 5.21 Comparison of displacement for cracked and uncracked slab condition at top of
girder at midspan (for a 90-day old girder at the time of casting of the slab)....................... 86
Figure 5.22 Stress in the prestressing strands for different ages of the girder .............................. 88
Figure 5.23 Comparison of stress in prestressing strands for a 10-day old girder........................ 88
Figure 5.24 Comparison of stress in prestressing strands for a 30-day old girder........................ 89
Figure 5.25 Comparison of stress in prestressing strands for a 60-day old girder........................ 89
Figure 5.26 Comparison of stress in prestressing strands for a 90-day old girder........................ 90
Figure 5.27 Comparison of stress by AASHTO refined loss method and ABAQUS analysis .... 91
x
List of Tables
Table 2.1 Details of positive moment connections in the stub specimens (American Association
of State Highway and Transportation Officials., 2012; Miller et al. 2004) .......................... 18
Table 2.2 Steps defined in the model (Stephen 2006) .................................................................. 23
Table 3.1 Part and number of instances (Single span system) ...................................................... 44
Table 3.2 Various steps defined in the analysis ............................................................................ 49
Table 3.3 Correction factors for the girder ................................................................................... 54
Table 3.4 Correction factors for the deck ..................................................................................... 54
Table 4.1 Default tolerance values for some parameters .............................................................. 66
xi
Acknowledgements
I sincerely express my profound gratitude to Dr. Richard Miller, my advisor, for his
guidance, encouragement, and suggestions throughout the period of my research. The joy and
enthusiasm he has for his research was contagious and motivational. It was his continuous
support and appreciation that kept me focused on my work.
My appreciation is extended to the committee members, Dr. Gian Rassati and Dr. Ala
Tabiei for reviewing my research work and providing helpful suggestions and comments. I
would like to thank Dr. Rassati for providing the ABAQUS computer program and helping out
with the problems during the modeling and analysis. I would also like to thank Dr. Tabiei for
guiding me on the issues I encountered with the finite element analysis of the problem.
My time at the University of Cincinnati was made enjoyable largely due to the many
friends and groups that became a part of my life. I thank all my friends at the University of
Cincinnati, especially, Abhik, Avdhesh, Kunal, Shekhar, and Suryanarayana.
I greatly appreciate the help provided by the Academic Writing Center, University of
Cincinnati, especially, Dani Clark, for proofreading my thesis.
A special thanks to Mr. Aaron Schwartz from Dassault Systemes for taking keen interest
in the analysis and helping in every possible way to resolve the convergence issues in the model
during the hands-on workshop in the ABAQUS Convergence class.
I would like to express the deepest appreciation to Mr. Umesh Rajeshirke, my previous
employer and director at Spectrum Techno Consultants Pvt. Ltd., Navi Mumbai, India, and my
supervisors, Mr. Nirav Mody and Mr. Rakesh Varadkar , who always supported my decision to
pursue higher education.
I have the heartiest gratitude for my dear friends, Abhimanyu, Aditya, Alisha, Deepti and
Sumeet. Thank you all for always being there for me and for being so awesome.
Finally, I would like to thank my family members for always offering me love and
support throughout my life.
This work was supported in part by an allocation of computing time from the Ohio
Supercomputer Center.
xii
Dedication
xiii
Chapter 1. Introduction
The concept of prestressed concrete goes back to 1888 when P.H. Jackson floated the
idea and was granted the patent for a prestressed concrete design for the first time in the United
States. Due to the unavailability of proper materials such as low relaxation steel, the design
concept did not become popular at that time. It was in the 1950s when prestressed/post-tensioned
concrete experienced a beginning of a new era. The idea of using prestressed concrete as a
structural building system was defined by Eugene Freyssinet. The first major prestressed
concrete bridge in the USA was the Walnut Lane Bridge in Philadelphia designed by Gustave
Magnel (Dinges 2009). It was completed in 1951 and had a girder length of 160 feet. Since then,
prestressed concrete has been one of the most sought after construction processes in the bridge
industry.
The following figure demonstrates the increase in the use of prestressed concrete in
bridges since 1950.
Figure 1.1 Percentage of bridges built annually with three major construction materials
(National Bridge Inventory Study Foundation,. 2003)
1
The following figures show the use of prestressed concrete bridges in recent times.
Reinforced
2000 Steel
Concrete
1500
Steel Prestressed
concrete
1000
Other
500
Other
0
2003 2004 2005 2006 2007 2008 2009 2010
Year Built
Figure 1.2 Number of new and replaced bridges (both Federal and Non-Federal aid) using
different materials from 2003 to 2010 (The National Bridge Inventory Database 2011)
Reinforced
Concrete
Prestressed
34%
Concrete
42%
Steel
22%
Reinforced Concrete Steel Prestressed Concrete Other
Figure 1.3 Building material for new and replaced bridges in percentages between 2003
and 2010 (The National Bridge Inventory Database 2011)
2
It is clear from Figure 1.2 and Figure 1.3 that prestressed concrete is the most used material for
the construction of bridges. As shown in Figure 1.3, more than 40% of recently built bridges are
made of prestressed concrete. This is due to the advantages of using prestressed concrete
elements for the construction of bridges and other structures. The construction process is fast,
reduces the use of formwork, and improves quality because of the possibility of better quality
control at the casting yard. A large amount of research has focused onto understanding the
behavior of prestressed concrete under different kinds of loading, and especially time-dependent
loading. These studies carried out on prestressed concrete helped understand its behavior over
time.
The research herein reported aims to understand the long term behavior of a bridge made
from precast pre-stressed concrete girders. The pre-stressed concrete bridge behavior is
significantly affected by time dependent effects like creep, shrinkage and ambient temperature
during its service life (Debbarma, 2011). This study focuses on the response of prestressed
concrete bridges to creep in the girder due to prestress, to the effect of concrete shrinkage in the
girder, and to differential shrinkage between the girder and the deck. The behavior of the bridge
is studied using a three-dimensional finite element model of a single span bridge built using the
ABAQUS Standard 6.11-2 computer program (Hibbitt et al. 2011b).
While this study focuses on single span bridges, the motivation was to create a model
which could eventually be used to study “continuous for live load” bridges.
3
1. The prestressed girder is cast in the construction yard as any other prestressed concrete
girder. The only exception is, some of the strands are left extended at the end of the
girder to be used them as positive reinforcement at the connection. The length of
extension depends on the design parameters. The extended strands are bent to make a
positive moment connection. The same connection can be achieved by embedding mild
steel bars instead of strands or by use of mechanical connectors.
2. The girder then is transported to the construction site. The age at which the girder is
erected on piers can vary for different projects. The girder is designed to act as simply
supported for self-weight and deck dead load.
3. Continuity is achieved by joining the ends of the girder at the pier by pouring the
diaphragm and the deck slab concrete. Positive and negative reinforcement are provided
to counteract the moment developing at the connection. Negative reinforcement is
provided in the slab as a longitudinal continuity bars while positive reinforcement is
provided at the bottom part of the girder at the support. The configuration of positive
reinforcement varies, most commonly consisting of bent bar or bent strand. The positive
moment connection is discussed in detail in the literature review section of this
document. Figure 1.4 throughFigure 1.6 depict the construction sequence of a
‘continuous for live load’ bridge.
4
Prestressing strands Positive reinf.
Diaphragm
Figure 1.6 Complete bridge structure with slab, diaphragm and negative reinforcement
(Typical longitudinal sections at pier. Details not shown for clarity)
5
shrinkage in concrete over time, the effects of daily/seasonal temperature changes, relaxation of
the steel and time dependent change to the modulus of elasticity of concrete.
As previously noted, the overall effect of the time dependent effects may give rise to
positive moments at the bottom of the diaphragm at the support. The age of the girder at the time
of the casting of the deck also plays an important role for the development of the positive
moment at the bottom of the support. Current specifications require an age of 90 days for the
girder in order to minimize the development of a positive moment at the bottom of the girder and
reduce cracking. The construction sequence of the bridge plays an important role in determining
its long-term behavior.
The experimental methods to study the long-term effects are time-consuming and require
many resources. The available analytical methods fail to accurately predict the behavior in most
cases. Therefore, finite element analysis can be a quick and reliable technique to study the
behavior of the ‘continuous for live load’ bridges subjected to dead load, creep and shrinkage,
temperature loads, and live loads. Finite element programs may or may not have a built-in
method to simulate these conditions as needed, but modifications can be made to develop an
approach to analyze such problems. An approach was developed previously by Eldhose Stephen
(Stephen 2006) in his thesis to analyze such problems using available resources in the finite
element program ABAQUS Standard (Hibbitt et al. 2011b). A similar approach is used in the
current project to simulate creep and shrinkage in a single span prestressed concrete bridge
girder.
The current project focuses on simulating the effects of creep and shrinkage on a simply
supported single span bridge girder with a deck slab over it. The idea is to provide a basis for the
analysis of such bridges using the finite element method. The effect of age variation of the
girders at the time of casting the slab on the overall system is also studied in the current thesis.
The role of differential shrinkage between the girder and the deck will be investigated. ABAQUS
Standard 6.11-2, a general purpose finite element software, is used to create a three-dimensional
finite element model of a single span bridge. The procedure followed is documented in detail
along with the results and can be used in the future to develop a multiple span model for
‘continuous for live load’ bridges.
6
Chapter 2. Background and Literature Review
Continuous for live load bridges have been used in the construction industry since the
1960s. This was done to allow the use of shallower sections. During the early 1960s, the
performance of these bridges was studied by researchers at the Construction Technology
Laboratory (CTL), at the Portland Cement Association (PCA). Various aspects of this kind of
bridge were studied: feasibility of the continuity connection, shear connections, flexural strength,
creep and shrinkage studies, and performance under live load (Kaar et al. 1960). The study on
creep and shrinkage was conducted on a two-span continuous girder by Mattock et al. in 1961.
The report was published as “Precast-Prestressed Concrete Bridges 5: Creep and shrinkage
studies.” Earlier prestressed concrete bridges were made continuous by providing negative
reinforcement in the slab across both the girders. There was no provision for positive steel at the
bottom of the girder at the support. The design detail for the positive moment connection was
published by PCA in a design guide, “Design of Continuous Highway Bridges with Precast,
Prestressed Concrete Girders” (Freyermuth 1969). The design details were based on the
outcomes of various long and short term tests carried out during the research. National
Cooperative Highway Research Program (NCHRP) report 322, “Design of Simple-Span Precast
Prestressed Bridge Girders Made Continuous” was published in 1989 by Oesterle et al. to
address uncertainties in the PCA research. This report concluded that the positive moment
developed at the support negated any benefit from continuity.
Because bridge engineers thought this type of bridge was useful and did not entirely agree
with the conclusion that the positive moment negated the benefit, further research was done by
NCHRP and the reports were published as NCHRP Report 519, “Connection of Simple Span
Precast Girders for Continuity,” (Miller et al. 2004). The scope of the NCHRP Report 519
included a survey of the then existing construction practice followed by an Excel-based
analytical model, RESTRAINT, which was used to analyze connection behavior. The
experimental models, which included six stub specimens and two full-size specimens, were
tested for different connection configurations and construction methods as part of the project.
The results of the experimental study were used to create the current language in the American
Association of State Highway and Transportation Officials (AASHTO) Load and Resistance
Factor Design (LRFD) Bridge Design Specifications (AASHTO 2014).
7
The Bridge Deck Task Force set up by the New York State Department of Transportation
conducted an evaluation of the cracking in the bridge decks of continuous bridges. The reasons
were mainly attributed to thermal stress due to the restraint of the deck while it cools, live load
stresses on continuous bridge, and stress due to concrete shrinkage while restrained by the
superstructure. Various other studies were carried out to study the effects of creep, shrinkage and
temperature variation on bridges and some of them are presented here.
8
Figure 2.1 Positive moment in the connection due to creep in the girder (Mattock 1961)
Figure 2.2 Negative moment in the connection due to the differential shrinkage in girder
and slab (Mattock 1961)
In the study by Mattock, the reaction and strain were measured in the negative continuity
reinforcement and the deflections at the midpsan were also measured. The variation of support
reaction with time showed an increase in the reaction for the first 35 days after the removal of the
deck formwork. Henceforth, the reaction kept on decreasing and became negative at around 220
days and continued to decrease before levelling out. Increase in reaction is caused by the
differential shrinkage between the slab and the precast. A similar response was observed with
respect to strain in the continuity reinforcement. Strain in the negative continuity reinforcement
was initially tensile and reached its peak at around 35 days and started decreasing and eventually
became compressive. This is an indication of the slowing down of a negative restraint moment,
which gradually becomes positive. This is caused by the dominance of creep due to prestressing
in the girder, which eventually negates the effect of differential shrinkage. The report concluded
that creep and shrinkage significantly affect the structure but the ultimate load carrying capacity
of the structure remains unaffected.
