Connection Details for Adjacent Precast Box Beam Bridges

NCHRP
Project 12-95
Connection Details of Adjacent

Precast Box Beam Bridges
Interim Report I
2/21/2014
TRANSPORTATION RESEARCH BOARD

NAS-NRC
LIMITED USE DOCUMENT
This Work Plan is furnished only for review by members of the

NCHRP project panel and is regarded as fully privileged.
Dissemination of information included herein must be approved by the
NCHRP.
Table of Contents
Table of Contents ........................................................................................................................... i
List of Figures ............................................................................................................................... iii
List of Tables ................................................................................................................................ iv
1. Introduction .............................................................................................................................. 1
2. Project Survey .......................................................................................................................... 2
2.1 Summary of Reponses ........................................................................................................ 2
2.2 Best Practices ...................................................................................................................... 4
2.3 Information for States Using Large Numbers of Adjacent Box Beam Bridges ............... 5
3. Focus Group ........................................................................................................................... 13
4. Literature Search ................................................................................................................... 14
4.1 NCHRP Synthesis 393 ...................................................................................................... 14
4.2 Publications Before 2008 ................................................................................................. 17
4.3 Publications After 2008 .................................................................................................... 23
4.4 Literature Search and Survey Synthesis and Summary .................................................. 30
5. Updated Work Plan ............................................................................................................... 36
5.1 Experimental and Analytical Evaluations ....................................................................... 36
5.1.1 Introduction ............................................................................................................. 36
5.1.2 Collection of Material Properties and Selection of the Joint Material and Keyway
Preparation ............................................................................................................. 38
5.1.3 Crack Development and Resistance Investigation using Small-Scale Testing and FE
Simulations .............................................................................................................. 40
5.1.4 Investigation of Cracking in Bridges with Variable Attributes ............................... 45
5.1.5 Influence of Shear Key Geometry on Keyway Behavior ......................................... 47
5.1.6 Cracking and Resistance Investigation using Full-Scale Testing and FE
Simulations .............................................................................................................. 50
5.1.7 Parametric study ..................................................................................................... 52
5.2 Specification Development ............................................................................................... 53
6. Data Archiving and Sharing Plan......................................................................................... 55
6.1 Background and Significance .......................................................................................... 55
6.2 Expected Data Formats .................................................................................................... 55
6.3 Description of Data Archiving and Quality Assurance Plan .......................................... 55
6.4 Description of Data Sharing Plan .................................................................................... 56
6.5 Schedule for Data Archiving and Public Release of Data .............................................. 56
6.6 Milestones for the Implementation of the Plan ............................................................... 56
i
6.7 Resources and Budget ...................................................................................................... 56
7. Bibliography ........................................................................................................................... 57
Appendix A - Project Survey to State DOTs ............................................................................ 59
Questionnaire ........................................................................................................................... 59
Summary of Responses ............................................................................................................ 61
Connection Details ................................................................................................................... 67
Appendix B - Focus Group......................................................................................................... 70
ii
List of Figures
Figure 1 Basic Keyway Geometries ............................................................................................... 4
Figure 2 States Frequently using Adjacent Box Beam Bridges ...................................................... 6
Figure 3 Basic Shearkey Shapes ................................................................................................... 22
Figure 4 Keyway Geometries for PCI and TxDOT Style-Box Girder Bridges ............................ 23
Figure 5 Typical Michigan Keyway Geometry and Post-tensioning............................................ 24
Figure 6 Common shearkey locations........................................................................................... 25
Figure 7 Common Bearing Pad Details ........................................................................................ 26
Figure 8 Connection Details Proposed by Hanna et al. [2011]..................................................... 28
Figure 9 Connection Details Proposed by Hansen et al. [2012] ................................................... 29
Figure 10 Typical Box Beam Cross-section ................................................................................. 38
Figure 11 Small Scale Specimen Keyway Geometries (all units are inches) ............................... 40
Figure 12 Basic Configuration for Step B Testing ....................................................................... 41
Figure 13 Small Scale Specimen Test Setup During Curing ........................................................ 42
Figure 14 Small Scale Specimen Test Setup during Post-tensioning ........................................... 43
Figure 15 Shear Key Geometries for Medium Scale Specimens .................................................. 48
Figure 16 Medium Scale Specimen Test Setup ............................................................................ 49
Figure 17 Setup for Large Scale Testing ...................................................................................... 52
iii
List of Tables
Table 1 Structural Design and Details [Russell 2009] .................................................................. 15
Table 2 Specifications and Construction Practices [Russell 2009] .............................................. 16
Table 3 Recommended Practices [Russell 2009].......................................................................... 17
Table 4 Design and Construction Attributes ................................................................................. 31
Table 5 Summary of FE Analysis ................................................................................................. 33
Table 6 Summary of Laboratory Testing ...................................................................................... 34
Table 7 Summary of Field Testing ............................................................................................... 35
Table 8 Initial Joint Material Property Testing ............................................................................. 39
Table 9 Keyway Interface Preparation Testing ............................................................................ 39
Table 10 Time Dependent Material Property Testing .................................................................. 39
Table 11 Small Scale Specimens .................................................................................................. 41
Table 12 Step D Parametric Study Matrix .................................................................................... 47
iv
1. Introduction
This report serves as Interim Report 1 for NCHRP Project 12-95 “Connection Details of
Adjacent Precast Box Beam Bridges”. This report documents activities completed during Phase
I of the project including: information collection (project survey, industry survey, focus group,
and literature review), development of proposed analytical and testing programs to investigate
cracking in the longitudinal joints between adjacent beams, and the establishment of a data
archiving and sharing plan. Given the scope of the analytical and experimental programs, the
information of principal interest in Interim Report 1 is the information collection and the
proposed methodologies for the remainder of the project.
1
2. Project Survey
2.1 Summary of Reponses
A web-based survey was distributed to the U.S. Departments of Transportation to gather
information about their experience with the performance of adjacent precast box girders and to
collect specific information on details and specifications used by these owners. Twenty eight
responses were received of which twenty indicated that they had experience with precast box
girder bridges.
In addition to contact information, agencies were asked to identify the type of keyway they used,
to rate the cracking and leaking performance of the joint between box girders, to describe their
greatest performance problem, and, based on their experience, to identify the best keyway
construction practices (keyway preparation, grout material, transverse post-tensioning, and cast-
in-place deck). The questionnaire used in the survey is given in Appendix A - Project Survey to
State DOTs. State-by-state responses are presented in Appendix A - Project Survey to State
DOTs. A few states provided links to plan drawings that illustrate some of the details they use
for box girder connections, as shown in Appendix A - Project Survey to State DOTs. A
summary of the responses is given below.
Typical keyway detail usage (see Figure 1):
 Ten states use keyway detail Type III with some variations in dimensions and with some
reporting problems with performance.
 No states use keyway details Type I and II, five use keyway detail Type IV, and one uses
keyway detail Type V.
Cracking and leaking performance rating for keyway detail Types I, II, III, IV and V (1 is no
leakage, 5 is major cracking and excessive leakage):
 None of the details performed to the owner’s expectations.
 Type IV and V (the best rated) averaged a score of about 2.5; Type I, II, III averaged
around 3.5 (note that no states reported actually using Types I or II. Thus, these scores
may be based upon prior/older details or speculated performance).
Performance related to cracking and leaking:
 Three states reported that they had no particular performance issues (Vermont,
Minnesota, and Texas). Follow-up phone discussions were conducted with these three
states to gain more insight. A brief summary for each is as follows:
o Vermont
 Uses a 5 in., double reinforced, composite, cast-in-place structural deck
and attributes the lack of performance issues to its usage.
 Bridges have been in-service for approximately 15 years.
 Recently decreased the spacing of post-tensioning locations to
approximately 20 ft.
o Texas
 Uses a 5-in., double reinforced, composite, cast-in-place structural deck
and what was described as a “relatively substantial keyway” and attributes
the lack of performance issues to these details.
 Basically see no water intrusion after changing to their current concepts.
2
o Minnesota
 Opinion offered in the survey was based upon a relatively small number of
bridges, on local roads, with no/minimal salt application that have been in
service for only “a few” years.
 Details are based upon a scan of bridges in New York and, to a lesser
extent, Wisconsin.
 Post-tensioning is used at the quarter points for most bridges.
 Utilize a full-depth shear key.
 Several others noted that seepage leads to corrosion of reinforcement and prestressing
strands that can eventually cause deck cracking and beam damage.
 Reflective cracking and rocking and uneven bearing where also cited as problems.
 Some individual states briefly described various attempts they had undertaken to solve
their problem, such as:
o Tar/roofing topcoat with asphalt overlay (did not seem to work very well)
o Waterproof membrane with asphalt overlay (membrane often failed)
o “Some” post tensioning (helped somewhat)
o Poly-urea (helpful but did not totally stop leakage)
o Cast-in-place overlay (worked in some cases)
o Post tensioning with steel plate connectors (no assessment of performance given)
Keyway preparation:
 Respondents were about equally split between no preparation, power-washed rough
preparation, and mechanical rough preparation.
Grout materials:
 Nine states use non-shrink grout.
 Five states use mortar, epoxy grout or resin.
Concrete topping:
 Only two states said they used a concrete topping (assumed to be a thin unreinforced
concrete as opposed to a cast-in-place deck).
Post-tensioning:
 Three states do not use transverse post-tensioning.
 Two states indicated that they use transverse tie rods that are installed snug tight.
 The remainder of the respondents were equally split between single stage and multi-stage
post-tensioning.
Cast-in-place deck:
 Only five states do not use a cast-in-place deck.
Other points of interest include:
 North Carolina has found two mats of reinforcement in the cast-in-place deck perform
better than one.
3
 Minnesota places post-tensioning at the ends of the beams and at the quarter points
(grouted tubes).
 Massachusetts does an initial post-tensioning of 5,000 pounds, then grouts, then performs
final post-tensioning of the system.
 Wisconsin experimented with a single layer of reinforcement in decks. There is an
ongoing research program at Western Michigan University.
Type I Type II Type III
Wide and deep longitudinal joint

