Steel Bridge Design Handbook

CHAPTER 13

Bracing System Theory

and Design for I-Girders
and Tub Girders
February 2022
 © AISC 2022

by

American Institute of Steel Construction

All rights reserved. This book or any part thereof must not be
reproduced in any form without the written permission of the publisher.
The AISC and NSBA logos are registered trademarks of AISC.

The information presented in this publication has been prepared following recognized principles of design
and construction. While it is believed to be accurate, this information should not be used or relied upon
for any specific application without competent professional examination and verification of its accuracy,
suitability and applicability by a licensed engineer or architect. The publication of this information is not a
representation or warranty on the part of the American Institute of Steel Construction, its officers, agents,
employees or committee members, or of any other person named herein, that this information is suitable for
any general or particular use, or of freedom from infringement of any patent or patents. All representations
or warranties, express or implied, other than as stated above, are specifically disclaimed. Anyone making
use of the information presented in this publication assumes all liability arising from such use.

Caution must be exercised when relying upon standards and guidelines developed by other bodies and
incorporated by reference herein since such material may be modified or amended from time to time sub-
sequent to the printing of this edition. The American Institute of Steel Construction bears no responsibility
for such material other than to refer to it and incorporate it by reference at the time of the initial publication
of this edition.

Printed in the United States of America

Foreword

The Steel Bridge Design Handbook covers a full range of topics and design examples to provide bridge
engineers with the information needed to make knowledgeable decisions regarding the selection, design,
fabrication, and construction of steel bridges. The Handbook has a long history, dating back to the 1970s
in various forms and publications. The more recent editions of the Handbook were developed and
maintained by the Federal Highway Administration (FHWA) Office of Bridges and Structures as FHWA
Report No. FHWA-IF-12-052 published in November 2012, and FHWA Report No. FHWA-HIF-16-002
published in December 2015. The previous development and maintenance of the Handbook by the
FHWA, their consultants, and their technical reviewers is gratefully appreciated and acknowledged.
This current edition of the Handbook is maintained by the National Steel Bridge Alliance (NSBA), a
division of the American Institute of Steel Construction (AISC). This Handbook, published in 2021, has
been updated and revised to be consistent with the 9th edition of the AASHTO LRFD Bridge Design
Specifications which was released in 2020. The updates and revisions to various chapters and design
examples have been performed, as noted, by HDR, M.A. Grubb & Associates, Don White, Ph.D., and
NSBA. Furthermore, the updates and revisions have been reviewed independently by Francesco Russo,
Ph.D., P.E., Brandon Chavel, Ph.D., P.E., and NSBA.
The Handbook consists of 19 chapters and 6 design examples. The chapters and design examples of the
Handbook are published separately for ease of use, and available for free download at the NSBA website,
[Link]/nsba.
The users of the Steel Bridge Design Handbook are encouraged to submit ideas and suggestions for
enhancements that can be implemented in future editions to the NSBA and AISC at solutions@[Link].
 TECHNICAL REPORT DOCUMENTATION PAGE

1. Title and Subtitle 2. Report Date

Steel Bridge Design Handbook February 2022
Chapter 13: Bracing System Theory and Design
for I-Girders and Tub Girders

3. Original Author(s) 4. Revision Author(s)

Todd Helwig, Ph.D. and Joseph Yura, Ph.D. Michael Grubb, P.E.
(University of Texas at Austin) (M. A. Grubb and Associates, LLC)

5. Sponsoring Agency Name and Address 6. Revision Performing Organization Name

National Steel Bridge Alliance, a division of the and Address
American Institute of Steel Construction HDR, Inc.
130 E. Randolph, Suite 2000 301 Grant Street, Suite 1700
Chicago, IL 60601 Pittsburgh, PA 15219

7. Supplementary Notes
The previous edition of this Handbook was published as FHWA-HIF-16-002 and was developed to be
current with the 7th edition of the AASHTO LRFD Bridge Design Specifications. This edition of the
Handbook was updated to be current with the 9th edition of the AASHTO LRFD Bridge Design
Specifications, released in 2020.

8. Abstract
This volume discusses the design of bracing systems for the superstructures of straight and curved
girder systems. I-girder and box shaped members are covered. Bracing for other types of bridges, such
as truss, arch or towers is not specifically addressed; however much of the information included in this
volume may be applicable. Bracing systems serve a number of important roles in both straight and
horizontally curved bridges. The braces provide stability to the primary girders as well as improving
the lateral or torsional stiffness and strength of the bridge system both during construction and in
service. Depending on the geometry of the bridge, braces may be designated as either primary or
secondary members.

The Engineer needs to recognize the importance of the bracing systems and bracing member design for
appropriate construction and in-service stages. This volume provides an overview of the design
requirements for the braces so that engineers can properly size the members to provide adequate
strength and stiffness.

This volume provides: a) an overview of bracing utilized for I-girders, b) a discussion of the bracing
systems for tub girders, and c) design requirements for the members and connections of bracing
systems.

9. Keywords 10. AISC Publication No.

Steel Bridge, Bracing, Bracing System, Bracing B913-22
Design, I-girder, Tub-girder, Curved Girder, Top
Flange Lateral Bracing, Lean-on Bracing
 Steel Bridge Design Handbook:
Bracing System Theory and Design for I-Girders and
Tub Girders
Table of Contents
1.0 INTRODUCTION ................................................................................................................. 1
1.1 Torsional Behavior of Open and Closed Girders ............................................................. 2
1.2 Lateral Torsional Buckling .............................................................................................. 5
1.3 Categories of Bracing ...................................................................................................... 5
2.0 BRACING OF I-GIRDERS ................................................................................................... 7
2.1 General Requirements ...................................................................................................... 8
2.1.1 Cross-Frame Spacing and Proportions.................................................................... 8
2.1.2 Top and Bottom Flange Lateral Systems .............................................................. 10
2.2 Cross-Frame Forces in Horizontally Curved Girders .................................................... 11
2.3 Stability Bracing of I-Girders ........................................................................................ 12
2.3.1 Torsional Bracing Design Requirements, T ........................................................ 14
2.3.2 Stiffness of Cross-Frame and Diaphragm Systems b .......................................... 16
2.3.3 Web Distortional Stiffness, sec ............................................................................ 18
2.3.4 In-Plane Stiffness of Girders, g ........................................................................... 21
2.3.5 Connection Stiffness, conn .................................................................................... 22
2.4 Effects of Support Skew ................................................................................................ 23
2.5 Lean-On and Staggered Bracing .................................................................................... 27
2.5.1 Lean-On Bracing ................................................................................................... 27
2.5.2 Staggered Bracing ................................................................................................. 32
2.6 System Buckling of Interconnected Girders .................................................................. 32
2.7 Lateral Bracing Systems ................................................................................................ 36
2.8 Continuous Bracing ....................................................................................................... 38
3.0 BRACING OF TUB GIRDER SYSTEMS.......................................................................... 39
3.1 Top Flange Lateral Truss ............................................................................................... 40
3.1.1 Top Lateral Brace Forces ...................................................................................... 41

i
 3.1.2 Selecting the Top Lateral Bracing Layout ............................................................ 45
3.1.3 Determining the Brace Forces .............................................................................. 46
[Link] Torsion ................................................................................................ 47
[Link] Sloping Webs ...................................................................................... 48
[Link] Vertical Bending ................................................................................. 49
3.1.4 Top Flange Truss Details ...................................................................................... 51
3.1.5 Controlling Global Lateral Buckling .................................................................... 52
3.2 Intermediate Internal Cross-Frames ............................................................................... 55
3.2.1 Tub Girder Distortion ........................................................................................... 56
3.2.2 Internal Cross-Frame Details ................................................................................ 59
3.3 End Diaphragms............................................................................................................. 59
3.3.1 Diaphragm Strength Design Requirements .......................................................... 61
3.3.2 Diaphragm Stiffness Design Requirements .......................................................... 62
3.4 Intermediate External Cross-Frames.............................................................................. 64
3.4.1 Analysis Approaches for Intermediate External Diaphragms .............................. 65
3.4.2 Spacing of Intermediate External K-frames ......................................................... 66
3.4.3 Forces in Intermediate External K-frames ............................................................ 66
4.0 BRACING MEMBER DESIGN AND CONNECTION DETAILS ................................... 69
4.1 Design of Tees and Angles ............................................................................................ 69
4.1.1 Tension Members.................................................................................................. 69
4.1.2 Compression Members ......................................................................................... 69
[Link] Single angles ....................................................................................... 69
[Link] Tees and Double Angles ..................................................................... 70
4.2 Fatigue Design of Cross-frame Members ...................................................................... 71
4.3 Welded and Bolted Connection Details ......................................................................... 71
5.0 SIMPLIFIED GEOMETRIC PROPERTIES FOR TUB GIRDERS ................................... 76
5.1 Shear Center, ey, for Open and Quasi-Closed Sections [61].......................................... 76
5.2 Monosymmetry Coefficient, x – Open section only [62] ............................................. 77
5.3 Warping Moment of Inertia, Cw – Open section only [62] ............................................ 77
6.0 REFERENCES .................................................................................................................... 78

ii
 List of Figures

Figure 1 Warping Stiffness is Related to the Bending Stiffness of the Plate Elements................. 2
Figure 2 Shear Flow in Tub Girder Due to Saint-Venant Torsion ................................................ 4
Figure 3 Categories of Bracing ...................................................................................................... 6
Figure 4 Collapse of a Bridge over the Tennessee River due to Insufficient Bracing Provided
During Steel Erection ...................................................................................................................... 7
Figure 5 Web Distortion .............................................................................................................. 13
Figure 6 Restraining Forces ......................................................................................................... 13
Figure 7 Bending Stresses in Singly Symmetric Section............................................................. 16
Figure 8 Diaphragm Stiffness, b................................................................................................. 17
Figure 9 Stiffness Formulas for Twin Girder Cross-Frames [21]................................................ 17
Figure 10 Web Stiffener Geometry.............................................................................................. 19
Figure 11 Diaphragm and Cross-Frame Geometry for Rolled Sections ...................................... 19
Figure 12 Beam Load from Braces .............................................................................................. 22
Figure 13 Plan View of Bridge with Skewed Supports ............................................................... 23
Figure 14 Brace Orientations for Bridges with Skewed Supports ............................................... 24
Figure 15 Bent Plate Connection Detail Frequently Used in Bridges with Skewed Supports .... 26
Figure 16 Half-Pipe Web Stiffener .............................................................................................. 26
Figure 17 Summation of P Concept for Sideway Frame Stability ............................................ 27
Figure 18 Beams Linked Together at Compression Flange......................................................... 27
Figure 19 Graph of P Concept for Beams .................................................................................. 28
Figure 20 Lean-on Cross-Frame Bracing ..................................................................................... 29
Figure 21 Plan View of Bridge with Lean-On Cross-Frame Bracing .......................................... 29
Figure 22 Lean-on Bracing Layout in Bridge with Large Numbers of Girders ........................... 30
Figure 23 Stiffness and Strength Requirements for Lean-On Cross-Frames ............................... 31
Figure 24 Staggered Cross-Frame Layout ................................................................................... 32
Figure 25 System Buckling of a Twin Girder Widening, where the system has buckled out of
plane nearly 10 inches during deck placement ............................................................................. 33
Figure 26 Comparison of Individual Buckling Mode and System Buckling Mode ..................... 34
Figure 27 Plan View of Typical Lateral (Relative) Bracing System ............................................ 37

iii
Figure 28 Twin Tub Girder System ............................................................................................. 39
Figure 29 Types of Bracing Systems for Tub Girders ................................................................. 39
Figure 30 Geometric Layout and Equivalent Plate Thickness of Top Lateral Systems [40] ...... 40
Figure 31 Top Lateral Truss Forces for Various Tub Girder Bracing Systems........................... 44
Figure 32 Deformations of different box girder bracing systems ................................................ 45
Figure 33 Diagonal Lateral Brace Forces Due to Torsion in a Tub Girder ................................. 47
Figure 34 Strut Forces from Torsion ........................................................................................... 48
Figure 35 Strut Forces from Top Flange Loads ........................................................................... 48
Figure 36 Tub Girder Vertical Bending Stresses ......................................................................... 49
Figure 37 Bending Induced Truss Forces .................................................................................... 51
Figure 38 Strut eccentricity in a tub girder cross section............................................................. 52
Figure 39 Effect of teq on global buckling on a tub girder section using a X-type lateral bracing
system ........................................................................................................................................... 54
Figure 40 Effect of Partial End Panel Bracing on Girder Buckling Stress .................................. 55
Figure 41 Internal Intermediate Cross-Frame Layouts for Tub Girders ...................................... 56
Figure 42 Sources of Torsion in a Tub Girder ............................................................................. 56
Figure 43 Pure Torsional and Distortional Components in a Tub Girder.................................... 57
Figure 44 Sign Convention for Eccentricity ................................................................................ 57
Figure 45 Approximate Distortional Forces in Intermediate Cross-Frames in a Tub Girder ...... 58
Figure 46 Typical End Diaphragm Geometry ............................................................................. 60
Figure 47 Non-continuous flanges illustrated by a 3D FEM of the end diaphragm of a twin box
girder model .................................................................................................................................. 60
Figure 48 Girder end torsional demand that acts on diaphragms and the resulting shears .......... 61
Figure 49 Idealized Rectangular Diaphragm Plate ...................................................................... 62
Figure 50 Girder and Diaphragm Deformations .......................................................................... 63
Figure 51 Relative Deformation between Adjacent Girders........................................................ 64
Figure 52 K-frame Geometry....................................................................................................... 67
Figure 53 External K-Frame Forces ............................................................................................ 67
Figure 54 Eccentrically loaded WT section ................................................................................. 71
Figure 55 Improper orientation and sizing of WT section used for top flange lateral bracing
member ......................................................................................................................................... 72

iv
Figure 56 Strut offset below diagonal .......................................................................................... 73
Figure 57 Large gusset plates used so working lines of bracing members intersect ................... 74
Figure 58 Shim plates can significantly increase the eccentricity of WT sections ...................... 74
Figure 59 Partial depth end diaphragm. ....................................................................................... 75
Figure 60 Calculation of Shear Center for Open and Quasi-Closed Sections ............................. 76

v
1.0 INTRODUCTION

Bracing systems serve a number of important roles in both straight and horizontally curved bridges.
The braces provide stability to the primary girders as well as improving the lateral or torsional
stiffness and strength of the bridge system both during construction and in service. Depending on
the geometry of the bridge, braces may be designated as either primary or secondary members. In
the AASHTO LRFD Bridge Design Specifications (hereafter referred to as the AASHTO LRFD
BDS) [1], the member designation as primary or secondary is typically assigned depending on
whether or not the member or component transmits gravity loads through a necessary as-designed
load path. For example, a first-order analysis on a straight bridge with little or no skew during
construction will often result in little or no force in the cross-frame members, which are designated
as secondary members in such bridges. However, in some situations during construction, the
removal of one or more of these bracing members can potentially result in a partial or complete
collapse of the structure due to instabilities that can develop as a result of the larger unbraced
lengths that result from the removal of these members. In cases such as this, the engineer needs
to recognize the importance of the brace and design the members accordingly despite their
designation as secondary members. This volume provides an overview of the design requirements
of the braces so that engineers can properly size the members to provide adequate strength and
stiffness.

In general, this volume discusses the design of bracing systems for the superstructures of straight
and curved girder systems. I-girder and box shaped members (e.g., tub girders) are covered
throughout this volume. Bracing for other types of bridges, such as truss, arch or towers is not
specifically addressed; however much of the information included in this volume may be
applicable.

The volume has been divided into five primary sections. Following this introduction, an overview
of bracing utilized for I-girders is covered. A discussion of the bracing systems for tub girders is
then provided. The next section of this volume outlines the design requirements for the members
and connections of bracing systems. The final section contains simplified solutions for the
calculation of geometric properties for tub girders.

Regardless of whether the bracing systems are utilized in straight or horizontally curved girders, a
clear understanding of the torsional behavior of both I-shaped and tub girder sections is important.
The need for torsional stiffness in horizontally curved girders is relatively obvious since the girders
are subjected to large torques due to the geometry of the bridge. However, understanding the
necessity of adequate torsional stiffness in straight girders is also important since lateral-torsional
buckling often controls the design of the girders during construction. In many sections, such as
tub girders, the presence of bracing dramatically impacts the torsional stiffness of the section.
Lateral instability of flexural members always involves torsion of the cross section. Therefore, the
remainder of this introductory section is focused on the torsional stiffness of open and closed cross
sections as well as a discussion of the buckling behavior of steel bridge systems.

1
1.1 Torsional Behavior of Open and Closed Girders

Torsional moments are resisted by the shear stresses on the girder cross section. The torsional
resistance in thin-walled structures is usually categorized as either Saint-Venant torsional stiffness
or warping torsional resistance. The Saint-Venant stiffness is often referred to as uniform torsion
since the stiffness does not vary along the length and is also not sensitive to the support conditions
of the section. St. Venant torsion results in a pure shear deformation in the plane of the plates that
make up the member.

The warping torsional resistance, on the other hand, is often referred to as non-uniform torsion
since the stiffness is associated with the bending deformation in the plane of the individual plates.
The warping stiffness of a section is related to the member’s resistance to warping deformation.
Two I-shaped sections subjected to a torque at the ends are shown in plan in Figure 1 to illustrate
warping deformation and also warping stiffness. Figure 1a shows that warping deformation
consists of a twist of the flanges relative to each other about a vertical axis through the web.
Warping deformation distorts the cross section such that it no longer is a plane section because the
two flanges have distorted relative to each other. Twist about the longitudinal axis of the member
in Figure 1a is prevented at one end, however the warping deformations are not restrained. Since
the section is free to warp along the entire length, the flanges remain straight as they twist relative
to each other and the member only possesses St. Venant torsional stiffness.

Figure 1 Warping Stiffness is Related to the Bending Stiffness of the Plate Elements.

The I-shaped section in Figure 1b has both twist and warping deformation prevented at one end.
With warping restrained at just one location along the length, the member cannot twist without
bending the flanges. Since the flanges must bend if the member twists, the section therefore has
warping stiffness. The warping torsion produces nonuniform longitudinal stresses in the flanges
of the member; these stresses are also sometimes called flange lateral bending stresses.

Many members do not have a physical restraint preventing warping as shown in Figure 1b,
however the member still has warping stiffness if twist is prevented at a minimum of two points
along the longitudinal axis. The twist restraint can come from sources such as cross-frames that
prevent the section from rotating about the longitudinal axis, but otherwise do not specifically

2
restrain warping deformation of the section. Since the bending stiffness is very sensitive to the
unsupported length, the warping stiffness is highly variable with the unbraced length.

In general, both Saint-Venant and warping torsional stiffness are developed when thin-walled
members are twisted. The torsional moment resistance, TT, of a section is a function of the uniform
torsional (TUT) and warping torsional (TW) components as follows:

TT = TUT + TW (1)

The uniform torsional component can be expressed as follows:

d
TUT = GJ (2)
dx

where G is the shear modulus, J is the torsional constant,  is the rotation of the cross section, and
x denotes the longitudinal axis of the member. The torsional constant of a thin-walled open section
is given by the following expression:

1
J= 
3 i
bi t i3 (3)

where bi and ti are the respective width and thickness of the plate elements that make up the cross
section of the girder. The torsional constant for single cell box or tub girders is given by

2
4A 0
J= (4)
 bi / ti
i

where A0 is the enclosed area of the cross section of the box girder, and the variables bi and ti in
the summation are the respective width and thickness of the ith plate that make up the cross section.
For example, in a box or tub girder with a cross section made up of four plates, the denominator
in Equation 4 is calculated by simply summing the width-to-thickness ratios of the four plate
elements. A0 is typically defined by the area enclosed from the mid-thickness of the plates that
make up the cross section.

The warping torsional component can be expressed as follows:

d 3
TW = EC w (5)
dx 3

where E is the modulus of elasticity, and Cw is the warping constant. For I-shaped sections bent in
the plane of the web, the warping constant is given by the expression:

CW = It ho2  (1-) (6)

3
 I yc
= (7)
Iy

where Iyc and Iy are the respective moments of inertia for the compression flange and the entire
section about an axis through the web, and ho is the spacing between flange centroids. For a doubly
symmetric section, the value of  is 0.5 and Equation 6 reduces to the following expression:

I y ho 2
Cw = (8)
4

A rigorous theory for warping torsion was established by Vlasov [2]. he warping torsional stiffness
often plays an important role in the total stiffness in girders with an open cross section such as I-
shaped girders. For open sections with a relatively long length, the St. Venant stiffness dominates
the total stiffness, while for shorter segments the warping torsional stiffness plays a much more
significant component in the total stiffness. Closed box or tub girders are usually dominated by
Saint-Venant torsion due to the closed cross section and the longitudinal normal stresses due to
warping torsion are usually negligible [2]. The large Saint-Venant stiffness of a box or tub girder
provides a torsional stiffness that may be 100~1000 times that of a comparable I-section.