The accuracy of the study is questionable because half scale models were used by PCA.
In a half scale model, the volume decreases with the cube while the surface decreases as the
square of the size. Since the volume-surface ratio is an important parameter to predict creep and
shrinkage, scaled models can produce erroneous results.
The PCA published a design guide, “Design of Continuous Highway Bridges with
Precast, Prestressed Concrete Girders,” (Freyermuth 1969), based on the outcomes of various
tests conducted during the study. The design method to determine the prestressing details and
positive moment connection details were provided in the guide. The prestressing steel was
9
designed for simple spans, considering the dead load, live load, and impact loads. The design of
the positive moment connection was based on the restraint moments set up at the supports due to
creep, differential shrinkage, and live/impact loads, after the continuity is established. These
restraint moments increase the positive moments at midpsan as well, but the design guide stated
that the additional moment requirement was satisfied by the prestressing detail designed for
gravity loads (Freyermuth 1969).
The restraint moment set up at the support is an important feature of ‘continuous for live
load’ bridges, and the detailed guidelines were outlined by the PCA to calculate them. The
calculation process can be divided into the following stages:
1. The girder and the deck are assumed to be cast simultaneously and the prestressing and
dead loads are applied to a monolithic structure.
2. An elastic analysis of the indeterminate continuous structure provides the moments at the
mid-spans and supports.
3. The restraint moments due to creep are calculated by multiplying the elastic moments by
a creep factor accounting for the actual difference in age between the girder and concrete.
4. The restraint moments due to differential shrinkage are calculated using a simplified
equation for the moment setup along the length of the structure,
𝑡
𝑀𝑠 = 𝜀𝑠 𝐸𝑏 𝐴𝑏 (𝑒2 + ) Eqn. (2.1)
2
𝜺𝒔 = Differential shrinkage
𝐸𝑏 = Elastic modulus of deck concrete
𝐴𝑏 = Cross-sectional area of deck slab
𝑡
(𝑒2 + 2) = Distance between mid-depth of the slab and centroid of the composite section
An indeterminate analysis using this moment provides the support moments in the continuous
structure.
5. The final restraint moment at the support is calculated as,
−𝜙
1 − 𝑒 −𝜙
𝑀𝑟 = (𝑌𝑐 − 𝑌𝐷𝐿 )(1 − 𝑒 ) − 𝑌𝑠 ( ) + 𝑌𝐿𝐿 Eqn. (2.2)
𝜙
(𝑌𝑐 − 𝑌𝐷𝐿 ) = Elastic moment in continuous, monolithic girder and deck, cast and
prestressed simultaneously
10
(1 − 𝑒 𝜙 ) = Factor converting an elastic moment to a restraint moment due to creep in
precast girder after continuity is established
𝑌𝑠 = Restraint moment due to differential shrinkage
𝑌𝐿𝐿 = Positive live load plus impact moment
1−𝑒 −𝜙
( ) = Adjusts the moment due to differential shrinkage to account for creep
𝜙
The positive moment connection at the support was to be designed for the final restraint
moment obtained by the above equation. Similarly, a negative reinforcement at the connection
was designed for the moment occurring due to live load, impact load, and dead load arising after
continuity is established. This design procedure led to the addition of Article 9.7.2 ‘Bridges
Composed of Simple-Span Precast Prestressed Girders Made Continuous,’ in the AASHTO
Standard Specifications (American Association of State Highway and Transportation Officials.,
2002) to design the highway bridges. This was the only design methodology available for
designing precast/prestressed bridge girder connections at that time. The economy of the
‘continuous for live load’ bridges along with the convenience of the PCA design procedure led to
widespread acceptance of this analysis/design process. Most of the research involving analytical
models carried out since then has been based on or validated using the PCA model and data.
The PCA study failed to take into consideration certain behavioral aspects of ‘continuous
for live load’ bridges:
1. The free shrinkage of the deck is restrained by the top of the girder and the
reinforcement, which results in development of tensile stresses in the deck. These stresses
can be large enough to cause cracking in the deck. Once the deck cracks, the assumption
of the structure being monolithic becomes invalid and the restraint moment developed
due to differential shrinkage is released. The negative camber induced by differential
shrinkage also gets negated due to deck cracking, and camber in the girder reverses to
cause positive camber.
2. The PCA model did not take into account the possibility of tensile creep in the deck
caused by the tensile stresses setup due to restraint of differential shrinkage.
3. The effect due to daily temperature variation was not considered in the PCA study.
11
4. The assumption that the deck and the girder have similar creep and shrinkage properties
was incorrect.
The above issues needed to be addressed in order to create an effective and efficient design
methodology for ‘continuous for live load’ bridges. They were addressed in the further
research work carried out by NCHRP and other transportation organizations. Some of them
are discussed in the following sections.
12
The moments due to live load plus impact were calculated using another computer program,
BRIDGELL.
Various conclusions were drawn through this study. It was concluded that the advantages
of this kind of bridges have made them a common choice of construction regardless of the
difficulty, time consumption and costly positive moment connection. The AASHTO
specifications are vague for positive moment connection design for continuous for live load
bridges and most States used the PCA method to design the positive moment connection. It was
also concluded that the creep and shrinkage results based on the calculation of ACI 209 report
(ACI Committee 209-Creep and Shrinkage. 2008) gives reasonably accurate predictions. The
service moment parametric study concludes that there is no structural benefit of a positive
moment connection. The positive moment connection gives rise to a restraint moment at the
support which in turn increases the midspan moment making it almost similar to the value as
expected in simply supported beam. The age of the girders plays an important role in the
continuity of the structure. A delay in the casting of the deck and the diaphragm leads to a high
level of positive moment continuity due to higher differential shrinkage. However, too much
delay can increase a negative restraint moment at the support demanding more negative
reinforcement, which can cause transverse cracking in the deck. The construction sequence
affects the development of a restraint moment. A deck cast prior to the diaphragm increases the
resultant positive moment at the midspan while the opposite sequence of casting slightly
decreases the resultant positive moment at the midspan. Hence there is no economic advantage
of casting in sequence and the simultaneous casting of the deck and the diaphragm is the simplest
procedure.
The conclusions drawn in NCHRP Report 322 identified areas which needed further
research to understand and improve the behavior of continuous for live load bridges. It suggested
studying the effectiveness of an approach to provide special types of preformed joints at the
support and unbonding the deck reinforcement or unbonding the deck to girder interface for a
certain length on each side of the joint. Another potential area of research can be a hybrid girder
with partial post-tensioning designed to take more dead load moments after continuity being
established. The effect of temperature and moisture gradient in the deck and girder section was
not considered in this research. Both of these, along with other time-dependent loading, can give
rise to a positive moment at the connection and needs further study. There is also a need to
13
develop more appropriate and simplified design procedures for shear in continuous prestressed
bridge girders. Along with the above suggestion, it is also important to note that most of the
work presented in the report was analytical in nature. While creep and shrinkage tests were
performed on the concrete specimen leading to an improved database for the future, the
measurement of creep and shrinkage from real time monitoring data of an existing bridge or
bridge specimen was not included. The effect of the time-dependent loading needs to be studied
experimentally and using finite element methods to help in improving the design of the positive
moment connection.
14
The survey indicated that the use of positive moment connection reinforcement at the
support was prevalent in most states and the most used connection detail was the bent bar and
bent-strand type connection. Three-fourths of the respondents overlapped the bar or strand in the
diaphragm and half of the respondents provided transverse reinforcement through the beam into
the diaphragm. I-girders and bulb-T girders were the ones where these types of connections were
mostly used with some of the respondents using it for box girders and other state-specific shapes.
Reponses regarding embedment of girder into the diaphragm were also a part of the survey and it
was found out that the majority of the responses included embedment of girder into the
diaphragm with embedment depth varying anywhere from 2 to 12 inches. The 28-day concrete
strength varied from 4000 to 9000 psi for girders while deck/diaphragm concrete had a strength
of 3000 to 5000 psi at 28 days. High-performance concrete was also used in some cases. The
support conditions in 80% of the responses included bearings placed under the girder ends. The
cost to provide a positive moment connection was found to be $200 per girder, which is quite
low as compared to the overall cost the girder.
The sequence of construction is an important aspect of the continuity connection. The
majority of the survey respondents cast the diaphragm and the slab together while few cast the
diaphragm or part of it before casting the slab. The minimum age for the girder varied between
28 and 90 days. The common problem was congestion of reinforcement in the diaphragm area
leading to improper consolidation of concrete and bending of the strands/bars after the girder is
cast. However, the problems were rated as being of minor significance (Miller et al. 2004).
The complete study was conducted in three major parts. First, analytical studies were
performed by creating a standard spreadsheet program called RESTRAINT, which was a
modified version of BRIDGERM. The analytical study was important to get an idea about the
configuration of specimens for the experimental studies. Six different types of stub specimen
were constructed with different connections. Along with the six stub specimens, two full size
specimens were constructed and their responses to different kinds of loading such as prestress
load, live loads, time-dependent loads such as creep and shrinkage and temperature loads were
documented thoroughly. Three different finite element models (no connection, bent-strand and
bent-bar) of the stub specimen were also modeled using finite element program, ANSYS, as a
part of the study.
15
2.3.1 Analytical study
The RESTRAINT program models a two-span continuous structure assuming that there
is a support at each end of the girder as shown below.
The program calculates the internal moments due to creep in the prestressed girder and
shrinkage in the girder and the deck on the basis of the creep and shrinkage model documented
in the American Concrete Institute 209 report (ACI Committee 209-Creep and Shrinkage. 2008).
The loss of prestress force is also calculated using the method outlined in the Precast/Prestressed
Concrete Institute Handbook. The result of RESTRAINT was compared to the results of the
original PCA tests showing a reasonable agreement.
Figure 2.4 Comparison of RESTRAINT with PCA study (Miller et al. 2004)
16
It was concluded from the parametric studies that the age of the girders at the time when
continuity is established is the most important factor in the behavior of the connection. The
amount of positive reinforcement at the connection plays an important role on the performance
of the continuity. There has to be a balance in the amount of positive reinforcement to achieve
continuity and limit the restraint moment. An increase in the restraint moment can increase the
positive moment at the mid-span to such an extent that the girder has to be designed for simply
supported condition. It was concluded that the positive reinforcement should be limited, so that
the capacity of the connection is at least 0.6 Mcr and does not exceed 1.2 Mcr; where Mcr is the
positive cracking moment based on the gross cross section of the girder and slab but the strength
of the diaphragm concrete
Figure 2.5 Connection capacity (stub girder) specimen (Miller et al. 2004)
Six stub specimens with different configurations of the positive moment connections
were tested. The positive moment connections with bent-strand or bent-bar were designed for a
capacity of 1.2 Mcr obtained from the analytical study where Mcr is the positive cracking
17
moment. The details of the positive moment connections in the stub specimens are shown in
Table 2.1.
Table 2.1 Details of positive moment connections in the stub specimens (Miller et al. 2004)
Specimen Type of Specimen Diaphragm Girder End Special Cycles to
Number Width (in.) Embedment (in.) Feature failure
1 Bent strand 10 0 None 16,000
2 Bent bar 10 0 None 25,000
3 Bent strand 22 6 None 55,000
4 Bent bar 22 6 None 11,600
5 Bent bar 22 6 Extra stirrups 56,000
in diaphragm
6 Bent bar 26 8 Web bars 13,3000
All six specimens were subjected to cyclic loading until failure. The first specimen
survived for 16,000 cycles before the concrete on the bottom of the diaphragm split and popped
off indicating a slipping and pull-out of the strand/s. Specimen 2 had a bent-bar configuration
and it was more difficult to construct. This specimen lasted for 25,000 cycles when diagonal
cracks started forming on the faces of the diaphragm and part of the diaphragm spalled off. The
bars fractured, showing signs of failure due to fatigue. The third specimen had a bent-strand
configuration with the ends of the girder embedded into the diaphragm. It lasted 55,000 cycles
before failing. It exhibited cracking and spalling on the face of the diaphragm. Specimen 4 was
similar to specimen 2 except the embedment of the girder ended into the diaphragm. It only
lasted 11,600 cycles and the reason for early failure was uncertain. There might have been
uneven stresses in the bent-bar due to bad construction leading to early failure. Specimen 5 was
similar to specimen 4 with additional stirrups in the diaphragm, close to the outer edge of the
bottom flange. The specimen failed after 56,400 cycles. The additional stirrups were spanning
the crack and prevented failure providing additional ductility at the connection. The sixth
specimen consisted of bent-bars and horizontal bars passed through the web of the beam. This
certainly added stiffness and capacity to the connection making it last 133,000 cycles. Upon
close inspection it was found that the web of the beam had cracked, which is not desirable.
