Coletti, Fan, Holt, Vogel 1

Practical Steel Tub Girder Design

Submission Date: November 15, 2005
Word Count: 7,492

Authors
Domenic Coletti, P.E., HDR Inc., 3733 National Drive, Suite 207, Raleigh, NC 27612, Phone:
(919) 232-6614, Fax: (919) 785-1187, e-mail: [email protected]. Corresponding
author.

Zhanfei (Tom) Fan, Ph.D., P.E., Texas Department of Transportation, 8100 Washington Ave.,
Houston, TX 77007, Phone: (713) 802-5366, Fax: (713) 802-5350, e-mail: [email protected].

John Holt, P.E., Texas Department of Transportation, 125 East 11th Street, Austin, TX 78701,
Phone: (512) 416-2212, Fax: (512) 416-2354, e-mail: [email protected].

John Vogel, P.E., Texas Department of Transportation, 8100 Washington Ave., Houston, TX
77007, Phone: (713) 802-5325, Fax: (713) 802-5350, e-mail: [email protected].

TRB 2006 Annual Meeting CD-ROM Paper revised from original submittal.
 Coletti, Fan, Holt, Vogel 2

Abstract
This paper addresses the entire design process for a steel tub girder bridge and offers a list of
pertinent references for each phase. The paper presents preliminary design considerations,
including appropriate applications for steel tub girders; preliminary girder sizing and spacing
guidelines; framing plan layout considerations; and suggestions related to preliminary
(approximate) design. Also discussed are issues related to final, detailed design, including
available analysis tools; specifics on design of numerous components of a steel tub girder bridge;
suggestions on steel tub girder detailing; and special considerations for construction of a steel tub
girder bridge.
While this paper is not a stand-alone design manual for steel tub girders, it endeavors to
cover the entire design process in outline form in one document, while providing an extensive
list of references the designer can consult. Because nearly every steel tub girder bridge is unique
in terms of configuration, site conditions, local fabrication and construction preferences, and
myriad other factors, the paper typically will outline issues and considerations related to specific
questions, rather than attempting to provide hard and fast rules, in the hope of enabling designers
to make better-informed final decisions for their projects.

TRB 2006 Annual Meeting CD-ROM Paper revised from original submittal.
 Coletti, Fan, Holt, Vogel 3

INTRODUCTION
Steel tub girder use is becoming more commonplace in modern infrastructure design. They offer
advantages over other superstructure types in terms of span range, stiffness and durability—
particularly in curved bridges. In addition, steel tub girders have distinct aesthetic advantages
due to their clean, simple appearance (Figure 1). However, steel tub girder design is in many
ways more complex than steel plate girder design, especially for construction loading stages.
This paper addresses the entire design process for a steel tub girder bridge and offers a
list of pertinent references for each phase. The paper presents preliminary design considerations,
including appropriate applications for steel tub girders; preliminary girder sizing and spacing
guidelines; framing plan layout considerations; and suggestions related to preliminary
(approximate) design. Also discussed are issues related
to final, detailed design, including available analysis
tools (the M/R Method, grid analysis, and three-
dimensional finite element analysis); specifics on
design of numerous components of a steel tub girder
bridge; suggestions on steel tub girder detailing; and
special considerations for construction of a steel tub
girder bridge.
While this paper is not a stand-alone design
manual for steel tub girders, it endeavors to cover the
entire design process in outline form in one document,
while providing an extensive list of references the
designer can consult. Because nearly every steel tub
girder bridge is unique in terms of configuration, site
conditions, local fabrication and construction
preferences, and myriad other factors, the paper
typically will outline issues and considerations related
to specific questions, rather than attempting to provide
hard and fast rules, in the hope of enabling designers to
make better-informed final decisions for their projects. FIGURE 1 A bold statement in
Reference 1 and the anticipated Reference 2 modern highway bridge design, steel
also provide extensive, current guidance on the topic of tub girders offer advantages in span
steel tub girder design. range, ability to accommodate
curvature, and aesthetics. IH 35/US
290 Interchange, Austin, Texas.
STEEL TUB GIRDER APPLICATION ISSUES
There are many reasons to choose steel tub girders—which offer several distinct advantages over
steel plate girders and other superstructure types—for a project. However, steel tub girders are
not a panacea. Designers should carefully consider each bridge on a case-by-case basis to
determine if steel tub girders are an appropriate superstructure choice. Several keys issues to
evaluate include:
• Span ranges—Steel tub girders generally are economical in middle span ranges, from
about 150' to 500'.
• Curvature—As closed cell structures, the torsional strength and rigidity of tub girders
make them excellent choices for curved structures with tight radii of curvature.
• Aesthetics—Their smooth uncluttered form and the reduced number of girders give steel
tub girder bridges a clean, simple appearance.

TRB 2006 Annual Meeting CD-ROM Paper revised from original submittal.
 Coletti, Fan, Holt, Vogel 4

• Durability/Maintainability—Since many elements of tub girders are located inside the

box section, protected from the elements, they exhibit greater durability.
• Economy—Tub girders are not inexpensive, but in appropriate applications offer
economically competitive solutions to tough design situations. Furthermore, as the use of tub
girders increases, their unit cost should decrease, making them more cost effective for a wider
range of situations.
References 1, 3, 4, 5 and 6 offer good discussions on many of these key issues.