The shear stress due to Saint-Venant torsion can be solved using Prandtl’s membrane analogy [2].
For example, for girders with a single cell cross-section, a uniform shear flow, q, develops along
the perimeter of the box and can be determined using the Bredt’s equation:

TT
q = t = (9)
2A 0

in which t is the thickness of the plate, and  is the shear stress, which is essentially uniform through
the thickness of the plates. The distribution of torsional shear stress is demonstrated for a tub girder
in Figure 2.

q= t


q
Figure 2 Shear Flow in Tub Girder Due to Saint-Venant Torsion

Although the torsional warping stresses in the box or tub girder are usually negligible, significant
warping stresses due to the cross-sectional distortion of tub girders may develop, as discussed later
in this volume. The large torsional stiffness of box or tub sections in bridges is the result of the
closed cross section once the concrete deck cures. During construction of tub girders, the steel

4
girder is an open section and requires bracing to be designed by the engineer that will stiffen the
tub girder. The bracing systems for tub girders are covered later in this volume.

1.2 Lateral Torsional Buckling

The overall stability of the girder system can be improved by either altering the geometry of the
individual girders or by providing braces to reduce the unsupported length of the girders.
Providing bracing is usually the more efficient solution and there are a variety of bracing systems
that are provided as is discussed later in this volume. The elastic buckling solution for doubly-
symmetric beams is given in the following solution derived by Timoshenko [3]:

2
π  π E
M cr = EIy GJ +   I y C w
Lb  Lb  (10)

where, Lb is the unbraced length, and the other terms are as defined above. The first term under
the radical in Eq. 10 relates to the St. Venant torsional stiffness, while the second term within the
radical reflects the warping stiffness of the beam. Equation 10 was derived for the case of uniform
moment loading. Most design specifications make use of solutions derived for uniform moment
and then use a moment gradient modifier (Cb) applied to the uniform moment solution to account
for the benefits of variable moment. In the derivation of the buckling expression, Timoshenko
assumed that the ends of the sections were restrained from twist. Although restraint against lateral
translation of the section was stated in the original derivation, the assumed support condition was
never applied or required to derive the expression. Therefore, effective bracing of beams can be
achieved by restraining twist of the section, which is the primary means of bracing I-shaped
members in bridges with the use of cross-frames or diaphragms. Twist of the section can also be
restrained by preventing lateral translation of the compression flange of the section, which
therefore introduces another means of bracing. Both lateral and torsional bracing requirements are
discussed later in this volume.

Lateral torsional buckling of closed box girders is not typically a concern due to the extremely
large torsional stiffness of the closed cross section. During construction of tub girders, a quasi-
closed section is typically created by using bracing that simulates the stiffness of a top plate (known
as top flange lateral bracing). Global buckling failures of tub girder sections have occurred during
construction when proper top flange lateral bracing was not provided [4].

1.3 Categories of Bracing

Bracing systems that are used to increase the stability of structural systems can be divided into the
four categories represented in Figure 3. This section introduces the basic bracing categories, which
are covered in more detail in the remainder of this volume. Although the focus of this document
is on bracing for the superstructure elements of steel bridges, the basic categories also apply to
columns and frames, which is demonstrated in Figure 3. Diagonal bracing such as that depicted in
Figure 3a fits into the category of relative bracing since the braces control the relative movement
of two adjacent points at different lengths along the main members. The lateral trusses that are
used to create quasi-closed tub girders and the bottom flange bracing on I-girder systems to

5
improve the lateral stiffness fit into the category of relative bracing. Another very common type
of bracing in steel bridges are nodal systems such as those depicted in Figure 3b. Nodal braces
control the deformation of a single point along the length of the member. Cross-frames or plate
diaphragms that frame between adjacent girders fit into the category of nodal torsional bracing
since the cross-frames restrain girder twist at a single location. The unbraced length of the girders
is defined as the spacing between adjacent nodal braces. Figure 3c illustrates continuous bracing
in which the bracing is connected along the entire length of the main members. One form of
continuous bracing is the concrete deck that is often connected to the girder top flange through the
welded shear studs. The slab provides both continuous lateral and torsional restraint to the girders.
In typical bridge design, the lateral restraint provided by the concrete deck is taken into account;
but the torsional restraint offered by the deck is ignored. Lean-on bracing concepts such as those
depicted in Figure 3d are commonly used in framed structures where lightly loaded members can
provide bracing to other heavily loaded members by providing struts to connect the main members.
The heavier loaded members lean on the other members for stability. The lean-on concepts also
work by providing bracing between one set of girders and then leaning several girders on a single
brace.

Some bracing systems may fit into multiple categories; however, the bracing design is usually
based upon selecting one design concept for a specific category. The design philosophies for the
different categories are covered in more detail in the remainder of this volume. The bracing
behavior for I-shaped girders is discussed first followed by the behavior for tub girders.

Figure 3 Categories of Bracing

6
2.0 BRACING OF I-GIRDERS

Bracing in steel bridges serves the dual purpose of providing overall stability of the girders as well
as directly increasing the stiffness and strength of the system. Typically, bridge girders are braced
by truss-type cross-frames or solid web diaphragms, while lateral bracing (bracing in a horizontal
plane between adjacent girders) is typically used when it is necessary to resist larger lateral loads
and limit lateral deflections during construction. In addition to resisting lateral and/or torsional
loads, the cross-frame type bracing also aids in distributing gravity loads in the structure.
Additionally, from a stability perspective, effective bracing must have sufficient stiffness and
strength. Therefore, in subsequent discussions of stability bracing both stiffness and strength
requirements are outlined.

Most reported failures of I-girder bridges occur during the construction stage prior to the hardening
of the bridge deck. Except for lifting problems with curved girders, the majority of construction
failures can be traced to bracing issues. The collapse of the partially erected State Highway 69
plate girder bridge over the Tennessee River on May 16, 1995 shown in Figure 4 illustrates the
importance of properly designed bracing. In this case, the three 14 ft. deep plate girders with a
main span of 525 ft had been successfully in place for a few days. There was a substantial bottom
flange relative lateral bracing system (a lateral truss) to help resist wind and other lateral forces.
There was no top flange relative lateral system provided because the bridge deck performs that
function after the concrete hardens. Cross-frames with double angle members in a K-frame
configuration between girders were specified to stabilize the top flange of the girders, with the
highest flange stresses occurring during the planned deck pour. To facilitate erection of the third
girder, during the successful lifting operation the contractor had only a few of the cross-frames in
place to support just the dead weight of the girders. During the process of installing the missing
cross-frames of the third erected girder, one of the previously erected cross-frames was removed
because an angle member had sustained some damage. With the cross-frame removed, the
unbraced length of the top flange was too large to support the dead weight of the steel and the
girder buckled leading to a total collapse of the bridge as shown in Figure 4.

Figure 4 Collapse of a Bridge over the Tennessee River due to Insufficient Bracing
Provided During Steel Erection

7
Bracing systems for I-girders may consist of combinations of cross-frames, solid diaphragms, as
well as top and/or bottom flange lateral truss systems. In this volume, the term cross-frame will
generally be used to also represent solid diaphragms since their functions are similar. For straight-
girder bridges, the bracing system design is typically dominated by stability issues and differential
deflection issues resulting from support skew. In horizontally curved-girder bridges, the effects of
torsion and lateral flange bending generally control the bracing design.

This section discusses bracing on two levels: 1) Bracing needed to transfer loads within a bridge
system, and 2) Bracing required to provide stability to the bridge system. In the following
subsections, the design requirements and geometric arrangements for bracing systems affected by
torsion, stability and skew are presented. Sections 2.1 and 2.2 provide general design requirements
necessary to properly transfer the static and transient loads within a bridge system. Section 2.3
provides details for the computations associated with determining the stability requirements of a
given bridge system. Also discussed within this section are the effects that support skew has on
bracing systems, the use of lean-on and staggered bracing, system buckling of interconnected
girders, lateral bracing systems, and continuous bracing systems.

The details and equations provided in Section 2.3 can be used to determine the stability bracing
forces. These equations and methods are usually sufficient for typical I-girder bridges, including
straight, curved, and skewed bridges. Using these equations, the stability bracing forces are
additive to the bracing forces resulting from a first-order type of analysis (dead load, live load,
etc.). For more complex bridges, or as alternative to using the equations discussed in Section 2.3,
a large displacement analysis can be used to determine the bracing forces. In this type of analysis,
the bracing forces will include the bracing forces required to transfer loads within the bridge
system and the bracing forces required for stability. When a large displacement analysis is used,
the effects of imperfections must be considered in order to achieve the desired analysis results.
Furthermore, the equations provided in Section 2.3 will generally yield conservative bracing
forces, as compared to those that result from a large displacement analysis.

2.1 General Requirements

For I-girder bridge systems the most common bracing is a discrete torsional system consisting of
cross-frames with a K- or X-configuration. Solid plate or channel diaphragms are also used. The
braces are usually fabricated from angles or of solid diaphragms constructed with channel-type
sections for ease in attachment to girder stiffeners. In some cases, top or bottom lateral truss
bracing (a relative brace system) may be needed as temporary bracing during construction or
permanent bracing to mainly resist wind loads. The requirements given below for cross-frames
and lateral flange bracing are generally taken from the AASHTO LRFD BDS [1].

2.1.1 Cross-Frame Spacing and Proportions

Cross-frames are necessary at all supports of straight and curved I-girder bridges to transfer lateral
loads from the superstructure to the bearings, to provide no-twist boundary conditions for lateral
buckling evaluation and transmit torsional overturning and uplift forces to the foundation. For
straight girders, previous bridge specifications required that intermediate cross-frames be spaced

8
at no more than 25 feet. Since the first publication of the AASHTO LRFD BDS in 1994, this
requirement has been replaced with the statement that the cross-frame spacing should be
determined by a rational analysis [1]. The elimination of the specified maximum spacing for
straight girders is intended to result in a reduction in the number of fatigue-prone attachments.
However, various Owner-agencies may have their own requirements and preferred practices
regarding cross-frame spacing that may supersede the AASHTO LRFD BDS.

To determine the spacing of intermediate cross-frames, at a minimum, a rational analysis should

consider the following:

• The need for cross-frames during all stages of the assumed construction staging, as well as
in the final condition.
• Lateral support to bottom flange for deck overhang construction brackets.
• Sufficient transfer of lateral wind loads from the bottom of the girder to the deck.
• Stability of the bottom flange for loads producing compression in the bottom flange.
• Stability of the top flange for loads producing compression, especially during the
construction stage or for non-composite systems.
• Control of flange lateral bending effects.
• Distribution of vertical dead and live loads applied to the structure.

Typically, cross-frames play a more active role in horizontally curved steel girder bridges
compared to straight girder bridges without significant skew. Curved girders are subjected to
combined bending and torsion. Without cross-frames, the flanges of the I-section would have to
be prohibitively large to control the flange lateral bending stresses (warping normal stresses) that
are combined with the major-axis bending stresses. Cross-frames allow the girders to work
together as a system to resist the torsion on the curved bridge and they limit the lateral bending
stresses by supplying torsional supports along the span. Therefore, cross-frame members in curved
bridges are considered primary members and should be designed for forces computed by
appropriate analysis methods (see Section 2.2).

In curved I-girder bridges, the cross-frames should be orientated in a radial manner throughout the
span, whenever possible (In curved and skewed bridges, cross-frames at the supports may be
placed along the skew or in a radial manner and are often orientated in a radial manner within the
span.) The spacing of the cross-frames, Lb, must control lateral buckling of the compression flange
and limit the magnitude of the flange lateral bending stresses. Davidson et al. [5] developed an
equation for the spacing required to limit the flange lateral bending stresses to a specified
percentage of the flange major-axis bending stress. However, AASHTO specifies a maximum Lb
limit as shown in the following expression:

E
L b  0.1R  π rt  30 ft (11)
0.7Fy

where R is the radius of curvature and rt is the radius of gyration for lateral buckling. The reasoning
for these limits is given in NSBA’s Steel Bridge Design Handbook: Structural Behavior of Steel
[6].

9
For straight girders with skewed supports, the relative displacement of the two ends of a cross-
frame or diaphragm system can introduce significant live load forces into stiff bracing systems,
especially near supports. The skew also affects the stability brace stiffness and strength
requirements. As specified in AASHTO LRFD BDS Article [Link], where the supports are not
skewed more than 20 degrees, intermediate diaphragms may be placed in contiguous skewed lines
parallel to the skewed supports. Where the supports are skewed more than 20 degrees, cross-
frames must be placed perpendicular to the girders in contiguous or discontinuous lines. In cases
where supports are skewed more than 20 degrees, it may be advantageous to place cross-frames in
discontinuous lines to reduce the transverse stiffness of the bridge, particularly near interior
supports. Placing the cross-frames in discontinuous lines can decrease cross-frame forces but
increase flange lateral bending effects (see Sections 2.3.2 and 2.3.3). Recommended beneficial
framing arrangements utilizing discontinuous cross-frame lines are discussed further in AASHTO
LRFD BDS Article C6.7.4.2.

Diaphragms and cross-frames for rolled beams and plate girders should be as deep as practicable,
but as a minimum should be at least 0.5 times the beam depth for rolled beams and 0.75 times the
girder depth for plate girders. Cross-frames should contain diagonals and both top and bottom
chords even if analysis shows that a chord force is zero. The flexural stiffness of a cross-frame
without a top or bottom chord is substantially reduced and may become ineffective as a stability
brace. Several orientations are possible for a cross-frame, such as an X-shape with top and bottom
chords, and K-shape where the diagonals intersect the bottom chord, or a K-shape where the
diagonals intersect the top chord. Cross-frame truss assemblies are preferably field delivered as a
single unit rather than individual pieces for erection efficiency as well as assisting the erector with
girder alignment. Efficient cross-frames are typically as deep as practical so that the diagonals of
the cross-frame have large enough angles to prevent the gusset-type plates at the ends of the cross-
frame from becoming too large. For cases requiring relatively shallow cross-frames, the diagonals
of X-systems may be subjected to large axial forces with large unbraced lengths. In these cases,
K-frame systems should be considered.

2.1.2 Top and Bottom Flange Lateral Systems

In steel I-girder bridges, the need for lateral bracing should be investigated for all stages of
construction, and the final condition. Lateral bracing may be required to resist lateral forces from
wind and during construction, when the deck is not in place. When lateral bracing is required, it
should be placed either in or near the plane of the flange being braced. Connecting the lateral
bracing directly to the flange with a bolted connection (with or without a connection plate) is a
preferred practice, as it eliminates the need for connection elements on the girder web that can be
sensitive to fatigue issues. In addition, connecting directly to the flange provides a direct load path
that improves the structural efficiency.

To help prevent lateral movement of the structural system during construction, especially in spans
greater than 250 feet, it may be desirable to consider providing either temporary or permanent
flange level lateral bracing. Flange level lateral bracing may also be needed in deck replacement
projects on long span bridges. In the final condition, the concrete deck resists lateral wind loads
and prevents significant horizontal movement of the structure. However, if the deck requires

10
replacement and is removed, lateral deflections due to wind can be excessive in long span bridges
without lateral bracing. The large lateral flexibility may make the construction workers
uncomfortable or can result in system instability. Essentially, a lateral bracing system will stiffen
a non-composite structure significantly, as compared to one without any lateral bracing.

As noted above, top flange level lateral bracing is preferred. When located in the same plane as
the top flange, the bracing is near the neutral axis of the final composite structure. As such, the
bracing located at the top flange will not be subjected to significant live load forces in the final
condition. In general, forces during construction related to wind and dead load will govern the
design of top flange level lateral bracing. If top flange lateral bracing is subjected to live load
forces in the finished structure, fatigue aspects of the detailing should be considered. Also, when
top flange lateral bracing is connected directly to the top flange, the deck formwork needs to be
detailed to avoid interference with the bracing members. A top lateral bracing system also
interferes less with future inspection access using traditional underbridge inspection equipment.

Bottom lateral bracing can provide a similar function as top lateral bracing, but the lateral trusses
can experience large forces induced by vertical bending of the I-girders, similar to those reported
in tub girders by Fan and Helwig [7]. These live load forces that result from the vertical bending
need to be considered by the designer. In I-girder bridges, bottom lateral bracing creates a pseudo-
closed section formed by the I-girders connected with bracing and the concrete deck. In curved
bridges where torsion is always present, the lateral truss will contribute significantly to the
torsional stiffness of the bridge system. In addition to significant bottom flange level bracing forces
caused by the torsion and pseudo-box effects, cross-frame forces will also be larger as the cross-
frames act to retain the shape of the pseudo-box section.

2.2 Cross-Frame Forces in Horizontally Curved Girders

Cross-frames are primary members in horizontally curved I-girder bridges. The cross-frame forces
from bending and torsion in all phases of construction and loading can be determined directly from
a first-order structural analysis of the bridge system. (Here, first-order structural analysis refers to
a typical design analysis for static and transient loads.) Guidelines for proper modeling of the
cross-frames in 3D-FEA or grid analyses are described in NSBA’s Steel Bridge Design Handbook:
Structural Analysis [8] and in the AASHTO/NSBA [9]. The availability of computer programs to
determine the forces during erection and staged deck pours is improving. Stability brace forces can
be determined using equations and methods discussed in Section 2.3. Alternatively, brace forces
can be determined by a large displacement analysis on straight and curved girders, provided the
effects of imperfections are considered. The necessity of including the imperfections in the
analysis is generally dependent on the degree of horizontal curvature. For straight and mildly-
curved girders (radius approximately greater than 1200 feet), stability forces will typically
dominate. Initial imperfections are not important for girders with significant curvature.

Cross-frame forces can be determined directly from 3D analysis methods, and somewhat directly
from 2D analysis methods. Prior to the widespread use of refined analysis methods for curved-
girder bridges, cross-frame forces in curved I-girder bridges could be determined by the
approximate V-load method mentioned in Ref. [8]. The development of the V-load method is
documented by Poellot [10]. In the V-load method, the lateral flange bending curvature effects

11
are approximated by applying equal and opposite lateral loads, q = M/Rho, to each flange of an
equivalent length straight girder, where R is the radius of curvature of the bridge and M is the
girder moment due to gravity loading. Vertical V-loads that are equivalent to the overturning
torsional moment in the curved system are also applied at the cross-frames. The cross-frames
provide lateral support to each flange with equal and opposite reactive forces qLb,, where Lb is the
spacing of the cross-frames. The distribution of the cross-frame end moments and shears across a
transverse section of the bridge is determined by equilibrium considerations only. The distribution
of cross-frame end moments and shears have been conveniently summarized by Liu and Magliola
[11] for systems ranging from 2-girder to 8-girder bridges. Design examples applying the V-load
method are available [10, 12]. The V-load method cannot be used if there is a flange lateral truss
system present, as the lateral truss resists the lateral flange bending effects and thus the V-load
method will yield inaccurate results. [13]. Other potential cross-frame forces from stability
requirements, wind, overhang brackets and other lateral loads should be added algebraically to the
cross-frame forces determined via the various analysis methods.

In relatively flexible systems, second order effects can be significant. For example, curved I-
girders with flange to depth ratios (bf/D) near the AASHTO LRFD BDS limit of 1/6 [1] specified
in Article [Link] will be relatively flexible and are likely to experience significant second order
effects, particularly during erection when the full bracing is not yet installed. Research on curved
girders by Stith et al. [14] found that proportioning girders with bf/D ratios of approximately 1/4
or greater significantly reduces second order deformations during steel erection. In these cases,
cross-frame forces due to horizontal curvature often can be predicted from a first-order structural
analysis of the curved system with sufficient accuracy. These forces can be determined directly
using commercial structural analysis programs or by using the approximate V-load method.

2.3 Stability Bracing of I-Girders

For bridges with straight I-shaped girders, some type of bracing system will most likely be
necessary to control lateral buckling of the compression flange(s) during construction. Note that
the inflection point should not be considered to act as a brace point, in accordance with Article
C6.[Link] of the AASHTO LRFD BDS [1].

This section provides a discussion of design recommendations for torsional and lateral bracing,
related to the required stiffness of the various bracing components. While engineers historically
have not typically performed these calculations, they are provided so that engineers can verify that
the stiffness provided by the cross-frame and connection details is sufficient. The details and
equations provided in this section and the related subsections can be used to determine the stability
bracing forces. Using these equations, the stability bracing forces are additive to the bracing forces
resulting from a first-order type of analysis (dead load, live load, etc.). For more complex bridges,
or as an alternative to using the equations discussed in Section 2.3, a large displacement analysis
can be used to determine the bracing forces.

