You are on page 1of 268

12-58 COPY NO.

__
EFFECTIVE SLAB WIDTH
FOR
COMPOSITE STEEL BRIDGE MEMBERS
PRELIMINARY DRAFT INTERIM REPORT
Prepared for
National Cooperative Highway Research Program
Transportation Research Board
National Research Council
S.S. Chen, A.J. Aref, I.-S. Ahn and M. Chiewanichakorn
Department of Civil, Structural and Environmental Engineering
State University of New York at Buffalo
Buffalo, New York
5 November 2001
TRANSPORTATION RESEARCH BOARD
NAS-NRC
PRIVILEGED DOCUMENT
This interim report, not released for publication, is furnished only
for review to members of or participants in the work of the
National Cooperative Highway Research Program (NCHRP). It is
to be regarded as fully privileged, and dissemination of the infor-
mation included herein must be approved by the NCHRP
ii
ACKNOWLEDGEMENT OF SPONSORSHIP
This work was sponsored by the American Association of State Highway and Transportation
Officials, in cooperation with the Federal Highway Administration, and was conducted in the
National Cooperative Highway Research Program, which is administered by the Transporta-
tion Research Board of the National Research Council.
DISCLAIMER
This is an uncorrected draft as submitted by the research agency. The opinions and conclu-
sions expressed or implied in the report are those of the research agency. They are not neces-
sarily those of the Transportation Research Board, the National Research Council, the Federal
Highway Administration, the American Association of State Highway and Transportation
Officials, or the individual states participating in the National Cooperative Highway Research
Program.




iii
TABLE OF CONTENTS
LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii
ACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
SUMMARY OF FINDINGS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii
CHAPTER 1 Introduction and Research Approach. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1
CHAPTER 2 Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Analytical and numerical studies
Experimental Findings
Design criteria
Box-girder, truss, arch and cabled-stayed bridges
CHAPTER 3 Interpretation, Appraisal and Applications. . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Introduction
Key Assumptions
The 12t Limitation
Philosophies Behind Effective Width Formulations
Experimental Verification
CHAPTER 4 Conclusions and Suggested Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
Conclusions
Suggested research
REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
APPENDIX A Survey Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-1
iv
APPENDIX B Shear Lag Phenomenon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-1
APPENDIX C AASHTO Specification and Historical Background . . . . . . . . . . . . . . . . . . C-1
APPENDIX D Effective Flange Width of Foreign Specifications. . . . . . . . . . . . . . . . . . . . D-1
APPENDIX E Recommended Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E-1
APPENDIX F Plans For NCHRP 12-50 Compliance . . . . . . . . . . . . . . . . . . . . . . . . . . . . F-1
APPENDIX G Annotated Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G-1
v
LIST OF FIGURES
Figure 2-1 : Two-Plate Girder Bridge and Its Evolution in Switzerland (Lebet 1990) 37
Figure 2-2 : Example of a Simply Supported Bridge 38
Figure 2-3 : Example of a Continuous BridgeExample of a Continuous Bridge 39
Figure 2-4 : Plan and Elevation of I-78 Highway Bridge (Yen et al. 1995) 40
Figure 2-5 : Cross-Section View of I-78 Highway Bridge (Yen et al. 1995) 40
Figure 2-6 : Example of Plate Girder and Deck Slab Strain Gages (Yen et al. 1995) 41
Figure 2-7 : Example of Strain Gage Locations on Deck Slab (Yen et al. 1995) 42
Figure 2-8 : Open Type Cross Section (Haensel 1997) 43
Figure 2-9 : Weserbridge at Porta-Westfalica (Haensel 1997) 43
Figure 2-10 : Kalkgrabenbridge at Rudersdorf (Haensel 1997) 44
Figure 2-11 : Domitz Bridge (Hanswille 1997) 45
Figure 2-12 : Tied-Arch Bridge crossing the creek Mulde (Kuhlmann 1997) 46
Figure 2-13 : Tied-Arch Bridge in the motorway A96 (Kuhlmann 1997) 46
Figure 2-14 : Highway-Railway Bridge at Ciudad Guayna in Venezuela (Saul 1997) 47
Figure 2-15 : Railway Bridge across River Main at Nantenbach in Germany (Saul 1997) 48
Figure 2-16 : Road Bridge across River Elbe at Torgau in Germany (Saul 1997) 48
Figure 2-17 : Types of Connection of Stay Cables to Deck Systems (Cremer 1990) 49
Figure 2-18 : The Hoogly River bridge in India (Cremer 1990) 49
Figure 2-19 : The Annacys bridge in Canada (Cremer 1990) 50
Figure 4-1 : Failure surfaces of concrete (a) Failure surface and (b) Collapsed failure surface 95
Figure 4-2 : ANAMAT cyclic compressive stress-strain model behavior 95
vi
Figure 4-3 : 3D-View of steel girders, cross-frames of 1/4 scaled laboratory model (Span 1) 96
Figure 4-4 : Cross-section of 1/4 scale laboratory model in finite-element analysis 96
Figure 4-5 : One half (Span 1) of finite-element model for 1/4 scaled laboratory model 97
Figure B-1 : Concept of Effective width B-16
Figure B-2 : Force Boundary Conditions (Allen & Severn 1961) B-17
Figure B-3 : Force Boundary Conditions (Lee 1960) B-17
Figure B-4 : Force Boundary Conditions (Adekola 1968) B-18
Figure B-5 : Force Boundary Conditions (Fan & Heins 1974) B-18
Figure B-6 : Distorted cross-section B-19
Figure D-1 : Overhang Width for Various Cross-Sections (CSA 2000) D-20
Figure D-2 : Assumed Points of Inflection under Dead Loads (CSA 2000) D-20
Figure D-3 : Equivalent Span Length (J RA 1996) D-21
Figure D-4 : One-side Effective Flange Width (J RA 1996) D-21
Figure D-5 : Effective Width of Concrete Flange (Australia 1996) D-22
Figure D-6 : Equivalent spans, for effective width of concrete flange (Eurocode 4) D-23
Figure E-1 : Plan and Cross section of Specimen 4GQT E-13
Figure E-2 : Plan and Cross section of Specimen 4GHF E-14
Figure E-3 :Comparison of Stresses between 2-wheel and 3-wheel Loading E-15
Figure E-4 : Loading Position for Service Limit State at Positive Moment Region E-16
Figure E-5 : Loading Position for Service Limit State at Negative Moment Region E-18
Figure E-6 :Loading Position for Ultimate Limit State at Positive and Negative Moment Regions
E-18
Figure E-7 : Loading Scheme for Service Limit State and Ultimate Positive Moment Test E-19
vii
Figure E-8 : Loading Scheme for Ultimate Negative Moment Tests E-20
Figure E-9 : Location of Strain and Displacement Measurement E-21
Figure F-1 : Comparison scheme F-4
Figure G-1 : Four highway bridge test G-55
Figure G-2 : Stations of measurements G-55
Figure G-3 : Position and numbering of mechanical strain measurements G-56
Figure G-4 : Bridge Number 2 - Framing plan G-56
Figure G-5 : Bridge Number 2 - Cross section at Test Lines 1 and 3 G-57
Figure G-6 : Set-up of loading system for Pratt bridge G-57
Figure G-7 : Bridge details G-58
Figure G-8 : Placing sequence of loading in Southern segment of bridge G-58
Figure G-9 : Cross-section of selected bridge G-59
Figure G-10 : Lateral distribution of deflections at midspan for center loading G-59
Figure G-11 : Bridge section G-59
Figure G-12 : Girder distribution of deflections and stresses for center lane loading G-60
Figure G-13 : Cross-section of I-40 bridge at floor beam location G-60
Figure G-14 : Tindall Bridge G-61
Figure G-15 : Example of stresses at midspan of panel 4-5 stringers G-61
viii
LIST OF TABLES
Table 2-1: Effective Width of Simply-Supported Bridge Example 34
Table 2-2: Effective Width for a Continuous Bridge Example (mm) 34
Table 2-3: Selected Cable-stayed bridges with composite main girders 35
Table 4-1: An Example of 1/4 Scaled Model 92
Table 4-2: Parametric Study Analysis Cases 92
Table 4-3: Descriptions of Field Test Bridge (I-78) 94
Table A-1: Survey Results for Effective Slab Width Criteria A-3
Table D-1: Effective breadth ratios y for simply supported beams (BS 5400) D-18
Table D-2: Effective breadth ratios y for cantilever beams (BS 5400) D-18
Table D-3: Effective breadth ratios y for internal spans for continuous beams(BS 5400) D-19
Table D-4: One-Side Effective Flange Width (JRA 1996) D-19
Table E-1: Specimens E-11
Table E-2: Usage of Specimens E-11
Table E-3: Test Sequences E-11
Table E-4: Tentative Experiment Schedule for Each Specimen E-12
ix
ACKNOWLEDGEMENTS
The research reported herein was performed under NCHRP Project 12-58 by the Depart-
ment of Civil, Structural, and Environmental Engineering at the University at Buffalo (UB), State
University of New York (SUNY). UB was the contractor for this study, with the Research Foun-
dation of SUNY serving as Fiscal Administrator.
Dr. Stuart S. Chen, P.E., Associate Professor of Civil Engineering at UB, was the Project
Director and co - Principal Investigator. The other authors of this report are Dr. Amjad J . Aref,
Assistant Professor of Civil Engineering at UB and co Principal Investigator; Il-Sang Ahn,
Research Assistant and Ph.D. Candidate at UB, and Methee Chiewanichakorn, Research Assis-
tant and Ph.D. Candidate at UB. The work was done under the general supervision of Professors
Chen and Aref at UB.
Others assisting in contributing or pointing to material used in this report, all of whose
assistance is gratefully acknowledged, include the following:
Various state bridge engineers and TRB representatives who responded to a survey seeking
information relevant to the present study,
Mr. Arun Shirole of the National Steel Bridge Alliance, Mr. Ayaz Malik of the New York
State Department of Transportation, and Mr. Peter Stapf of the New York State Thruway
Authority, who served as members of the Industry Advisory Panel,
x
Mr. Y. Kitane of UB and Dr. S. Unjoh of the Public Works Research Institute of J apan, who
provided an English translation of the Effective Width criteria found in the J apanese code
and information about girder spacing practices in J apan, respectively,
Mr. G. Booth, Dr. J .-P. Lebet, and Mr. J oel Raoul, who provided information about British,
Swiss/Eurocode, and French effective width practices, respectively.
Dr. H. Gil of the Korea Highway Corporation, who provided information about recent effec-
tive width studies on a cable-stayed bridge in Korea,
Mr. I. Savage of Parsons Transportation Group, who provided information about a recent
effective width study on a major box girder crossing, and
Profs. B. T. Yen of Lehigh University and K. H. Frank of the University of Texas, and Mr. J .
Schulz of Bridge Diagnostics on bridge field test results and planning.
xi
ABSTRACT
This report documents and presents the results of the initial stages of a study of the effec-
tive slab width of steel bridge girders acting compositely with concrete decks. With the effective
slab width concept, values of maximum deflection or stress are obtained from elementary beam
theory. Shear lag effects are accounted for indirectly by replacing the actual slab width by an
appropriate reduced (effective) width. The determination of effective slab width directly affects
the computed moments, shears, torques, and deflections for the composite section and also affects
the proportions of the steel section and the number of shear connectors that are required.
An extensive literature review concerning issues related to effective slab width of steel-
concrete composite bridges was undertaken and is reported herein. This review was necessarily
quite broad, encompassing historical, analytical and experimental studies both domestically and
internationally of not only conventional girder and stringer-type superstructures but also tub-
girder systems, cable-stayed bridge floor systems, and material models for use in finite element
modeling and analysis of reinforced concrete bridge decks. Promising methodologies are identi-
fied for further investigation.
The findings of the preliminary study suggest that the 12t limitation on effective slab
width in the current AASHTO LRFD and Standard Specifications could be eliminated and that
companion laboratory and field experiments be conducted to provide verification of finite ele-
ment based investigations of effective slab width planned to provide the basis for new effective
criteria to result from the subsequent portions of the NCHRP 12-58 project.
xii
SUMMARY OF FINDINGS
Findings relevant to shear lag and effective width formulations of steel bridge girders in
composite action with concrete decks are grouped into the following categories, the first three
focusing principally on I - girders:
Analytical findings,
Laboratory and field experimental findings,
Domestic and international design criteria, and
Findings specific to tub/box, truss, tied-arch, and cable-stayed structural systems.
In the latter, the principal complication is the presence of axial load effects (tension in arch
ties and compression in cable-stayed composite-girder floor systems under live loads) in addition
to the flexural effects normally considered in effective width formulations. It is postulated that an
axial effective width, distinct from the flexural effective width, should be employed in such cases.
Replies to a survey distributed to the states indicate no leads regarding other studies inves-
tigating effective slab width. A small number of replies indicate a few recently constructed
bridges with large girder spacings (on which field testing may be possible). Where maximum
girder spacing policies are explicitly stated, they are generally quite conservative, with a 3 3.6 m
(10 12) limit being common. In some cases, the stated reason for this limit is to facilitate
eventual deck replacement. Where more liberal limits are stated, those limits are based on, e.g.,
maximum spans of stay-in-place forms (approximately 4.6 m =15) or maximum girder spacings
xiii
allowed for the use of empirical deck design (4.1 m =13.5). The relevance of the girder spacing
issue is that the 12t limitation (where t =deck slab thickness) typically governs effective slab
width for the larger girder spacings that the industry advocates for improved economy of this type
of bridge.
Analytical and numerical studies and laboratory and field experimental investigations
have consistently called into question thickness based limitations on effective slab width for con-
crete decks acting compositely with supporting steel girders. Unfortunately, however, there is a
paucity of experimental data suitable to assess shear lag in decks placed compositely on widely
spaced girders (for which the 12t limitation would easily govern effective width).
Internationally, European and J apanese limits on girder spacing are more liberal than in
the U.S. In J apan, for example, girder spacings are permitted up to 6 m when supporting pre-
stressed decks (although no field test results on bridges with girder spacings larger than 3 m are
available). In the few specifications where any dependence on slab thickness at all still appears in
the effective width formulation, that dependence is more liberal than it is in AASHTO. The for-
mulations that do not depend on slab thickness vary from the simple and straightforward (e.g., in
the Canadian code) to relatively complex.
Thus, the principal finding is that there is little to no justification for retaining the 12t lim-
itation in the AASHTO effective width formulation. Additionally, subsequent finite element
based analytical investigation of effective width which was contemplated in the original NCHRP
12-58 project scope should be augmented with laboratory and field experiments on structural con-
figurations with widely spaced supporting girders if experimental verification of the finite ele-
ment results for widely spaced girder configurations is desired.
1
CHAPTER 1
INTRODUCTION AND RESEARCH APPROACH
Shear lag can result in the stresses and strains at the web-flange intersections of a girder
being underestimated in calculations based on line-girder analysis and the elementary theory of
bending (in which it is assumed that plane sections remain plane during bending). It is traditional
to obtain correct values of maximum deflection or stress from the elementary theory by utilizing
an effective slab width concept in which the actual width of each flange is replaced by an appro-
priate reduced (effective) width (Moffatt 78). The determination of effective slab width directly
affects the computed moments, shears, torques, and deflections for the composite section and also
affects the proportions of the steel section and the number of shear connectors that are required.
The effective slab width is particularly important for serviceability checks (e.g., fatigue, overload,
and deflection), which can often govern the design.
In AASHTO bridge design specifications, the effective slab width for all types of compos-
ite bridge superstructures (interior girders) except for segmental concrete structures, is specified
as the least of: (1) 12 times the least thickness of the deck plus one-half the top flange width, (2)
one-fourth the span length of the girder, and (3) the girder spacing. For girder spacings 8 feet or
less, the effective width computed according to this provision generally includes all of the deck.
With the ever-increasing use of wider girder spacing, the contribution of the additional width of
deck is not fully recognized by the current criteria. The AASHTO Guide Specifications for Seg-
mental Concrete Bridges recognize the entire deck width to be effective unless shear lag adjust-
ments become necessary. Field measurements of modern composite steel bridges indicate that
2
recognition of more of the concrete deck is often necessary to better correlate actual with calcu-
lated deflections.
The above criteria are currently applied to all types of composite interior and exterior steel
bridge members with any combination of the following:
Conventional or High-Performance Steel Girder System:
- Tub-girder
- Two-girder system
- Conventional multi-girder system,
Deck System:
- Conventional Cast-in-Place
- Conventional concrete (e.g., =21 MPa or 28 MPa)
- High-Performance Concrete (HPC)
- Prestressed (e.g., Berger 83), either constant depth or variable depth, and often prestress lon-
gitudinally as well as transversely, on potentially very wide girder spacings,
Alignment:
- Right
- Skew
Span Location:
- Positive Moment Region
- Negative Moment Regions considered composite where sufficient shear studs and longitudi-
nal reinforcing steel are supplied, and
Applicable Limit State, e.g.,
- Service II and Fatigue (elastic), and
f

c
3
- Strength I and perhaps Strength II (possibly inelastic).
A further variation on the theme arises with composite deck systems that participate struc-
turally with tied arches or cable-stayed bridges. Thus, it is quite possible that the effective width
of the slab may well be different among some of these cases from more conventional multi-
stringer I-girder bridges. The effective width of decks using high-strength concrete may also be
affected by the larger elastic and shear moduli of the concrete. As listed, distinctions may be
needed between the effective width of slab to be used in positive and negative bending for both
analysis and design at the AASHTO serviceability and strength limit states considering both con-
ventionally reinforced and prestressed decks.
In accordance with the NCHRP 12-58 Project Statement, the objectives of the research
undertaken are to investigate both new and existing approaches for effective slab width and to
develop and validate the most promising of these. The research products will consist of criteria,
recommended specifications and commentary, and worked examples addressing applicable
AASHTO LRFD limit states in AASHTO LRFD format.
This interim report describes the results of the completed first three tasks of the NCHRP
12-58 project, which are described next.
The Task 1 statement is as follows: "Review relevant domestic and foreign field and labo-
ratory test results, analytical studies, and specifications regarding the effective slab widths for all
types of steel and concrete composite structures.
An extensive literature review concerning issues related to effective slab width of steel-
concrete composite bridges was undertaken and is reported herein. This review was necessarily
4
quite broad, encompassing historical, analytical and experimental studies both domestically and
internationally of not only conventional girder and stringer-type superstructures but also tub-
girder systems, cable-stayed bridge floor systems, high-performance concrete (HPC), and mate-
rial models for use in finite element modeling and analysis of reinforced concrete bridge decks.
Specifications for building structures were partially investigated as well.
The Task 2 statement is as follows: Using the findings from Task 1, summarize applica-
ble methodologies for determining the effective slab width for different types of composite steel
bridge superstructures typical of those in use today. As a minimum, I-girder and tub-girder cross-
sections should be considered. Both interior and exterior girders should be considered. Composite
floor systems that participate structurally with tied arches, cable-stayed bridges, and deck or
through trusses should also be considered, along with variable- or constant-depth composite, pre-
cast posttensioned decks. Consider the effects of the larger elastic and shear moduli of high-
strength deck concrete on the computed effective width. This work was undertaken and the
results reported herein.
Work on Task 3 utilized material coming out of Tasks 1 and 2. The Task 3 statement is as
follows: Prepare an interim report documenting the findings from Tasks 1 and 2. Provide practi-
cal recommendations for promising methodologies to determine effective slab width that can be
further developed and validated. Prepare an expanded work plan for the remainder of the project
describing the type of investigations needed to develop and validate the recommended methodol-
ogies. The work accomplished in Task 3 is contained in the present report. Relevant combina-
tions of key parameters are considered to include the following: (i) both positive and negative
moment regions, (ii) both interior and exterior girders, (iii) both linear (Fatigue and Service) and
5
nonlinear (Strength) realms of material behavior, (iv) both distributed and concentrated loads (the
effective width theoretically depends on the type of load), as well as (v) the different girder types
(I vs. tub) and floor systems used both domestically and internationally.
The expanded work plan is based primarily on Tasks 4 through 7 in the original work plan,
augmented by recommended laboratory and field experimental studies.
6
CHAPTER 2
FINDINGS
This chapter is organized as follows. A description of the survey sent to gather informa-
tion from various bridge owning jurisdictions is provided, followed by an overview of steel con-
crete composite bridge construction with an international flavor. Sections 2.1, 2.2, and 2.3 cover
analytical and experimental findings and design criteria with a focus on I girder superstructures.
Findings on other types (tub/box, truss, tied arch, and cable stayed) are grouped in section 2.4.
A survey was distributed to the state bridge engineers and TRB representatives in all 50
states. Replies were received from approximately 40 of these. The replies are tallied in Appendix
A along with a copy of the survey form itself. Replies indicate no leads regarding other studies
investigating effective slab width. A small number of replies indicate a few recently constructed
bridges with large girder spacings (on which field testing may be possible). Where maximum
girder spacing policies are explicitly stated, they are generally quite conservative, with a 3 3.6 m
(10 12) limit being common. In some cases, the stated reason for this limit is to facilitate even-
tual deck replacement. Where more liberal limits are stated, those limits are based on, e.g., maxi-
mum spans of stay-in-place forms (approximately 4.6 m =15) or maximum girder spacings
allowed for the use of empirical deck design (4.1 m =13.5).
Further direct inquiries to international sources were made, including personal contacts in
Europe (Switzerland, France, and the U.K.) and J apan. European and J apanese limits on girder
spacing are more liberal than in the U.S., as described further later in this chapter. In J apan, for
example, girder spacings are permitted up to 6 m when supporting prestressed decks (although no
field test results on bridges with girder spacings larger than 3 m are available).
7
Cremer (1990) defined the characteristics of steel-concrete composite bridges as: (1) The
cross-section of the bridge involves both materials; (2) Each of the two materials contributes sig-
nificantly to the load carrying capacity of the structure; (3) The steel part of the structure is fabri-
cated either by welding or bolted. Therefore, cable-stayed bridges or tied-arch bridges with only
the stays or suspenders in steel are excluded; (4) Each of the two materials contributes to the total
cost in similar proportions. In Germany, modern steel concrete composite bridges rely on two
basic improvements of the design, crack width control instead of stress control in concrete and
using a concrete bottom chord instead of steel, compared to traditional composite bridges (Saul
1997). As a very popular composite configuration, two-girder bridges that have two main girders
without any supplementary stringers in the floor system, as shown in Figure 2-1, have been used
in several countries in the world, including Germany, France, Switzerland, and Japan (Lebet
1990; Cremer 1990; Haensel 1997; Saul 1997).
Lebet (1990) pointed out the trends in composite design as the simplification of the steel
structure and construction details, and generosity in the design of individual elements (web thick-
ness and slab thickness). Also, the same figure shows the evolution of composite bridge in Swit-
zerland emphasizing reduction of longitudinal stiffeners and thicker web with fewer vertical
stiffeners. Other examples of composite bridges can be found in several European countries,
including Secheval viaduct, Remouchamps viaduct in Belgium, Layrac bridge in France, Syre
viaduct in Luxembourg (Cremer 1990).
8
2.1 ANALYTICAL AND NUMERICAL STUDIES
The effective width or length is an engineering term which is introduced in order to utilize
an already prepared analysis or design procedure with modification of parameters used in that
procedure. For example, effective span length of continuous bridges is typically defined as the
distance between contra flexural points, which can be used with simply-supported beam analysis
for simplicity. In studies of concrete T-beams, the effective width is used to represent some por-
tion of concrete slab that can behave integrally with the concrete web. By introducing effective
flange width, it is possible to use the familiar rectangular beam design methodology, which is
advantageous in its simplicity.
Due to the action of in-plane shear strain in the flange plate of a girder under bending
loads, the longitudinal displacements in the parts of the flanges remote from the webs lag behind
those near the webs. This phenomenon, termed shear-lag, can result in inaccurate estimation of
the deflections and stresses at the flange based on the elementary theory of beam bending (Moffatt
1978). Shear lag also has to be considered in the design of steel structure connections, where
effective net area incorporates consideration of shear-lag (Salmon and J ohnson 1996). For the
shear-lag phenomenon in slab on beam structures, effective flange width approaches were intro-
duced from the early studies. Along with elementary beam bending theory it produced simple and
efficient methods for deflection and stress calculations. However, other approaches, e.g., the ana-
log-beam method (Gjelsvik 1991), can be used for the same purpose.
Since the 1920s there have been many analytical or numerical studies for shear-lag effect
of stiffened thin plate, T-beam structures, and composite steel-concrete structures based on con-
tinuum mechanics to develop more realistic and reasonable definition of effective flange width.
9
These studies assumed in-plane plate behavior for slabs and elementary beam bending behavior
for beams (von Krmn 1924; Metzer 1929; Chwalla 1936; Winter 1940; Schade 1951; Schle-
icher 1955; Timoshenko and Goodier 1970). For non-homogeneous materials, J ohnson and Lewis
showed that the effective width expression based on von Krmn is valid for beams in which the
flange and web are of different materials
1
(Johnson and Lewis 1966). Marguerre (1952) consid-
ered not only in-plane behavior but also bending behavior for the deck plate and developed two
effective widths for a slab: one due to plane stress effect in the plate and the other due to bending
action of the plate. Several research studies were continued based on these two plate behaviors
with variation of loading conditions and force equilibrium and boundary conditions (Allen and
Severn 1961; Lee 1962; Brendel 1964; Adekola 1968). Heins and Fan studied effective width of
composite I-beams considering stress redistribution after yielding of its members. They used
orthotropic plate governing equations in their study and concluded that the effective width in the
1973 AASHTO Specification, as applied to Load Factor Design, overestimated the widths for the
interior girders and underestimate the widths for the exterior girders (Heins and Fan 1976).
With development of computing capacity, composite beams could be analyzed much more
easily with various loading conditions and geometric shapes. Moffatt and Dowling (1978) used
the finite element method for developing flange effective width of box girder bridges. Their study
was used as the basis for effective breadth rules of BS 5400. Cheung and Chan (Cheung and Chan
1978)
2
used finite strip methods to analyze realistic effective width for over 300 bridge models
with Ontario highway bridge loadings and AASHTO HS-25 wheel loads. They found that for
most practical bridges, girder size and deck thickness have very little effect on the effective width
1. C.G. Salmon and J .E. J ohnson, Steel Structures: Design and Behavior (4th Ed.) (HarperCollins College Publishers,
1996).
2. The range of model parameters: girder spacing: 1.52-10.97m, number of girders: 2-9, span length: 10.67-67.06m
10
calculations. As the major contributing factors, they used span length and girder spacing. Subse-
quently version of the Ontario Highway Bridge Design Code and the current Canadian bridge
code adopted their results for defining effective width of composite and concrete bridges (CSA
2001).
One of the aspects that must be considered in an effective flange width study is that the
conventional effective width concept is developed as a simplified method to calculate stresses and
deflections, not capacity or resistance. Therefore, effects of shear-lag phenomenon at the ultimate
limit state can be different from those at the service limit state. J ohnson (1970) studied effective
width related to the ultimate strength design of a continuous composite floor system under distrib-
uted load and concluded that except when flanges are very wide, the effective span of the slab can
be taken as the clear distance between the edges of the flanges. Also, shear-lag effects are not con-
sidered at the ultimate limit state in BS 5400: Part 5 (BSI 1979). In the current AASHTO LRFD
specification, the same effective flange width is used for both service limit state and ultimate limit
state. The current Canadian code also uses the same effective width for stress and capacity calcu-
lation (CSA 2000). Underestimation of effective flange width is conservative for the resistance of
a beam, so this may be advantageous for simplicity of design, even though inaccurate. This is also
the basic idea behind using elastic analysis for effective width evaluation with uncracked con-
crete, because effective widths are increased both by inelasticity and by cracking of concrete
(J ohnson and Anderson 1993).
For bridges and buildings, effective width of flange in the negative moment region is one
of the critical issues, and this is closely related with crack control. Pantazopoulou and Moehle
(1990) introduced nonlinear strain distribution, which means plane sections do not remain plane
11
after deformation. They showed that effective width of flanges can be considered as active in neg-
ative bending moment regions of concrete T-beams. Practically, most specifications including
AASHTO LRFD use effective span length for evaluating effective flange width in the negative
moment region. By introducing effective span length, the same rules defined for simply-supported
spans can be used for continuous spans (J RA 1996 [89, 90]; CSA 2000).
12
2.2 EXPERIMENTAL FINDINGS
2.2.1 Laboratory Experimental Findings
Laboratory experiments are recognized to be one of the key ways to determine the actual
behavior of new and existing slab-on-girder bridges. Most experiments usually showed that many
existing bridges had some reserve capacity. There has been considerable research activities in this
area. As one of the early experimental studies for shear-lag or effective flange width of composite
steel-concrete structures, Siess (1949) used a 60 inch long isolated composite structure with 1 3/4
in slab thickness. After measuring strains at the top surface of the concrete slab induced by con-
centrated 5.2 kips load applied at the mid span of the simply supported structure, he concluded
that the distribution of stress on the top surface of the slab of an isolated composite T-beam is sub-
stantially uniform, both at a section near the load and at a section some distance removed.
Mackey and Wong tested four composite structures, which had the same length and slab
thickness (2 inch) with different girder spacing (Mackey and Wong 1961). Their b/L ratios are
0.26, 0.56, 0.70, and 0.84, where b =girder spacing and L =span length. Concentrated load was
applied at mid-span of each girder, and effective widths for each specimen were calculated based
on measured strains at the slab. After comparing with analytical studies, they concluded that the
effective width determined experimentally is larger than that predicted by theory and is much
larger than that recommended in AASHTO 1957. Hagood and others used a similar size specimen
as used in Mackey and Wongs study, but they applied much higher loads, beyond their yielding
stresses, on their specimen (Hagood, Guthrie, and Hoadley 1968). The b/L ratios of their speci-
mens were 0.33, 0.5, and 0.67. For the specimen with 0.67 ratio, shear-lag effects were increased
13
with each increment of loading level. However, for the specimen with 0.33 ratio, shear-lag was
not definitely observed.
Many researchers conducted full-scale tests to study the structural behavior of slab-on-
girder bridges. Corrado and Yen (1973), Labia et al. (1997) and Roshke and Pruski (2000) per-
formed failure tests on composite bridges. These include box-girders as well as posttensioned slab
bridges. Moore and Viest (1993) proposed guidelines for inelastic design of continuous girders
with non-compact elements as the results of a scaled model experiment.
Skew effect is another area where changes to the structural behavior of composite bridges
are observed. Bakht and Agawal (1995) studied the effect of skew supports on the minimum
requirement for the amount of reinforcement required in the end zones near the skew supports.
Ebeido and Kennedy (1996) tested continuous skew composite bridges to come up with simpli-
fied formulas for computing the moment distribution factors.
Kathol et al. (1995) designed, constructed, and tested a full-scale bridge model in
Nebraska up to the ultimate limit state. Razaqpur and Nofal (1998) tested a simple span bridge to
determine the ultimate load capacity. Both results were utilized by Deng and Ghosn (2000) for the
non-linear analysis of composite steel girder bridges using a modified grillage (grid) analysis
method.
One of the limitations is that inelastic response (i.e. Strength I Limit State) can be
achieved in laboratory testing but not in most field testing. Therefore, the results are more reliable
in terms of unintended factors such as curbs and railings.
14
The most important part of the findings on laboratory experiment is the methodologies of
how test specimens were instrumented and tested. Even though there are not many test results
available that explicitly focused on effective width, there are many related experimental studies
involving composite girders. A number of these are summarized in Appendix G.
Even though there are some related experimental results that can be used for service limit
state, only a small number of bridge tests and model tests at the ultimate limit state have been con-
ducted (e.g. Newmark, Siess, and Penman 1946; Siess and Viest 1953; Chapman 1964; Slutter
and Driscoll 1965; Daniels and Fisher 1967; J ohnson 1970; Burdette and Goodpasture 1973;
Hamada and Longworth 1976; Botzler and Colville 1979; Kennedy and Soliman 1992). These
studies primarily focused on understanding the composite structural behavior without any signifi-
cant consideration of shear-lag or effective flange width. To evaluate effective flange width based
on strain measurements, sufficient strain measurements at the deck are necessary. Such measure-
ments have typically which has not been done for most composite structure tests. However,
Mackey and Wong (1961), Hagood et al. (1968), Heins (1980) reported strain measurements on
the concrete deck.
Needless to say, full-scale tests deliver the most realistic behavior of bridges. Although
some researchers have conducted such tests, scaled model tests are typically preferred not only
because of their accuracy, but because of their economy and feasibility. Also, it has been shown
that carefully planned scaled model tests can represent closely the behavior of full-scaled bridges
(Richart 1949; Newmark 1949). Considerable previous research has been conducted based on
either beam model tests or slab-on-beam model tests. Isolated beam models and their results have
been used to study effective flange width and elastic/plastic behavior of composite beams, espe-
15
cially for building structures with uniformly distributed loadings (Chapman 1964; J ohnson 1970;
J ohnson 1997). However, for the purpose of model verification and validation of full bridge mod-
els, reconstructing the behavior of multi girder systems based on test results on isolated beam can
introduce severe error in the interpretation of test results.
Studies in the 1950s by Siess and Viest (1953) revealed that moments in both the negative
and positive moment regions of two continuous composite bridge models were nearly the same.
Based on this research, it was concluded that no special provisions were needed for the design of
continuous composite bridges (Daniels and Fisher 1966). For continuous bridges, the current
AASHTO Specification also recommends using the same effective flange rules as for simply-sup-
ported bridges with the span length defined by the distance between contra-flexural points. From
the early studies, there was no objection to using the same rules of effective flange width for sim-
ply-supported span and for the continuous span between contra-flexural points. However, com-
pared to the AASHTO Specification, most other countries codes specify the ratio of span length
for evaluation of effective flange width, distinguishing exterior span, interior span, and interior
support span (BSI 1979; J RA 1996; CSA 2000; Eurocode 4 1992). This change is a possible ave-
nue that AASHTO Specification can follow to produce an easy and more efficient evaluation of
effective flange width for continuous bridge design. Therefore, evaluating the ratio of span length
that can be used in effective width calculation combined with experimental verification is
required.
Compared with behavior at the ultimate limit state, the service limit state is relatively well
known. Effective flange width at the ultimate limit state is greater than that at the service limit
state, which results in safer or conservative design for the ultimate limit state if the same effective
16
width is used (J ohnson and Anderson 1993). Also, compared with behavior in the composite neg-
ative moment region of a continuous span, behavior of the positive moment region of simply sup-
ported and continuous spans is well understood. Generally, local or lateral bottom flange buckling
is thought to be the general failure mode in the negative moment region, where the slab may be
cracked in tension throughout its depth (Chapman 1964; Daniels and Fisher 1966; J ohnson 1970;
Subcommittee 1974; Hamada and Longworth 1976). Several research studies supported by exper-
iments showed that the full girder spacing could be used as the effective flange width in the nega-
tive moment regions at the ultimate limit state, also, using more conservative effective width at
the ultimate limit state can compensate for unknown behavioral effects in that region (J ohnson
1997).
2.2.2 Field Experimental Findings
Nowadays, field experiments are commonly used for load rating purposes at the service
limit state. Chajes et al. (1999) developed methods for integrating bridge field testing and in-ser-
vice monitoring into DelDOTs bridge management efforts. In Switzerland, Moses et al. (1994)
tested a large number of bridges. The test information was statistically used in evaluation of the
bridge. Many other researchers (Saraf et al. 1996, Reid et al. 1996 and Kathol et al. 1995) per-
formed Proof Load testing to evaluate the structural integrities of the bridges.
In the past, there were not many failure tests on highway bridges. On of the few consists of
the destructive tests conducted by Burdette and Goodpasture (1972). Four highway bridges in
Tennessee were tested to failure. The measured capacities were compared with the computed
capacities, determined on the basis of strain compatibility relations using the actual stress-strain
relations of the material in the structure.
17
Bakht and J aeger (1988) performed an ultimate load test of a slab-on-girder bridge in Can-
ada. The final report was very well documented, including the locations of strain gages, LVDTs
and loadings. Unfortunately, this bridge was designed as non-composite without any shear con-
nectors.
Destructive testing of two 80-year-old truss bridges was conducted by Aktan et al. (1994).
An ultimate load test of an old reinforced concrete slab bridge was done by Azizinamini et al.
(1994). Continuous skew composite bridges were tested to collapse by Helba and Kennedy [77,
78] (1994). None of the above have wide-girder spacing characteristics. However, the instrumen-
tation methods were similar.
As mentioned previously, the inelastic response is not possible to obtain in field testing for
most cases. This becomes a major concern that limits the amount of test data obtained from field
testing situations. Another concern is that most field-tested bridges have been in service for a
number of years. Deterioration gives rise to uncertainties regarding material properties, structural
redundancies and boundary conditions.
According to the literature review, a steel-plate-girder bridge which carries the east bound
traffic on the I-78 highway over the Delaware River near Easton, Pennsylvania is perhaps the
closest to what is needed for the present study. This is a seven-span bridge with a total length of
1222-ft. The overall deck slab width, from outside to outside of the parapet wall is 51-6. The
span lengths are 100-ft., 169k-ft., 228-ft., 228-ft., 228-ft., 169-ft. and 100-ft. The structural deck
thickness is 10-in. The main superstructure components consist of four welded plate girders,
spaced at 14-3 center to center, which qualifies as a wide-girder spacing. Unfortunately, original
18
data is no longer available. Plan and cross section of the bridge are shown in Figure 2-4 and Fig-
ure 2-5, respectively. Locations of strain gages are also depicted in Figure 2-6 and Figure 2-7.
Some additional information on the relevant experimental literature appears in Appendix
E.
19
2.3 DESIGN CRITERIA
AASHTO SPECIFICATIONS
AASHTO LRFD Specifications (AASHTO 1998) specify effective flange width of com-
posite bridges in section 4: Structural Analysis and Evaluation. In particular, the specified effec-
tive width can be used for determining resistance for all limit states. As far as the calculations of
effective slab width for deflection are concerned, the full flange width shall be used. Distinguish-
ing interior girders from exterior girders, effective width in the absence of a more refined analysis
is explicitly specified as outlined below:

Even though the differences in effective width due to loading types have been studied
(Hambly 1991), i.e. uniformly distributed vs. concentrated loads, the differences from the various
loading conditions are not defined in the Specifications, except for the case of axial forces, which
are applied to the cross section due to post-tensioning and other concentrated forces. The Specifi-
For Interior beams, the effective flange width may be taken as the least of:
- One-quarter of the effective span length;
- 12.0 times the average thickness of the slab, plus the greater of web thickness
or one-half the width of the top flange of the girder; or
- The average spacing of adjacent beams.
For Exterior beams, the effective flange width may be taken as one-half the
effective width of the adjacent interior beam, plus the least of:
- One-eighth of the effective span length;
- 6.0 times the average thickness of the slab, plus the greater of half the web
thickness or one-quarter of the width of the top flange of the basic girder; or
- The width of the overhang.
(AASHTO 1998)
20
cations define the effective span length as the total span length for simply-supported span and as
the distance between permanent load inflection points for continuous spans. Based on the effec-
tive span length, the same rules are applied for negative moment regions. Effective width rules are
defined for determining resistance for all limit states, however, it can be inferred from the effec-
tive span definition that they mainly concern the flexural behavior. Effective flange width for
axial forces that is needed in composite cable-stayed bridges, for example, may not be based on
these rules. Actually, the AASHTO Standard Specification uses the expression as ... effective
width of the slab as a T-beam flange..., which limits their application to flexural behavior
(AASHTO 1996).
FOREIGN SPECIFICATIONS
The British Standards Institution publishes BS5400 Steel, Concrete and Composite
Bridges Specifications as its bridge standard (BSI 1979). Effective width for composite bridges is
specified in Part 3. Code of practice for design of steel bridges and Part 5. Code of practice for
design of composite bridges. Following part 3, the effect of shear lag in the flange may be
neglected at the ultimate limit state, but not at the serviceability limit state. Tables in Part 5 distin-
guish among simply supported, cantilever, and internal spans of continuous beams as shown in
10.38.3.1 In composite girder construction the assumed effective width of the slab as a
T-beam flange shall not exceed the following:
(1) 1/4of the span length
(2) the distance center to center of girders
(3) 12 times the least thickness of the slab
(AASHTO 1996)
21
Appendix D. For each case, effective width ratios for mid span, quarter span, and support are
specified based on the girder spacing and span length ratio, b/l.
The current Canadian Highway Dridge Design Code (CSA 2000) adopted effective width
provisions from the previous Ontario Highway Bridge Design Code. In calculating bending resis-
tance and bending stresses in slab-on-girder and box girder bridges having a concrete slab
(whether girders are steel or concrete), a reduced cross-section defined by the effective width for-
mulation shall be used. The reduced section is not only used for service limit state, but also for
ultimate limit state.
In J apan, one-side effective flange width, , is used to evaluate the effective flange width
of beams and/or stringers in calculating stresses and displacements (J RA 1996). This effective
The effective width consists of a left hand overhang, a central portion, and a
right-hand overhang. The overhang, Be, shall be determined as follows:
where
(CSA 2000)
B
e
B
----- 1 1
L
15B
----------
)
`

3
= for
L
B
--- 15
B
e
B
----- 1 = for
L
B
--- 15 >
L = span, for simply supported spans
length of positive or negative region of dead load moment
B
e
= the effective width
B = lefthand or right-hand overlang
22
slab width parameter illustrated in tabular format is shown in Appendix D. The following equa-
tions are used for effective width calculations. Equation I is based on the analysis of simply-sup-
ported beams under uniformly distributed load, whereas Equation II is based on a concentrated
load applied at the midspan.
In Australia, two different effective flange width criteria appear in the Specification (AS
2327.1:1996) for evaluating the section strength of (a) solid slabs and (b) composite slabs. Solid
slab represents the concrete flange that is supported by steel girders, while composite slab is the
(Equation I)
(Equation II)
where
(JRA 1996)
b =
b
l
--- 0.05
1.1 2
b
l
---
\
| |
b = 0.05
b
l
-- - 0.30 < <
0.15 l = 0.30
b
l
---
b =
b
l
--- 0.02
1.06 3.2
b
l
---
\
| |
4.5
b
l
---
\
| |
2
+ b = 0.02
b
l
-- - 0.30 < <
0.15 l = 0.30
b
l
---
= one-side effective flange width (cm)
b = either one-half the center-to-center girder spacing (cm)
or a length of the overhang (cm)
l = equivalent span length (cm)
23
combination of the in-situ concrete slab and the profiled steel sheeting spanning between two
steel girders.
In Eurocode 4, action effects may be calculated by elastic analysis, even where the resis-
tance of a cross section is based on its plastic or nonlinear resistance (Eurocode 4 1992). For ser-
viceability limit states, elastic analysis should be used with appropriate corrections for nonlinear
effects such as cracking of concrete. When elastic analysis is used, a constant effective width may
be assumed over the whole length of each span. This value may be taken as the value
beff,1
at
midspan for a span supported at both ends, or the value b
eff,2
at the support for a cantilever. The
total effective width b
eff
may be determined from the following equation.
Where the concrete flange is a solid slab, its effective width, b
eff
, shall be calculated as
the sum of the distances, bc, measured on each side of the center-line of the steel
beam, where b
e
is in each case the smallest of:
(a) Lef / 8, where Lef is the effective span of the beam
(b) in the case of a concrete slab with a free edge (i.e. an edge beam situation). either
the perpendicular distance to the edge measured from the center-line of the beam,
or 6 times the overall depth, Dc, of the concrete slab plus half the width of the steel
beam flange, bsf1; and
(c) in the case of a concrete slab which spans between two steel beams (i.e. either an
edge beam or internal beam situation), either half the center-to-center distance
between the steel beams or 8 times the overall depth, Dc, of the concrete slab plus
half the width of the steel beam flange, bsf1.
(AS 2327.1:1996)
24
where:
b
0
is the distance between the centres of the outstand shear connectors or, for
angles, the width of the connector,
b
ei
is the value of the effective width of the concrete flange on each side of the
web and taken as L
e
/8 but not greater than the geometric width b
i
. The
value b
i
should be taken as the distance from the outstand shear connector
or, for angles, the edge of the connector to a point mid-way between adja-
cent webs, measured at mid-depth of the concrete flange, except that at a
free edge b
i
is the distance to the free edge.
L
e
The approximate distance between points of zero bending moment. For
typical continuous composite beams, where the design is governed by a
moment envelope from various load arrangements, and for cantilevers, L
e
may be assumed to be as shown in figure, in Appendix D.
b
i
is equal 1.0 for the effective width b
eff,1
at midspan and for the effective
width b
eff,2
at internal supports and cantilevers. For the determination of
b
eff,0
at end supports the value b
i
should be determined from the below
equation where b
ei
is the effective width of the end span at midspan.

(Eurocode 4 1992)
b
eff
b
0

i
b
ei
+ =

i
0.55 0.025
L
e
b
i
----- +
\
| |
1.0 =
25
Numerical Comparison
To compare effective flange width for various specifications, two examples are intro-
duced. The first one is a single span simply-supported bridge, as shown in Figure 2-2, the design
of which described in AASHTO Design examples (HDR Engineering 1995). The second example
is a three-span continuous bridge taken from Barker and Puckett (1997). Figure 2-3 shows the
cross section and a plan view of this bridge. The effective widths for flexural stress evaluation are
summarized in Table 2-1 and Table 2-2. Because shear-lag is considered only for stress check of
cracked concrete in BS 5400, direct comparison with another value is not appropriate. It may be
better to use the total width for the purpose of comparison.
Larger girder spacing and longer span used in the simply-supported bridge example made
the effective width from AASHTO Specification smaller than the other values. For this case, the
total width is used for the other specifications. Because of the smaller girder spacing of the contin-
uous bridge example, total width is used for most specifications. However, the J apanese Code
delivers remarkably smaller effective widths at the interior support than the other codes.
26
2.4 BOX-GIRDER, TRUSS, ARCH AND CABLED-STAYED BRIDGES
Shear lag phenomena in box-girder type sections have been recognized by many research-
ers. Different methods were proposed for analyzing box-girder bridges. One of the fundamental
concepts was the use of the principle of minimum potential energy for shear lag analysis by Reiss-
ner (1946). Foutch and Chang (1982) reviewed and made use of this concept in a case study to
compare the deflection of a cantilever box girder with shear lag and without shear lag by Euler-
Bernoulli beam theory. It was found that the shear lag intensity function is highly dependent on
the type of loading that is applied to the box girder type of structures. The concept of effective
slab width emerged in the studies of thin-walled steel box sections where the primary stress result-
ants are membrane forces rather than bending moments (Ansourian 1983). Kuzmanovic and Gra-
ham (1981) investigated the effective slab width using a simplified method, which was developed
from the principle of minimum potential energy. The effective slab width was found to be a func-
tion of geometrical factors such as slab width, effective span length and type of loading. Finite
element modeling of box-beam bridges was also studied by Kostem et al. (1984). Four planar
finite elements on each side of the beam through the depth of the beam were used. However,
higher order finite elements were recommended in the web of box-girders. It was also suggested
that in the verification of the accuracy of any given finite element model, the comparisons should
not be restricted to the deflection, but should be extended to the stresses as well.
Practically, there are many recently constructed box-girder bridges in the country. For
example, the Woodrow Wilson Bridge is one such steel box-girder bridges which was designed to
be composite with the concrete deck (Savage and Nasim 2001).
27
In this study (Savage and Nasim 2001), a finite element investigation was conducted in
order to investigate suitability of AASHTO effective width criteria for tub girders with webs 4.6
m (15) apart. Both top and bottom flanges were investigated, the latter having two longitudinal
stiffeners. Deck thickness increases to 0.4 m (15-1/2) where the supporting box girders splay out
to a 7.6 m (25) girder spacing. 3D solid elements were used, and stresses across the flanges and
concrete decks were examined. From peak stresses that were recorded, effective width was
obtained by back-computing equivalent net stresses. A sensitivity study was not undertaken for
various slab thicknesses. The result of the investigation provided justification for using AASHTO
effective width criteria in the design (Savage 2001).
The Storrow Drive Bridge (Kumarasena and McCabe 2001) is one of the largest steel box-
girder bridges in North America. This bridge was designed to support the 76 wide road way con-
necting Bostons Storrow Drive and Leverett Circle with I-93. Because of the large web spacing,
this bridge may provide an important point of reference for upcoming parametric studies.
As typical types of modern composite bridge superstructures, composite open type steel
girders, composite twin-box steel girders, composite box girders, composite tied-arches (bow-
string arches), composite truss girders, and composite cable-stayed can be considered. The most
common cross section of an open type girder with up to 17m width in Germany is shown in Fig-
ure 2-8 (Haensel 1997). The steel structure consists of two plate girders stabilized by cross
frames. Raoul and Hoorpah (2000) describe a similar type of bridge in France. As an example of
composite twin-box steel girders, the Weserbridge at Porta-Westfalica is shown in Figure 2-9.
Figure 2-10 shows the Kalkgrabenbridge as an example of composite box girders.
28
Composite tied-arches, a steel arch under compression and a concrete slab as tension tie,
can be thought to be unreasonable. But mainly due to economical and functional reasons a con-
crete slab as road deck is often preferred over an orthotropic steel deck (Kuhlmann 1997). Some
older tied-arch bridges had been designed to reduce the contribution of the concrete slab as ten-
sion tie with prestressing used to limit tensile stresses in deck slab. As the first tied-arch with a
reinforced concrete deck acting as a tension member, Germany constructed a bridge in 1991
shown in Figure 2-11 (Hanswille 1997). In this type of bridge, deck concrete cracking and the
effect of tension stiffening change the distribution of internal forces. Therefore, those effects must
be considered at both service and ultimate limit states. Following this new design concept, load
transmission between main steel girder and concrete slab is achieved by studs as shown in Figure
2-12. Horizontal studs as well as conventional vertical studs can be used, and Figure 2-13 shows a
example of a tied-arch bridge with longitudinally oriented composite deck structure.
Three bridges are described as examples of using concrete in the bottom chord. They are a
highway-railway bridge across the Caroni river at Ciudad Guayna in Venezuela (Figure 2-14), a
railway bridge across the river Main at Nantenbach in Germany (Figure 2-15), and a road bridge
across the river Elbe at Torgau in Germany (Figure 2-16) (Saul 1997). The first and the last ones
are composite box girders with concrete bottom chord, and the second one is a composite truss
girder bridge.
Most cable-stayed bridges built before the1970s have steel decks. However, the develop-
ment of computing makes it possible to implement composite cable-stayed bridges, and cable-
stayed bridges are typically not built with steel decks nowadays (Cremer 1990). By using a con-
crete deck, this type of bridge can reduce cost of fabrication and vulnerability to fatigue compared
29
with steel orthotropic decks. By reducing weight of girders, these bridges are also competitive
with concrete girder types in mid or long span ranges. Generally, composite cable-stayed bridges
have two planes of stays to support horizontal forces from the stays applied before the concrete
deck is finished. The stays can be anchored at the ends of an appropriately strengthened cross-
frame as shown in Figure 2-17(a), or they can be connected to the main girder directly as shown in
Figure 2-17(b). Table 2-3 shows selected composite cable-stayed bridges that have been con-
structed.
For cable-stayed bridges of large spans (larger than 400 or 500m), the cost of the stays
increases with the square of the span, for the same deck. Therefore, the benefits that come from
using concrete girders rapidly decreases, and composite construction can reduce the cost by using
lighter girders (Cremer 1990). The Hoogly River bridge in India (Figure 2-18) and the Annacys
bridge in Canada (Figure 2-19) are good examples of composite long span cable stayed bridges.
Actually, the composite alternate design of the Sunshine Skyway bridge in 1982 had all the main
characteristics of modern composite cable-stayed bridges as described in Svensson (1997):
(1) Open steel grid from I-girders
(2) Outside main girders to which the stay cables are directly connected
(3) Precast concrete slab elements, spanning longitudinally between floor beams, con-
nected by lap spliced reinforcement in CIP joints
(4) Erection in small elements possible
(6) Crack control in the slab by rebar only, without post-tensioning
These concepts were successfully applied to the Annacys bridge and initiated a series of
cable-stayed bridges with composite decks all over the world. For composite main girders, a main
30
concern is to limit tensile stresses in the deck slab. Transversely, this is achieved by using two
outer cable planes, such that the slab can be considered as the top flange of simply-supported
composite beam. In the longitudinal direction, the slab sustains low bending stresses because the
center of gravity is located close to its underside, and negative moments from live loads are rela-
tively small. These tensile stresses can be reduced by normal forces from inclined stay cables and
by proper camber control during construction (Svensson 1997).
In the composite girder construction of a cable-stayed bridge, a section is typically built
by: (1) installing steel deck segments; (2) erecting and partially stressing cables; (3) placing and
post-tensioning concrete decks longitudinally and/or transversely; (4) forming the steel-concrete
composite section using cast-in-place concrete; and (5) adjusting the cable forces (Cai, Zhang,
and Nie 1998). During incremental cantilevering of girders supported by stay cables, casting of
the concrete slab is delayed compared to the erection of the steel skeleton. As a result, the hori-
zontal component of the force in the stay cables must temporarily be sustained by the steel girder,
as concrete slab strength still does not contribute (Cremer 1990). Future deck replacement is thus
also made possible.
As in conventional multi-girder composite bridges, concrete decks of composite cable-
stayed bridges act not only as a continuous surface for vehicles but also as a part of the girders for
live load and additional dead load that are applied on the composite section. However, the normal
forces in the deck from stay cables that do not exist in conventional bridges are important for
design of cable-stayed bridges. Permanent loads as well as vehicles applied on the deck change
tension of stay cables, which result in change of normal axial forces in the girders. Including com-
posite action of girders for supporting permanent loads, however, may be suspicious, especially
31
when bridges are constructed following the above described construction procedures, because
stay cables are tensioned for non-composite sections. Specifically, this aspect has been studied
during the preparation of design of the bridge at Hoogly, which was one of the early major appli-
cations of composite deck in cable-stayed bridges. The delicate key problem of ascertaining to
what extent the participation of the concrete in resisting normal forces would be reduced by
shrinkage and creep was the subject of long debate. It was finally stipulated that this normal force
should be considered to act only on the steel, calling for stiffening of the main beams (Walther et
al.1988). However, composite action is definitely developed for live loads. Not only bending
moment but also axial normal force are developed in girders under live loads.
Generally, as with a conventional structure, the analysis of a cable stayed bridge consists
of several stages. In the first stage, preliminary sizes of the deck system, pylons, and stay cables
are determined to check the feasibility of the work and to estimate the quantities required. In the
second stage, final calculations are prepared, determining the strength and deformations based on
the final dimensions. This stage requires much more detailed calculations. It is then generally nec-
essary to take into account second order effects as well as material non-linearity in the service
limit state and in the ultimate limit state (Walther et al. 1988). To evaluate final stresses and
deflections, it is necessary to consider all the construction procedures, time-dependent effects, and
nonlinear effects that come from beam-column behavior of girders, cable sag, the large deflection
of superstructure, and creep/shrinkage. In particular, consideration of construction steps is impor-
tant, because almost each cable-stayed bridge is built following unique construction steps. For
conventional multi-girder composite girder bridges, the construction steps are much more stan-
dardized, and changes of stress for each step are reasonably predictable. Additionally, nonlinear
32
characteristics of cable-stayed bridges can change their stress distribution of members for differ-
ent sequences of construction procedures.
The main reason for using effective flange width in design of composite girders in cable-
stayed bridges is for evaluating stresses, deflections, and capacity based on simplified design
methodologies, such as 2-dimensional plane analyses. This model can be used in the second stage
of design. However, assuming the existence of full 3-dimensional analyses methodologies with
consideration of non-linear behavior for the final design, the very notion of effective flange width
may not be sufficiently rigorous. Also, results from refined analyses for nonlinear effects must be
used according to the AASHTO LRFD Specifications [S.4.6.3.7]. Therefore, the following dis-
cussion is restricted to situations that require definition of effective flange width ignoring exist-
ence of refined analyses.
As discussed before, composite action can be expected for both bending and axial forces
due to live loads. Therefore, not only the conventional definition of effective width for bending
moment, but also definition of effective width for axial normal force are required for design of
composite cable-stayed bridges. As such, the use of this method might be restricted to calculation
of stresses of girders, and not be applicable for deflection and capacity. Additionally, it is desir-
able to use a somewhat conservative value of effective width for both bending moment and axial
forces in the simplified models in order not to deliver what a more refined analysis would deter-
mine to be an unsafe design.
Literature review reveals two methods for evaluation of effective width for axial normal
forces that have been developed. Effective flange width for axial normal forces is defined and
expressed by effective flange width for bending based on equilibrium on the cross section in one
33
of them (Cai, Zhang, and Nie 1998). The other used finite element analyses for several cable-
stayed bridges and delivered combined effective width (Byers 1999). Even though using com-
bined effective flange width may be simple, it can be undesirable in designing bridges. Overstress
of girder section cannot indicate which force, i.e. moment or axial force, is dominant for revision
of design. Also girders behave as beam-columns with axial compressive forces, which require
consideration of P-M interaction. Therefore, separating bending from axial force is necessary.
Thus, separating effective width for bending from effective width for axial force is also required.
For the bending effective width, it is thought that the same specification of effective flange
width for multi-girder bridges can be used with minor modification for bending effective width of
cable-stayed bridges, if the specification for multi-girder bridges is developed for such a wide
girder spacing, as is used for composite cable-stayed bridges. The modification needed is the def-
inition of effective span length. For conventional continuous composite bridges, the effective span
length is defined as the distance between inflection points under permanent loads. So with proper
definition of effective span length for a cable-stayed bridge, the same specification can be applied
for evaluation of bending effective width (Kim and Park 2001). As for the axial effective width,
more research is required to define general specifications. In related research, studies on post-ten-
sioning of composite bridges can be utilized to investigate the transverse distribution of axial
forces (Dunker el al. 1990; Dunker, Klaiber, and Sanders 1986).
34
Table 2-1: Effective Width of Simply-Supported Bridge Example
Specification
Interior
Girder(mm)
Exterior
Girder(mm)
Remark
AASHTO 3227 2903 Ignore barrier contribution
BS 5400 3835 3000 at the midspan
Canadian 3900 3240
J apan 3900 3240
Eurocode 4 3900 3240 (J ohnson and Anderson 1993)
Table 2-2: Effective Width for a Continuous Bridge Example (mm)
Location AB C DE
Girder Interior Exterior Interior Exterior Interior Exterior
AASHTO 2440 2210 2440 2210 2440 2210
BS 5400 2395 2025 2063 1761 2381 2017
Canadian 2440 2210 2440 2210 2440 2210
J apan 2436 2208 1959 1816 2408 2194
Eurocode 4 2440 2210 2440 2210 2440 2210
35
Table 2-3: Selected Cable-stayed bridges with composite main girders
Ref
No.
Name
Location
Completed
Main Span
Deck Width
Deck Depth
Slab Type
Post-
Tension
Tower Type
Cable Planes
1 Stromsund Bridge
Sweden (not composite)
1995
183 m
14.30 m
3.00 m
CIP, 0.20 m 2 H inclined
2
2 Buchenauer Bridge
Bruchsal, Germany
1956
58.80 m
20.80 m
1.40 m
CIP, 0.25 m
longitudinal
2 H w/o struts
2
3 Pont des lles
Expo 67, Canada
1967
2 x 105 m
28.65 m
2.82 m
CIP, 0.19 m
none
1 H w/ struts
2
4 2nd Hooghly River Br.
India
(1980) 1992
457 m
25.00 m
2.33 m
CIP, 0.23 m
none
2 H inclined
2
5 Sitka Harbor Bridge
Alaska, USA
1972
157 m
11.00 m
1.80 m
CIP, 0.20 m
none
2 H w/o struts
2
6 Heer-Agimont Bridge
Belgium
1975
123 m
14.50 m
2.05 m
CIP, 0.19 m
none
2 H w/ struts
2
7 Steyregger Donau Br.
Linz, Austria
1979
161.2 m
24.86 m
4.07 m
CIP, 0.20 m
longitudinal
2 A, unsym.
2
8 Sunshine skyway Br.
Florida, USA: Design
(1982)
366 m
27.50m
2.34 m
PC, 0.23 m
none
2 Diamond
2
9 Annacis Bridge
Vancouver, Canada
1986
465 m
28.00 m
2.10 m
PC, 0.27 m
none
2 H inclined
2
10 Saint Maurice
Switzerland
1986
105 m
2 x 11.75 m
1.01 m
CIP, 0.22 m
none
1 A incl. longit.
2
11 Quincy Bridge
Mississippi River, USA
1987
274 m
13.80 m
2.10 m
PC, 0.23 m
longitudinal
2 H inclined
2
12 Kemjoki Bridge
Finland
1989
126 m
25.50 m CIP, varies
transverse
1 center mast
2
13 Weirton-Steubenville
Br.
Ohio River, USA
1990
250 m
28.00 m
2.74 m
CIP, 0.22 m
none
2 A, unsym.
2
14 Nan Pu Bridge
Shanghai, Chaina
1991
423 m
25.00 m
2.10 m
PC, 0.26 m
longitudinal
2 H inclined
2
36
15 Burlinton Bridge
Ohio River, USA
1993
195 m
25.70 m
1.85 m
PC, 0.25 m
longitudinal
2 H inclined
2
16 Tahtiniemi (Heinola) Br.
Finland
1993
165 m
22.00 m
3.20 m
CIP, varies
transverse
2 H, 1 strut
2
17 Utsjoki Bridge
Finland
1993
155 m
12.00 m
1.76 m
PC, 0.26 m
none
2 H, inclined
2
18 Kamali River Bridge
Nepal
1993
325 m
11.30 m
3.00 m
PC, 0.23 m
none
1 H, unsym.
2
19 Mezcala Bridge
Mexico
1993
300/311 m
18.10 m
2.79 m
CIP, 0.20 m
none
3 H towers
2
20 El Canon
Mexico
1993
166 m
21.00 m
2.11 m
0.20 m 1 H inclined
2
21 El Zapote
Mexico
1993
176 m
21.00 m
2.11 m
0.20 m 1 H inclined
2
22 Yang Pu Bridge
Shanghai, China
1993
602 m
32.50 m
3.00 m
PC, 0.26-
0.40 m
long.+trans
2 inverted Y
2
23 Clark Bridge
Mississippi River, USA
1994
230 m
30.50 m
1.90 m
PC, 0.27 m
longitudinal
2 center masts
2
24 Baytown Bridge
Texas, USA
1995
381 m
2 x 23.83 m
1.83 m
PC, 0.20 m
none
2 twin diamond
4
25 2nd Severn Bridge
UK
1996
456 m
34.60 m
2.70 m
PC, onto
grid, 0.25 m
none
2 H w/ 2 struts
2
26 Kap Shui Mun Br.
Hong Kong
1996
430 m
35.20 m
7.46 m
PC, onto
grid, 0.25 m
none
2 H inclined
2
27 Ting Kau Bridge
Hong Kong
1997
448/475 m
43.00 m
1.75 m
PC, 0.23 m
none
3 center masts
4
Table 2-3: Selected Cable-stayed bridges with composite main girders
Ref
No.
Name
Location
Completed
Main Span
Deck Width
Deck Depth
Slab Type
Post-
Tension
Tower Type
Cable Planes
37
Figure 2-1: Two-Plate Girder Bridge and Its Evolution in Switzerland (Lebet 1990)
38
Figure 2-2: Example of a Simply Supported Bridge
39
Figure 2-3: Example of a Continuous Bridge
40
Figure 2-4: Plan and Elevation of I-78 Highway Bridge (Yen et al. 1995)
Figure 2-5: Cross-Section View of I-78 Highway Bridge (Yen et al. 1995)
41
Figure 2-6: Example of Plate Girder and Deck Slab Strain Gages (Yen et al. 1995)
42
Figure 2-7: Example of Strain Gage Locations on Deck Slab (Yen et al. 1995)
43
Figure 2-8: Open Type Cross Section (Haensel 1997)
Figure 2-9: Weserbridge at Porta-Westfalica (Haensel 1997)
44
Figure 2-10: Kalkgrabenbridge at Rudersdorf (Haensel 1997)
45
Figure 2-11: Domitz Bridge (Hanswille 1997)
46
Figure 2-12: Tied-Arch Bridge crossing the creek Mulde (Kuhlmann 1997)
Figure 2-13: Tied-Arch Bridge in the motorway A96 (Kuhlmann 1997)
47
Figure 2-14: Highway-Railway Bridge at Ciudad Guayna in Venezuela (Saul 1997)
48
Figure 2-15: Railway Bridge across River Main at Nantenbach in Germany (Saul 1997)
Figure 2-16: Road Bridge across River Elbe at Torgau in Germany (Saul 1997)
49
Figure 2-17: Types of Connection of Stay Cables to Deck Systems (Cremer 1990)
Figure 2-18: The Hoogly River bridge in India (Cremer 1990)
50
Figure 2-19: The Annacys bridge in Canada (Cremer 1990)
51
CHAPTER 3
INTERPRETATION, APPRAISAL AND APPLICATIONS
3.1 Introduction
The findings reported in Chapter 2 lead to the following insights and implications for con-
duct of the remainder of the work in the NCHRP 12-58 project:
There are some key assumptions underlying the use of the notion of effective slab width that
should be kept in mind by bridge engineers,
The 12t limitation that governs effective slab width in wide girder spacing designs is archaic and
has little to defend its continuing retention in AASHTO,
Several distinct philosophies are evident in international effective slab width formulations that
merit careful consideration, and
The paucity of suitable existing bridge field experimental results points to the need for further
field and laboratory experimental work focused explicitly on generating data needed to validate
companion FEM studies investigating effective slab width.
3.2 Key Assumptions
Some key, almost paradoxical, assumptions underlie the use of the notion of effective slab
width, e.g.,
1.Even with the sophisticated computer-aided analysis available for bridge design in the 21
st
cen-
tury, traditional line-girder analysis provides an ongoing useful context for analyzing steel gird-
ers acting compositely with concrete decks.
52
2.Since the context of effective width based analyses is single member line girder analysis, effec-
tive width based analyses are not appropriate for use in situations where a line girder analysis is
insufficient and a system analysis is thus considered necessary.
3.2.1 Some Implications of Line Girder Analysis Limitations
Beyond line-girder analysis, system analysis is generally considered necessary in, e.g., the
following kinds of situations:
Highly skewed and Curved girder bridge analysis and design,
After - fracture redundancy and load redistribution analysis where required [e.g., Daniels et al.
89], and
Detailed design stages for systems where second-order effects can be important, e.g., cable-
stayed bridges.
Since a simplistic line-girder analysis is insufficient for such situations, the use of ideali-
zations that inherently assume line-girder analysis (like effective slab width) may be considered
questionable at best in these kinds of situations.
3.2.2 Some Implications of Wide Girder Spacing
Since the challenge to the 12t limitation arises, for typical deck thicknesses, in the context
of girder spacings wider than 3 m (10) or so, additional implications of wide girder spacings are
also of interest. Some of these implications are listed as follows:
1.The empirical method of deck design is currently prohibited by AASHTO for use beyond a
girder spacing of 4.1 m (13.5 ft.) [AASHTO LRFD S9.7.2.4]. Thus, methods of traditional
design, prestressed design, and system analysis of decks that go beyond line girder analysis
(e.g., grillage, finite strip, etc.) must be used to design the actual decks. Since these methods
53
typically go beyond line girder analysis, some may ask why it should still be permissible to use
line girder analysis for the in plane analysis of the composite girders supporting the resulting
decks.
2.The use of the line-girder oriented distribution factor formulas is currently prohibited by
AASHTO for use beyond a girder spacing of 4.9 m (16.0) [AASHTO LRFD S4.6.2.2].
3.The possible interaction of plate bending with in-plane effective width behavior increases.
Investigating the significance of this possible interaction needs to be a concern of the Task 4
analytical studies planned for this project.
4.Longitudinal shear forces that get funneled into the shear connectors increase. Whether the
current AASHTO shear stud design criteria still apply, in HPS and HPC composite combina-
tions, may need to be revisited.
Some of these implications of the use of wider girder spacings in conjunction with high-
performance steels and concretes point to the need to re-examine existing AASHTO design crite-
ria and the limits of their applicability (e.g., shear connector design criteria, limits of applicability
of empirical deck design methods and transverse load distribution factors for line girder analyses).
For the purposes of the present effective slab width investigation, it is assumed that present
AASHTO criteria for such concerns apply, including skew corrections.
3.3 The 12t Limitation
The historical review presented in Chapter 2 and Appendix C indicates that the 12t limita-
tion in the AASHTO effective width formulation (AASHTO LRFD S4.6.2.6, AASHTO Standard
Specs 10.38.3.1) has been in AASHTO (then AASHO) since the 1940s, the early days of com-
posite beams, and is based on empirical research published in the World War I era. That research
54
was for reinforced concrete T-beams, not steel beams with composite concrete decks. Almost all
building and bridge codes for steel-concrete composite beam members internationally in the last
few decades have departed from any kind of thickness limitation in their effective slab width for-
mulations.
That is not to advocate a blind copy-cat approach. The background for the various inter-
national code provisions typically consists primarily of parametric analytical work which most
recently is finite element based. Little companion experimental work is cited by these codes to
accompany the analytical background.
The limited test data points to the desirability of experimental verification of analytical
results.
3.4 Philosophies Behind Effective Width Formulations
The review of design criteria presented in Section 2.3 (and augmented in Appendix D)
brought to light several distinct philosophies underlying the various effective width code formula-
tions being used internationally, ranging from simple (e.g., Canada, which presumes line-girder
analysis) to relatively complex (e.g., the British BS 5400, which does not and which also distin-
guishes between point loads and distributed loads). As one might expect, there is an inevitable
trade-off between simplicity and accuracy especially when the full spectrum of possibilities
must be accommodated even within the context of line-girder analysis (e.g., interior and exterior
girders, positive and negative moment regions, linear and nonlinear realms of behavior, box and I
girders, and absence or presence of axial load, the latter being the case for cable-stayed and tied-
arch structures).
55
Implications for the possible results to arise from the remainder of the present project
appear to be as follows:
Distinct development of dual Method A and Method B approaches, where one is simplistic
and conservative while the other is more complicated and more precise, or
A single straightforward approach to cover not only the common cases but also the cases where
owners wish to push the envelope of large girder spacings.
Ideally, a dual Method A and Method B approach is undesirable in the interest of sim-
plicity. It remains to be seen, however, based on the outcomes of Tasks 4 6, whether a single
simple method will accomplish project objectives. Such a single simple method would presum-
ably be based on effective width at the service limit state (J ohnson and Anderson 1993). Although
the research team understands the desirability of simplistic criteria for effective width, it also con-
siders as of first importance not to prejudice the investigation as it embarks on Task 4. Most inter-
national codes also incorporate consideration of the ratio of adjacent span lengths in their b
eff
criteria. This kind of consideration should also be adoptable within AASHTO approaches.
3.5 Experimental Verification
3.5.1 Field Experimental Results
Although there is a considerable even vast body of research results pertaining to field
experiments on bridges, very few if any of these results appear to be of much use to the present
investigation. Results from the nationwide survey indicated that no known experiments have been
conducted under state sponsorship explicitly to investigate effective slab width in composite steel
bridge members. This means that the field experiments have been conducted for other reasons.
These other reasons typically include the following:
56
Load rating of older bridges (which are typically not on wide girder spacings!) that have insuffi-
cient posting capacity based on traditional line-girder based rating methods,
Quantifying differences (and reasons for those differences) between idealized notions of bridge
behavior and actual behavior (e.g., unintended composite action, lateral resistance provided by
roller bearings, etc.), and
Re-examining other AASHTO analysis provisions that researchers seemingly cannot bring
themselves to leave well enough alone, e.g.,
- The traditional or Zokaie-Imbsen transverse load distribution factors,
- The dynamic load allowance (IM factor) to account for vibratory effects of live load, and
- The traditional L/800 live load deflection limit.
Many of these that do not appear to have direct applicability to the present effective slab
width investigation are not included in the reference list or the annotated bibliography in Appen-
dix G, for brevity.
One of the very few bridge field test results that have been found for bridges built with
wide girder spacings concerns bridges carrying Interstate Route 15 recently built in Salt Lake City
[Iaquinta 2001] that were instrumented in order to quantify in-field fatigue stress ranges [Frank
2001]. Unfortunately,
No strain gauges were placed on the concrete deck that could be used to measure shear lag
effects, and insufficient instrumentation was placed on cross sections to infer neutral axis
heights, and
The negative moment region was designed noncomposite.
57
Thus, results regarding effective slab width are not inferable for either the positive or neg-
ative moment regions from this study.
No results at all could be found that have all of the following desiderata.
Girder spacing sufficiently long and wide to cause the 12t limitation to control effective width
by a significant margin,
Multiple strain gauges on the deck on each of several different cross sections, or at least multiple
strain gauges on each of many cross sections sufficient to infer neutral axis location,
Composite negative moment regions as well as composite positive moment regions instru-
mented as described above,
Deployment of HPC in the deck.
The one that appears to come the closest is the report of field experiments of the I-78
bridge over the Delaware River [Yen et al.1995]. For those experiments, unfortunately, no data
beyond what appears in the printed report still exists. Those results could be used for FEM valida-
tion/verification studies, but they fall short of what would be desired.
In particular, the principal shortcoming of field experiments, even if they satisfy the fore-
going list of desiderata, is that they very seldom afford the opportunity of destructive loading to
collapse. As such, the nonlinear range of material behavior relevant to assessment of effective
width for the Strength I limit state cannot be investigated in field studies. Most destructive bridge
experiments historically have been on structures for which the girder spacings are not wide
enough for the 12t limitation to govern effective width. Interestingly, one of the bridges in the
well-known Burdette and Goodpasture destructive tests (1972) had a girder spacing wide enough
58
for the 12t limitation to govern effective width, but the reported instrumentation is insufficient to
make inferences about experimental effective width.
Thus, companion laboratory experiments on widely spaced girders exercised well into the
nonlinear range are needed in a thorough protocol. But no such experimental results have been
found.
3.5.2 Summary
Firstly, the 12t limitation need not be preserved. Secondly, international precedents pro-
vide a good indicator of possible alternate formulations for effective slab width criteria, although
cultural and philosophical differences prevent blind wholesale application of their formulations to
US-based practice without further investigation. Thirdly, regarding that further investigation, if it
is desired to have experimental confirmation for FEM based analytical predictions for the limit
states and girder spacings of interest, there is a need for all three of the following:
properly instrumented field experiments, and
properly correlated laboratory experiments, in order to provide principled verification of
finite element model based parametric studies in both the linear and nonlinear ranges of material
behavior.
The first two of these three (the field and laboratory experiments) were not included in the
original scope of work of the NCHRP 12-58 project and are described further in Chapter 4 and
Appendix E.
59
CHAPTER 4
CONCLUSIONS AND SUGGESTED RESEARCH
4.1 CONCLUSIONS
Findings from analytical and numerical studies and from laboratory and experimental
investigations indicate that the slab thickness limitation in the formulation of effective slab width
should be removed and replaced with more rational criteria. The current Canadian formulation of
effective slab width is suggested as a starting point for what that alternate set of rational criteria
might consist of. That formulation is presented in Section 2.3 and Appendix D. Further investiga-
tion using finite element based parametric studies accompanied by proposed additional laboratory
and field experimentation is described in the next section.
In addition, related topics considered to be beyond the scope of the NCHRP 12-58 project
merit re-examination in the context of envisioned construction practices utilizing the wider girder
spacings that can be expected if current trends continue. These topics include the following:
Axial effective width notions (as in arch ties and cable-stayed bridge floor systems) that are dis-
tinct from flexural effective width notions,
Girder spacing limits on the use of the empirical deck design method,
Girder spacing limits on the use of the line-girder oriented distribution factor formulas, and
Shear connector design criteria in HPS/HPC composite systems that did not even exist when
those criteria were developed over 30 years ago.
60
4.2 SUGGESTED RESEARCH
4.2.1 Analytical Investigations
Finite Strip Method
Bridge decks which have a constant cross-section along the entire span length can be ana-
lyzed with a simple type of finite element called a finite strip. The method is very similar to folded
plate analysis. The structure is made up of finite elements called strips which extend from one
end of the deck to the other. The displacement functions for in-plane and out-of-plane displace-
ment of the strips are defined as: -
(4-1)
where x is the longitudinal direction along the structure and y is the transverse direction
across the strip. Harmonic analysis can be greatly simplified if the deck is assumed to have right
end supports with diaphragms to prevent any end movements in the plane of the diaphragms. Har-
monic components can be summed to give the total stress distribution. In the finite strip method,
the transverse functions f(y) are assumed to be simple polynomials so that in effect the method is
an approximation to the rigorous elastic folded plate method in which these functions have a
complicated hyperbolic form.
Computer programs employing harmonic analysis can consider only a limited number of
harmonics, for instance ten or one hundred, and ignore the higher harmonics. If the sum of the
considered harmonics of load or shear force, etc., is plotted in the region of a discontinuity, the
result violently oscillates close to the discontinuity. Increasing the number of considered harmon-
w u or v , , f y ( )
nx
L
---------
\
| |
sin

=
61
ics moves the oscillation closer to the discontinuity, but the amplitude is not reduced. At the limit
the oscillation still exists but is infinitely narrow. This characteristic of harmonic analysis, called
Gibbs phenomenon, can lead to significant errors in the output from finite strip computer pro-
grams in the region to each side of a discontinuity within two or three wavelengths of the highest
harmonic considered. One result of this approximation is that calculated stresses are discontinu-
ous transversely at strip interfaces.
Finite Difference Method
A classical approach to finding a numerical solution to the governing equations of a math-
ematical continuum model is to use finite differences. In a finite difference solution, the deriva-
tives in the differential equations of equilibrium are replaced by finite difference approximations,
and the differential and variational formulations of mathematical models can be solved. It is nec-
essary to approximate by finite differences and impose on the coefficient matrix both the essential
and the natural boundary conditions. The finite difference method can also be used to generate
stiffness matrices. In some cases the resulting equations obtained in a Ritz analysis and in a finite
difference solution are identical or almost identical.
However, for complex geometries the imposition of the natural boundary conditions can
be difficult to achieve since the topology of the finite difference mesh restricts the form of differ-
encing that can be carried out, and it may be difficult to obtain a symmetric coefficient matrix in a
rigorous manner.
The difficulties associated with the use of the differential formulations give rise to the
development of finite difference analysis procedures based on the principle of minimum total
62
potential energy, called finite difference energy method. The displacement derivatives in the
total potential energy, , of the system are approximated by finite differences. The minimum
condition of is used to calculated the unknown displacement at the finite difference stations.
Furthermore, a symmetric coefficient matrix is always obtained. An advantage of the finite differ-
ence energy method lies in the effectiveness with which the coefficient matrix of the algebraic
equations can be generated.
Finite Element Method
The finite element method is a technique for analyzing complicated structures by notion-
ally cutting up the continuum of the prototype into a number of small elements which are con-
nected at discrete joints called nodes. For each element, approximate stiffness equations are
derived relating the displacements of the nodes to the node forces between elements. A computer
is used to solve the very large number of simultaneous equations that relate node forces and dis-
placements. Since the basic principle of subdivision of the structure into simple elements can be
applied to structures of all forms and complexity, there is no logical limit to the type of structure
that can be analyzed if the computer software is written in the appropriate form. Therefore, finite
elements provide the most versatile method of analysis available at present, and for some struc-
tures the only practical method. However, the quantity of computation can be enormous and
expensive so that often the cost cannot be justified for this type of analysis. Furthermore, the
numerous different theoretical formulations of element stiffness characteristics all require approx-
imations which in different ways affect the accuracy and applicability of the method. Conse-
quently, finite element method will be employed in determining the effective slab width analysis
using ABAQUS. Application of the method to slab-on-girder bridge modeling is described next.