Narrow longitudinal (Cast-in-place concrete)
joint
Type IV Type V
Figure 1 Basic Keyway Geometries
2.2 Best Practices

Of the responses received from the survey, the following summary seems to be a consensus of
current best practices (i.e., what works best and what doesn’t seem to work very well). The
items identified with an asterisk (*) were also listed as practices that affect the likelihood of
longitudinal cracking in box girder bridges in the NCHRP Synthesis report [Russell 2009].
What works:
 Type IV or V full depth keyway*
 Cast-in-place deck*
 Post tensioned system*
 Keyway roughening*
 Non-shrink grout or mortar*
4
What doesn’t work:
 No keyway prep*
 Waterproof membrane
 No cast-in-place deck*
 Asphalt overlay*
 Type I, II and III keyway*
Inconclusive:
 Power-washed vs. mechanical rough keyway preparation
 Single stage vs. multi-stage post tensioning
 Required post tensioning amount and location
 Single layer vs. two layer deck reinforcement
 Steel plate shear connectors
 Epoxy grout
2.3 Information for States Using Large Numbers of Adjacent Box Beam Bridges
Based on the National Bridge Inventory (NBI) data from 2002-2012, the ten states having the
most adjacent box beam bridges are shown in Figure 2(a) and the ten states having constructed
the most adjacent box beam bridges in the last ten years are shown in Figure 2(b). From these
figures one can clearly see that those that have had high numbers of adjacent box beam bridges
in the past seem to continue to build them now. Also of interest is that the majority of these
bridges tend to be in locations that experience snow (and thus likely have deicing material
application). For easy reference the specific survey responses for the states in these top ten lists
that responded to the survey are given in the following pages. Some interesting trends (and lack
of consistency in some cases) exist when one examines the responses collectively. For example,
high-use respondents tend to:
 Use the Type III joint (one used Type IV)
 Believe that the Type IV or V would perform the best
 Have a wide variety of concerns, including:
o Shorter design lives
o Cracking due to bearing rocking
o Crack development even before the bridge is open to traffic
Also, there is very little consistency in opinions as to what leads to good performance with one
exception – the use of a cast-in-place deck.
5
10,000
Total Bridges up to 2012
9,000
8,000
Numver of Adjacent Box Beam Bridges
7,000
6,000
5,000
4,000
3,000
2,000
1,000
Michigan
California
Indiana
Pennsylvania
Ohio
Kentucky
West Virginia
Illinois
Texas
New York
State in US
(a) Total number of adjacent box beam bridges
1,600
Constructed in the Last Ten Years
1,400
Numver of Adjacent Box Beam Bridges
1,200
1,000
800
600
400
200
0
New York
Missouri
West Virginia
Ohio
Michigan
Indiana
Illinois
Kentuchy
California
Texas
State in US
(b) Number of adjacent box beam bridges constructed in the last ten years
Figure 2 States Frequently using Adjacent Box Beam Bridges
6
Illinois has 26,514 total bridges, has 8,775 adjacent box beam bridges, and has constructed 1,060
adjacent box beam bridges in the last 10 years.
Q1. Does your agency use adjacent precast prestressed concrete box beams for bridges?
Response: Yes
Q2. Five generic types of common keyway geometries are illustrated below. Do you currently
use a keyway geometry similar to the five generic types? If yes, please describe the keyway
detail or provide an internet link where we may obtain detail information.
Response: we use the Type III version shown below, we have 2 versions one for shallow
beams and one for deep beams. see the following link for base sheet details
http://www.dot.il.gov/cell/PPC_deckbeam.pdf.
Q3. Based upon your agency’s experience, please rate the above geometries in terms of
performance related to cracking and leakage at the longitudinal joints, where “1” represents
no cracking and leakage and “5” represents major cracking and excessive leakage.
Response:
Type I Type II Type III Type IV Type V
5 5 4 2 2
Q4. What is your greatest performance problem related to cracking and seepage at these
joints? What documentation do you have with regard to this problem (data, specifications,
construction practices, etc.)? In addition, do you believe that cracking is related to
environmental conditions and/or policy (such as road salt and/or restriction on use)?
Response: the greatest performance problem would be a shorter design life. No readily
available data, but we have replaced a considerable amount of these type structures in the
past 10 yrs, all originally built in the 1960s-70s the cracks are related to loading and bearing
rocking or uneven bearing.
Q5. Based upon your experiences, which are the best practices for keyway construction as
related to the box beam performance?
Response:
“Mechanical rough”; Mortar, epoxy grout or resin; Multi-stage post-tensioning; Cast-in-
place Deck
7
Ohio has 27,045 total bridges, has 7,139 adjacent box beam bridges, and has constructed 1,349
Response: Yes
Response: Type III
Response:
4 4 4 4 4
Response: ODOT research shows that these cracks occur prior to opening the bridge to
traffic. When we use asphalt wearing surface on top of a waterproofing membrane, on many
bridges the membrane fails and the joint has excessive leaking. The concrete will spall and
the prestressing wire is exposed. NOTE: IN ORDER TO CONTINUE FILLING OUT THIS
FORM, I HAD TO ANSWER QUESTION 4 FOR ALL TYPES. WE DO NOT HAVE
EXPERIENCE WITH ALL TYPES. DISREGARD MY ANSWER FOR TYPE I, II, IV, & V.
YOU SHOULD HAVE GIVEN ME THE OPTION OF N/A.
Response:
No preparation; Non-shrink grout; Single-stage post-tensioning; Cast-in-place Deck
8
Texas has 55,260 total bridges, has 2,554 adjacent box beam bridges, and has constructed 576
Response: Yes
Response: Type III
Response:
5 5 5 5 4
Response: Amount and location of transverse post-tensioning has varied over the years.
Current details appear to have adequately addressed cracking/leakage.
Response: Other preparation; Concrete topping; Single-stage post-tensioning; Cast-in-place
Deck
9
Michigan has 11,000 total bridges, has 2,501 adjacent box beam bridges, and has constructed
462 adjacent box beam bridges in the last 10 years.
Response: Yes
Response: Type III
Response:
3 3 3 3 3
Response: Long term performance.
Response: No preparation; Non-shrink grout; Single-stage post-tensioning; Cast-in-place
Deck
10
New York has 17,420 total bridges, has 1,926 adjacent box beam bridges, and has constructed
Response: Yes
Response: Full depth keyway similar to Type IV: https://www.dot.ny.gov/main/business-
center/engineering/cadd-info/bridge-details-sheets-repostitory-usc/BD-PA7E.pdf
Response:
5 5 5 2 3
Response: Reflective cracking in deck above the keyways. Chlorides (road salt) penetrate the
cracks, eventually reaching the beams. If left unchecked, the result is deck deterioration, and
eventually beam deterioration. These issues are documented in oue bi-anual inspection
reports.
Response: “Mechanical rough”; Other grout; Multi-stage post-tensioning; Cast-in-place
Deck
11
West Virginia has 7,093 total bridges, has 1,803 adjacent box beam bridges, and has constructed
Response: Yes
Response: 3/4" opening at top flares out to 1-1/2" similar to Type III. The keyway goes about
8" to 12" down from top of beam depending on the depth of beam.
Response:
3 3 3 3 3
Response: truck loads and salt exposure.
Response: No preparation; Non-shrink grout; Single-stage post-tensioning
12
3. Focus Group
To brainstorm possible techniques for improving the performance of adjacent box beam bridges,
a focus group meeting was convened on September 27, 2013 from 11am to 1pm at the Iowa State
University Bridge Engineering Center. Those attending the meeting represented consulting
engineers, contractors, precast concrete fabricators, an Iowa DOT engineer, and Bridge
Engineering Center personnel. During the focus group meeting, background information,
historical performance issues, current practices and the objective of the project were introduced
and then the local stakeholders were encouraged to talk about experiences/success/failures
designing, constructing, and maintaining adjacent box beam systems and brainstorm ways to
improve the design and construction of the systems. General questions such as “what works?”,
“what does not work?”, and “how could it be improved?” were asked about topics such as
keyways, bearings, grout timing, grout material, post-tensioning, use of toppings, etc. In
addition, the attendees were encouraged to share any experiences, ideas, and suggestions on any
topics related to the subject of this project. A summary of the Group discussion pertinent to this
project follows:
 Keyway geometries vary widely. A more uniform keyway geometry would be more cost
effective for all stakeholders.
 There seems to be three choices for keyway grout--high strength grout, concrete, high
strength concrete.
o Grout is expensive in large volumes.
o The availability of high strength/high performance concrete can be an issue in
some rural locations.
o The size of the keyway may impact the material of choice.
 Transverse post-tensioning procedures vary widely.
o Force levels vary (e.g., 30 to 120 kips)
o Longitudinal locations vary (e.g., four per 50 ft)
o At least two sequence procedures are used: (1) single application after grouting
(2) initial application for alignment, then grouting, then final application
o It is easier to install post-tensioning when it is oriented parallel to skew.
o Strands are easier to work with for crown and misalignment but can be more
expensive because of the volume that must be purchased vs. what is actually used.
 Topping/overlay
o It may be necessary on paved road for salt protection.
o Grinding of topping may compensate for differential camber.
Other items that were discussed but are not directly related to this project included tolerances,
camber, railroad bridge applications, web-to-web shear connectors, match-cast keyways, use of
embedded plate with field weld connection, and the concept of letting water flow through rather
than trying to achieve a leak-tight system.
13
4. Literature Search
A comprehensive literature search was conducted to collect information relevant to the project.
These data were gathered, categorized, and summarized so as to guide subsequent tasks. It is
worth noting that a complete understanding of the current state-of-art and the state-of-practice is
extremely important and invaluable at finalizing the plans for the analytical and experimental