Guidance on the practical implementation of stability bracing forces in bridge design is provided
in the Pennsylvania Department of Transportation’s Bridge Design Standard Drawing BD-619M
[15]. BD-619M provides design guidance on combining the stability bracing forces with first-order
analysis bracing forces in the context of the AASHTO LRFD BDS limit state load combinations.
More recent guidance is provided in NCHRP Report 962 [16].

12
For continuous straight spans, bottom flange compression will generally require cross-frames or
diaphragms to provide discrete bracing of the otherwise unbraced compression flange. In
composite construction, the hardened deck prevents the top flange from twisting and lateral
movement and also provides a bracing effect to the bottom flange if web distortion (Figure 5) is
considered. For rolled sections, web distortion is not a significant issue because the web
slenderness ratio is low (d/tw < 60).

For plate girders with stiffeners designed to control web distortion that are in contact with or
welded to the top flange, no live load lateral buckling can theoretically occur in the negative
moment region. For unstiffened plate girders in continuous composite construction, web
distortional buckling as shown in Figure 5 occurs along the bottom flange, not lateral buckling.
Span length is not a significant variable for distortional buckling, so the classical lateral buckling
moment given by Equation 10 does not apply. The web distortional buckling moment for
unstiffened bridge girders in negative moment regions can be conservatively predicted by the
formulas given in Ref. [17]. However, the practice of considering only the transverse stiffeners
and web distortion as discussed above is currently not a common practice in bridge engineering
nor is it in conformance with the AASHTO LRFD BDS. It may only be potentially suitable for
special situations. For example, a bridge in service could be hit by an over height vehicle and the
cross-frames could be damaged. In this case, it may be necessary to remove the cross-frame and
determine the buckling strength without the cross-frame in accordance with the formulas given in
Ref. [18].

Figure 5 Web Distortion

Figure 6 Restraining Forces

13
Stability bracing of beams is provided by lateral (L), torsional (T) and warping (W) restraints as
shown in Figure 6. Any one of these restraints alone can increase stability of the compression
flange. Relative, nodal and lean-on systems as described earlier provide lateral and/or torsional
restraint to single girders. Warping restraint is present only in continuous bracing systems. Lateral
and warping restraints control the lateral movement of the flange to which they are attached.

Permanent metal deck forms (PMDF) that act as shear diaphragms and are attached directly to the
top flange of a girder can also improve the lateral stability. Such systems provide mainly warping
restraint to the top flange rather than lateral or torsional restraint. Stiffness and strength design
recommendations for PMDF-braced beams are given elsewhere [19]. The diaphragm strength
requirement, which is limited by the fastener capacity, generally controls the design. However,
permanent metal deck forms in steel-girder bridges are not typically attached directly to the top
flange and instead often bear on flexible clip angles welded to the edges of the girder top flanges,
which limits the lateral stability provided by the system. Therefore, AASHTO LRFD BDS Article
[Link] states that metal stay-in-place deck forms should not be assumed to provide adequate
stability to top flanges in compression prior to curing of the deck.

Top or bottom flange lateral bracing in I-girder bridges is relative or lean-on; torsional bracing is
nodal, continuous or lean-on. If two adjacent beams are interconnected by a properly designed
cross-frame or diaphragm at midspan, that point can be considered a torsionally-braced point when
evaluating the beam buckling strength. Since the beams can move laterally at midspan, the
effectiveness of such a torsional bracing system is sometimes questioned. As long as the two
flanges move laterally the same amount, there will be no twist. If twist is prevented, the beam can
be treated as braced at that point. Tests and theory confirm this approach [20, 21].

A general discussion of beam lateral and torsional bracing and the development of the design
recommendations with design examples for bridge girders are presented elsewhere [18]. The
design recommendations for torsional and lateral bracing given in Sections 2.3.1 and 2.3.5 have
been adopted by the AISC Specification. The Commentary on the AISC Specification should be
consulted for discussion on implementing the stability bracing requirements. The provisions are
limited to doubly- and singly-symmetric members loaded in the plane of the web. Beam loads are
assumed to be applied at the top flange, which is typical in bridges. Stability braces must have
sufficient stiffness and strength to be effective.

2.3.1 Torsional Bracing Design Requirements, T

The most common form of bracing in steel bridge systems are cross-frames or diaphragms that
restrain the twist of the girders and are thereby typically classified as torsional braces. Concrete
bridge decks in composite systems also provide torsional resistance to the girders. As noted earlier,
effective stability bracing must possess sufficient stiffness and strength. The required flexural
stiffness and strength requirements for torsional bracing recommended herein are given by Eqs.
(A-6-11a) and (C-A-6-11), respectively, taken from Appendix 6 of the AISC Specification [22] as
follows:

14
 βT L 2.4 LM r2
Stiffness: βT = = (13)
n  nEI yeff Cb2

0.005Lb LM r2
Strength: M br = Fbr hb = (14)
nEI yeff Cb2 ho

where  = 0.75, Mr is the largest factored major-axis bending moment for the limit-state load
combination under consideration, Iyeff = Iyc + (t/c)Iyt, t and c are as defined in Figure 7, L is the
span length, Lb is the unbraced length, n is the number of span braces, ho is the distance between
flange centroids, and Cb is the moment gradient modifier for the full bracing condition. For a
singly-symmetric section, Iyc and Iyt are the out-of-plane moments of inertia of the compression
and tension flanges, respectively. If the cross section is doubly symmetric, Iyeff becomes Iy. All
torsional bracing (nodal and continuous) uses the same basic design formulas. T and  T are
defined as the torsional stiffnesses of the nodal and continuous bracing systems, respectively (for
continuous bracing, L/n in Equation 13 is to be taken equal to 1.0). Mbr is the required strength of
the nodal torsional brace (for continuous bracing, Lb in Equation 14 is to be taken as the maximum
unbraced length permitted for the beam to resist Mr). For cross-frames, the moment is converted
to chord forces, Fbr, by dividing Mbr by hb, the distance between the chords. When the values of
the variables in the two unbraced segments adjacent to a nodal brace are different, the brace can
be designed for the average of values of the strength and stiffness determined for both segments.
It is conservative to use Cb = 1.0.

Equation 13 and a simplified form of Equation 14 appeared in the 2010 AISC Specification. The
2016 AISC Specification introduced an even simpler strength equation to replace Equation 14 (as
it originally appeared in the 2010 specification), but its applicability for use in the design of steel
bridges has not yet been conclusively demonstrated. Therefore, the original form of the strength
equation without any simplification, as given above by Equation 14 (and as presently given by Eq.
C-A-6-11 in the Commentary to the 2016 AISC Specification), is recommended herein for use in
bridge design until further studies are completed. This original form of the strength equation gives
less conservative results for singly symmetric sections than the simplified form of the strength
equation that appeared in the 2010 AISC Specification. Further, research conducted under NCHRP
Project 12-113 recommends an increase in the coefficients in Equations 13 and 14 to 0.36 and
0.008, respectively, to provide three times the ideal stiffness to limit the out-of-plane deformations
to a value equal to the initial imperfection as the critical buckling load is approached [16]. The
ideal stiffness is the brace stiffness required for a perfectly straight element to reach a specified
buckling capacity between the brace points. Three times the ideal stiffness is felt to be more
appropriate for flexural applications; the AISC equations are based on twice the ideal stiffness. As
of this writing, Equations 13 and 14 (with the modified coefficients) are being considered for
possible adoption into the next edition of the AASHTO LRFD BDS.

15
 Figure 7 Bending Stresses in Singly Symmetric Section

In the development of the design recommendations outlined in this section, Yura et al. [21]
extended the work of Taylor and Ojalvo [23] and showed that a torsional brace is equally effective
if it is attached near the tension flange or the compression flange. A moment diagram with
compression in both flanges (reverse curvature) does not significantly alter the torsional brace
requirements. On the other hand, the stiffness of a torsional brace system T is greatly affected by
web cross-section distortion at the brace point, as illustrated in Figure 5, and by the in-plane
stiffness of the girders and is given by:

1 1 1 1
= + +
T b  sec g
(15)

where b is the attached brace stiffness, sec is the distortional web stiffness and g is the in-plane
girder system stiffness (see Section 2.3.4). Because all of the stiffness terms are in series, the
effective T is always less than the smallest of b, sec and g. Brace member sizes that satisfy the
torsional brace stiffness and strength criteria are usually small, but connection details must be
carefully considered to control distortion.

2.3.2 Stiffness of Cross-Frame and Diaphragm Systems b

The b of some common torsional brace systems are given in Figure 8 and Figure 9. The choice
between the two diaphragm cases shown in Figure 8 depends on the deck details. If the distance
between the flanges of adjacent girders is maintained constant by the attachment of decking in
addition to the diaphragm, then all the girders must sway in the same direction and the diaphragm
stiffness is 6EIb/S. On the other hand, if adjacent flanges can separate as shown for the through
girders, then the diaphragm stiffness will be 2EIb/S. For regions of the girders with the top flange in
compression, placing a diaphragm above mid-height will typically cause the two compression
flanges to displace laterally in the same direction bending the diaphragm in reverse curvature and
resulting in stiffness of 6EIb/S. Values of the torsional bracing stiffness shown in Figure 9 assume
that the connection between the girder and the brace can support a bracing moment Mbr. Elastic truss
analyses were used to derive the stiffness of the cross-frame systems shown in Figure 9. If the
diagonals of an X-system are designed for tension only, then horizontal members are required in the

16
system. Although the top chord of the K-brace system has zero force, a top strut is still recommended
to link the top girder flanges together (for the development of the stiffness 6EIb/S).

Figure 8 Diaphragm Stiffness, b

Figure 9 Stiffness Formulas for Twin Girder Cross-Frames [21]

17
Designs that utilize X-framing, such as the top two systems depicted in Figure 9Error! Reference
source not found., may be treated as tension-only systems in which only one of the diagonals is
relied upon for resisting the forces. The diagonal in compression is conservatively neglected.
Structural analysis models should account for the tension-only concept when utilized. There have
been cases where engineers conducting peer reviews may have been unaware of the tension-only
philosophy used in the original design. As a result, the peer reviewer has requested costly retrofits
because they claimed the designs were inadequate because of an apparent compressive force in
one of the diagonals that exceeded the buckling capacity of that member. If one of the diagonals
can support the entire load in tension, then the design is satisfactory.

In X-frame systems that rely on the compression strength of the diagonals and are connected at the
intersection point, the out-of-plane and in-plane compressive strength can potentially be
determined using an effective length of one-half the total diagonal length, as discussed further in
Article C6.9.4.4 of the AASHTO LRFD BDS [1]. It has also been shown [24, 25] both theoretically
and experimentally for T > 0.6C and no compression member splice that the effective length is
0.5L. These studies neglect catenary action of the tension member, which is also effective in
helping to provide the necessary bracing force at the intersection. However, Article C6.9.4.4
currently takes a more conservative approach and recommends that the full length of the diagonal
be used for the effective length until further validation studies are completed.

2.3.3 Web Distortional Stiffness, sec

In Figure 5, the top flange is prevented from twisting by the bridge deck, but web distortion may
permit a relative displacement between the two flanges. A stiffener (i.e., connection plate) at the
brace location as shown in Figure 10 can be used to control the distortion. Note that in most bridge
applications, the connection plate is to be full depth and positively attached to both flanges to
account for potential fatigue issues related to distortion-induced fatigue. .

Cross-sectional distortion significantly impacts the behavior of beams with relatively shallow
braces with respect to the beam or girder depth. The portion of the web that impacts distortion is
the region above and below the brace. Therefore, web distortion essentially cannot occur for cross-
frames or diaphragms that are nearly the full depth of the web. For braces that are less than the full
web depth, distortion is controlled with the connection plates. The impact of cross-sectional
distortion stiffness is accounted for with the term, βsec, in the calculation of the actual overall
bracing system stiffness. The sec effect on the torsional brace system stiffness, related to the out-
of-plane bending stiffness of the web plus any web stiffening, is given by the following expression:

18
 Figure 10 Web Stiffener Geometry

E  (1.5D ) tw3 ts bs3 

sec = 3.3  +  (16)
D  12 12 

where tw is the thickness of web, ts is the thickness of the connection plate, D is the web depth,
and bs is the total width of the connection plate(s) on either side of the web as shown in Figure
10. The two terms within the parenthesis are the moments of inertia of the web, with an effective
width of 1.5D, and the connection plate(s). For continuous bracing use a unit width instead of
(1.5D) in Eq. (16) and the torsional brace stiffness per unit length ( β b ) in place of b in Eq. (15)
to determine the required continuous brace system stiffness, βT . Equation (16) is similar to the
expression for the bending stiffness of a member with the far end pinned, 3EI/L.

For rolled sections, Figure 11 shows some of the geometrical decisions on the layout that need to be
made in detailing diaphragms and cross-frames.

Figure 11 Diaphragm and Cross-Frame Geometry for Rolled Sections

19
The detailing can significantly affect the stiffness of the bracing system. The portion of the web
along the depth of the brace (i.e., within the brace depth, hb) will not affect the stiffness of the
brace since that portion of the web cannot distort. AASHTO LRFD BDS Article [Link] requires a
diaphragm or cross-frame in an I-girder bridge to extend at least ½ of the beam depth for rolled
beams and ¾ of the girder depth for plate girders; however, with adequate web stiffening effective
bracing can be achieved with smaller depth braces. For example, in through-girder systems the
floor beams are relatively shallow compared to the girders. With proper web stiffening, the floor
beams provide good torsional restraint to the girders

Diaphragms are usually W-shapes or channel sections connected to the beam web through
connection plates or end angles. Typically, a full-depth connection plate positively attached to both
beam flanges is used. However, AASHTO LRFD BDS Article [Link].1 does permit less than full-
depth end angles or connection plates to be bolted or welded to the beam web to connect
intermediate diaphragms in straight rolled-beam bridges with limited skew under certain
conditions. This provision reflects the fact that less than full-depth end angles or connection plates
have been bolted or welded to the webs of rolled beams to connect intermediate diaphragms for a
number of years without noted issues. Rolled beams typically have thicker webs resulting in larger
resistance to out-of-plane distortion and larger lateral-torsional buckling resistance. The end angles
or plates must be at least two-thirds the depth of the web to provide some additional torsional
resistance to the beam. A minimum gap of 3.0 inches is to be provided between the top and bottom
of the end-angle or plate welds, or top and bottom bolt holes (as applicable), and each flange to
preclude potential problems with distortion-induced fatigue.

For cross-frames or diaphragms whose depth is at least 0.8 times the beam or girder depth, attached
to full-depth connection plates positively attached to both flanges as specified in AASHTO LRFD
BDS Article [Link].1, the term, βsec, is sufficiently large such that web distortion is generally not
an issue [19]. In such cases, βsec is taken equal to infinity. Otherwise, as illustrated in Figure 11,
the web is broken up into segments that depend on the stiffening conditions. For example,
considering the section depicted in Figure 11, the web can be divided into an unstiffened
compression region (c), an unstiffened tension region (t), a stiffened region above the brace (s),
and a stiffened region below the brace (u). The stiffness values of the various portions of the web,
hi =hc, hs. ht and hu, are evaluated separately by:

 (1.5hi ) tw3 ts bs3 

2
3.3E D
 c ,  s , t , u =    +  (17)
hi  hi   12 12 

where ts is the thickness of the connection plate or projecting leg of the end angle, bs is the total
combined width of the connection plates or projecting leg sizes of the end angles on each side of
the web, and 1/ sec = (1/ i). For continuous bracing, replace 1.5hi with a unit width and
neglect the ts term if there is no connection plate at that location. The portion of the web within hb
can be considered infinitely stiff. For less than full-depth connection plates or end angles used to
connect diaphragms as shown in the left-hand side of Figure 11, hi is taken as the distances along
the web from the top and bottom of the connection plate or end angle to the adjacent flange (i.e.,
hi = hc and ht); otherwise, if the connection plate is full-depth, hi is taken as the distances along

20
the web from the top and bottom of the diaphragm member to the adjacent flange [i.e., hi = (hs +
hc) and (hu + ht)]. For the shallow cross-frame shown in the right-hand side of Figure 11 with the
full-depth connection plate, hi is taken as hc and ht.

The diaphragm does not have to be located close to the compression flange. As noted above, the
location of the diaphragm may affect the brace stiffness (2EIb/S vs. 6EIb/S); however, for a given
brace stiffness value, the location of a diaphragm or cross-frame on the cross section is not very
important. The most effective cross-section location for diaphragm/cross-frames to minimize
distortion and stiffener sizes is centered about the beam mid-height. The minimum depth
requirements for diaphragms and cross-frames in I-girder bridges, mentioned previously, make the
placement effect somewhat less significant.

2.3.4 In-Plane Stiffness of Girders, g

In cross-frames and diaphragms the brace moments Mbr are reacted by vertical forces on the main
girders as shown in Figure 12. The vertical couple causes a differential displacement in adjacent
girders that reduces the torsional stiffness of the cross-frame system. These forces increase some
main girder moments and decrease others and cause a relative vertical displacement between
adjacent girders. The effect is greater for the two linked “twin girder” systems shown in Figure 12B
compared to the fully interconnected system depicted in Figure 12A. For a brace only at midspan
in a multi-girder system, the contribution of the in-plane girder flexibility to the brace system
stiffness is [21]:

24(n g − 1) 2 S 2 EI x
g = (18)
ng L3

where Ix is the strong axis moment of inertia of one girder, ng is the number of girders connected by
the cross-frames, and L is the span length. As the number of girders increase, the effect of girder
stiffness will be less significant. For example, in a two-girder system the term 24(ng-1)2/ng is 12
while for a six-girder system the factor becomes 100. Helwig et al. [26] showed that for twin girders
the strong axis stiffness factor g is significant and Eq. (18) can be used even when there is more
than one brace along the span. If g dominates the torsional brace stiffness in Eq. (15), then a system
mode of buckling that is discussed later in this section is possible.

21
 M br Beam Load

M br 2 Mbr
2 Mbr Brace Load
S S
S

4 Mbr
3S

2 Mbr
S

Figure 12 Beam Load from Braces

2.3.5 Connection Stiffness, conn

The diaphragm and cross-frame stiffnesses given in Figure 8 and Figure 9 assume that the
attachment connections are not flexible. Clip angles welded only along the toe and tee stubs with
bolted flanges will flex when tension is applied to the outstanding leg or tee stem. This flexibility,
conn, will reduce the system stiffness. If partially restrained connections are used, the flexibility of
the two connections should also be included in determining the system stiffness by adding the term,
2/conn, to the right side of Eq. (15). Field studies [27] have reported a reduction of 40-70% in the
stiffness of the non-permanent external cross-frames between tub girders due to tee stub flange
flexibility. Bent gusset plates or connection plates used as an option to connect skewed cross-frames
to girders when the skew angle exceeds 20 degrees from normal can also reduce the system stiffness.

The brace force design requirements are directly proportional to the magnitude of the initial out-of-
straightness of the girders [28]. The brace force design requirements above are based on an out-of-
straightness of 0.002L. If oversize holes are used in the bracing details, the brace forces will be
increased if slip occurs in the connection. This can be considered in design by adjusting the
magnitude of the lateral and torsional brace force requirements by the modification factor, (1 +
oversize / (Lb/500)).

The attachment of a single-angle or a tee-section member to a connection plate whether by direct

connection or through an intermediate gusset plate introduces an eccentricity into the member and
the connection simultaneously, which reduces the stiffness of the member. AASHTO LRFD BDS
Article [Link].4 states that the influence of the end-connection eccentricities is to be considered
in the calculation of the equivalent axial stiffness of single-angle and flange-connected tee-section
cross-frame members. Article C4.[Link] further states that in lieu of a more accurate analysis, the
equivalent axial stiffness, (AE)eq, of equal leg single angles, unequal leg single angles connected
to the long leg, and flange-connected tee-section members may be taken as 0.65AE. Engineers

22
should consider this guidance in the evaluation of the cross-frame stiffness, βb, and in the
development of finite-element models that are often used to predict forces in curved and skewed
bridges. An outcome of many such analyses is the prediction of significant cross-frame forces.
These forces typically result due to the size of the member input into the model and the fact that
traditionally engineers have modeled these members as truss-type elements using their full AE/L
stiffness contribution. Using the modified stiffness generally results in lower forces being attracted
to the cross-frame members, lesser demands, and thus potentially smaller member sizes and
connections.

2.4 Effects of Support Skew

Due to geometric requirements with either intersecting roadways or the terrain of the job site, the
support lines of bridge systems often must be offset as depicted in the plan view in Figure 13.

Figure 13 Plan View of Bridge with Skewed Supports

Since skew angles increase the interaction between the steel girders and the braces, the behavior
of bridges with skewed supports becomes more complicated than that in bridges with normal
supports. The interaction between the girders and braces often results in large live load forces in
the cross-frames or diaphragms, which can lead to fatigue problems around the brace locations.
The severity of the fatigue problem is dependent on the details that are used for the bracing. Figure
14 illustrates the two different orientations of braces for skewed bridges.