63
4.2.1.1 Finite Element Analysis
4.2.1.1.1 Geometric Modeling
Composite sections are made up of the reinforced concrete slab supported by a number of
steel girders in either longitudinal or transverse directions. The reinforced concrete deck is con-
nected to the steel girders by means of shear connectors to ensure the presence of fully or partially
composite action. The mechanical phenomena governing the behavior of composite steel-con-
crete girders can be analyzed using the finite element method (ABAQUS). In this section, proce-
dures for modeling bridge components, by MSC Patran, to represent the prototype will be
discussed. Several assumptions associated with the modeling method will be addressed.
Steel Girders
Generally, steel girders for bridges are constructed from I-sections of hot rolled sections or
welded plate girders, which entirely consist of flanges and web. There are several different ways
to model steel girders in finite element analysis depending on the degree of structural complexity.
The simplest way is to represent the steel girders by 1-dimensional beam elements. Girder dimen-
sions and sectional properties, i.e. cross-sectional area, second moment of inertia in both direc-
tions, torsional stiffnesses, etc., must be computed for stiffness matrix formulations. Effects of
warping restraint and girder depth may not be captured by the model.
A second method is to model the steel girders as the composition of 2-dimensional shell
elements for both flanges and web. Only the width and thickness of the shell need to be specified.
These shell elements may allow transverse shear deformation. Thick shell theory can be used as
the shell thickness increases. However, discrete Kirchhoff thin shell elements suffice as the thick-
64
ness decreases. The transverse shear deformation also becomes very small as the shell thickness
decreases.
Finally, 3-dimensional solid elements could be used to model the entire steel girder. The
reduced-integration, second-order elements are recommended for linear and smooth nonlinear
problems. However, using 3-dimensional solid elements would lengthen the analysis time which
can become very expensive. Hence, using 2-dimensional shell elements is typically preferable.
Depending on model verification, a decision will favor the element type that best match the exper-
imental results. Even though, solid elements produce larger model, this use in the model is the
most likely choice.
Concrete Deck
In United States, the thickness of concrete deck in most bridge structures lies in a range
between 159 mm (6 1/4-in.) and 197 mm (7 3/4-in.). For example, New York State uses deck
thickness of 240 mm (9.5), which is reasonably large compared to the thickness of steel girder
flanges and web. The AASHTO LRFD code limits deck thickness to a minimum of 175 mm (7).
Biggs et al. (2000) employed the four-node reduced-integration shell elements to model the rein-
forced concrete deck for a three-span continuous steel-concrete bridge. Cracking and nonlinear
behavior were incorporated into the analysis.
Three-dimensional solid elements can also be employed to represent the concrete deck.
Reinforcing steel can be modeled as layers of smeared solid elements in the deck. The advantage
of using solid elements for the concrete deck is that the longitudinal strain distributions can be
easily captured. Consequently, the longitudinal stress distributions can be plotted across the width
of the slab. From these stress distributions, it is possible for effective slab width to be determined.
65
Another advantage of employing solid elements is when concrete strains at the top and bottom of
the deck are to be compared with the measured strains from experiments during the model valida-
tion process.
Shear Connections
The specific behavior of the steel and concrete composite girders is related to the mechan-
ical properties of the interface between the steel girder and the concrete slab. Flexural members
generally require large transfer of forces, so that discrete mechanical connectors are the most suit-
able devices. Ultimate strength of shear connectors is usually obtained by push-out tests. How-
ever, the mechanical behavior and the modes of failure of shear connectors are beyond the scope
of this research. Therefore, the assumption of fully composite action in the flexural members will
be made for this purposes of this research. Current AASHTO LRFD shear connector design crite-
ria will be assumed to be valid, even though some parameter combinations investigated will push
the envelope beyond current bounds of AASHTO-based practice.
In the case of using beam or truss elements as the steel girder, the connection between the
concrete deck and steel girders can be modeled as a rigid-link elements. Both components are
assumed to have the same rotation and curvature when subjected to bending. When the steel gird-
ers are modeled as shell elements or solid elements, the deck can be attached to the top of girders
by means of a thin shell element with a very high stiffness.
Boundary Conditions
Linear constraints or boundary conditions are of practical significance. They may not only
increase the computational efficiency, but also eliminate potential numerical difficulties. Bound-
ary conditions that represent structural supports specify values of displacement and rotation vari-
66
ables at appropriate nodes. To facilitate a more economical solution, finite element meshes may
also use symmetry, which can be implemented with symmetric boundary conditions. Linear
springs may also be considered supports where a component rests on a bearing pad.
It was shown by Kostem et al. (1988) that if the longitudinal restraints at the supports can-
not be accurately quantified, then peak stresses and deflections under consideration can be off by
100%. It must be noted that if the supports on one end of the bridge are pins, with rollers at
the other end, then due to differential longitudinal elongation of the bottom flanges of the girders,
forces of varying magnitude will be applied a the support points.
For a simply-supported bridge in this report, at the pin support, the displacements are
restrained in the x-,y-, and z-directions, and rotations are restrained about the y- and z-axes at the
pinned connection. The girders are allowed to rotate only about the x-axis (transverse axis). At the
roller connection, zero displacement is imposed on the x- and z-degrees of freedom along with
zero rotation about the y- and z-axes (vertical axis).
Bearing restraining forces from the end bearings were examined by Fu and DeWolf
(2001). In their analysis, the finite element model was verified against the experimental results
and modified to account for changes in the vibrational behavior due to the eccentrically applied
bearing forces, which occurred when the bearings were partially restrained in colder weather.
However, these effects could be minimized in laboratory conditions.
67
4.2.1.2 Material Models
Reinforcing and Structural Steel
For reinforcing steel the Ramberg and Osgood formulation will be employed. The analyti-
cal expression is given by
(4-2)
where E
sc
=longitudinal modulus of elasticity of the steel, whereas the parameters n and B
are: -
(4-3)
where f
t,sc
=tensile strength; f
y,sc
=yield stress; and
u,sc
is the ultimate strain of reinforce-
ment.
For structural steel, many stress-strain relationships can be introduced. A simple elastic-
plastic model with strain hardening will be used. It depends on the yield stress, the ultimate stress,
the limit strain of the plastic plateau and the ultimate strain. It is assumed that steel shows the
same mechanical properties both in compression and in tension. This type of modeling is suitable
to fit the mechanical behavior of so-called compact steel sections, which fully develop the plastic
bending moment. Lateral-torsional buckling phenomena are not considered and must be avoided.

sc

sc
E
sc
-------

sc
B
-------
\
| |
n
+ =
n
ln
u sc ,
f
t sc ,
E
sc
( ) 0.002 ( )
ln f
t sc ,
f
y sc ,
( )
----------------------------------------------------------------------- B ;
f
y sc ,
0.002
1 n ( )
------------------------ = =
68
Concrete
The nonlinear response of reinforced concrete material requires the development of accu-
rate and computationally efficient models for the analysis of concrete slabs under the ultimate
loads. The computational complexity and cost of finite element models results in high demands
on knowledge about numerical methods. However, these difficulties will be overcome as more
acceptable constitutive material models are developed.
An important feature of a concrete constitutive model, in addition to the essential require-
ment of accurately representing the actual mechanical behavior, is the clarity of formulation and
the efficient implementation in robust and stable nonlinear algorithms. In the finite element analy-
sis used in this research, a classical 3-D plasticity will be used in the concrete model called
ANAMAT. This concrete model has been used to represent the concrete material by Hose (2001).
Brittle concrete cracking under low tensile stresses is represented by the failure surface.
This implies isotropic behavior for all stress states within the region enclosed by the failure sur-
face (see Figure 4-1). Unlike metal yield surfaces, this failure surface is not changed upon reload-
ing. Once a failure stress state is reached, the failure surface immediately collapses to a lower
surface. In the ANAMAT concrete model, the existence of the yield function is expressed by
(4-4)
where is the plastic strain and K is a material parameter. The plastic strain increment
must obey a flow rule in the form of Equation (4-5).
F
ij

ij
p
K , , ( ) 0 =

ij
p
69
(4-5)
where d is a positive scalar and is a function of the state of stress, strain, temperature and
work hardening. For hardening material, the Drucker stability postulate in Equation (4-6) must be
satisfied.
(4-6)
where:
=stress state on current yield surface
=previous stress state lying inside or on the yield surface
=incremental stress state as the stress state changes from to .
A specific form of work hardening is introduced to help simplify the incremental stress-
strain law. A Mohr-Coulomb yield condition, which is a modified form of the Drucker-Prager
yield function, is used with the Prager kinematic hardening rule extended to include isotropic
hardening on generally nonlinear stress-strain curves. The yield surface is assumed to translate in
the direction of the plastic strain increments. The final 3-dimensional plasticity equations for the
ANAMAT concrete model consists of Equation (4-7) to Equation (4-10).
(4-7)
d
ij
p
d

ij

F
\
| |
=

ij
'
ij
( )d
ij
p
0 or d
ij
d
ij
p
0

ij
'
ij
d
ij

ij
'
ij

n
H
n

n
f
( ) 2
KK
Q
------------
n
2J
I
I + ( ) + =
70
(4-8)
(4-9)
(4-10)
where:
=incremental strain vector
=incremental stress vector
=free expansion incremental strain vector
=deviatoric stress vector,
I =identity matrix
D =elasticity matrix,
E =Youngs modulus
=Poissons ratio
H M

Q
----
\
| |
M
n
2J
I
I + ( )
n
2J
I
I + ( )
T
M =
Q
n
2J
I
I + ( )
T
M
n
2J
I
I + ( )
2
3
---E'
n
2J
I
I + ( )
T

n
2J
I
I + ( ) + =
M I DC + ( )
1
D =

n
f

n

ij
'
ij
1
3
---'
kk

ij
=
D
ijkl
E
1 +
------------
ik

jl
E
1 + ( ) 1 2 ( )
--------------------------------------
ij

kl
+ =
71
=Kronecker delta, where repeated index implies summation within context of indicial
notation
=change in compliance matrix C due to age and temperature
=constant which governs the plastic material volume change
J
I
=first invariant of the stress tensor,
=0 when inside the yield surface
=1 when outside the yield surface
=slope of stress-plastic strain curve in compression, which is a function of the plastic
strain and the history of deformation
K =material parameter
In the ANAMAT concrete model, a crack is a mechanism that transforms the material
behavior from isotropic to orthotropic, where the material stiffness normal to the crack surface
becomes zero while the full stiffness parallel to the crack is maintained. The cracks can follow
independent histories. In this smeared crack model, a smooth crack should close and all the mate-
rial stiffness in the normal direction to the crack is recovered. Strain-strain relationship of the
ANAMAT concrete model can be shown in Figure 4-2.
Since crack surfaces are typically rough and irregular, ANAMAT takes into consideration
the mechanism of shear transfer in cracked concrete by retaining a reduced shear modulus in the

ij
C
'
ij
E'
72
stress-strain matrix. Tension stiffening of cracked concrete, which is the ability of cracked con-
crete to share the tensile load with the reinforcement, is also considered in the ANAMAT concrete
model. The addition of tension stiffening to the smeared crack model improves the numerical sta-
bility of the solution.
Because the ANAMAT concrete constitutive model does not include confined concrete
properties, stress will decrease dramatically as soon as concrete starts crushing. This phenomenon
is very important for the computation of section capacity, especially at the positive moment
region.
The ANAMAT concrete model will be employed as the user material model via the
UMAT subroutine which is available in the general-purpose finite element program ABAQUS.
Concrete cylinder laboratory tests could be conducted to validate the actual behavior of the con-
crete model.
For analysis of high-strength concrete, it is important to use appropriate material proper-
ties, such as tensile strength and modulus of elasticity. In most cases, high elastic modulus for
High Performance Concrete members is often desirable to control serviceability. Therefore,
appropriate estimation of elastic modulus becomes more important. Myers and Yang (2001) dis-
cussed such issues, and summarized several models for elastic modulus evaluation. Such models
will be considered in the finite element modeling of systems with HPC decks.
73
4.2.1.3 Analysis and Results
4.2.1.3.1 Static Analysis
This section briefly describes the analysis procedures that will be used in the finite ele-
ment analysis. Linear and nonlinear analysis will be included in the discussion. Most finite ele-
ment analysis software is based on the direct stiffness method of analysis, which requires the
execution of the following steps.
1. Computation of the stiffness matrices for all elements in the structures.
2. Computation of statically equivalent nodal loads for any loads that may be distributed
over an element.
3. Transformation of element stiffness matrices and nodal load vectors to a common glo-
bal coordinate system.
4. Assemblage of structure stiffness matrix and load matrix which together constitute the
global equilibrium equations.
5. Application of boundary conditions and linear constraints.
6. Simultaneous solution of the equilibrium equations for all nodal displacements.
7. Computation of element stresses or internal forces and moments.
One of the important steps is to enforce both compatibility and equilibrium at all nodal
points of a structure. Compatibility requires that corresponding displacement components of all n-
elements joined at some node i are equal.
(4-11) u
i
u
i
1 ( )
u
i
2 ( )
u
i
m ( )
u
i
n ( )
= = = = = =
74
where , m =1,..., n are the corresponding displacement components of the elements
meeting at node i. Equilibrium requires that:
(4-12)
where is the externally applied load at node i, and
(4-13)
are corresponding nodal force components due to the deformations of the m-th element,
exerted on node i. Substitution of Equation (4-13) into Equation (4-12) gives with Equation (4-
11),
(4-14)
provided all are expressed in the same coordinate system. Equation (4-14) can also
be written as:
(4-15)
where [K] is the structure stiffness matrix, which is found by appropriately summing up all
element stiffnesses [k
(m)
]. {R} contains all load contributions from elements with locally applied
loads, which have been converted to equivalent nodal loads.
u
i
m ( )
R
i
R
i
1 ( )
R
i
2 ( )
R
i
m ( )
R
i
n ( )
+ + + + + =
R
i
R
i
m ( )
k
ij
m ( )
u
j
m ( )
m
j

1 n , , = =
R
i
k
ij
1 ( )
k
ij
2 ( )
k
ij
n ( )
+ + + ( )u
j
j

=
k
ij
m ( )
R ( ) k
m ( )
| ]
m

\
|
| |
u { } K | ] u { } = =
75
After the global equations of equilibrium in the form of Equation (4-15) have been assem-
bled with the appropriate boundary conditions and nonsingular stiffness matrix, a Gauss elimina-
tion scheme is normally used to solve the equilibrium equation for the n-displacements, {u}.
(4-16)
where [L] is a unit lower triangular matrix, and [D] a diagonal matrix.
The solution of the structure equilibrium equations provides values for all joint displace-
ments due to the applied loads. Stresses and internal forces throughout the structure can be com-
puted with the help of the element stiffness properties, once the nodal displacements for the
element are known.
(4-17)
where =stress in element m
=nodal strains of element m in local coordinate system
=stiffness matrix of element m in local coordinate system
Local strains can be derived from a common global coordinate by a transformation matrix.
(4-18)
where =coordinate transformation matrix for element m
u { } D | ] L | ]
T
( )
1
L | ]
1
R { } L | ] D | ] L | ]
T
( )
1
R { } = =
s
m ( )
{ } k
m ( )
| ] v
m ( )
{ } =
s
m ( )
{ }
v
m ( )
{ }
k
m ( )
{ }
v
m ( )
{ } a
m ( )
| ] B
m ( )
{ } u
m ( )
{ } =
a
m ( )
| ]
76
=strain-displacement coefficient in global coordinates
=element nodal displacements in global coordinates
The entire static analysis can be divided in to two major sections, i.e. linear and nonlinear
behaviors. Nonlinear static analysis involves the specification of load cases and appropriate
boundary conditions. However, the effects of any nonlinearity present in the model can be
included. The starting condition for each step is the ending condition from the last step with the
state of the model evolving throughout the history of the analysis steps as it responds to the his-
tory of loading. Sources of nonlinearity can be from material nonlinearity, geometric nonlinearity
and boundary nonlinearity. Each analysis step has its own step time, which begins at zero in each
step. It is possible to include the geometric nonlinearity effect from during large-displacement
stages. Most elements will then be formulated in the current configuration using current nodal
positions. Elements therefore distort from their original shapes as the deformation increases.
Additionally, for nonlinear analysis, ABAQUS offer a tool for automatically controlling
the increment size. ABAQUS uses Newtons method to solve the nonlinear equilibrium equa-
tions. If the increments are too large, more iterations will be required. Furthermore, Newtons
method has a finite radius of convergence, meaning that the solution will not be too far away from
the equilibrium state.
4.2.1.3.2 Finite Element Model Verifications
After conducting the finite element analysis for a specific type of bridge and loading con-
figuration, the analytical results will be compared with experimental results for model verifica-
tion. Such results are midspan displacement and longitudinal strain distributions. Model
B
m ( )
{ }
u
m ( )
{ }
77
validation might be required to ensure that the finite element model maintains a good representa-
tion of the actual bridge behaviors. In addition, convergence studies will be conducted to optimize
th use of element types in the model.
Many different factors could give rise to differences between the analytical and experi-
mental results. These include unintended composite actions, bearing restraint, actual dimensions
of bridge components, actual material properties, etc. It must be noted that strains are the second
derivative of the displacements which are very difficult to correctly capture in the field or labora-
tory tests. Displacements are preferable and reliable for the model verification stage. Hence, it is
recommended to install a sufficient number of LVDTs to measure the vertical displacements of
concrete deck at several locations. More detail discussion on the experimental instrumentations
appear in the experimental sections of this chapter and Appendix E.
In this project, experimental results will be obtained from both laboratory and field tests.
For laboratory test, four different test specimens will be tested for Service II and Strength I limit
states. Two half scaled and two one-quarter scaled specimen results will be utilized in finite ele-
ment model verification process. Examples of finite element models can be shown in Figure 4-3,
Figure 4-4 and Figure 4-5. Detailed discussion will be given in the next section.
Field testing is required to verify whether the test results from scaled models contain any
scale effects. Generally, field test results are required for finite-element model verification. How-
ever, it is impossible to achieve Strength I limit state in most real bridge testing. This is one of the
major reasons why the laboratory scaled model testing is necessary. Fortunately, one of few that
fits into the scope of this project is a steel plate girder bridge which carries the east bound traffic
on the I-78 highway over the Delaware River near Easton, Pennsylvania, as described earlier.
78
Descriptions of this bridge are shown in Table 4-3. Instrumentation locations will be further dis-
cussed in the following section.
4.2.1.4 Parametric studies
Once a well-developed finite element model is completed, the effects of the different
design parameters will be investigated via parametric studies. The parameters can be categorized
into major three groups, i.e. geometric properties, material properties and initial conditions. More
details will be presented in this section.
4.2.1.4.1 Matrix of Analysis Cases
Geometric Properties
Geometric properties represent a few dimensional parameters such as girder spacing (s),
span length (l) and slab thickness (t). By keeping an overall width of the bridge constant, the num-
ber of girders can be changed to give a different girder spacing. However, the slab thickness is
theoretically a function of the girder spacing. For simplicity of the analysis, slab thickness can be
kept constant at a minimum value in order to maximize shear lag phenomenon as long as it is a
feasible thickness.
Material Properties
Material property parameters such as concrete compressive strength (f

c
) and yield
strength of steel (f
y
) are expected to have some contribution to the effective slab width determina-
tion. Different material failure modes are expected between the normal strength concrete and the
high performance concrete (HPC).
79
Initial Conditions
Initial conditions of the applied loads prior to linear and nonlinear static analysis, for
instance axial forces due to the prestressing process and compressive forces from cable-stayed
bridges, will be considered as variables in the parametric study.
Base Model
The base model represents the simplest type of bridge which will be used for the purposes
of comparison in the parametric study. This model will be frequently referenced in the investiga-
tion of each design variables.
A prototype bridge, which will be used as the base model, has the geometrical properties,
material properties and initial conditions as shown in Table 4-1.
Table 4-2 gives an overview of the analysis cases that will be performed in the parametric
study of this research. Data reductions of the results from the parametric study will be done and
regression analysis will be conducted.
4.2.1.4.2 Effective Slab Width Determination Methodology
In the parametric study, the different cases specified in Table 4-2 will be analyzed by the
general-purpose finite element analysis software ABAQUS. In order to better discern the shear-
lag phenomenon, a more refined mesh will be applied in the region of flange-to-web intersection.
This would provide better results in terms of the longitudinal strains and stresses at each trans-
verse section, especially at the area of maximum deflection.
The first method of determining the effective slab width is to use the transverse distribu-
tion of longitudinal strains at slab surface. The effective slab width can be obtained by summing
80
up the area under the curve by numerical integration and dividing by the maximum strain ordinate
in that area. For a girder spacing of b in the y-direction, subjected to the longitudinal strain of
x
,
the effective slab width of b
eff
can be computed as shown in Equation (4-19):
(4-19)
where
x,max
is the maximum longitudinal strain at a specific flange-to-web junction.
The second method is to define the effective slab width from strain distribution in the
composite sections. When the position of the neutral axis is known, the effective slab widths can
then be obtained. Furthermore, the determined effective slab width can be used to compute the
second moment of inertia in order to reconfirm the analytical results with the value of the associ-
ated deflections.
The third method utilizes the measurement of moments and shear forces from experiments
to compute longitudinal stress profiles in the composite sections. Subsequently, internal couple
forces can be determined with associated effective slab width for both positive and negative
moment regions.
In order to obtain the accurate effective slab width using any of the three methods men-
tioned above, we should be aware that the deck plate bending could have some contributions to
the actual in-plane stress distribution and hence the effective slab width (b
eff
). To overcome this
problem, a couple of analyses can be conducted to study this effect by applying loads (1) at girder
lines and (2) between girder lines.
b
eff
2
x
y d
0
b 2

x max ,
------------------------- =
81
4.2.2 Experimental Investigations
4.2.2.1 Laboratory Test
The purpose of the experimental study is, firstly, to verify composite bridge finite element
modeling scheme at both service and strength limit states, and secondly, to understand more real-
istic behavior of composite bridges at the service and ultimate limit states. Because it is almost
impossible to consider all aspects of structural behavior, essential characteristics and basic
assumptions have to be selected and to be considered in the analyses. The numerical studies are
based on those assumptions, which can affect the modeling detail of bridges, including material
modeling, meshing, loadings, and boundary conditions. These assumptions and modeling details
should be verified by comparing analytical results with measurements from experiments to assure
that results from further numerical analyses are reliable. Even though it is not easy to use slab-on-
beam model tests, those results can be used directly for verification and validation of analytical
models. In particular, more realistic loading cases can be applied for those models, i.e. design
truck loadings. Additionally, using the same rules for reinforcement of slabs as specified in the
AASHTO Specification, additional global failure modes including plate failure and punching fail-
ure can be reevaluated during model construction and tests.
In the present study, two bridge types are used. One is the conventional multi-girder two
span continuous bridge, and the other is the two-girder bridge. For each bridge, a 1:4 scaled whole
bridge model and a 1:2 scaled internal support sub-model are proposed. Considering slab thick-
ness and carefully scaled reinforcement in it, it is thought that the 1:4 model maintains the optimal
balance between economy and reality. Longitudinal rebars in the effective width of negative
moment regions can be used in evaluating section capacity, so reinforcement details specified by
current code (e.g. the 1% requirement added to AASHTO LRFD S6.10.3.7 in the 2000 Interims)
82
are also a main contributing factor in strength and ductility evaluation of that section. Therefore,
to capture more realistic behavior in the negative moment section at the ultimate limit state, a
larger scaled model, 1:2 model, is considered appropriate. Based on 1:4 model tested at the ser-
vice limit state and positive moment region ultimate limit state, the location of contra-flexural
positions can be known. Therefore, it is possible to build a 1:2 scaled sub-model representing the
internal support region between two contra-flexural positions. This increases model size without
exceeding budget and experimental capacity limitations.
To avoid secondary failures during ultimate limit state tests, such as longitudinal shear
failure, unintended local buckling of the beam, shear connector failure, longitudinal splitting,
bond failure, and other local failures, longitudinal and transverse reinforcement details and num-
ber and spacing of shear connectors have to be designed and built with extreme caution. Details of
specimen design, laboratory experimental plans, contemplated instrumentation, test procedures,
and accompanying schedule and budget are presented in APPENDIX E. It should be noted that
commencement of form work construction should coincide with (or, better yet, precede) com-
mencement of Task 4 in order for results to be available when needed.
4.2.2.2 Field Test
Finite element analysis of steel-concrete composite bridge for two or multi-girder system
requires a well-represented model to give the accurate results. As the model will be used to pre-
dict the bridge response at the Strength I limit state, modeling procedures are critical.
From a practical point of view, a perfect match between the finite element model and the
prototype bridge is impossible to achieve due to a number of factors. Consequently, engineers are
able to accept the approximate results within a certain range of deviation.
83
One of the most reliable methodologies is to compare the results from finite element anal-
ysis to the experimental results, which can be either from the laboratory or field testing. As far as
the level of accuracy is concerned, laboratory tests would offer a more controllable testing envi-
ronment compared to field tests. The results will be more accurate due to reduced effects of
unknown factors. For instance, guide railing is said to be one source of contributions to the stiff-
ness if not strength of composite structures and is difficult to quantify. Another advantage of labo-
ratory tests is that the model can be tested until it reaches failure at Strength I limit state unlike the
field test which typically cannot be loaded beyond a serviceability limit state. But one of the dis-
advantages of most laboratory tests is the scale or size effect in the scaled model.
Field testing can certainly give good results, but it requires some amount of effort in a
planning process.
In this project, field testing is required for verifying the finite element models at the initial
stage of loading. One of the difficulties is to find a suitable prototype bridge for field testing. The
bridge must at least have a wide-girder spacing characteristic. It should also be in good condition
with no asphalt wearing surface (in order to accommodate strain transducers on the deck).
Field tests conducted on a steel plate girder bridge which carries the east bound traffic on
the I-78 highway over the Delaware River near Easton, Pennsylvania are perhaps the closest to
what is needed for the present study. This bridge is a seven-span bridge with a total length of
1222-ft. The overall deck slab width, from outside to outside of the parapet wall is 51-6. The
span lengths are 100-ft., 169k-ft., 228-ft., 228-ft., 228-ft., 169-ft. and 100-ft. The structural deck
thickness is 10-in. The main superstructure components consist of four welded plate girders,
spaced at 14-3 center to center, which qualifies as a wide-girder spacing.
84
Compared to the I-78 final test report, it is suggested that the more strain gages on con-
crete deck would be needed such that they can capture the longitudinal strain distribution at cer-
tain locations such as the negative moment region.
For the purposes of finite element model verifications, a number of measurements are
required for the comparisons, such as applied loading, longitudinal and transverse strains, dis-
placements and interface slippage. Details of recommended field testing are presented in Appen-
dix E.
The remainder of the Expanded Work Plan closely follows Tasks 5, 6, and 7 in the original
project statement and previous work plan, as described below.
Task 5. Perform parametric studies of different composite steel-bridge superstruc-
ture configurations using the current provisions of the AASHTO LRFD
Bridge Design Specifications and the proposed effective-width criteria.
This task will utilize the findings of the work completed under Tasks 1 through 4 and con-
duct parametric studies of various composite steel-bridge superstructure configurations. These
studies will compare the proposed effective-width criteria to those of the current provisions of the
AASHTO LRFD Bridge Design Specifications. It is envisioned that the software listed above in
Task 4 will also be utilized to facilitate the work in Task 5. As such, the work will progress effi-
ciently, and a maximum number of parameters variations will be possible to pursue. Areas of
excessive conservatism and of unconservatism (if any) in the current provisions will be high-
lighted.
85
Appendix F describes further detail regarding plans to ensure a systematic investigation of
parameters and impacts, consistent with the approaches being developed in the NCHRP 12-50
project. In addition, the Advisory Panel (AP) will be involved in reviewing the parameter matrix
and samples of the work itself while it is ongoing in order to suggest possible mid-course changes
of emphasis.
Task 6. Propose recommended revisions to the specifications and provide design
examples demonstrating their use. For each case considered, develop sug-
gested general guidelines for designing the slab reinforcement and shear
connectors to effectively transfer the calculated shear forces between the
girder and slab at each limit state.
In task 6, further use of the previously developed MathCad worksheets embodying design-
office calculations will be formatted for hardcopy presentation as design examples. These will
also be posted on our web site for downloading and use (since they are executable programs as
well).
The examples will include applicable shear connector design checks, guidelines, and
detailing suggestions, even if no change to current provisions occurs. It is possible that there will
be no change since fatigue typically governs their design. Slab reinforcement design and detailing
will be included in the examples as well.
86
Draft Specifications and Commentary
The format for the draft LRFD Specification provisions and Commentary language will be
fully compatible with the current AASHTO LRFD specification, i.e., written following the two-
column format and standard language of the specification. Since provisions already exist in the
LRFD specification explicitly dealing with effective width, the draft provisions are likely to be in
the form of changes to existing provisions.
The project team considers that it is not within the scope of this project to provide full cal-
ibration for load and resistance factors for whatever effective width provisions result from the
investigation, due to a likely insufficient statistical database. However, since familiar materials
and limit states are employed, it will be considered reasonable to use load and resistance factors
consistent with those materials and limit states particularly the strength limit states for which
fully calibrated factors are already available.
Worked Design Examples
Examples covering both I-girder and box-girder superstructures will be prepared which
demonstrate good design and detailing practice and illustrate key features of the recommended
methodologies and specifications. Project Director Stuart Chen will work closely with the gradu-
ate student(s) in this part of the effort to ensure coordination, accuracy and consistency of presen-
tation, minimal duplication of effort, and readiness of the products for web-based dissemination to
the bridge engineering community. Advisory Panel (AP) members will review an interim draft of
the examples as well, considering cost issues as well as clarity and relevance of presentation.
The principal sources envisioned both for reference and as a starting point in preparing
these worked examples include extensive illustrations of design and code-check procedures for
87
the three principal gravity-related LRFD limit states (Strength I & II, Fatigue/fracture, and Ser-
vice II) as contained in the following example documents:
the Four LRFD Design Examples of Steel Highway Bridges document published by
AISC/NSBA (e.g., Eaton and Grubb 96) augmented with calculations suitable for the
preliminary design stage, and
extensive MathCad-based AASHTO LRFD calculation packages from the best design
teams participating in the NSBA-sponsored intensive graduate course in Steel Bridge
Engineering at UB (e.g., Russo et al. 98).
These calculations in the worked examples will be prepared in dual-unit format, with both
an SI unit format and a customary U.S. units format. In addition to traditional hardcopy format,
these calculations will be prepared in such a way as to be readily converted to web-accessible
interactive forms (e.g., using computer-based design aids such as spreadsheets and web-accessi-
ble interactive design worksheets) as described earlier and in the Implementation Plan.
Since the first above-listed source is considered to be widely available and widely cited by
the U.S. steel bridge engineering community, a tabulated comparison of results and approaches
will be prepared at key points in the analysis and design procedure to show the differences
between the existing design examples (based on simple-span and continuous girder superstruc-
tures seated on bearings over the pier) and the new design examples. Comparison examples will
include design forces, section-property determination, deck and girder comparisons, and so forth.
Comparisons will be summarized in a technical memo for each example. The memo will address
88
similarities and differences at key points in the analysis and design procedure with special empha-
sis on the design output.
Evolving work products from this task will be circulated to members of the Advisory
Panel (AP) for their review and comments. Their involvement will provide independent review
and critique on the technical completeness, quality, and usability of these work products.
Task 7. Submit a final report documenting the entire research effort. The recom-
mended specifications shall be provided in an appendix to the report and
must be in a format suitable for consideration by the AASHTO Highway
Subcommittee on Bridges and Structures.
The results of the project will be documented in a final report, which will summarize the
work performed under the project. A draft of the final report will be submitted 3 months prior to
the scheduled end of the project. Based on review comments received from the NCHRP panel, the
revised final report will be submitted by the end of the 24 month time period designated for this
project. Appendix A of the final report will contain the recommended specifications and commen-
tary. These will be in a form that is ready for integration into the AASHTO LRFD specifications.
Appendix B will contain the recommended design methodologies and worked design examples
with links to downloadable executable versions thereof.
The units used in the final report will be dual (both SI and customary U.S. units). The
Specifications, commentary, and design examples contained in Appendices A and B in the final
report will be prepared in two unit versions each: one complete set in SI units, and one complete
89
set in customary U.S. units. As such, they will be compatible with the two currently available ver-
sions of the AASHTO LRFD specifications.
The budget increment for the recommended laboratory and field experiments is $156,000
and is discussed further in Appendix E. It is assumed that the bridge owner would provide the fol-
lowing for the bridge selected for field testing:
An asphalt-free deck,
Lane closure and associated MPT (maintenance and protection of traffic) for one lane,
A suitable test truck and driver, and
Access to the bridge (e.g., snooper) for installing and removing gauges, including on the
top surface of the deck.
Anticipated research results and Implementation Plan are also in accordance with the pre-
vious workplan and are not repeated here for brevity.
90
4.2.3 Recommended Criteria
The revised criteria must be sufficiently accurate but simple to use. They must be flexible
to be used with minor modification to consider such influences as barrier contribution, skew
effect, and changing of material characteristics of steel and concrete. Ideally, they should also be
comparable to criteria used for other types of bridges, i.e. concrete T-beams, concrete box girders,
etc. Also they have to maintain consistency with other parts of the Specifications. Finally, they
should be usable for bridge evaluation as well as design.
Based on the study of foreign specifications (Appendix D), it is thought that the current
Canadian Bridge Design Code (CSA 2000) has the characteristics described above. Basically, the
Canadian Code uses two basic formulas with only two parameters, girder spacing (overhang for
exterior girders) and effective span length. The same criteria are applied for concrete and steel
girders, and the girders can be beam type or box type. Also, the same effective width can be used
for both service and strength limit states, which is similar to the current AASHTO criteria. It also
should be noted that the Ontario highway bridge loadings and AASHTO HS-25 were used in the
study of effective width performed by Cheung and Chan (Cheung and Chan 1978). That serves as
the basis of the CSA criteria.
Following notations in AASHTO Specification, S.4.6.2.6.2 Segmental Concrete Box
Beams and Single-Cell, Cast-in-Place Box Beams, the recommended criteria based on the cur-
rent Canadian Bridge Code will look like this:
91
For the negative moment regions, the distance between points of permanent load inflec-
tion is used for the current AASHTO Specification. The same basic concept is used for the other
foreign specifications, but they usually define the percentage of span length that can be used as
effective span length for the purpose of determining effective width. For example, 80% of span
length is used for the end span of continuous beams in the Canadian Code without calculation of
inflection points. For internal span of continuous bridges, 60% of span length is used, whereas
25% of the sum of adjacent span lengths is used as effective span length in the negative moment
regions. Figure D-2 in Appendix D shows this idea. A similar percentage of physical span length
is used in Figure 4.6.2.6.2-1 of AASHTO Specifications except for the internal supports.
The effective width of one-side of a girder, be, may be determined as follows:
where
b
e
b
----- 1 1
l
i
15b
---------
)
`

3
= for
l
i
b
--- 15
b
e
b
----- 1 = for
l
i
b
--- 15 >
l
i
= a notional span length to determine effective flange width
b
e
= effective flange width
b = physical flange width on each side of the web
92