From the results, it was evident that bent-strand with embedment can be beneficial. When
non-embedded bent-strand was used, the girders separated from the diaphragm without
damaging the face of the diaphragm, while an embedded specimen showed a pull-out type of
18
failure. Embedded specimens also have lower strain than their counterpart. It was also clear that
an asymmetrical configuration of the bars is undesirable at the connections.
19
diaphragm as seen from Figure 2.6. A similar response was observed at other locations in the
system.
Figure 2.6 Variation of strain at bottom of diaphragm with time (Miller et al. 2004)
After the construction process was completed, thermal effects played an important role in
the response of the structure. The end reactions due to temperature variation during the day were
significant. The changes in reactions caused a positive moment in the diaphragm in the range of
+- 250 kip-ft per day. This was more than 50% of the cracking moment value. The differential
shrinkage between the girder and the deck was not observed as expected from analytical studies.
The observed behavior was consistent with field studies of bridges.
The girders were tested for continuity after the monitoring period. It was expected that
the cracks formed at the girder-diaphragm interface would reduce the continuity or eliminate it.
The reactions and the strains obtained from the continuity test were consistent with the results
expected from a continuous system. So cracking at the girder-diaphragm interface did not
significantly affect the continuity of the structure.
The second full-size specimen was tested for positive moment/continuity and negative
moment capacity. The result of the test concluded that continuity was maintained in the system
even after positive moment cracking at the joint. Loss of continuity was observed when the
connection was about to fail due to cracks in the slab and in the diaphragm.
20
The negative moment capacity was tested on the same specimen after testing for the
positive moment. The negative moment reinforcement was provided in the deck as per AASHTO
LRFD specifications. Cracking due to negative moment occurred at a value lower than the
calculated value. This was attributed to the fact that the deck was already cracked due to positive
moment testing. The bottom of the girder crushed at an applied moment which exceeded the
failure moment capacity calculated using actual material properties. So it was concluded that
positive moment testing reduced the negative cracking moment but the negative moment
capacity of the structure was not affected.
21
some creep and shrinkage before the establishment of continuity and reduce the formation of
positive moment at the connection. The study also concluded that bent strand positive moment
connections designed using the equations developed by Salmons et al. (1974) performed
adequately and were adopted in the AASHTO LRFD Bridge Design Specifications. The bent-bar
connections also performed adequately if they are embedded into the girder with extended hooks
embedded into the diaphragm as per the AASHTO LRFD Specifications. The embedment of
girder into the diaphragm reduces the positive moment stress but it is difficult to quantify its
effects. The placement of additional stirrups in the diaphragm just outside the girder improves
the ductility of the connection. The horizontal bars through the web of the girder increase the
strength and the ductility of the connection but cause significant cracking in the girders at failure.
The idea of limiting the tensile stress at the bottom by partial pouring of the diaphragm was
found out to be only partially effective. The report also concluded that temperature effects are
significant on the structure. The end reaction can vary +-20% due to the daily temperature
variation causing a positive moment at the connection, almost equal to 2.5 times the positive live
load moment. The study also concluded that the positive moment cracking in the diaphragm does
not necessarily reduce the continuity as opposed to the predictions made from analytical studies.
The negative moment capacity of the connection is also not affected by the positive moment
cracking until the crack extends into the slab. Based on the findings of the research, various
revisions were proposed to the AASHTO LRFD Bridge Design Specifications.
22
compatibility with FORTRAN subroutines. The results obtained were compared to the existing
experimental results to validate the usage of ABAQUS to model such systems.
The non-linear material properties were used for the girder and the deck concrete to
simulate the tension softening and the compression hardening in the concrete. Different concrete
material models were used for the girder concrete and the slab concrete. The Extended Drucker
Prager model and the Concrete Damaged Plasticity Model were used for the girder and the deck
concrete, respectively. Elastic properties were used for the steel, since the steel stresses were
expected to be within the elastic limit. The loads considered for the analysis were prestress load,
dead loads of the girder and the deck, and time dependent effects such as creep in the girder and
shrinkage in the deck. Creep and shrinkage are concurrent properties but to simplify the analysis,
creep was applied only to the girder elements and shrinkage was applied to the deck elements.
The sequence of construction of the bridge was taken into consideration in the analysis by using
“Model Change” keyword in the ABAQUS program. The details of different steps defined in the
model are shown in Table 2.1.
Table 2.2 Steps defined in the model (Stephen 2006)
Name Type Time Stabilizatio Maximum Increment size
period n no. of Initial Minimum Maximum
(days) increments
DEACTIVATE Static 1 No 100 1 1 1
RELEASE Static 1 No 100 1 1 1
CREEP Static 164 No 1.00E+06 1 1.00E-08 5
DECKDL Static 1 No 100 1 1 1
POURDECK Static 1 No 100 1 1 1
SHRINKAGE Static 500 Yes 1.00E+06 0.1 1.00E-20 5
FORTRAN subroutines were used along with ABAQUS CAE to simulate different
behaviors in the model. Subroutine UEXPAN( ) was used to calculate creep and shrinkage in the
girder and the deck, respectively, on the basis of the creep and shrinkage model reported in ACI-
209 (ACI Committee 209-Creep and Shrinkage. 2008). The results obtained from the analysis
were compared to the study conducted by the Federal Highway Administration (FHWA 2001)
and found to be in agreement with the data provided in the FHWA study on a bridge in New
Hampshire. A comparison of total strain vs time is shown in Figure 2.7.
23
Figure 2.7 Comparison of ABAQUS and FHWA data (Stephen 2006)
The results from the ABAQUS analysis validated its usage to simulate the response of
such structures. A similar approach to study the behavior of a single span prestressed bridge
girder is taken in the current project and details about the analysis are discussed in the next
chapter.
2.5 NYSDOT Bridge Deck Task Force Evaluation of Bridge Deck Cracking on
NYSDOT Bridges (Curtis 2007)
Deck cracking is a common phenomenon. A Bridge Deck Task Force (BDTF) was set up
by the New York State Department of Transportation (NYSDOT) to evaluate the cracking of the
deck on the NYSDOT bridges. The BDTF studied the historical designs of the bridges and
conducted a survey on existing decks which included newly constructed decks as well. The
decks of two different kinds of girder configurations were investigated. Decks on adjacent
prestress concrete girders and decks on spread steel girders were considered for the study. The
various factors influencing the deck cracking were noted. Since concrete cracks easily in tension,
causes of tensile stresses in the deck were studied and their effects on bridge deck durability
24
were analyzed. Two case studies were investigated and recommendations were made to reduce
bridge deck cracking and its effects on the bridge structure.
The survey of the existing decks on prestressed concrete box girders demonstrated two
types of cracking: longitudinal cracking, which follows the beam joints, and random cracking,
with relatively large spacing of the cracks (Curtis 2007). The longitudinal cracks followed the
beam keyway and were caused by the differential movement of the beam at the keyways. The
random cracking was attributed to the fact that the wet concrete of the deck was placed on dry
precast beams. Observations on concrete core samples showed that cracks were initiating and
propagating from the bottom of the concrete decks (Curtis 2007).
The survey of the then existing bridge decks, which included newly constructed bridges,
showed serious cracking issues. 38% of single span bridges and 67% of multiple span bridges
had significant cracking (Curtis 2007). Based on the database created from the research
conducted on the construction records, as-built plans, and materials records, it was determined
that concrete strength, concrete cover, and pour temperature were some of the most influential
factors in concrete bridge deck cracking. The deck concrete should meet minimum strength
criteria (3000 psi for NYSDOT) but it should not be too strong, to minimize deck cracking. The
concrete cover directly controls cracking in the deck and the crack width. To minimize corrosion
of the reinforcement from the salt laden water during the winter, increased concrete cover is
provided on the top surface of the deck; however, this results in more cracks and larger crack
widths. A warmer temperature on the day of pouring the deck concrete can reduce the cracking,
but it is difficult to control this parameter.
The cause of cracking in the deck was the development of tensile stresses. There are three
causes for the development of tensile stresses in the deck (Curtis 2007):
1) Thermal stress due to restraint of the deck while it cools.
2) Live load stresses on the continuous bridge.
3) Stress due to concrete shrinkage while restrained by the superstructure.
Thermal stresses develop in the deck as early as the time of initial set of concrete. There is a
significant temperature variation in the deck concrete in the first couple of days from the time of
pour. There is an increase in the concrete temperature due to the release of heat from the
chemical reaction of hydration. The increase in temperature starts from the point when the
concrete gains its initial strength but is still not set. The temperature of the top flange of the steel
25
girder also rises with the increase in concrete temperature but it is still cooler than the concrete,
as seen from Figure 2.8.
26
Several recommendations were made to reduce cracking in the concrete bridge deck
along with their advantages and disadvantages. Some of the recommendations to reduce the deck
cracking are mentioned below:
1) Minimize size of the top flange
2) Reduce effective strength of the deck concrete
3) Shrinkage reducing agents
4) Two course decks
5) Decrease concrete cover
In addition methods to treat the bridge deck cracking were also outlined in the paper.
Figure 2.9 Cross section of single cell precast pre-stressed box girder, Unit = mm
(Debbarma and Saha 2011)
27
The time variation of shrinkage and creep strain predicted from the ACI 209 model (ACI
Committee 209-Creep and Shrinkage. 2008)and the CEB-FIP 90 model (Comité euro-
international du béton., 1993) using a computer program and are shown in Figure 2.10 and
Figure 2.11 respectively. The strain data due to the combine effect of creep and shrinkage was
collected in soffit, web and deck slab of the girder. The comparison of predicted data and
observed data showed the occurrence of maximum strain in the structure nearly at the same time
that is around 350 days from the concreting of the girder. The maximum deflection in the bridge
profile was also observed at around 390 days, as seen from Figure 2.13. The deflection at the
mid-span was hogging in nature due to the presence of prestressing in the girder.
Figure 2.10 Prediction of shrinkage strain using model codes (Debbarma and Saha 2011)
Figure 2.11 Prediction of creep compliance using model codes (Debbarma and Saha 2011)
28
Figure 2.12 Development of strain vs. time in concrete box girder section (Debbarma and
Saha 2011)
Figure 2.13 Deflection profile of prestressed concrete box girder (Debbarma and Saha
2011)
Thermal effects have an influence on the behavior of the box girder bridge. The effect of
change in temperature on concrete creep and shrinkage is basically two-fold. First, they directly
influence the rate of creep and shrinkage with time. Second, they affect the rate of aging of the
concrete, i.e. the change of material properties due to progress of cement hydration (Debbarma
and Saha 2011). The deflection pattern for one of the girders was obtained after 600 days of
concreting under no traffic condition. The maximum deflection was observed at the mid-span
and the deflection was highest for the maximum value of temperature, as seen in Figure 2.14.
29
Figure 2.14 Pattern of deflection of box girder G-13 at different temperature (Debbarma
and Saha 2011)
The strain values obtained from the vibrating wire gauges installed in the mid-span in
soffit and the deck slab showed development of compression in the soffit and tension in the deck
slab due to temperature variation. For a temperature increase of 5oC (41o F), the girder deflected
by 9.5mm (0.37 inches) at the mid-span. This daily variation of temperature in the structure may
give rise to the formation of cracks and causes deterioration at a latter age (Debbarma and Saha
2011) but quite a bit before the end of the design life.
Figure 2.15 Compressive strain vs temperature in soffit slab of box girder bridge at the
mid-span (Debbarma and Saha 2011)
30
Figure 2.16 Tensile strain vs temperature in deck slab of box girder bridge at mid-span
(Debbarma and Saha 2011)
The study concluded that creep and shrinkage cause continuous deformation in the bridge
with age. This final deformation depends on the initial deformed configuration of the bridge that
is after the application of prestress. The daily temperature variation can cause thermal cracks in
the structure and can reduce the service life of the structure. Hence, it is important to consider
these effects while designing the bridge.
31
Figure 2.17 Temperature readings in deck and top-and bottom flanges (Okeil et al. 2013)
32
Figure 2.19 Strains in hairpin bars at both sides of continuity diaphragm (Okeil et al. 2013)
The girder end rotations on both sides of the connection followed the same trend; hence
the diaphragm was able to provide continuity to the structure. Similarly the strain in hairpin bars
at both sides of the continuity diaphragms is similar. So the continuity detail was able to transfer
force across the diaphragm from one girder to another due to long-term effects and live loads.