DEPTH, WIDTH AND SPACING OF TUB GIRDERS

Depth-to-Span Ratios
The traditional rule of thumb for steel bridge girder depth of L/25 (with L measured between
points of contraflexure) is a good starting point for steel tub girders. The L/25 guideline is
mentioned in Section 12 of the latest version of the AASHTO Guide Specifications for
Horizontally Curved Steel Girder Highway Bridges (7) and Section 2.5.2.6.3 of the AASHTO
LRFD Bridge Design Specifications (8). However, designers should not be afraid to exceed this
ratio; tangent steel tub girders have approached L/35 and still met all code requirements for
strength and deflection control. Keep in mind, the minimum recommended web depth for tub
girders is 5', as cited by several writers (4, 9). Note that steel tub girder depth affects more than
just commonly considered implications, such as vertical clearance below a bridge. Because most
tub girders use sloped webs (typically sloped at 1H:4V), web depth also directly affects bottom
flange width (discussed later).

Tub Girder Width and Spacing

As a lower bound, at least one writer suggested a minimum width of 4' be maintained for tub
girders (9) to allow sufficient room for workers to complete fabrication of various internal girder
details. As an upper bound, there are no hard and fast rules. Upper limits on tub girder width
(and tub girder spacing) often are controlled indirectly by three central issues:
• Bottom Flange Width—Extremely wide bottom flanges likely will require the addition of
longitudinal stiffeners and/or transverse stiffeners for stability (see further discussion below).
• Deck Transverse Span—Wide girder spacings can offer some significant opportunities
for taking advantage of some of the inherent efficiencies of tub girders and reduce total steel
weight, reduce construction efforts, and lower initial construction costs. However, some owners
(e.g., state departments of transportation) are uncomfortable with wide girder spacings, both
within each tub girder and between girders. One concern is constructability. Wider deck spans
are more difficult to form, and some owners have very economical deck details, including
permanent metal deck forms or stay-in-place precast concrete deck panels, which may not be
suited for wider girder widths and spacings. Another issue is the ability of wide-span decks to
accommodate certain overload conditions. Finally, redecking while maintaining traffic may be
quite difficult on bridges with wider girder widths and spacings.
• Transportation Limits—Note that tub girder widths above approximately 12' present
transportation difficulties. Widths above 14' may require provisions for a longitudinal bottom
flange splice so the tub can be transported in two sections.
The ratio of the width of an individual tub girder to the spacing between tub girders also
should be considered in accordance with Article 10.39.1.1 of the AASHTO Standard
Specifications for Highway Bridges (10) or with Article 6.11.2.3 of the AASHTO LRFD Bridge
Design Specifications (8). However, exceeding these limits may be acceptable, provided a more

TRB 2006 Annual Meeting CD-ROM Paper revised from original submittal.
 Coletti, Fan, Holt, Vogel 5

refined analysis that more rigorously quantifies live load distribution is performed. Designers
should take care to consider all implications if they choose to exceed these limits.
Overall width of a bridge also is a controlling factor in setting individual tub girder
widths and spacings. In many narrower bridges carrying only one lane of traffic plus shoulders
(such as direct connector ramp structures in multilevel interchanges), total out-to-out deck width
may be as narrow as 28' or less. In longer spans requiring deep girders, two tub girders with
webs sloped at 1H:4V may end up with overly narrow bottom flange widths. Single tub girder
bridges have been successfully used in these situations, but beware that using a single tub girder
may result in an excessively wide bottom flange as well as design, detailing and construction
complexities associated with providing two bearings per tub girder. Also, during construction,
care must be taken to ensure that a single tub girder is sufficiently braced at all stages of erection
to avoid the potential for an unstable system. See Reference 1 for more information.
Designers are advised to evaluate these considerations early on, before even preliminary
structural design calculations are performed.

ANALYSIS AND DESIGN METHODS

Final design of steel tub girder bridges is a detailed and intensive process. Care taken during the
preliminary design phase will pay dividends many times over later on. Thorough consideration
of design, detailing and construction issues up front will result in better, more efficient and
easier-to-construct bridges. Before beginning to run numbers, designers are advised to consider
numerous framing plan issues discussed later.
Full discussion of each level of analysis mentioned below is beyond the scope of this
paper. The reader is instead directed to References 1 and 11 for more detailed treatment of this
topic.

Tangent Tub Girder Design Techniques and Tools

Straight bridges containing multiple tangent tub girders can be designed using a single girder
model. Design can be facilitated using any of several commonly available tangent girder
programs, some of which are developed specifically for tub girder design. Others normally are
only applicable to plate girder or rolled beam design, but can provide sufficient design
information to shortcut some tedious analysis steps.
For shear design, designers must remember that all tub girders carry torsion, which
increases effective design shear in one web. If a single girder model is used, torsional moments
can be approximated by hand. If a multiple girder computer analysis model is used, torsional
moments should be available from the analysis and additional web shears can be derived.
In addition, because most tub girder webs are sloped, the designer must account for the
increase in resultant web shear due to web slope, as well as the increase in web depth along the
slope.