investigations of this project and ensuring that the work completed in this project does not
duplicate work already completed. It should be noted that NCHRP Synthesis 393 by Russell
[2009] described current concrete box beam practices from multiple Departments of
Transportation (DOT) at multiple levels and also provides extensive literature search results from
before 2008.
The literature search for this project will be summarized as follows. First, NCHRP Synthesis 393
will be summarized including the conclusions and recommendations. Second, literature
published before 2008 will be reviewed to take note of the important information beneficial to
this project. Finally, literature published after 2008 will be reviewed especially those with a
connection to the results of NCHRP Synthesis 393. Finally, literature will be summarized,
synthesized, and then categorized as they relate to laboratory testing, field testing, and Finite
Element (FE) analysis. Note that to provide a brief summary of each piece of literature, “Take
Away” points for each are provided after each general summary (with the exception of NCHRP
Synthesis 393).
4.1 NCHRP Synthesis 393
The NCHRP Synthesis report by Russell [2009] summarized the observed types of distress
associated with the joints used in adjacent box girder bridge systems including longitudinal
cracking along the joint material and box beam interface, water and salt leakage through the
joint, cracking within the grout, spalling of the grout, spalling of the girder corners, differential
vertical movement, corrosion of transverse ties and longitudinal prestressing strands, freeze-thaw
damage to the grout and concrete near the joint. Note that the most common types of distress are
longitudinal cracking along the grout and box beam interface, water and salt leakage through the
joint, and reflective cracks that are commonly observed in the road surface.
Based on the survey of state DOTs and the literature search, Russell [2009] also began the
process of identifying factors impacting the long-term performance of adjacent box beam bridge
systems. In the synthesis, practices for structural design and detailing for adjacent box girder
bridges from state DOTs and the literature were summarized as shown in Table 1. Specifications
and construction practices for adjacent box girder bridges from state DOTs and the literature
were also summarized as tabulated in Table 2. Finally, the recommended and not-recommended
design and construction practices were summarized as tabulated in Table 3.
Russell [2009] indicates that keyway configurations consist of partial depth and full depth
keyways. In the United States, three typically used generic partial depth keyway geometries are
the Types I, II and III keyways and one generic full depth keyway geometry is the Type IV
keyway as shown in Figure 1 (note that in Figure 1 the box beams have been shown to be in
direct contact – this may or may not always be the case; however, sweep is typically removed
with the application of post-tensioning). Conversely, the typically used Japanese keyway is the
full depth keyway Type V shown in Figure 1. El-Remaily et al. [1996] reported that longitudinal
cracking was seldom found in the adjacent box beam bridges with the Type V full-depth keyway.
14
Table 1 Structural Design and Details [Russell 2009]
Practices Survey summary Literature cited by Russell [2009]
Girder Around 50% of states use
cross AASHTO/PCI-shaped box beams
sections
Span Below 20 ft to above 80 ft 40 to 140 ft [PCI 1997; 2004]
lengths
Bridge skew 0º-60º; Most common: 30º
Composite  Most states use simple spans with The use of a deck does not eliminate differential rotation of girders
deck composite deck (3-9 in. depth); and is not an economically and structurally efficient solution [El-
 Bridges with multi-span and Rmaily et al. 1996]
composite deck are usually designed
continuous for live load.
Keyway Most states use partial depth keyways;  Longitudinal cracks were found in 54% of bridges with 12 in.
geometries some use full depth keyways partial depth keyway and 6 in. depth concrete deck and in 23%
of the bridges with full depth keyways, concrete deck and more
transverse ties [Lall et al. 1997; 1998].
 No longitudinal cracks were found in Japanese bridges with 6 in.
wide full depth keyway, cast-in-place concrete grout and 2-3 in.
concrete or asphalt wearing surface [Remaily et al. 1996]
 The full depth keyway hinders the joint from opening [Miller et
al., 1999]
 Wider full-depth keyways improves the interaction between
adjacent girders and the contact between grout and girders, but
forms are needed to contain the fresh grout during placement
[Nottingham, 1995]
Transverse  Most states use unbonded post-  Illinois DOT equation for the number ties per span (Anderson
ties tensioned strands or bars; some states 2007): N  span  1  1
use bonded post-tensioned strands or 25
bars; other states use non-  Less longitudinal cracking: Three transverse tie locations for the
presteressed reinforcements span less than 50 ft five for the span more than 50 ft [Lall et al.
 The number of tie locations: 1-5 per 1997; 1998]
span  Durable system in Japan: 4-7 evenly spaced transverse
 Most states placed ties at mid-depth diaphragms with post-tensioning ties and post-tensioning is
of girders (one tie per location) determined by flexural design [Yamane et al. 1994]
 Ties are typically placed at the third  Partial depth keyway: Due to eccentricity of post-tensioning,
points when two ties are used at a cracks may be induced by post-tensioning ties at the girder mid-
single location depth
 Full depth keyway: Good with post-tensioning ties at the girder
mid-depth
Post-  Most states specify the required post-  4-14 kip/ft [El-Remaily et al. 1996]
tensioning tensioning force without extensive  7-14 kip/ft [Hanna et al. 2007]
force calculations  27 kip/ft for 15 in. beam depth [Badwan and Liang et al. 2007]
 For 11 states: 0.5-12.5 kip/ft  21 kip/ft per AASHTO LRFD specification [AASHTO 2007;
2008]
 4-11 kip/ft per PCI bridge design manual [PCI 1997; 2004]
 Average of 11 kip/ft is Japanese practice [Yamane et al. 1994]
Exterior  Most of states have the same design  No concrete barriers were used by Illinois DOT for box girder
girders for exterior and interior girders system because of the increased stiffness of exterior girders
might cause increased differential deflections [Macioce et al.
2007]
 The barrier load could be counteracted by the increased exterior
girder section property [Harries 2006].
15
Table 2 Specifications and Construction Practices [Russell 2009]
Practices Survey summary Literature cited by Russell [2009]
Standard No guidelines are provided for the design and
specifications construction of the connection details of adjacent
(AASHTO 2002) box girders
LRFD  A compression depth (≥7 in.) should be
specifications provided with a transverse post-tensioning ≥
(AASHTO 2007; 0.25 ksi
2008)  Post-tensioning ties are required to be placed at
the centerline of the keyway
Bearing types  Plain elastometic bearing: ¾ of respondents
 Laminated elastomeric bearing: ¼ of respondents
 Full-width support or full-point support on ends: 42% of
states for each; Two-point support and one-point
support: the other states
 Uneven seating: half the respondents (especially for a
full-width support)
Construction  One stage construction: Erect all beams and connect Greuel et al. [2000] reported that spalling of
sequence them at one time beam bottom flanges occurred near the shear key
 Two stage construction: a variety of sequences for the two half bridges when the shear key was
 Grout before or after post-tensioning: 50% of states for not grouted prior to post-tensioning
each
 Grout after post-tensioning shows higher cracking
resistance
 Construction sequence is affected by the skew of the
bridge and intermediate diaphragm locations
Differential  Restrictions for differential camber: 1/3 of respondents
camber  Maximum differential camber: 0.5 in. (½ of
respondents)
 Others: 0.25 in. in 10 ft; 0.75 in. maximum; 1 in.
relative deflection for high and low beams in one span
 Improving methods: load high beam before grouting
and post-tensioning; adjust bearing seat elevations;
concrete or asphalt topping; preassemble girders before
shipment
Keyway  Sandblast keyway: 45% of states  Poor adherence of keyway mortar [Attanayake
preparation  Sandblast and powerwash keyway:1/3 of respondents and Aktan 2008]
Grout materials  Nonshrink grout: 40% of respondents; mortar: 25% of  High-quality joint: prepackage mix with
and practices respondents; epoxy grout, epoxy resin, or concrete predetermined amount of water (e.g.,
topping: other respondents prepackaged magnesium-ammonium-
 No curing: 40% of respondents; curing compounds: 5%; phosphate grout with pea gravel) [Nottingham
wet curing: around 45% of states 1995]
 Most of states manually place the grout  Improvements by West Virginia DOT [El-
Remaily et al 1996]: a pourable epoxy
replacing a nonshrink grout; sandblasting
surfaces; post-tensioning ties.
 Andover Dam Bridge in Maine: wider shear
key rapidly grouted with shrinkage-restrained
self-consolidating concrete [Russell 2009]
 Illinois DOT [2008]: use a mechanical mixer
for mixing nonshrink grout; place with a pencil
vibrator; smooth surface; cover with the cotton
mats for more than 7 days
16
Table 3 Recommended Practices [Russell 2009]
Practices Recommended Not recommended
Design practices  Full depth keyway: grouted easily  Non-tensioned transverse ties: no crack
 Post-tensioning transverse ties: eliminating tensile resistant ability
stresses in the shear key
 Cast-in-place reinforced concrete deck (compressive
strength of more than 4 ksi and thickness of more than
5 in.): restrains longitudinal deck cracking
Construction  Form the void using stay-in-place expanded  Use asphalt wearing surface with non-
practices polystyrene water proofing membrane: water
 Sandblast the keyway surface before shipment: gathers under the asphalt
ensuring a better bonding surface for the grout  Use non-prepackaged products for the
 Powerwash the keyway surfaces (compressed air or keyway grout
water) before erection of girders: ensuring a better
surface for the grout
 Grout keyways before post-tensioning: the grout under
compression
 Grout with high bond strength: limit cracking
 Provide suitable curing for the grout: developing
desired strength and minimize shrinkage effects
 Provide suitable wet curing for the concrete deck
(more than 7 days): ensuring durable surface and
minimize shrinkage cracks
4.2 Publications Before 2008