Table 4-1: An Example of 1/4 Scaled Model
Description
Overall Width, W 10.2 m (33.5 ft.)
Span Length, L 27 m (88.6 ft.)
Number of Span, N
span
2
Number of Girder, N
girder
4
Girder Spacing, s 3 m (9.8 ft.)
Slab Thickness, t 202 mm (8 in)
Skew Angle, 0 degree
Concrete Compressive Strength, f

c
28 MPa (4 ksi)
Yield Strength of Steel Reinforcement, f
y,rebar
345 MPa (50 ksi)
Yield Strength of Steel Girder, f
y
345 MPa (50 ksi)
Prestressing Force, P 0 kN (0 kip)
Loading HL-93
Table 4-2: Parametric Study Analysis Cases
Category Variables
Boundary
Condition*
Skewness Limit State
Simply
Support
Conti-
nuous
Non
Skew
Skew Fatigue
Service
II
Strength
I
Two-Girder System
Geome-
try
Girder Spacing/
Span Length
(s/l)
x x x x x x x
Material Concrete
Compressive
Strength (f

c
)
x x x x x x x
Steel Yield
Strength (f
y
)
x x x x x x x
93
* Continuous spans take in to account of the behavior in positive and negative moment regions.
** Initial conditions represent axle forces from prestressing action and cable-stayed bridge.
Multi-Girder System
Geome-
try
Girder Spacing/
Span Length
(s/l)
x x x x x x x
Material Concrete
Compressive
Strength (f

c
)
x x x x x x x
Steel Yield
Strength (f
y
)
x x x x x x x
Box-Girder System
Geome-
try
Girder Spacing/
Span Length
(s/l)
x x x x x x x
Material Concrete
Compressive
Strength (f

c
)
x x x x x x x
Steel Yield
Strength (f
y
)
x x x x x x x
Initial
Condi-
tion**
Applied Axial
Forces (P)
x x x x x x x
Table 4-2: Parametric Study Analysis Cases
Category Variables
Boundary
Condition*
Skewness Limit State
Simply
Support
Conti-
nuous
Non
Skew
Skew Fatigue
Service
II
Strength
I
94
Table 4-3: Descriptions of Field Test Bridge (I-78)
Description
Overall Width, W 15.6 m (51 ft.-6 in)
Span Length, L 30.5 m - 34.2 m
(100 ft. - 228 ft.)
Number of Span, N
span
7
Number of Girder, N
girder
4
Girder Spacing, s 4.3 m (14 ft.-3 in)
Slab Thickness, t 250 mm (10 in)
Skew Angle, 0 degree
Concrete Compressive Strength, f

c
28 MPa (4 ksi)
Yield Strength of Steel Reinforcement, f
y,rebar
345 MPa (50 ksi)
Yield Strength of Steel Girder, f
y
345 MPa (50 ksi)
Prestressing Force, P 0 kN (0 kip)
Loading HL-93
95
Figure 4-1: Failure surfaces of concrete (a) Failure surface and (b) Collapsed failure surface
Figure 4-2: ANAMAT cyclic compressive stress-strain model behavior
(a) (b)
96
Figure 4-3: 3D-View of steel girders, cross-frames of 1/4 scaled laboratory model (Span 1)
Figure 4-4: Cross-section of 1/4 scale laboratory model in finite-element analysis
97
Figure 4-5: One half (Span 1) of finite-element model for 1/4 scaled laboratory model
98
REFERENCES
1. AASHO. 1944. Standard Specifications for highway bridges (4th Ed.). Washington, D.C:
American Association of State Highway Officials.
2. AASHO. 1953. Standard Specifications for highway bridges (6th Ed.). Washington, D.C:
American Association of State Highway Officials.
3. AASHTO. 1996. Standard Specifications for Highway Bridges (16th Edition with annual
updated Interims). Washington DC.
4. AASHTO. 1998. AASHTO LRFD Bridge Specifications (2th Edition with annual updated
Interims through 2001). Washington DC.
5. ACI Committee 318. 1989. Building Code Requirements for Reinforced Concrete. ACI.
6. Adekola, A.O. 1968. Effective Widths of Composite Beams of Steel and Concrete. The
Structural Engineer 46, 9: 285-289.
7. AISC. 1936. Specification for the Design, Fabrication and Erection of Structural Steel for
Buildings. American Institute of Steel Construction.
8. AISC. 1961. Specification for the Design, Fabrication and Erection of Structural Steel for
Buildings. American Institute of Steel Construction.
9. AISC. 1998. Manual of Steel Construction: Load & Resistance Factor Design (2nd ed.).
American Institute of Steel Construction.
99
10. Aktan, A.E., K.L. Lee, R. Naghavi and K. Hebbar. 1994. Destructive Testing of Two 80-
Year-Old Truss Bridges. Transportation Research Record 1460: 62-72.
11. Allen, D.N. de G., and R.T. Severn. 1961. Composite Action Between Beams and Slabs
Under Transverse Load. The Structural Engineer 39, 5: 149-154.
12. Amer, A., M. Arockiasamy and M. Shahawy. 1999. Load Distribution of Existing Solid
Slab Bridges Based on Field Tests. Journal of Bridge Engineering-ASCE 4,3: 189-193.
13. Ansourian, P. 1983. The Effective Width of Continuous Beams. Civil Engineering
Transportation CE25,1: 63-70.
14. Ansourian, P. 1984. On the Design of Continuous Composite Beam. Composite and
Mixed Construction - Proceedings of the U.S./Japan Joint Seminar. 1-12.
15. ASCE. 1979. Structural Design of Tall Steel Buildings. New York: ASCE.
16. Australian Bridge Design Code, Section 6, Steel and Composite Construction. 1996. Stan-
dards Australia, Austroads Inc. and The Australasian Railway Association Inc.
17. Australian Standard - Composite Structures, Part 1: Simply Supported Beams (AS
2327.1:1996). 1996. Standards Association of Australia, New South Wales.
18. Ayyub, B.M., Y.G. Sohn, and H. Saadatmanesh. 1992. Prestressed Composite Girders. II:
Analytical Study for negative moment. Journal of Structural Engineering-ASCE 118, 10:
2763-2783.
19. Ayyub, B.M., Y.G. Sohn, and H. Saadatmanesh. 1992. Prestressed Composite Girder. I:
100
Experimental Study for Negative Moment. Journal of Structural Engineering-ASCE 118,
10: 2743-2762.
20. Azizinamini, A., S. Kathol and M. Beacham. 1995. Effect of Cross Frames on Behavior
of Steel Girder Bridges. Fourth International Bridge Engineering Conference. 117-124.
Transportation Research Board.
21. Azizinamini, A., T.E. Boothby, Y. Shekar and G. Barnhill. 1994. Old Concrete Slab
Bridges I: Experimental Investigation. Journal of Structural Engineering-ASCE 120,11:
3284-3304.
22. Bakht, B. and A.C. Agawal. 1995. Deck Slabs of Skew Girder Bridges. Canadian Jour-
nal of Civil Engineering 22,3: 514-523.
23. Bakht, B. and L.G. J aeger. 1988. Ultimate Load Test of Slab-On-Girder Bridge (Structural
Research Report SRR-88-03), Ministry of Transportation of Ontario, Canada.
24. Bakht, B. and L.G. J aeger. 1990. Bridge Testing - A Surprise Every Time. Journal of
Structural Engineering-ASCE 116,5: 1370-1383.
25. Barker, M.G. 2001. Quantifying Field-Test Behavior for Rating Steel Girder Bridges.
Journal of Bridge Engineering-ASCE 6,4: 254-261.
26. Barker, R.M., and J .A. Puckett. 1997. Design of Highway Bridges: Based on AASHTO
LRFD Bridge Design Specifications. J ohn Wiley & Sons.
27. Bathe, K.J . 1996. Finite Element Procedures. Prentice-Hall International, Inc.
101
28. Biggs, R.M., F.W. Barton, J .P. Gomez, P.J. Massarelli and W.T. McKeel. 2000. Finite Ele-
ment Modeling and Analysis of Reinforced-Concrete Bridge Decks (Report No. VTRC 01-
R4). Virginia Transportation Research Council.
29. Botzler, P.W., and J. Colville. 1979. Continuous Composite-Bridge Model Tests. Jour-
nal of the Structural Division-ASCE 105, ST9: 1741-1755.
30. Brendel, G. 1964. Strength of the Compression Slab of T-Beams Subject to Simple Bend-
ing. ACI Journal - Proceedings 61, J anuary: 57-76.
31. Bridge Manual - Manual Number: SP/M/014. 2000. Transit New Zealand, Wellington,
New Zealand.
32. Bridge, R.Q. and Patric, M. 1996. Research on Composite Structures in Australia 1960-
1985. Composite Construction in Steel and Concrete III (Proceedings of an Engineering
Foundation Conference, Irsee, Germany, June 1996), ASCE.
33. BSI. 1979. Part 5: Code of Practice for Design of Composite Bridges. In BS 5400 Steel,
Concrete and Composite Bridges. London: British Standards Institution.
34. BSI. 1982. Part 3: Code of Practice for Design of Steel Bridges. In BS 5400 Steel, Con-
crete and Composite Bridges. London: British Standards Institution.
35. Burdette, E.G., and D.W. Goodpasture. 1972. Comparison of Measured and Computed
Ultimate Strengths of Four Highway Bridges. Highway Research Record No.382.
36. Burdette, E.G., and D.W. Goodpasture. 1973. Tests of Four Highway Bridges to Failure.
102
Journal of the Structural Division-ASCE 99, ST3: 335-348.
37. Byers, D.D. 1999. Evaluation of Effective Slab Width for Composite Cable-Stayed Bridge
Design. Ph.D. diss., The University of Kansas.
38. Cai, C.S., Y. Zhang, and J . Nie. 1998. Composite Girder Design of Cable-Stayed
Bridges. Practice Periodical on Structural Design and Construction-ASCE 3, 4: 158-163.
39. Chajes, M.J ., H.W. Shenton and D. OShea. 1999. Use of Field Testing in Delawares
Bridge Management System. The 8th International Bridge Management Conference -
Transportation Research Circular, Transportation Research Board.
40. Chapman, J .C. 1964. Composite Construction in Steel and Concrete-the Behaviour of
Composite Beams. The Structural Engineer 42, 4: 115-125.
41. Cheung, M.S., and M.Y.T. Chan. 1978. Finite Strip Evaluation of Effective Flange Width
of Bridge Girders. Canadian Journal of Civil Engineering 5: 174-185.
42. Chwalla, E. 1936. Die Formeln zur Berechung der voll mittragenden Breite, dner Gurt-
und Rippenplatten. (in German) Der Stahlbau 2, 10: 73-78.
43. Corrado, J .A. and B.T. Yen. 1973. Failure Tests of Rectangular Model Steel Box Gird-
ers. Journal of the Structural Division-ASCE 99,7: 1432-1455.
44. Cremer, J .-M. 1990. Case Studies in Composite Bridges. In Composite Steel-Concrete
Construction and Eurocode 4 (IABSE short course, Brussels 1990). IABSE.
45. CSA. 2000. Canadian Highway Bridge Design Code (CAN/CSA-S6-00). CSA Interna-
103
tional.
46. CSA. 2001. Commentary on CAN/CSA-S6-00, Canadian Highway Bridge Design Code.
CSA International.
47. Daniels, J .H., and J .W. Fisher. 1966. Fatigue Behavior of Continuous Composite Beams.
Fritz Engineering Laboratory Report No.324.1, Dept. of Civil Engineering, Lehigh Uni-
versity.
48. Daniels, J .H., and J.W. Fisher. 1967. Static Behavior of Continuous Composite Beams.
Fritz Engineering Laboratory Report No.324.2, Dept. of Civil Engineering, Lehigh Uni-
versity.
49. Deng, L. and M. Ghosn. 2000. Nonlinear Analysis of Composite Steel Girder Bridges.
Engineering Journal-AISC 37,4: 140-156.
50. Dilger, W., J .C. Beauchamp, M.S. Cheung, and A. Ghali. 1981. Field Measurements of
Muskwa River Bridge. Journal of the Structural Division-ASCE 107, 11: 2147-2161.
51. Dorton, R.A., M. Holowka and J .P.C. King. 1977. The Conestogo River Bridge - Design
and Testing. Canadian Journal of Civil Engineering 4: 18-39.
52. Dunker, K.F., F.W. Klaiber, and J r. W.W. Sanders. 1986. Post-Tensioning Distribution in
Composite Bridges. Journal of Structural Engineering-ASCE 112, 11: 2540-2553.
53. Dunker, K.F., F.W. Klaiber, F.K. Daoud, and Sanders J r., W.W. 1990.Strengthening of
Continuous Composite Bridges. Journal of Structural Engineering-ASCE 116, 9: 2464-
104
2480.
54. Ebeido, T. and J .B. Kennedy. 1996. Girder Moments in Continuous Skew Composite
Bridges. Journal of Bridge Engineering-ASCE 1,1: 37-45.
55. Elhelbawey, M., Chung C. Fu, M.A. Sahin, and David R. Schelling. 1999. Determination
of Slab Participation From Weigh-In-Motion Bridge Testing. Journal of Bridge Engi-
neering-ASCE 4,3: 165-173.
56. Eurocode 4. 1992. Design of Composite Steel and Concrete Structures, Part 1.1:General
Rules and Rules for Buildings (ENV 1994-1-1:1992). European Committee for Standardi-
sation.
57. Eurocode 4. 1996. Design of Composite Steel and Concrete Structures, Part 2: Bridges
(Draft). European Committee for Standardisation.
58. Eurocode 4. 2000. Design of Composite Steel and Concrete Structures, Part 2: Composite
Bridges (German version ENV 1994-2:1997). European Committee for Standardisation.
59. Fan, H.M., and C.P. Heins. 1974. Effective Slab Width of Simple Span Steel I-Beam Com-
posite Bridges at Ultimate Load, Dept. of Civil Engineering, Maryland Univ., College
Park.
60. Farhey, D.N., R. Naghavi, A. Levi, A.M. Thakur, M.A. Pickett, D.K. Nims and A.E.
Aktan. 2000. Deterioration Assessment and Rehabilitation Design of Existing Steel
Bridge. Journal of Bridge Engineering-ASCE 5,1: 39-48.
105
61. Ferro-concrete. 1910. Engineering: An Illustrated Weekly Journal (London) 90 (Oct.
28): 607.
62. Final Report of the Special Committee on Concrete and Reinforced Concrete. 1916.
ASCE Proceedings 42 (November): 1657-1708.
63. Foutch, D.A., and P.C. Chang. 1982. A Shear Lag Anomaly (Technical Notes). Journal
of the Structural Division-ASCE 108, ST7: 1653-1658.
64. Fu, Chung C., M. Elhelbawey, M.A. Sahin, and David R. Schelling. 1996. Lateral Distri-
bution Factor from Bridge Testing. Journal of Structural Engineering-ASCE 122,9: 1106-
1109.
65. Fu, Y. and J .T. DeWolf. 2001. Monitoring and Analysis of a Bridge with Partially
Restrained Bearings. Journal of Bridge Engineering-ASCE 6,1: 23-29.
66. Gjelsvik, A. 1991. Analog-Beam Method for Determining Shear-Lag Effects. Journal of
Engineering Mechanics-ASCE 117, 7: 1575-1594.
67. Goldbeck, A.T., and E.B. Smith. 1916. Tests of Large Reinforced Concrete Slabs. ACI-
Proceedings 12: 324-333.
68. Haensel, J . 1997. Recent Trends in Composite Bridge Design and Construction in Ger-
many. Composite Construction in Steel and Concrete III (Proceedings of an Engineering
Foundation Conference, Irsee, Germany, June 1996), ASCE.
69. Hagood, T.A. J r., L. Guthrie, and P.G. Hoadley. 1968. An Investigation of the Effective
106
Concrete Slab Width for Composite Construction. Engineering Journal - AISC 5, 1 (J an-
uary): 20-25.
70. Hamada, S., and J . Longworth. 1976. Ultimate strength of composite beams. Journal of
Structural Engineering-ASCE 102, 7: 1463-1478.
71. Hambly, E.C. 1991. Bridge Deck Behavior (2nd ed.). E&FN SPON.
72. Hanswille, G. 1997. Composite Bowstring Arches with Reinforced Concrete Decks Act-
ing as Tension Members in the Main System. Composite Construction in Steel and Con-
crete III (Proceedings of an Engineering Foundation Conference, Irsee, Germany, June
1996), ASCE.
73. Hassan, M., O. Burdet and R. Favre. 1995. Analysis and Evaluation of Bridge Behavior
Under Static Load Testing Leading to Better Design and J udgment Criteria. Fourth Inter-
national Bridge Engineering Conference 1. 296-303.
74. HDR Engineering. 1995. Four LRFD Design Examples of Steel Highway Bridges (Vol.II,
Chap. 1A, Highway Structures Design Handbook). American Iron and Steel Institute.
75. Heins, C.P. 1980. LFD Criteria for Composite Steel I-Beam Bridges. Journal of the
Structural Division-ASCE 106, ST11: 2297-2312.
76. Heins, C.P., and H.M. Fan. 1976. Effective Composite Beam Width at Ultimate Load.
Journal of the Structural Division-ASCE 102, ST11: 2163-2179.
77. Helba, A. and J.B. Kennedy. 1994. Collapse Loads of Continuous Skew Composite
107
Bridges. Journal of Structural Engineering-ASCE 120,5: 1395-1414.
78. Helba, A. and J .B. Kennedy. 1994. Parametric Study on Collapse Loads of Skew Com-
posite Bridges. Journal of Structural Engineering-ASCE 120,5: 1415-1433.
79. Hibbit, Karlsson & Sorensen. 2000. ABAQUS/Standard Users Manual - Version 6.1 Vol.2.
Hibbitt, Karlsson & Sorensen, Inc.
80. Hose, Y.D. 2001. Seismic Performance and Failure Behavior of Plastic Hinge Regions in
Flexural Bridge Column. Ph.D. Dissertation, University of California, San Diego.
81. Hulsey, J .L. and D.K. Delaney. 1993. Static Live Load Tests on a Cable-Stayed Bridge.
Transportation Research Record No. 1393. 162-174.
82. Idriss, R.L., K.R. White, C.B Woodward., and D.V. J auregui. 1995. After-Fracture
Redundancy of Two-Girder Bridge: Testing I-40 Bridges Over Rio Grande. Fourth Inter-
national Bridge Engineering Conference 2: 316-326.
83. J ohnson, J .E., and A.D.M. Lewis. 1966. Structural Behavior in a Gypsum Roof-Deck
System. Journal of the Structural Division-ASCE 92, ST2 (April): 283-296.
84. J ohnson, R.P. 1970. Research on Steel-Concrete Composite Beams. Journal of the
Structural Division-ASCE 96, ST3: 445-459.
85. J ohnson, R.P. 1992. EC4:Composite Structures of Steel and Concrete. In Structural
Eurocodes, IABSE Conference, Davos, 1992. 197-202. IABSE.
86. J ohnson, R.P. 1997. Some Research on Composite Structures in the UK 1960-1985.
108
Composite Construction in Steel and Concrete III (Proceedings of an Engineering Foun-
dation Conference, Irsee, Germany, June 1996), ASCE.
87. J ohnson, R.P., and D. Anderson. 1993. Designers' Handbook to Eurocode 4 (Part 1.1:
Design of Composite Steel and Concrete Structures). London: Thomas Telford.
88. J ohnson, R.P., and R.J . Buckby. 1986. Composite Structures of Steel and Concrete, Vol 2:
Bridges. London: Collins.
89. J RA. 1996. Design Specifications for Highway Bridges (Part I - General). J apan Road
Association.
90. J RA. 1996. Design Specifications for Highway Bridges (Part II - Steel Bridges). J apan
Road Association.
91. Krmn, T.v. 1924. Die Mittragende Breite. Beitrge zur technischen Mechanik 114-
127.
92. Kathol, S, A. Azizinamini and J . Luedke. 1995. Strength Capacity of Steel Girder Bridges
- Final Report (Report No. RES1(0099) R469), Nebraska University, Lincoln.
93. Kennedy, J .B. and N.F. Grace. 1983. Load Distribution in Continuous Composite
Bridge. Canadian Journal of Civil Engineering 10,3: 384-395.
94. Kennedy, J .B., and M. Soliman. 1992. Ultimate Loads of Continuous Composite
Bridges. Journal of Structural Engineering-ASCE 118, 9: 2610-2623.
95. Kennedy, J .B., and N.F. Grace. 1982. Prestressed Decks in Continuous Composite
109
Bridges. Journal of the Structural Division-ASCE 108, ST11: 2394-2410.
96. Khalil, A, L. Greimann, T.J . Wipf, and D. Wood. 1998. Modal Testing for Nondestructive
Evaluation of Bridges: Issues. Transportation Conference Proceedings. 109-112,
97. Kim, S., A.S. Nowak and R. Till. 1996. Verification of Site-Specific Live Load on
Bridges. Probabilistic Mechanics and Structural Reliability and Geotechnical Reliability,
Proceedings of the seventh specialty conference. 214-217.
98. Kim, W.-J., and C.-M. Park. 2001. Design and Construction of Precast Panels for the
Composite Cable-Stayed Bridge. Cable-Supported Bridges-Challenging Technical Limits
(IABSE Conference, Seoul, Korea), IABSE.
99. Kindmann, R. 1990. Composite Girders. IABSE Short Course Composite Steel-Con-
crete Construction and Eurocode 4, IABSE Reports Vol.61. 117-145.
100. Klaiber, F.W., T.J . Wipf and B.M. Phares. 1997. Precast Units for Simple Span Bridge.
Recent Advances in Bridge Engineering. 223-230.
101. Kostem, C.N., T.D. Hand, and E.S. de Castro. 1984. Finite Element Modeling of Box-
Beam Bridges. Computing in Civil Engineering. 223-232.
102. Kostem, C.N. and T.A. Lenox. 1988. Idealization of Support Conditions of Highway
Bridge. Conference Proceedings: Computing in Civil Engineering-5th conference. 108-
114.
103. Kuhlmann, U. 1997. Design, Calculation and Details of Tied-Arch Bridges in Composite
110
Constructions. Composite Construction in Steel and Concrete III (Proceedings of an
Engineering Foundation Conference, Irsee, Germany, June 1996), ASCE.
104. Kumarasena, S.T. and R.J. McCabe. 2001. Bostons Storrow Drive Connector Bridge.
Informational: Brochure. National Steel Bridge Alliance.
105. Labia, Y., M.S. Saiidi and B. Douglas. 1997. Full-Scale Testing and Analysis of 20-Year-
Old Pretensioned Concrete Box Girders. ACI Structural Journal 94,5: 471-482.
106. Lampe, N. and A. Azizinamini. 2001. Steel Bridge System, Simple for Dead Load and
Continuous for Live Load, NaBRO Official Website.
107. Lebet, J .-P. 1990. Composite Bridges. In Composite Steel-Concrete Construction and
Eurocode 4 (IABSE Short Course, Brussels 1990). IABSE.
108. Lee, J .A.N. 1962. Effective Widths of Tee-Beams. The Structural Engineer 40, 1: 21-
27.
109. Loo, Y.-C., and T.D. Sutandi. 1986. Effective Flange Width Formulas for T-beams. Con-
crete International-ACI 8, 2: 40-45.
110. Mackey, S., and F.K.C. Wong. 1961. The Effective Width of a Composite Tee-Beam
Flange. The Structural Engineer 39, 9 (September): 277-285.
111. Maeda, Y. 1997. Research and Development of Steel-Concrete Composite Construction
in J apan from 1950 to 1986. Composite Construction in Steel and Concrete III (Proceed-
ings of an Engineering Foundation Conference, Irsee, Germany, June 1996), ASCE.
111
112. Maeda, Y., and S. Matsui. 1985. Load-Carrying Capacity and Remaining Life of First-
Built Composite Girder Bridge in J apan. Composite and Mixed Construction (Proceed-
ings of the US/Japan Joint Seminar, July 18-20, 1984), ASCE.
113. Marguerre, Von K. 1952. Uber die Beanspruchung von Plattentrgem. Der Stahlbau 8.
114. Mensch, L.J ., G.S. Bergendahl, and E.S. Martin. 1917. Discussion on Final Report of the
Special Committee on Concrete and Reinforced concrete. ASCE Proceedings 43: 1873-
1890.
115. Metzer, von W. 1929. Die Mittragende Breite. Luftfarhtforschung 4.
116. Meyer, C. ed. 1987. Finite Element Idealization For Linear Elastic Static and Dynamic
Analysis of Structures in Engineering Practice. American Society of Civil Engineers.
117. Military Handbook (Structural Engineering-Steel Structures). 1987. MIL-HDBK-1002/3,
Superseding NAVFAC DM-2.03.
118. Miller, A. 1929. ber die mittragende Breite. Luftfarhtforschung 4, 1.
119. Miller, R.A., A.E. Aktan, and B.M. Shahrooz. 1994. Destructive Testing of Decommis-
sioned Concrete Slab Bridge. Journal of Structural Engineering-ASCE 120,7: 2176-
2198.
120. Moffatt, K.R., and P.J. Dowling. 1978. British Shear Lag Rules for Composite Girders.
Journal of the Structural Division-ASCE 104, ST7: 1123-1130.
121. Moore, M. and I.M. Viest. 1993. American Iron and Steel Institute Federal Highway
112
Administration Model Bridge Test. Transportation Research Record 1380. 9-18.
122. Moses, F., J .P. Lebet and R. Bez. 1994. Applications of Field Testing to Bridge Evalua-
tion. Journal of Structural Engineering-ASCE 120,6: 1745-1762.
123. Mufti, A.A., L.G. J aeger and B. Bakht. 1997. Field Performance of Steel-Free Deck Slabs
of Girder Bridges. Recent Advances in Bridge Engineering. 239-246.
124. Myers, J .J. and Y. Yang. 2001. Practical Issues for the Application of High-Performance
Concrete to Highway Structures. Journal of Bridge Engineering-ASCE 6,6:613-627
125. Newmark, N.M. 1949. Design of I-Beam Bridges. ASCE Transactions 114: 997-1022.
126. Newmark, N.M., C.P. Siess, and R.R. Penman. 1946. Studies of Slab and Beam Highway
Bridges: Part I. Tests of Simple-Span Right I-beam Bridges. Bulletin No.363, University
of Illinois Engineering Experiment Station.
127. NZS 3101:1995. 1995. Code of Practice for the Design of Concrete Structures. Standards
Association of New Zealand, Wellington, New Zealand.
128. NZS 3404:1997. 1997. Steel Structures Standard. Standards Association of New Zealand,
Wellington, New Zealand.
129. Pantazopoulou, S.J., and J .P. Moehle. 1990. Identification of Effect of Slabs on Flexural
Behavior of Beams. Journal of Engineering Mechanics-ASCE 116, 1: 91-105.
130. Preliminary Report of the Committee on Reinforced Concrete Bridges and Culverts.
1916. ACI-Proceedings 12: 419.
113
131. Proposed Revised Standard Building Regulations for the use of Reinforced Concrete.
1916. ACI-Proceedings 12: 178.
132. Proposed Standard Building Regulations for the use of Reinforced Concrete. 1917. ACI-
Proceedings 13: 416.
133. Ramey, G.E. 2001. Deck Thickness Considerations for CIP Reinforced Concrete Bridge
Decks. Practice Periodical on Structural Design and Construction. 35-47.
134. Raoul, J., and W. Hoorpah. 2001. Common Road Steel-Concrete Composite Bridges in
France Evolution and Future Applications. Proceedings of 2001 World Steel Bridge Sym-
posium, NSBA, FHWA.
135. Razaqpur, A.G. and M.Nofal. 1988. Transverse Load Distribution at Ultimate Limit States
in Single Span Slab-on-Girder Bridges with Compact Steel Girders (MISC-88-01).
Ontario Ministry of Transportation.
136. Razaqpur, A.G., and M. Nofal. 1990. Analytical Modeling of Nonlinear Behavior of
Composite Bridges. Journal of Structural Engineering-ASCE 116, 6: 1715-1733.
137. Reid, J .S., M.J . Chajes, D.R. Mertz and G.H. Reichelt. 1996. Bridge Strength Evaluation
Based on Field Test. Probabilistic Mechanics and Structural Reliability. 294-297.
138. Reinforced-Concrete Construction. 1913. Engineering:An Illustrated Weekly Journal
(London) 95 (March 28): 430.
139. Reissner, E. 1946. Analysis of Shear Lag in Box Beams by the Principle of Minimum
114
Potential Energy. Quarterly of Applied Mathematics 4, 3 (Oct.): 268-278.
140. Richart, F.E. 1949. Laboratory Research on Concrete Bridge Floors. ASCE Transactions
114: 980-996.
141. Roik, K. 1997. Review of the Development of Composite Structures in Germany 1950-
1990. Composite Construction in Steel and Concrete III (Proceedings of an Engineering
Foundation Conference, Irsee, Germany, June 1996), ASCE
142. Roschke, P.N. and K.R. Pruski. 2000. Overload and Ultimate Load Behavior of Postten-
sioned Slab Bridge. Journal of Bridge Engineering-ASCE 5,2: 148-155.
143. Salmon, C.G., and J .E. J ohnson. 1996. Steel Structures: Design and Behavior (4th Ed.).
Harper Collins College Publishers.
144. Saraf, V.K, A.S. Nowak and R. Till. 1996. Proof Load Testing of Bridges. Probabilistic
Mechanics and Structural Reliability and Geotechnical Reliability, Proceedings of the
seventh specialty conference. 526-529.
145. Saraf, V.K. and A.S. Nowak. 1997. Field Evaluation of Steel Girder Bridge. Transporta-
tion Research Record 1594. 140-146.
146. Saul, R. 1997. Design and Construction of Long Span Steel Composite Bridges. Com-
posite Construction in Steel and Concrete III (Proceedings of an Engineering Foundation
Conference, Irsee, Germany, June 1996), ASCE.
147. Savage, I. and M.Nasim 2001. The Woodrow Wilson Bridge - Innovation at Work. Pro-
115
ceedings of 2001 world Steel Bridge Symposium: 27.1-27.10.
148. Savage, I. 2001. Personal Communication with S. Chen, October 2001.
149. Schade, H.A. 1951. The Effective Breadth of Stiffened Plating under Bending Loads.
Trans. SNAME 59.
150. Schleicher, F. 1955. Taschenbuch fur bauingenieure. Vienna.
151. SETRA. 1990. Ponts mixtes acier-beton bipoutres - Guide de conception. Service
D'etudes Techniques des Routes et Autoroutes, France, Octobre 1985 (Rempression
mars).
152. Siess, C.P. 1949. Composite Construction for I-Beam Bridges. ASCE Transactions 114:
1023-1045.
153. Siess, C.P., and I.M. Viest. 1953. Studies of Slab and Beam Highway Bridges: Part V.
Tests of Continuous Right I-Beam Bridges. Bulletin No.416, University of Illinois Engi-
neering Experiment Station.
154. Slutter, R.G., and G.C. Driscoll. 1965. Flexural Strength of Steel-Concrete Composite
Beams. Journal of the Structural Division-ASCE 91, ST2: 71-99.
155. Song, Q., and A.C. Scordelis. 1990. Formulas for Shear-Lag Effect of T-, I-, and Box
Beams. Journal of Structural Engineering-ASCE 116, 5: 1306-1318.
156. Song, Q., and A.C. Scordelis. 1990. Shear-Lag Analysis of T-, I-, and Box Beams. Jour-
nal of Structural Engineering-ASCE 116, 5: 1290-1305.
116
157. Stallings, J .M., and C.H. Yoo. 1993. Tests and Ratings of Short-Span Steel Bridges.
Journal of Structural Engineering-ASCE 119,7: 2150-2168.
158. Steel Structures for Building: Limit State Design (CAN/CSA-S16.1-M89). 1989. Ottawa.
159. Subcommittee. 1974. Composite Steel-Concrete Construction. Journal of the Structural
Division-ASCE 100, ST5: 1085-1139.
160. Svensson, H. 1997. The Development of Steel-Concrete Composite Cable-Stayed
Bridges. Composite Construction in Steel and Concrete III (Proceedings of an Engineer-
ing Foundation Conference, Irsee, Germany, June 1996), ASCE.
161. Talbot, A.N. 1907. Tests of Reinforced Concrete T-Beams (Series of 1906). Engineering
Experiment Station Bulletin No.12, University of Illinois.
162. Tentative Specifications for Concrete and Reinforced Concrete. 1921. ASCE Proceed-
ings 47, 6: 59-124.
163. Timoshenko, S.P., and J .M. Gere. 1963. Theory of Elastic Stability. McGraw-Hill.
164. Timoshenko, S.P., and J .N. Goodier. 1970. Theory of Elasticity. McGraw-Hill.
165. Unwin, W.C. 1911. Simplification of Formula for Ferro-Concrete Beams with Single
Reinforcement. Engineering:An Illustrated Weekly Journal (London) 91 (March 3): 267.
166. Vallenilla, C.R., and R. Bjorhovde. 1985. Effective Width Criteria for Composite
Beams. Engineering Journal - AISC 22, 4: 169-175.
117
167. Viest, I.M. 1956. Investigation of Stud Shear Connectors for Composite Concrete and
Steel T-Beams. Journal of the American Concrete Institute-Proceedings 52 (April): 875-
891.
168. Viest, I.M. 1960. Review of Research on Composite Steel-Concrete Beams. Journal of
the Structural Division-ASCE 86, ST6: 1-21.
169. Viest, I.M. 1997. Studies of Composite Construction at Illinois and Lehigh; 1940-1978.
Composite Construction in Steel and Concrete III (Proceedings of an Engineering Foun-
dation Conference, Irsee, Germany, June 1996), ASCE.
170. Waldram, P.J . 1913. Some Notes on the Proposed L.C.C. Regulations for Reinforced
Concrete-No.II. Engineering:An Illustrated Weekly Journal (London) 95 (J une 20): 836-
838.
171. Walther, R., B. Houriet, W. Isler, and P. Moa. 1988. Cable Stayed Bridges. London: Tho-
mas Telford.
172. Wegmuller, A.W. 1977. Overload Behavior of Composite Steel-Concrete Bridges. Jour-
nal of the Structural Division-ASCE 103, ST9: 1799-1819.
173. Winter, G. 1940. Stress Distribution in an Equivalent Width of Flange of Wide Thin-Wall
Steel Beams, National Advisory Committee for Aeronautics (Tech. Note NO.784).
174. Yakel, A., P. Mans and A.Azizinamini. 1999. Negative Bending Tests of HPS-485W
Bridge Girders. Materials and Construction. 559-565.
118
175. Yen, B.T., T. Huang and D.A. VanHorn. 1995. Field Testing of a Steel Bridge and a Pre-
stressed Concrete Bridge: Final Report Field Study of a Steel Plate-Girder Bridge - Vol. 2,
Research Project No. 86-05. Fritz Engineering Laboratory Report No. 519.2, Lehigh
University.
176. Youn, Seok-Goo and Sung-Pil Chang. 1998. Behavior of Composite Bridge Decks Sub-
jected Static and Fatigue Loading. ACI Structural J ournal 94,3: 249-258.
A-1
APPENDIX A
SURVEY RESULTS
Survey form is shown on the next page and survey results are shown in Table A-1.
A-2
Survey of Current Practice:
Effective Slab Width for Composite Steel Bridge Members
NCHRP Project 12-58
Please answer each of the following questions
1. What criteria are used by your agency for effective slab width for composite steel bridge mem-
bers? Circle one: AASHTO other If other, briefly describe:_________________________
_________________________________________________________________________
_________________________________________________________________________
2. Is your agency conducting (or has it recently conducted) any research or demonstration projects
on effective slab width? Circle one: Yes No If yes, briefly describe and list contact informa-
tion if different than below: _____________________________________________________
_________________________________________________________________________
_________________________________________________________________________
3. Is your agency conducting (or has it recently conducted) any field testing on bridges with steel
girder spacings greater than 10 ft. (3 m)? Circle one: Yes No If yes, briefly describe and list
contact information if different than below: ________________________________________
________________________________________________________________________
________________________________________________________________________
4. Has your agency recently constructed any steel girder bridges with girder spacings greater than
12 ft. (3.6 m)? Circle one: Yes No If yes, briefly describe and list contact information if
different than below:
____________________________________________________________
__________________________________________________________________________
__________________________________________________________________________
5. Do you currently have a policy limiting maximum girder (stringer) spacings in new designs for
steel girder bridges? Circle one: Yes No If yes, briefly describe that pol-
icy:________________________________________________________________________
__________________________________________________________________________
6.Survey form Completed By:________________________________________________(name)
___________________________________________________________(affiliation,division)
________________________________________(FAX number, e-mail address)
Please return survey to:
Stuart S. Chen, Ph.D., P.E.
Dept. of Civil, Structural, and Environmental Engineering
212 Ketter Hall, SUNY at Buffalo, Buffalo NY 14260
FAX 716-645-3733
Thank you!
A-3
Table A-1: Survey Results for Effective Slab Width Criteria
S > 12-ft. S > 10-ft. Policy Comments(" Yes" only)
No. State
1,2
(3.6 m) (3 m) Limitation Max. Girder Min. Girder Reason Remark
built tested Spacing (ft.) Number
1 Alabama - - - - - - -
2 Alaska N N N - - - -
3 Arizona - - - - - - -
4 Arkansas N N Y 11' - - -
5 California Y N N - - - White Hill Sidehill Viaduct, 3 girders
spacing 15-ft. Designed in 10/27/00
6 Colorado N/A N N - - - -
7 Connecticut N N N - - - -
8 Delaware - - - - - - -
9 Florida - - - - - - -
10 Georgia N N Y 10'-6" - - -
11 Hawaii N N N - - - -
12 Idaho N N N - - - -
13 Illinois N N N - - - -
14 Indiana N N N - - - -
15 Iowa N N N - - - -
16 Kansas N N Y 10' - For deck replacement -
17 Kentucky N N N - - - -
18 Louisiana N N N - - - -
19 Maine - - - - - - -
20 Maryland N N Y 10' - For deck replacement -
21 Massachusetts N N N - - - -
22 Michigan N N Y See Reason Follow AASHTO -
23 Minnesota - - - - - - -
24 Missisippi N N N - - - -
25 Missouri N N Y 10"-0" to 10"-5' - Limits for 3"thk -
Precast/Prestressed deck
26 Montana N N N/A - - - -
27 Nebraska - - - - - - -
28 Nevada N N Y 12'-0" to 14'-0" - Follow NDOT bridge -
design manual
29 New Hampshire N N N - - - -
30 New J ersey N N N - - - -
31 New Mexico - - - - - - -
32 New York N N Y 11'-6" - - Larger spacing in special cases
subject to approval
33 North Carolina N N Y 12'-0" - - -
34 North Dakota N N N - - - -
35 Ohio N N Y - 4 - -
36 Oklahoma N N Y - - Follow AASHTO -
37 Oregon Y N N - - - 4-girder hybrid w/ 12.75-ft. spacing
38 Pennsylvania Y Y Y 15'-0" - Due to SIP form I-78 Bridge done by Lehigh Univ.
39 Puerto Rico - - - - - - -
40 Rhod Island - - - - - - -
41 South Carolina N N N - - - -
42 South Dakota - - - - - - -
43 Tennessee N N N - - - -
44 Texas N N N - - - "Prefer" 10-ft. max. spacing
45 Utah Y N N - - - -
46 Vermont N N Y 10'-0" - Larger spacing if economical -
47 Virginia N N Y 12'-0" - - -
48 Washington N N N - - - -
49 West Virginia - - - - - - -
50 Wisconsin N N N - - - -
51 Wyoming N N Y 10-6" - Standard girder spacing -
based on deck width
Note:
1. All respondents follow AASHTO for b
eff
.
2. All respondents report no b
eff
research.
- means "No Answer"
B-1
APPENDIX B
SHEAR LAG PHENOMENON
In order to evaluate effective width of flanges to be used with elementary theory of bend-
ing of beams, real behavior of combined structures of slabs and beams must be understood. Gen-
erally, slab plate in-plane and/or plate bending behavior are described by governing partial
differential equations while beam bending behavior is described with elementary bending theory.
Because historically there were no computers, reducing this problem to something that could be
solved by hand calculation was inevitable in the early studies.
von Karman developed a solution for the combined plate and beam structure based on sev-
eral assumptions such as:
(1) The structure is infinitely long and continuous on equidistant supports
(2) All spans are equally loaded by loads symmetrical with respect to the middle of span,
(3) the width of the flange is infinitely large and its thickness is very small in comparison
with the depth of the beam
(4) Bending of the flange as a thin plate can be neglected (von Karman 1924).
Timoshenko and Goodier also developed the same solution as Karmans (Timoshenko and
Goodier 1970). Figure B-1 shows effective flange width concept of continuous beam used in their
study. Due to applied loads, stresses and resultant forces were developed in a deck and a support-
ing beam. Evaluation of effective width, , was the major concern of their studies.
For plate in-plane behavior, the basic two dimensional field equation of elasticity is
expressed as Equation (B-1), with the geometric compatibility condition described in Equation
2
B-2
(B-2). Displacement-strain and stress-strain relation in two-dimensional plane stress problems are
summarized in Equation (B-3).
(B-1)
where