Approximate/Preliminary Curved Girder Design

In slightly curved bridges, it is helpful to know that using the straight girder methods previously
mentioned in this paper would result in a design very close to the final design where torsion is
considered and resulting cross-section sizes are good candidates for the trial member sizes in the
final design model. In cases where the curvature is more severe and would affect primary
bending moments, an approximate analysis can still be performed using straight girder methods
as the basis, but with the straight girder results amplified using factors to account for curvature

TRB 2006 Annual Meeting CD-ROM Paper revised from original submittal.
 Coletti, Fan, Holt, Vogel 6

found in Reference 12. When analyzing a curved bridge using tangent girder methods, developed
span lengths of the tub girders should be used.

Effects of Curvature, M/R Method

Designers must remember that tangent girder programs do not account for load shifting and
torsional effects caused by horizontal curvature. It is up to designers to account for these effects
in order to correctly quantify design loads on a tub girder.
In general terms, an overturning moment (moment about the longitudinal axis of the
bridge) occurs in curved bridges, since the centers of gravity of each girder, and of the bridge as
a whole, are offset from a chord line drawn between support points of each span. When girders
act independently of each other (for example, self-load of an individual girder, or deck placement
when no intermediate external diaphragms are provided), this effect results in torsion in each tub
girder. These torsional moments can be determined using the M/R Method, an approximate, hand
calculation method for torsional analysis of single curved tub girder bridges developed by Tung
and Fountain (5, 12 and 13).
The M/R Method can be applied to effectively calculate the torsional moment in single
girders or in girders under construction before adjacent tub girders are tied together by
intermediate diaphragms and slab. Because cross-sectional rotations also can be obtained using
the M/R Method, it is possible that the method can be applied to multiple girders connected by
intermediate diaphragms to solve the load shifting by iterations. However, the process would be
too complicated without additional computational aids. More effective methods to consider the
interaction of the girders are grid analysis and Three-Dimensional Finite Element Analysis (3-D
FEA).

Grid Analysis
Multiple girders and diaphragms can be modeled in a grid analysis; most commercial programs
for curved tub girder design are based on this method. A two-dimensional grid is used to define
the overall geometry, with line elements used to model the girders and diaphragms. Most
programs distribute live loads to girders using the lever rule.
Grid analysis is popular because building a grid model is relatively simple and fast.
However, designers are cautioned that the many simplifying assumptions required to model a
complex structure in a 2-D grid model requires experience and good judgment. Also, grid
analysis does not directly account for all related deformations, including bending, shear, torsion,
axial deformation and cross-sectional distortion.
For final design of curved or severely skewed tub girders, a grid analysis is the minimum
level of analysis that should be considered.

3-D FEA
3-D FEA is a method in which girder plates are modeled using plate/shell elements, while a
concrete slab is modeled using either plate/shell elements or solid brick elements. Bracing
members also are included in the analysis, usually modeled with either truss or beam elements.
3-D FEA methods are intended to create highly inclusive and accurate models that can
replace all other analysis methods. However, building, running and post-processing a 3-D FEA
model requires significant effort and may result only in marginal improvement in refinement of
the analysis.

TRB 2006 Annual Meeting CD-ROM Paper revised from original submittal.
 Coletti, Fan, Holt, Vogel 7

Three-dimensional analysis programs and services exist that include code-checking

procedures for main girders. These are powerful analysis tools if used properly; however, design
procedures still require other supplemental structural analysis, as demonstrated in the design
example in the current AASHTO Guide Specifications for Horizontally Curved Steel Girder
Highway Bridges (7).

GIRDER DESIGN
The girder design process for tub girders should begin with development of a framing plan that
includes consideration of all of the many parts of a tub girder (Figure 2). Many decisions made
early in the design process can have significant impacts later on; some basic issues are presented
below. Structural analysis as discussed above must be performed to derive the load effects,
which are subsequently checked with the code provisions.
Tub girder bridges must always be designed considering construction sequence. Some
structural members, such as lateral bracing, are provided only for construction purposes. Their
design is consequently controlled by
construction loading. The analysis must
therefore be performed on the partially
completed structure, simulating the
sequence of construction loading. Total
stresses are the sum of those generated
due to loads on the complete bridge and
those locked-in during construction.
Numerous guides and design
examples are available that specifically
discuss girder design and code
provisions in detail (3, 5, 6, 7, 10, 14
and 15).
Designers are cautioned that
narrow tub girder bridges with only one FIGURE 2 The various parts of a typical tub girder.
or two girders in their cross-section may
be considered fracture-critical. However, recent revisions to the AASHTO LRFD Bridge Design
Specifications (8) allow for demonstration of redundancy by means of rigorous analysis.