23
 Figure 14 Brace Orientations for Bridges with Skewed Supports

If the skew angle is less than 20 degrees, the AASHTO LRFD BDS [1] allows the bracing to be
parallel to the skew angle. For skew angles greater than 20 degrees, the AASHTO LRFD BDS
requires the bracing to be perpendicular to the longitudinal axis of the girder. For braces parallel
to the supporting abutments, points A and B at the ends of the brace will have similar vertical
displacements during truck live load. However, when braces are normal to the girder lines, the
two ends of the braces will have different vertical displacements during truck loading. This
differential vertical displacement can result in large brace forces, which can lead to fatigue
problems. Alternative bracing layouts to help minimize live load induced forces are to use either
lean-on bracing or a staggered cross-frame layout as discussed in the next section. Alternatively,
AASHTO LRFD BDS Article C6.7.4.2 discusses omitting individual cross-frames in perpendicular
contiguous cross-frame lines, particularly near the obtuse corners of skewed supports, to reduce
forces. Additionally, for perpendicular frames, the commentary discusses offsetting the first
intermediate cross-frame line from the skewed support and provides recommend distances. This
offset can be seen in Figure 14b.

When the cross-frames are oriented perpendicular to the longitudinal axis of the girders as shown
in Figure 12(b), the provisions outlined in the previous sections for the stability stiffness and
strength requirements are directly applicable with no correction required for the skew angle. The
braces will develop additional forces due to differential displacement from the skew angle. For
skew angles larger than approximately 45 degrees, the forces induced due to the differential
displacement will be of similar magnitude or even larger than stability induced forces. During
construction of the concrete bridge deck, these forces can be predicted with reasonable accuracy
from a first-order analysis on a relatively simple computer model of the steel girders and bracing
system. The forces from such an analysis are additive to the stability forces predicted from Eq.
(14) and Figure 9.

24
When the cross-frames are oriented parallel to the skew angle as depicted in Figure 12(a), the skew
angle has an impact on both the stability stiffness and strength requirements of the bracing. Wang
and Helwig [29] present expressions for the stiffness and strength requirements of braces in bridges
with skewed supports. The required stiffness of the braces is given in the following expression:

b
 bSkew = (19)
cos 2 

where, b Skew is the stiffness requirement of the skewed brace, b is the required stiffness that
results from Eq. 13, and  is the skew angle. Once the required skewed brace stiffness is
determined, the stiffness equations given in Figure 9 can be used to size the diagonals and struts
of the cross-frame. Although “s” in the stiffness equations is typically thought of as the girder
spacing, for a skewed brace the value of “s” should be taken equal to the length of the cross-frame
in the skewed orientation (equal to the girder spacing divided by cos).

The strength requirement of the skewed brace is given in the following expression:

M br
M br Skew = (20)
cos

where Mbr Skew is the brace moment applied to the skewed brace and Mbr is the required brace
moment from Eq. (14).

If the cross-frame is properly sized for the stiffness and strength requirements considering the
skewed geometry, there is no technical reason why a cross-frame cannot be oriented parallel to the
skew for angles larger than 20 degrees. The engineer needs to recognize that the stiffness of the
cross-frame is affected by the skew as a function of the above equations and also that the parallel
brace can becomes relatively long for larger skew angles. If the longer geometry is considered in
the equations the brace will have the proper stiffness and strength.

For larger skew angles, the intermediate cross-frame lines will typically be oriented perpendicular
to the longitudinal axis of the girders, but at the supports the cross-frames are usually parallel to
the skew angle. One detail that can reduce the provided stiffness occurs when cross-frames parallel
to the skew angle use a bent gusset plate at the end of the cross-frame member. Many designers
depict a bent plate to make the connection between the brace and the connection plate (web
stiffener), as illustrated in Figure 15 for a bridge with support skew of nearly 60 degrees. Such a
detail allows the fabricator to utilize a connection plate that is perpendicular the web plate;
however, the bent plate connection can dramatically reduce the effectiveness of the brace due to
the flexibility introduced by the eccentric connection. One solution to eliminating the bent plate is
orienting the connection plate parallel to the skew angle; however, such a detail can be complicated
for larger skew angles as access to the weld in the acute angle becomes problematic. In addition,
fatigue tests and analytical investigations on stiffeners orientated parallel to the skew and welded
to the flanges showed a lower fatigue life compared to perpendicular stiffeners [30, 31]. Another
option is to use a full depth bent connection plate on the girder, where the bend line is located at
the outer edge of the flange, reducing or eliminating any welds in the longitudinal direction. This
option is more appropriate when using a full-depth diaphragm.

25
Figure 15 Bent Plate Connection Detail Frequently Used in Bridges with Skewed Supports

Although currently not a standard practice, an alternative to the bent plate detail in skewed bridges,
Quadrato et al. [30] proposed the detail depicted in Figure 16 which shows a plan view of an I-
shaped girder with a half-pipe (or half-round) bearing stiffener. The round pipe allows a
perpendicular connection between the skewed support cross-frame and stiffener for any skew
angle. The split pipe stiffener serves as both the bearing stiffener and connection plate. The pipe
stiffener increases the warping resistance of the girder and thus improves the buckling resistance
of the girder. Further information regarding the increase in warping resistance can be found in
Quadrato et al. [30]. Additionally, with regard to fatigue behavior, analytical and experimental
research has shown that the pipe stiffener weld detail to the girder flanges is no worse than a typical
plate stiffener welded to the girder flanges [32]. However, the Engineer needs to be aware that
currently, the pipe material may not be approved for use in bridges and may require special
approval from the Owner.

Half-Pipe
Stiffener

Cross Frame
Figure 16 Half-Pipe Web Stiffener

26
2.5 Lean-On and Staggered Bracing

2.5.1 Lean-On Bracing

A common practice in the design of frames in building applications is to provide lateral stability
by using lightly loaded columns to restrain other columns such as the case depicted in Figure 17.

Figure 17 Summation of P Concept for Sideway Frame Stability

Column A has pins at both ends and therefore has no lateral stiffness. However, the column can
be laterally stabilized by leaning on Column B, provided that column is designed to possess
adequate lateral stiffness to support the total frame load. The figure demonstrates the P concept
that was presented by Yura [33] in which the frame is laterally stable in the sidesway buckling
mode provided the sum of the applied load is less than the sum of the sway mode contribution of
the columns in the plane of the frame. In the case depicted in Figure 17, Column A contributes no
lateral stiffness to the frame and therefore Column B must be able to support the entire frame load.
However, leaning columns such as Column A, must be able to support their axial force in the no-
sway mode.

The P concept also applies to beam systems such as the two beams depicted in Figure 18.
PA PB

PA> PB

A B

Figure 18 Beams Linked Together at Compression Flange

The beams are simply supported with gravity load that causes compression in the top beam flange.
The two beams are connected at the top flange through a “truss bar” that does not restrain the
flange rotation, but instead causes the two flanges to have essentially the same lateral displacement.

27
The respective loads on the two beams are PA and PB, in which the load on beam B is less than the
members buckling load. Beam A can therefore rely, or “lean on” Beam B for stability and the P
concept, would simply require that the sum of the two applied loads are less than the sum of the
two beams buckling loads. The spacing between the links must be close enough that beam A cannot
buckle between the links.

The P concept for beams is demonstrated numerically from the graph shown in Figure 19 [21].

3
Buckling Load
TOTAL LOAD P (kips)

2 P
0.5P 0.5P

1
W12x14 Span =24 ft.
Initial Deflection = 0.16 in.
0
0 1 2 3 4 5
LATERAL DEFLECTION (in.)
Figure 19 Graph of P Concept for Beams

The graph shows results from a three-dimensional finite element analysis for two beams linked
together at the top flange. The buckling load is indicated by the horizontal line in the graph that
that was determined from a critical load analysis (eigenvalue buckling analysis). The critical load
does not reflect the impact of imperfections on the behavior. The two solid lines represent the
results from a large displacement analysis on an imperfect system. The curves approach the critical
load results (buckling load) at relatively large displacements. In one of the large displacement
graph cases the two beams are equally loaded with 0.5P, while in the other case only one of the
beams was loaded with a load of P. The graphs show that the total load both beam systems can
support is approximately 2.5 kips despite extremely different load distributions.

The above cases demonstrate the bracing scenario where lightly loaded beams can provide bracing
to other, more heavily loaded beams. The lean-on concepts also apply to cases where beams can
lean-on braces such as cross-frames or diaphragms. Cross-frames and diaphragms represent
relatively costly structural components in steel bridges from the perspective of both fabrication
and erection. The braces can often be difficult to install in the bridge due to fit-up problems and
also may attract significant live load forces, particularly in bridges with large support skews.
Therefore, minimizing the number of cross-frames on the bridge can lead to better overall bridge
behavior as well as reduced maintenance costs. The typical practice in steel-bridge behavior is to
place cross-frames between each of the girders at a uniform spacing along the length of the girders.
Although this practice results in effective braces for providing overall stability to the bridge

28
girders, the resulting system is not necessary structurally efficient. Cross-frames and diaphragms
fit into the category of torsional braces since they resist twist of the girders. Improved structural
efficiency is possible by utilizing lean-on bracing concepts in which several girders can be braced
across the width of the bridge by a single cross-frame. Lean-on bracing systems allow the designer
to eliminate cross-frames on parts of the bridge where the brace is difficult to install or where large
forces in the finished bridge may result from truck traffic, thereby potentially leading to poor long-
term fatigue behavior. In a given bracing line, a cross-frame may be selectively positioned, and 3
or more girders can lean-on that brace as depicted in Figure 20. Girders that lean on the brace
require top and bottom struts to control girder twist.

Full Cross-Frame

Only Top and Bottom Struts

Figure 20 Lean-on Cross-Frame Bracing

Helwig and Wang [34] developed recommendations for straight girders with skewed supports so
that intermediate cross-frames (between the supports) can be selectively located to minimize forces
induced in the cross-frames. For example, the plan view of the bridge in Figure 21 shows a possible
layout that will reduce the number of cross-frames and minimize the live load induced forces.
124 ft.

24 ft. 24 ft. 24 ft. 24 ft. 24 ft. 4'

A B C D E

33.7

Figure 21 Plan View of Bridge with Lean-On Cross-Frame Bracing

29
In a given cross-frame line, the full cross-frame is located between the girders that will place the
cross-frame as far away from the support as possible. In cross-frame lines A and B, this puts the
cross-frame near the top of the figure, while in lines D and E, the braces are near the bottom of the
plan view. Near midpsan, at least one full line of cross-frames should be provided to link the
girders together and control differential displacement. In addition, the cross-frame lines near the
supports (lines A and E) should not frame directly into the support, but instead be offset by
approximately 4 or 5 feet. Offsetting the bracing line from the skewed support reduces the forces
induced in the cross-frame, while still producing effective bracing.

In systems with a large number of girders across the width, a contiguous line of cross-frames near
midspan may not be necessary. Instead, the cross-frames can be distributed across the width of the
bridge as shown in Figure 22, which shows a plan view of a straight two-span bridge with a 54-
degree skew. The bridge was one of three straight bridges in Lubbock, Texas that was constructed
using lean-on bracing. The circled cross-frames were cross-frames that were necessary to provide
stability to the partially erected bridge.

ground F.S. air F.S.

B A
G9
G8
G7
G6
54 G5
G4
G3
G2
G1
B A
bent bent
Span 1 Span 2
(180 ft.) (166 ft.)

Section A-A Section B-B

Figure 22 Lean-on Bracing Layout in Bridge with Large Numbers of Girders

Because there are several girders restrained by a single cross-frame in lean-on systems, the
individual cross-frames need to be sized for the increased demand on the bracing. Figure 23 [34]
demonstrates the distribution of forces across the bridge in a cross-frame system with lean-on
bracing.

30
 F F F F
-F 2F F

A B C D

-3F -2F -F

Lean-On Cross Frame Stiffness and Strength Requirements

Define ngc as the number of girders per cross-frame.

Brace Strength:
Fd
ngc FLd Fs hb
Fd =
S
M br
F=
hb
Fs = (ngc − 1) F s

Brace Stiffness:

ES 2 hb2
b =
ngc L3d S 3
+ (ngc − 1) 2
Ad Ac

Figure 23 Stiffness and Strength Requirements for Lean-On Cross-Frames

The cross-frame is idealized as a tension-only system. The expressions in the figure show the
corresponding stiffness of the cross-frame as well as the maximum forces in the struts and
diagonals of the cross-frame. In most situations, although the forces vary, the same size struts will
be used throughout.

One other modification that is necessary with lean-on bracing is a reduction in the in-plane stiffness
of girders. Since the bridge is not fully connected throughout, Helwig and Wang [34]
recommended that the in-plane stiffness of the girders be reduced by 50% in the brace system
stiffness calculation. This leads to the following expression for systems with lean-on bracing:

12(n g − 1) 2 S 2 EI x
 g Lean−on = (21)
ng L3

Note that ng in the above cross section is the total number of girders across the width of the bridge
and should not be confused with ngc in the lean-on calculations.

31
2.5.2 Staggered Bracing

Another cross-frame configuration that is sometimes utilized in straight bridges with large support
skews is the staggered layout depicted in Figure 24.

Figure 24 Staggered Cross-Frame Layout

In the staggered layout, the cross-frames are oriented perpendicular to the longitudinal axis of the
girders; however, the individual cross-frames are staggered along a line parallel to the skew angle.
The advantage of this layout is that the differential deflections at the ends of the cross- frame lines
are relatively similar since the centers of the individual cross-frames are located at approximately
the same longitudinal location on the bridge. Therefore, the behavior is similar to the parallel
layout; however perpendicular connections can be used. The basic stiffness and strength
expressions for the bracing are essentially the same as outlined for the cross-frames along
contiguous lines.

Additionally, a staggered cross-frame arrangement as shown in Figure 24 will result in flange

lateral bending, as the cross-frame members apply lateral loads to the girder flanges. In cases of
smaller bridges or bridges with small skews, the lateral flange bending effects will be less than
those that result from longer span bridges, or bridges with larger skews. The differential deflections
between the adjacent girders where the cross-frames connect play a significant role in the cross-
frame forces and the subsequent flange lateral bending effects.

2.6 System Buckling of Interconnected Girders

Cross-frame and diaphragm systems provide bracing by restraining twist of the interconnected
girders. The generally accepted belief among designers is that reducing the spacing between these
braces will improve the buckling capacity of the girder systems. This belief holds true for many
applications. However, there have been a number of applications in which the buckling behavior
of the girder system can be relatively insensitive to the spacing or size of the braces. For example,
the two-girder widening shown in Figure 25 had relatively close cross-frames, however the girder
experienced significant twisting during placement of the concrete deck as evidenced by the 10-
inch lateral deformation of the bottom flange relative to the plumb line. The load on the twin girder
system was balanced and did not have an eccentricity. The mode exhibited by the bridge widening
is a buckling failure of the entire girder system as described below.

32
 Figure 25 System Buckling of a Twin Girder Widening, where the system has buckled out
of plane nearly 10 inches during deck placement

The buckling mode that is typically envisioned in a properly braced girder system is depicted in
Figure 26a, which shows a plan view of a twin girder system. By reducing the spacing between
the braces, the engineer can reduce the size of Lb and thereby improve the buckling capacity of the
girders that results from lateral-torsional buckling expressions such as Eq. (10). However, in girder
systems with a relatively large length to width ratio (Lg/s), the controlling mode is the buckled
shape depicted in Figure 26b. In the system buckling mode, the girder system behaves as a unit
and the resulting resistance is not significantly affected by the spacing or size of the braces.

33
 S

(a) Individual girder buckling, plan view Lb

A
Lg
(b) Global system buckling , plan view

x x (c) System cross section A-A

y
S
Figure 26 Comparison of Individual Buckling Mode and System Buckling Mode

Yura et al. [35] presented the following solution for doubly-symmetric girders that can be used to
evaluate the buckling capacity of a girder in the system buckling mode:

 2 SE
M gs = IyIx (22)
L2g

where: s is the girder spacing, Lg is the total length of the girder, E is the modulus of elasticity of
the steel girder, and Iy, and Ix are the respective moments of inertia of a single girder about weak
and strong axes. The expression estimates the capacity of one of the girders for comparison with
the girder design moment.

For singly-symmetric girders, Iy in Eq. 20 can be replaced with Ieff [35]:

t
I eff = I yc + I yt (23)
c

where Iyc and Iyt are the respective moments of inertia of the compression and tension flanges about
an axis through the web, and c and t are the respective distances from the centroidal axis to the
compression and tension flanges. For a doubly-symmetric section, Ieff given by Eq. (23) reduces
to Iy since c = t.

34
Equation 20 provides a closed form solution that can be used to evaluate the system buckling
capacity of twin girder systems. For a three-girder system, replace Iyc in Eq. (23) with 3/2Iyc, and
define S in Eq. (22) as 2S, which is the distance between the two exterior girders. For four girders,
replace the corresponding values of the Iyc and S terms with 2Iyc and 3S. Equation (22) shows that
for a given girder span (Lg), the system buckling mode can be improved by either increasing the
stiffness of the individual girders or by increasing the girder spacing. Alternative methods of
improving the buckling capacity include adding a top and bottom flange lateral truss near the ends
of the girders as is discussed later in this volume.

Equation (22) is written as follows in Article [Link].2 of the AASHTO LRFD BDS [1]:

 2 wg E
M gs = Cbs I eff I x
L2
(24)

where Cbs is a system moment gradient modifier, wg is equal to the girder spacing for a two-girder
system or the distance between the two exterior girders of the unit under consideration for a three-
girder system, and L is the length of the span under consideration. Ieff is to be taken as Iy for doubly
symmetric girders and from Eq. (23) for singly symmetric girders.

Equation (24) is for application to spans of straight I-girder bridge units with three or fewer girders,
interconnected by cross-frames or diaphragms, in their noncomposite condition during the deck
placement. The equation applies to units that are not braced by other structural units and/or by
external bracing within the span, and to units that do not contain any flange level lateral bracing
or lateral bracing from a hardened concrete deck within the span. Such units are particularly
susceptible to significant global lateral-torsional amplification of the lateral and vertical
displacements of the unit during the deck-placement operation. Situations exhibiting potentially
significant global second-order amplification include phased construction involving narrow
unsupported units with only two or three girders and possibly unevenly applied deck weight. Once
a concrete deck is acting compositely with the steel girders, a given span of a bridge unit is
practically always stable as an overall system. Eq. (24) is not intended for application to I-girder
bridge spans in their composite condition. Eq. (24) is also not applicable to I-girder bridge units
with more than three girders, which are typically not susceptible to excessive global lateral–
torsional amplification during the deck placement.

Considering all the girders across the width of the unit within the span under consideration, the
sum of the largest total factored girder moments during the deck placement within the span under
consideration should not exceed 70 percent of Mgs given by Eq. (24). Limiting the sum of the
largest total factored girder moments across the width of the unit within the span under
consideration to 70 percent of the elastic global buckling resistance of the span theoretically limits
the amplification under the corresponding nominal loads to a maximum value of approximately
2.0. Should the sum of the largest total factored girder moments across the width of the unit exceed
70 percent of Mgs, the following alternatives may be considered: 1) the addition of flange level
lateral bracing adjacent to the supports of the span may be considered; 2) the unit may be revised
to increase the system stiffness; or 3) the amplified girder second-order displacements of the span
during the deck placement may be evaluated to verify that they are within tolerances permitted by

35
the Owner. Yura et al. [35] suggest adjustments to be made when estimating the elastic global
lateral–torsional buckling resistance of the system where a partial top-flange lateral bracing system
is present at the ends of the span, along with some associated bracing design recommendations.

For cases where the girders are nonprismatic and/or the girder cross sections vary across the unit,
it is recommended that length-weighted average moments of inertia within the positive-moment
sections of all the girders in the span under consideration be used for Ix, Iy, Iyc and Iyt, as applicable,
in calculating the elastic global lateral–torsional buckling resistance from Eq. (24). Also, in cases
where the girder spacing is less than the girder depth, it is recommended that the more general
elastic global lateral–torsional buckling equation provided in Yura et al. [35] be used, as Eq. (24)
becomes more conservative in this case.