Huckelbridge et al. [1995] revealed that precast prestressed adjacent box beams have been
mostly used for the construction of bridges with short and medium spans ranging from 30 ft to
100 ft. The authors conducted field testing of several adjacent box girder bridges and the test
results from two of these bridges were summarized in the 1995 report; one for a simply
supported bridge and one for a four-span continuous bridge. A dump truck with a front axle
weight of 12 kips and tandem axles weighing 38 kips was used to conduct on-site, controlled
tests. During those tests, deflection transducers were installed on the bottom of adjacent beams
near the keyway so as to record the relative deflections between those box beams; flexural strains
were also measured on the girder bottom. The maximum relative deflection was found to be 0.2
and 0.15 in. for the two bridges, respectively. According to results from FE analysis (details of
FE analysis were not given) and field tests, the authors pointed out that intact shear keys, should
not permit relative deflection of more than 0.001 in. between adjacent girders. As they expected,
reflective cracks were found around the shear keys on both bridges. Partially fractured shear keys
were generally found close to the daily wheel positions and driving lanes with heavy truck
traffic. However, they did note that the partially fractured shear keys still displayed adequate
lateral live load distribution characteristics. The addition of lateral tie bars was found to have
insignificant influence on shear key performance. Note that the transverse tie bars used in the
tested bridges wer made of 1 in. diameter mild steel and spaced at no more than 25 ft and were
not post-tensioned.
Take Away Points:
 Intact shear keys (i.e., crack free) should not permit relative deflection
between adjacent box beams.
17
 Partially fractured shear keys still have adequate strength to distribute live
loads laterally.
 Mild steel lateral tie-bars have insignificant influence on shear key
performance.
In the experimental work by Gulyas et al. [1995], the performance of grouted keyways using
non-shrink grouts and magnesium ammonium phosphate mortars were studied and compared.
Three types of tests were conducted including a direct vertical shear test considering truck loads
on the bridge, a direct transverse tension test considering transverse creep and shrinkage effects,
and a direct longitudinal shear test considering longitudinal creep and shrinkage effects. All the
16 tested specimens had small dimensions and grout strengths ranging from 5.9 to 7.3 ksi. They
found that the composite keyway specimens using magnesium ammonium phosphate mortars
showed higher direct tensile bond strengths, vertical shear, and longitudinal shear than those of
the non-shrink grout keyway specimens. They also found that magnesium ammonium phosphate
mortars showed significantly lower chloride absorption ability, which is of benefit for roadways
exposed to salts or sea sprays. Finally, the authors recommended not using non-shrink grouts for
the keyway unless the tensile and shear strengths satisfy the requirements in their study.
Take Away Points:
 Mortars used in shearkeys consisting of ammonium phosphate displayed high
bond and shear strengths and also had low chloride absorption.
El-Remaily, et al. [1996] compared the American and Japanese approaches to designing adjacent
concrete box beam bridges primarily because longitudinal cracking was very rarely associated
with Japanese box beam bridges. It was found that the primary differences between American
and Japanese designs were: (1) the size and shape of longitudinal joints and (2) the amount of
transverse post-tensioning. After further review, the authors proposed a new precast prestressed
box girder bridge design along with a design methodology suitable for U.S. practice. The
proposed design methodology takes the transverse diaphragms as the only components which
sustain the post-tensioning forces from the post-tensioning ties. The transverse diaphragms are
connected at the joints and laterally distribute live loads among those box girders. A grillage
analysis was performed using beam elements with common nodes for the diaphragms and beams
and considering dead and live loads (including barriers). Working stress methodologies are used
to compute the transverse stresses in the top and bottom of the diaphragms after the bending
moments in the diaphragms are derived from the grillage model. The post-tensioning is
determined to counteract the calculated stresses in the diaphragms such that no lateral tensile
stress is induced in the diaphragms. The author’s parametric studies indicated that the needed
transverse post-tensioning remains constant per unit span length and varies significantly with the
bridge width. This method was adopted by the Precast/Prestressed Concrete Institute (PCI)
Bridge Design Manual [PCI 2003]. The authors described a design example but provided no
information on neither experimental validation nor analytical evaluations using a rigorous finite
element approach.
Take Away Points:
 The primary differences between American and Japanese designs are: (1) the
size and shape of longitudinal joints and (2) the amount of transverse post-
tensioning.
18
 The amount of post-tensioning remains constant on a per foot basis (for
constant width of bridge); the amount of post-tensioning needed varies with
bridge width.
A study conducted in the State of Ohio examined the performance of the State’s standard box
beam shear key design, investigated the problem causing shear key failure and developed new
types of keyway connection details [Huckelbridge and El-Esnawi, 1997]. Initially, a 3D FE
model of a three-box beam bridge with a length of 40 ft and a width of 12 ft was established. A
concentrated load simulating a truck wheel load was applied on the center of the interior beam.
The analytical results indicated that transverse tensile stresses in the bridge top flange are the
main factor causing many shear key failures. To deal with the issue, a new type of shear key was
proposed by placing the shear key at the neutral axis of the beam cross section. FE results
showed that the proposed shear key sustained much smaller tensile stresses which would not
cause shear key cracking nor failure. To complete the examination, small scale testing of a multi-
beam bridge cross section was conducted. The small scale specimens are slices of the three-beam
assembly with a length of 12 in., a width of 144 in., and a depth of 33 in. Static and cyclic loads
were applied at the center of these specimens. The experimental results showed that the mid-
depth shear key design (only the shear key was grouted instead of the whole keyway) had
significantly improved the static load carrying capacity and provided a longer fatigue life than
the previous shear key design. In the end, the authors also proposed a water-proofing shear key
design with a mid-depth shear key, which uses water-proofing membrane, asphalt topping and
foam filler above the shear key. The test results indicated that this shear key design maintained
watertightness after fatigue testing in the laboratory environment. However, further evaluations
at real bridge sites were noted to be needed.
Take Away Points:
 A shear key placed at mid-depth of the beam must resist much smaller tensile
stresses than that which would cause cracking.
Research conducted by Lall et al. [1998] compared the long-term performance of a partial depth
shear key system and a modified, full-depth shear key/transverse tie system based on a survey of
bridges in New York State. The modified full-depth shear key/transverse tie system was
developed based on the results of bridge inspections in the State of New York and information
from other states - in particular the State of Michigan. Note that the new system possesses two
post-tensioning ties located at the third points of the girder depth instead of one tie at the girder
mid-depth. Survey results indicated that the new full-depth shear key/transverse tendon system
showed superior cracking prevention ability and reduced the frequency of reflective cracking in
the deck. As a result of the work, the authors recommended using the new full-depth shear key
for future adjacent box beam bridges. Additionally, the authors recommended the use of full-
width bearing pads, more reinforcement in the concrete topping, higher transverse post-
tensioning forces and two ties at each post-tensioning location.
Take Away Points:

 Two ties at each post-tensioning location are preferred to single ties.
 Full-depth shear keys show improved performance.
19
 Additional reinforcement in a cast-in-place topping also resulted in improved
performance.
 Higher transverse post-tensioning also led to improved performance.
Miller et al. [1999] evaluated the performance of box girder shear keys with different shear key
locations and different grouting materials. Three types of specimens, made of four box beams,
were fabricated with a top shear key plus non-shrink grout, a mid-depth shear key plus non-
shrink grout, and a top shear key plus epoxy grout. The specimens were fabricated and tested
outside under real environmental conditions and thus experienced continuous temperature
gradients. For each specimen a total of 1,000,000 cycles of load (20 kips) were applied on one
interior beam and then moved to the other interior beam. The cracks that developed in the shear
keys were inspected using ultrasonic pulse velocity. A static load (20 kips) was also applied on
the interior beams separately or simultaneously to check the live load distribution characteristics
before and after the development of cracks caused by cyclic loads. The test results indicated that
temperature induced stresses – when a shear key was located near the top of the beam – were
consistently high enough to cause significant cracking of the shear key material. These cracks
significantly propagated from the two ends near the supports to the bridge mid-span after cyclic
loads. Conversely, when the shear key was placed at member mid-depth, the shear key did not
experience significant cracking under neither thermal nor live loads. They also found that live
loads would not cause new cracking but appeared to propagate existing thermal cracks. In
addition, static load test results showed that the cracking in the shear key had no remarkable
effect on the live load distributions among box beams, but did cause leakage in the joints. In the
end, Miller et al. [1999] recommended the use of a grout material with high bond strength for the
joints of the adjacent box girders even though this results in some concerns such as thermal
compatibility due to the high thermal expansion coefficient of the epoxy, undesired failure in the
concrete rather than the epoxy, inconvenience, and the use of poisonous methylethyketone
(MEK) for the epoxy.
Take Away Points:
 Shear keys located near the top of the beam can experience stresses high
enough to induced cracking from temperature changes.
 Cracking tends to start near the ends of the beams.
 Shear keys located near the beam mid-depth did not experience cracking of
the joint material.
Follow-up work by Greuel et al. [2000] studied the field performance of a bridge constructed
with a mid-depth shear key. Only the shear key was grouted and the gap above the shear key was
filled with compacted sand with a sealant encapsulating the exposed longitudinal joint. Non-
prestressed tie rods were used to connect the box beam together before grouting. Field testing
was conducted using four Ohio DOT dump trucks - with a total weight ranging from 27 to 32
kips - at various transverse positions. In addition to the static load test, the bridge responses were
continuously collected when trucks travels cross the bridge at a speed of around 50 miles per
hour. The results indicated that there was no appreciable differential displacement between
girders. The authors further concluded that the shear key and transverse rod system adequately
resisted the applied live loads.
20
Take Away Points:
 A bridge with only the shear key grouted and non-tensioned transverse rods
can result in a bridge that shows no differential displacement under live loads.
Issa et al. [2003] conducted small scale tests of keyway specimens to investigate the performance
of four grout materials using direct shear, direct tension, and flexural tests. The chloride
permeability and shrinkage of the four grouts were also measured. The test results indicated that
the polymer concrete showed the highest shear, tensile and flexural strengths. The polymer
concrete also had superior chloride resistance and less shrinkage compared to the other grouts
while set grout had significant shrinkage due to its high water content. In addition, FE analysis of
tension test specimens showed that the polymer concrete specimens sustained the highest load
with a minimum of cracking and crushing compared to others.
Take Away Points:
 Polymer concrete has good strength and chloride resistance characteristics.
Badwan and Liang [2007a] performed a grillage analysis to determine the needed transverse
post-tensioning for a precast adjacent, solid, multi-beam deck. The grillage model was
established using beam elements for the beams while also considering the stiffness at the keyway
locations. Parametric studies were performed to investigate the importance of factors such as
skew, deck width, thickness, and span length on the design of such a system. The results indicate
that the required post-tensioning stress decreases with an increase in the deck width, deck
thickness, and skew angles (especially for skew angles greater than 30 degrees). The authors note
that the influence of skew is due to the fact that transverse bending in the skew direction
decreases with skew angle. The span length affects the needed post-tensioning stress when the
bridge skew is very large. In the end, they concluded that it is adequate to design the needed
post-tensioning for such a system (especially with high skew) based on current AASHTO
specifications.
Take Away Points:
 The amount of post-tensioning decreases with an increased deck width,
thickness, and skew.
 Span length affects the needed post-tensioning when the skew is very large.
A literature search conducted by Badwan and Liang [2007b] revealed that little research has been
conducted to study the performance of full depth keyways even though testing has been
conducted to investigate the behavior of partial depth keyways. Thus, the authors implemented
field testing and associated FE analysis of a post-tensioned adjacent solid box girder bridge with
full depth keyways, mid-depth shear keys, and transverse post-tensioning. The 3D FE model was
established using solid elements for the concrete and grout and link elements were used for the
post-tensioning tendons. During testing, longitudinal strains in the girders were recorded. The
adequacy of the FE model was validated using the strain data. Based upon the testing and
analytical results, the authors concluded that the lateral load distribution was not affected as long
as no cracks were induced in the shear keys. It should be noted that serviceability issues caused
by shear key cracking were not addressed by the authors.
21
Take Away Points:
 Lateral load distribution is not impacted by keyway geometry as long as no
cracks are induced in the shear keys.
Dong et al. [2007] established 3D finite element models to investigate and compare the behavior
of the three types of joints shown in Figure 3. Finite element models were established using solid
elements for both the concrete and grout. Parametric studies were then conducted considering the
three types of joints and three strengths of grouts. The results showed that no cracking was found
in the FE model of Joint A but significant stress concentrations and cracking occurred in Joints B
and C. They concluded that cracks developed in Joints B and C were due to the significant
change of the keyway shape. In addition, they also found that higher strength grout material does
not reduce the cracks.
Take Away Points:
 Radical changes in shearkey geometry (i.e., very sharp corners) may result in
higher stress levels.
(a) Joint A (b) Joint B (c) Joint C

Figure 3 Basic Shearkey Shapes
Sharpe [2007] conducted extensive FE element analysis of Precast/Prestressed Concrete Institute
(PCI) style and Texas Department of Transportation (TxDOT) style box girder bridges to
investigate the performances of the shear keys. FE models were established using solid elements
for the beams, diaphragms, and keyways and elastomeric bearing pads were modeled using
spring elements whose vertical and lateral stiffnesses were determined based on the material
properties of the bearing pad and basic mechanics of materials. The AASHTO HS-25 truck load,
strains due to shrinkage, and a temperature gradient were applied to those bridge models. Sharpe
considered two types of failure in the shear keys: debonding and cracking (with different failure
stresses). The FE analysis results indicated that reflective cracking was due to high tensile
stresses in the shear keys caused by temperature gradients and shrinkage strains instead of live
loads. It was further found that these cracks usually developed near the supports instead of at the
bridge mid-span. Analytical results showed that composite slabs are most effective at alleviating
high tensile stresses in the shear keys although post-tensioning and full-depth keyways also
22
reduce the tensile stresses. It should be noted that the full-depth keyways shown in Figure 4(c)
and Figure 4(d) and examined by Sharpe extend the partial-depth keyways shown in Figure 4(a)
and Figure 4(b) to the beam bottom.
Take Away Points:
 Cracking is due to shrinkage strains and temperature and not live loads.
 Cracks usually develop near the end of the bridge first.
 Composite slabs are the most effective means of alleviating high tensile
stresses.
(a) PCI partial depth keyway (b) AASHTO partial depth keyway
(c) Full depth keyway (d) Full depth keyway

(revised from PCI partial depth keyway) (revised from AASHTO partial depth keyway)
Figure 4 Keyway Geometries for PCI and TxDOT Style-Box Girder Bridges
4.3 Publications After 2008