Bottom Flanges
Bottom flange thickness and b/t ratios may seem like minor details better left to final design.
However, designers of steel tub girders are well-advised to consider this issue up front while
developing a framing plan, because many potentially complex and costly ramifications may
result from the simple choice of bottom flange width and thickness.
The AASHTO Standard Specifications for Highway Bridges (10) suggest that for b/t
ratios greater than 45, “longitudinal stiffeners be considered” and that b/t ratios greater than 60
are not permitted for compression flanges. The AASHTO LRFD Bridge Design Specifications
(8) replace these provisions with more general guidance in Article 6.11.8.2.3 and its associated
commentary. The b/t limit of 45 for consideration of longitudinal stiffeners is intended to be a
rule of thumb; with a b/t above 45, it may be more economical to add a longitudinal stiffener to
the bottom flange to increase bottom flange capacity without thickening the flange. Because they
are costly additions and their use may result in undesirable fatigue details, careful consideration

TRB 2006 Annual Meeting CD-ROM Paper revised from original submittal.
 Coletti, Fan, Holt, Vogel 8

should be given before adding longitudinal stiffeners. Engineers often find it more economical to
simply thicken the bottom flange in lieu of using longitudinal stiffeners.
Even in positive moment regions, there are lower bound limits for bottom (tension)
flange thickness. As stated in Preferred Practices for Steel Bridge Design, Fabrication, and
Erection prepared by the Texas Steel Quality Council (16):
“For wide bottom flanges, plate distortion during fabrication and erection can be a
problem. Designers should check with fabricators when using bottom tension flange plates less
than 1" thick to determine whether practical stiffness needs are met. In no case should bottom
tension flanges be less than 1/2" thick. Another suggested guideline is that the bottom tension
flanges have a b/t ratio of 80 or less.”
Other writers have suggested a maximum b/t ratio of 120 for bottom flanges in tension
(17).
For extremely wide and/or slender bottom flanges, transverse stiffeners may be required.
Bottom flange transverse stiffeners serve several purposes, including bracing bottom flange
longitudinal stiffeners and stiffening bottom flanges for torsional shear stresses. Again, care
should be taken before adding transverse stiffeners to bottom flanges, since they will be costly
and may result in constructability and fatigue problems if not carefully detailed.
Some detailing guides, such as Preferred Practices for Steel Bridge Design, Fabrication,
and Erection (16), provide more detailed suggestions regarding tub girder bottom flange
thickness and b/t ratios. These issues also are discussed in the Commentary to Section 10.4.2.4 of
the AASHTO Guide Specification for Horizontally Curved Steel Girder Highway Bridges (7).
Designers should carefully review the issues presented in these documents and seek guidance
from local steel bridge fabricators with tub girder experience regarding relative cost issues before
making hasty assumptions regarding bottom flange thickness.

Webs
Although tub girders carry vertical shear similar to plate girders, they also carry torsional shear
stress. Tub girders are very efficient in carrying torsion, so this generally does not present a
significant design challenge. The shear flow can be obtained from the torsional moment using
the formula q = T/2A, where A is the area enclosed by the tub girder webs, flanges and slab (or
lateral bracing, if investigating girder prior to deck hardening). However, designers should
remember that the webs will carry more shear than what might be predicted by an approximate,
tangent girder analysis, and thus increased thickness or additional transverse stiffeners may be
required. Recent experience has shown that providing a reasonable number of transverse
stiffeners is currently more economical than providing either a thinner web with extensive
transverse stiffeners or a thicker web without.
It should be noted that all tub girders have torsion; even tangent tub girders will be
subject to some level of torsion from a variety of causes. Some potential sources of torsion in
tangent (and curved) tub girders include:
• Skew—Skew increases torsion in tub girders, because web span positions relative to
various load points are no longer symmetrical from one web to the next.
• Asymmetrical Non-Composite Loading—External girders in particular can be subject to
asymmetrical loading during deck placement, since overhang widths often are not equal to the
tributary deck width between adjacent girders and at phased construction lines. This effect can be
controlled/reduced by use of intermediate external diaphragms and/or lateral bracing.

TRB 2006 Annual Meeting CD-ROM Paper revised from original submittal.
 Coletti, Fan, Holt, Vogel 9

• Asymmetrical Live Loading—The Commentary for Article 9.7.2.4 of the AASHTO

LRFD Bridge Design Specifications (8) offers a good discussion of this issue. Because tub
girders have very high torsional stiffness, they can develop torsional loading during
asymmetrical application of live load. Plate girders do not experience this phenomenon, since
they are very flexible torsionally and twist or rotate to accommodate the deck behaving as a
simple span. For tub girders, this phenomenon is more important for deck design (it can control
the design of the deck slab), but it is worth noting here as another source of torsion in tangent
and curved tub girders.