The study concluded that the girder cambers up at the mid-span due to creep and thermal effects,
which caused positive moment at the continuity connection. The thermal effects can cause large
restraint moment due to seasonal and daily temperature variation and is an important factor in the
design. Thermal effects along with creep should be considered to determine the magnitude of the
positive restraint moment. The positive restraint moment may cause cracking at the diaphragm
and girder interface, which can affect the performance of the girder. The results from the study
validate the performance of the positive moment connections recommended by NCHRP Report
519 and also support the idea that the same connection can be used for different configuration of
bridges constructed using different kind of girders.
33
Chapter 3. Finite Element Modeling and Analysis
In order to eventually study continuous for live load bridges, it is necessary to first
determine how to model creep, shrinkage and deck cracking in a finite element model. Because
of the complexity of modeling a two span system, a non-linear three dimensional finite element
model of a single span bridge is created in ABAQUS Standard 6.11-2 as a first step. There are
various ways to create a working model in ABAQUS, such as using ABAQUS CAE, keyword
version, and Python script. This study uses the ABAQUS interactive version in the form of
ABAQUS CAE to create the model. The feature to edit the keywords directly in the input file
created using the CAE is also used when needed.
The model database is divided into two parts: Model and Analysis. They are further divided into
several modules covering different aspects of modeling in each module. The figure below gives a
better understanding of the model database division. The various modules in the model database
definition are parts, materials, sections, assembly, steps, field
output requests, history output requests, interactions, constraints,
fields, loads, boundary conditions, and predefined fields. Some
of the steps involved in creating a working model are creating
parts, defining material properties, defining sections, assembling
the parts to create the required structure, defining various steps
of analysis, adding boundary conditions, and submitting the job
for analysis, etc. After submitting the job, we can monitor the
progress of the job. Upon completion of the job, the
visualization module can be used for the post processing of the
model.
The current model is a single span prestressed concrete
bridge girder with a deck slab on it. The girder is a 50 foot long
Type III AASHTO I girder and represents half the specimen
used in NCHRP 519 (Miller, et al. 2004). The steps to be
followed to create a successful model in ABAQUS are discussed
in this chapter.
2
end 2
1 end 1
35
The C3D8 element is a general purpose, fully integrated linear or first-order brick
element with 2x2x2 integration points. The node numbering follows the convention shown in
Figure 3.2. These elements are stress/displacement elements with three displacement degrees of
freedom per node. The C3D8 elements have a better convergence rate in a regular mesh. The
T3D2 elements are three-dimensional truss elements having two degrees of freedom per node.
They can be to model slender, line-like structures, such as strands, reinforcement, etc. The
geometry of parts are shown here.
36
Figure 3.4 Three dimensional extruded view of girder and deck part
37
Figure 3.6 Typical steel part
38
3.3.1 Steel
An elastic material property is assigned to steel as the stresses in the steel during the
analysis are expected to remain well under the elastic limit. The same material definition is used
for both reinforcement steel and prestressing strands, as their properties are the same. The cross-
section area of the reinforcement and the strand differ and hence have different section
properties. The steel material is defined by the following characteristics in the model:
Elastic modulus
Poisson’s ratio
Density
3.3.2 Concrete
An inelastic material model is used for simulating the behavior of concrete in the current
project. It is expected that concrete will reach its stress limit in tension and hence it is important
to choose a material that will successfully capture its behavior after it reaches the stress limit in
tension.
39
are many features of the CDP model, which makes it appropriate for the current analysis. Some
of the features of the CDP model are listed below. The concrete damaged plasticity model:
a) provides a general capability for modeling concrete and other quasi-brittle materials
in all types of structures
b) uses concepts of isotropic damaged elasticity in combination with the isotropic tensile
and the compressive plasticity to represent the inelastic behavior of concrete
c) is intended primarily for the analysis of reinforced concrete structures
d) can be used in conjunction with a viscoplastic regularization of the constitutive
equations in ABAQUS/Standard to improve the convergence rate in the softening
regime (Hibbitt et al. 2011a)
Concrete behaves differently to tensile and compressive loads. The strength of concrete in
compression is much greater than its strength in tension. The CDP model assumes that the
concrete material fails mainly in tensile cracking and compression crushing. The failure surface
𝒑𝒍 𝒑𝒍
is controlled by two independent hardening variables: 𝝐𝒕 and 𝝐𝒄 , which are referred to as tensile
and compressive equivalent plastic strains, respectively. The non-linear concrete response to
uniaxial loading in tension and compression is shown in Figure 3.7 and Figure 3.8, respectively.
The stress-strain response of concrete is linearly elastic under the uniaxial tensile load
until the stress reaches a point of failure. Beyond the failure point that is the plastic region, a
softening stress-strain response is seen in the concrete. This induces strain localization in the
concrete because of formation of micro-cracks in the concrete. To simulate the tensile behavior
the user needs to input post failure stress-strain data in the CDP model. The cracking stress vs
cracking strain data corresponds to the post failure data. This data can be obtained from
experimental results on the concrete material or from numerical models.
40
Figure 3.7 Response of concrete to uniaxial loading in tension (Hibbitt et al. 2011a)
41
Figure 3.8 Response of concrete to uniaxial loading in compression (Hibbitt et al. 2011a)
The behavior of concrete under uniaxial compressive load is linear up to the point of
initial yield. After concrete has reached the peak stress value, the plastic behavior of concrete is
characterized by stress hardening followed by strain softening beyond the point of ultimate
stress. Similar to tensile data, cracking stress vs. inelastic strain needs to be provided by the
user, which can be obtained from experimental results or numerical models.
Although it is a simplified approach, the CDP model captures the important features of
concrete behavior in a structure. The parameters needed to define the concrete damaged
plasticity model are:
– dilation angle in degrees measured in the p-q plane at high confining pressure
– Flow potential eccentricity, default is 0.1
bo/co – ratio of the initial equibiaxial compressive yield stress to the initial uniaxial
compressive yield stress, default value is 1.16
Kc – ratio of the second stress invariant on the tensile meridian to that on the compressive
meridian at initial yield, 0.5 < Kc <= 1.0, default is 2/3
42
– viscosity parameter representing the relaxation time of the viscoplastic system,
default is 0.0
The failure surface in the Concrete Damaged Plasticity model is governed by parameter
Kc. Kc is the ratio of the cracking stress in triaxial tension to the yield stress in triaxial
compression. ABAQUS manual specifies a default value of 2/3 for the parameter, Kc. A similar
value was reported by Kupler (1969) from his experimental data. Since conducting laboratory
experiments to determine the material parameters was beyond the scope of this project, the
default value of 2/3 is used in the analysis. The plastic potential surface in the meridional plane is
in a form of a parabola in the CDP model. The eccentricity parameter can be used to adjust the
shape of the plastic potential surface. A default value of 0.1 is defined for the concrete materials
in the current analysis. The dilation angle characterizes the performance of concrete under
compound stress (Kmiecik and Kamiński 2011). It is the angle of inclination of the failure
surface towards the hydrostatic axis, measured in the meridional plane. The unquestionable
advantage of the CDP model is the fact that it is based on parameters having an explicit physical
interpretation (Kmiecik and Kamiński 2011).
43
3.3.2.3 Deck concrete
Tensile behavior Compressive behavior
Cracking stress Cracking strain Peak stress (psi) Inelastic strain
300
(psi) 0 8500 0.0
225 0.00005 9100 0.0001
150 0.0001 Other Parameters
112.5 0.00013 56
75 0.00017 0.1
56.25 0.0002 bo/co 1.16
37.5 0.00025 Kc 0.6667
30 0.0003 0.25
26.25 0.0004
3.4 Assembly
The next step in modeling is to assemble the different parts to create the desired
geometry. The number of instance/s of the same part/s can be created at the assembly stage, such
as the number of strands or reinforcements in a structure. The instance/s of the part/s are
positioned relative to each other in a global coordinate system. The constraints available in
ABAQUS can be used to position the instances relative to each other. After the desired geometry
is created, the elements, geometry or nodes can be grouped together into sets, which can be
helpful for the application of material properties, loads, boundary conditions, and constraints,
etc. at a later stage. The assembly for the current project comprises different parts with multiple
instances of some of the parts. The number of instances of each of the parts is listed inError!
Reference source not found..
Table 3.1 Part and number of instances (Single span system)
Part Number of instances
Girder and deck 1
Strand 20
Transverse deck reinforcement 118
Longitudinal deck reinforcement 24
Longitudinal reinforcement at girder top 6
44
Figure 3.9 Girder and deck mesh in cross section
45
Top reinforcement in the girder
46
Figure 3.13 Wireframe of the complete model
47
Figure 3.15 Strands
48
3.5.1 Steps
The various steps defined in the current model are listed in Table 3.2.
Table 3.2 Various steps defined in the analysis
Step Name Type Time Maximum Increment size No. of
no. period no. of Initial Minimum Maximum increments
(days) increments
1 Initial
2 Prestress Static 1 400 0.1 1E-10 0.2 129,
release 137,129,
and 137
3 Girder creep Static 10, 30, 500 0.1 1E-10 1 136,195,
and shrinkage 60 and 208,and
90 260
4 Slab dead Static 1 100 0.1 1E-05 1 6
load on girder
5 Deck Static 1 100 0.1 1E-10 1 6
activated
6 Slab creep Static 187, 1000 0.1 1E-10 1 192, 206,
and shrinkage 177, 137 185, and
and 107 156
The number of increments corresponds to 10-days, 30-days, 60-days, and 90-days old
girder at the time of casting of the slab. The prestress load can be applied using various methods
in ABAQUS. It can be applied by using the predefined field in the Initial step of the analysis by
selecting the strand elements and providing the value of the stress. Additionally, stress value can
be converted to temperature and the temperature loading can be applied to the strands in the first
step to simulate initial stress in the strands.
In the current model, the first method is used to apply a prestressing force to the strands
using ABAQUS CAE. The input file code for the same is written below:
** PREDEFINED FIELDS
*Initial Conditions, type=STRESS
Strands_geometry, 202500., 0., 0.,
where, Strands_geometry is a set representing all the strand elements in the model and
the numerical value 202500 is the magnitude of the prestressing force applied in the axial
direction. A value of 202500 psi is calculated by using the following formula.
Prestress load = 0.75fpu = 0.75*270 = 202.5 ksi =202500 psi
𝑓𝑝𝑢 = yield strength of prestressing strands = 270 ksi
49
Adequate boundary conditions need to be provided to simulate the release. The strands
are connected to the girder using the embedded region constraint. A perfect bond is assumed
between the strands and concrete elements since they share the same nodal coordinates.
50
is easy to implement and it does a satisfactory job in predicting the creep and shrinkage strain in
the concrete.
51
𝑡
(𝜀𝑠ℎ )𝑡 = (𝜖 ) Eqn. (3.3)
35 + 𝑡 𝑠ℎ 𝑢
(𝜀𝑠ℎ )𝑡 = shrinkage strain at any time, t
𝑡 = time from the end of initial curing
(𝜀𝑠ℎ )𝑢 = ultimate shrinkage strain
52
Creep, 𝛾𝑠 = 0.82 + 0.067𝑠 Eqn. (3.10)
Shrinkage, 𝛾𝑠 = 0.89 + 0.041𝑠 Eqn. (3.11)
𝑠 = observed slump in inches
The correction factor for percent of aggregate is given by:
Creep, 𝛾Ψ = 0.88 + 0.0024𝜓 Eqn. (3.12)
3.5.3.1.1 Correction factor for creep and shrinkage coefficient for girder and deck
Table 3.3 and Table 3.4 give the values of different correction factors for creep and shrinkage,
respectively, for this project.
53
Table 3.3 Correction factors for the girder
Condition Creep factor Shrinkage factor
Loading age 𝑡𝑙𝑎 = 1 day 1.250 -
Relative humidity 𝜆 = 70% 0.801 0.686
Volume-surface ratio 𝑣/𝑠 = 4.004in 0.753 0.742
Slump 𝑠 = 2.5in 0.988 0.993
Percent of fine aggregate 𝜓 = 60% 1.024 1.02
Cement content 𝑐 = 752 lbs/cu yd - 1.021
Air content 𝛼 = 7% 1.090 1.006
Applicable correction factor (𝜸𝒄 )𝒈𝒊𝒓𝒅𝒆𝒓 = 0.831 (𝜸𝒔𝒉 )𝒈𝒊𝒓𝒅𝒆𝒓 = 0.529
The plot of creep coefficient vs. time and shrinkage strain vs. time is shown in Figure 3.16 and
Figure 3.17, respectively, for the girder and the deck slab.