The system moment gradient modifier, Cbs, in Eq. (24) accounts for the beneficial effect of the
moment gradient within the span on the elastic global lateral–torsional buckling resistance of the
span acting as a system, which is particularly significant for continuous-span units. A Cbs value of
1.1 applies to simply supported units and should also be applied if investigating continuous-span
units that are in a partially erected condition. A Cbs value of 2.0 applies to fully erected continuous-
span units.
Mgs should only be used as a general indicator of the susceptibility of horizontally curved I-girder
systems to second-order amplification under noncomposite loading conditions. Narrow
horizontally curved I-girder bridge units as defined herein in their noncomposite condition during
the deck placement may be subject to significant second-order amplification and should instead be
analyzed using a global second-order load-deflection analysis to evaluate the behavior. As an
alternative, the addition of flange level lateral bracing adjacent to the supports of the span may be
considered, or the unit can be braced to other structural units or by external bracing within the
span.

2.7 Lateral Bracing Systems

Common bracing systems that may be used in bridges consist of lateral trusses that control the
relative moment of two points along the girders. Relative bracing systems generally consist of a
combination of struts and diagonals as depicted in the plan view shown in Figure 27. As noted in
the previous section, the lateral truss-type bracing depicted in Figure 27 may be used on the bottom
flanges of some I-girder systems to enhance the resistance to lateral loads from sources such as
wind. As is covered in detail later in this volume, these types of lateral trusses are extremely
important to stiffen tub girders during construction. The struts are oriented perpendicular to the
longitudinal axis, while the diagonals connect two points at different lengths along the bridge. The
spacing between the two points defines the unbraced length. As the name implies, the bracing
system controls the relative movement of the two points. For example, the diagonal and two struts
in Figure 27 combine to form a relative bracing system.

The stiffness of the lateral bracing is a function of both the strut and the diagonal sizes, and the
bracing can be designed to control the movement of Point C relative to Points A and B. The actual
bracing system may vary in terms of the number or orientation of the diagonals. In some cases,
two diagonals may be used, and the stiffness of the system is dependent on the buckling capacity
of the diagonals. A “tension-only” system with two diagonals is sometimes specified such that the

36
compression resistance of the diagonal is conservatively neglected since (depending on the type
of member selected) the buckling resistance may be low. However, since the lateral truss often is
connected near the top or bottom flange of the girders, the designer needs to be aware of forces
that can be induced in the bracing due to stresses from vertical bending of the girders. These
stresses develop due to longitudinal strain compatibility between the bracing system and the girder
flanges that experience compression or tensile stresses due to girder flexure. The nature of the
stress induced in the lateral truss is the same state of stress in the girder where the truss is
connected. For example, if the lateral truss is located at the top flange near the midspan region,
gravity load stresses in the girders will induce compression in the truss. Generally, if a
perpendicular cut everywhere along the unbraced length passes through the brace, then the brace
system is relative.

Girder Flange
A B C
Strut

Girder Flange
Figure 27 Plan View of Typical Lateral (Relative) Bracing System

Refer to the discussion at the end of Section 2.3.2 for information that may be relevant to the design
of lateral bracing systems that utilize X-framing.

Winter [36] demonstrated that effective stability bracing must satisfy both stiffness and strength
requirements. He demonstrated the concept with a simple rigid link model that could be used to
determine the ideal stiffness requirements of the bracing as well as the impact of imperfections on
the brace strength requirements. The bracing requirements for relative bracing based on Winter’s
approach [36], are:

Stiffness: L = 4Mf Cd /Lbho (25)

Strength: Fbr = 0.008MfCd/ho (26)

where  = 0.75, Mf is the maximum moment within the unbraced length (Lb), ho is the distance
between flange centroids and the constant Cd = 1.0 for single curvature bending and 2.0 for reverse
curvature. The flange compressive force is conservatively approximated as Mf /ho.

37
These provisions are applicable for lateral bracing attached near the compression flange (except
for cantilevers where top flange bracing is more effective). Braces that are adjacent to an inflection
point must be attached to both flanges and the stiffness and strength requirements are greater as
given by the Cd factor. It should also be noted, that in most I-girder bridges, the design of the lateral
bracing members will be governed by the applied loads, and not necessarily the bracing
requirements presented above.

2.8 Continuous Bracing

In a continuous bracing system with sufficient strength and stiffness, compression flange lateral
buckling cannot occur because of the very close spacing of the connection points. A concrete
bridge deck attached to the top flange through shear studs is an example of a continuous system.
The hardened concrete bridge deck has very large in-plane shear stiffness and bending stiffness
that effectively prevents top flange twist and lateral movement of the girders when attached by
shear studs or flange embedment.

During a deck pour, the concrete has no stiffness, but the permanent metal deck forms (PMDF) do
have significant stiffness and strength. In building construction, it is standard practice to use the
PMDF as beam bracing when the ribs are perpendicular to the beam because the PMDF is attached
directly to the top flange through the field welding of the shear studs.

In bridge construction, the connection of the deck forms requires leveling angles to account for
flange transitions or differential camber. The leveling angles introduce flexible connections that
reduce the effectiveness of the forms for bracing; however, the forms still do provide some help to
the girder stability. A report [4] on a stability failure during a deck pour indicated that PMDF
increased the girder buckling capacity 50% compared to an unbraced girder, but the increase was
insufficient to support the entire deck weight. Field tests [27] on a U-shaped girder with the PMDF
attached directly to the flanges with powder-actuated fasteners showed good performance.
Improved PMDF attachment details have been successfully implemented on short span bridges
[37, 38] that eliminated all the intermediate cross-frames. Currently, AASHTO does not allow the
Engineer to consider the stability that can be provided by the PMDF’s, as discussed in Section
[Link] of the AASHTO LRFD BDS.

38
3.0 BRACING OF TUB GIRDER SYSTEMS

U-shaped steel girders (tub girders) composite with a hardened concrete deck form a closed-box,
torsionally strong bridge system. A typical cross section of a twin tub girder system is shown in
Figure 28. Single tub girder systems are sometimes used for single-lane bridges. Prior to the
development of composite action, the tub girder itself is a torsionally-weak, open steel section that
must be braced to support the erection and construction loads. The three typical types of bracing
systems are: interior diaphragms (ID), a top flange lateral truss (LT) and external intermediate
cross-frames (EC) between adjacent tub girders (see Figure 29).

Figure 28 Twin Tub Girder System

The three bracing systems used with tub girders are designed to achieve one or more of the
following objectives mainly during the construction stage:
1. Control box girder distortion (ID)
2. Control lateral buckling of the individual top flanges (ID, LT)
3. Increase the torsional stiffness and strength (LT, EC)
4. Control global lateral buckling of the tub girder (LT)
5. Support sloping webs in trapezoidal cross sections (ID, LT)
6. Control warping normal stresses (ID)
7. Maintain alignment in multi-girder systems (EC)

Top lateral bracing system Internal diaphragm

bracing system

External intermediate
cross- frame system

Figure 29 Types of Bracing Systems for Tub Girders

39
In the following sub-sections, the three types of bracing systems will be discussed, and design
methods presented for spacing and proportioning the brace members.

3.1 Top Flange Lateral Truss

Horizontally curved tub girders are subjected to significant torsional loading that open steel
sections cannot support by warping strength alone. By tying together the two top flanges of a tub
girder with diagonals and struts to form a top flange lateral truss, the cross section becomes quasi-
closed. When designing the top lateral bracing system, two major criteria must be considered,
torsional rigidity and torsional strength. Torsional rigidity is related to the torsion constant J which
is greatly enhanced by the top lateral system. A diagonal with an area of a few square inches will
increase J by more than a thousand times (Eq. 4 divided by Eq. 3) if the top flange lateral truss is
converted to an equivalent plate thickness, teq. As indicated by the expressions, the equivalent
thickness is a function of the area of the diagonals, struts, the top flange areas, and the web areas.
The top lateral members must have sufficient areas so that warping normal stresses can be
neglected and torsional deformations can be kept small. The areas of the members must also be
sufficient to resist the torsional forces imposed on the system. Vertical bending of the tub girders
during construction can also develop forces in the top lateral system.

Figure 30 Geometric Layout and Equivalent Plate Thickness of Top Lateral Systems [40]

Three common geometric arrangements for the top flange lateral truss as shown in Figure 32 are
the Warren, the Pratt and the X-type. The Pratt system is usually oriented so the diagonals are in
tension. In the Warren system the diagonal at the location of maximum torque is usually oriented
to be in tension. To maximize teq, the slope of the diagonals relative to the longitudinal axis of the
girder should be between 35 and 40 degrees [39]. The truss arrangement should have an even
number of panels within the span for improved performance especially for the Pratt and Warren
types. The panel spacing is controlled by geometry or the unbraced length of the top flange.

40
The formulations for teq given in Figure 32 were derived by Kollbrunner and Basler [40]. If teq of
all three geometric arrangements are the same, the angle of twist will be the same but the forces in
the top lateral systems will vary as discussed in the next sub-section. To achieve the same torsional
rigidity, the X-type requires the smallest total Ad within the panel and the Pratt system the largest.
If the X-system diagonals are designed for tension only, then only one diagonal should be
considered effective and teq is determined using the Pratt formula. Frequently, publications
incorrectly define At in the teq formulations as the area of the top flange, Af. When the top lateral
truss is located below the top flanges, w is redefined as the width of the truss.

Article C6.7.5.3 of the AASHTO LRFD BDS [1] suggests as a guideline that Ad ≥ 0.03w (all units
in inches) in curved tub girders, based on work done by Heins [41] for X-type systems with two
effective diagonals within the panel. Making the same Heins’ assumptions for Pratt and Warren
systems would require that Ad ≥ 0.054w. The full derivation of the requirement for X-type systems
can be found in Heins [41].

Wind and other lateral forces during the construction stage also can cause torsion in a tub girder
because the shear center of the quasi-closed section is generally located below the bottom flange.
The location of the shear center for a single tub or quasi-closed section is given in the Section 4.3.
The applied torque is the resultant lateral force times its distance to the shear center. In straight tub
girders with perpendicular supports, torsional loads do not dominate so a full length top lateral
system may not be necessary. Lack of a top lateral system, however, makes the tub girder more
susceptible to global lateral buckling as discussed later. If the supports are skewed, torsion must
be considered.

3.1.1 Top Lateral Brace Forces

The forces in the members of a top lateral truss system can come from torsion, vertical bending,
lateral bending, cross-section geometry, distortion and stability effects. Except for relatively
straight girders, torsion effects dominate the forces in the bracing members. Figure 31 shows the
forces determined from 3D-FEM analyses for a curved 180 ft simply-supported girder with three
different top lateral truss arrangements [42]. The three different layouts consist of an X-type, a
Pratt, and a Warren truss layout. The teq = 0.05 in. is the same for all three systems. The brace
member sizes within each arrangement are different in order to achieve a similar teq but are constant
along the span. Only gravity load (steel and concrete deck self-weight) during construction was
applied and there are no distortion forces, lateral forces or stability effects included in this
comparison. The torsional forces in a lateral truss system are not significantly affected by the truss
member sizes but the plots show that the forces in the three systems are different.

In the X-system two diagonals are considered effective within each panel. For clarity, the forces
in the two diagonals within each panel, designated as diagonal 1 and diagonal 2, are plotted
separately at the center of each panel along the span. For torsional loads only, the two diagonals
should have the same magnitude of force, one in tension and the other in compression. At the left
support, which is the location of maximum torque, the magnitudes of the two diagonal forces differ
slightly but the disparity increases away from the supports because of in-plane bending. Although
the tub girder alone is usually designed for the total bending gravity force during construction by

41
neglecting the top lateral truss, the top flange bending strain will induce forces in the diagonals for
strain compatibility. The bending forces are maximum at midspan and small near the simple
supports. On the other hand, torsional moments are maximum near the supports and zero at
midspan so the diagonal forces at midspan are caused mainly by bending. Both diagonals within
the panel have a 30-k compressive force at midspan. Note also that the maximum compressive
force in a diagonal does not occur at the supports or midspan but at the fourth panel from each end.
The largest compressive force is 25% higher than the maximum tension force in this example.
Design approaches for X-type systems will be discussed later in this section.

The diagonals in the Pratt and Warren systems show similar force distributions along the span.
The diagonals of the Warren truss alternate between tension and compression in adjacent panels,
whereas the Pratt system has only tension in this example. The maximum diagonal force occurs at
the end panels due to torsion and is approximately twice that in the X-type system since there is
only one diagonal in each panel. At midspan the Warren system has a maximum compressive
force of 10 kips and the Pratt system almost zero so the bending effect in these two systems is
much smaller than in the X-type arrangement. In both of these single diagonal systems, lateral
movement of the two top flanges occurs in the same direction due to bending, which reduces the
in-plane axial stiffness of the top lateral truss. In the Warren system the lateral displacements (0.08
in.) occur in a local two-panel zigzag pattern along the span whereas the lateral displacements of
each panel in the Pratt system accumulate in the same direction with a maximum lateral
displacement at midspan (0.96 in) [43] as depicted in Figure 32. These lateral displacements induce
local lateral bending stresses in the two top flanges of the tub girder at each strut location in the
Warren system [6]. In the Pratt system, the local top-flange lateral bending effect is concentrated
only at midspan where the truss diagonals change their orientation.

The strut force magnitude and distribution vary among the three systems. The struts are also
usually part of the internal diaphragm system designed to control cross-section distortion and/or
flange lateral buckling. No internal diaphragms were used in the span analyzed for Figure 31 in
order to isolate certain types of forces in the top lateral system. In this case the strut forces can be
due to torsion, bending and sloping webs. In all three systems the horizontal tension force that
prevents the two flanges from spreading apart due to wet concrete load applied at the top flanges
is a constant along the span and is usually quite small. The main strut forces are developed from
torsion and bending. The strut forces due to torsion are zero in the X-type, are related to the
algebraic sum of the transverse component of the two diagonal forces adjacent to the strut in the
Warren system and are equal to the transverse component of the larger diagonal force framing into
ends of each strut in the Pratt system (except at midspan). In the Warren system the strut forces
due to torsion alternate from tension to compression along the span and are relatively small
compared to the compressive strut forces in the Pratt system. Tub girder bending causes the large
tensile forces in the struts near midspan of the X-type system (30 kips), 10-kip tensile forces in the
Warren system and almost zero force in the Pratt system. In the Warren system the strut forces
from all three sources (sloping webs, torsion and bending) are small and fairly constant along the
length compared to the large strut forces generated in the X-type and Pratt systems.

The forces shown in Figure 31 do not include stability brace forces because a first order structural
analysis was performed. Stability brace forces develop from the initial out-of straightness of the
structural components. These forces and deformations are not included in a first order structural

42
analysis. A top flange lateral truss system is a relative brace system that defines the unbraced
length of the top flanges as the distance between panel points during the construction stage. For
design, the stability brace force requirements shown previously should be added to the first order
analysis forces from torsion and bending. Usually, the stability brace requirements will not alter
the top lateral truss design because the largest stability forces occur at the location of the highest
moment where the torsional forces are small. The stability brace requirements will only affect the
top lateral design when the girders are relatively straight.

43
 X - type
60

diagonal1 1
diagonal strut diagonal 2
40

20

strut

Forces (kip)
0
0 20 40 60 80 100 120 140 160 180

-20

-40

-60 diagonal 2

80
Warren

60

40
diagonal 1

20
strut
Forces (kip)

0
0 20 40 60 80 100 120 140 160 180

-20

-40

-60
diagonal 2
-80

80
Pratt
60

40
diagonal
20
Forces (Kip)

-20
strut
-40

-60
0 20 40 60 80 100 120 140 160 180

Distance Along Span (ft)

Figure 31 Top Lateral Truss Forces for Various Tub Girder Bracing Systems

44
 0.019

0.558

A) X-type lateral bracing

system

0.083

0.604

B) Alternating-Diagonal lateral bracing

system

0.956

0.613
C) Single-Diagonal lateral bracing
system
Figure 32 Deformations of different box girder bracing systems

3.1.2 Selecting the Top Lateral Bracing Layout

The layout of the top lateral bracing system is based primarily on effectively resisting the torsional
forces during construction. Usually, the member sizes are kept constant along the span to minimize
fabrication and detailing costs so the panels with the largest torsional moment near supports
controls the initial member sizes. The diagonal forces within each panel due to torsion are similar
for all three lateral systems. The Pratt system with the diagonals arranged for tension will require
the smallest diagonal area. The X-type diagonals, one in tension and the other in compression, and
the Warren system diagonals are controlled by compression. The two X-type diagonals will have
a smaller total weight than the single Warren diagonal because the force and the unbraced length
of the X- diagonal is one half that in a single diagonal system due to the bracing effect of the
tension diagonal. The Warren compression diagonal would be designed for the slightly lower
torsional moment in the panel adjacent to the one with the highest torsion where the diagonal is in
tension. On the other hand, the strut forces from torsion are the highest in the Pratt truss and are
zero in the X-type. The net effect based on brace system weight alone favors the Pratt tension
system by approximately twenty percent over the Warren system. An X-type system over the entire
span is the costliest system because of the greater number of pieces and connections.

The Pratt system is attractive for simple spans because it appears that the diagonals can be oriented
in a tension only arrangement as shown in Figure 31 and the bending compatibility forces are
negligible. However, even in a simple span, compression can develop if the pouring sequence
starts at one end of the span. For example, the girder in Figure 31 will develop compressive top
lateral loads (maximum value of 10 kips) in the three panels near midspan when half the span is
loaded. Evaluation of the pouring sequence is important in Pratt systems. In continuous spans it is
more difficult to guarantee that tension will always control in each panel. This dilemma can be

45
overcome by using a few X-type panels in locations where compression may develop in
conjunction with the Pratt arrangement. The Warren diagonal design, which is controlled by
maximum compression, offers more flexibility to handle variations in the pouring sequences that
often occur in the field.

Torsional rigidity, which is affected by teq, should also be evaluated when designing the top lateral
bracing system. As discussed earlier, for a targeted teq the X-type arrangement requires the least
bracing weight, followed by the Warren and the Pratt systems. Comparing the Warren and the Pratt
system designs for the maximum loads given in Figure 32, the Pratt was 17% lighter. The teq from
the formulas in Figure 29 were 0.023 for the Pratt truss and 0.039 for the Warren truss indicating
less rigidity for the Pratt system. For the Pratt truss design, a 3D-FEA analysis of the uniformly-
loaded simple span gave a midspan rotation 2.3 times greater than the rotation with the Warren
design. This rotation of the Pratt-system girder gave a 1.6 in. relative vertical displacement
between the two top flanges of the tub girder. When torsional rigidity is a principal concern, both
the Warren and X-type systems will be lighter than the Pratt arrangement with the Warren being
more cost effective.

The X-type top lateral truss attracts larger bending compatibility forces than the Warren or Pratt
systems as shown in Figure 31 because its geometric symmetry provides greater in-plane lateral
stiffness. If these girder-bending induced forces are considered in design, the member sizes may
be controlled by bending, not torsion. At locations of the highest girder bending stress where the
torsional moment may be small, both diagonals within the panel can have compressive loads. In
this case a diagonal is not braced at the intersection point because both diagonals can bend out-of-
plane. The unbraced length is the full length of the diagonal.

3.1.3 Determining the Brace Forces

The forces in the top lateral system due to torsion, girder bending and sloping webs can be
determined directly if a 3D-FEA is used as illustrated in Figure 31. Field tests have shown [44]
that the top flange lateral truss and external intermediate cross-frames most important function is
during the construction stage. For construction stages, the designer should consider the forces in
these top flange lateral truss and external intermediate cross-frames at the end of steel erection and
during the deck placement sequence analysis. Most commercially available analysis software
should have the ability to analyze the bridge for these conditions. Additionally, UTrAp is a no-
cost 3D FEM tool that was specifically developed to analyze single and twin tub girder systems
during construction [45]. The struts in the top lateral system may also function as the top chord of
an intermediate cross-frame used primarily to control distortion. Distortional forces are discussed
with intermediate cross-frames later in this volume. Structural models employed by grid analyses
do not directly model the top flange lateral trusses. In this section analytical methods for
determining the top lateral truss forces and the top flange lateral deflection bending stresses
developed mainly by Fan and Helwig [6] will be summarized. The comparison between the forces
from the analytical methods and 3D-FEM is very good.

46
[Link] Torsion

An approximate torsional analysis of a quasi-closed box girder can be performed using the M/R
method (see reference [8]) to determine the torsional moment along the span coupled with the
equivalent plate method for determining the geometric properties of the cross section. The
resulting torsional properties are used in the structural analysis to determine the torsional moments
in the girders. Once the distribution of torsional moment, T, is known, the shear flow, q, within
each panel can be determined from Eq. 9 and used to determine the forces in the top flange lateral
truss. The shear flow acting on the fictitious plate is then transformed into diagonal member forces
in the lateral truss as demonstrated in Figure 33. The type of force (compression or tension) is
important with regard to superimposing the torsionally-induced force with the other force
components that will be discussed subsequently. Although a Warren truss is shown in Figure 34b,
the same expression would be used for the Pratt truss except the forces would generally be all
tension provided the diagonals are oriented properly.