Top Flanges
Top flanges of tub girders are designed primarily to carry the girder bending stresses. Additional
longitudinal stresses due to torsion exist, as the flanges also serve as members of the truss system
used for lateral bracing. Top flanges also are subjected to significant lateral bending stresses.
These lateral bending stresses can be generated by horizontal girder curvature (6), sloping webs
(18), and temporary supporting brackets for slab overhangs (7). In addition, forces from lateral
bracing systems may represent a major source of lateral flange bending stresses and should also
be considered in design. See the recent article by Fan and Helwig (18) for a more detailed
discussion.
Design provisions (7) suggest that lateral bending of top flanges can greatly affect the
portion of capacity allocated to bending stresses. Increasing top flange width is generally more
effective for resisting the lateral bending stresses than increasing top flange thickness. Sufficient
top flange width also is necessary to provide room for connection of lateral bracing members.
However, the recommended b/t ratio for top flanges in tension or compression should be
followed carefully.
Critical design stages for top flanges often occur during construction prior to the deck
curing, when the flanges are laterally braced only at the K-frame locations. Lateral bending
stresses due to live load effects can be neglected in the capacity check when top flanges are
embedded in the hardened concrete with shear studs, except in areas where shear studs are not
provided. However, curved tub girders typically are provided with shear studs along the entire
girder length to achieve the desired torsional rigidity from a closed cross-section.
In addition, top flanges also are subjected to erection loads, as most contractors lift steel
girder sections by clamping the top flanges. Ideally, local stresses in the top flanges, including
stresses in the weld between the flanges and web, should be checked for these erection loads.

DIAPHRAGMS AND BRACING

Internal Intermediate Diaphragms
Internal intermediate diaphragms are provided in tub girders to control cross-sectional distortion,
which introduces additional stresses in box girders and should be minimized. Cross-sectional
distortion is caused by torsional loads that do not act on boxes in the same pattern as the St.-
Venant shear flow, which is uniformly distributed around the perimeter of the tub girder cross-
section. Unfortunately, most torsional loads on box girder bridges—such as eccentric vertical
loads and curvature-induced torsion—fall in this category and will introduce distortion. Because
a true 3-D FEA analysis using plate/shell elements will give total stresses induced by bending,
torsion and distortion, a separate distortional analysis is not needed. However, all design
programs based on grid analysis using line elements are unable to predict distortional behavior of
box girders.

TRB 2006 Annual Meeting CD-ROM Paper revised from original submittal.
 Coletti, Fan, Holt, Vogel 10

Distortional stresses can be neglected in design, if a sufficient number of internal

diaphragms with adequate stiffness are provided. Research efforts in the early 1970s resulted in
guidelines for spacing and stiffness requirements of internal diaphragms. While there might be
conceptual deficiencies in these previous investigations, experience in the past three decades has
proven that the guidelines are, in general, conservative. Those studies focused primarily on X-
type cross frames. Most cross frames in modern box girder bridges are K-frames, allowing better
access for construction and inspection. With very narrow boxes, consideration of X-frame or Z-
frame configurations might be warranted. Until more studies on K-frames are available, previous
data for X-type cross frames can be judiciously used for K-frames. It should be noted that the
current AASHTO Guide Specifications for Horizontally Curved Steel Girder Highway Bridges
(7) and AASHTO LRFD Bridge Design Specifications (8) have omitted these previously
published guidelines.
Spacing of internal diaphragms is part of the framing process, and should be determined
by also considering factors such as the angle and length of lateral bracing members, as discussed
below. The stiffness requirement often results in very small sizes of cross frame members, which
typically are structural angles. Members are sometimes selected based on handling
considerations; cross frame members must also be checked to meet strength requirements. Large
brace forces can develop inside cross frames, particularly when they are spaced relatively far
apart in curved bridges. Approximate design equations have recently been developed to estimate
these forces (19). Strength verification should be conducted on diaphragms with controlling
(largest) brace forces, usually located in the largest bending moment regions. The above-
mentioned guides (5, 7, 19, 20 and 21) offer good discussions on sizing and spacing issues.
Internal cross frames also act as braces to prevent girder top flanges from buckling.
However, the flange stability issue has not been well addressed in previous research. Current
design guidelines based on distortional stresses likely will continue to be the primary approach in
years to come.
Internal cross frames should be detailed to minimize fatigue concerns. Stiffeners to which
bracing members are attached should be connected to girder flanges to reduce local distortion.

External Intermediate Diaphragms

It is widely known that intermediate diaphragms in curved I-shaped plate girder bridges are
primary structural members and are always needed. Intermediate diaphragms prevent individual
girders from rotating independently under torsional loads and tie the girders together such that
torques are resisted by a structural system with a large rotation arm and are thus much stiffer
than individual girders. Individual I-shaped plate girders have low torsional stiffness and can
develop significant rotation if not restrained.
Tub girders, on the other hand, possess larger rotational stiffness and are much more
capable of carrying external torques individually. In finished bridges, when tubs are fully closed
and the concrete deck effectively ties the girders together, transverse rotations of tubs are
expected to be small and external intermediate diaphragms typically are not warranted. During
construction, however, the rotational rigidity of tub girders is not as high. Steel tubs at this stage
are quasi-closed, and their rotational behavior is not fully understood or predictable. Research
recently performed at the University of Texas will provide more insight into this behavior;
publication is expected by fall 2005 (2). The main reason for concern is the consequence of
transverse rotations. Because a girder’s two top flanges are spaced apart but still rotate together,