54
Variation of creep coefficient with time
1.6
1.4
1.2
Girder
Creep coefficient
1 creep
coefficient
0.8
Deck
creep
0.6 coefficient
0.4
0.2
0
0 50 100 150 200
Time (days)
Figure 3.16 Creep coefficient vs. time for the girder and deck slab
0.00035
0.0003
Girder
Shrinkage strain
0.00025 shrinkage
strain
0.0002
Deck
shrinkage
0.00015 strain
0.0001
0.00005
0
0 50 100 150 200
Time (days)
Figure 3.17 Shrinkage strain vs. time for the girder and deck slab
55
3.5.3.2 User subroutine to calculate creep and shrinkage strain
The use of subroutines provides an additional advantage to simulate complicated
conditions in ABAQUS. All the subroutines for a particular analysis can be combined into a
single text file and saved with an extension .f or .for. ABAQUS does not a have built in ACI-209
(ACI Committee 209-Creep and Shrinkage. 2008) creep and shrinkage model to calculate creep
and shrinkage strain for concrete, therefore it is achieved by using a FORTRAN subroutine.
Creep and shrinkage strain is calculated in a simplified manner by means of UEXPAN
subroutine in ABAQUS. UEXPAN subroutine simulates the thermal expansion/contraction in a
material, but it can be manipulated to calculate any kind expansion/contraction in a material. The
subroutine is called for all integration points for each iteration of an increment in an element
whose material definition includes, user-specified thermal expansion, UEXPAN. The strain is
calculated as:
𝑡𝑛 0.60 𝑡𝑛−1 0.60
Δ𝜀𝑐𝑟 = 𝜐𝑢 𝜀𝑖 ( − ) Eqn. (3.20)
10 + 𝑡𝑛 0.60 10 + 𝑡𝑛−1 0.60
∆𝜀𝑐𝑟 = creep strain increment
𝜐𝑢 = ultimate creep coefficient
𝜀𝑖 = instantaneous elastic strain from end of previous increment
𝑡𝑛 = time since loading at current increment
𝑡𝑛−1 = time since loading at previous increment
The instantaneous elastic strain from the end of the previous increment is saved as a state
variable using USDFLD subroutine and passed on to the UEXPAN subroutine.
Increment in shrinkage strain at time 𝑡𝑛 is calculated as:
𝑡𝑛 𝑡𝑛−1
Δ𝜀𝑠ℎ = 𝜀𝑠ℎ𝑢 ( − ) Eqn. (3.21)
35 + 𝑡𝑛 35 + 𝑡𝑛−1
∆𝜀𝑠ℎ = shrinkage strain increment
𝜀𝑠ℎ𝑢 = ultimate shrinkage strain
𝑡𝑛 = time since curing in current increment
𝑡𝑛−1 = time since curing at previous increment
The total strain increment due to creep and shrinkage is calculated as:
Δ𝜀 = Δ𝜀𝑐𝑟 − Δ𝜀𝑠ℎ Eqn. (3.22)
56
The shrinkage strain is deducted from creep strain due to the sign convention in ABAQUS.
Tensile stresses and strains are positive in ABAQUS, hence shrinkage strain is considered
negative while the sign of creep strain increment is taken care of by the instantaneous elastic
strain allowing for both compressive and tensile creep.
57
3.8 Post processing in ABAQUS using the visualization module
ABAQUS has a very user friendly graphical user interface for post processing the results
from the analysis. The results can be viewed in the visualization module by opening the .odb
(Output Database) file. The .odb file is generated as the analysis progresses, but it is advisable to
open the file once the analysis is completed or terminated. Accessing the output database file
while the analysis is in progress will slow down the analysis and is not recommended. The
visualization module allows the viewing of the results as well as provides tools for diagnosing
the errors and warnings in a job. Some of the most commonly used post-processing features are:
1) Contours – The variables, such as stress, strain, and displacement, can be
represented as contour in the model for any increment in a step. This makes it
easier to understand the behavior of the model. The contours can be plotted on
both the deformed and the undeformed shapes.
2) Deformed shape – The nodal deformation pattern of the model can be examined
in detail using the deformed shape tool. The scale can be changed to view an
exaggerated deformation pattern of the model, which is helpful when the
deformation in the model is small.
3) X-Y data – The X-Y data can be read from the field output in the output database
obtained from analysis results. The field outputs requested during the analysis are
available and can be recorded for an element, nodes, and integration points.
Tabular data from the X-Y data can be created, saved and/or copied to Excel or
any other program for further analysis.
4) X-Y plots – The plots created from the X-Y data can be formatted using the X-Y
plot tool. The title of the plot, axis title, and legends can be labeled and their fonts
can be edited as required.
5) Time history animation –The behavior of the structure over time can be displayed
as a series of plots shown rapidly, emulating a movie. This feature can be useful
to present the result for the whole or part of the model in a presentation.
6) Job diagnostics – ABAQUS terminates the analyses when it encounters severe
instabilities or an error. The job diagnostic is a very helpful tool to investigate the
reason for the termination. The job history in the job diagnostic tab lists
summaries of all the iterations in each increment in a step. The summary of each
58
iteration can be investigated to locate any instabilities/errors in the model. The
diagnostic gives the nodes/elements numbers causing instability/error in an
iteration. All the warnings and errors at different iterations can be found here and
the model can be fine-tuned as required by investigating these warnings/errors.
59
Chapter 4. Analysis approach
60
4.1.1 Newton-Raphson Method in ABAQUS (Hibbitt et al. 2011a)
In non-linear analysis, the total time period in each step is divided into small increments
so that the nonlinear path can be followed. The idea is to break the total load into small values.
Since the stiffness of the structure changes with deformation, it is not possible to calculate the
solution by applying the total load in one attempt. An initial increment size is provided by the
user and the subsequent increment is automatically chosen by ABAQUS. The initial increment
size should be small so that a stable solution can be achieved in the first increment. An initial
increment size of 0.01 or 0.001is acceptable in most of the cases but can vary as per the demand
of the problem. The structure is in approximate equilibrium after the end of each successful
increment during the analysis. For each increment, iterations are performed until convergence is
obtained. If ABAQUS fails to converge during an iteration, it starts a new iteration with a
smaller increment size. The smallest increment size possible in an analysis can be controlled by
the user. It is important to use the minimum increment size parameters wisely so that the analysis
is completed within an acceptable amount of time.
The concept of a body to be in equilibrium requires the net force acting at every node to
be zero or very close to zero. The internal forces, I, and the external forces, P, must balance each
other (Hibbitt et al. 2011a).
P
P
Ib
Id
P = External load
Ib and Id = Internal forces
Figure 4.1 Internal and external loads on a body (Hibbitt et al. 2011a)
P–I=0
61
There are various numerical methods to solve problems by an iterative procedure. The
Static, General solver in ABAQUS uses the Newton-Raphson numerical analysis method, which
is a robust iterative method to solve a nonlinear problem. In this technique, each iteration
involves the formulation and solution of linearized equilibrium equations. Each iteration consists
of defining the terms in the equilibrium equations (forming the stiffness matrix) and solving the
resulting system.
Ktangent Cu = P – I Eqn. (4.1)
Where Ktangent = tangent stiffness matrix
Cu = correction to displacement, u
If the solution in an iteration falls within a defined level of tolerance, then it is accepted
as a sufficiently accurate solution.
P - I = R(u) Eqn. (4.2)
R(u) is the residual or the out of balance force at u for the iteration and it is nonlinear. If
the structure is not in equilibrium at the displaced position u then,
𝑅(𝑢) ≠ 0
The displacement correction factor, cu is found so that u +cu is in equilibrium,
𝑅(𝑢 + 𝑐𝑢 ) = 0 Eqn. (4.3)
On expanding the R(u) in a Taylor series about the current displacement u, we get
𝜕𝑅
𝑅(𝑢 + 𝑐𝑢 ) = 𝑅(𝑢) + 𝜕𝑢 |𝑢 𝑐𝑢 + ⋯ = 0 Eqn. (4.4)
62
forces acting on each node should be very small and the displacement correction should be very
small as well. The tolerance for the above values is set to a small value by default in ABAQUS
and can be edited by the user, if needed.
P1
R(u1) R(u2)
R(u0)
K(u2) R(u3)
K(u1)
K(u0)
P0
u
u0 u1 u2 u3 u4
Du1 Du2 Du3 Du4
P = load, u = displacement
Figure 4.2 Newton-Raphson approach to solve a nonlinear problem (ABAQUS convergence
class, Dassault Systemes)
In the above figure, it can be seen that iterations are performed until a point where the residual
R(u3) is small enough to approximate the solution as converged. When the residual is less than
the defined force residual tolerance at all the nodes, ABAQUS Standard accepts the system to be
in equilibrium. The default tolerance value in ABAQUS is 0.5% of the average force in the
63
structure, averaged over time. This tolerance value depends on the accuracy of the results
intended for the problem and can be changed by the user. The solution procedure followed by
ABAQUS for a problem without contact is shown in Figure 4.3. When the solution for a given
increment cannot be found by ABAQUS, a new attempt is made by reducing the magnitude of
the load increment (by default to 25% of the previous increment size). If too many attempts are
made to reduce an increment (by default 5 attempts), ABAQUS stops the analysis with the
following error: Too many attempts made in an increment. When two consecutive increments
converge within 5 iterations, ABAQUS automatically increases the size of the increment by 50%
(default). ABAQUS controls the automatic incrementation and helps reduce the wastage of the
CPU time. In case, the solution does not converge within five (default) iterations, the default
value can be increased to a higher value. Increasing the maximum number of cutbacks allowed
for an increment can be useful if the problem is highly nonlinear. However, it is important to
monitor the analysis intermittently to make sure that the increment size is not so small that there
is no significant progress in the analysis for a long period of time. This results in consumption of
more CPU resources.
64
Begin
Solve for Cu
Update u
Compute
residuals
Yes
Yes
No No
End of Yes
End of step Output results
analysis
Yes
End
Figure 4.3 Flowchart showing solution procedure for analysis without contact (ABAQUS
convergence class, Dassault Systemes)
All the default values for tolerances can be changed by the user in the step module under
the Other → General solution controls → Manager/Edit and choose the step/s for which the
default values need to be changed. The screenshot of the menu instruction is shown in Figure
4.4.
65
Figure 4.4 Screenshot for changing default solution control
Some of the default parameters and their values are shown in Table 4.1.
Table 4.1 Default tolerance values for some parameters
Parameters Function Default
value
𝑅𝑛𝑎 Convergence criterion for the ratio of the largest residual to the 5 x 10-3
corresponding average flux norm for convergence.
𝐶𝑛𝑎 Convergence criterion for the ratio of the largest solution correction to 10-2
the largest corresponding incremental solution value.
𝐼0 Number of equilibrium iterations (without severe discontinuities) after 4
which the check is made whether the residuals are increasing in two
consecutive iterations.
𝐼𝑅 Number of consecutive equilibrium iterations (without severe 8
discontinuities) at which logarithmic rate of convergence check begins.
𝐼𝐴 Maximum number of cutbacks allowed for an increment or number of 5
attempts to be made to reduce the time increment.
𝐼𝐶 Upper limit on the number of consecutive equilibrium iterations 16
(without severe discontinuities), based on prediction of the logarithmic
rate of convergence.
66
softening. Since both of the above nonlinear features add to the complexity of the problem, there
were convergence issues during the analysis. The most common issue was the convergence of
the solution when the stress in the concrete reaches the peak tensile stress. The analysis would
run up to a point where the material behaved as linearly elastic while being in tension. The
increment where tensile stress in the concrete reaches the maximum value, the solution of the
analysis failed to converge and the analysis terminated. This was due to the fact that the stress-
strain data has a negative slope in the tension softening area. This causes the formation of
negative tangent stiffness matrix, which is undesirable in a finite element analysis.
The notes, warnings, and errors in the message (.msg), data (.dat), and status (.sta) files
created after an analysis can be useful in interpreting the issues in the model. Further details can
be obtained from the job diagnostic tool in the visualization module. Even for a successful
analysis, the above files can provide more information about the step by step increments, which
can be helpful to verify the accuracy of the results and refining the model. Most of the time,
warnings in the message file provide a good insight about the issues in the model. Some typical
notes, warnings, and errors encountered in the current project with their interpretation are
discussed here. Most of them can be termed as classical convergence errors in a nonlinear
analysis.
67
convergence. So before increasing the number of unsuccessful iterations for the analysis,
message and data files should be studied for any other warnings and notes.