(B-2)
(B-3)
where

After introducing a stress function, , which satisfies Equation (B-4), equilibrium is auto-
matically satisfied, and the compatibility condition can be expressed as shown in Equation (B-5)
without considering body force.

x
--------

xy

y
---------- B
x
+ + 0 =

xy

x
----------

y

y
-------- B
y
+ + 0 =

x

y
and
xy
: the normal and shear stresses, respectively , ,
B
x
B
y
: body forces ,

x

y
+ ( )
2
x
2
2

y
2
2

+
\
|
| |

x

y
+ ( ) 1 + ( )
B
x

x
--------
B
y

y
-------- +
\
| |
= =

x
u
x
-----
1
E
---
x

y
( ) = =

y
v
y
-----
1
E
---
y

x
( ) = =

xy
u
y
-----
v
x
-----
1
G
----
xy
= + =
: Poisson ratio
E : Modulus of Elasticity
G : Shear Modulus
F
B-3
(B-4)
and
(B-5)
In order to select a stress function, the two conditions in Equation (B-6) and Equation (B-
7) below are imposed to satisfy geometric conditions. Then, the stress function, Equation (B-8), is
introduced with additional condition that . Therefore, the strain energy of a plate, V
1
,
is expressed as Equation (B-9).
(B-6)
(B-7)

x
F
2

y
2

-------- + =
y
F
2

x
2

-------- + =
xy
F
2

x y
----------- =
B
x

x
------- = B
y

y
------- =
F
4

F
4

x
4

-------- 2
F
4

x
2
y
2

----------------
F
4

y
4

--------
2
+ + 0 = =
f
n
( ) 0 =

xy
F
2

x y
----------- 0 = =

xy
u
y
-----
v
x
----- + 0 = =

xy
y
---------
u
2

y
2

--------
v
2

x y
----------- + 0 = =

x
x
-------
v
2

x y
-----------
1
E
---

y

x
--------

x

x
--------
\
| |
0 = = =
F
3

x
3

--------
F
3

x y
2

-------------- 0 =
B-4
(B-8)
(B-9)
After dividing the resultant moment, M, into a contribution due to the resultant shear force
at the interface, M, and a beam contribution, M, two force boundary conditions, Equation (B-
10) and Equation (B-11), can be developed. Based on the resultant forces applied in the beam, the
strain energy stored in the beam can be evaluated as Equation (B-12). The total strain energy sum
of two elements, Equation (B-14), can be specified with two unknown coefficient, X
n
and Y
n
, due
to the applied moment M(x) given in Equation (B-13).
(B-10)
(B-11)
(B-12)
F f
n
y ( )
nx
l
--------- cos
n 1 =

=
A
n
e
ny ( ) l
B
n
1
ny
l
--------- +
\
| |
e
ny ( ) l
+
nx
l
--------- cos
n 1 =

=
V
1
2
h
2 E
-----------
x
2

y
2
2
x

y
2 1 + ( )
xy
2
+ + | ] x d y d
0
2l

=
2h
n
3

3
l
2
-----------
B
n
2
E
------
A
n
B
n
2G
------------
A
n
2
2G
------- + +
\
|
| |
n 1 =

=
N 2h
x
y d
0

+ 0 =
M x ( ) M' M'' + M' N e M' 2he
x
y d
0

= = =
V
2
N
2
2AE
-----------
M'
2
2EI
--------- + x d
0
2l

=
B-5
(B-13)
(B-14)
The two coefficients, Xn and Yn, in the total strain energy can be evaluated by applying the
minimum potential energy theorem, as shown in Equation (B-15). If the applied moment is
assumed as Equation (B-16), then the relation between M and M can be explicitly evaluated as
Equation (B-17).
(B-15)
(B-16)
(B-17)
On the other hand, the stress at the middle plane of the flange, , can be calculated by
using the stress at the centroid of beam, , and M based on elementary beam theory as shown in
Equation (B-18). The resultant moments based on these stresses can be expressed as Equation (B-
19), and their ratio can be expressed as Equation (B-20). Finally, the effective width, in Equation
(B-21), can be determined by comparing Equation (B-17) with Equation (B-20).
(B-18)
M x ( ) M
0
M
n
nx
l
--------- cos
n 1 =

+ =
V X
n
Y
n
, ( ) V
1
V
2
+ = X
n
2h
n
l
------A
n
= Y
n
2h
n
l
------B
n
=
V
X
n

--------- 0 =
V
Y
n

-------- 0 =
M x ( ) M
1
x
l
------
\
| |
cos =
M''
M
1 I Ae
2
( ) I he
2
l ( ) 3 2
2
+ ( ) 4 | ] + +
--------------------------------------------------------------------------------------------------------- =

e

c
M' e
I
------------- =
B-6
(B-19)
(B-20)
(B-21)
In the study of Allen and Severn (1961), not only the in-plane plate behavior but also plate
bending behavior was considered. They decomposed the original three-dimensional composite
slab-beam problem into two-dimensional problems by applying compatibility conditions between
thin-plate behavior for a slab and simple bending behavior for a beam, which is the same
approach as the previous studies by Karman and Timoshenko. Following the sign convention of
Figure B-2, the governing equations for a plate can be expressed as depicted in Equation (B-22),
considering plate bending behavior.
(B-22)
where
h
e

c
A + 0 = M'
I
e
-- 1
2h
A
--------- - +
\
| |

e
=
M'' N e 2h
e
e = =
M''
M
-------
M''
M' M'' +
-------------------
1
1 I Ae
2
( ) I 2he
2
( ) + +
-------------------------------------------------------------- = =
I
2he
2
---------------
I
he
e
l
----------
3 2
2
+
4
--------------------------- =
2
4l
3 2
2
+ ( )
----------------------------------- =
F
4
0 =
D w
4
q =
D
Eh
3
12 1
2
( )
------------------------- =
q =uniformly distributed load
a
2
-- - x
a
2
--- 0 y L
B-7
The first equation of Equation (B-22) is the governing equation for in-plane behavior of
the plate and is identical to Equation (B-5). The second equation governs thin-plate bending
behavior, which is defined in terms of the vertical displacement function, w. Considering geomet-
ric conditions at the assumed simply-supported edge and symmetry, the stress function, F, and the
displacement function, w, can be defined as in Equation (B-23) based on a series form:
(B-23)
where
In order to determine the four unknown coefficients, A
n
, B
n
, P
n
, and Q
n
, two geometric
boundary conditions specified in Equation (B-24) - namely, equilibrium conditions at the intersec-
tion of the beam and the slab as shown in Equation (B-25), and strain compatibility condition at
the intersection as depicted in Equation (B-26) were used.
(B-24)
F
n
L
------y A
n
n
L
------x
\
| |
B
n
x
n
L
------x
\
| |
sinh + cosh sin
n 1 =

=
w
n
L
------y q
n
P
n
n
L
------x
\
| |
Q
n
x
n
L
------x
\
| |
sinh + cosh + sin
n 1 =

=
q
n
4qL
4
n ( )
5
D
------------------ =
n 1 3 5 , , , =
w
x
------ 0 =
u 0 =
)

on x
a
2
-- - = any y

x
1
E
---
x

y
( )
u
x
----- 0 = = = on x 0 = any y
B-8
(B-25)
where
(B-26)
Because the geometric boundary condition and strain compatibility condition are evalu-
ated at the mid-plane of the plate, stress and strain contribution from plate bending behavior were
not considered in compatibility conditions. This study also considered multiple beams with sym-
metric loading conditions, and the stress distribution of slab deck was calculated. However, any
extended work for effective width evaluation was not tried.
Based on a similar approach, Lee studied the effective width problem considering force
boundary conditions between slab and beam as shown in Figure B-3 (Lee 1962). He considered
resultant axial forces and bending moment developed in the slab and beam due to interactive
forces, p and n, by applied force F. Shear force can be replaced with resultant axial force, N, and
moment, Nh and Nt/2. For the plate, axial force, N, is resisted by plate in-plane behavior, while
reaction force, p, and moment Nt/2 are resisted by plate bending behavior. He used stresses for in-
q
1
dy dF 2Q
x
dy + 0 =
dN 2h
xy
dy 0 =
dM Fdy 2h
xy
l dy ( ) + 0 =
l
1
2
--- d h ( ) =
M EI
w
2

y
2

--------- =
Q
x
D
x
w
2

x
2

---------
w
2

y
2

--------- +
\
|
| |
=

y
beam ( )
y
plate ( ) =
N
Ebd
----------
w
2

y
2

---------
1
2
--- d h ( ) +
1
E
---
y

x
( ) =
B-9
plane behavior as specified in Equation (B-27), which were derived from a stress function, and
satisfy geometric boundary conditions at the simply-supported edges. Different displacement
function for plate bending behavior were needed for each loading case; distributed line loading
and moment couple. These appear in Equation (B-28) and Equation (B-29).
(B-27)
(1) for a line load on y = 0
(B-28)
(2) for a couple
(B-29)
The stress and deflection of the beam for a line load, p, and a couple, Nh, can be evaluated
based on elementary bending behavior. Equation (B-30) shows stresses, and Equation (B-31) and
Equation (B-32) are used for beam deflections.

x
N
m
t 4 b' 3 + ( ) + ( )
------------------------------------------ 3 1 + ( )y + { }e
y
x sin
m 1 =

y
N
m
t 4 b' 3 + ( ) + ( )
------------------------------------------ 1 1 + ( )y { }e
y
x sin
m 1 =

=
where m 2a =
p P
m
x sin
m 1 =

=
w
p
2a
3
P
m
m
3

3
---------------- 1 y + ( )e
y
x sin
m 1 =

=
Nt
2
------
w
m
ta
2
N
m
2m
2

2
---------------- 1 y + ( )e
y
x sin
m 1 =

=
B-10
(B-30)
(B-31)
(B-32)
To solve the two unknown coefficients, Pm and Nm, the geometric condition and strain
compatibility condition on the line of y = 0 were used expressed in Equation (B-33) and Equation
(B-34). Because strain compatibility was evaluated at the interface surface, not in the mid-surface,
strain and stress due to plate bending motions had to be included. The resulting effective width is
defined as Equation (B-35).
(1) deflection of the beam and the plate are the same
(B-33)
(2) the strain at the lower surface of the slab is equal to the strain of the top of the beam
(B-34)

x
N
m
A
s
--------- x sin
m 1 =

=
w
p
16a
4
P
m
m
4

4
E
s
I
s
----------------------- x sin
m 1 =

=
w
m
8a
3
N
m
h
m
3

3
E
s
I
s
----------------------- x sin
m 1 =

=
w
p
w
m
+ ( )
beam
w
p
w
m
+ ( )
plate
=

x N ( )

x p ( )

x m ( )
+ + ( )
beam

x in plane ( )

x p ( )

x m ( )
+ + ( )
plate
=
B-11
(B-35)
Effects from the resultant force, Q, are the focus of Adekolas (1968) study of effective
width of composite beams. As shown in Figure B-4, resultant force, Q, was used in the equilib-
rium condition instead of interaction force, p. The plate displacement function was considered
only for uniformly distributed loading in Equation (B-22). Basically, his approach was very simi-
lar to the Allens except that equilibrium conditions were as specified in Equation (B-36).
(B-36)
The strain of plate at the mid-surface, e(ms), was calculated based on the strain of top of
the beam, e(j), and assumed linear variation across the section as shown in the Figure B-4. Finally,
the same form as Equation (B-26) was used to solve four unknown coefficient, A
n
, B
n
, P
n
, and Q
n
as in Equation (B-23), with Equation (B-36) and geometric boundary conditions such as Equation
(B-24). Effective width was then calculated based on Equation (B-37), which is the same as Equa-
tion (B-35) except that it excludes the beam width.
(B-37)
Fan and Heinss study is also based on the same governing equation as in Equation (B-22),
but they introduced orthotropic plate theory in their formulation (Fan and Heins 1974). Therefore,
the two governing equations were modified as shown in Equation (B-38):
2b
2t
x
y d
0

b't
x
y 0 =
+
t
x
y 0 =
------------------------------------------------------ =
Q
s
Q
c
+ p
1
x =
M
c
tF Q
c
x + 0 =
M
c
M
s
h F Q
c
Q
s
+ ( )x + + 0 =
\

|
b
2
x
y d
0
b 2

x
y b 2 =
-------------------------- =
B-12
(B-38)
where
The procedures to determine unknown coefficients are the same as other approaches, but
they used a more detailed free-body diagram to investigate equilibrium as shown in Figure B-5.
Equilibrium specified in Equation (B-39) with compatibility conditions between left and right
plate, Equation (B-40), were used in their study. Equation (B-41) derived from strain compatibil-
ity at the interface between beam and slab, was also used as other studies did.
(B-39)
1
E
y
-----
F
4

x
4

--------
1
G
----

x
E
x
-----

y
E
y
-----
\
| |
F
4

x
2
y
2

----------------
1
E
x
-----
F
4

y
4

-------- + + 0 =
D
x
w
4

x
4

--------- 2H
F
4

x
2
y
2

---------------- D
y
w
4

y
4

--------- + +
q N
x
w
2

x
2

--------- N
y
w
2

y
2

--------- 2N
xy
F
2

x y
----------- + + + =

x
t
F
2

y
2

-------- =
y
t
F
2

x
2

-------- =
xy
t
F
2

x y
----------- =
D
x
E
x
t
3
12 1
x

y
( )
------------------------------- = D
y
E
y
t
3
12 1
x

y
( )
------------------------------- =
H D
1
2D
xy
+ = D
1

y
D
x

x
D
y
= = D
xy
Gt
3
12
-------- =
F
x
0 = N
x
r
N
x
l
0 =
F
y
0 = N
xy
r
N
xy
l
+ S
y
=
F
z
0 = p R
x
r
R
x
l
+ N =
M
x
0 = m
t
2
--- S
y
=
M
y

0 = M
x
r
M
x
l
T =
B-13
(B-40)
(B-41)
They used resultant compressive force, Equation (B-42), which was calculated between c1
and c2, where the sum of forces between them became equal to the tensile force in the girder for
defining effective width of flange. The longitudinal stress in the middle surface, not in the inter-
face, was used in Equation (B-42). Based on numerical analyses utilizing the finite difference
method, they produced simple equations for effective width that are specified in Equation (B-43)
and Equation (B-44), in which b represents the girder spacing.
(B-42)
(1) Interior Girders
(B-43)
w
r
w
l
=
w
r

x
--------
w
l

x
-------- =
v
r
v
l
=
u
r
u
l
=

y
( )
beam

y
( )
plate
=

y
( )
beam

y
( )
bending

y
( )
axial
h
2
---
y
2
2

F
AE
------- + = + =

y
( )
plate

y
( )
in plane

y
( )
bending
1
t
---
N
y
E
y
------
x
N
x
E
x
------
\
| |
t
2
-- -
y
2
2

w
= + =
2b
e
F
g

y
( )
max
------------------ =
F
g

y
x d
c1
c2

=
2b
e
b
-------- 0.617
b
L
--- 0.702 + = 2b
e
slab width
No. of girders
--------------------------------- >
B-14
(2) Exterior Girders
(B-44)
Pantazopoulou and Moehle (1990) introduced higher-order terms in the description of the
transverse strain distribution as shown in Equation (B-45), which has the profile in Figure B-6.
Based on the virtual work principle equating internal work evaluated based on assumed strain
profile with external work, the relation between unknown coefficients were developed. Especially
for the negative bending moment region of building structures, they concluded that an effective
width of flange should be approximately equal to 1.5 beam depths on each side of the web for
preyielding response, and equal to 2.0 beam depths for moderate post-yielding response.
(B-45)
Song and Scordelis (1990) developed series solutions of stress functions for top flange,
bottom flange, and edge, which can be applied to T-, I-, and box beams with various loading con-
ditions including axial forces and continuous structure. The following Equation (B-46), which
was derived for simple I- or box beams under various loadings, was the result of simplifications of
analytical results. For the T-beams, the equivalent I-beam is used for the effective width evalua-
tion.
(B-46)
where
b
e
b
----- 1.98 ( )
b
L
--- 0.873 + = b
e
slab width
No. of girders
--------------------------------- <

xx
x y z , , ( )
xx
x 0 0 , , ( ) c
0
z c
1
y c
2
z
2
c
3
zy c
4
y
2
+ + + + + =
3.77l
1.9
1 3.1 99l
3
( ) 0.5 x
1.5
+ | ] + = for l 4 ( )
0 = for box or I-beam with l 30 >
0.088 0.0455l
0.2
= for I-beam with l 30
B-15
Before computers could be used for analysis of partial differential equation, the main con-
cern was their simplification and solution based on hand calculation. Since the 1960s, series solu-
tions for slab-beam composite problems can be evaluated for more general types of loading and
boundary conditions. However, it is only after development of rigorous numerical methodologies
such as the finite element method, finite strip method, that these problems can be solved for more
realistic loading and structural systems.
1
b
e
b
----- = l
L
b
--- =
L total length of beam
b the half-width of the flange from the edge of the web
b
e
the effective flange width
x is distance from the left support to the section
B-16
Figure B-1: Concept of Effective width
z

x
y
xy

xy

N
N
M,M'
e
Cw
Cs
h
x
z
x,u
y,v
l
B-17
Figure B-2: Force Boundary Conditions (Allen & Severn 1961)
Figure B-3: Force Boundary Conditions (Lee 1960)
h
M
N
F
N+dN
F+dF
q1
h
d
b
M+dM y

xy h y
Qxy
Mxy
a
a
x, u
y,v
z,w
b'
Beam
Slab
AsIs Es
D Ec
h
x
y
2a
t
N
Nt/2
Nh
Nt/2
N
N N
Nh
F
p
n
n
p
z,w
B-18
Figure B-4: Force Boundary Conditions (Adekola 1968)
Figure B-5: Force Boundary Conditions (Fan & Heins 1974)
Beam
Slab
h
n
n
x
y
2t
b
F
Mc
Qc
p1
F+dF
Qc+dQc
Mc+dMc
F
Ms
Qs
F+dF
Ms+dMs
Qs+dQs
dx
N.A.
e(j)
e(ms)
dc
t
x
y
z
Nx(l)
Mx(l)
Rx(l)
Nxy(l)
Sy
Rx(r)
Nx(r)
Mx(r)
Nxy(r)
m
T
N
P
h
Sy
T
m
N
B-19
Figure B-6: Distorted cross-section
C-1
APPENDIX C
AASHTO SPECIFICATION AND HISTORICAL BACKGROUND
AASHTO LRFD Specifications (AASHTO 1998) specifies effective flange width of com-
posite bridges in section 4 - Structural analysis and evaluation [4.6.2.6.1]
3
. In the absence of a
more refined analysis and/or unless otherwise specified, the width of a concrete slab shall be
determined as specified. In particular, the specified effective width can be used for determining
resistance for all limit states. These rules have appeared since the 1944 Specifications, which used
the allowable stress method in general. During migration to the load factor method and load and
resistance factor method, the same rules have been used with extended applicable scope so that
they are now used not only for stress calculation but also for capacity evaluation.
To distinguish interior girders from exterior girders, their differences are explicitly
described in the specifications as follows:
For Interior beams, the effective flange width may be taken as the least of:
- One-quarter of the effective span length;
- 12.0 times the average thickness of the slab, plus the greater of web thickness or one-half
the width of the top flange of the girder; or
- The average spacing of adjacent beams.
For Exterior beams, the effective flange width may be taken as one-half the effective
width of the adjacent interior beam, plus the least of:
- One-eighth of the effective span length;
- 6.0 times the average thickness of the slab, plus the greater of half the web thickness or
one-quarter of the width of the top flange of the basic girder; or
3. Brackets are used to specify the part of specifications.
C-2
- The width of the overhang.
Even though the differences of effective width due to loading types have been studied
(Hambly 1991), i.e. uniformly distributed and concentrated loads, the differences from the differ-
ent loading conditions are not recognized in the Specifications, except of the case for normal
forces, which are applied on cross section due to post-tensioning and other concentrated forces
[Figure 4.6.2.6.2-4]. To consider negative moment regions, Specification defines the effective
span length as the total span length for simply-supported span and the distance between points of
permanent load inflection for continuous spans. Based on the effective span length, the same rules
are applied for negative moment regions.
Even though effective width rules are defined for determining resistance for all limit
states, it can be inferred from the definition of effective span that the Specifications mainly con-
cern flexural behavior. Effective flange width for axial forces that is needed in composite cable-
stayed bridges, for example, may not be based on these rules. Actually, AASHTO Standard Spec-
ification uses the expression ... effective width of the slab as a T-beam flange..., which limits
their application to flexural behavior (AASHTO 1996). Following is a brief historical background
of U.S.-based effective flange width rules.
As one of the early studies, Talbot used 4 different flange sizes in a study of concrete T-
beams (Talbot 1907). Even though he understood the shear lag phenomenon of the concrete deck,
he concluded that the changes of compressive stress across the flange were not so significant and
the changes could be ignored for T-beams with the ordinary amount of reinforcement
4
. His con-
clusion is as follows:
4. This experiment was done to evaluate ultimate capacity of T-beams. Thus the conclusion does not necessarily also
apply to the service limit state.
C-3
It seems clear from the general behavior of these beams that for calcula-
tions on strength the T-section may be considered to be the equivalent of
rectangular beams of the size of the inclosing rectangle for widths of
flange equal at least to four times the width of stem. ... It seems proba-
ble that this relation may be applicable to even greater widths of flange.
However, the actual value of this limit cannot be of great practical impor-
tance, since a greater width would not materially change the value of the
calculated resisting moment, and since the amount of steel will at any rate
be limited by the space in the stem and by practical considerations in plac-
ing and bedding it to an amount which will keep the maximum compres-
sive stress within a reasonable limit
Based on other studies and engineering practices, the Final report of the special commit-
tee on concrete and reinforced concrete (ASCE 1916) introduced the notion of effective width of
concrete T-beams as the minimum of one-fourth of the span length of the beam and twelve times
the thickness of the slab. Some researchers had objections to this rule, and Mensch et al discussed
that
5
:
... the writer would like to know the experimental facts on which the
rules for the width of slabs in T-beams are based. ... The tests of the
French and German Committees show that one can count on a wider slab
acting together with the stem. (Mensch, Bergendahl, and Martin 1917)
At this time, several empirical rules of effective width for T-beam design existed due to
inexperience and insufficient knowledge of concrete material and behavior. These rules used as
the principal parameters span length, girder spacing, the width of stem, and the thickness of slab.
Some of them show consistency with each other, but quite different rules were also used (Ferro-
5. Dr. Talbot also said that wider flange might be used as effective width, even though he presented no experimental
evidence for that assertion.
C-4
Concrete 1910; Preliminary Report of the Committee on Reinforced... 1916; Proposed Stan-
dard Building Regulations for the... 1917; Goldbeck and Smith 1916; Unwin 1911; Waldram
1913). After 1/4 of the girder length and 12 times the thickness of slab were published, however,
such variations of effective width rules could not be found. It may be concluded that these rules
were successfully applied for engineering practice since they were adopted by several specifica-
tions.
Some research indicated that the effective width specification related to slab thickness had
been taken from earlier specifications concerning protection against crippling or buckling in steel
compression members (Committee 1974; Vallenilla and Bjorhovde 1985). However, from the
early T-beam studies, including Talbots work, first yielding of reinforcement was intended in
design. Thus, it can be inferred that buckling criteria of steel flange was not considered as a major
concern in concrete T-beam design. Instead, it appears that thickness criteria for effective width of
concrete T-beams was developed to apply to rectangular concrete beam design methods, which
was verified through early experimental work.
As the first specifications for steel and concrete composite beams in US, Specification
for the Design, Fabrication and Erection of Structural Steel for Buildings (AISC 1936) and
Standard Specifications for Highway Bridges (AASHO 1944) adopted the first provisions for
composite beams. In the 1950s, 1956 interim revisions of the 6th edition of the AASHO Specifi-
cation (AASHO 1953), which were comprehensive provisions for the design of composite
bridges, were adopted. In its 1961 specification (AISC 1961), AISC adopted a completely new,
greatly increased edition of its specification
6
. However, there was no significant change in the
6. I.M. Viest, Studies of Composite Construction at Illinois and Lehigh; 1940-1978, Composite Construction in Steel
and Concrete III (ASCE, 1997).
C-5
definition of effective width since its first provisions. Since 1944, AASHO Specifications have
used the same definitions of effective width: the minimum of one-fourth of the span length of the
beam, the distance center to center of beams, and twelve times the least thickness of the slab,
which are also the same for the concrete T-beams.
However, applying the definition of effective width for general concrete T-beam structures
to bridges, especially to composite bridges, can be inappropriate in several respects. Firstly, steel
behavior is more well-understood than concrete behavior. Therefore, effective width definition for
rectangular concrete beam design may not be applicable to design of composite structures. Sec-
ondly, different construction procedures are used for concrete T-beams than for composite struc-
tures. Generally, unshored construction is used for composite structures. Composite action is only
applicable for some part of dead load and live load, not for full loads. Therefore, effective width
for live load is the major concern for composite structures. Thirdly, wider girder spacing with or
without deeper girders is usually used for bridge construction. This can cause a more severe
shear-lag effect in the slab, because relatively thin plate is involved compared with general con-
crete T-beam structures.
As described, the current specification provisions for effective width of composite beams,
which emerged from effective width definition for concrete T-beams and not changed since the
1910s, are not based on reliable theories, even though they have been safely applied for engineer-
ing practice because of their conservatism. Effective width rules in AASHTO LRFD Specifica-
tions have a long history of successful application. The beauty of these rules is simplicity, not
accuracy. However, as seen in the effective width rules for segmental concrete box beams
[4.6.2.6.2], simplicity cannot be the only requirement anymore.
C-6
In contrast to AASHTO Specifications, there were some changes in AISC rules. The same
rules defined in the ACI code were used for composite beams prior to the 1986 LRFD Specifica-
tion; however, the new AISC rules do not use slab thickness and beam flange width any more
7
.
For concrete T-beam design, the effective width concept for applying the same design method as
in rectangular concrete beams is currently used, and most of the concrete design specifications
includes slab thickness related definition among them (Loo and Sutandi 1986). The ACI code
(ACI 1989) has long used the effective flange widths for interior T-sections as the minimum of (1)
L/4, (2) bo - for equal beam spacing, and (3) bf+16ts.
7. C.G. Salmon and J .E. J ohnson, Steel Structures: Design and Behavior (4th Ed.) (HarperCollins College Publishers,
1996).
D-1
APPENDIX D
EFFECTIVE FLANGE WIDTH OF FOREIGN SPECIFICATIONS
British (BSI 1982) (BSI 1979)
The British Standards Institution publishes BS5400 Steel, concrete and composite bridges
specifications as its bridge standard. This describes the application of the limit state principles
adopted and includes sections on analysis and foundation design. Among them, effective width
for composite bridges is specified in Part 3. Code of practice for design of steel bridges and
Part 5. Code of practice for design of composite bridges. Following part 3, the effect of shear
lag in flange may be neglected for ultimate limit state [Part 3: 9.2.1.3]
8
, but not in the serviceabil-
ity limit state [Part 3: 9.2.3.2]. To consider shear lag phenomenon in stress calculation by elastic
theory, effective width (effective breadth) of flange is defined, assuming that there is full interac-
tion between the steel beam and the concrete in compression [Part 5: 5.2.2].
The following discussion foreign specifications focus considerations on (1) interior vs.
exterior girder; (2) uniformly distributed load vs. point load; (3) positive moment regions vs. neg-
ative moment regions; (4) additional considerations.
(1) Interior vs. Exterior Girders
For portions between webs, effective width ratio is multiplied by half the distance between
the center lines of webs, i.e. . For portions projecting beyond an outer web, effective
width ratio is multiplied by 85% of the distance, b, from the free edge of the projecting portion to
8. Brackets are used to specify the part of specifications.
b
e
b =
D-2
the center line of the outer web i.e. . The total effective width of flange should be
taken as the sum of the effective widths of the portions of flange considered separately on each
side of that web [Part 5.2.3.1].
(2) Uniformly Distributed vs. Point Loads
Different ratios are used for uniformly distributed and point load in part 5, however, as the
effective width ratios to be used in stress calculations on structural elements subjected to standard
highway or railway loading, it is specified that ratios from uniformly distributed cases should be
used [Part 5: 5.2.3.2].
(3) Positive vs. Negative Moment Region
Three tables in Part 5 distinguish among simply supported, cantilever, and internal spans
of continuous beams as shown in Table D-1, Table D-2, and Table D-3. For each case, effective
width ratios for mid span, quarter span, and support are specified based on the girder spacing and
span length ratio, b/l. For the end span of continuous beams, the ratios from the simply supported
case with span length 0.9l may be used [Part 5: 5.2.3.5]. The effective width ratio at an internal
support is obtained as the mean value of ratios obtained at that support for each span adjacent to
that support. For a cracked concrete flange, the ratio may be modified [Part 5: 5.2.3.7]
(4) The effective width ratios defined here take no account of the increase in shear lag in
the flange of a composite beam due to the presence of longitudinal stiffening members [Part 5:
5.2.3.4]. Also, ratios take no account of any contribution to the in-plane shear stiffness of a flange
that may be made by transverse members connected to it [Part 5: 5.2.3.6].
b
e
0.85b =
D-3
The first British code of practice for composite bridge beams appears in CP117. Compos-
ite Construction in Structural Steel and Concrete, Part 2: Beams for Bridges in 1967 (J ohnson
1970). It took account of the AASHTO, German code, and other codes, and included new features
based on research of the CP 117 Committee. The effective widths for concrete flanges proposed in
this code are based on theory and tests done by the Committee and lie between the values given in
the German code DIN 1078 for simply-supported and for continuous beams. The current specifi-
cations in BS 5400 described above are based on the parametric study of the shear lag phenome-
non for steel box-girder bridges (Moffatt and Dowling 1978). During this study, finite element
analysis and experimental research were performed to verify the proposed table and methodolo-
gies. In fact, there are two rules for effective width in BS. One of them is described above for the
most general case of design, and the other is specified in Appendix A in BS 5400 part 5 for more
specific loading and support conditions to cover all possible combination of parameters. J ohnson
describes research efforts for composite structures in the U.K. with comments on the Eurocode
(J ohnson 1997).
Testing of composite structures had been conducted since 1920s. Shear connectors were
used in the early 1930s and some composite bridges were built in Tasmania before 1936. At
Cambridge University, where plastic hinge analysis of steel structures had been developed, study
of composite frame structures for buildings was started. Most structures was designed based on
elastic analysis up until the 1960s. This leads to results strongly influenced by assumptions about
cracking, creep and shrinkage. Researchers then deviated to focus more on ultimate limit states
and the use of welded shear stud connectors. In the UK, design was based on 80% of shear stud
resistance (Pu) rather than 100% because of a castrophic failure.
D-4
The first British code of practice was published in 1965, for simply-supported beams only.
At that time, it was not known whether the shear connection design rules would be unconservative
for the negative moment with wide cracks. Push test was conducted and was found that the
shear resistance of stud shear connection was not significantly reduced by the tensile strain. How-
ever, partial loss of interaction occurred once the slip at 80% of maximum load was imposed.
Therefore, it was concluded that the design strength of studs in the negative moment regions
should be reduced by 20% to compensate for the extra flexibility (i.e., 64% for design). Tension
push tests also provided informations on tension stiffening in cracked concrete slabs, which had
been neglected in the design. However, tension stiffening was allowed in the draft Eurocode for
bridges 30 years later.
In 1960s, the design of shear connection of serviceability (SLS) and fatigue was based on
the classical elastic analysis (=VAy/Ib), but using the envelope of vertical shears V rather than
the distribution due to a particular loading. It was argued in 1971 that full-interaction design for
SLS would always provide enough connectors for ULS (except in a few special situations) and
Part 5 of BS 5400 (1978) was drafted on this basis.
For vertical shear, in 1965, BS 5400 had assumed that the steel web alone resists the verti-
cal shear. Moment-shear interaction in composite beams of compact section was studied in 1968.
Tests have shown that the slab resisted significant vertical shear, even if cracked, except in the
region close to the support where its reinforcement had yielded. In that region, the steel element
did not transfer the entire shear, but if the web is compact enough not to buckle, interaction
between resistance to bending and shear can be neglected.
D-5
Canada (CSA 2000)
The current Canadian Highway Bridge Design Code adopted the previous Ontario High-
way Bridge Design Code and uses the same specifications for effective width of flanges. In calcu-
lating bending resistances and bending stresses in slab-on-girder bridges and box girder bridges
having a concrete slab, whether girders are steel or concrete, a reduced cross-section defined by
effective width formulation shall be used [5.8.2.1]. The reduced section is used not only for the
service limit state but also for the ultimate limit state. The effective width consists of a lefthand
overhang, a central portion, and a right-hand overhang. The overhang, Be, shall be determined as
follows:
where
(1) Interior vs. Exterior Girders
The same formula will be used for interior and external girders. For interior girders, half
of girder spacing is used for calculation, and the distance between girder web and free edge is
used for external girders as shown in Figure D-1.
(2) Uniform vs. Point Loading
B
e
B
----- 1 1
L
15B
----------
)
`