TRB 2006 Annual Meeting CD-ROM Paper revised from original submittal.
 Coletti, Fan, Holt, Vogel 11

resulting differential deflections at the top flanges would be large, even with a small girder
rotation. Reasonable fit-up may become impossible and differential slab thickness will result at
the top flange locations if girder rotation is not controlled.
Consequently, external intermediate diaphragms are used to control differential
displacement and rotation of individual tub girders during slab placement. In tangent tub girders,
external intermediate diaphragms can sometimes be omitted or minimized if individual tub
rotation and differential displacements are expected to be relatively small and manageable.
External intermediate diaphragms often utilize a K-frame configuration, with depth closely
matching girder depth for efficiency and simplification of supporting details.
The necessity of external diaphragms has been a debated topic for years, and tub girder
bridges have been built with and without external diaphragms. Problems with slab placement
have occurred on some curved girders without external intermediate diaphragms.
Many previous and current research investigations report that external intermediate
diaphragm forces are small and concluded that they are not needed. Yet girder rotations are
frequently observed in curved bridges. For example, bolt holes do not align during installation.
Many engineers and contractors tend to be conservative with this issue and like the additional
stiffness and rotation control that external intermediate diaphragms bring to structures. It is likely
they will continue to be used unless strong evidence suggests otherwise, or a convenient, reliable
tool becomes available to predict girder rotation.
There also is ongoing debate regarding external intermediate diaphragm removal after the
deck has cured. Some advocate removal of these diaphragms for aesthetic reasons. However,
care should be taken in evaluating the effects of removing "temporary" external intermediate
diaphragms; at a minimum, five issues should be addressed:
• Safety—Removing temporary external intermediate diaphragms can be difficult and
potentially hazardous, due to falling debris concerns or if members are carrying significant load.
Crane access is limited once the deck is in place.
• Deck Stresses—Loads carried by temporary external intermediate diaphragms will shift
to the deck when the diaphragms are removed, adding to deck stresses and potentially leading to
deck cracking if this effect is not carefully evaluated in the deck design.
• Future Redecking—In locations where future redecking of the bridge is likely (for
example, to address deck deterioration in regions where the heavy use of deicing salts is routine),
consideration should be given to retaining external intermediate diaphragms, since they would
likely be required during redecking for the same reasons they were required during the original
construction.
• Traffic Control—If lane closures are required to remove external diaphragms, salvage
value of removed diaphragms will be far less than costs to close lanes.
• Fatigue—Leaving external diaphragms in place may present fatigue concerns, since
connection of diaphragm bottom chords to the girder bottom flanges is problematic and
connection to the web with stiffener backups may result in a poor fatigue detail.
In Texas, external K-frames currently are provided every two to three internal K-frames,
and typically are removed after slab construction.

Internal Diaphragms at Supports

Pier and end diaphragms generally are full-depth plate girder sections. Particular care should be
taken in detailing end diaphragms for constructability, because the presence of abutment
backwalls, other girders or pier cap stems will limit access to one side of the diaphragm during

TRB 2006 Annual Meeting CD-ROM Paper revised from original submittal.
 Coletti, Fan, Holt, Vogel 12

erection. Note that pier diaphragms (and sometimes end diaphragms) require access manholes
for future inspection.
Internal diaphragms at supports are designed as deep beams subjected to 1) bending
loads, which are the shear forces from the girder webs; and 2) torsion-induced shear flow around
the perimeter of the diaphragms, due to torsional moment reactions on box girders.
Internal diaphragms typically consist of a vertical plate. Top flanges can be provided for
the diaphragm to increase the bending capacity of the diaphragms, as pointed out above, and also
can provide support for the slab at girder ends. Bearing stiffeners can be attached to diaphragms
and are designed as columns subjected to an axial load equal to the reaction. Access holes should
be provided for inspection purposes; in determining the location and size of the holes, it is
important to consider the stress flow due to bending and torsional moments.
Diaphragms are supported by one or two bearings. Two-bearing supports provide better
torsional resistance and induce less bending stress in the internal diaphragms. However, two-
bearing supports are not often recommended, due to width limitations of bottom flanges and high
demand for construction accuracy. Single supports are more widely used, and the torsional
resistance—as well as the distributional reactions—resulting from the large bearing contact area
often is conservatively neglected. Bearing design is discussed in more detail in Reference 1.

External Diaphragms at Supports

If dual bearings or other measures (such as anchor bolts or shims) under the girder are able to
prevent transverse rotation, external diaphragms could, theoretically, be stress-free. If single-
point support is used, however, torsional rotations at the ends of the girders must be resisted by
external diaphragms.
Large torsional reactions may occur at the girder supports, leading to the use of solid
plate girders for diaphragms in many curved bridges. These solid plate girder diaphragms are
subjected to bending and shear as torsion in individual girders is resolved via a force-couple
between the supports of adjacent girders. The bending moments and shear forces in the
diaphragms can be determined by assuming that external and internal diaphragms are combined
to form a beam supported at the bearing locations. The loads on the beam are moments equal to
the resisting torques for girders.
Since these moments are acting at the supports of the beam while the middle span
portions of the beam represent the external diaphragms, this approach typically yields bending
moments on external diaphragms that are smaller than resisting torsional moments on the girders.
However, the accuracy of this model could be compromised by potential eccentricities of the
bearing supports and other uncertainties. Alternately, it is reasonable and usually conservative to
directly use the torsional reactions of the box girders as the design moments for the external
diaphragms. Shears on the diaphragms can be evaluated using the same model (1). Torsional
reactions at girder supports should be available from global bridge analysis results. The largest
torsion is very likely to occur during the construction stage.
The loads in these diaphragms may suggest the use of a moment connection to the girder
(where the diaphragm flanges are connected to the girder); the connection may be designed in a
similar manner to the splice design for the girder.
Torsional moments in straight girders are significantly smaller, so a plate girder may not
be warranted for external diaphragms; a truss-type diaphragm may be sufficient in these cases.