68
4.3 Solution to the convergence issues
Many attempts were made to solve the convergence issues in the analysis. Different
material properties were used, other solvers available in ABAQUS were tried, boundary
conditions were modified, stabilization techniques in ABAQUS were used, etc. None of the
above attempts at solving the convergence issues were successful. The automatic stabilization
feature is available in ABAQUS to solve convergence issues in an analysis. The automatic
stabilization provides a mechanism for stabilizing unstable quasi-static problems through the
automatic addition of volume-proportional damping to the model. The applied damping factors
can be constant over the duration of a step, or they can vary to account for changes over the
course of a step. The latter, adaptive approach is typically preferred (Hibbitt et al. 2011a). The
automatic stabilization technique was used for the current analysis, but it failed to address the
convergence issue.
Finally, the convergence issue was dealt with by introducing the viscosity regularization
in the concrete material definition. The concrete damaged plasticity model is equipped to include
a viscosity parameter which helps to overcome the convergence problem due to stiffness
degradation and softening behavior in a material. Viscosity regularization helps to make the
consistent tangent stiffness of the softening material to become positive for sufficiently small
time increments (Hibbitt et al. 2011a). It stabilizes the material model by allowing stresses
outside the yield surface. ABAQUS uses a generalization of the Duvaut-Lions regularization,
according to which the viscoplastic strain rate tensor, 𝜀̇𝜐𝑝𝑙 , which is defined as
1 𝑝𝑙
𝜀̇𝜐𝑝𝑙 = (𝜀 − 𝜀𝜐𝑝𝑙 ) Eqn. (4.5)
𝜇
𝜇 = viscosity parameter representing the relaxation time of the viscoplastic system
𝜀 𝑝𝑙 = plastic strain evaluated in the inviscid backbone model
A small value of viscosity parameter compared to the typical time increment helps improve the
rate of convergence of the model in the softening regime, without the compromising results
(Hibbitt et al. 2011a). Values of 0.009 and 0.25 are used for the girder concrete material and the
deck concrete material, respectively, for the current analysis. The above values were reached by
creating various analysis models with different values of viscosity parameter. A post tensile yield
stress vs. cracking strain plot was made for each of the analysis results. The value used in the
analysis which gave the closest post tensile yield stress vs. cracking strain curve to the one
69
initially provided was chosen for the final analysis. The post tensile yield stress vs cracking
strain plot for different analysis results is shown in Figure 4.5.
700 40 - 0.05
40 - 0.01
600
Tensile stress (psi)
40 - 0.02
500 40 - 0.03
40 - 0.02
400 40 - 0.005
56 - 0.001
300
Viscosity parameter = 0.009 40 - 0.015
200 40 - 0.017
56 - 0.005
100
56 - 0.009
0 Input curve
0 0.0001 0.0002 0.0003 0.0004
Cracking strain
50 (Dilation angle) – 0.01 (Viscosity parameter)
Figure 4.5 Calibration of Viscosity regularization parameter for the girder concrete
70
Chapter 5. Results and Discussion
The results obtained from the analysis are presented in this section. Stress, strain, and
displacements in the slab and the girder are plotted. Various models were created for the
different age of each girder at the time of the casting of the slab. The different girder ages
considered are 10, 30, 60, and 90 days. The analysis was run for a total time period of 200 for all
the models, simulating the behavior of the bridge for the first 200 days from the day of casting
the girder in the casting yard. The effects of the different age of the girder at the time of the
casting of the deck slab on the system are compared and presented here. It is observed that the
age of the girder at the time of the casting of the slab affects the overall system in a significant
manner. The behavior of the girder without creep and shrinkage in the slab was also simulated
and compared with the system with creep and shrinkage in the slab. The effect of creep and
shrinkage on the stress in the prestressing steel is also presented here. The Figure 5.1 shows the
stress in the girder after the transfer of the prestressing force to the girder (End of step 1). It can
be seen that the bottom of the girder experiences compressive stress (negative stress value in
ABAQUS) while the top of the girder is in tension (positive stress value in ABAQUS) after step
1.
Figure 5.1 Stress distribution in the girder after release of prestress (camber exaggerated).
71
Figure 5.2 Zoomed view of stress distribution near the midspan
Figure 5.3 Stress in the girder and the slab at the end of 200 days for 90 -day old girder
72
Figure 5.4 throughFigure 5.7 shows the stress distribution at the end section and the
midspan section for the girder age of 90 days at the time of the casting of the slab. The stresses
are shown for all the major steps in the analysis.
Before release at end After release (End of Step 1) 90 days of creep and shrinkage
in the girder (End of Step 2)
Figure 5.4 Stress distribution at the end section up to 90 days of creep and shrinkage in the
girder
73
After addition of slab load (End of Step 3)
After creep and shrinkage in slab (End of Step 3, Total time = 200 days)
Figure 5.5 Stress distribution at the end section with slab dead load and creep & shrinkage
in slab (Girder age is 90 days at the time of casting of the slab)
74
Before release After release (End of Step 1) 90 days of creep and
shrinkage in the girder (End of
Step 2)
Figure 5.6 Stress distribution at the midspan section up to 90 days of creep and shrinkage
in the girder
75
After addition of dead load of slab (End of Step 3)
After creep and shrinkage in slab (End of Step 4, Total time = 200 days)
Figure 5.7 Stress distribution at the midspan section with slab dead load and creep &
shrinkage in the slab (Girder age is 90 days at the time of the casting of the slab)
76
The girder cambers up (positive camber) after the first step of analysis because of the
compressive stresses at the bottom flange of the girder due to the transfer of the prestressed force
from the strands to the concrete. A gradual increase in positive camber is observed in the girder
for the first few days after the release of the prestressing force. The positive camber in the girder
begins to slow down eventually due to the decrease in the creep effect. After the slab is cast, the
girder cambers down due to the dead weight of the slab. Once the slab is set, it begins to
influence the girder. Then a gradual cambering down of the girder is observed due to the
shrinkage of the deck slab concrete. This happens due to differential shrinkage between the
girder and the deck slab which is prominent for older girders such as the 60-day and 90-day
girder. Shrinkage in the slab is restrained by the reinforcement in the slab and the interaction
between the slab bottom and the girder top flange. Due to the restraint shrinkage, tensile stresses
start to develop at the bottom of the slab and compressive stress is experienced by the top flange
of the girder. The value of the tensile stress at the bottom of the slab depends upon the
differential shrinkage between the girder and the slab. When the value of tensile stress in the slab
reaches the tensile elastic limit, the slab cracks. The cracking in the slab is simulated by the
tension softening behavior of the concrete. As the portion of the slab cracks, the stresses in those
areas are released. The cambering down of the girder ceases after the cracking in the slab and the
girder starts to camber up again as seen in Figure 5.8. This behavior is observed in the girder,
which are 60 days and 90 days old at the time of casting of the slab. This is because of the
significant differential shrinkage between the slab and the girder. The cracking in the slab is
observed at approximately 60 days and 42 days after the casting of the slab in 60-day and 90-day
old girder, respectively. From the above observation, we can say that a slab cast on a relatively
old girder tends to experience cracking sooner as compared to slabs cast on relatively young
girders. This can be based on the fact that differential shrinkage between the girder and the deck
will be more for relatively old girders and hence causing higher tensile stresses in the deck slab.
The above fact is also supported by the observations made from the model with girder
age of 10 days and 30 days. From Figure 5.9, we can see that the stresses in the deck slab for 10
days and 30 days old girder are well within the tensile stress limit of 300 psi. This can be
attributed to smaller differential shrinkage between the slab and the girder. The smaller
differential shrinkage also results in the girder and the slab shrinking together causing minimal
displacement to the top of the flange of the girder at the midspan. It is observed that the
77
differential shrinkage causes downward camber at the girder midspan. The magnitude of the
deformation of the top flange of the girder at the midspan was found to be relatively higher for
older girders few days after the activation of the slab, supporting the fact that higher differential
shrinkage causes more deformation.
The total stress in the top flange of the girder at the midspan is shown in Figure 5.10. The
top flange of the girders at the midspan show compressive stresses after the application of dead
load of slab in the model. An increase in compressive stress is observed in all the girders due to
the shrinkage the deck. Higher compressive stress is experienced by the older girder after the
activation of the deck slab in the model. However, these compressive stresses are eventually
relieved in older girders because of the cracking in the deck slab, as seen for the girder age of 60
and 90 days. The compressive stress in the top flange of the girder with an age of 10 and 30 days
begins to level out after a few days of creep and shrinkage in the girder and the slab together. It
can be seen that the final stress state in all the girders is compressive at the end of 200 days of
simulation which is desirable as far as the girder concrete is concerned.
1.6
1.4
Displacement (in.)
1.2
90 days
1
60 days
0.8
30 days
0.6 10 days
0.4
0.2
0
0 50 100 150 200
Time (days)
Figure 5.8 Displacement at girder top flange at midspan due to creep and shrinkage
78
Total stress at bottom of slab at midspan
300
250
200
90 days
Stress (psi)
60 days
150
30 days
10 days
100
50
0
0 50 100 150 200
Time (days)
Figure 5.9 Stress at bottom of slab at midspan for different age of girders
600
400 90 days
200 60 days
30 days
0
Stress (psi)
-400
-600
-800
-1000
-1200
Time (days)
Figure 5.10 Stress at top flange of girder for different age of girders
79
A condition without the effect of creep and shrinkage in the deck slab was also simulated
for all ages of the girders and the results are compared to the model with creep and shrinkage
effect in the deck slab. The dead load of the slab was added in both conditions. Figure 5.11
through Figure 5.14 shows the comparison of stresses in the top flange of the girder at midspan
for both the above conditions while Figure 5.15 throughFigure 5.18 depicts the total
displacement in the top flange of the girder at midspan. For all ages of girder, it was observed
that the stress in the top flange of the girder at the midspan does not vary much after the
instantaneous drop in stress due to the slab dead load for the condition without creep and
shrinkage effect in the slab. Since there are no long-term effects occurring in the deck slab, it
does not influence the behavior of the girder. The displacement plots for the top flange of the
girder at midspan without the creep and shrinkage effects in the deck slab shows an increase in
positive camber for all the girders after the instantaneous decrease in displacement due to slab
dead load. This increase in displacement can be attributed to the domination of creep due to
prestress in the girder concrete.
600
Creep and
400 shrinkage in
slab
Stress (psi)
No creep
200
and
shrinkage in
slab
0
0 50 100 150 200
-200
-400
Time (days)
Figure 5.11 Stress in the top flange of girder due to variation in creep and shrinkage in the
slab (for 10-day old girder at the time of casting of the slab)
80
Total stress comparison for 30-day old girder at top flange at
midspan
800
600
Creep and
400 shrinkage in
slab
200
No creep
Stress (psi)
0 and
0 50 100 150 200 shrinkage in
-200 slab
-400
-600
-800
-1000
Time (days)
Figure 5.12 Stress in the top flange of girder due to variation in creep and shrinkage in the
slab (for a 30-day old girder at the time of casting of the slab)
200
No creep and
Stress (psi)
0
0 50 100 150 200 shrinkage in
-200 slab
-400
-600
-800
-1000
-1200
Time (days)
Figure 5.13 Stress in the top flange of girder due to variation in creep and shrinkage in the
slab (for a 60-day old girder at the time of casting of the slab)
81
Total stress comparison for 90-day old girder at top flange at
midspan
800
200
No creep and
Stress (psi)
0
shrinkage in
0 50 100 150 200 slab
-200
-400
-600
-800
-1000
-1200
Time (days)
Figure 5.14 Stress in the top flange of girder due to variation in creep and shrinkage in the
slab (for a 90-day old girder at the time of casting of the slab)
1 No creep
and
shrinkage
0.8
in slab
0.6
0.4
0.2
0
0 50 100 150 200
Time (days)
Figure 5.15 Total displacement in the top flange of girder at mid-span due to variation in
creep and shrinkage in the slab (for a 10-day old girder at the time of casting of the slab)
82
Total displacement comparison for 30-day old girder at top
flange at midspan
1.6
Creep and
1.4 shrinkage in
slab
1.2
Displacement (in.)
No creep
1
and
shrinkage in
0.8
slab
0.6
0.4
0.2
0
0 50 100 150 200
Time (days)
Figure 5.16 Total displacement in the top flange of girder at mid-span due to variation in
creep and shrinkage in the slab (for a 30-day old girder at the time of casting of the slab)
1.2 No creep
and
1 shrinkage
in slab
0.8
0.6
0.4
0.2
0
0 50 100 150 200
Time (days)
Figure 5.17 Total displacement in the top flange of girder at mid-span due to variation in
creep and shrinkage in the slab (for a 60-day old girder at the time of casting of the slab)
83
Total displacement comparison for 90-day old girder at top
flange at midspan
1.8
Creep
1.6 and
shrinkage
1.4 in slab
Displacement (in.)