3
= for
L
B
--- 15
B
e
B
----- 1 = for
L
B
--- 15 >
L span, for simply supported spans
length of positive or negative region of dead load moment
B
e
the dimension shown in Figure 2.1
B the dimension shown in Figure 2.1
D-6
In the basic study for delivering this formula design truck loading, Ontario highway bridge
loadings and AASHTO HS-25, were used. Therefore this code is based on point loading. Because
the effective width for point loading is smaller than that of uniformly distributed loading, the
resultant effective width delivers conservative values for ultimate limit state.
(3) Positive vs. Negative Moment Region
When slab reinforcement is continuous over interior supports, or when the slab is pre-
stressed longitudinally, and shear connectors are provided, the factored moment resistance of the
section shall be based on composite section in the negative moment region [10.11.5.3.1 &
10.11.6.3.1]. Different span length is used for positive and negative moment regions (Figure D-2).
For the end span of continuous beams, 80% of span length is used, and for internal span, 60% of
span length is used. The span length for negative moment region is 25% of sum of adjacent span
length, while 20% was used for the previous Ontario code. By this change, the effective width of
negative moment regions is increased, but the effective width for positive moment regions of
internal spans is decreased.
(4) The major advantage of this code approach is simplicity. Only two simple formulae are
used for calculation of effective width, and these formulae are applied for any type of girders, and
for slab-on-girder and box girder bridges.
The formulae given in the code are based on the research done by Cheung and Chan
(1978). In their study, the finite strip method was used for analyzing effective width of a total of
392 theoretical bridges, which were simply-supported structures. These bridges are made by
applying design guidelines used at that time. A few hypothetical bridges were also included in
D-7
that bridge population. As mentioned before, Ontario highway bridge loading and AASHTO HS-
25 loadings were applied to their numerical models. After comparing their results with DIN 1078
and AASHTO, they concluded the AASHTO values can be quite conservative for beam-slab type
bridges with small L/B ratios. Based on the analysis results, they also concluded that variation of
the effective width along the span is relatively uniform, which is different from BS 5400 and
mainly comes from the difference of loading conditions.
Japan (JRA 1996)
In J apan, one-side effective flange width, , is used to obtain the section properties of
beams and/or stringer in calculating stresses and displacements. The following two equations and
Table D-4 are used for calculation of one-side effective width.
(Equation I)
(Equation II)
where
b =
b
l
--- 0.05
1.1 2
b
l
---
\
| |
b = 0.05
b
l
-- - 0.30 < <
0.15 l = 0.30
b
l
---
b =
b
l
--- 0.02
1.06 3.2
b
l
---
\
| |
4.5
b
l
---
\
| |
2
+ b = 0.02
b
l
-- - 0.30 < <
0.15 l = 0.30
b
l
---
= one-side effective flange width (cm)
b = either one-half the center-to-center girder spacing (cm)
or a length of the overhang (cm)
l = equivalent span length (cm)
D-8
(1) Interior vs. Exterior Girders
Effective width of each side of the girder is based on one-side rules. Half of the girder
spacing is used for interior girders and inner side of external girders. The distance between the
girder face and the outer edge is used for outer side of the exterior girder (Figure D-4).
(2) Uniform vs. Point Loading
The J apanese code is developed based on the theoretical value of be/b ratio at the midspan
of simply-supported beams for the two types of moment distributions; one is for uniformly dis-
tributed load and the other is for a point load. Two loading cases make two different be/b ratios
with variation of b/L. When the b/L ratio is less than 0.3, the assigned be/b ratio is between two
values made from two loading conditions. For larger b/L, the point loading case is used. It is
understood that the moment distribution of the stringer is controlled by a concentrated load case
when b/L is increased.
(3) Positive vs. Negative Moment Region
To consider continuous and Ghelber girders (girders with internal hinges), two approaches
are used. The first one is considering different span lengths for different portions of girders as
shown in Table D-4. For example 20% of the sum of adjacent span lengths is used for the negative
bending moment region. The second approach is to use a formula based on triangular moment dis-
tribution, which can be used to represent moment distribution at the internal support of a continu-
ous span. This formula was developed to represent the theoretical effective width ratio for the
concentrated load case.
D-9
(4) As discussed in the Commentary of the code, the effective flange width should vary
along the span. However, the value at the midspan is used for any locations over span length. This
will overestimate the effective width especially at the support following the BS5400 table for the
simply-supported span case.
The same rules are also used in Korea without any modification. Maeda summarized
research of composite construction in J apan (Maeda 1997) and destructive testing results of the
first steel concrete composite bridge in Japan built in 1953 (Maeda and Matsui 1985). The
research, design and construction of simply-supported steel-concrete composite bridges started
early in the 1950s. In 1953, the first composite girder highway bridges, the Kanzaki Bridge in
Osaka and Enoki Bridge in Tokyo, were built (Maeda 1996). In 1959, the Code of Practice for
Design of Composite Girder Highway Bridges was established. The majority of composite
bridges in J apan were built for simple spans. However, the tendency toward continuous composite
girders could be observed. Then, in 1965, the provisions of prestressed continuous composite
girder bridges were added to the Standard Specifications for Highway Bridges. Fatigue tests on
composite beams subjected to negative bending moment was conducted in 1971. The results
showed that if sufficient longitudinal reinforcement in the negative moment region were pro-
vided, premature fatigue failure in the tension flange plate may not occur and the concrete slab
cracking could be restrained within 0.2 mm in width. The J apanese Highway Bridge Specification
(1973) specified that the cross-sectional area of longitudinal reinforcing bars in concrete in ten-
sion shall not be less than 2% of the area of the concrete slab and the ratio of peripheral length of
the reinforcing bars shall not be less than 0.045 cm/cm
2
.
D-10
Although prestressed continuous composite girders and fully-composite continuous gird-
ers without prestressing were adopted in the J apanese Standard Specifications for Highway
Bridges (1973), partially-composite continuous girders were not included in specifications. At
that time, the committee was very nervous about cracking in bridge slabs and did not want to
accept the proposal of non-prestressed continuous composite girder bridges even if this type of
girder can keep a crack width of reinforced concrete slab within 0.2 mm. After a while, the com-
mittee decided to decline the proposal for the non-prestressed continuous composite girders.
In 1982, the J apanese Industrial Standards established a new standard JIS B1198-1982
for headed studs based on the results of the extensive surveys and tests. This standard was also
included in the J apanese Specifications for Design of Composite girder Highway Bridges.
Australia (AS 2327.1:1996)
In 1965, Lysaght developed a profiled steel sheeting that would act compositely as com-
bined formwork and reinforcement with a concrete slab. Hence, two different effective flange
width criteria are specified in AS 2327.1:1996 for evaluating the section strength of (a) solid slab
and (b) composite slab. Solid slab represents the concrete flange that is supported by steel girders,
while composite slab is the combination of the in-situ concrete slab and the profiled steel sheeting
spaning between two steel girders.
Where the concrete flange is a solid slab, its effective width, bef, shall be calculated as the
sum of the distances, bc, measured on each side of the center-line of the steel beam as
shown in Figure D-5, where be is in each case the smallest of:
(a) Lef/8, where Lef is the effective span of the beam
D-11
(b) in the case of a concrete slab with a free edge (i.e. an edge beam situation). Either the
perpendicular distance to the edge measured from the center-line of the beam, or 6
times the overall depth, Dc, of the concrete slab plus half the width of the steel beam
flange, bsf1; and
(c) in the case of a concrete slab which spans between two steel beams (i.e. either an edge
beam or internal beam situation), either half the center-to-center distance between the
steel beams or 8 times the overall depth, Dc, of the concrete slab plus half the width of
the steel beam flange, bsf1.
It must be noted that the Australian code takes into account the slab thickness in the effec-
tive slab width computations.
(1) Interior vs. Exterior Girders
For solid slab [5.2.2.1], each side of effective slab width criteria can be calculated for both
interior and exterior girders. In wide girder spacing bridges, the effective slab width on each side
of the interior girder is larger than the exterior girder. The total effective width for an interior
girder is specified to be 16 times the slab thickness in bridges with very wide girder spacing.
For composite slab [5.2.2.2], the effective width of the portion of the slab above the ribs is
computed the same as for solid slab. However, a multiplication factor needs to be applied in order
to consider the portion of the slab within the depth of the ribs. The multiplication factor varies
depending on the acute angle between the sheeting ribs and the longitudinal axis of the steel
girder.
(2) Uniform vs. Point Loading
D-12
The same rules are applied to both uniform and point loadings.
(3) Positive vs. Negative Moment Region
In continuous bridge girders, the flexural strength design in the positive moment region
can be considered to be simply supported with an effective span [5.3.3]. The effective span of a
composite beam can be considered as the distance between the lines of action of the vertical reac-
tions at the ends of the beam, i.e. bending moment is assumed to be zero. However, there are no
separate provisions for the flexural design in the negative moment region.
Australian research on composite construction intensively began in 1961 at the School of
Civil Engineering in the University of Sydney (Bridge and Patric 1996). The studies covered a
variety of topics such as the behavior of stud shear connectors, single and continuous span girders
subjected to static and cyclic loadings, etc. In 1965, a profiled steel sheeting, known as Bondek,
was developed to be used as a concrete slab formwork. In 1971, a complete review of shear con-
nection was made by Bridge. Those results, together with the values used in American, European
and British specifications, formed the basis for the static stud strengths in Australian Standard AS
2327.1:1980.
In 1982, a complete study of continuous beams including effective widths was made by
Ansourian (1982). The complexity of analytical methods for determining effective width led to
the use of finite element analysis. A number of factors were found to be significant including: the
panel proportions of the slab bounded by girders and their supports; type of loading including uni-
formly distributed slab loading, girder line loading and concentrated loading at midspan; and the
position along the span. In the Australian Code (1980) for simply-supported beams, the simple
D-13
approach of the AISC had been adopted where the effective width was taken as one quarter of the
effective span with limitations on the overhang. For L-sections, it was specified as the steel flange
plus 0.083 times the span. In the code for continuous beams, Ansourian (1984) had proposed to
use the same approach for midspan section, but with 33 percent of width reduction applied at the
interior support or at the location where the loading is highly concentrated. However, composite
action was not allowed at the exterior support.
New Zealand (NZS 3404:1997, NZS 3101:1995)
The Code of practice for bridge engineers in New Zealand is called The Bridge Manual
(2000), which was published in Transit New Zealand in 2000. This document provides many
aspects of the design requirements such as loading criteria, design requirements, etc. However, it
focuses more on the design loadings. The Bridge manual must be used as a guideline document in
conjunction with different other design codes, for instance NZS 3101:1995 - Code of Practice for
the Design of Concrete Structures (1995) and NZS 3404:1997 - Steel Structures Standard (1997).
In section 4 - Analysis and design criteria of the Bridge Manual can be applied to all
bridge types, except steel box-girder bridges. However, the Bridge Manual recommends the use
of different other codes to design box-girder bridges, which are Australian Bridge Design Code:
Section 6 - Steel and Composite Construction (1996) or BS 5400 - Steel Concrete and Composite
Bridges: Parts 3, 5 and 10.
According to NZS 3404:1997, effective slab width criteria are based on the Canadian
Code CAN3-S16.1-M89 (1989) and have been calculated from consideration of shear lag effect.
(1) Interior vs. Exterior Girders
D-14
The effective slab width for the interior girder is 20 percent larger than the exterior girder
for short-span bridges.
(2) Uniform vs. Point Loading
The same formula applies to both uniform and point loadings.
(3) Positive vs. Negative Moment Region
There are no specific requirements for the design of composite sections in the positive and
negative moment regions.
Eurocode 4 (Eurocode 4 1992)
Eurocode 4 is one of several series in Eurocode collection that deals with the design and
construction of composite girders (Kindmann 1990). The first draft of Eurocode 4, Design of
composite steel and concrete structures, Part 1.1 was published in 1983/1984 based on a Model
Code prepared between 1971 and 1980 by the IABSE/CEB/ECCS/FIP J oint Committee for Com-
posite Structures (J ohnson 1992). In the General Rules in the title of Part 1.1, it was stated: The
basic principles apply to all types of composite structures or elements consisting of a steel compo-
nent and a reinforced or prestressed concrete part mechanically interconnected so as to act
together to resist loads. However they do not cover all aspects relevant to special structures,
certain types of bridges . Therefore the decision that there should be a Part 2, Bridges was
made in 1989 (J ohnson and Anderson 1993). Part 2 was expected to deal with particular aspects,
such as fatigue of shear connectors, composite plates, composite box girders, etc. However, publi-
cation of Part 2 was delayed, and the final code is not available until now, except in draft form
(Eurocode 4 1996). Following the General Rules of Eurocode 4, as described above, it is not
D-15
believed that the general rules of effective width in Part 1.1 will be in Part 2. The following dis-
cussion, therefore, is based on the effective flange width rules in Part 1.1.
Generally, action effects may be calculated by elastic analysis, even where the resistance
of a cross section is based on its plastic or non linear resistance. For serviceability limit states,
elastic analysis should be used with appropriate corrections for non linear effects such as cracking
of concrete. When elastic analysis is used, a constant effective width may be assumed over the
whole of each span. This value may be taken as the value beff,1 at midspan for a span supported
at both ends, or the value beff,2 at the support for a cantilever. The total effective width beff may
be determined from the below equation, where b
eff
is shown in Figure D-6 for a typical cross-sec-
tion:
where:
b
0
is the distance between the centres of the outstand shear connectors (Figure D-6) or,
for angles, the width of the connector,
b
ei
is the value of the effective width of the concrete flange on each side of the web and
taken as L
e
/8 but not greater than the geometric width b
i
. The value b
i
should be taken
as the distance from the outstand shear connector or, for angles, the edge of the con-
nector to a point mid-way between adjacent webs, measured at mid-depth of the con-
crete flange, except that at a free edge b
i
is the distance to the free edge. The length L
e
should be taken as the approximate distance between points of zero bending moment.
b
eff
b
0

i
b
ei
+ =
D-16
For typical continuous composite beams, where the design is governed by a moment
envelope from various load arrangements, and for cantilevers, L
e
may be assumed to
be as shown in Figure D-6 in which values at supports are shown above the beam,
and midspan values below the beam. The distribution of the effective width between
internal supports and midspan regions may be assumed to be as shown in Figure D-6,

i
is equal to 1.0 for the effective width b
eff,1
at midspan and for the effective width b
eff,2
at internal supports and cantilevers. For the determination of b
eff,0
at end supports the
value
i
should be determined from the equation given below where b
ei
is the effec-
tive width of the end span at midspan and L
e
is the equivalent span of the end span
according to Figure D-6:
Military Handbook (MIL-HDBK-1002/3)
The Military Handbook (1987) has been developed from an evaluation of facilities in the
shore establishment, from surveys of the availability of new materials and construction methods,
and from selection of the best design practices of the Naval Facilities Engineering Command
(NAVFACENGCOM). The Military Handbook has been used as a reference document for pro-
curement of facilities construction.
Section 2 of the Military Handbook provides the standard design criteria (Class A struc-
tures) for bridge type structures. This provision includes the design of steel beams, girders and
box-girders with composite flanges. For the design of steel highway bridges, the Military Hand-

i
0.55 0.025
L
e
b
i
----- +
\
| |
1.0 =
D-17
book recommends the use of American Association of State Highway and Transportation Offi-
cials (AASHTO) - Standard Specifications for Highway Bridges, Section 10 of Division 1 (1996).
This specification is also recommended for the design of composite bridge structures.
D-18
Table D-1: Effective breadth ratios for simply supported beams (BS 5400)
b/l Loading uniformly distributed over a
length not less than 0.5 l
Point loading at midspan
Midspan Quarter
Span
Support Midspan Quarter
Span
Support
0 1.0 1.0 1.0 1.0 1.0 1.0
0.02 0.99 0.99 0.93 0.91 1.0 1.0
0.05 0.98 0.98 0.84 0.80 1.0 1.0
0.10 0.95 0.93 0.70 0.67 1.0 1.0
0.20 0.81 0.77 0.52 0.49 0.98 1.0
0.30 0.65 0.60 0.40 0.38 0.82 0.85
0.40 0.50 0.46 0.32 0.30 0.63 0.70
0.50 0.38 0.36 0.27 0.24 0.47 0.54
Table D-2: Effective breadth ratios for cantilever beams (BS 5400)
b/l Loading uniformly distributed over a
length not less than 0.5 l
Point loading at midspan
Support Quarter
Point near
support
Free end Support Quarter
Point near
support
Free end
0 1.0 1.0 1.0 1.0 1.0 1.0
0.05 0.82 1.0 0.92 0.91 1.0 1.0
0.10 0.68 1.0 0.84 0.80 1.0 1.0
0.20 0.52 1.0 0.70 0.67 0.84 1.0
0.40 0.35 0.88 0.52 0.49 0.74 1.0
0.60 0.27 0.64 0.40 0.38 0.60 0.85
0.80 0.21 0.49 0.32 0.30 0.47 0.70
1.00 0.18 0.38 0.27 0.24 0.36 0.54
D-19
Table D-3: Effective breadth ratios for internal spans for continuous beams(BS 5400)
b/l Loading uniformly distributed over a
length not less than 0.5 l
Point loading at midspan
Midspan Quarter
Span
Internal
support
Midspan Quarter
Span
Internal
support
0 1.0 1.0 1.0 1.0 1.0 1.0
0.02 0.99 0.94 0.77 0.84 1.0 0.84
0.05 0.96 0.85 0.58 0.67 1.0 0.67
0.10 0.86 0.68 0.41 0.49 1.0 0.49
0.20 0.58 0.42 0.24 0.30 0.70 0.30
0.30 0.38 0.30 0.15 0.19 0.42 0.19
0.40 0.24 0.21 0.12 0.14 0.28 0.14
0.50 0.20 0.16 0.11 0.12 0.20 0.12
Table D-4: One-Side Effective Flange Width (JRA 1996)
Region Effective Flange Width Figure
Symbol, Equation Eqv. Span
length, l
Simply-Sup-
ported Girder
(1) L Eq. I L Figure D-3(a)
Continuous
Girder
(1) L1 Eq. I 0.8L1 Figure D-3 (b)
(5) L2 Eq. I 0.6L2
(3) S1 Eq. II 0.2(L1+L2)
(7) S2 Eq. II 0.2(L2+L3)
(2)(4)(6)(8) Linear interpolation between two end pt.
Ghelber Girder (1) L1 Eq. I L1 Figure D-3 (c)
(4) L3 Eq. I 0.8L3
(2) S2 Eq. II 2L2
(3) Linear interpolation between two end pt.
D-20
Figure D-1: Overhang Width for Various Cross-Sections (CSA 2000)
Figure D-2: Assumed Points of Inflection under Dead Loads (CSA 2000)
D-21
Figure D-3: Equivalent Span Length (JRA 1996)
Figure D-4: One-side Effective Flange Width (JRA 1996)
(1)
L
L
L
L

L

(1)
1 L
L

1 2 . 0 L

(2)
(3)
(4) (5) (6) (8)
(7)
2 L
L

3 L
L

1 L

1 S

2 S

2 L

3 L

2 2 . 0 L

2 2 . 0 L

3 2 . 0 L

(1)
1 L
L

(2) (3) (4)
2 L
L

3 L
L

1 L

2 S

3 L

3 2 . 0 L

(a) Simply-Supported Girder
(b) Continuous Girder
(c) Ghelber Girder
D-22
Figure D-5: Effective Width of Concrete Flange (Australia 1996)
D-23
Figure D-6: Equivalent spans, for effective width of concrete flange (Eurocode 4)
E-1
APPENDIX E
RECOMMENDED EXPERIMENTS
E.1 LABORATORY TEST
Bridge Models
Until the field-test prototype is identified, a two-span continuous bridge described in the
LRFD Design Examples is proposed for this study (HDR Engineering 1995). The bridges in this
publication have been studied by several researchers, and they have served as a key reference for
designing similar types of bridges. This is the main reason of using this particular bridge in this
study.
The two-girder bridge type can be considered as the limiting case of the slab-on-girder
bridge form regarding the wider girder spacing and reduced number of girderlines recommended
for improving cost-effectiveness in such bridges. Although this type of bridge has not been popu-
lar in the U.S., for the experimental prototype, one typical two-girder bridge will be designed for
the proposed laboratory experimentes based on AASHTO LRFD Specification and SETRA
guidelines (SETRA 1990; Raoul and Hoorpah 2001). To compare the performance of the two-
girder bridge, similar geometric parameters to those in the two-span continuous example bridge
will be used.
In this laboratory experimental study, four specimens are proposed as specified in Table
E-1. Table E-2 shows the usage of specimens for each limit state and for each moment region of
interest. The following figures describing specimens and testing schemes illustrate only the 4G
E-2
specimens; those describing the 2G specimens will be analogous. Figure E-1 shows details of the
contemplated quarter scaled 4-girder bridge (4GQT) specimen, and Figure E-2 depicts the con-
templated half scaled 4-girder bridge (4GHF) specimen intended for ultimate limit state test at
negative moment regions.
Loading and Measurements
The characteristics of loadings can be classified as single point load, partial or complete
design truck load, and uniformly distributed load. It has been shown that effective flange width is
very sensitive to loading characteristics and locations. The effective flange width of a uniformly
distributed load is generally larger than that of the point loads. Design truck load is chosen in this
experimental study, because it is more realistic for bridges. Two different loading situations are
considered: complete design truck loading and two rear axles of the design truck. To compare
their differences in terms of resulting stress distributions, numerical analyses were performed
using SAP2000, and the results show variations less than 1% in the large stress regions and less
than 10% in the small stress regions, as shown in Figure E-3. Based on this observation, two rear
axles loading situation is recommended for experimental loading, which can avoid difficulties in
applying several loading positions simultaneously.
Tentatively, four loading positions are selected for service limit test in the positive
moment region. These are shown in Figure E-4. Specifically, there are two sub cases in case 1 and
case 2 to determine the most severe loading position. Their selected positions are also used for
negative moment regions. Cases 3 and 4 are not exact truck loading configurations, but they can
deliver a more regular pattern of stress distribution, which will be advantageous for verification of
numerical models.
E-3
Eight loading positions are considered for the service limit state test in the negative
moment region. Four of them have the same features as the positive moment case except that their
loading position along the span will be varied. Another four loading positions are considered to
maximize negative moments at the interior support by applying additional loadings to the other
span. The loading positions for negative moment regions are shown in Figure E-5. For ultimate
limit state tests, concentrated load is applied at each girder position as shown in Figure E-6. If two
wheel loads are applied, the most probable failure of girder will occur at the girder between two
wheels, which is unrealistic and unpredictable. Therefore, if one concentrated load is applied on
each girder, the bridge will develop a classical continuous span failure mode. Such a failure mode
is more reasonable and reliable for model verification.
Figure E-7 shows the loading scheme for quarter scaled models, which is used for service
limit state tests in both positive and negative moment regions and ultimate limit state in the posi-
tive moment region. Basically, actuators mounted in the strong floor pull down the loading beam
to apply four point loads on the specimen. Simple plastic analysis results for one girder show
1082 kN of applied load at the first plastic hinge formation with 126mm deflection at the loading
position (0.4L). When the second plastic hinge develops, the applied load will be about 1303 kN
with 342mm deflection at the loading position. In ideal laboratory tests, if four girders show the
same failure mechanism at the same time, the required loads and stroke will be 136 kN and
31.5mm at the first hinge formation, and 163 kN with 85.5mm deflection at the second hinge for-
mation.
Figure E-8 shows the contemplated loading scheme for the ultimate limit state in the neg-
ative moment region. The half scaled sub-model is inverted and reaction forces at the internal sup-
E-4
port position will be applied in a similar manner to that used in Figure E-7. Simple plastic analysis
for one girder with two concentrated loads, which are applied at the 0.6L position from each end
of the span, indicated that 1163 kN for each loading position will develop the first plastic hinge
with 1840 kN of reaction force at the internal support, and 1489 kN for each loading position will
be required to develop the second plastic hinge, which results in a 2168 kN reaction force. If the
same failure mechanism occurs for all four girders, and following loading scheme and scaling fac-
tor, the required actuator capacity will be 921 kN for the first hinge formation and 1084 kN for the
second hinge formation.
Basic measurements of these tests are strain and deflection. To acquire a good resolution
for service and ultimate limit state loading, 12 strain gages on the top concrete along one cross
section may be needed. Also, 4 strain gages and one LVDT for each girder at the same section are
needed. Two LVDTs on the loading side should suffice for deflection measurements. Therefore 28
strain gages and 2 LVDTs for each measuring section would be needed. For ultimate limit state
tests, 4 strain gages are added in each girder near maximum moment sections. The tentative loca-
tion of strain gages and LVDT are shown in Figure E-9. For positive moment region tests, 0.2L,
0.4L and 0.6L locations are selected. 0.9L, 1.0L, and 1.1L locations are selected for negative
moment region tests. For negative moment region at the ultimate limit state, not only concrete but
also rebar strains need to be measured. To consider the effects of slip between slab and girder at
ultimate limit state, additional slip measurements are required. However, it is very difficult to pre-
dict the slip location during ultimate loading tests. It is noted that gauge location and/or actuator
locations may need to be fine-tuned to the need to comply with St. Venants Principle.
E-5
Test Procedures and Schedule
The basic sequence of test procedures is service limit state test, ultimate positive moment
test, and negative moment test. For service limit state tests and ultimate positive moment test,
quarter models will be used, while half-scaled models will be used for ultimate negative moment
tests. During positive moment region test, strains and deflections are measured at the 0.2L, 0.4L
and 0.6L positions, with four different loading positions. The final loading position for case 1 and
case 2 needs to be decided for negative moment tests. Responses for negative moment regions are
measured at 0.9L, 1.0L, and 1.1L positions. Firstly, loading cases 1 through 4 are applied to span
1 where responses are measured. Secondly, loading cases 5 through 8 are applied, locating a mass
block in span 2 to maximize negative moment at the section. After service limit state tests are fin-
ished, ultimate loading which develop the maximum positive moment is applied with the ultimate
loading position configuration. For ultimate negative moment tests, half-scaled sub-models are
used. The above procedures are summarized in Table E-3, and the following is the test procedure
considering specimens and limit state.
a. 4GQT: Service limit
b.4GQT: Ultimate at Positive moment region
c. 2GQT: Service limit
d. 2GQT: Ultimate at Positive moment region
e. 4GHF: Ultimate at Negative moment region
f. 2GHF: Ultimate at Negative moment region
Table E-4 shows the tentative schedule according for one specimen. The schedule indi-
cates a duration of 36 weeks for 4 specimens, approximately.
E-6
E.2 FIELD TEST
Strain Measurements
Steel Girder
Flexural strains of the steel plate girders can be monitored using strain gages mounted on
the top and bottom flanges. At each instrumented section of each plate girder, two gages will be
mounted on the upper surface of the top flanges, and two gages on the lower surface of the bottom
flange. The top flange gages will be placed near the edges of the flange plate. The bottom flange
gages will be placed as close as possible to the web plate.
Longitudinal strains in the web can be monitored by attached one strain gage on each side
at mid-height of the web. Strains induced by the out-of-plane deflections of the web are likely to
be very small compared to the longitudinal strains. Strain gages in the vertical direction would not
be required.
Concrete Slab
Strain gages will be mounted on the concrete deck slab, along the transverse cross-sec-
tions. Deck slab gages should be arranged as 90
o
rosettes on long-gage-length transducers, with
individual gages placed in the longitudinal and transverse directions, respectively. Strain gages
will be attached to the top and bottom surfaces of the deck slab with protection against weather
and traffic. The surface of concrete deck is a preferable location for strain gaging purposes
because it is where the maximum compressive strains would occur. Large variations in strain
reading usually take place where the measured strains are small. Hence, more accurate strain data
can be obtained at the concrete deck top surface.
E-7
Displacements
Vertical deflections are important measurements for the finite element model verification
process as they are the most reliably accurate results. The reason for this is that displacement is a
direct measurement compared to strain measurements, which are the first derivative of the dis-
placement. Horizontal displacement can also be measured for any steel-concrete interface slip-
page be locally removing stay-in-place forms.
Results and Comparisons
As the measured longitudinal strains are directly comparable, they will be thoroughly
investigated to evaluate how much the deck slab participates in the longitudinal bending actions.
The concrete slab can be initially assumed to behave elastically at the serviceability limit state.
Further assuming the longitudinal strain distribution to be linear between the bottom flange of the
girder and the top surface of the deck, the longitudinal strain at deck top can be calculated by
extrapolation from known girder flange strains. Strains at the longitudinal deck gage locations,
meanwhile, can be directly measured. The number of strain gages installed on the deck top must
be sufficient in the region near the girder locations and where the maximum positive and negative
moments are, in order to capture the effects of any shear-lag phenomenon on the girders. The
effect of lateral position of trucks can be investigated by placing wheel loads at different locations
in the transverse direction.
In 1987, the I-78 highway bridge over Delaware River was monitored for determination of
stresses and deflections developed during construction and under traffic loads after completion.
The primary purpose of the study was to use stress data based on measured strains to evaluate the
E-8
adequacy of several commercially available finite element analysis programs which were being
used for bridge superstructures analysis and design.
Results and comparisons were documented in final report (Yen et al. 1995) with a fair
amount of information for the purpose of finite element model verifications. Location of neutral
axes was reported at different construction stages. Strains of girders and concrete deck were mea-
sured as well as longitudinal strain distribution across the transverse direction.
Unfortunately, original data is no longer available. Instrumentation plan can be shown sys-
tematically in Figure 2-6 and Figure 2-7 for strain gages and LVDTs.
Field test results will be compared with the results from laboratory scaled model testing
and analytical results from finite element analysis. Finite element model calibration will be done
when necessary. Consequently, the finite element model will be modified to serve the purpose of
effective slab width studies.
E-9
E.3 BUDGET FOR PROPOSED EXPERIMENTATION
Contingency Testing: Budget Discussion and Justification
Both field and laboratory experiments are recommended, as discussed earlier. It is
assumed that a widely-spaced-girder bridge appropriate for field testing will not be available near
UB, so that travel to a possibly distant site is required. The budget estimate for this work is as fol-
lows:
Additional faculty time and laboratory staff technician time, including fringe benefits: $29,000
Additional graduate student and part-time undergraduate student time, including fringe benefits:
$24,000, plus $8,500 tuition,
Supplies, specimen preparation and disposal, and equipment usage including data acquisition for
laboratory experiments: $28,000
Field experiment subcontract, site travel, and customized preparations for field experiments:
$17,500
Indirect Costs: $49,000
Total additional: $156,000
Additional faculty time (Chen and Aref) is budgeted at approximately 10% effort for each.
This is required not just for the conduct of the experimental work itself but also to ensure close
coordination of that work with the companion analytical/numerical studies.
Subcontract costs are estimated utilizing the services of Bridge Diagnostics, Inc. for the
following aspects of the field work:
Instrumentation
E-10
Data acquisition for passages of test trucks
Adjustment/calibration of initial model parameters to match test results
Reporting of field results and provision of all field test data for use by the UB team.
Reasons for exceeding the original estimated budget amount ($156,000 vs. $120,000) for
contingency testing stem primarily from three sources:
A new university policy requiring cost recovery of tuition for graduate research assistants if per-
mitted by the sponsor,
New laboratory cost recovery policies coming into effect after 15 November 2001, and
The need to construct new scale-model laboratory specimens rather than utilizing the existing
full-size girders described in the original proposal, due to limitations in laboratory equipment
capacity.
E-11

Table E-1: Specimens
Specimens Prototype Scale
4GQT LRFD Example (4 girders) 1:4
4GHF LRFD Example (4 girders) 1:2
2GQT Two-Girder Bridge 1:4
2GHF Two-Girder Bridge 1:2
Table E-2: Usage of Specimens
(+) Moment Region (-) Moment Region
Service Limit State 4GQT, 2GQT 4GQT, 2GQT
Ultimate Limit State 4GQT, 2GQT 4GHF, 2GHF
Table E-3: Test Sequences
Sequence Specimen
Limit
State
Positive/
Negative
Loading
Measurement
(ratio to L)
1 4GQT Service Positive Figure E-4 0.2,0.4,0.6
2 4GQT Service Negative Figure E-5 0.9,1.0,1.1
3 4GQT Service Negative Figure E-5 0.9,1.0,1.1
4 4GQT Ultimate Positive Figure E-6 0.2,0.4,0.6
5 4GHF Ultimate Negative Figure E-6 0.9,1.0,1.1
E-12
Table E-4: Tentative Experiment Schedule for Each Specimen
Week 1 2 3 4 5 6 7 8 9
Form Work
Reinforcing Steel
- Strain Gaging
- Cage Tieing
Concrete
- Casting Set-up
- Pouring
Test Set-up
- Instrumentations
- Data Acquisition
Testing
- Concrete Cylinder
- Loading Tests
E-13
Figure E-1: Plan and Cross section of Specimen 4GQT
260 260
(Unit : mm)
3@750=2250
2770
6
7
5
0
m
m
6
7
5
0
m
m
1
3
5
0
0
m
m
G
i
r
d
e
r

A
G
i
r
d
e
r

B
G
i
r
d
e
r

C
G
i
r
d
e
r

D
E-14
Figure E-2: Plan and Cross section of Specimen 4GHF
Girder A
Girder B
Girder C
Girder D
(Unit : mm)
3750mm
7500mm
3750mm
520 520 3@1500=4500
5540
E-15
Figure E-3:Comparison of Stresses between 2-wheel and 3-wheel Loading
E-16
Figure E-4: Loading Position for Service Limit State at Positive Moment Region
SPAN1
2
7
0
0
m
m
1
0
7
5
m
m
Internal Support
1075mm
450mm
9.07kN
9.07kN
9.07kN
9.07kN
Case1(a)
Case1(b)
Case2(a)
Case2(b)
SPAN1
2
7
0
0
m
m
1
0
7
5
m
m
Internal Support
Case3
Case4
E-17
Figure E-5: Loading Position for Service Limit State at Negative Moment Region
SPAN1
1
0
7
5
m
m
2
9
7
5
m
m
SPAN1
1
0
7
5
m
m
2
9
7
5
m
m
1
0
7
5
m
m
2
9
7
5
m
m
SPAN2 SPAN2
Case1
Case5
Case2
Case6
with
Case5 & 6
Case3
Case7
Case4
Case8
with
Case7 & 8
E-18
Figure E-6:Loading Position for Ultimate Limit State at Positive and Negative Moment Regions
E-19
Figure E-7: Loading Scheme for Service Limit State and Ultimate Positive Moment Test
Actuator (1112kN, 203mm
Stroke)
6
7
5
0
m
m
450
2
7
0
0
m
m
3
7
7
5
m
m
L
o
a
d
i
n
g
B
a
r
Support
Strong Floor
Internal Support
E-20
Figure E-8: Loading Scheme for Ultimate Negative Moment Tests
Actuator (1112kN,
203mm Stroke)
Strong Floor
Support
Loading Beam
Bottom Flange
Deck
Girder
S.S. S.S.
Loading Beam
E-21
Figure E-9: Location of Strain and Displacement Measurement
0.2L
0.4L
0.9L
1.0L
1.1L
6
7
5
m
m
6
7
5
m
m
Girder Section
at 0.4L, 1.0L at other pt.
Strain Gage
Displacement gage
Negative Moment Test : 0.9L, 1.0L, 1.1L
1
3
5
0
m
m
1
3
5
0
m
m
1
3
5
0
m
m
0.6L
2
0
2
5
m
m
Positive Moment Test : 0.2L, 0.4L, 0.6L
F-1
APPENDIX F
PLANS FOR NCHRP 12-50 COMPLIANCE
This part of the project briefly addressed the plan for the use of the most recent research of
National Cooperative Highway Research Program No.12-50. NCHRP 12-50 is a research project
to investigate current software validation procedures and to provide an improved method of veri-
fying bridge design and analysis software related to LRFD bridge design.
The bridge analysis and design process can be divided into smaller computational
domains, called Subdomains, such as dead load distribution, live load distribution, live load
actions on a girderline and resistance and specification checks for the girderline. A number of
bridge software packages are used for comparing the results to one another. The explanations for
result differences are also given.
It is recommended that specification writers may use the proposed process to investigate
the effect of a change during the development of the specifications. In other words, they can
investigate a large number of bridges be investigated to help formulate the specification language,
intent and application (commentary) with an awareness of the impacts that specification changes
will have.
In order to incorporate this proposed software-validation procedure into this project, i.e.
NCHRP 12-58, the part of the research scheme should be described for the ease of understanding
parametric study will be employed as a tool to formulate and explore impacts of new proposed
effective slab width criteria. Different cases will be considered to ensure the entire populations are
F-2
completely covered within the well-defined scopes and limitations. Statistical methods, such as
regression and sensitivity analyses, will be employed to explore effects of candidate equation(s)
for effective slab width criteria.
By using commercial bridge engineering design software such as OPIS (which is already
installed and operational at UB), the proposed effective slab width criteria can be applied as a
user-defined code provision. A base model of a bridge will be designed by the two different effec-
tive slab width design criteria from the current AASHTO LRFD Specification and the proposed
criteria, respectively. A flow chart (see Figure F-1) shows the major components of the compari-
son scheme to evaluate the impacts of the proposed effective slab width criteria to the other
required specifications. Different design parameters will be considered.
Hypothetically, one of the most important parameters which might see some impact from
the change in the effective slab width criteria could be the design of the shear connection. The
ultimate strength of the shear connection is dependent upon the concrete modulus of elasticity
(E
c
). Additionally, a substantial slippage along the concrete-steel interface could be expected at
the Strength I limit state. In addition, the empirical formula used in the AASHTO Specification
regarding the design of shear connection is based on experimental results with the use of normal
strength concrete. As higher concrete strength is introduced in bridge design, i.e. high-perfor-
mance concrete (HPC), the concrete modulus of elasticity might not lie in the specified range of
applicability. Hence, the empirical formula would technically not apply. However, in some cir-
cumstances where fatigue limit state governs shear connection design, the concrete modulus of
elasticity might not contribute any design impacts.
F-3
Another aspect that would be affected from the change in the effective slab width criteria
is the deck design, especially in skew bridges. With the application of the proposed effective slab
width criteria, the girder spacing is expected to be increased. Hence, thicker slabs would also be
required. According to the empirical deck design method in AASHTO Specifications, there are
some limitations which must be satisfied. For a bridge where the skew exceeds 25
o
, the required
reinforcement in both directions shall be doubled in the end zones of the deck.
Therefore, utilizing relevant aspects of the NCHRP 12-50 methodology provides a sys-
tematic means to explore how many other provisions that must be checked to ensure the number
of violated sections are kept to the minimum.
This project is planned to include the validation methodology proposed in NCHRP 12-50
as a guideline to assess proposed code changes. Different design parameters will be studied and
their impacts will be determined.
F-4
Figure F-1: Comparison Scheme
Bridge Engineering
Software
e.g. OPIS, MDX
Specifications

Recent
Specifications

Proposed
Girder sizes

MathCAD, etc.