TRB 2006 Annual Meeting CD-ROM Paper revised from original submittal.
 Coletti, Fan, Holt, Vogel 13

Avoid skews in tub girder bridges if at all possible—especially at girder ends—since

skewed end diaphragms are enormously complicated to detail, fabricate and erect. Pier
diaphragms can be detailed as pairs of perpendicular diaphragms to avoid problems; note
however that end moments in these types of paired perpendicular diaphragms may be high since
they may be subject to significant racking due to differential deflection of adjacent girders along
the skew.

Top Flange Lateral Bracing

Top flange lateral bracing is required in tub girders to form the “fourth side” of the box until the
slab is in place. The previously mentioned guides (5, 7, 18, 20 and 21) offer good discussions on
sizing and spacing issues. This bracing typically is formed with WT or angle sections, often
configured in a single diagonal (Warren truss) or double diagonal (X-bracing truss) arrangement.
The purpose of providing a lateral bracing system is to increase the torsional stiffness of
tub girders. However, the relationship between torsional stiffness and lateral bracing systems
(including connection) has not been well studied. Analytical formulas are available to calculate
the thickness of an equivalent plate converted from the horizontal truss, although there is no
experimental verification on the stiffness aspect of this method. Rotation of tub girders also is
likely to occur if bolted connections of the lateral bracing members experience slip. A better
understanding of the quasi-closed tub girder stiffness would impact the analytical method and the
use of external diaphragms.
Design of a top lateral bracing system consists of three major steps:
1. Framing—The angle between diagonal lateral bracing members and the top flange, , is
an important design parameter. Ideally, an around 45 degrees is desired. A small leads to
larger cross frame spacing and fewer bracing panels and connections. However, larger brace
forces and girder stresses result from a small . Engineers should plan the framing using both
structural and economical considerations.
Diagonal lateral bracing members are framed into the intersection of the girder top flange and
internal diaphragm. For narrow girders, the guidelines discussed earlier may suggest a cross
frame spacing much larger than the girder width, which would result in a small . Therefore,
location of the internal diaphragms should be planned with consideration of lateral bracing.
An alterative is to divide the spacing between two consecutive cross frames into more than
one lateral bracing panel. Single lateral members (typically angle sections) are used at those
dividing lines. Attaching two more members to the single lateral to form a vertical K-frame does
not significantly increase steel weight, but does add cost, due to increased fabrication. Note that
doing so also makes the structure stiffer, and brace forces in both the lateral bracing members
and internal diaphragms may be slightly larger.
It is suggested to keep the plane of the top flange lateral bracing reasonably close to the
plane of the top flanges, which increases the torsional efficiency of tub girders and avoids
excessive bending loads in web stiffeners.
2. Analysis—Torsional moments induced by dead loads and construction loads will result in
brace forces in lateral bracing members. These forces can be derived from the St.-Venant shear
flow at girder cross-sections, assuming the horizontal truss acts as an imaginary plate.
Lateral brace members, transverse struts in internal diaphragms, and top flanges form a
geometrically stable horizontal truss. It is therefore possible to take axial forces representing the
top flange component of girder bending moments and calculate member forces in the lateral

TRB 2006 Annual Meeting CD-ROM Paper revised from original submittal.
 Coletti, Fan, Holt, Vogel 14

bracing due to box girder bending. Design equations have been developed to evaluate this
component of force for different truss types (18).
Vertical loads on the top flanges also induce brace forces due to web slope. However, the
majority of these forces are resisted directly by lateral struts. Thus, the total forces in lateral
bracing members are essentially the sum of torsional and bending components. Because bending
forces are a function of member size, trial truss member sizes must be assumed before the force
calculation. External diaphragms also contribute load to the bracing system. If a 3-D FEA
program is available, total forces can be obtained directly from the analysis.
3. Member Size Selection and Connection Detailing—Once forces are determined, brace
members can be designed as beam columns. Depending on the connection details, bending
moments resulting from any eccentricity must be considered. The effective length factor (k) used
in design of brace members should reflect end connection conditions.

MISCELLANEOUS DETAILS
Many other details associated with steel tub girder design have unique aspects. These details
include:
• Flange-to-web welds
• Shear studs
• Bearings
• Bridge decks
• Field splices
• Steel detailing
• Material selection
• Painting
The reader is directed to Reference 1 for in-depth discussions of these details.