The work done by Attanayake and Aktan [2008] summarized the evolution of the Michigan
design procedures for adjacent box-beam bridges and their performance since the 1950s. The
Michigan Bridge Design Guide had adopted many recommended practices provided in NCHRP
Synthesis 393 such as higher transverse post-tensioning forces, full-depth keyways, top shear
keys, and using a 6 in. thick cast-in-place concrete deck as shown in Figure 5. They found that
reflective cracks were still found in the Michigan adjacent box-beam bridges. In order to identify
the main source of the formation of longitudinal reflective cracks, they monitored an adjacent
box-beam bridge starting from construction (note that the bridge has narrow, full depth keyways
with top shear keys). Inspection results revealed that cracks were found at the interfaces between
beams and keyways before and after the post-tensioning was applied. They also found reflective
cracks were found in the concrete deck (mostly near supports) 15 days after placement even
23
before construction the barrier or applying live loads. They concluded that reflective cracks are
due to effects such as hydration heat and drying shrinkage.
Take Away Points:

 Cracking forms at the interface between the joint material and the box beam
concrete.
 Reflective cracks are principally due to shrinkage.
Figure 5 Typical Michigan Keyway Geometry and Post-tensioning

Kim et al. [2008] presented recent applications of precast adjacent box-beam bridges with full-
depth keyways with mid-depth shear keys grouted with cast-in-place concrete and transverse
post-tensioning in South Korea. The authors performed 2D FE analysis of three box beam
sections without transverse post-tensioning to investigate the performance of four placement
conditions for the shear key (i.e., no shear key, top shear key, mid-depth shear key, and bottom
shear key) as shown in Figure 6(a), Figure 6(b), Figure 6(c), and Figure 6(d). Various loading
and boundary conditions were applied. Based on the beam differential deflections results, it was
indicated that the top shear key, the mid-depth shear key, and the bottom shear keys all show
superior performance than that of no shear key and the mid-depth shear key was the best of the
four configurations. Sang [2010] confirmed their results and concluded that the location of the
shear key does not significantly affect the performance of full-depth keyways. To verify the
feasibility of the proposed full depth keyway (with the mid-depth shear key grouted with cast-in-
place concrete and high transverse post-tensioning), Kim et al. [2008] conducted flexural testing
and 3D FE modeling of a three box beam specimen. The failure and cracking loads both
exceeded the ultimate load and service load based on the Korea design code which is similar to
the AASHTO bridge design specifications. No longitudinal cracks were found in the joints when
the specimen sustained service and ultimate loads. The measured relative displacements
indicated that effective load transfer by the shear key connections was occurring. Kim et al.
[2008] conducted fatigue testing (2 million cycles) of the three-box beam specimen. The test
results indicated that no cracks were found in the longitudinal joints and the specimen exhibited
excellent fatigue resistance with the residual deflection being recovered 24 hours after fatigue
testing. Finally, Kim et al. [2008] applied the proposed full depth keyway to a real bridge. Field
tests were conducted using static and moving dump trucks on the bridge. They concluded that the
box-beam bridge performed well structurally under static and moving dump truck loads. Further,
no longitudinal cracking in the keyway joints was reported by the authors. It should be pointed
24
out that long term behavior of the three box beam specimens and the constructed bridge were not
evaluated.
Take Away Points:
 Mid-depth shearkey placement results in the best performing joint – especially
when used with high post-tensioning and cast-in-place concrete.
(a) No shear key (b) Top shear key
(c) Mid-depth shear key (d) Bottom shear key

Figure 6 Common shearkey locations
Attanayake and Aktan [2009] developed a simple analytical model consisting of plate elements
based on the macromechanics concept. In this model, the plate element represents a combination
of two half box-beam sections, one shear key and concrete deck. Namely, the cross-section of the
plate element has the identical section properties as those of the combination cross-section. The
stiffnesses of the box-beam sections, shear keys and concrete deck are calculated and then
incorporated into the plate elements. The transverse moments along the longitudinal joints
between the adjacent beams were determined from the macromechanical model based on the
AASHTO LRFD bridge design specifications. These calculated moments were then used to
determine the needed transverse post-tensioning. Further, the authors demonstrated a design
example in their paper. However, the macromechanical model fails to simulate the interaction
between the keyways and beams due to different material properties (i.e., grout and concrete)
and bond at the interfaces.
Take Away Points:
 Machromechanical modelling fails to simulate the interaction between the
keyway and the beam.
Follow-up work by Ulku et al. [2010] proposed a rational design procedure utilizing the
macromechanical model developed by Attanayake and Aktan [2009] to calculate the transverse
moments along the transverse joints and thus determine the required transverse post-tensioning.
The concept is to use multi-stage post-tensioning to minimize the longitudinal cracking in the
keyway and reflective cracking in the concrete deck. A 3D FE model was established using solid
elements for the beams, keyways, diaphragms and deck. Multi-stage post-tensioning after
grouting the keyway and after the deck placement was simulated. They concluded that the two
stage post-tensioning process is effective at reducing cracking issues for the bridge subjected to
25
dead and live loads and temperature effects. However, in their designs, the tensile stresses in the
deck near the fascia beams due to live loads are significant and may not be easily offset by two-
stage post-tensioning. They also found that the temperature gradient is the main factor causing
the cracks which developed at the interface of the top shear keys. Another cause of cracks is that
the post-tensioning is not uniformly distributed at the keyway because of shear lag.
Take Away Points:
 Two stage post-tensioning may minimize longitudinal cracking.
 Temperature gradient is the main factor causing cracks to develop at the joint
interface.
Sang [2010] performed grillage analysis of adjacent box girder bridges subjected to live loads so
as to determine shear forces and moments that must be sustained by the shear keys.
Subsequently, the performance of the keyway joint was investigated using a 2D FE model which
sustained the loads equivalent to the shear forces and moments derived from the grillage model.
The FE model was established using plane strain elements for the concrete and the grout which
share common nodes at the interfaces. Shear tests were conducted to examine the failure modes
of the keyway joints grouted with cementitious grout and epoxy. The test results were also used
to validate the adequacy of the FE model. Finally, parametric studies were performed using the
validated FE models to investigate the influences of keyway geometry, grouting materials, post-
tensioning, and bearing locations on the performance of the shear key. Note that fiber reinforced
cementitious material was recommended by the author to grout the shear key due to its high
tensile strength and was also used in their FE shear key models, although no previous research
was found in the literature using fiber reinforced concrete for grouting the shear key. Based on
the FE analysis results, the authors concluded that cracks developed in both the full depth and
partial depth keyways using cementitious grout while cracks was found in only the partial depth
keyways but not in the full depth shear key using the epoxy grout and fiber reinforced
cementitious grout. They also found that the vertical locations of the shear key did not affect its
behavior. They recommended using a higher transverse post-tensioning force since they found
the post-tensioning specified by the PennDOT was not enough to provide crack resistance. The
FE results indicated that the shared bearing pad (bearing under the shear key as shown in Figure
7(a)) reduces the cracks in the shear key relative to isolated bearing pads (bearing under the
beam flanges as shown in Figure 7(b)).
Take Away Points:
 Epoxy grout and fiber reinforced cementitious materials perform well when
used in a full-depth shear key.
 High post-tensioning may be needed to completely eliminate cracking.
(a) Isolated bearing pad (b) shared bearing pad

Figure 7 Common Bearing Pad Details
26
Fu et al. [2011] proposed an approach to designing the required post-tensioning for solid, multi-
beam bridge system based on the shear friction concept and FE modeling techniques. The FE
models were established using solid elements, link elements, and contact elements for the beams,
post-tensioning ties and interfaces between the shear key and the beam, respectively. The
adequacy of the FE models were validated against the strain data measured during field tests
using an onsite controlled dump truck. Based on the FE results, the author recommended
different levels of post-tensioning for bridges with different span lengths. The authors found that
the boundary conditions had great influence on the predicted bridge response. They found that
the post-tensioning does not affect the live load distribution until cracks develop in the keyway
and/or concrete topping. Finally, the authors gave some recommendations for improving the use
of shear keys in Maryland (e.g., using a two-staged construction sequence {e.g., 16.7% and
100% of the designed post-tensioning of design level before and after grouting the keyways} and
using full-depth shear keys).
Take Away Points:
 Bridges of different span lengths may require different amounts of post-
tensioning.
 Two-stage post-tensioning may help reduce the development of cracks.
With the goal of achieving simple and economic fabrication and construction of precast adjacent
box girder systems, Hanna et al. [2011] developed and evaluated two types of non-post-
tensioned transverse connection details that don’t use diaphragms nor a concrete deck (i.e., the
wide joint system and the narrow joint system shown in Figure 8(a) and Figure 8(b)). The two
systems were developed based on the AASHTO/PCI and the Illinois DOT box beam connection
details, respectively. The wide joint system incorporates a wide full-depth keyway joint filled
with cast-in-place concrete and utilizes top and bottom reinforcement placed in the top and
bottom flanges of the box beams to resist transverse tensile stresses. The narrow joint system
incorporates a narrow joint with a partial depth keyway, top shear key and non-shrink grout and
utilizes top and bottom threaded rods placed in the top and bottom flanges of the box beams to
resist the transverse tensile stresses. 3D FE element models were established using shell elements
for the beam flanges and webs and frame elements for the reinforcement and threaded rods.
Design charts were developed for determining the needed tension force at the connection (i.e.,
the required amount of reinforcement or threaded rods). Two-beam specimens using the two
systems were fabricated and tested under cyclic loads. Water dams were constructed on the top
surface of the specimens so as to monitor for crack development and water leakage. Test results
indicated that, for the two system specimens, neither cracks nor water leakage were found in the
keyway after 2 million cycles and the differential deflections were found to be below 0.07 in.
after 3 million cycles. However, in their study, no apparent consideration was given to
performance under thermal loads.
Take Away Points:
 It may be possible to design a bridge without transverse post-tensioning that
performs adequately.
27
(a) Wide joint
(b) Narrow joint