Fatigue
Shear Conn. Deck Design
Reinforcement
in Skew Deck
Web Stability
Flange Stability
Compact Section Empirical Deck
Design
Inelastic
Deformation


Ductility
Stiffeners
and
Cross-Frames
Local Buckling
Member
Buckling
Deflection at
Serviceability
Limit State
Load Effects
e.g. DFs,
Moment Envelopes
G-1
APPENDIX G
ANNOTATED BIBLIOGRAPHY
G.1 EXPERIMENTAL FINDINGS
G.1.1 Laboratory Experiment
Azizinamini, A., S. Kathol and M. Beacham.1995. Effect of Cross Frames on Behavior of
Steel Girder Bridges, Fourth International Bridge Engineering Conference, Transportation
Research Board. 117-124.
Azizinamini et al. (1995) investigated the performance of steel girder bridges that use dif-
ferent types and spacing of cross frames. The experimental investigation included construction
and testing of a full-scale steel girder bridge in the laboratory. Elastic and ultimate load tests were
carried out, and punching shear tests were conducted after the ultimate load tests. Results of the
research indicate that for bridges with zero skew, the influence of cross frames is minimal. Ulti-
mate load tests indicate that steel girder bridges have large reserve capacities. Very large punching
shear capacity of the slab was also observed.
Bakht, B. and A.C. Agawal. 1995. Deck Slabs of Skew Girder Bridges, Canadian Journal
of Civil Engineering, 22,3: 514-523.
G-2
Bakht and Agawal (1995) conducted the experimental testing of the deck slabs of skew
girder bridges. According to the AASHTO code, minimum requirement for the amount of rein-
forcement must be doubled in the end zones near the skew supports when skew angles are greater
than 20 degrees. This experiment showed that the need for such an increase can be eliminated by
providing composite end diaphragms with high flexural rigidity in the horizontal plane. The
results showed that there was no effect from the skew supports in the region well away from the
supports.
Botzler, P.W. and Colville, J. 1979. Continuous Composite-Bridge Model Tests, Journal of
the Structural Division 105,9:1741-1755.
Botzler and Colville (1979) conducted ultimate-load tests on two continuous span slab-on-
girder bridge models. One of the models was a three-girder two-span structure and the other was a
three-girder three-span structure. Both models were tested to failure with a single concentrated
load applied over the center girder. The purpose of the tests was to determine the variations in
both the distribution of moment throughout the structures and the effective slab width acting with
each girder as the load was increased to failure. In the conclusion, the effective slab widths and
distribution factors at failure loads are significantly different from values obtained at service
loads.
Corrado, J.A. and B.T.Yen. 1973. Failure Tests of Rectangular Model Steel Box Girders,
ASCE Structural Division Journal 99,7: 1432-1455.
G-3
The use of thin-walled steel box girders as the main load-carrying members in bridge
superstructures has recently gained increasingly popularity in this country. Most research on steel
box girders had been concerned with the behavior of the member in the linear range. In 1973,
Corrado and Yen (1973) investigated, through the use of model tests, the post-buckling behavior,
modes of failure and load-carrying capacity of simply supported, rectangular steel box girders.
Two rectangular model box girders were fabricated and tested. The box-girder test specimen con-
sisted of a box section of 76 mm (3 in.) deep by 100 mm (4 in.) wide with a span length of 610
mm (24 in.). The results indicated that failure of the box girder occurred when large vertical
deformation of the flanges followed extensive tension field yielding, again similar to the failure
mechanism of a plate girder. It was also noted that because of the small scale of the models,
extrapolation of the test results to prototype girders such as in bridges should be made with cau-
tion.
Ebeido, T. and J.B.Kennedy. 1996. Girder Moments in Continuous Skew Composite
Bridges, Journal of Bridge Engineering 1,1: 37-45.
Ebeido and Kennedy (1996) tested three two-span bridge models with continuous sup-
ports. One of the bridges was a right continuous composite steel-concrete bridge model while the
other two bridges were skewed at 45 degrees. Strain gages of 10-mm length were installed at sev-
eral critical locations on the steel beams and on the concrete deck slab. Mechanical dial gages
with travel sensitivity of 0.025 mm were used to measure deflections of the steel beams. Test
results from the three tests were used to verify a finite-element analysis for such bridges. The
finite-element analysis was then employed to conduct an extensive parametric study on more than
G-4
600 prototype continuous composite bridge cases. The parametric study was used to deduce
expressions for both span and support moment distribution factors for AASHTO truck loading.
The derived formulas were based on the bridge elastic response which did not take into account
the concrete cracking over the interior bridge piers.
Klaiber, F.W., T.J. Wipf and B.M. Phares. 1997. Precast Units for Simple Span Bridge,
Recent Advances in Bridges Engineering by Merier, U. and Betti, R., 223-230.
Klaiber et al (1997) conducted bridge model tests to study the effects of the shear connec-
tors and diaphragms on load distribution factors. The model bridge was comprised of three 2.13 m
(7) wide precast units. The model girder spacing was 1.1 m (3.5). The results indicated that the
five connector arrangement did improve the distribution relative to the three connector scheme.
However, there was minimal improvement in lateral load distribution when seven and nine con-
nector arrangements were used. The results also showed that diaphragms were ineffective for the
load distribution.
Labia, Y., M.S. Saiidi and B. Douglas. 1997. Full-Scale Testing and Analysis of 20-Year-Old
Pretensioned Concrete Box Girders, ACI Structural Journal 94,5: 471-482.
Full-scale laboratory testing of 20-year-old pretensioned concrete box girders was con-
ducted by Labia, Saiidi and Douglas (1997). The study was to evaluate the structural parameters
in two girders, to determine the actual prestress loss, to study the ultimate flexural behavior of the
girders and to assess the adequacy of codes in predicting the behavior. The girder was preten-
G-5
sioned with thirty 12.7 mm (0.5) diameter Grade 270 (f
u
=1862 MPa or 270 ksi) stress relieved
strands. The loss estimate include elastic shortening, creep, shrinkage and relaxation losses. Two
point loads were applied to the girders to create a constant moment region. Wire displacement
transducers were used to measure deflection. Strain profile LVDTs (linear variable differential
transformers) and crack re-opening LVDTs were mounted across prominent cracks to determine
crack reopening. Crack detection electrical gages were mounted in series on the bottom flange
across the central 3.048 m (10) of the girder in order to detect the first crack. Eight strain gages
were mounted on the prestressing strands. The measured and theoretical deflections at ultimate
limit state were compared.
Lampe, N. and A. Azizinamini. 2001. Steel Bridge System, Simple for Dead Load and Con-
tinuous for Live Load. NaBRO Official Website.
Lampe and Azizinamini (2001) proposed the concept of designing the bridge system as
simple for dead loads and continuous for live loads. By making the girders continuous for love
load only had several advantages. In such cases, the decrease in negative moment coupled with an
increase in positive moment allowed use of the same cross-section for the entire length of the
girder. This potentially could eliminate the need for full penetration welds. This behavior also
made use of rolled sections an attractive alternative for short span bridges. A full-scale specimen
representing a portion of a two span continuous steel bridge system was constructed and a series
of tests were conducted to evaluate its behavior. Test specimens were also prepared for fatigue
tests. The ultimate capacity of the specimen was well above the predicted values using AASHTO
provisions.
G-6
Moore, M. and I.M.Viest. 1993. American Iron and Steel Institute Federal Highway
Administration Model Bridge Test, Transportation Research Record 1380. 9-18.
Federal Highway Administration (FHWA) and American Iron and Steel Institute (AISI)
jointly sponsored a project to conduct a comprehensive research program by Moore and Viest
(1993). The study involved a large experimental test to evaluate the behavior of a 0.4-scale model
of a two-span continuous plate-girder bridge with modular precast prestressed deck panels. The
purpose of this study was to extend inelastic design procedures to continuous girders with non-
compact elements. Two load levels were considered, i.e. serviceability, overload and maximum
load levels. Girder spacing for the test model was approximately 2.1 m (6-10). Load-deflection
response at midspan of girders were also shown in the paper.
Roschke, P.N. and K.R. Pruski. 2000. Overload and Ultimate Load Behavior of Postten-
sioned Slab Bridge, Journal of Bridge Engineering 5,2: 148-155.
Roschke and Pruski (2000) tested three-tenth-scaled prestressed slab bridge models in the
overload and ultimate load ranges. In addition to uniformly distributed longitudinal posttension-
ing, a band of tendons was located in a narrow region directly above the supporting columns. Ulti-
mate positive bending moment was imposed on one span and an ultimate negative moment was
applied to the slab in the region of the supporting columns. A total of seven external strain gages
with gage length equal to 50.8 mm (2) were attached to the top and bottom surfaces of the slab.
Strain gages attached to the reinforcing steel were also monitored. The experimental and numeri-
cal results were compared, and it was found that the cracking model used for the concrete over-
predicted the amount of tension of the concrete could resist. The prestressing tendons helped to
G-7
maintain structural integrity and avoided the total structural collapse, though the concrete failed
by crushing. However, the experimental posttensioned slabs were not longitudinally supported by
any steel girders that would result in the shear-lag phenomenon.
Youn, Seok-Goo and Sung-Pil Chang.1998. Behavior of Composite Bridge Decks Subjected
to Static and Fatigue Loading, ACI Structural Journal 94,3: 249-258.
Five one-third-scale reinforced concrete panels simulating a typical composite bridge
deck were manufactured and tested by Youn and Chang (1998). This research was aimed at inves-
tigating the variations in punching shear strength and fatigue strength of composite bridge decks
at various positions. The loading positions were carefully selected based on the relative magni-
tude of the sectional forces (compressive in-plane forces and shear forces) in the lab, which are
obtained by three-dimensional finite element analysis. Test results showed that both the punching
shear strength and the fatigue strength near the support area of the panels are less than that mea-
sured at the center of the panels. Girder spacing for the models was 700 mm (2-4) with 200 mm
(8) deck overhang. The applied load was measured and controlled by a load cell and a hydraulic
pressure transducer, which were integral parts of the hydraulic loading system. Vertical deflec-
tions of the concrete panels and the steel girders along the central line were measured using elec-
trical dial gages to calculate the net slab deflections. Electrical resistance strain gages were
mounted on the principal reinforcements and the top surface of the concrete panels to measure the
strain profiles. The net slab deflections at the center of the loaded areas were plotted as a function
of the applied loads.
G-8
Khalil, A, L. Greimann, T.J. Wipf, and D. Wood. 1998. Modal Testing for Nondestructive
Evaluation of Bridges: Issues. Transportation Conference Proceedings. 109-112.
Nondestructive bridge evaluation is one of the methods that have been utilized during
recent years. The concept of the method is that modal characteristics (frequencies and mode
shapes) of the structure are directly related to its stiffness properties which change as the structure
deteriorates. Roone River bridge on IA-17 in Hamilton County, Iowa, was tested and model in a
three-dimensional finite element program by Khalil et al.(1998). The bridge was constructed in
1972. It was a three-span, steel welded plate girder bridge. Composite action was provided
between the deck slab and the steel girders using shear studs. Accelerometer stations were along
the bridge shoulders with distances between consecutive stations varying between 3.90 m (12-
10) and 7.00 m (23).
G-9
G.1.2 Field Experiment
Aktan, A.E., K.L. Lee, R. Naghavi and K. Hebbar. 1994. Destructive Testing of Two 80-
Year-Old Truss Bridges, Transportation Research Record 1460. 62-72.
A destructive load test of two truss bridges was conducted by Aktan et al. (1994). It was
intended to explore truss bridge behavior by nondestructive and destructive tests and to correlate
the experimentally measured responses with corresponding analytical predictions. The floor sys-
tem for the Pratt truss bridge consisted of main transverse girders, longitudinal stringers, and a
wood deck. Interior stringers were I-sections 460 mm (18-1) deep spaced 1.22 m (4) center to
center. The responses that were measured in the static tests included (a) the global vertical and
horizontal displacements at each truss connection at the lower chords, including the horizontal
and vertical displacements at the bearings; (b) local strains and axial distortions along selected
members and connections; and (c) Variations in elongation along the length of critical members as
well as strain distributions within built-up cross sections.
Amer, A., M. Arockiasamy and M. Shahawy. 1999. Load Distribution of Existing Solid
Slab Bridges Based on Field Tests, Journal of Bridge Engineering 4,3: 189-193.
Amer et al. (1999) used the term Equivalent width for Effective slab width. The main
parameters affecting the equivalent width were identified using the grillage analogy method. As
the results from this research, the equivalent widths based on the grillage analogy and field tests
were higher than those based on the AASHTO and LRFD codes. The Equivalent width was com-
G-10
puted as the ratio of total area under the moment curve to the maximum moment. In their paramet-
ric study, the main parameters considered were span length, bridge width, slab thickness, edge
beam and number of lanes. It was found that the equivalent width is independent of the element
length and width. The variation of slab thickness had very little effect on the equivalent width. A
simplified equation for the equivalent width was proposed for solid slab bridges with and without
edge beams.
Azizinamini, A., T.E. Boothby, Y. Shekar and G. Barnhill. 1994. Old Concrete Slab Bridges
I: Experimental Investigation, Journal of Structural Engineering 120,11: 3284-3304.
An ultimate load test of a reinforced concrete slab bridge, built in 1938, located in north-
western Nebraska over the Niobrara River was conducted by Azizinamini et al. (1994). The
superstructure was a cast-in-place, five-span reinforced concrete slab comprised of three continu-
ous spans and two simply supported end spans. The curb on each side of the roadway was cast
monolithically with the slab and the tie bars extended from the slab to the curb. The continuous
span was loaded using eight 1,780-kN-capacity hydraulic rams and the simple span was loaded
using four of the same capacity rams. Strains and deflections were measured at various locations
of the slabs for the continuous and simple spans. The strain measurements were make both for
concrete and steel (rebars). Strains were measured by using electric wire resistance bonded strain
gages. Deflections were measured using potentiometers. In the same research, six additional con-
crete slab bridges were tested under truck loads at service load levels, with the weight of each
truck selected such that response of the bridges would be confined to the elastic regime only.
G-11
Again, experimental results indicate that reinforced concrete slab bridges possess much higher
strength than that indicated by current rating procedures.
Bakht, B. and L.G. Jaeger. 1988. Ultimate Load Test of Slab-On-Girder Bridge, Structural
Research Report SRR-88-03, Ministry of Transportation of Ontario, Downsview, Ontario,
Canada.
An ultimate load test on a short-span simply supported Stoney Creek Bridge (see Figure
G-7) in the city of London, Ontario, Canada, was tested by Bakht and J aeger (1992). The purpose
of the test was to study the transverse load distribution pattern of the bridge as the ultimate limit
state was approached, and to examine the degree of composite action between the girders and
noncomposite deck slab at different load levels. The bridge consisted of six rolled-steel girders
and a concrete slab. There was no mechanical shear connection between the deck slab and the
girders. Longitudinal strains of each girder were measured at midspan by three unidirectional
strain gages. One of these gages was installed at the underside of the bottom flange, one on the
web at midheight, and another also at the web very close to the top flange. Deflections of the two
outer girders were measured at midspan by means of deflection transducers. The bridge was
loaded by means of concrete blocks (see Figure G-8). Load-deflection response and moment dis-
tribution were plotted at different load levels. It must be noted that bearing restraint force usually
results in the reduction of applied moments with increase the load carrying capacity. The final
report (1988) for this research iis a good example of well documented literature with sufficient
information for field testing. However, because of the lack of test data from which the boundary
G-12
conditions at the supports could be determined, no attempt was made to analyze the bridge rigor-
ously.
Bakht, B. and L.G. Jaeger. 1990. Bridge Testing - A Surprise Every Time, Journal of
Structural Engineering 116,5: 1370-1383.
Bridge engineers agree that there is no better way to understand the shortcomings of the
mathematical models used for design or evaluation of bridges than to investigate the behavior of
bridges through field testing. Many bridges proof tested in Ontario were able to sustain safely
much larger loads than their estimated capacities shown in the posting loads. Bakht and J aeger
(1990) reviewed the fundamental issues of field testing with two examples of tested bridges.
Barker, M.G. 2001. Quantifying Field-Test Behavior for Rating Steel Girder Bridges,
Journal of Bridge Engineering 6,4: 254-261
Barker (2001) proposed a method for separating and quantifying contributions from vari-
ous factors, which tend to increase the load capacity in real bridges, through field testing. These
factors are important for the bridge owner to remove the unwanted contributions. The paper pre-
sents the results of a comprehensive field test of a three-span steel girder bridge. Systematic field
test procedures that separate and quantify the contributions are demonstrated to determine a safe
load-carrying capacity.
G-13
Burdette, E.G., and D.W. Goodpasture. 1972. Comparison of Measured and Computed
Ultimate Strengths of Four Highway Bridges, Highway Research Record No.382.
In summer 1970, four highway bridges were tested to failure in the area near Winchester,
Tennessee by Burdette and Goodpasture (1972). All bridges (see Figure G-1) were subjected to
vibratory loads produced by Federal Highway Administration equipment, truck loads up to almost
double the AASHO-HS20 loading and static loads to failure. The computed and measured ulti-
mate strengths of each of the four bridges were compared. Two values of computed ultimate
capacity were obtained for each bridge: (a) The ultimate bridge capacity was determined by sum-
ming the ultimate capacities of each longitudinal girder in the bridge, as calculated on the basis of
the 1971 Interim Specifications of AASHO and (b) the capacity of each bridge was calculated on
the basis of strain compatibility relations, using the actual stress-strain relations of the material in
the structure. The ultimate capacities calculated agreed quite closely with the values obtained
through field testing. However, the ultimate capacities computed on the basis of AASHO Specifi-
cations did not compare as closely with the actual ultimate bridge capacities because no redistri-
bution of moments at ultimate load was considered.
The four bridges had girder spacings ranging from 2.1 m (6-10) to 2.7 m (8-10). Dur-
ing the static test, instrumentation was installed in order to measure deflections, strains and
applied loads. Test results provided limited information on strain measurements. However, useful
load-displacement relationships were presented by Burdette and Goodpasture (1972).
G-14
Chajes, M.J., H.W. Shenton and D. OShea. 1999. Use of Field Testing in Delawares
Bridge Management System, The 8th International Bridge Management Conference -
Transportation Research Circular, Transportation Research Board.
Delaware Department of Transportation (DelDOT) and researchers at the University of
Delaware repeated developing methods for integrating bridge field testing and in-service monitor-
ing into DelDOTs bridge management efforts (1999). The differences between Diagnostic Load
Testing and In-Service Monitoring were discussed. Field tests performed in Delaware were
summarized and classified according to different types of application, i.e. posted bridges, permit
vehicles, non-standard designs, etc. However, field testing information was not presented in the
paper.
Dilger, W., J.C. Beauchamp, M.S. Cheung, and A. Ghali. 1981. Field Measurements of
Muskwa River Bridge, ASCE Structural Division 107,11: 2147-2161.
In 1981, the Muskwa River Bridge, located in British Columbia, Canada, was tested
(Dilger et al. 1981). The bridge was a two-lane steel-concrete composite bridge, continuous over
five spans. The testing was done on the two 91.5 m (300). spans. It was composed of two steel
boxes 2.6 m (8-6) wide. The reinforced concrete deck had a minimum thickness of 203 mm
(8). Deflections of the main girders were measured along the length of the west box. Strain gages
were installed along the two girders at 22.86 m (75) intervals (see Figure G-3). At each station
(see Figure G-2), 17 measurements were taken on each of the two steel boxes, 16 on the concrete
deck and 9 on the top surface. The relative displacement at the interface between concrete deck
and steel girder was also monitored. Additionally, mechanical measurements were taken across
G-15
one splice, located at distance of 45 ft. (13.72 m) from support C, between stations 5 and 6 in the
west box. The objectives of this study were to provide information about strains and deflections
developed during the different stages of construction of the concrete deck. Hence, the applied
loads were entirely due to dead loads of box-girders and concrete slab at different construction
stages. There was not any live load imposed from test trucks.
Dorton, R.A., M. Holowka and J.P.C. King. 1977. The Conestogo River Bridge - Design
and Testing, Canadian Journal of Civil Engineering 4: 18-39.
Dorton et al. (1977) presented the background research data incorporated in the Conestogo
River Bridge, the details of the design, a description of the instrumentation and test programme,
and an evaluation of some of the testing. The proposed three span layout at the Conestogo River
crossing was ideal for such a structure, with a central span of 44.20 m (145) and side spans of
34.75 m (114). The bridge had no skew with four girders at 2.64 m (8-8) spacing. The slab was
longitudinally prestressed to prevent cracking under negative moments. The structure thus acted
compositely for the full length of the bridge under live load and superimposed dead load. Static
and dynamic tests were performed and partially described in Dorton et al. (1977). This reference
appears to be one of the best in the field testing literature of composite bridge structures.
Elhelbawey, M., Chung C. Fu, M.A. Sahin, and David R. Schelling. 1999. Determination of
Slab Participation From Weigh-In-Motion Bridge Testing, Journal of Bridge Engineering,
4,3: 165-173.
G-16
Elhelbawey et al. (1999) also developed a refined method for determining the neutral axis
location from bridge testing results. Strain measurement from the testing of four bridges (see Fig-
ure G-4), which were used in Elhelbawey et al. (1999), were used in the analysis. Furthermore,
the effective flange width was determined by analyzing the composite cross section as it is being
subjected to flexural normal stresses. By knowing the location of the neutral axis, assuming linear
strain throughout the cross section and by balancing the tensile and the compressive forces pro-
duced in the cross section (see Figure G-5), the amount of force sustained by the concrete slab
was obtained. Elhelbawey et al. (1999) suggested that neutral axis can be determined from strain
data of no more than two or three typical trucks. This is one of many ways for inferring the effec-
tive slab width from field test results.
Farhey, D.N., R. Naghavi, A. Levi, A.M. Thakur, M.A. Pickett, D.K. Nims and A.E. Aktan.
2000. Deterioration Assessment and Rehabilitation Design of Existing Steel Bridge, Jour-
nal of Bridge Engineering 5,1: 39-48.
A steel truss bridge was diagnostically tested to evaluate strains, internal forces and
deflections under controlled static and moving truck loading by Fahey et al. (2000). The Tindall
Bridge was composed of two identical, simple spans of 50.6 m (166). The original deck was
replaced by an open steel grate in 1961 (see Figure G-14). Strains, temperatures and displace-
ments were measured at selected locations. Sensors at selected truss joints enabled checking the
equilibrium of forces. A representative floor panel, consisting of stringers and a girder, was also
instrumented to investigate the load distribution characteristics of the steel-grate floor as a dia-
phragm. Wire potentiometers (WP) installed at the truss panel points measured the global vertical
G-17
deflections and direct current displacement transducers (DCDT) were used to measure the hori-
zontal support displacements. Load-deflection, member axial stresses, and stringer strain profiles
were plotted (see Figure G-15).
Fu, Y. and J.T. DeWolf. 2001. Monitoring and Analysis of a Bridge with Partially
Restrained Bearings, Journal of Bridge Engineering 6,1: 23-29.
Some researchers utilized results from field testing to develop finite-element models and
evaluate the behavior of bridges at different temperatures. The flexural bending frequencies and
corresponding mode shapes from the analytical model would change in colder weather as the
results of partially restrained bearings. Fu and DeWolf (2001) tested a two-span, slightly skewed
continuous bridge with seven nonprismatic steel plate girders and composite reinforced concrete
slab. Rocker bearings that did not allow longitudinal displacement were used at the center pier.
The natural frequencies of the bridge corresponding to the first three mode shapes were consid-
ered. Field testing results showed that the natural frequencies associated with he first three mode
shapes started increasing as the temperature decreased. Analytically, the finite-element models
were calibrated to represent the actual behavior of the bridge at 60 degrees F, i.e. when the field
data indicate that the bearings were unrestrained. Subsequently, the validated finite-element mod-
els were modified to account for the thermal forces induced by the partial restraint of bearings.
Fu, Chung C., M. Elhelbawey, M.A. Sahin, and David R. Schelling. 1996. Lateral Distribu-
tion Factor from Bridge Testing, Journal of Structural Engineering 22,9: 1106-1109.
G-18
Some researchers (Fu et al. 1996) have done analyses by studying variation of the distribu-
tion factors due to the variation of predetermined bridge parameters, such as span length, girder
spacing, longitudinal stiffeners and so forth. In order to obtained the lateral distribution factors,
four existing bridge structures were tested. Girder spacing for Bridge Number 2 was 2.4 m (8)
Strain measurements were used to determined the location of the neutral axes and to verify the
existence of composite action. Strain gages were installed at the web and top and bottom flanges
as well as at the crossframes. Results from the proposed formulas for the distribution factors were
close to the test results.
Helba, A. and J.B.Kennedy. 1994. Collapse Loads of Continuous Skew Composite
Bridges, Journal of Structural Engineering 120,5: 1395-1414.
Helba and Kennedy (1994) conducted a laboratory test on simply supported and continu-
ous two-span skew composite bridge models. The predictions of the collapse load for skew com-
posite bridges were presented as the most probable yield-line patterns of failure. All models were
of 45 degrees skew with a span of 2.1 m (6-10). The longitudinal steel I-beams spaced at 305
mm (12) were welded to the transverse I-beams (diaphragms) forming an orthogonal arrange-
ment. Surface strains were measured by means of strain gages 10 mm (0.4) long installed on the
steel beams and 30 mm (1.5) long on the concrete deck slab. Deflections were recorded using
mechanical dial gages with a travel sensitivity of 0.010 mm. All models were initially tested elas-
tically placing the load at several positions. The model test results compared very well with the
analytical results from finite-element analysis and yield-line theory. The equations derived for the
ultimate load proved to readily account for the composite action between the deck slab and the
G-19
transverse steel diaphragms by means of the moments of resistance. Helba and Kennedy (1994)
also performed a parametric study to relate the main design parameters and the geometries of the
failure patterns for the minimum collapse load. For the ultimate load analysis, the interaction
between the transverse diaphragms and the longitudinal girders was accounted for by the
moments of resistance in the longitudinal and transverse directions at collapse. The various effec-
tive resisting moments along failure yield lines can be accurately predicted using the proposed in
Helba et al. (1994). Positive moment signified tension at the bottom of the steel section, whereas
negative moment signified tension in the top rebars in the concrete slab, ignoring the tensile
strength of the concrete.
Hulsey, J.L. and D.K. Delaney. 1993. Static Live Load Tests on a Cable-Stayed Bridge,
Transportation Research Record No. 1393. 162-174
Hulsey and Delaney (1993) performed field testing to evaluate the behavior of a cable-
stayed bridge in Canada. The Captain William Moore Creek bridge was a two-lane bridge (see
Figure G-11). The deck consisted of 171.8 mm (7-in.) laminated timber planks supported by
transverse floor beams spaced at 3.66 m (12) on center. The purpose of reference Hulsey and
Delaney (1993) is to give static test results for a control vehicle and ore truck loading, show an
experimental comparison with a 2-D analysis and present the general condition of the structure.
Strains at the extreme inside face of the girders, exterior face of the pylon bases (above the stiffen-
ers) and the exterior face of the upstation left column were monitored during testing. The mea-
sured maximum girder deflection matched very well with the computed value.
G-20
Idriss, R.L., K.R. White, C.B Woodward. and D.V. Jauregui. 1995 After-Fracture Redun-
dancy of Two-Girder Bridge: Testing I-40 Bridges Over Rio Grande, Fourth International
Bridge Engineering Conference 2: 316-326.
Idriss et al. (1995) conducted field testing of the I-40 bridges over the Rio Grande in Albu-
querque, New Mexico in the fall of 1993. These medium-span steel bridges represented a com-
mon design in the United States and were classified by AASHTO as non-redundant fracture
critical two-girder steel bridges. The bridge was field tested to determine the impact of a near
full-depth girder fracture on the redistribution of loads, the load capacity and the potential for col-
lapse. A reinforced concrete deck of 178 mm (7) thick was supported on steel stringers at 2.3 m
(7-6) center spacing (see Figure G-13). Strain gages were placed at the top flange, bottom flange
and the center of the web near the girder critical location. The results showed that when a mid-
depth fracture was imposed on the north girder of the three-span unit, the bridge did not collapse.
Most of the load was redistributed longitudinally through the damaged girder and stringer deck
system to the interior supports. The main load path proved to be the fractured girder itself as it
redistributed the load longitudinally to the interior supports through cantilever action. The floor
beams, lateral bracing system and deck transferred the load to the intact girder, through torsional
stiffness of the system.
Kathol, S, A. Azizinamini and J. Luedke. 1995. Strength Capacity of Steel Girder Bridges -
Final Report, Report No. RES1(0099) R469, Nebraska University, Lincoln.
Kathol et al. (1995) investigated the global and local behavior of steel bridges with and
without diaphragms. The investigators also considered the feasibility of using the new AASHTO
G-21
LRFD empirical method of concrete slab design, which requires a minimal amount of reinforcing
bars, when relatively large girder spacing is prescribed. Field test on composite steel girder
bridges was conducted under service loads to verify the findings of the laboratory tests.
Kennedy, J.B. and N.F. Grace. 1983. Load Distribution in Continuous Composite Bridge,
Canadian Journal of Civil Engineering 10,3: 384-395.
Testing of two continuous composite bridge models was conducted by Kennedy and Grace
(1983). The models had two equal spans with Model II post-tensioned longitudinally in the region
of the intermediate support. Each span was 1.52 m (5) long and 1.32 m (4-4) wide with 64 mm
(2.5) thick concrete deck in composite action with five longitudinal steel girders. They were rig-
idly connected transversely by five steel I-beams acting as diaphragms. The continuous shear con-
nection between the concrete deck and the girders was provided. The top and bottom surface
strains in the concrete deck and steel girders of each bridge model were measured by means of a
total of 73 linear strain gages, installed at midspan and in the vicinity of the intermediate support.
The deflection profile at the middle of each span was determined by means of the mechanical dial
gages. The horizontal displacement between the concrete deck and steel girder were also mea-
sured by dial gages. The elastic response of continuous composite bridges subjected to static loads
was examined by means of orthotropic plate theory using realistic estimates of the various rigidi-
ties. The study showed that the presence of I-beam diaphragms, when rigidly connected to the
longitudinal girders, enhanced significantly the transverse load distribution as well as the effec-
tiveness of the orthotropic theory in predicting the elastic response of a continuous composite
G-22
bridge. However, an increase in the number of diaphragms beyond a certain limit did not signifi-
cant improve the transverse load distribution.
Kim, S., A.S. Nowak and R. Till. 1996. Verification of Site-Specific Live Load on Bridges,
Probabilistic Mechanics and Structural Reliability and Geotechnical Reliability, Proceedings
of the seventh specialty conference, pp. 214-217.
Seven bridges in the Detroit area were monitored using a weigh-in-motion system to
determine the truck data such as axle weights, gross vehicle weight and axle spacings. Kim et al.
(1996) conducted the load test and found that the traffic on I-94 over M-10 was heavier than on
any other tested bridges. Girder spacing for I-94/M-10 was 2.7 m (8-10) with the span length of
23.2 m.
Miller, R.A., A.E. Aktan and B.M. Shahrooz. 1994. Destructive Testing of Decommissioned
Concrete Slab Bridge, Journal of Structural Engineering 120,7: 2176-2198.
Destructive testing of a decommissioned concrete slab bridge was conducted by Miller et
al. (1994). Testing was initiated to determine if the beam strip design of the bridge portrayed the
true load distribution and resistance of the slab. Hydraulic cylinders were used to simulate truck
traffic in one lane. Results indicate that the boundary conditions and the geometry of the bridge
significantly influence bridge behavior. Slab failure occurred due to punching shear, the capacity
of which may have been reduced because of deterioration. The strip design model was shown to
G-23
overestimate structural demand and underestimate structural supply. Nonlinear finite element
models provide better estimates of structural supply and demand.
Moses, F., J.P. Lebet and R. Bez. 1994. Applications of Field Testing to Bridge Evaluation,
Journal of Structural Engineering 120,6: 1745-1762.
Full-scale testing is a common practice as a large number of bridges have been field tested
in Switzerland (Moses et al 1994). A bridge test was required of every major bridge in Switzer-
land by the Federal and Cantonal Administration before it was opened to traffic. The Viadic du
chene bridge, for example, was tested in Switzerland in 1982. The bridge test information was sta-
tistically used to consider an evaluation of the bridge. This research fully described the evaluation
procedure by using field testing information. Hassan et al. (1995) presented the advantages of
load testing on the accuracy of the results when compared to the design calculations. Many unex-
pected parameters such as reinforced-concrete parapets or an asphalt wearing surface, which had
some contributions to the stiffness of the bridge during the load test, would be experimentally
captured.
Mufti, A.A., L.G. Jaeger and B. Bakht. 1997. Field Performance of Steel-Free Deck Slabs
of Girder Bridges, Recent Advances in Bridges Engineering by Merier, U. and Betti, R.,pp.
239-246.
Canadian researchers introduced an empirical design method for deck slabs based on the
recognition of arching action in 1979. This method leads to significantly smaller amounts of rein-
G-24
forcement than required by flexural designs. Mufti et al. (1997) tested two Canadian bridges
which were partly comprised of steel-free deck. The Salmon River bridge and The Kent County
Road No. 10 bridge had girder spacings of 2.7 m (8-10) and 2.1 m (6-10), respectively. Strain
gages were used to capture the linearity of strain distribution over the depth of the girders at mid-
span of the interior and exterior girders.
Reid, J.S., M.J. Chajes, D.R. Mertz and G.H. Reichelt. 1996. Bridge Strength Evaluation
Based on Field Test, Probabilistic Mechanics and Structural Reliability, pp. 294-297.
Reid et al. (1996) pointed out that load tests of bridges often reveal that the load carrying
capacity is higher than estimated, thereby eliminating the need for costly rehabilitation or replace-
ment. A case study of Christina Creek Bridge, which is located on Interstate 95 in Delaware, was
discussed. The bridge was tested in the fall of 1995. The girders of the bridge were instrumented
using 32 strain transducers, positioned at 16 locations with a gage on the upper and lower flange
at each location. Emphasis was placed on the midspan of the bridge. A strain history plot mea-
sured by transducers on a girders top and bottom flanges was shown. The test results were used to
identify (1) flexural section properties and the nature of the composite action between the girder
and deck; (2) transverse load distribution; (3) support restraint and (4) impact effects. The neutral
axis location was identified using measured strain data from transducers placed on the top and
bottom flanges of the girders.
Saraf, V.K. and A.S. Nowak. 1997. Field Evaluation of Steel Girder Bridge. Transportation
Research Record 1594, pp. 140-146
G-25
A 70-year-old bridge in Michigan was nondestructively tested to evaluate the load-carry-
ing capacity of the bridge as the steel girders had been extensively corroded (Saraf et al. 1997).
The bridge was a simply supported steel girder bridge carrying state route MI-50 over Grand
River in Jackson County, Michigan (see Figure G-9). Before the measurements, the bridge was
calibrated with a truck with known weight and dimensions. The Weigh-In-Motion (WIM) mea-
surements were carried out for 2 days. As the results, load-deflection relations were illustrated for
different load cases (see Figure G-10).
Saraf, V.K, A.S. Nowak and R. Till. 1996. Proof Load Testing of Bridges, Probabilistic
Mechanics and Structural Reliability and Geotechnical Reliability, Proceedings of the sev-
enth specialty conference, pp.526-529.
Four other simply supported bridges with spans ranging from 6.5 m (21-4) to 15.2 m
(50) were proof-load tested by Saraf et al. (1996). One selected bridge was a simply supported
bridge with ten steel girders and a reinforced concrete slab. The distance between girders was
approximately 1.5 m (5). Two M60A3 military tanks on flat bed trailers were used to achieve the
high target load level. Strain transducers were mounted on lower flanges of all steel girders at
selected locations. LVDTs were used to measure deflections of steel girders at midspan. The
results showed that even though the bridge was designed to behave as a non-composite structure,
unintended composite action between concrete slab and steel girders was present. Girder distribu-
tion of deflections and stresses for center line loading case is illustrated below (see Figure G-12).
G-26
Stallings, J.M., and C.H. Yoo. 1993. Tests and Ratings of Short-Span Steel Bridges, Jour-
nal of Structural Engineering 119,7: 2150-2168
A series of diagnostic tests were performed on three short-span, two-lane, steel-girder
bridges by Stallings and Yoo (1993). Tests were performed with stationary and moving test trucks.
Girder spacings of the three bridges ranged from 1.78 m (5-10) to 2.34 m (7-8). Strain gages
were placed at midspan on top and bottom flanges of selected girders. Midspan deflections of the
interior girders were also measured to provide an overall indicator of the bridge response during
testing. The positions of the truck in the static tests were also given.
Yakel, A., P. Mans and A.Azizinamini. 1999. Negative Bending Tests of HPS-485W Bridge
Girders, Materials and Construction, pp. 559-565.
High performance steel has been developed to address the two main concerns often asso-
ciated with high strength structural steel, weldability and toughness. Two steel plate girders con-
structed of high performance steel (HPS) were tested by Yakel et al. (1999). Each specimen was
tested as a simply supported beam with a single point load applied at midspan. Instrumentation
included strain gages along the flanges and over the depth of the web within the test regions. Two
pairs of potentiometers were used at each end of the girder to obtain end rotations. Both speci-
mens reached moment capacities exceeding the plastic moment capacity of the section calculated
based on the actual material properties. Significant inelastic rotational capacity was observed for
the same specimen even though AASHTO LRFD provisions did not imply any inelastic rotational
capacity for non-compact sections.
G-27
G.2 BEHAVIOR OF COMPOSITE GIRDERS
Newmark, N.M., C.P. Siess, and R.R. Penman. 1946. Studies of Slab and Beam Highway
Bridges: Part I. Tests of Simple-Span Right I-Beam Bridges. Bulletin No.363, University of
Illinois Engineering Experiment Station.
In this experimental study, scale relations were studied to have the similar real behavior
from quarter-scale models; that is, they were geometrically similar to full-sized bridges and their
linear dimensions were one-fourth as great. In order to obtain equal stresses in the model and the
full-sized structure or prototype, the loads on the two structures must be related as follow: (1)
Concentrated loads should be (1/4)
2
=1/16 as large for the model as for the prototype; (2) Loads
distributed over a length, such as the weight of a beam per foot of length, should be as large for
the model as for the prototype; (3) Loads distributed over an area, such as the weight of the slab
per square foot, should have the same magnitude per unit of area for the model as for the proto-
type. In each of the above cases, the total load on the models 1/16 as great as that on the prototype.
For loads related as indicated above, deflections and other linear movements such as slip between
the slab and the beams will be as large in the model as in the prototype.
Richart, F.E. 1949. Laboratory Research on Concrete Bridge Floors. ASCE Transactions
114: 980-996.
The laboratory structures tested have been models, generally with scales of 1:2 to 1:5.
Most of these models were first designed in the full-sized prototype, using standard design meth-
G-28
ods and loadings specified by the AASHTO. Great care was taken to keep the materials and
dimensions of the models in true proportions to those of the prototype, and many studies were
made to determine whether any scale effect existed between large-scale and small-scale models.
Newmark, N.M. 1949. Design of I-Beam Bridges. ASCE Transactions 114: 997-1022.
As stated, the experimental work was entirely on quarter-scale models. Initial studies indi-
cated that duplication of the behavior of full-scale prototype structures could be obtained in quar-
ter-scale models if attention were paid to the proper scaling of the aggregate and reinforcement.
Extensive studies indicated good correlation, over the entire range of action, from elastic behavior
at low loads, the subsequent action at working loads, and the behavior after considerable yielding
of the reinforcement, to and including the failure of the slab by punching. The reinforcement used
in the models consisted of bars, 1/8 in. square, specially heat treated to have stress-strain charac-
teristics corresponding to the full-sized reinforcement used in an actual structure. The effect of the
relative stiffness of the beam on the moments resulting from a single concentrated load is large;
but the effect on the moments caused by truck loads is small. Results emphasize the importance of
the slab thickness in distributing the loads to the beams (also the effect of thickness for shear lag).
Little difference appears in the results before and after cracking of the concrete (not much effect
from cracking in mid span). When several loads are on the bridge, corresponding to trucks in
adjacent lanes, the moments in the beams are much more uniform than those for a single load in
the bridge. In the negative moment regions of continuous bridges the advantages of composite
action are not so marked. Indeed, it is questionable whether there is a disadvantage in having the
slab subjected to appreciable tensile stresses as the upper flange of a composite beam. Failure of
G-29
shear connection can be catastrophic, and is not to be tolerated; consequently, it is reasonable to
be extremely conservative in the design of shear connectors. To take advantage of the increased