CONCLUSION
Steel tub girders often are an excellent choice for modern highway bridge superstructures. They
offer advantages over other superstructure types in terms of span range, stiffness and
durability—particularly in curved bridge applications. Steel tub girders also offer distinct
aesthetic advantages, due to their clean, simple appearance. The increased use of tub girders in
the future could address many demands for improved bridge aesthetics, efficiency, and
durability.
However, the layout, design, detailing, fabrication and erection of steel tub girder bridges
is in many ways different and often more complicated than for steel plate girder bridges.
Designers are advised to understand and evaluate all these issues very early in the design
process.
Designers should also realize that in many cases, there are no hard and fast rules
associated with these issues. Each tub girder project is unique and a solution that worked for one
bridge might not necessarily work for another. Instead, it is imperative that the designer of a tub
girder bridge explore all options on a case-by-case basis, often in consultation with local
fabricators and/or erectors.
A full understanding and careful consideration of all these issues will result in a more
successful tub girder bridge project.

TRB 2006 Annual Meeting CD-ROM Paper revised from original submittal.
 Coletti, Fan, Holt, Vogel 15

REFERENCES
1. Coletti, D.A., Z. Fan, J. Holt, W. Gatti, and J.P. Vogel. Practical Steel Tub Girder Design
National Steel Bridge Alliance, 2005.
2. Yura, J.A., T.A. Helwig, R.H. Herman, and E. Williamson. Trapezoidal Box Girders—State
of the Art, Texas Department of Transportation, Research Report 0-4307-1, 2005
(anticipated).
3. Hedefine, A., J. Swindlehurst, and M. Sen. Structural Steel Designer’s Handbook, Section
12, Beam and Girder Bridges, 3rd Edition, (Brockenbrough, R.L., and F.S. Merritt, Editors),
McGraw Hill, Inc., 1999.
4. Hall, D. H. Why Steel Box Girders? Modern Steel Construction, April 1997.
5. Poellot, W.N. Structural Engineering Handbook, Section 21, Curved Steel Box Girder
Bridges, 4th Edition, (Gaylord, E.H., C.N. Gaylord, and J.E. Stallmeyer, Editors), McGraw-
Hill, Inc., 1997.
6. AISC Marketing, Inc. Highway Structures Design Handbook, ADUSS 88-8535, 1981.
7. American Association of State Highway Transportation Officials (AASHTO), Guide
Specifications for Horizontally Curved Steel Girder Highway Bridges, GHC-4, 2003.
8. American Association of State Highway Transportation Officials (AASHTO). LRFD Bridge
Design Specifications, 3rd Edition, LRFD-3, 2004 with interim revisions through 2005.
9. Kase, R.A. Twelve Commandments for Economic Steel Box Girders, Modern Steel
Construction, August 1997.
10. American Association of State Highway Transportation Officials (AASHTO). Standard
Specifications for Highway Bridges, 17th Edition, HB-17, 2002.
11. Coletti, D.A, and J.M. Yadlosky. Behavior and Analysis of Curved and Skewed Steel
Bridges. 2005 World Steel Bridge Symposium, November 29-December 1, 2005.
12. Richardson, Gordon, and Associates (now Pittsburgh, Pa., office of HDR). Curved Girder
Workshop Lecture Notes, prepared under Contract No. DOT-FH-11-8815, Federal Highway
Administration, 1976.
13. Tung, D.H.H, and R.S. Fountain. Approximate Torsional Analysis of Curved Box Girders by
the M/R Method. AISC Engineering Journal, Vol. 7, No. 3, July 1970.
14. American Iron and Steel Institute. Four LRFD Design Examples of Steel Highway Bridges,
Highway Structures Design Handbook, Vol. II, Chapters 1A and 1B, 1996.
15. American Iron and Steel Institute and National Steel Bridge Alliance. Four LFD Design
Examples of Steel Highway Bridges, Highway Structures Design Handbook, Vol. II, Chapter
2B, 1999.
16. Texas Steel Quality Council. Preferred Practices for Steel Bridge Design, Fabrication, and
Erection, November 2000.
17. Wolchuk, R., and R.M. Mayrbaurl. Proposed Design Specifications for Steel Box Girder
Bridges, Report No. FHWA-TS-80-205, Federal Highway Administration, January 1980.
18. Fan, Z., and T.A. Helwig. Behavior of Steel Box Girders with Top Flange Bracing, ASCE
Journal of Structural Engineering, Vol. 125, No. 8, pp 829-837, August 1999.
19. Fan, Z., and T.A. Helwig. Distortional Loads and Brace Forces in Steel Box Girders, ASCE
Journal of Structural Engineering, Vol. 128, No. 6, pp 710-718, June, 2002.
20. Heins, C.P., and D.H. Hall. Designer’s Guide to Steel Box Girder Bridges, Booklet No. 3500,
Bethlehem Steel, February 1981.
21. United States Steel. Steel/Composite Box-Girder Bridges—A Construction Manual, ADUSS
88-7493-01, December 1978.

TRB 2006 Annual Meeting CD-ROM Paper revised from original submittal.