After the diagonal forces from torsion have been established, the strut forces from torsion, Fs, are
determined for the Pratt and Warren truss systems as shown in Figure 34. In the Pratt system the
strut forces are equal to the transverse component of the larger diagonal force framing into the
ends of each strut except where the diagonals in adjacent panels meet at one point (midspan in
Figure 31). That particular strut would be in a Warren configuration. The simple expression for
the strut force in the Warren truss shown in Figure 34 conservatively neglects the effect of the top
flange lateral flexibility. More complex expressions that consider flange flexibility in the Warren
system have been developed [46] but since the strut forces are small in typical bridges, this
conservatism will not affect the design. The strut forces due to torsion are zero in the X-type. The
signs of the forces (tension or compression) must be maintained so truss forces from sources can
be properly superimposed.

Fictitious -X1
Plate F1 X1
w  
X2
q
q -X2
-F2

q=T/(2A0) F1,2 = ± q1,2w/sin  X1,2 = ± q1,2w/(2sin )

(a) (b) (c)

Figure 33 Diagonal Lateral Brace Forces Due to Torsion in a Tub Girder

47
 F1 F1
F2
Fs = F1sin
Fs =(F1-F2)sin

 

(a) Pratt Truss (b) Warren Truss

Figure 34 Strut Forces from Torsion

[Link] Sloping Webs

The sloping webs of trapezoidal girders also induce a lateral load component on the top flange.
This lateral load component causes additional top flange lateral bending stress as well as axial
forces in the struts of the lateral truss. The struts are typically designed to carry the horizontal
component due to the sloping webs. Historically, some past design aides [8] provided
recommendations that the top and bottom flanges each support half of the horizontal web
components of the applied load. Based on this assumption, the half acting on the bottom flange
does not generate any top flange lateral bending stress or forces in the struts. While this assumption
would be relatively accurate for the girder self-weight, the sloping web component from external
loads from sources such as the fresh concrete deck must be resisted by the top flange lateral truss.
This can be demonstrated by considering a free body diagram of the top flange with an externally
applied distributed load of p/2 applied to each flange. Figure 35 demonstrates the transformation
of the vertical load into a web shear and a horizontal component, f (force per unit length). For a
truss panel length of s, the recommended design tensile force for the struts is equal to (f × s). The
maximum lateral flange bending moment due to the top flange loading is (fs2)/12, assuming the
top flange behaves as a continuous beam supported at the strut locations.

p/2 p/2 p/2 p/2 w

f = 2 tan
p/2 p/2 p/2 p/2 f f
p/2 p/2 f f
 

(a) (b) (c) (d)

Figure 35 Strut Forces from Top Flange Loads

48
[Link] Vertical Bending

In addition to torsionally-induced forces, the top flange truss also develops forces due to vertical
bending of the box girder. When the lateral truss system is attached to the top flanges of the tub
girder, the longitudinal top flange deformations between panel points from bending stresses
produces a compatible longitudinal deformation and corresponding force in the truss diagonals.
As shown in Figure 36, the tub girder and the top lateral truss together resist the vertical bending.
If the tub girder alone was designed to support the bending forces, the stress distribution through
the depth of the tub girder would be represented by the dashed line. The forces in the diagonals
reduce the tub bending stresses shown by the solid line. These strain- compatibility truss forces
are generally undesirable since the primary purpose of the lateral truss is for torsional stiffening.

Top flange stress, fx

Horizontal F –Truss
yt Truss Force

Neutral Axis Girder

bending
stress
Figure 36 Tub Girder Vertical Bending Stresses

The bending compatibility forces can be significant as illustrated in Figure 32, especially for the
X-type top lateral system. If the X-type redundant lateral system is not considered when
proportioning the tub girder for bending, then the compatibility forces determined from an elastic
analysis need not be considered in design. The diagonals can be designed for torsion alone. Within
a panel the compatibility forces increase the force in one diagonal and decrease the force in the
other diagonal by the same amount as shown in Fig 34. Torsion alone develops the same absolute
magnitude of force in both diagonals. If the diagonals are designed only for torsion, one of the
diagonals will reach its design limit first because of the added compatibility force, say 10 kips.
The axial stiffness of that panel is then reduced and additional compatibility forces will also be
reduced as additional load is applied to the bridge. The other diagonal in the panel has a smaller
force than expected; the torsion force minus 10 kips. As additional bending and torsional forces
are applied to the tub girder, the force in the highest stressed diagonal will not change (it is at its
strength limit). The diagonal with the lower force will resist the torsion alone but it can support an
additional torsion force equivalent to the compatibility force (10 kips). The compatibility forces
do not affect the ability of a panel with two diagonals to resist the torsional moments. There may
be some initial sag in both diagonals at locations of low torsion, but this does not affect structural
performance of the tub girder system. Designing for compatibility forces in redundant systems is
somewhat self-defeating. When the brace size increases, the compatibility forces also increase.

In the Warren truss and Pratt systems with a single diagonal within each panel, the bending
compatibility forces must be considered because the system is not redundant. When the single
diagonal in compression reaches its strength limit by forces from torsion and bending
compatibility, only small additional forces can be applied.

49
Equations for predicting the truss forces induced due to vertical bending of the tub girder are
available [6] and summarized in Figure 37. The Warren truss layout also results in a lateral load
on the flanges that cause the flange stress denoted by fL bend in the figure. The formulations for the
diagonal forces are related to the strut and diagonal sizes; the larger the members, the larger the
forces in those members from tub girder bending. The top lateral compatibility forces will be
compressive in the positive moment regions and tensile for negative moments.

The expressions given in Figure 37 for the Warren system were developed for the specific case of
internal cross-frames positioned in every other panel, which is a spacing of 2s, where s is the
spacing between the struts of the top flange truss. Bending induced forces in the top flange Warren
truss are sensitive to the spacing between the internal K-frames. When internal cross-frames are
spaced at every panel point of the top flange truss (spacing of s), the bending induced top lateral
forces are actually larger [46, 47, 48] The 2s (every other panel point) spacing of the internal K-
frames in the Warren truss system is recommended. At the truss panel points between the internal
K-frames only a strut is provided. The intermediate internal cross–frame spacing does not affect
the forces in the X-type system.

There are currently no direct analytical solutions for the Pratt arrangement but 3D-FEA has
indicated that the diagonal and strut forces due to bending are much smaller than those in the
Warren system over most of the span. However, at locations where the Pratt diagonals change
their orientation (maximum bending moment locations), the two adjacent diagonals meet at one
point in a Warren configuration. At these locations, the bending compatibility forces are maximum
and similar in both the Pratt and Warren orientations.

50
 Dbend = diagonal force due to girder bending
Sbend = strut force due to girder bending
fx top = top flange bending stress in panel
s =panel length (spacing between struts)
 = Angle between diagonal and flange
Ld = diagonal length
w = strut length
Ad, As = respective area of diagonal or strut
bf, tf = respective width and thickness of girder flange
fL bend = lateral bending stress in girder top flange

Diagonal
w
Strut
 

s
f xTop s cos  f xTop s cos 
Dbend = D bend =
K1 K2
Ld 2wsin 2 
3
Ld w s
K1 = + sin 2  + 3 sin 2  K2 = +
Ad As 2b f t f Ad As
S bend = − D bend sin  S bend = − 2 D bend si 

1.5 s
f L bend = S
b f 2 t f bend

WARREN, (Pratt) X-TYPE

Figure 37 Bending Induced Truss Forces

3.1.4 Top Flange Truss Details

The number of panels and the orientation of the diagonals can have a significant effect on the
efficiency of the design as well as the performance of the girder system. In specifying the number
of panels along the span length, the angle of the diagonals  defined in Figure 39 should be kept
within the range 35° <  < 50°. The upper limit on this range is related to economics since larger
values of  will lead to more panels which results in more connections and larger fabrication costs.
The lower limit on this range is related to the compression behavior of diagonals from both
torsional and vertical bending. With a smaller angle of inclination, the diagonals become relatively
long and therefore possess a lower buckling capacity. In general, diagonals with orientations
outside of the recommended range are inefficient and should be avoided.

51
Structural T-sections are often used for the diagonals, while angles are commonly used for the
struts. For practicality of the connections and safety of the construction workers, the T-sections
should be oriented with the stem pointing downwards. The construction personnel often must
walk on these members during erection and early stages of construction. In addition, the stem
should be pointed downward to avoid clearance issues with the metal deck forms. In detailing the
connections for the diagonals, care should be taken not to employ excessively thick connection
plates or shims that will increase the eccentricity of the connection. The thickness of the
connection plate should be approximately equal to the thickness of the WT flange.

The strut for the top flange truss frequently serves as the top chord member of an internal cross-
frame if one is provided at the panel point. To avoid congestion at the intersection of the struts
and the diagonals, some designers connect the strut to the web stiffener at an eccentricity denoted
as e in Figure 38.

e
Strut

Figure 38 Strut eccentricity in a tub girder cross section

This eccentricity generally has an insignificant effect on the performance of the top flange truss;
however, the eccentricity should be limited to a maximum value of 3 or 4 inches. In many cases,
lowering the strut due to concerns about congestion between the diagonals and the struts is
unnecessary because of the inclination of the diagonals and the length of the connection. In cases
where the Pratt truss geometry is specified, the effects of e are more significant than for the Warren
and X-type layouts because the strut forces are much higher. With an eccentric connection such as
the one depicted Figure 38, forces from the diagonal in the Pratt truss would be transferred into
the web/stiffener of the girder, down to the strut and across the girder, back up the web/stiffener
of the girder and into the diagonal of the adjacent panel.

3.1.5 Controlling Global Lateral Buckling

Lateral buckling of an I-shaped girder is a well-documented limit state included in the AASHTO
LRFD BDS. Either intermediate diaphragms or top flange lateral truss systems are used for braces
to establish the unbraced length of the compression flange. Global lateral buckling of the tub girder
as a whole is not as well understood and there are no AASHTO provisions for this phenomenon.
There have been two total collapses from global buckling of straight tub girders during the deck
pour [4]. In both cases the girders had frequently spaced internal diaphragms but no top flange
lateral truss system.

Open-section tub-girder cross sections are susceptible to lateral buckling due to the location of the
shear center (see Appendix), which is well below the bottom flange. Global buckling of tub girders

52
is discussed by Yura and Widianto [49]. It is very unconservative (by approximately a factor of
five) to use formulas developed for single web I-shaped members such as Eq. 10 for checking
lateral buckling of twin web tub girders. Another unconservative approach is to assume that lateral
buckling cannot occur if the tub girder is bent about its smallest principal axis. The Marcy bridge
that collapsed had Iy/Ix = 1.75. A girder with a trapezoidal shape has reduced lateral buckling
resistance compared to a rectangular girder.

The lateral buckling capacity of tub girders can be determined from a 3D-FEM buckling analysis
(a free download is available for one such program [45]) or from the classic lateral buckling
formula for singly-symmetric cross sections [28],

 2 EI y   x  x2 GJ  Lb  Cw 
2

M cr =   + + 
EI y   
(27)
2 Iy 
2
Lb 4
 

where x is the monosymmetric constant. Formulations for x and Cw for an open-section tub girder
are given in Section 5.2 and 5.3. If Eq. (27) indicates that an open-section tub girder is inadequate
during the construction stage, a partial or full-length top flange lateral truss system or external
bracing will be required. The number of internal diaphragms does not affect global lateral buckling.

The effect of installing a top lateral X-type truss system along the entire length of a uniformly-
loaded, straight 170 ft simple span tub girder to improve the global buckling strength is illustrated
in Figure 39. The trapezoidal cross section used in the analysis was similar to the Marcy Bridge.
The area Ad of the truss diagonals was varied, and a 3D-FEM buckling analysis performed. For
an open section (no lateral system), the girder buckled at a top flange bending stress of 13 ksi. The
addition of a top lateral system increases the buckling strength linearly as the size of the truss
diagonals increases. Only a very small area for the top lateral bracing diagonals is required to
increase the global buckling strength to adequate levels. For this particular girder, if the bracing
area of the X-type system is larger than 0.08 in2 (teq = 0.0015 in.), then global LTB will occur at a
stress higher than 50 ksi. Single diagonal bracing systems give similar results provided that the
area of the diagonal is two times the area shown for the X-type system.

53
 Ad of X-type System (in)2

0 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16

100
90
80
70
Fcr 60
(ksi)50
40
30
L=170 ft
20
10
0
0 0.0005 0.0010 0.0015 0.0020 0.0025 0.0030
t eq (in)

Figure 39 Effect of teq on global buckling on a tub girder section using a X-type lateral
bracing system

Installing a top lateral bracing system along the entire span length for a condition that occurs during
construction may be expensive. A study [49] indicated that bracing of the end panels where the
flange stress is low is much more effective than bracing near midspan where the stress is
maximum. The effect of the end panel bracing on the global buckling strength is shown in Figure
40.

54
 b/L
0 0.05 0.1 0.15 0.2
80

70

60

50
Fcr (ksi)

40

30

20 X-type, Ad=1.05 in^2

10 Single-Diagonal, Ad=2.1 in^2

0
0 1 2 3 4 5 6
Number of braced panels at each end
X-type top lateral bracing system

6.3  Single-Diagonal top lateral bracing system 18.9

b L=170 ft b

Figure 40 Effect of Partial End Panel Bracing on Girder Buckling Stress

Both the number of braced panels and the ratio of the length of the braced panels at each end, b,
to the length of the girder, L, are shown. The diagonal areas of the X-type and Single-Diagonal
systems are 1.05 in2 and 2.1 in2, respectively, corresponding to the teq of 0.019 inches and 0.012
inches. To achieve Fcr higher than 50 ksi so that yielding will govern, only bracing of four panels
at each end is required. Since the teq of both the X-type and the Single-Diagonal systems are about
the same, the effectiveness of both systems is almost the same. A global buckling parameter study
with span, number of braced panels, cross-section proportions and type of truss layout as variables
and a constant teq = 0.02 in. in the braced panels indicated that Mcr is linearly proportional with the
increase of b up to b/L = 0.2. Additional braced panels do not have a large impact on the global
lateral-torsional buckling strength.

3.2 Intermediate Internal Cross-Frames

The primary role of intermediate internal cross-frames/diaphragms in tub girders is to maintain the
shape of the cross section against torsional forces that tend to distort the shape of the box girder.
Typical geometric arrangements, commonly called K-, X- or Z-frames, employed as intermediate
cross-frames are shown in Figure 41. Solid plate intermediate diaphragms are typically reserved
for support regions. In the absence of a top lateral system, internal cross-frames act as torsional
braces to control lateral buckling of the top flanges. Torsional bracing stability requirements were
presented earlier. This section outlines the design requirements for internal cross-frames to

55
properly control distortion and provides recommendations on detailing practices for the internal
cross-frames.

K-Frame X-Frame Z-Frame Plate

Figure 41 Internal Intermediate Cross-Frame Layouts for Tub Girders

3.2.1 Tub Girder Distortion

Depending on the distribution of the applied torsional loads, the cross-section of a quasi-closed
tub girder may distort from its original shape. This distortion of the cross-section can lead to
significant warping stresses, which are in addition to torsional warping stresses. Warping stresses
that develop as a result of distortion of the cross-section are appropriately referred to as distortional
warping stresses. While torsional warping stresses in box girders may be relatively small, without
proper bracing distortional warping stresses can be quite significant.

Forces develop in the intermediate cross-frames and other bracing members due to the distortion
of the box section. Torsion in box girders is usually the result of either horizontal curvature of the
girder or unbalanced gravity loading that results in an eccentricity of the load on the cross-section.
Depending on the type of loading, the torque on girders can be visualized as either a horizontal or
vertical couple as depicted in Figure 42.

Figure 42 Sources of Torsion in a Tub Girder

The torsional moments in curved tub girders determined from the M/R method can be visualized
as a horizontal couple. In the cases of unbalanced gravity loading, the effective eccentric loading
can be idealized as two couples, pure flexural load plus a torque consisting of a vertical couple.

56
Cross-sectional distortion of box girders is induced by the components of the two torsional loads
that are not directly distributed in proportion to a uniform Saint-Venant shear flow on the cross-
section. All practical loading cases cause some form of cross-section distortion since the load
application is never distributed in proportion to the Saint-Venant shear flow. Each torsional load
can be divided into pure torsional components (kips per unit length), qh,v,T, and distortional
components, qh,v,D as shown in Figure 43.

Figure 43 Pure Torsional and Distortional Components in a Tub Girder

The pure torsional components are distributed around the cross section in proportion to the St.
Venant shear stresses. The distortional components of the applied loads on the right side of the
figure yield zero net torque on the cross section. The horizontal and vertical torsional couples
produce distortional components that are in opposite directions even though both torsional couples
produce clockwise moments. Therefore, a distortional analysis requires a separation of the
horizontal and vertical components and a sign convention for the eccentricity (see Figure 44).

Figure 44 Sign Convention for Eccentricity

An approximate distortion analysis was developed by Fan and Helwig [50] for determining the
forces, H and V, (see Figure 45) applied to intermediate cross-frames. H and V are equal to the
qhD and qvD distortional unit forces, respectively, given in Figure 43 multiplied by the spacing

57
between intermediate cross-frames, SK. These applied forces are applicable for any of the cross-
frame arrangements shown in Figure 41. For the specific K-frame which is the most common, the
H and K forces are converted to diagonal and strut forces as follows:

Figure 45 Approximate Distortional Forces in Intermediate Cross-Frames in a Tub Girder

s K L dK M b
D=± ( - ep) (28)
2A 0 R w

sK b b M
S=± ( ep - ) (29)
4A0 w R

where: D = distortional induced force in the K-frame diagonal;

S = distortional induced force in the K-frame strut;
sK = spacing between internal K-frames measured along the girder length;
LdK=length of the K-frame diagonal;
A0=area enclosed by box=(w+b)/2d;
w and b = box girder dimensions as depicted in Figure 45;
e = effective eccentricity of resultant distributed load;
p = distributed load (weight/ft.);
M= box girder bending moment; and
R = radius of horizontal curvature of box.

The M/R term in the parentheses is directed at the torsional effects of horizontal curvature while
the ep term in the parentheses captures the effects of eccentric gravity loading. The plus/minus
sign on the expressions indicates that the distortion induces tension and compression as indicated
in Figure 45b. One diagonal experiences compression while the other experiences tension. In the
case of the strut, equal magnitudes of tension and compression are induced on either side of the
two diagonals. Since the struts serve as members of both the internal K-frames and the top lateral
truss, these members have torsional components from box girder bending, torsion, and distortion.
The components due to torsion and bending are uniform across the strut while the distortional
components have equal magnitudes of tension and compression as indicated by the plus/minus
sign in Eq. (29). The distortional component can therefore be isolated from the bending and

58
torsional component by averaging the magnitudes of the strut force on either side of the two
diagonals.

Examples demonstrating the use of these preceding equations can be found in Fan and Helwig [50]
and Helwig, et. al [51].

3.2.2 Internal Cross-Frame Details

As discussed in the Section [Link], the spacing of the intermediate internal cross-frames, should
be sk = 2s, where s is the panel spacing in the top lateral system in single diagonal systems. Not
only is the cost reduced by using fewer intermediate cross-frames, but the forces in the top lateral
system are also reduced. The panels adjacent to the cross-frames only have a single top flange
strut.

A K-frame composed of angles is the preferred layout arrangement based mainly on its
convenience when the interior of the closed box must be inspected. With an X-frame it is difficult
to travel within the interior. Z-frames with only one diagonal provide less interference but the
diagonal may be in compression. Since its diagonal is almost twice the length of a diagonal in a
K-frame, a heavier frame will be required. The intermediate cross-frame should also be detailed
to minimize fatigue issues since live loads can cause distortion.

3.3 End Diaphragms

Diaphragms are provided at the supports of the tub girders. The diaphragms are required for
torsional equilibrium of the girder system. Diaphragms resist the girder twist at the ends of the
girders based primarily upon the shear stiffness of the diaphragm if the aspect ratio, Ld/hd, is less
than 4 [1]. The ends of the girders are typically closed by solid plates and the diaphragm that
connects the adjacent girders is may be trapezoidal in shape. The detailing requirements of the
support diaphragms depend on the aspect ratio of the end diaphragms. Although the end
diaphragm is typically trapezoidal in shape and is often bolted to the two adjacent girders, for
establishing the detailing requirements the effective length of the diaphragm can be assumed to be
measured as the spacing between the center of the bearings of two adjacent girders as shown in
Figure 46. For most tub girder geometries, the practical range of spacing between the bearing
centerlines and therefore the effective diaphragm length is in the range of 14 ft. < Ld < 20 ft. The
lower range of the diaphragm depth so that shear stiffness dominates the behavior is 3.5 feet at the
lower Ld range (14 ft.) and 5 feet at the upper Ld range (20 ft.). The end diaphragms are often
deeper than these lower-bound values so shear stiffness will govern the behavior of these braces.