strength of the composite structure, not only must the shear connectors be conservatively
designed, but the concrete slab must be adequately reinforced in the longitudinal direction to pre-
vent serious opening of shrinkage cracks.
Siess, C.P. 1949. Composite Construction for I-Beam Bridges. ASCE Transactions 114:
1023-1045.
It is evident from the curves that the effectiveness of the shear connectors was increased
considerably be an increase in the compressive strength of the slabs. The increase, however, was
proportionately less for the higher strengths than for the lower ones
Siess, C.P., and I.M. Viest. 1953. Studies of Slab and Beam Highway Bridges: Part V. Tests
of Continuous Right I-Beam Bridges. Bulletin NO.416, University of Illinois Engineering
Experiment Station.
They used three test bridge models, which were two-span continuous structures with two
spans of 15 ft., each span consisting of five steel I-beams spaced at 1.5 ft. intervals and supporting
a reinforced mortar slab 1 in. thick.For composite bridges, final failure occurred by punching of
the slab at a load a little below what was expected by calculated ultimate flexural capacities
G-30
Mackey, S., and F.K.C. Wong. 1961. The Effective Width of a Composite Tee-Beam
Flange. The Structural Engineer 39, 9 (September): 277-285.
Four specimens had the same length of 9ft-6in. with varying girder spacing; 2-6, 5-4,
6-8, and 8. 4 or 6 beams were used for each specimen. All specimens were tested under equal
concentrated loads applied through steel blocks direct to the top flange of each joist at the mid-
span points. For a single concentrated load, the effective width is least at the loaded section and
increases towards the supports; for uniformly distributed load, it is maximum at midspan and
decreases slightly towards the ends, while for sinusoidal loading, the effective width is constant
over the span. As their conclusions: (1) effective width of a tee-beam flange determined experi-
mentally is larger that predicted by theory and is much larger than that currently recommended for
use in design. (2) Both the test results and analysis agree that values of effective width vary with
beam spacing over beam span ratio (2b/L) (3) Loading conditions and slab thickness, as shown by
the analysis, also have effects on the effective width, but experimental confirmation is necessary.
(4) Experimental effective widths vary along the span, but in an opposite sense to that obtained by
analysis
Chapman, J.C. 1964. Composite Construction in Steel and Concrete - the Behaviour of
Composite Beams. The Structural Engineer 42, 4: 115-125.
Bending of the isolated composite T-beam If sufficient shear connectors are provided for
the steel and concrete sections to act compositely, the maximum moment carried may exceed the
nominal fully plastic moment, due to strain hardening in the beam. In this case the plastic neutral
axis is located within the concrete slab. However, if the plastic neutral axis is close to the beam
G-31
flange, full plasticity cannot developed in the steel before crushing at the surface of the slab
occurs. The resulting reduction in the ultimate moment appears to be small, but the rotational duc-
tility of the composite section required further study. Although there is a considerable gain in elas-
tic section modulus due to composite action, the gain in ultimate strength is much greater.
The effect of slab bending on the composite beam If failure of the T-beam occurs when the
slab deformations are large, the two failure modes will clearly interact. On the other hand, if one
mode of failure occurs well in advance of the other, then it may be that the earlier failure can be
treated in isolation. Transverse slab bending may also be expected to increase the splitting ten-
dency. It is possible that the shear connectors may be adversely affected in parallel beam systems
where the loaded beam deflects more than the adjacent beams, so that the concrete at the root of
the connectors would be in transverse tension. An uplift force due to transverse bending of the
slab may occur at a beam remote from the loaded beam or slab.
Behavior of shear connectors The shear connectors are required to transmit horizontal
shear between the beam and slab, and also to prevent the slab lifting away from the beam. Break-
down of the shear connection can occur by failure of the shear connectors or by crushing of the
concrete, or both. One difficulty in determining the strength of the shear connector by calculation
lies in determining the behavior of concrete under the triaxial stress which exists at the root of the
shear connector. A characteristic of all shear connectors is that some slip must occur before their
resistance can be mobilized.
Slip and Uplift The slip is due partly to deformation of the concrete in the region of the
shear connectors and partly to deformation of the shear connector. For design purposes it seems
G-32
wholly reasonable to take no account of bond resistance. Nevertheless bond resistance must be
taken into account in the interpretation of experimental results.
The greatest slip tends to occur around quarter-span. If sufficient shear connectors are pro-
vided to develop the fully plastic moment of the composite section, the effect of slip in the work-
ing range is unlikely to be sufficiently great to necessitate consideration in design. When slip
begins the slab also begins to lift from the beam and the lifting occurs even where a uniformly dis-
tributed load is applied to the slab along the length of the beam. Bending and axial forces which
are associated with the slip and uplift then develop in the shear connectors. The uplift forces are
considerable and if the shear connectors are of insufficient length they may eventually pull out of
the slab.
Distribution of shear connectors These considerations suggest that it may generally be
preferable to space the shear connectors uniformly along the length of the beam. Experiments at
Lehigh University and at Imperial College on uniformly loaded beams have indicated that beams
with uniformly spaced shear connectors behave as well as beams with shear connectors spaced
triangularly. Where the sign of the bending moment changes, so that the concrete slab is in tension
or cracks, the determination of the horizontal shearing forces on an elastic basis becomes difficult,
since the distribution of bending moment and the extent of cracking are independent. The maxi-
mum possible total shear to be transmitted at collapse is easily determined, since this is given by
the sum of the tensile force in the steel beam at mid-span and the tensile force in the reinforce-
ment at the support. The determination of the tensile force in the reinforcement requires the defi-
nition of an effective breadth of slab in the cracked inelastic condition. The development of the
plastic support moment depends on the rotational ductility of the mid-span section.
G-33
Prestressing A bridge deck slab may be transversely prestressed, and this could affect the
performance of the shear connectors.
Creep and Shrinkage Although significant stresses and deflections can arise from these
effects the collapse load of a composite beam would normally not be significantly affected.
The design of the composite section To prevent the formation of a longitudinal crack on
the line of shear connectors, sufficient transverse reinforcement must be provided to resist hori-
zontal shear, transverse direct stresses, and local stresses in the region of the shear connectors, as
well as transverse bending stresses.
Slutter, R.G., and G.C. Driscoll. 1965. Flexural Strength of Steel-Concrete Composite
Beams. Journal of the Structural Division-ASCE 91, ST2: 71-99.
Twelve simple span composite beams of 15 ft. span, one two-span continuous beam. All
beam specimens consisted of a 12 WF 27 beam connected to a concrete slab 4 in. thick by 4 ft.
wide. It was concluded that if the neutral axis is located within the concrete slab at ultimate load,
then the ultimate flexural capacity of a beam can be determined, even if the number of shear con-
nectors provided is less than the number required to develop the theoretical ultimate bending
moment M
u
.
G-34
Daniels, J.H., and J.W. Fisher. 1966. Fatigue Behavior of Continuous Composite Beams.
Fritz Engineering Laboratory Report No.324-1, Dept. of Civil Engineering, Lehigh Univer-
sity.
Viest, Fountain and Siess discussed the negative moment region of continuous composite
beams in some detail. This discussion was based on the static behavior of two composite model
bridges that were reported by Siess and Viest. These models differed in that one had shear connec-
tors throughout the beam while the second had shear connectors in the positive moment regions
only. From these studies, it was concluded that (1) in the negative moment regions, only the slab
reinforcement can act compositely with the steel sections, (2) in beams with shear connectors
throughout the beam, the slab reinforcement was fully effective; when connectors were omitted
from the negative moment region the slab reinforcement was only partly effective, and (3) the
action of both of these continuous composite bridges was about the same as the distributions of
strains and also moments in both the negative and positive moment regions was nearly the same.
Based on these studies, it was concluded that the use of an elastic analysis in conjunction with the
usual load distribution factors is justified, and no special provisions are needed for the design of
continuous composite bridges. Slutter and Driscoll summarized the behavior of a single continu-
ous composite beam statically to its ultimate load. It was noted that the theoretical plastic collapse
load was exceeded in the test even though the member had inadequate shear connection through-
out its length. This test also indicated that on the basis of static tests alone, there was not a great
deal of need for shear connectors in the negative moment region of continuous composite beams.
Barnard and J ohnson presented the results of a study on the plastic behavior of continuous com-
posite beams. It was shown that simple plastic theory gave a good estimate of the flexural strength
G-35
of the beams tested, provided that various types of secondary failure could be prevented by appro-
priate detailing.
These preliminary studies were followed up by further studies designed to provide more
information on these secondary failure. In particular, the effects of cracking and transverse bend-
ing of the slab and the behavior of the longitudinal reinforcement over the supports was exam-
ined. This study indicated that simple plastic theory can be used for the design of continuous
composite beams and that full use can be made of the tensile strength of longitudinal reinforce-
ment in the slab. The basic variables investigated in this test program were then: (1) the effect of
stud shear connectors in the negative moment region on the fatigue behavior: shear connectors are
required in the negative moment regions of continuous composite beams which have continuous
longitudinal slab reinforcement. (2) the effect of increased amounts of longitudinal reinforcing
steel in the negative moment regions: Increased percentages of longitudinal reinforcement in the
negative moment regions of continuous composite beams, over that presently allowed by the
AASHO specifications, appears desirable in order to control the number and widths of slab cracks
as well as to improve interaction and flexural conformance in that region.
Hagood, T.A. Jr., L. Guthrie, and P.G. Hoadley. 1968. An Investigation of the Effective
Concrete Slab Width for Composite Construction. Engineering Journal - AISC 5, 1 (Janu-
ary): 20-25.
Three specimens which had the same length of 6 with varying girder spacing of 6, 4-6,
and 3 were tested. Each test structure was loaded with concentrated loads at the third points of
each I beam. Effective slab width requirements for composite construction are conservative for
G-36
interior beams. On the other hand, the results indicate that present specification requirement are
reasonable for exterior beams.
Johnson, R.P. 1970. Research on Steel-Concrete Composite Beams. Journal of the Struc-
tural Division-ASCE 96, ST3: 445-459.
A thorough study of simple flexure was first made. This included tests on simply-sup-
ported beams with flange breadth-to-thickness ratios below 4.8. It was found that the elastic the-
ory gave accurate predictions of flexural rigidity and of curvature at first yield. The bending
moment at first yield was over-estimated by about 13%, due to shrinkage and residual stresses.
Very close estimates of ultimate moment of resistance were obtained if allowance was made for
partial plasticity of the steel. It was shown that if full plasticity is assumed, the resulting error in
M
n
rarely exceeds 5%. Four exploratory tests were then made on continuous beams, steel with
narrow concrete flanges. In these beams, the top longitudinal reinforcement at the supports con-
sisted of high-yield steel.
The test showed that moment-curvature relations in hogging bending agreed closely with
the computed curves until large curvatures were reached, when some form of secondary failure
usually occurred. Due to strain hardening, some shear connectors were severely overloaded, and
slips were greater than in the earlier tests. This increased deflections, but had little effect on ulti-
mate moments of resistance. It seemed likely that a simple plastic design method could be devised
for continuous beams in buildings; but the tests showed that further research was needed on hog-
ging moment regions and on the secondary failures, particularly longitudinal shear failure in the
slab, and plastic buckling of the steel beam adjacent to a support.
G-37
Yam made a theoretical study of the influence of interface slip and other variables on the
ultimate-load behavior of composite T-beams. His research suggested that the behavior of com-
posite beams is not influenced much by uplift, which was neglected in the analysis. he application
of plastic theory to the design of continuous composite beams was studies and tests done by
Teraszkiewicz were analyzed. In drafting the code for CP117, the 80% design method of the code
was chosen to ensure that flexural failure should precede shear failure and to limit the loss of
interaction at ultimate load. Two series of tests on continuous beams were carried out by Kemp
and Kipps to provide data on the behavior of hogging moment regions. The type of failure that
occurred in series TB included primary flexural failure, buckling and shear failures of the joist,
and flexural and shear failures of the slab. In series CS it was found that flexural failure could be
obtained in both the slab and the beams. The principal conclusions are given. (1) They relate to
the ultimate strength design of a continuous composite floor system under distributed load. (2)
They expected that when joist flanges are very wide, the effective span of the slab can be taken as
the clear distance between the edges. The full strength of top longitudinal reinforcement at a sup-
port can be developed for force ratios at least as high as 0.4, and improvements in both stiffness
and overall economy can be achieved by its use. (3) Sufficient bottom transverse reinforcement
should be provided to resist a shear force. (4) The ultimate sagging moment of resistance of a
composite beam is reduced by the transverse bending and horizontal shear stresses in the slab, as
well as by shear lag. Allowance for all these effects may be made by assuming that at failure the
effective breadth of the slab Be is given by Be/B=1-B/2Ls.
Van Dalen also tested 17 specimens. Conclusions were: (1) when adequate shear connec-
tion and transverse reinforcement are provided, the resistance to bending of a hogging moment
region with longitudinal slab reinforcement continues to increase with curvature until buckling
G-38
occurs in the steel joist, even though the slab may be cracked in tension throughout its depth. (2)
the tests showed that as ultimate load was approached, the transverse variation of longitudinal ten-
sile strain at the level of the slab reinforcement was of the form =
0
(1-y/9t).
Burdette, E.G., and D.W. Goodpasture. 1973. Tests of Four Highway Bridges to Failure.
Journal of the Structural Division-ASCE 99, ST3: 335-348.
The first failure tests of full-size bridges performed in the US. One of four bridges was a
four span continuous bridge, with composite action in the positive moment region only. Loads
were applied to both two lanes, which was similar to design loads.
The behavior of this continuous steel bridge was almost linearly elastic up to yielding at
the section under the applied loads nearest the center of the span. After yielding, a plastic hinge
formed at a section near the center pier at the end of the cover plates on the side of the pier away
from the loaded span. Local buckling of the compression flange of one of the exterior girders
occurred at the formation of the plastic hinge. Shortly after this hinge formed, a secondary com-
pression failure of one of the curbs occurred at the section of maximum positive moment, and the
test was terminated.
Subcommittee. 1974. Composite Steel-Concrete Construction. Journal of the Structural
Division-ASCE 100, ST5: 1085-1139.
Composite beams and Girders with a homogeneous steel I cross section and a solid cast-
in-place slab The elastic behavior of composite beams was well understood by the end of the
G-39
1950s and elastic design methods served as the basis of the 1957 AASHO Specifications. In
1960, the J oint ASCE-ACI Committee on Composite Construction published tentative recom-
mendations for building design which introduced certain modifications based on ultimate strength
concepts. However, elastic design methods for composite beams have been retained in American
specifications to this day.
Ultimate Strength 1957 AASHO specifications and tentative recommendations of the joint
ASCE-ACI based the design of shear connectors on the concept of limiting the slip between the
slab and the beam. However, studies by Slutter and Driscoll showed that such a limitation was
unnecessary because composite beams developed the full flexural capacity of the cross section as
long as the sum of the ultimate strengths of the individual connectors between the points of zero
and maximum moment was at least equal to the total horizontal shear. The tests have shown that
the ultimate strength of a composite beam is essentially independent of the history of loading.
Accordingly, the ultimate strength of a beam built with temporary supports is the same as that of
the same beam built without temporary supports. However, the presence or absence of temporary
supports during construction can have a pronounced effect on the magnitude of deflections. In
other studies, Barnard investigated the effect of the shape of the stress block on the ultimate flex-
ural capacity of a composite beam. Daniels and Fisher reported on tests of composite beams with
simulated moving loads. Lew examined the effects of shear connector spacing on the ultimate
strength of beams. Reddy and Hendry made a through review of the results of British studies of
simply-supported composite beams and derived equations for the ultimate bending strength of
such beams.
G-40
Effective slab width Mackey and Wong, Lee, Severn, and Adekola have studied the effec-
tive slab width for simple span beams. Reddy and Hendry have reviewed the various rules and
formulas and proposed a new effective width formula for simple span composite beams.
Studies of the effective slab width in the negative moment regions of continuous compos-
ite beams with continuous slab reinforcement have been conducted by Garcia and Daniels at
Lehigh University. It was concluded that the effective slab width is approximately constant
throughout the beam length. A similar conclusion was drawn by Tachibana.
Transverse and Longitudinal Reinforcement Analytical studies by Adekola showed that
transverse tensile stresses were developed in the slab, and transverse reinforcement must be pro-
vided to prevent premature failure. Roderick, Hawkins, and Lim made an analytical and experi-
mental investigation of the various modes of failure, i.e., pullout of the studs, shearing of the studs
and longitudinal splitting which can occur when the transverse slab reinforcement is insufficient.
The principal studies of the longitudinal reinforcement requirements for the negative moment
regions of continuous composite beams have been carried out by J ohnson in England and by
Daniels, Fisher and Slutter in the US. Recent studies at Lehigh University of the results of static
and fatigue on large size continuous composite beams have indicated the desirability of requiring
that the area of continuous longitudinal reinforcing in the negative moment regions of bridge
members equals or exceeds 1% of the total cross-sectional area of the slab.
Precast elements for bridge decks Normally, the adjacent edges of longitudinally pre-
stressed elements are constructed in such a way as to provide a formwork for the concrete to be
placed in the joint. After the joints are filled, the deck may be prestressed in the negative moment
region by posttensioning. The posttensioning force is applied before there is composite action
G-41
between the slab and the girders. The behavior of carefully designed and constructed precast slabs
is the same as that of cast-in-place slab.
Hamada, S., and J. Longworth. 1976. Ultimate Strength of Composite Beams. Journal of
Structural Engineering-ASCE 102, 7: 1463-1478.
Conventionally, the ultimate strength of composite beams under positive bending has been
evaluated on the basis of yield stress for steel and equivalent stress block for concrete. Significant
contributions to an understanding of the behavior of continuous composite beams have been made
by Barnard and J ohnson, Park, and Daniels and Fisher. In continuous composite beam, the ulti-
mate condition under positive moment occur due to the yielding of steel and crushing of concrete,
but conditions under negative moment may be affected by buckling. In tests of isolated simple
beams under negative bending at the University of Alberta and at the University of Cambridge,
the majority of beams failed as a result of local buckling. The University of Alberta tests have
shown that the width-thickness ratio of the flange of the steel section and the amount of longitudi-
nal slab reinforcement are significant for local flange buckling. If other failure conditions such as
shear connector failure, longitudinal splitting, and longitudinal shear failure of the concrete slab
are prevented by appropriate provisions, two major failure modes may be considered for continu-
ous composite beams, i.e., crushing of concrete in the positive moment region and local flange
buckling in the negative moment region. Research at Lehigh University and at the University of
Alberta indicated the concrete crushing is the dominant failure condition for simple span beams in
positive moment. Research at the University of Cambridge and the University of Alberta has
shown local flange buckling to be significant in beams subjected to negative moment.
G-42
Wegmuller, A.W. 1977. Overload Behavior of Composite Steel-Concrete Bridges. Journal
of the Structural Division-ASCE 103, ST9: 1799-1819.
Very little information is available, either on full-scale testing or on methods of analysis,
for determining the ultimate load-carrying capacity of highway bridges. Towards the late 1940s,
the University of Illinois published the results of a series of tests of simple span and continuous I-
beam bridges. A series of comprehensive tests on large-scale model bridges was performed in
conjunction with the AASHTO Road Test. However, it was not until 1970 that Burdette and
Coodpasture performed the first failure tests of full-size bridges in US. To date, only a few meth-
ods are known to have been developed for composite bridge systems under ultimate loads. FEM
for finding the elastic-plastic response of eccentrically stiffened plate was applied in the previous
study by modeling the concrete deck as elastic-plastic material. The present report describes an
extension of this approach by including cracking and crushing of the concrete deck and the effect
of the reinforcement.
Botzler, P.W., and J. Colville. 1979. Continuous Composite-Bridge Model Tests. Journal of
the Structural Division-ASCE 105, ST9: 1741-1755.
If the load-factor method is to be consistent in its application, the load-distribution factor
and effective-width values used should reflect the inelastic behavior of the structure, which
includes the redistribution of moment in the post-elastic stress range. A few plastic analysis meth-
ods have also been developed for composite-bridge systems under ultimate loads. Very little
information is available on full-scale testing for determining the ultimate load-carrying capacity
G-43
of highway bridges, although considerable testing of single flexural members has been per-
formed.
Two model bridge structures were tested in the laboratory. One of the models was a multi-
girder two-span composite structure and the other was a multi-girder three-span composite struc-
ture. A single jack used to apply a concentrated load to the top surface of the concrete slab in each
test. If S/L is very large, equal load distribution is not possible and the effective width approaches
that of a single isolated girder with a wide flange. The effective width is also affected by the load-
ing configuration. For example, a somewhat distributed loading, such as several AASHTO truck
loadings side by side, tends to induce more uniform effective widths in a multigirder system than
would a single concentrated load. If the slab strength and loading configuration of a highway
bridge are such that a general collapse mode consisting of full plastic hinges for each girder could
form, then the use of plastic distribution factors and plastic effective widths in calculations of
the bridge resistance to factored loads would result in a consistent load-factor design method.
Heins, C.P. 1980. LFD Criteria for Composite Steel I-Beam Bridges. Journal of the Struc-
tural Division-ASCE 106, ST11: 2297-2312.
During the past decade bridges can be designed using a pseudo plastic concept, designated
as Load Factor Design. This method considers effects of plastification on the girder, but utilizes
elastic distribution and elastic effective width. Three composite beams were load tested. The load
was applied to mid span of the center girder. Proposed plastic effective width was presented for
positive moment region based on single point load analysis and experiment results.
G-44
Kennedy, J.B., and N.F. Grace. 1982. Prestressed Decks in Continuous Composite
Bridges. Journal of the Structural Division-ASCE 108, ST11: 2394-2410.
In 1953, Siess and Viest investigated the behavior of two continuous composite bridge
models and the effects of slab reinforcement and shear connectors on the distribution of strains
and moments in the sagging and hogging moment regions. Similar studies were carried out by
several investigators with emphasis on crack control in the hogging moment region. In 1972,
Fisher et al. recommended increasing the percentage of longitudinal reinforcing steel in the slab
over the hogging moment region to at least 1%. The objective of this investigation was to examine
the effect of the interaction between prestressing a portion of the slab and shear connectors on the
elastic response and crack control of continuous composite bridges.
Two 1/8 scale models were builtof a two-span continuous bridge of composite construc-
tion. Bridge Models I and II were identical except that a portion of the concrete deck of Bridge
Model II was post-tensioned longitudinally in the region of the intermediate support. The struc-
tures were scaled one-to-eight in plan and one-to-three in the vertical direction. Below the crack-
ing-load condition, each bridge was loaded by a single concentrated load at three positions; (1)
midspan of an exterior girder (2) midspan of the first interior girder and (3) midspan of the middle
girder. Finally, each bridge model was tested to near collapse by means of two equal concentrated
loads applied one in each span through a heavy distributed beam to the midspan of the middle
girder.
Test results indicate that transverse cracking at the intermediate support does not signifi-
cantly influence the transverse distribution of deflections, nor longitudinal and transverse
moments at midspan of the continuous structure. Composite action of the transformed section can
G-45
be realized in the vicinity of the hogging moment region by providing adequate shear connections
throughout the region, coupled with a suitable prestressing force applied to a portion of the con-
crete deck. Prestressing of the concrete deck can eliminate cracking of the deck under service load
conditions, thus increasing substantially the cracking load as well as the stiffness of the composite
bridge.
Razaqpur, A.G., and M. Nofal. 1990. Analytical Modeling of Nonlinear Behavior of Com-
posite Bridges. Journal of Structural Engineering-ASCE 116, 6: 1715-1733.
Yam and Chapman tested a number of simply supported and continuous composite beams
at Imperial College. Using different shear connectors, Hamada and Longworh tested a continuous
beam. The bridge is symmetric with four continuous spans, with the end spans equal to 21.34m
and the center spans equal to 27.43m. The deck was 178mm thick with a total width of 10.52m.
The girder spacing is wider than 12 times slab thickness limitation. This bridge appears to be one
tested by Burdette and Goodpasture.
Kennedy, J.B., and M. Soliman. 1992. Ultimate Loads of Continuous Composite Bridges.
Journal of Structural Engineering-ASCE 118, 9: 2610-2623.
Yield-line theory has been applied successfully to predict the collapse load of plate-like
structures. Some attempts have been made to apply the yield-line theory to predict the failure
loads of composite bridges. However, none of these attempts include the effect of transverse
beams (diaphragms) on the collapse load. It has been shown that diaphragms in a composite
G-46
bridge of the slab-on-I-steel girder type have a significant effect on the failure load of a composite
bridge. In this paper, the prediction of the most probable yield-line patterns of failure for rela-
tively wide composite bridges was based on a parametric study by means of a finite element anal-
ysis, as well as on test results from four composite bridge models.
Bridge models I and II were simply supported single spans while bridge model III and IV
were simply supported at their ends and continuous over a symmetrically placed interior support.
In bridge models I and III, diaphragms were connected with bolts, transferring shear only. In
bridge models II and IV, the connections were welded, transferring both shear and moment.
Bridge models I and II represented 1/5 scale single-span two-lane composite highway bridge. The
continuous bridge models III and IV were constructed to 1/10 scale chosen to comply with the
laboratory space limitations and loading frame capacity. To allow adequate rotation for the failed
segments, it is further assumed that: (1) The composite section is underreinforced, (2) shear,
punching, or bond failure are precluded, and (3) strain hardening of the steel is neglected, with the
steel having elastic perfectly plastic characteristics. The ultimate transverse negative moment of
resistance in the welded diaphragms is relatively large since both the transverse steel diaphragms
and the concrete deck slab contribute in a quasi-composite action.
Zhou and Novak obtained a collapse load of 3338kN when subjecting a 60ft X 27 ft. com-
posite bridge model to an eccentric HS-20 truck loading. McCarth and Melham tested a 6ftX4ft
composite bridge model loaded eccentrically and the failure load was 93kN.
Results from this investigation show that the yield-line method of analysis can be used to
reliably predict the collapse load of simple-span and continuous-span composite bridges sub-
jected to AASHTO truck loading. It is shown that the manner in which the transverse diaphragms
G-47
are connected to the main longitudinal beams or girders will have a significant influence on the
ultimate load-carrying capacity of relatively wide bridges.
Ayyub, B.M., Y.G. Sohn, and H. Saadatmanesh. 1992. Prestressed Composite Girders. I:
Experimental Study for Negative Moment. Journal of Structural Engineering-ASCE 118,
10: 2743-2762.
In 1975, Sarnes tested two composite steel concrete beams with prestressed slabs in a neg-
ative moment region. The test results showed that slab prestressing in the negative moment region
eliminated slab cracking and reduced stresses in the tension flange of the steel beams under ser-
vice loads. Kennedy and Grace studied the effect of the interaction between prestressing a portion
of the deck slab and shear connectors in a negative moment region analytically as well as experi-
mentally through the test of two one-eighth scale, two-span continuous bridges of composite con-
struction. The results showed that prestressing the deck increased the cracking load and stiffness
of the girders. Basu et al. studied experimentally and analytically the behavior of two-span par-
tially prestressed composite beams. They concluded that partial prestressing increased the load
capacity of the beam by about 20% and eliminated the problem of concrete deck cracking in the
negative moment region.
Johnson, R.P. 1997. Some Research on Composite Structures in the UK 1960-1985. Com-
posite Construction in Steel and Concrete III (Proceedings of an Engineering Foundation
Conference, Irsee, Germany, June 1996), ASCE.
G-48
Simply-supported encased T-beams It had been found at Lehigh and Imperial College that
beams with uniformly-spaced studs and distributed load behaved as well as those with a stud
spacing based on elastic theory. In UK, design was based on 0.8 P
u
in the belief that with 100%
design, the shear connection could fail without warning. The bending strength of T-beams with
the neutral axis in the slab was known to exceed the full-plastic value, and to be insensitive to the
properties of the concrete.
Early tests on continuous beams for buildings The four three-span beams tested at Cam-
bridge in 1962 had short side spans and were loaded only in the center span. It was known that for
plastic hinge analysis to be applicable, region of hogging moment would need large rotation
capacity, so one objective was to study the influence of fairly heavy slab reinforcement on the
behavior of there regions. It was found that slip had little influence on the overall behavior, and
the plastic hinge global analysis was applicable, provided that non-flexural modes of failure were
prevented. These could include longitudinal splitting, bursting of a deep haunch, local buckling,
vertical shear, and distortional lateral buckling. The point loads for these beams were applied
above the steel web, so that transverse bending moments in the concrete flange were low. In prac-
tice, floor slabs would be designed for distributed loading, perhaps by yield-line theory, so the
effects of this needed study, as did the influence of cracking of concrete on shear connectors, and
on deflections in service
Research on hogging moment regions of composite beams Fifteen push tests were done,
and it was found that the shear resistance of the studs was not much reduced by the tensile strain,
probably because the less stiff containment from the concrete reduced the axial tension in the
studs. Successive UK codes for composite bridges have found a different solution. The longitudi-
G-49
nal shear in a hogging region is calculated assuming the concrete to be uncracked. This usually
exceeds the cracked value by much more than 20%. These tension push tests also provided much
data on tension stiffening in cracked concrete slabs. It was thought that its effects could be
neglected in design. Almost 30 years later, the draft Eurocodes for bridges are allowing for it.
Thirteen double-cantilever composite beams were tested to failure. Hinge rotations were larger
than necessary for plastic hinge analysis. Fourteen more double-cantilever tests were reported in
1972. The objective was to determine moment-rotation curves for locally-buckling regions, and
hence to check whether the slenderness limits at the class boundaries in use for structural steel-
work were also applicable to hogging regions of composite beams. Where cantilever cross-girders
are used in a continuous composite bridge deck, a region of the slab near an internal support can
be subjected to almost uniform biaxial tension, significant biaxial horizontal shear, and punching
shear from wheel loads. The tests showed that the normal rules for transverse reinforcement and
crack-width control were satisfactory for the biaxial stressed region. Punching-shear tests on three
of the four quadrants of the slab gave most unexpected results. It was most reassuring to discover
that biaxial tension had no effect on resistance to punching shear.
Beams with transverse bending For several of the specimens, loading was distributed over
the slab. Nine two-span T-beams were then subjected to line loads along the edges of the concrete
flange, such that both the slab and the beam were expected to fail at the same load. In isolated T-
beams, transverse flexural failure of a concrete flange slightly reduced the bending resistance of
the composite beam, but there was compensation from the fact that effective flange breadth at
ultimate load usually exceeded the values given in codes of practice, which are based on elastic
analyses. In continuous floor systems, any weakness caused by biaxial straining of concrete above
a steel web is outweighed by membrane action in the slab, because hinge mechanisms in continu-
G-50
ous beams are incompatible with yield-line failures of slabs spanning between them. The behavior
of these specimens showed that hogging transverse bending above a steel web effectively pre-
stressed the concrete at the base of the shear connectors, so that little or no reinforcement was
needed. It was argued in 1971 that full-interaction design for SLS (Service Limit State) would
always provide enough connectors for ULS (Ultimate Limit State) and Part 5 of the UK bridge
Code was drafted on this basis. Work on Eurocode 4 in 1995 has shown that rational design of
shear connection at ULS for traveling loads is still an unsolved problem.
Vertical shear It has been assumed that the steel web alone resists the vertical shear in the
Code. A study of moment-shear interaction in continuous composite beams of compact section
was commenced in 1968. Tests showed that the slab resisted significant vertical shear, even if
cracked, except in the region close to the support where its reinforcement had yielded. In that
region, the steel element does resist the whole shear, but if the web is compact enough not to
buckle, interaction between resistance to bending and the shear can be neglected.
Closure When examined closely, many types of structural behavior are found be become
very complex, as failure is approached. An example is the state of stress in a concrete flange
where the material properties are time-dependent and highly non-linear.Particular attention must
be paid to the scale dependency. It can be significant for: (1)initial imperfections and residual
stresses in structural steelwork; (2) aggregate-interlock effects in cracked concrete subjected to
shear; (3) the magnitude and time-scale of shrinkage of concrete. Examples of potential problems
are the interaction between yield-line analysis of a slab and plastic hinge analysis of beams for
which the slab acts as a flange, and the work on punching shear.
G-51
G.3 BOX-GIRDER BRIDGES
Foutch, D.A. and P.C. Chang. 1982. A Shear Lag Anomaly, Journal of the Structural Divi-
sion 108,7: 1653-1658.
The purpose of this note is to report an interesting phenomenon associated with shear lag
in the flanges of box girders, not to derive new analytical technique. The author considered the
cantilever box girder with shear lag effect, which can be represented by the approximate longitu-
dinal displacement function for the flange (Reissner 1946)
9
. From a case study, a plot of the cen-
terline deflection of a cantilever tube shows an increase of 11% in the tip deflection, which results
from shear deformation and shear lag effect, compared to the deflection of an Euler-Bernoulli
beam theory. It was verified that the shear lag intensity function is highly dependent on the type of
loading that is applied to the box girder type of structures.
Kuzmanovic, B.O. and H.J. Graham. 1981. Shear Lag in Box Girders, Journal of the
Structural Division 107,9: 1701-1712.
The writers developed a simplified method for an idealized box girder type bridges based
on the principle of minimum potential energy of a simply cross-section by Reissner (1946). The
effective slab width is a function of geometrical factors, such as slab width, effective span length,
and the type of applied loads. Boundary conditions at bridge supports and the type of loading,
e.g., point load, uniform load and prestressing also affect this effective slab width. In case of uni-
9. Reissner, E (1946)., Analysis of Shear Lag in Box Beams by the Principle of Minimum Potential
Energy, Quarterly of Applied Mathematics, vol.4(3), pp. 268-278.
G-52
form loading and using several simplifications, it appears that manual calculation, in case study, is
still practical and yields width reduction factors close to those given by DIN 1075, although less
conservative.
Branco, F.A. and R. Green. 1985. Composite Box Girder Bridge Behavior During Con-
struction, Journal of Structural Engineering 111,3: 577-593.
As steel-concrete composite box girder bridges may have a flexible open box section prior
to the slab placement, excessive twist or distortion can arise under construction loading. Two
types of loading were categorized: (1) bending and (2) torsional. The authors conducted model
studies experimentally and analytically. A one-quarter scale model of an open box girder was
built and tested. Several tests were carried out on the girder with concentric and eccentric loading
and for different types of bracing system.
The finite strip method was employed to study the effectiveness of bracing systems on the
open box girders. The load conditions considered in the finite strip analysis correspond to the
experimental situations. Various bracing systems were analyzed, using the force method.
The authors concluded that ties connecting the top flanges and located at the eighth points
proved to be effective in preventing the bending distortion. The effect of torsional distortion can
be reduced with adequate web stiffeners, and distortional bracing can be added if high concen-
trated torsional moments occur in the girder.
G-53
Dezi, L. and L. Mentrasti. 1985. Nonuniform Bending-Stress Distribution (Shear Lag),
Journal of Structural Engineering 111,12: 2675-2690.
The writers used the theorem of minimum potential energy to develop the governing equa-
tions assuming as unknown the displacement of the beam axis and three functions, which describe
the warping of the central flanges and of the lateral cantilever flanges. The writers have shown
that the shear lag phenomenon is entirely independent of the constraint conditions of the flanges
and webs at the ends of the beam. In effect, even if the distribution of the stress at the end sections
is made to coincide with the distribution according to the Euler-Bernoulli theory, warping in sec-
tions still appears, along with a consequent nonuniform distribution of normal longitudinal stress.
Kostem, C.N., T.D. Hand and E.S. deCastro. 1984. Finite Element Modeling of Box-Beam
Bridges, Computing in Civil Engineering, pp. 223-232.
The accurate stress analysis of box-beam bridges requires approaches more sophisticated
than those used in the design process. The box-beam can be simulated minimally by using four
planar finite elements on each side of the beam. through the depth of the beam a large strain gradi-
ent exists. The elements to be employed for the webs must be high order finite elements. The
accuracy of any developed model should not be limited to the comparison of the stresses, but it
also must include the determination of the total moment carried by the beam. Authors suggested
that in the verification of the accuracy of any given finite element model, the comparisons should
not be limited to the deflection of the key points, but should be extended to the stresses as well.
G-54
Curb and parapet sections of the bridge structures also contribute stiffnesses sufficiently large to
be considered as beams.
Savage, I. and M.Nasim. 2001. The Woodrow Wilson Bridge - Innovation at Work, Pro-
ceedings of 2001 world Steel Bridge Symposium, pp. 27.1-27.10.
The Woodrow Wilson Bridge, the steel box girders were designed to be composite with
the concrete deck. The configuration and geometry of the boxes did not comply to the AASHTO
requirements for center-to-center flange spacing with respect to center-to-center flanges of adja-
cent boxes. A study was conducted to ascertain which portion of the deck and the bottom flange
fully participated in resisting stress. A detailed sub-model was created using a combination of
plate and shell elements.This model was loaded and the peak stresses in several cross sections of
the deck were studied. These results were used to verify that to the formulae specified by code
were still valid in determining the correct effective width of deck to consider for analysis.
Kumarasena, S.T. and R.J. McCabe. 2001. Bostons Storrow Drive Connector Bridge,
Informational, National Steel Bridge Alliance (NSBA).
The Storrow Drive Bridge is one of the largest steel box girders in North America. This
single-cell trapezoidal steel box girder bridge is designed to support the 76 wide road way con-
necting Bostons Storrow Drive and Leverett Circle with I-93. The 10 thick, 4500 psi concrete
deck slab was designed to act compositely with the main box girder, the transverse floor beams
(including the cantilever outriggers) and the longitudinal fascia girders.
G-55
Figure G-1: Four highway bridge test
Figure G-2: Stations of measurements
G-56
Figure G-3: Position and numbering of mechanical strain measurements
Figure G-4: Bridge Number 2 - Framing plan
G-57
Figure G-5: Bridge Number 2 - Cross section at Test Lines 1 and 3
Figure G-6: Set-up of loading system for Pratt bridge
G-58
Figure G-7: Bridge details
Figure G-8: Placing sequence of loading in Southern segment of bridge
G-59
Figure G-9: Cross-section of selected bridge
Figure G-10: Lateral distribution of deflections at midspan for center loading
Figure G-11: Bridge section
G-60
Figure G-12: Girder distribution of deflections and stresses for center lane loading
Figure G-13: Cross-section of I-40 bridge at floor beam location
G-61
Figure G-14: Tindall Bridge
Figure G-15: Example of stresses at midspan of panel 4-5 stringers

You might also like