1.2 No creep
and
1 shrinkage
in slab
0.8
0.6
0.4
0.2
0
0 50 100 150 200
Time (days)
Figure 5.18 Total displacement in the top flange of girder at mid-span due to variation in
creep and shrinkage in the slab (for a 90-day old girder at the time of casting of the slab)
The ideal condition for a concrete bridge is no cracking in the deck slab throughout its
service life. Cracking is not only unsightly, but cracks allow water and other chemicals to seep
into the slab leading to undesirable effects such as corrosion, freeze thaw damage, leaching or
alkali silica reaction. The uncracked slab will also mitigate the formation of positive moment at
the diaphragm in a continuous system.
A single span bridge model was created where cracking in the slab was restricted by
increasing the tensile stress limit of the slab concrete to a higher value in order to see the effects
of a large differential shrinkage that was not mitigated by slab cracking. This simulation was
conducted only for girder age of 90 days. The comparison of the results from the above model
with the one where the slab does not crack is presented in Figure 5.19 throughFigure 5.21. It can
be seen that the tensile stress keeps increasing at the bottom of the slab while the top of the
girder experiences continuous compression. Continuous decrease in positive camber is observed
after the application of slab dead load in Figure 5.21.
In a continuous system, this would be advantageous because it would cause the end of the
girder to rotate into the diaphragm and increase compressive stresses at the bottom of the
84
diaphragm, prevent formation of positive moment at the bottom of the support and reduce or
eliminate tensile cracking at the girder-diaphragm interface. However, in a real system the slab
cracks and the beneficial effect of the differential shrinkage in the continuous system is lost.
This shows a major shortcoming of analysis methods for continuous bridges which do not
account for slab cracking.
Uncracked
300
slab
250
Stress (psi)
200
150
100
50
0
80 100 120 140 160 180 200
Time (days)
Figure 5.19 Comparison of stresses for cracked and uncracked slab condition at bottom of
slab at midspan (for a 90-day old girder at the time of casting of the slab)
85
Total stress comparison at top of girder at midspan for
90-day old girder
700
Cracking
400 in slab
100 Uncracked
slab
0 50 100 150 200
Stress (psi)
-200
-500
-800
-1100
-1400
Time (days)
Figure 5.20 Comparison of stresses for cracked and uncracked slab condition at top of
girder at midspan (for a 90-day old girder at the time of casting of the slab)
1.4 Uncracked
slab
Displacement (in.)
1.2
0.8
0.6
0.4
0.2
0
0 50 100 150 200
Time (days)
Figure 5.21 Comparison of displacement for cracked and uncracked slab condition at top
of girder at midspan (for a 90-day old girder at the time of casting of the slab)
86
5.1 Effect of girder age on prestressing force in the strands
As soon as the prestressing force is applied to the girder, the strands begin to lose stress.
There is an immediate loss due to elastic shortening of the girder and long term losses due to
creep and shrinkage of the concrete and relaxation of the strand. Stress gains are also possible
due to superimposed dead loads and the effects of differential creep and shrinkage. The gains
are controversial. One method for assessing loss of prestressing force in the AASHTO LRFD
Bridge Design Specifications is the “refined” method. One aspect of this method is the
calculation of a gain in prestressing force due to differential shrinkage of the deck. However, not
all states accept this. Some states do not allow inclusion of this gain while others limit it.
The stresses in the prestressing strands are shown in Figure 5.22 for different ages of the
girders at the time the slab is added. It can be seen that the prestress loss is higher in the first few
days for all girders. A gain in prestress is observed due to the dead load of the slab for each of
the girder. In addition to the above gain, a minor gain in prestress can be seen for a girder aged
60 or 90 days when the slab is added in the analysis. This gain is due to differential shrkinage of
the slab and was eventually lost because of the cracking in the deck slab. The prestress gain is
not significant when compared to the prestress losses in the model.
The results from the current analysis also show that the prestressing force in the strands is not
significantly affected by the creep and shrinkage of the deck slab. The prestressing force in the
strands for the model with creep and shrinkage in the slab and the model without creep and
shrinkage in the slab are found to be varying by less than 1% for all ages of girder after 200 days
of simulation. This can be seen in Figure 5.23 throughFigure 5.26. Cracking of the slab also does
not have any significant effect on the stress in the steel. This was observed for all girder ages of
10, 30, 60, and 90 days.
87
Stress in strands at the midspan for different ages of the girder
210000
10 days old
girder
200000 30 days old
girder
60 days old
190000 girder
Stress (psi)
90 days old
girder
180000
170000
160000
150000
0 50 100 150 200
Time (days)
Figure 5.22 Stress in the prestressing strands for different ages of the girder
200000
Creep and
190000 shrinkage in
slab
Stress (psi)
160000
150000
0 50 100 150 200
Time (days)
Figure 5.23 Comparison of stress in prestressing strands for a 10-day old girder
88
Stress in strands at the midpsan for 30-day old girder
210000
200000
Creep and
190000 shrinkage in
slab
Stress (psi)
160000
150000
0 50 100 150 200
Time (days)
Figure 5.24 Comparison of stress in prestressing strands for a 30-day old girder
200000
Creep and
190000 shrinkage in
slab
Stress (psi)
160000
150000
0 50 100 150 200
Time (days)
Figure 5.25 Comparison of stress in prestressing strands for a 60-day old girder
89
Stress in strands at the midspan for 90-day old girder
210000
200000
Creep and
190000 shrinkage in
slab
Stress (psi)
160000
150000
0 50 100 150 200
Time (days)
Figure 5.26 Comparison of stress in prestressing strands for a 90-day old girder
A hand calculation was performed to calculate the prestress losses in the system by the
AASHTO refined loss method. The transformed section properties were used to for the
calculation. The solution obtained from the hand calculation was compared to the result of the
simulation performed using the AASHTO creep and shrinkage model (American Association of
State Highway and Transportation Officials., 2012). The above analysis was only done for a
girder age of 90 days at the time the slab was cast and the total simulation time was kept at 700
days.
It was observed that the AASHTO refined loss method gives a conservative value of the
prestress losses in the system. The difference between the stress values in the strands after 700
days from the above methods was 5.43 ksi (3.32%) as shown in Figure 5.27. This can attributed
to the fact that the AASHTO refined loss method is a single step procedure in which the
continuous variation of stress in the system with time is not taken into account.
90
Stress in the prestressing strands
210000
Abaqus
analysis
200000
Hand
190000 calculation
Stress (psi)
180000
170000
160000
150000
0 100 200 300 400 500 600 700
Time (days)
Figure 5.27 Comparison of stress by AASHTO refined loss method and ABAQUS analysis
91
Chapter 6. Conclusions and Future Research
The key to creating a successful finite element model is to start with a simple model and
introduce complications such as loadings, boundary conditions, and nonlinearities one by one
and study the output after every new step is introduced in the model. This simplifies the process
of the identification of error and helps to efficiently debug the model. Too many steps at once
make the model and analysis complex and sometimes makes it difficult to find out the root cause
of the error. Studying the results after the introduction of every new parameter helps in
understanding the working of the model in a better way.
The current finite element model created in ABAQUS Standard 6.11-2 successfully
simulated the behavior of single span precast prestressed concrete I-girder bridge subjected to
long-term loading, such as creep and shrinkage. The results indicate that ABAQUS Standard
6.11-2 does a good job in simulating the behavior of the prestressed concrete, both in
compression and tension. The additional feature of the linking of the subroutine with the CAE
model can be used to write a code for various creep and shrinkage models, such as the ACI 209
model (ACI Committee 209-Creep and Shrinkage. 2008), the AASHTO LRFD model (American
Association of State Highway and Transportation Officials., 2012), and the CEB-FIP model
(Comité euro-international du béton., 1993).
From the analysis, it can be concluded that the cracking of the slab depends on the
amount of differential shrinkage between the girder and the deck, which depends on the age of
the girder at the time of the casting of the slab. For the parameters chosen in this simulation, a
girder older than 60 days causes the slab to crack within the first two months after casting of the
slab, whereas, girders age 10 days or 30 days at the time the slab is case do not seem to
experience cracking of the slab even after 6 months.
The current project shows that the age of the girder plays an important role in the long-
term behavior of bridges constructed using precast prestressed concrete girders. The older girder
will result in more prestress gain, but eventually the slab will crack due to higher differential
shrinkage. This causes release in stresses and hence the gain in prestress is lost. The cracking in
the slab may also result in the development of the positive moment at the bottom of the support
in a continuous girder. On the other hand, if the slab is cast when the girder is relatively young,
cracking in the slab is not observed. The girder does not camber up significantly which might be
92
desirable to avoid the development of the positive moment at the bottom of the support in a
continuous system. The absence of cracking in the deck slab minimizes the seepage of water in
the slab and helps in reducing the corrosion of reinforcement. Hence, a balance has to be found
with regard to the age of the girder at the time of the pouring of the deck slab. An appropriate
estimation of the desired age of the girder at the time of the casting of the slab from the finite
element model can increase the serviceability and durability of the bridge. From the current
results, that age may be in between 30 and 60 days.
However, it must be noted that the analysis is heavily dependent on the creep coefficients
of the girder and slab, the shrinkage potential of the girder and slab and the tensile strength of the
slab. Thus, re-running the analysis with different parameters would like result in very different
girder ages when the slab could be cast without cracking. The ages cited in this work are to
simply illustrate results from a single case and should not be extrapolated to other cases.
It can be concluded that the AASHTO refined loss calculation gives a conservative estimate
of the prestress losses in the structure. It can also be concluded that the gain in prestressing force
calculated by that method is small and may not occur if the deck cracks. Thus, the need for
including this calculation is questionable.
93
bridge monitoring data for a multiple span bridge. A complete bond between concrete and
reinforcement was assumed in the current research. More realistic bonding characteristics
between the concrete and the strands might be of interest in the future.
94
References or Bibliography
1. ACI Committee 209-Creep and Shrinkage. (2008). "Guide for modeling and calculating
shrinkage and creep in hardened concrete." Rep. No. 209.2R, American Concrete
Institution, Farmington Hills, MI.
2. American Association of State Highway and Transportation Officials.,. (2012).
"AASHTO LRFD bridge design specifications." American Association of State Highway
and Transportation Officials, Washington, DC.
3. American Association of State Highway and Transportation Officials.,. (2002).
"AASHTO LRFD bridge design specifications." American Association of State Highway
and Transportation Officials, Washington, DC.
4. ANSYS® Academic Research, Release 10.0. "ANSYS® Academic Research, Release
10.0." .
5. Comité euro-international du béton.,. (1993). CEB-FIP model code 1990 : design code.
T. Telford, London.
6. Curtis, R. H. (2007). "NYSDOT Bridge Deck Task Force Evaluation of Bridge Deck
Cracking on NYSDOT Bridges." .
7. Debbarma, S. R., and Saha, S. (2011). "Behavior of pre-stressed concrete bridge girders
due to time dependent and temperature effects." First Middle East Conference on Smart
Monitoring, Assessment and Rehabilitation of Civil Strucutres, .
8. Dinges, T. (2009). "The history of prestressed concrete: 1888 to 1963". Master of
Science. Kansas State University, Manhattan, Kansas.
9. Freyermuth, C. L. (1969). "Design of continuous highway bridges with precast,
prestressed concrete girders." Journal of the Prestressed Concrete Institute, also Printed
as PCA Engineering Bulletin EB014.01E, 14(2), 14-29.
10. Hibbitt, H. D., Karlson, B. I., and Sorenson, E. P. (2011a). ABAQUS 6.11 Analysis User's
Manual. Hibbitt, Karlson & Sorenson, Providence, RI.
11. Hibbitt, H. D., Karlson, B. I., and Sorenson, E. P. (2011b). "ABAQUS Version 6.11,
Finite Element Program." Hibbitt, Karlson & Sorenson, Providence, RI, 6.11.
95
12. Kaar, P. H., Kriz, L. B., and Hognestad, E. (1960). "Precast-prestressed concrete bridges:
1. Pilot tests of continuous girders." Journal of the PCA Research and Development
Laboratories, 2(2), 21-37.
13. Kmiecik, P., and Kamiński, M. (2011). "Modelling of reinforced concrete structures and
composite structures with concrete strength degradation taken into consideration."
Archives of Civil and Mechanical Engineering, 11(3), 623-636.
14. Mattock, A. H. (1961). "Precast-prestressed concrete bridges 5: Creep and shrinkage
studies." Journal of the PCA Research and Development Laboratories, 3(2), 32-66.