Figure 8 Connection Details Proposed by Hanna et al. [2011]
Follow-up work by Hansen et al. [2012] developed another joint system based on the narrow
joint system proposed by Hanna et al. [2011]. This system was developed without using
diaphragms nor concrete topping and utilizes post-tensioning to reduce the possibility of
cracking or leakage. As shown in Figure 9, the sleeves, located below the beam top flange and
above the bottom flange, are used to accommodate the duct, the post-tensioning rods and
couplers. The required post-tensioning was determined based upon the design chart for the
required tension force in the connection developed by Hanna et al. [2009]. Experimental testing
was conducted for four-box beam specimens placed in a cantilever and mid-span loading setups,
successively. In the cantilever setup, the specimen was supported at the transverse center and
edge and a load with 5 million cycles was applied on the joint. In the mid-span loading setup, the
specimen was supported at the two transverse edges and the load applied on the specimen center.
Results indicated that no significant strain change, cracking, nor leakage near the shear key
region occurred. The authors strongly recommend this system for real bridge construction.
However, temperature gradient and shrinkage effects were not considered in their study.
28
Take Away Points:
 A cast-in-place topping may further improve the performance of a non-post-
tensioned box beam bridge.
Figure 9 Connection Details Proposed by Hansen et al. [2012]

Grace et al. [2012] inspected a bridge in Michigan constructed based on recent Michigan design
procedures. The bridge has two simply supported spans of 122.5 ft, seven diaphragms with post-
tensioning bars that were highly post-tensioned before grouting, and full depth keyways with a
top shear key and a concrete deck. The inspection results found that significant longitudinal
cracks were formed in the shear key and deck even though the traffic on that bridge is light and
was judged to not likely have induced those cracks. In addition, inspection on some other
adjacent box girder bridges in Michigan also revealed that reflective cracks had formed in the
deck. To investigate the source of those cracks, the authors conducted an experimental test of a
bridge specimen in the lab. A four-point concentrated load up to the service load of 80 kips was
applied on the specimen, and no reflective cracks in the deck were found even when the
transverse post-tensioning decreased to zero. They concluded that the traffic loads are not the
main condition causing reflective cracking in the deck. Thus, the authors considered temperature
effects in subsequent FE analyses. The FE model was established using solid elements for the
beams, diaphragms and deck, and link elements for the post-tensioning ties. After the FE model
was validated against the results from the experimental tests, FE analyses of real bridges was
performed considering dead and live loads and temperature gradients according to the AASHTO
bridge design specifications. Based on the FE results, the required amount of transverse post-
tensioning required to mitigate reflective cracking for the real bridges was then established. For
practical applications, the required number of diaphragms and the required amount of post-
tensioning per diaphragm were given for the adjacent box-beam bridges in Michigan. The
authors found that the post-tensioning effects are mainly localized at the diaphragm regions due
to shear lag effects and the required amount of diaphragms for eliminating reflective cracks
increases with an increase in span length, while the required post-tensioning increases with
increased bridge width.
Take Away Points:

 Traffic loads are not the primary factor in the development of cracks.
 Temperature induced effects may be the primary source of crack development.
29
4.4 Literature Search and Survey Synthesis and Summary
A significant amount of information related to adjacent box beams was presented and
summarized in the preceding pages. Although there are many important facts to take-away from
these sources, the following synthesis and summary was formulated to provide a brief synopsis
of the information that had the greatest impact on the development of the research plan
summarized in the following pages.
Cracking of the shearkey between adjacent box beams appears to principally be a service related
problem as multiple sources indicate that even with a cracked joint, a bridge can continue to
effectively distribute loads throughout the primary load carrying members. Consistent
throughout the literature is the conclusion that joints that utilize full-depth keyways combined
with a shearkey located at mid-depth perform the best. In a related manner, transverse post-
tensioning seems to be the most effective when two ties are used at each location (e.g, one near
the top and one near the bottom). Further, it seems apparent that bridges perform better with high
amounts of transverse post-tensioning (note, however, that there have been some reported
instances where no post-tensioning was reported to perform well too). Further, systems that use a
structural cast-in-place deck tend to perform better over time and that the best performance tends
to be derived when the reinforcement amount is rather high. However, there is some evidence to
suggest that using higher post-tensioning levels may be able to compensate for not having a
structural deck.
Geometrically, it may be possible that the amount of post-tensioning required on a per foot basis
may be constant for bridges with variable lengths but a single width; also, skew may play a role
in the amount of post-tensioning required for a given span length. For a given span length, the
amount of post-tensioning required does vary with bridge width and also varies with skew.
With regard to cracking, it appears that cracking tends to be most prominent at the interface
between the joint material and the box beam. Further, cracking seems to first initiate near the
ends of beams. Cracking does not seem to be first initiated by the application of live loads.
There are, however, differing opinions on the relative contribution to cracking from shrinkage
and temperature. Nevertheless, once cracking is initiated by either shrinkage and/or temperature,
they can continue to grow with subsequent live load application.
To summarize information useful for the development of the analytical and testing plans,
design/construction aspects and FE analyses and testing on adjacent box beam bridges were
grouped. Design and construction attributes for the adjacent box girder bridges in studies
reported above are summarized in Table 4. FE analysis details are summarized in Table 5.
Laboratory tests of small scale, medium scale, and full scale specimens are summarized in Table
6. Field testing of adjacent box girder bridges are summarized in Table 7. These four tables were
found to be very helpful in developing the plan for the analytical and experimental evaluations.
For instance, the performance of various keyway geometries have been evaluated by other
researchers, including partial depth keyways and full depth keyway with (or without) top shear
key, mid-depth shear key or bottom shear key as shown in Table 4. The literature search results
indicated that the full depth keyways generally show better performance than partial depth
keyways, and mid-depth shear keys show better performances than top and bottom shear keys.
Thus, the proposed experimental and analytical studies will mainly focus on full depth keyway
with mid-depth shear key.
30
Table 4 Design and Construction Attributes
Transverse Tie Keyway Bearing Construction Concrete FE Laboratory Field
Refs. Keyway geometries Diaphragms Grout
details preparation details sequence Deck analysis testing testing
Girder mid-height;
Huckelbridge Non-post-tension
Partial depth keyway Yes NG NG NG NG None Yes No Yes
[1995] mild steel (1 in.
diameter)
Gulyas et al. Full depth keyway and top Non-shrink grout; Sandblast/wash
No No NA NA NA No Yes No
[1995] shear key; Narrow joint MAP mortars off
Partial depth keyway and Post-tensioning

El-Remaily, et Post-tensing
top shear key (pocket near (determined by 5 NG NG NG No Yes No No
al. [1996] after grouting
diaphragms) design calculations)
Huckelbridge Partial depth keyway and Non-shrink grout; Power grinder
and El-Esnawi top and mid-depth shear No No MAP mortars; and wire brush; NA NA No Yes Yes No
[1997] keys epoxy sand-blaster
Two post- Sandblast,

Lall et al. Full depth keyway and top Full width
tensioning ties at More than 3 NG cleaned, and pre- NA Yes No No No
[1998] shear key bearing
third points in depth wetted
2.5 in. asphalt
31
Greuel et al. Partial depth keyway and Non-post-tensioned Neoprene Grout after
5 NG NG wearing No No Yes
[2000] mid-depth shear key rods bearing pad installing rods
surface
Partial depth keyway and
Miller et al. Slightly post- Non-shrink grout; Post-tensioning
mid-depth shear key 5 NG NG No No No Yes
[1999] tensioned rods epoxy before grouting
(Pocket near diaphragms)
Sandblast; air
Set 45; set 45 HW;
Issa et al. Full depth keyway and mid- pressure and
No No set grout; polymer NA NA NA Yes Yes No
[2003] depth shear key high pressure
concrete
washing
Badwan and Full depth and mid-depth Bonded post- Post-tensioning
No NG NG NG No Yes No No
Liang [2007a] shear key tensioning tendons before grouting
Badwan and Full depth keyway and Bonded post- Post-tensioning

No NG NG NG No Yes No Yes
Liang [2007b] mid-depth shear key tensioning tendons before grouting
Full depth and mid-depth