59
 w g w

hd “Solid Plate
Diaphragm”

Ld
Centerline Centerline
of Bearing of Bearing

Figure 46 Typical End Diaphragm Geometry

The fact that most end diaphragms have aspect ratios less than 4 is important from a detailing
perspective. The end diaphragms usually have top and bottom stiffening plates that increase the
out of plane stiffness of the diaphragm plates. Many designers treat the top and bottom plates as
flanges of a beam and then associate the connection requirements with what is frequently required
in the beams of a frame. The flanges of beams in a frame are often fully connected to columns to
create a “moment connection” between the beam and column. In the case of the plate diaphragms,
the primary mechanism of restraint provided by the diaphragm comes in the shear stiffness of the
plates and connecting the flanges has very little effect on the behavior of the system. This is
illustrated in Figure 47, which shows a 3D-FEM of the end diaphragm in a twin box girder model.
The end connections of the solid diaphragm were modeled with both continuous top and bottom
plates as well as discontinuous plates. There was virtually no difference in the behavior of the two
girder systems [47]. Only in cases where a relatively shallow diaphragm with an effective aspect
ratio in excess of 4 should designers consider making the stiffening plates continuous across the
ends of the girders. In most applications, simply bolting the end diaphragm will provide exactly
the same behavior as if the top and bottom stiffening plates were connected. However, in cases
where redundancy considerations are a significant concern, consideration should be given to
making the stiffening plates continuous.

Figure 47 Non-continuous flanges illustrated by a 3D FEM of the end diaphragm of a twin

box girder model

60
3.3.1 Diaphragm Strength Design Requirements

There are two criteria that the designer should consider when proportioning the solid end
diaphragms for box girder applications. The most obvious consideration is the basic shear strength
of the plate diaphragm. The other consideration is related to excessive shear deformations at the
ends of the beams that can result in rigid body rotations of the girders along the length. The
strength limit state is relatively well-understood; however, an expression based upon stiffness
criteria is also presented in the following sub-section.

Figure 48 shows girder torsional demand that acts on diaphragms and the resulting shears that
develop as a result of these torsional moments.

T1 T2

Ld

V V
Figure 48 Girder end torsional demand that acts on diaphragms and the resulting shears

The moments T1 and T2 are the torsional moments that come as output from the results from a
grid model. The girders are generally subjected to vertical gravity loading and although the net
end reaction will typically be upwards, the diaphragm tends to redistribute the gravity load so that
more of the gravity load shifts towards the exterior girder. Figure 48 shows the redistribution in
the form of a downward shear on one girder and an upward shear on the other girder. The shear,
V, represents the design shear for the end diaphragm. In terms of the girder end torques, the shear
is:

T1 + T2
V= (30)
Ld

In a curved girder, Ld is the difference in the radii of the two girders. The design approach for
shear is based upon a uniform shear stress through the depth of the plate. Referring to the area of
the diaphragm plate as Ad = hdtd, the shear stress would be given as follows:

V T1 + T2 T1 + T2
= = = (31)
Ad Ld Ad Ld h d t d

Based upon a uniformly distributed load on a simply supported girder, the end torque (neglecting
the presence of intermediate K-frames) is:

61
 pL3i pL2 
Ti = = i o (32)
24 Ri 24

where p is the uniformly distributed load, Li and Ri are the respective chord length and radius of
curvature of the ith girder, and o is the subtended angle within the span. This leaves the designer
with several options to design the plate diaphragm for strength. The torque from Eq. (32) also
provides a reasonable estimate of the design torques. Although the equation was derived for a
simply supported girder, there is little torsional interaction between adjacent spans in box girders
since the diaphragms are relatively stiff and the St. Venant stiffness tends to dominate the behavior.
Therefore Eq. (32) provides reasonable estimates of the end torques on each girder. Alternatively,
analysis results can also be used such as getting the torques from a grid analysis. With the end
torques in the two adjacent girders, the diaphragm shear V can be found using Eq. (30) and
compared with the shear strength (i.e., Vn from the AASHTO LRFD BDS [1]). For example, if
the web slenderness satisfies the requirements for full shear yielding, the material shear strength
 = 0.58Fy can be applied with Eq. (31) to give:

T1 + T2
Ad =
Ld (0.58Fy )
(33)

where Fy is the material yield stress of the web of the diaphragm. For a diaphragm web not
satisfying the slenderness limits for full yielding, the appropriate expression for shear buckling can
be utilized.

3.3.2 Diaphragm Stiffness Design Requirements

Instead of the diaphragm strength limit, an alternative limit on the diaphragm may be shear
deformation. Relative vertical deformation between the adjacent girder flanges can also occur due
to deformations in the end diaphragms. To develop a deformational limit, some geometrical
approximations of the portion of the end diaphragm restraining girder twist must be established.
Although the diaphragm itself is usually viewed as a trapezoidal plate, since the plate is fully bolted
to the two girders with slip critical bolts, the portion of the diaphragm resisting girder twist can be
idealized as a rectangular plate extending from the middle of the two girder bearings as depicted
in Figure 49 for a total diaphragm effective length of Ld.

Figure 49 Idealized Rectangular Diaphragm Plate

62
Figure 50 shows the girder end deformations at the diaphragm and then the resulting deformations
of the idealized rectangular diaphragm plate. As discussed above, the girder deformations result
in a rigid body deformation that causes a relative vertical movement between the two flanges of
the box girder. The vertical flange deformations are f in Figure 50. The relative value between
the two flanges is then given by the expression:

Re l dia. = 2f = 2oxr (34)

a + bf
xr =
bf
2  f = o xr

 

Girder E Girder I
a) Girder Deformations
d

d
hd

Ld

b) Diaphragm Plate Deformations

Figure 50 Girder and Diaphragm Deformations

where o is the end twist due to the shear deformations in the end diaphragm, and the distance xr
is defined in the figure. Since the end diaphragm deformations result in rigid body movements of
the entire girder, these deformations are very undesirable and should be kept to a minimum. As a
result, the diaphragms should be made relatively stiff to avoid large relative movements between
the adjacent girder flanges. A value for the tolerable relative movement is a matter of judgment.
In reference [51] a value of 0.5 in. was selected to control variations in the slab thickness due to
girder flexibility along the span. Since a rigid body rotation at the supports contributes to the
midspan relative deflection, the support deformations must be kept to a minimum. A limit of 2f
< 5% x 0.50 in. = 0.025 in. at midspan was chosen. In reference [50], the f = 0.0125 in. limit was
transformed into the following stiffness limit for the area of the diaphragm:

(T1 + T2 ) x r
Ad = hd td  (35)
0.0125GLd

The controlling diaphragm area would be the larger value from Eq. 35, or the strength criteria
discussed in the previous sub-section.

63
3.4 Intermediate External Cross-Frames

The primary role of the intermediate external diaphragms is to control the relative deformation
between adjacent curved girders during casting of the concrete bridge deck. In cases where the
external bracing is not removed after construction, the external bracing will also contribute to the
load transfer between adjacent tub girders. In a twin girder system depicted at midspan in Figure
51, Girder E is the exterior girder with the longest individual span. The girders will twist
independently during construction if there are no intermediate external cross-frames connecting
the two girders with Girder E having the larger angle of twist and vertical deflection compared to
Girder I on the interior. Since the girders have some flexibility, there will be a relative vertical
displacement rel between points B and C in (c) that causes a variation in the deck thickness and
reinforcement cover. The dashed lines in (b) and (c) represent a straight line between the finishing
machine rails at points A and D. The displacement rel must be controlled within some practical
limit. Stiffer girders (larger teq) can be one approach but may be cost-prohibitive and very stiff
girders pose greater problems with bolted field splices during erection. External intermediate
cross-frames shown in (b) can be used to improve the constructability. Studies [48, Error! R
eference source not found.] have shown that only one or two cross-frames near midspan are
sufficient to control rel.

w g w

Girder E Girder I

a) Unloaded

b) Loaded with Cross Frame Present

rel

c) Relative Twist Without Cross Frame

Figure 51 Relative Deformation between Adjacent Girders

64
Cross-frames should be concentrated towards the middle of the span instead of spreading the
braces out equally along the bridge length. Although the deformations often do not differ
substantially with the addition of several braces, the forces induced in the intermediate cross-
frames will be smaller as more braces are added. However, the forces in the intermediate cross-
frames from the construction loading are often relatively small except for spans with sharp curves
(R < 250 ft). With skewed supports, the forces in the cross-frames do become larger. Unless the
support skew becomes large (greater than approximately 30 degrees), the forces can often be
handled with members satisfying slenderness limits.

External intermediate K-frames are primarily needed on horizontally curved girders. In straight
girders, if diaphragms are provided at the supports and a top flange lateral truss is utilized, the
girders are very stiff and intermediate external diaphragms can be omitted. Diaphragms would be
needed if a large unbalanced load (i.e., large torque) is applied or if the supports have significant
skew. The problem with the large support skew is that the ends of the girders may tend to twist
due to the angled diaphragm. The twist at the support results in a rotation of the girders that can
result in problems with the uniformity in the slab thickness as outlined earlier. Most practical
applications of straight girders do not need external bracing.

In most situations, external intermediate K-frames can be removed once the concrete deck hardens
because field tests have shown that external K-frames have very little impact on live load
distribution. The cross-frames are most often removed due to fatigue concerns.

3.4.1 Analysis Approaches for Intermediate External Diaphragms

The treatment of the external braces in the structural analysis as well as the accuracy of the
resulting forces depends heavily on the analysis method that is used. If a 3D-FEM is utilized, the
braces can be modeled relatively accurately and the member forces can be obtained directly from
the analysis. However, incorporating the external K-frames into a grid model poses a complicated
geometrical problem. Although the internal and external K-frames are trusses made up of several
members, the grid models treat these braces as a line element that spans between the centerlines
of the adjacent girders. Therefore, estimating the stiffness of these external braces can be difficult.
In addition, the analytical estimates of the intermediate diaphragm moments from a grid model
may often be of questionable accuracy.

Another possible analysis approach is to designate the intermediate external K-frames solely as
members to help control the constructability of the slab. With this approach, the analysis would
be carried out only modeling the girders and solid diaphragms at the supports. The girders and
support diaphragms would therefore be sized to support the entire load. This will often result in
larger member sizes for the top flange lateral truss, when compared to an analysis that includes the
external braces; however, the economics of the top flange truss should not change too dramatically.
The cost of the top flange truss is mainly related to fabrication costs, which are often primarily a
function of the number of pieces required to fabricate. Increasing the size of a member by a few
pounds per foot should not have too large of an impact on the design economics; however, the
behavior and safety of the design are much easier to predict with this approach. As mentioned
above, a few external cross-frames concentrated near the middle of the span often provide excellent
control over the relative twist between adjacent girders. In the following sections, approximate

65
methods developed in [51] will be summarized that can be used to determine the number of
external cross-frames needed to control constructability and to predict the member forces in the
external cross-frame.

3.4.2 Spacing of Intermediate External K-frames

The approximate approach outlined is for determining how many intermediate cross-frames are
needed. The spacing equation is based on the following assumptions: a simple- span twin curved
girder symmetrical system with radial supports, no relative twist between the two supports of the
span and the two girders have similar cross-section properties. The vertical and horizontal
deflections at the web-top flange intersection at midspan for points B and C from vertical load
applied at the top flanges of the quasi- closed sections were determined. These calculations
required that the angle of twist of each girder, which is a function of the individual span lengths,
be determined along with the vertical displacement of both girders. Since the cross-sectional
properties of the two girders is the same, the difference between the perpendicular displacements
of points B and C relative to a straight line between screed rails, rel, is mainly a function of the
difference in span lengths between the two girders and the identical cross-sectional properties. The
difference in span lengths is related to the spacing between the two girders, (w + g), and the
subtended angle o = L/R of the bridge plan. With some minor assumptions associated with the
relative location of the screed rails on typical tub girder geometries, the relationship of the span
length between external cross-frames and rel is

 
 1.2Δ rel 
L max =  
 5wβ 0 (w + g) )  EI
(36)

  − 3 
 384EI  GJ 

For design, the maximum permissible deviation in slab thickness rel can be set equal to some
maximum value, say 0.5 in., to determine Lmax. If Lmax is greater than the span length, no external
intermediate cross-frames will be required for constructability.

3.4.3 Forces in Intermediate External K-frames

To develop expressions for the forces in the external diaphragm, the case of a single external K-
frame located at midspan will be considered. The lengths, Li and Le, of the respective interior and
exterior girders in these derivations are taken as the total arc length of the respective girders
between the supports. The external diaphragm helps to restrain twist and vertical deflections of
the box girders at the location of the braces. The basic geometry of the external K-frame and box
girder system are shown in Figure 52. The angle of the diagonal of the K-frame is represented as
, while the depth of the K-frame is denoted as hK. The rotation of the two girders and the K-
frame system are assumed to be the same and are represented as . The distance from the center
of a girder to where the top chord of the K-frame connections is shown as LT. The distances hK
and LT will be used to represent the torque exerted by the external K-frame on the girders.

66
 Interior
Exterior w
g
w
hK


 LT
w+g
Figure 52 K-frame Geometry

Exterior Interior
Ve Vi
FT

He Hi FD − FD


FBe FBi
a) Forces on K-Frame b) Member Forces
Figure 53 External K-Frame Forces

The K-frames exert restraining forces H and V on the girders in the respective horizontal and
vertical directions. The equal and opposite forces that act on the K-frame are depicted in Figure
53a. The corresponding member forces that develop in the K-frames are depicted in Figure 53b.

Liw,ext + Lew,int − K1 w,rel

FD = 4GJ
K2 (37)

4GJ (w,ext − w,int ) − FD LK ( Le − Li )

FT =
hK ( Li + Le ) (38)

FB =  FD cos − FT (39)

The variables in the above equations can be calculated as follows:

LK = hK cos + LT sin

67
  EI   
K 0 = 1 + 1 + 1 − cos 0 
 GJ  2 

Li + Le
K1 =
w+ g

L3i + L3e
K 2 = K 0 K1 sin  + 2 Li Le LK
12( EI / GJ )

 w,rel = K 0
5w
384 EI
(L4e − L4i )

5wL4i  EI 
w,int = 1 + 
384EIRint  GJ 

5wL4e  EI 
w,ext = 1 + 
384 EIRext  GJ 

Comparison of the member forces with the forces from a 3-D FEA show reasonable agreement
[47].

68
4.0 BRACING MEMBER DESIGN AND CONNECTION DETAILS

4.1 Design of Tees and Angles

For tension and compression members, the 2020 AASHTO LRFD BDS [1] has adopted the limit
state provisions from the 2016 AISC Specification [22] and added mandatory L/r limitations. All
-factors in AASHTO [1] are generally larger than their counterparts in the 2016 AISC
Specification [22]. Therefore, the use of the design capacities in the AISC Manual [52] for tension
and compression members and their connections is conservative. If desired, engineers can adjust
AISC Manual [52] tabulated values by the ratio of the respective -factors (AASHTO/AISC).

4.1.1 Tension Members

Single angle, double angles, and tee-sections in tension are checked for the limit states of yielding
on the gross section, fracture on the net section, block shear at the end connections and limiting
slenderness ratio. When evaluating the slenderness ratio, any bracing member required to enable
a main member to support the applied loads should be treated as a primary member. That is, if the
main member cannot support the design loads without the bracing member, then the bracing is not
a secondary member. Single and double angle members are usually connected at the ends through
one of the legs. The eccentricity between the end connection and the angle centroid has little
structural significance and is typically ignored. Tees are mainly connected through the flange.
The bracing members are connected to the main member by connection plates on the webs or
directly to the flanges. The eccentricity between the connection and the member centroid has little
structural significance for tension members and can be ignored. Also, the U-factor reduction in
the net section fracture capacity in AASHTO Section [Link]-2 [1] was developed from test data
with specimens that had eccentricities.

4.1.2 Compression Members

[Link] Single angles

Single angles are commonly used in cross-frame members. Since the angle is typically connected
through one leg only, the member is subjected to combined axial compression and flexure about
both principal axes due to eccentricities of the applied compressive force. Additionally, the angle
may be restrained by differing amounts about its geometric axes. As a result, the prediction of the
nominal compressive resistance of these members can be difficult.

The methods specified in various design codes for the design of single angle compression members
loaded through one leg can be divided into two categories: column approaches and beam-column
approaches. The column approach, which is the simplest for design, was given in the first edition
(1971) of the ASCE Standard 10-97, Design of Latticed Steel Transmission Towers [53]. Based on
the results of many angle tests and full-size tower tests, angle design was based on axial load only
with an adjusted L/rz used with the normal column formula to account for eccentricities and end
restraint. The British, European, South African, and Japanese steel design specifications use
similar column approaches. The main variable among these different standards is the axis used to

69
define the critical column slenderness ratio. The ASCE method uses the minimum principal radius
of gyration, rz.

Prior to 2005, AISC [54] specified a beam-column approach that required the eccentricity to be
defined explicitly about both principal axes and that second-order effects also be considered. For
unequal-leg angles the orientation of the principal axes makes the geometry even more complex.
Over the past thirty years, numerous experimental studies have shown the AISC method to be
overly conservative [55]. In 2005, the AISC Specification [22] adopted a new design approach
based on the ASCE 10-97 [56] Standard for angles loaded through one leg. The ASCE method is
based mainly on tests with small (2.5 in.) equal leg angles. Tests with larger angles and unequal
legs [57] showed that the capacity with the short leg connected was approximately 80% greater
than tests with the long leg attached. The ASCE method predicted the same capacity for both cases
since rz is constant for both arrangements. Also, tests have shown that the dominant deformation
is bending about the centroidal axis parallel to the attached leg, which will always be called the x-
axis, so L/rx should be the controlling slenderness ratio. Taking rz/rx = 0.63 for all equal leg angles,
the ASCE effective slenderness of KL/rz for typical tower framing details (not highly restrained)
was directly converted to an equivalent KL/rx formulation [57]. There is an additional equation for
the effective slenderness ratio, (KL/r)eff, given for longer members. Using rx instead of rz made the
ASCE formulation reasonably accurate for both equal and unequal leg angles.

The AASHTO LRFD BDS has adopted the AISC method [22] for single-angle members in Article
[Link], which significantly simplifies the design of single-angle members that are subject to
combined axial and flexure, if they satisfy certain conditions. When these conditions are met, the
simplified provisions permit the flexural effects due to eccentric loading to be neglected, and
allows single angles to be designed as axial loaded compression members for flexural buckling
only using an appropriate specified effective slenderness ratio, (KL/r)eff in place of KL/rs, in
Articles [Link], [Link].1, and [Link].2 to determine the nominal compressive resistance, Pn.
Equations for (KL/r)eff for equal-leg angles and certain unequal-leg angles are provided in
AASHTO LRFD BDS Article [Link].

A summary of stability issues for angle members is given in Ref. [58] and Chapter 11 of the SSRC
Guide to Stability Design Criteria for Metal Structures [28].

[Link] Tees and Double Angles

A tee flange is normally attached to a gusset or top flanges by welding or bolting as shown in
Figure 54. The eccentrically loaded tee is subjected to a concentric load, P, and an end moment,
eP, where e is the eccentricity. Unlike tension members, the effects of the eccentricity must be
considered in the design of the brace because the compressive load amplifies the moments and
deflections. In tension members, the bending deformations and moments along the span are
reduced by the tension force. Designing tees and double angles for combined bending and axial
compressive load by directly applying all the required design checks per the AASHTO LRFD BDS
is a formidable task as tees and double angles do not have a similar effective slenderness ratio
method that is used for angles. As such, the chosen member needs to satisfy the interaction
equations for the combined effects of compression and flexure (Article [Link]).

70
 P
e
P

Figure 54 Eccentrically loaded WT section

The nominal compressive resistance, Pn, is computed in accordance with Article [Link], where the
elastic critical buckling resistance, Pe, is determined as the minimum of Equation [Link].2-1
(elastic flexural buckling) and Equation [Link].3-2 (elastic flexural-torsional buckling), as noted
in Table [Link].1-1 of the AASHTO LRFD BDS.

For tees and double angles loaded in the plane of symmetry, the nominal flexural resistance is
computed in accordance with AASHTO LRFD BDS Article [Link].4. The nominal flexural
resistance is taken as the smallest value based on yielding, lateral torsional buckling, flange local
buckling, or local buckling of tee stems and double angle web legs, as applicable. The equations
for computing the nominal resistance for each of these criteria are presented in Articles [Link].4b
through [Link].4e. Moment magnification should be considered as specified in Article [Link].2b.