15. Miller, R. A., Castrodale, R., Mirmiran, A., and Hastak, M. (2004). "NCHRP Report 519
Connection of Simple-Span Precast Concrete Girders for Continuity." Rep. No. 519,
Transportation Research Board, Washington, DC.
16. National Bridge Inventory Study Foundation,. (2003). "NBI Report 2003." .
17. Oesterle, R., Glikin, J., and Larson, S. (1989). "NCHRP Report 322: Design of Precast
Prestressed Bridge Girders Made Continuous." Transportation Research Board,
Washington, DC.
18. Okeil, A. M., Hossain, T., and Cai, C. (2013). "Field monitoring of positive moment
continuity detail in a skewed prestressed concrete bulb-tee girder bridge." PCI J., .
19. Stephen, E. (2006). "Simulation of the long-term behavior of precast/prestressed concrete
bridges". Master of Science. University of Cincinnati, Cincinnati, Ohio.
20. Tadros, Maher K., National Cooperative Highway Research Program., American
Association of State Highway and Transportation Officials., United States., Federal
Highway Administration., National Research Council (U.S.).,Transportation Research
Board.,. (2003). Prestress losses in pretensioned high-strength concrete bridge girders.
Transportation Research Board, Washington, DC.
96
Appendix A. Validation model for creep and shrinkage subroutine
The validation for the creep and shrinkage subroutine in ABAQUS is provided in this
appendix. A solved example from book titled, Reinforced concrete structures, by Omar Chaallal
and Mohamed Lachemi, chapter-1, Example 1.4, is taken for reference.
97
A-2 ABAQUS model and results
98
Calculated instantaneous stress in concrete = 9.91219 MPa
Calculated instantaneous strain in concrete = 473.481 x 10-6 mm/mm
Strain due to creep = 567.005 x 10-6 mm/mm
Strain due to shrinkage = 348.224 x 10-6 mm/mm
Table A.1 Comparison of the ABAQUS output with the calculated result
ABAQUS value Calculated value Percent Error
Creep strain 567.005 x 10-6 mm/mm 533 x 10-6 mm/mm 6.98
Shrinkage strain 348.224 x 10-6 mm/mm 348 x 10-6 mm/mm 0.06
99
Appendix B. Volume to surface ratio for the girder and the deck
4.5 7.0
4.5
19.0
7.0 45.0
7.5
7.5
7.0
22.0
Figure B.1 Cross-section of the AASHTO Type III, I girder
100
8.0 96.0
101
Appendix C. FORTRAN subroutine
SUBROUTINE USDFLD(FIELD,STATEV,PNEWDT,DIRECT,T,CELENT,
1 TIME,DTIME,CMNAME,ORNAME,NFIELD,NSTATV,NOEL,NPT,LAYER,
2 KSPT,KSTEP,KINC,NDI,NSHR,COORD,JMAC,JMATYP,MATLAYO,
3 LACCFLA)
C
INCLUDE 'ABA_PARAM.INC'
C
CHARACTER*80 CMNAME,ORNAME
CHARACTER*3 FLGRAY(15)
DIMENSION FIELD(NFIELD), STATEV(NSTATV), DIRECT(3,3), T(3,3),
TIME(2)
DIMENSION ARRAY(700), JARRAY(700), JMAC(*), JMATYP(*), COORD(*)
C
CALL GETVRM('THE',ARRAY,JARRAY,FLGRAY,JRCD,JMAC,JMATYP,MATLAYO,
LACCFLA)
THE = ARRAY(3)
STATEV(1)=THE
C
C Reading instantaneous elastic strain in direction 33(axial)
CALL GETVRM('EE',ARRAY,JARRAY,FLGRAY,JRCD,JMAC,JMATYP,MATLAYO,
LACCFLA)
EE = ARRAY(3)
STATEV(3)=EE
C
RETURN
END
C
SUBROUTINE UEXPAN(EXPAN,DEXPANDT,TEMP,TIME,DTIME,PREDEF,DPRED,
STATEV,CMNAME,NSTATV,NOEL)
C
INCLUDE 'ABA_PARAM.INC'
C
CHARACTER*80 CMNAME
C
DIMENSION EXPAN(*),DEXPANDT(*),TEMP(2),TIME(2),PREDEF(*),
DPRED(*),STATEV(NSTATV),ARRAY(15)
THE=STATEV(1)
EE=STATEV(3)
C
C Calculation of creep and shrinkage strain in girder concrete
C
IF ((CMNAME.EQ.'CONCRETE_GIRDER').AND.(TIME(2).GT.1)) THEN
T=TIME(2)-1
TL=T-DTIME
IF(TL.LT.0) THEN
102
TL=0
END IF
CREEPSTRAIN=1.800*EE*(T**0.6/(10+T**0.6)-TL**0.6/(10+TL**0.6))
SHRINKAGESTRAIN=-T/(35+T)*0.00061+TL/(35+TL)*0.00061
EXPAN(3)=CREEPSTRAIN+SHRINKAGESTRAIN
END IF
C
C Calculation of creep and shrinkage strain in slab concrete
C
IF((CMNAME.EQ.'CONCRETE_SLAB_AND_DIAPHRAGM').AND.(TIME(2).GT.63))
THEN
C
C 63 is used for 60 day old girder. Will change as per the age of
C the girder
C
T=TIME(2)-63
TL=T-DTIME
IF(TL.LT.0) THEN
TL=0
END IF
CREEPSTRAIN1=1.35*EE*(T**0.6/(10+T**0.6)-TL**0.6/(10+TL**0.6))
SHRINKAGESTRAIN1=-T/(35+T)*0.000608+TL/(35+TL)*0.000608
EXPAN(3)=CREEPSTRAIN1+SHRINKAGESTRAIN1
END IF
C
C Writing output for a single girder element in data file for
C verification
C
IF ((NOEL.EQ.75750)) THEN
WRITE (6,'(I5, F16.11,F16.11,F16.11,F16.11,F16.11,F16.11,
F16.11)') NOEL,T,DTIME,TL,EE,CREEPSTRAIN,SHRINKAGESTRAIN,
EXPAN(3)
END IF
C
C Writing output for a single deck element in data file for
C verification
C
C IF ((NOEL.EQ.9164)) THEN
WRITE (6,'(I5, F16.11,F16.11,F16.11,F16.11,F16.11,F16.11,
F16.11)') NOEL,T,DTIME,TL,EE,CREEPSTRAIN1,SHRINKAGESTRAIN1,
EXPAN(3)
END IF
C
RETURN
END
103
Appendix D. Script file for submission of job to the supercomputer
After submitting the script file, a job ID is created for each submission. The job ID can be
helpful in checking the status of the job, deleting the job, altering the walltime of the job, etc.
The #PBS statement in the above script is a special comment statement used to specify job
parameters to PBS. The PBS commands for the above functions are provided in section D-2.
104
D-2 Explanation of the script file
Role of each line in the script file
Command name Description
#PBS -l walltime=HH:MM:SS This directive specifies the maximum walltime (real time, not CPU
time) that a job should take. If this limit is exceeded, PBS will stop
the job. Keeping this limit close to the actual expected time of a job
can allow a job to start more quickly than if the maximum walltime
is always requested.
#PBS -l nodes=N:ppn=M This specifies the number of nodes (nodes=N) and the number of
processors per node (ppn=M) that the job should use. PBS treats a
processor core as a processor, so a system with eight cores per
compute node can have ppn=8 as its maximum ppn request. Note
that unless a job has some inherent parallelism of its own through
something like MPI or OpenMP, requesting more than a single
processor on a single node is usually wasteful and can impact the
job start time.
#PBS -N my_ABAQUS_job
#PBS -j oe Normally when a command runs it prints its output to the screen.
This output is often normal output and error output. This directive
tells PBS to put both normal output and error output into the same
output file.
#PBS -l software=ABAQUS+19 This specifies the number of tokens to be requested for running the
job.
module load ABAQUS/6.11-2 This command loads the ABAQUS 6.11-2 module.
cd $PBS_O_WORKDIR This changes the directory we submitted the script from.
cp *.inp $TMPDIR/ This copies the input file to the TMPDIR (temporary directory)
location.
cp *.f $TMPDIR/ This copies the Fortran file to the TMPDIR (temporary directory)
location.
cd $TMPDIR This changes the directory.
ABAQUS These commands specify the name of the input file for the analysis
job=<inputfilename.inp> and the Fortran subroutine file to be linked with the analysis. The
user=<Fortranfilename.f> number of CPUs to be used can be found by multiplying the
number of nodes requested by the number of processors available
cpus=<NxM> interactive
per node.
mv * $PBS_O_WORKDIR This command directs the computer to move the output files in the
working directory. The directory can be changed if needed.
105
PBS command for submitting a job, checking job status, and deleting a job
Command name Description
qsub <scriptfilename.sh> This command submits the PBS script to the PBS to make the script
eligible to run.
qstat This command shows the status of all the PBS jobs. The time
displayed is the CPU time used by the job
qstat-u <User ID> This command shows the status of all the PBS jobs submitted by
the user <User ID>. The time displayed is the walltime time used
by the job
qstat-f <Job ID> This command shows detailed information about the Job ID.
qdel <Job ID> This command deletes the job identified by the Job ID.
qdel $(qselect-u <UserID>) This command deletes all the jobs belonging to the provided User
ID.
PBS job states
State Meaning
Q It indicates that the job is queued and is waiting to start.
R It indicates that the job is currently running.
E It indicates that the job is currently ending.
H It indicates that the job has a user or system hold on it and will not
be eligible to run until the hold is removed.
106
Appendix E. MS Excel Macro to create plots
Sub Plots()
' Creatinging plots Macro
' Selects the cells for plotting
' Keyboard Shortcut: Ctrl+Shift+R
'ActiveCell.Offset(0, 2).Range("A1").Select
m=1
l = 40
Do While Cells(4, m).Value <> ""
Application.ScreenUpdating = False
'Dim 10day As Worksheet
'Set 10day = Wb.Sheets("10 day_plots")
Set objChart = ThisWorkbook.Sheets("90 day_plots").ChartObjects.Add(25, l, 500,
300).Chart
objChart.ChartType = xlXYScatterLines
'Set objSeries = objChart.SeriesCollection.NewSeries
q=m
Do While Cells(4, m).Value <> ""
ActiveSheet.Cells(4, m).Select
x=0
y=0
Do While ActiveCell.Value <> ""
x=x+1
y=y+1
ActiveCell.Offset(1, 0).Range("A1").Select
Loop
ActiveCell.Offset(-x, 0).Range("A1").Select
a = ActiveCell.Row
b = ActiveCell.Column
c=a+x-1
d=b+1
Set objSeries = objChart.SeriesCollection.NewSeries
107
objSeries.Name = Cells(2, m).Value
objSeries.XValues = Range(Cells(a, b), Cells(C, b))
objSeries.Values = Range(Cells(a, d), Cells(C, d))
objSeries.MarkerStyle = 0
m=m+2
Loop
objChart.SetElement (msoElementChartTitleCenteredOverlay)
objChart.ChartTitle.Caption = Cells(1, q).Value
With objChart.ChartTitle.Format.TextFrame2.TextRange.Font
.NameComplexScript = "Arial"
.NameFarEast = "Arial"
.Name = "Arial"
End With
objChart.ChartTitle.Format.TextFrame2.TextRange.Font.Size = 12
objChart.Axes(xlCategory).HasMajorGridlines = True
objChart.SetElement (msoElementPrimaryValueAxisTitleRotated)
objChart.SetElement (msoElementPrimaryCategoryAxisTitleAdjacentToAxis)
objChart.Axes(xlValue).AxisTitle.Caption = Cells(3, q + 1).Value
With objChart.Axes(xlValue).AxisTitle.Format.TextFrame2.TextRange.Font
.NameComplexScript = "Arial"
.NameFarEast = "Arial"
.Name = "Arial"
End With
objChart.Axes(xlValue).AxisTitle.Format.TextFrame2.TextRange.Font.Size = 10
objChart.Axes(xlCategory).AxisTitle.Caption = Cells(3, 1).Value
With objChart.Axes(xlCategory).AxisTitle.Format.TextFrame2.TextRange.Font
.NameComplexScript = "Arial"
.NameFarEast = "Arial"
.Name = "Arial"
End With
objChart.Axes(xlCategory).AxisTitle.Format.TextFrame2.TextRange.Font.Size = 10
objChart.Axes(xlCategory).MaximumScale = 200
objChart.PlotArea.Height = 250
objChart.PlotArea.Width = 400
objChart.PlotArea.Top = 30
objChart.PlotArea.Left = 20
m=m+1
108
l = l + 400
Loop
End Sub
109