For tees and double angles subject to combined axial compression and flexure in which the axial
and flexural stresses in the flange of the tee or the flange legs of the angles are additive in
compression, e.g., when a tee is used as a bracing member and the connection of this member is
made to the flange, a bulge in the interaction curve occurs. As a result, AASHTO LRFD BDS
Equations [Link].1-1 and [Link].1-2 may significantly underestimate the resistance in such cases.
Alternative approaches attempting to capture this bulge have proven to be generally inconclusive
or incomplete [1]. Therefore, it is recommended that Equations [Link].1-1 and [Link].1-2 be
conservatively applied to these cases.

4.2 Fatigue Design of Cross-frame Members

Fatigue design should be considered in the design of cross-frame members, especially in

horizontally curved and/or skewed I-girder bridges, where cross-frames can transmit significant
live load forces through the structure and the force effects in these members are determined from
a refined analysis. In these cases, the fatigue live load forces in the cross-frame members should
be considered in the design of the members and their connections. Refer to NCHRP Report 962
[16] for information on research that has been completed relative to the checking of fatigue in
cross-frame members. As of this writing, recommendations from this research are being
considered for possible adoption into the next edition of the AASHTO LRFD BDS.

4.3 Welded and Bolted Connection Details

The AASHTO/NSBA Collaboration publication, Guidelines for Design Details [59], shows
preferred details for connecting cross-frame and lateral bracing members to girders. Additionally,

71
the NSBA publication, Practical Steel Tub Girder Design [60], provides bracing connection
details in tub girders that are based on preferred detailing practices for cost-effective fabrication.
In both publications, the details provided do not reflect any specific actual design conditions but
are only preferred types of details. All details require careful analysis and consideration, and the
preferred details may not always be applicable.

This section contains a number of pictures of details that are not recommended as standard practice
details. Generally, if details such as these are used, special consideration must be given by the
designer with regard to load paths and load transfer through the various structural components.
Although the pictures focus on tub girders, many of the basic principles also apply to a variety of
different bracing systems. A brief description of the picture is given for each detail.

In Figure 55, the WT for the top flange lateral truss on the box girder had an improper orientation,
and an incorrect size. The stem of the WT was oriented upwards, which increases likely
interference with the metal deck forms. The stem of the WT should point downward so that the
flat surface of the flange is available for the construction personnel to walk on and also so that the
WT can be placed very near the top flange. Also, orientating the WT so that the flange is nearer
to the top flange decreases the eccentricity between the flange and the WT member, and thus
decreasing subsequent bending moments in the WT caused by eccentric loading. A larger
eccentricity results in a very inefficient load path for the lateral bracing system as the load from
the flange is applied to the WT member in a very eccentric manner. This eccentricity must be
considered in the design of the lateral bracing member. Furthermore, in the case shown in Figure
55, the diagonals were also improperly sized. Torsional loads on the box produced compression
in alternating diagonals that was unaccounted for, and members around the supports near the
regions of maximum torsion buckled with only the steel dead load acting on the girders.

A) WT – Stem up B) Buckled Diagonal

Figure 55 Improper orientation and sizing of WT section used for top flange lateral
bracing member

In Figure 56, the WT section is properly oriented; however, the top strut of the cross-frame is
located well below the flange to avoid interference between the two members. Refer to Section
3.1.4 for a further discussion on the recommended maximum eccentricity for these types of details.
The top strut of the internal cross-frame also participates in the lateral bracing system. This detail

72
results in a very inefficient load path between the two members since force components must be
transferred through the web between the two members. The designer must consider how this force
is transferred in the design of this connection. As an alternative, the top strut should be located as
close as possible to the same plane as the lateral bracing system, and ideally, as close as possible
to the top flange.

Figure 56 Strut offset below diagonal

Many poor bracing details are the result when working lines of the bracing members are forced to
intersect. An example of this is the top flange truss that is being fabricated in Figure 57. The
picture shows that very large gusset plates were used in an attempt to have the working lines of
the diagonals and the top struts intersect. In the case shown here, the diagonals can cause a
significant eccentric moment on the gusset plate connection to the top flange, which must be
considered in the design of this connection. Better behavior would result if the diagonals were
offset slightly and bolted directly to the underside of the flange, thereby eliminating the gusset
connection plates. Although the working lines may not intersect with this detail, many stability
bracing systems can be offset from the working points by as much as 10% of the unbraced length
with very little impact on the structural performance. For tub girders, the designer can usually
offset the working line by approximately a flange width with little effect on the performance.
Connection eccentricities in the plane of the truss have very little impact on the performance of
the bracing system.

73
 Figure 57 Large gusset plates used so working lines of bracing members intersect

As shown in Figure 58, eccentric connections can lead to poor behavior of the lateral bracing
members, unless properly accounted for in the design. In this case, a 1-inch thick shim plate was
utilized to lower the top flange lateral truss to avoid interference with the permanent metal deck
forms. Even though the bolted connections were fully tightened, the shim plate increased the
eccentricity of the WT section and significantly increased the bending of the member. Under steel
dead load, the compression diagonals experienced large bending deformations.

Eccentricity
increased by
shim plate

Figure 58 Shim plates can significantly increase the eccentricity of WT sections

The use of shim plates to lower top flange lateral bracing members in an effort to clear permanent
metal deck forms is not recommended herein. Instead, consider using a deeper concrete haunch if

74
possible, and reserve the use of shims as a last resort. If it is determined that shims will be used,
consider the effects of connection eccentricity on the top flange lateral bracing member design
moments, and address the potential for bolt bending in the bolted connection design.

Figure 59 shows a partial depth end diaphragm that results in a poor load path that cannot anchor
the top flange lateral truss. The top flange lateral truss must be sufficiently anchored to the end of
the tub girder. The partial depth end diaphragm was used to accommodate a thickened armor joint
at the end of the bridge. However, the thin web of the box girders may not be stiff enough to
anchor the very large diagonal forces in the top flange lateral truss. The torsional stiffness of the
box is therefore reduced, which can lead to large deformations. In this case, the load path must
be considered in the design of this lateral bracing system, or a full depth end diaphragm should be
used.

Figure 59 Partial depth end diaphragm.

75
5.0 SIMPLIFIED GEOMETRIC PROPERTIES FOR TUB GIRDERS

Symbols are defined in Figure 60.

5.1 Shear Center, ey, for Open and Quasi-Closed Sections [61]

Figure 60 Calculation of Shear Center for Open and Quasi-Closed Sections

𝑡𝑏 𝑏𝑏3 𝑑𝑡𝑤 2 2
𝑡𝑏 𝑏𝑏3 𝑡𝑡 𝑏𝑡 𝑤 2
𝐼𝑦 = + (𝑤 + 𝑤𝑏𝑏 + 𝑏𝑏 ) + ( + )
12 6 6 2

−𝑡𝑡 𝑏𝑡 𝑤 2
𝑡𝑒𝑞 + 0.875𝑑
𝑞𝑎 =
2 𝑤 𝑑 𝑏𝑏
( )
𝑤 𝑡𝑒𝑞 + 2 𝑡𝑤 + 𝑡𝑏

𝑑 𝑡𝑡 𝑏𝑡 𝑤𝑡𝑒𝑞 (2𝑤 + 𝑏𝑏 )𝑑𝑡𝑤 𝑏𝑏

𝑒𝑦 = [( + ) 𝑤2 − + 𝑞𝑎 (𝑤 + 𝑏𝑏 )]
𝐼𝑦 2 12 12

The shear center location ey is measured from the centroid of the bottom flange. A negative value
indicates that the shear center is below the bottom flange. For open-section tub girders, teq= 0 and
qa = -btttw.

76
5.2 Monosymmetry Coefficient, x – Open section only [62]

 1 2 2  1 2 w2 2 
 b b
y A bb + y b  − 2 At t
y bt + + yt  + 
1  12   12 4 
x =   − 2y
Ix   o


2tw  xc 3

 tan  sin   24
(
w − bb −
3
) 1 4
64
( 4 
)
w − bb  + w
t
 2cos
yb − yt
4 4
( ) 


where At = bttt and Ab = bbtb. All geometric functions have positive values but x will be negative.

5.3 Warping Moment of Inertia, Cw – Open section only [62]

Cw = +
( )
ca2 Ab ca2 + ca cb + cb2 Aw  cc2 cb2 
+ + A
12 6 6 2 t
 

where ca = eybb, cb = eyw + dbb, cc = bt(ey – d) and Aw = dtw

77
6.0 REFERENCES

1. AASHTO. LRFD Bridge Design Specifications, 9th Edition. American Association of State
Highway and Transportation Officials, Washington, DC, 2020.

2. Vlasov, V. Thin-Walled Elastic Beams, 2nd Edition. National Science Foundation,

Washington DC, 1961.

3. Timoshenko, S. and Gere, J. Theory of Elastic Stability. McGraw-Hill, New York, 1961.

4. Corr, D, McCann, D. and McDonald, B. “Lessons Learned from Marcy Bridge Collapse”.
Proc., ASCE Forensic Engineering Congress, 2009.

5. Davidson, J., Keller, M. and Yoo, C. “Cross-Frame Spacing and Parametric Effects in
Horizontally Curved I-Girder Bridges”, Journal of Structural Engineering, ASCE, Vol. 122,
No. 9, pp 1089-1096, 1996.

6. NSBA. Steel Bridge Design Handbook: Structural Behavior of Steel. National Steel Bridge
Alliance, 2021.

7. Fan, Z. and Helwig, T. “Behavior of Steel Box Girders with Top Flange Bracing”, Journal of
Structural Engineering, ASCE, Vol. 125, No. 8, pp. 829-837, 1999.

8. NSBA. Steel Bridge Design Handbook: Structural Analysis. National Steel Bridge Alliance,
2021.

9. AASHTO/NSBA Steel Bridge Collaboration. G13.1 – Guidelines for Steel Girder Bridge
Analysis, 3rd Edition. American Association of State Highway and Transportation Officials,
Washington, DC, 2019 .

10. Poellot, W. “Computer-Aided Design of Horizontally Curved Girders by the V-Load Method”.
Engineering Journal, AISC, Vol. 24, No. 1, pp. 42-50, 1987.

11. Liu, D. and Magliola, R. “End Forces on Cross-Frames in Horizontally Curved Steel I-Girder
Bridges”, Practice Periodical on Structural Design and Construction, Vol. 15, No. 1, pp. 21-
26, 2010.

12. Gaylord, E. and Gaylord, C. Structural Engineering Handbook, 3rd Edition, McGraw –Hill,
New York, 1990.

13. Grubb, M. A. “Horizontally Curved I-Girder Bridge Analysis: V-Load Method”,

Transportation Research Record, Record N982, Transportation Research Board, Washington,
DC, pp. 26-36, 1984.

14. Stith, J., Schue, A., Farris, J., Petruzzi, B., Helwig, T., Williamson, E., Frank, K., Engelhardt,
M. and Kim, J. “Guidance for Erection and Construction of Curved I-Girder Bridges”, TxDOT

78
 Report 0-5574-1, Center for Transportation Research, The University of Texas at Austin, pp.
240, 2010.

15. PennDOT. Cross-Frame and Solid Plate Diaphragms for Steel Beam/Girder Bridges Design
with Refined Methods of Analysis. Bridge Design Standard BD-619M, Pennsylvania
Department of Transportation, 2014.

16. NCHRP. National Cooperative Highway Research Project Report 962: Proposed Modification
to AASHTO Cross-Frame Analysis and Design, National Cooperative Highway Research
Program, Transportation Research Board, Washington, DC, 2021.

17. Yura, J. and Helwig, T. “Buckling of Beams with Inflection Points”, Proc., Structural Stability
Research Council, Annual Stability Conference, pp. 761-780, 2010.

18. Yura, J. “Fundamentals of Beam Bracing”, Engineering Journal, American Institute of Steel
Construction, Vol. 38, No.1, pp. 11-26, 2001.

19. Helwig, T. and Yura, J. "Shear Diaphragm Bracing of Beams II: Design Requirements",
Journal of Structural Engineering, ASCE, Vol. 134, No. 3, pp. 357-363, 2008.

20. Flint, A. “The Stability of Beams Loaded Through Secondary Members”, Civ. Eng. Public
Works Review, Vol. 46, No. 537-8, pp. 175-177, 259-260, 1951.

21. Yura, J., Phillips, B., Raju, S., and Webb, S. “Bracing of Steel Beams in Bridges,” Research
Report 1239-4F, Center for Transportation Research, Univ. of Texas at Austin, October, 80
pp., 1992.

22. AISC. Specification for Structural Steel Buildings. American Institute of Steel Construction,
Chicago, IL, 2016.

23. Taylor, A. and Ojalvo, M. "Torsional Restraint of Lateral Buckling", J. of Structural

Engineering, ASCE, Vol. 92 No. ST2, pp. 115-129, 1966.

24. Kitipornchai, S. and Finch, D. “Stiffness Requirements for Cross Bracing”, J. of Structural
Engineering, ASCE, Vol. 112, No. 12, pp. 2702-2707, 1986.

25. Picard, A. and Beaulieu, D. “Design of Diagonal Cross Bracing”, Part1: Theoretical Study,
Engineering Journal, AISC, Vol. 24, No. 3, pp.122-126, 1987 and Part2: Experimental Study,
Vol. 25, No. 4, pp. 156-160, 1988.

26. Helwig, T., Yura, J. and Frank, K. "Bracing Forces in Diaphragms and Cross-Frames", Proc.,
Structural Stability Research Council Conference “Is Your Structure Suitably Braced?”, pp.
129-140, 1993

79
27. Chen, B., Yura, J., Williamson, E. and Frank, K. “Top-Lateral Bracing System for Trapezoidal
Steel Box-Girder Bridges”, Report 0-1898-4, Center for Transportation Research, U. of Texas
at Austin, January, 116 pp., 2003.

28. Ziemian, R., Ed. Stability of Angle Members, Chapter 11, Guide to Stability Design Criteria
for Metal Structures, 6th Edition. J. Wiley & Sons, New York, pp. 493-530, 2010.

29. Wang, L. and Helwig, T.. “Stability Bracing Requirements for Steel Bridge Girders with
Skewed Supports”, Journal of Bridge Engineering, ASCE, Vol. 13, No. 2, pp. 149-157, 2008.

30. Quadrato, C., Wang, W., Battistini, A., Wahr, A., Engelhardt, M., Helwig, T. and Frank, K.
“Cross-Frame and Diaphragm Layout and Connection Details”, TxDOT Report 0-5701, Center
for Transportation Research, 393 pp., 2011.

31. Connor, R.J., and Korkmaz, C. Fatigue Categorization of Obliquely Oriented Welded
Attachments. Purdue University, West Lafayette, IN, 2020.

32. Wahr, A. “The Fatigue Performance of Cross-Frame Connections,” Thesis, University of

Texas at Austin, 228 pp., 2010.

33. Yura, J. A. “The Effective Length of Columns in Unbraced Frames”, Engineering Journal,
AISC, pp. 37-42, 1971.

34. Helwig, T. and Wang, L. “Cross-Frame and Diaphragm Behavior for Steel Bridges with
Skewed Supports”, Report 1772-1, The University of Houston, 231 pp., 2003.

35. Yura, J., Helwig, T., Herman, R., and Zhou, C. “Global Lateral Buckling of I-Shaped Girder
Systems”, J. of Structural Engineering, ASCE, Vol. 134, No. 9, pp. 1487-1494, 2008.

36. Winter, G. “Lateral Bracing of Columns and Beams,” ASCE Transactions, Vol. 125, pp. 809-
825, 1960.

37. Egilmez, Ozgur O.; Helwig, Todd A.; Jetann, Charles A.; and Lowery, Richard. “Stiffness and
Strength of Metal Bridge Deck Forms,” ASCE Journal of Bridge Engineering, Vol. 12, No. 4,
pp. 429-437, July/August 2007.

38. Helwig, Todd A.; Egilmez, Ozgur; and Jetann, Charles, “Lateral Bracing of Bridge Girders by
Permanent Metal Deck Forms,” Research Report 4145-1, Report for Texas Department of
Transportation, January 2005.

39. Klonaris, G. 2003. “Analytical and Computational Studies on the Torsional and Flexural
Response of steel Trapezoidal Box-Girder Bridges”, MS Thesis, Dept. of Civil Engineering,
U. of Texas at Austin, 204 pp., 2003.

40. Kollbrunner, C. and Basler, K. Torsion in Structures – An Engineering Approach, Springer-

Verlag, New York, 1969.

80
41. Heins, C. “Box Girder design – State of the Art”, Engineering Journal, AISC, Vol. 15, No.4,
pp. 126-142, 1978.

42. Topkaya, C., Widianto and Williamson, E. “Evaluation of Top Flange Bracing Systems for
Curved Box Girders”, J. of Bridge Engineering, ASCE, Vol. 10, No. 6, pp. 693-703, 2005.

43. Widianto. “General Behavior of a Steel Trapezoidal Box-Girder during Construction”, MS

Thesis, Dept. of Civil Engineering, U. of Texas at Austin, 398 pp., 2003.

44. Helwig, T., and Fan, Z. “Field and Computational Studies of Steel Trapezoidal Box Girder
Bridges”, TxDOT Report 0-1395-3, The University of Houston, 238 pp., August 2000.

45. Williamson, E., Topkaya, C., Popp, D. and Yura, J. UTrAp 2.0 Analysis of Steel Box Girders
during Construction, Department of Civil Engineering, Univ. of Texas at Austin, 2004.

46. Kim, K. and Yoo, C. “Interaction of Top Lateral and Internal Bracing Systems in Tub Girders”,
J. of Structural Engineering, ASCE, Vol. 132, No. 10, pp. 1611-1620, 2006.

47. Helwig, T., Herman, R., and Li, D. “Behavior of Trapezoidal Box Girders with Skewed
Supports”, TxDOT Report 0-4148-1, The University of Houston, 198 pp., 2004.

48. Kim, K. and Yoo, C. “Bending Behaviors of Quasi-Closed Trapezoidal Box Girders with X-
Type Internal Cross-Frames”, J. of Constructional Steel Research, Elsevier, Vol. 65, pp. 1827-
1835, 2009.

49. Yura, J. and Widianto. “Lateral Buckling and Bracing Of Beams – A Re-Evaluation after the
Marcy Bridge Collapse”, Proceedings, Structural Stability Research Council, Annual Stability
Conference, pp. 277-294, 2005.

50. Fan, Z. and Helwig, T. “Distortional Loads and Brace Forces in Steel Box Girders”, J. of
Structural Engineering, ASCE, Vol. 128, No. 6, pp. 710-718, 2002.

51. Helwig, T., Yura, J., Herman, R., Williamson, E. and Li, D. “Design Guidelines for Steel
Trapezoidal Box Girder Systems”, Report 0-4307-1, Center for Transportation research, U. of
Texas at Austin, April, 72 pp., 2007.

52. AISC. Steel Construction Manual, 15th Edition. American Institute of Steel Construction
Chicago, IL, 2017.

53. ASCE. Design of Latticed Steel Transmission Towers, Standard 10-97. American Society of
Civil Engineers, Reston, VA, 88 pp., 1971.

54. AISC. Load and Resistance Factor Design Specification for Single-Angle Members. American
Institute of Steel Construction, Chicago, IL, 20 pp., 2000.

81
55. Lutz, L. “Evaluating Single-Angle Compression Struts Using an Effective Slenderness
Approach”, Engineering Journal, AISC, Vol. 43, No. 4, pp. 241-246, 2006.

56. ASCE. Design of Latticed Steel Transmission Towers, Standard 10-97. American Society of
Civil Engineers, Reston, VA, 88 pp., 2000.

57. Mengelkoch, N. and Yura, J. “Single-Angle Compression Members Loaded Through One
Leg”, Proc., Structural Stability Research Council Annual Stability Conference, pp. 201-218,
2002.

58. Kennedy, J. and Madugula, M. “Buckling of Angles: State of the Art”, J. of Structural
Engineering, ASCE, Vol. 108, No. 9, pp.1967-1980, 1982.

59. AASHTO/NSBA Steel Bridge Collaboration. G1.4 – Guidelines for Design Details, 1st
Edition, American Association of State Highway and Transportation Officials, Washington,
DC, 2006.

60. Coletti, D., Fan. T., Gatti, W., Holt, J., and Vogel, J. “Practical Steel Tub Girder Design,”
National Steel Bridge Alliance, 2005.

61. Terry, P. “Steel Trapezoidal Box Girder Section Properties”, MS Thesis, Dept. of Civil and
Environmental Engineering, U. of Houston, Texas, 126 pp., 2004.

62. Popp, D. “UTrAp 2.0: Linerarized Buckling Analysis of Steel Trapezoidal Girders”, MS
Thesis, Dept. of Civil Engineering, U. of Texas at Austin, 121 pp., 2004.

82
 Smarter. Stronger. Steel.
National Steel Bridge Alliance
312.670.2400 | [Link]/nsba

B913-22