Dong et al. shear key; partial depth
No No Yes NG NG NA No Yes No No
[2007] keyway and mid- (bottom-)
shear key
MAP - Magnesium ammonium phosphate; NA - Not Applicable; CFRP - Carbon Fiber Reinforced Polymer; NG - Not Given
Table 4 Design and Construction Attributes (Continued)
Transverse Tie Keyway Bearing Construction Concrete FE Laboratory Field

Refs. Keyway geometries Diaphragms Grout
details preparation details sequence Deck analysis testing testing
Partial depth keyway and

Unbonded post- Spaced at 10 Elastomeric
Sharpe (2007) top shear key; Full depth Non-shrink grout NG NG Yes Yes No No
tensioning tendons ft bearing pads
keyway and top shear key
Type R-2, which is

Attanayake cement and
Full depth keyway and Bonded post- Post-tensioning
and Aktan 6 fine aggregate NG NG Yes No No Yes
top shear key (1.5-3 in.) tensioning tendons after grouting
(2008) mixture with 14 +/-
4% air
FEA: Full depth keyway

and no, top, mid-depth or
Kim et al. Bonded post- Cast-in-place Elastometric
bottom shear keys 5 NG NG Yes No No Yes
(2008) tensioning tendons concrete rubber pad
Test: Full depth and mid-
depth shear keys (2-4.8 in.)
Multi-staged
Attanayake
construction:
and Aktan
32
Full depth keyway and Unbonded post- Post-tensioning
(2009) and 5-7 NG NG NG Yes Yes No No
top shear key (1.5-3 in.) tensioning tendons after grouting
Ulku et al.
and after deck
(2010)
placement
Fiber reinforced
Full depth keyway and top cememtitious
Post-tensioning Placed under Post-tensioning
Sang (2010) shear key; Partial depth NG material; NG Yes Yes Yes No
tendons the keyway after grouting
and top shear key cememtitious
material; epoxy
Fu et al. Full depth keyway and top Post-tensioning Post-tensioning
No Non-shrink grout NG NG Yes Yes No Yes
(2011) shear key threaded rods before grouting
Non-post-
Full depth keyway and no tensioning Cast-in-place
Hanna et al.
shear key; Partial depth reinforcement; Non- No concrete; Non- Roughened NG NA No Yes Yes No
(2011)
and top shear key post-tensioning shrink grout
threaded rods
Non-post-
Jenna et al. Partial depth keyway and Neoprene Post-tensioning
tensioning threaded No Non-shrink grout Roughened No No Yes No
(2012) top shear key bearing pad after grouting
rods
Grace et al. Full depth keyway and top Unbonded post- Neoprene Post-tensioning
From FEA Non-shrink grout NG Yes Yes Yes No
(2012) shear key tensioning CFRP bearing pad after grouting
CFRP - Carbon Fiber Reinforced Polymer; MAP - Magnesium ammonium phosphate; NA - Not Applicable; NG - Not Given
Table 5 Summary of FE Analysis
Type of Box
Refs. Software Keyway Interface Diaphragm Deck Tie Bearing Load
Analysis Beam
Huckelbridge
NG NG NG NG NG NG NG NG NG NG
(1995)
Dead and live
El-Remaily, et Beam Common Common beam Simply
Grillage analysis NG None NA loads
al. (1996) elements nodes nodes elements supported
(including
Directly
Solid Solid Common Solid Simply Concentrated
Huckelbridge 3D SAP None apply
elements elements nodes elements supported load
and El-Esnawi forces
(1997)
Plane Plane Common Spring Concentrated
2D SAP None None None
elements elements nodes elements load
Issa et al. Solid Solid Common Concentrated
3D ANSYS None None None NA
(2003) elements elements nodes load
Badwan and Beam Common Common Beam Simply
Grillage analysis ANSYS None NA HS-25 truck
Liang (2007a) elements nodes nodes elements supported
Badwan and Solid Solid Common Link Simply

3D ANSYS None None Dump truck
Liang (2007b) elements elements nodes element supported
Dong et al. Solid Solid Common Concentrated

3D ABAQUS None None None NA
(2007) elements elements nodes load
HS-25 truck;
Sharpe Solid Solid Common Solid Solid Spring
3D ANSYS None shrinkage;
(2007) elements elements nodes elements elements elements
thermal
Plane Plane Common Simply Concentrated
2D DIANA None None None
elements elements nodes supported load
Kim et al.
(2008)
Solid Solid Common Solid Solid Bar Simply Concentrated
3D DIANA
elements elements nodes elements elements elements supported load
Attanayake Integrate Integrate Integrate

Maromechanical Programmi Simply
and Aktan d Plate d Plate NONE None d Plate None HL-93
model ng supported
(2009) elements elements elements
Ulku et al. Solid Solid Solid Solid Truss Simply
3D ABAQUS NONE HL-93
(2010) elements elements elements elements elements supported
Beam Common Common Beam Simply

Grillage analysis NG None NA HS-25 truck
elements nodes nodes elements supported
Sang (2010) Concrentratd
Plane Plane Common
2D ABAQUS None None None NA and
elements elements nodes
distributed
Hanna et al. Shell Common Common Frame Simply
3D SAP2000 None None HL-93
(2011) elements nodes nodes elements supported
Fu et al. Solid Solid Contact Link Simply

3D ANSYS None None HL-93
(2011) elements elements elements elements supported
HL-93 and
Grace et al. Brick Brick Contact Brick Brick Truss Simply
3D NG temperature
(2012) elements elements elements elements elements elements supported
gradient
NG - Not Given; NA - Not Applicable
33
Table 6 Summary of Laboratory Testing
Testing Number Concrete Relative Crack
References Specimens Length Skew Width Depth Grout strength Load Temperature Strain
scale of beams strength displacement detestion
Vertical shear;
Gulyas et al. Small Keyway Non-shrink grout: 5.9; Direct tension;
3.25 in. NA 6-6.5 in. 7-14 in. NA NG No NA NA Visually
(1995) scale specimens MAP mortars: 7.3 ksi Longitudinal
shear
Direct Current Foil-

Huckelbridge Non-shrink grout: 5.5 ksi;
Small Multu- Cyclic Differential backed
and El-Esnawi 12 in. 0º 144 33 in. 3 MAP mortars: 5 ksi; 6 ksi No Visually
scale beam slices concentrated load Transducer strain
(1997) epoxy: 13 ksi
(DCDT) gages
Set 45: 5.8 ksi;

Direct shear;
Issa et al. Small Keyway set 45 HW: 5.6 ksi;
5-6 in. 0 17-21 in. 17-26 in. NA 6.5 ksi Direct tension; No NG NG Visually
(2003) scale specimens set grout: 7.7 ksi;
Flexeral bending
polymer concrete:10.8 ksi
Static Linear
Multi-box comcentrated variable
Kim et al. Full Strain
34
beam 61 ft 0 95 in. 31.5 in 3 4.9 ksi 8 ksi load/ Cyclic No differential Visually
(2008) scale gauges
specimens concentrated load transducers
(Mid-span) (LVDTs)
Small Keyway Cementitious grout: 4.5
Sang (2010) 5 in. 0 7 in. 17 in. NA 11.3 Direct shear No NG NG Visually
scale specimens ksi; epoxy: 10 ksi
Multi-box
Hanna et al. Medium 27 in.; Cyclic A water
beam 8 ft 0 8 ft 2 6 ksi 8 ksi No Yes No
(2011) scale 32 in. concentrated load dam
specimens
Multi-box A water
Jenna et al. Medium Cyclic
beam 8 ft 0 16 ft 27 in. 4 Non-shrink grout: 10 ksi 8 ksi No Yes No dam;
(2012) scale concentrated load
specimens Visually
No
Multi-box Service (Recognize
Grace et al. Full Beam: 6 ksi;
beam 20 ft 0 75 in. 14 in. 4 Low-shrink grout: 8 ksi concentrated load importance of NG No Visually
(2012) scale Deck: 5.7 ksi
specimens up to 80 kips temperature
effects)

Table 7 Summary of Field Testing
Number Beam Concrete
Span Skew Width Grout Relative Crack
References Bridge Number of width × strength Temperature Load Strain
(ft) (degree) (ft) strength displacement detection
of Span beams depth (in.) (ksi)
No. 1 1 32.5 20 44 11 48 × 17 NG Relative

50 kips Dump
Huckelbridg displacement Vibrating
NA NA truck (12 +38 Visually
e et al. (1995) 40/54/ transducers wire gage
No. 2 4 17.4 68 17 48 × 27 NG kips)
54/40 (Own made)
4 (with Transverse
Yealy range: - Direct Current
same 20 kips on the omega clip Ultrasoni
Miller et al. 10-100 ºF; Differential
girders, 1 75 0 16 4 48 × 33 5 ksi Beam: 9.4 loaded interior gauges; c pulse
(1999) summer: 50-90 Transducer
different beam vibrating velocity
ºF (DCDT)
grouting) wire gauge
Linear
Vibrating
Ohio DOT truck variable
Greuel et al. wire gauge;
1 1 115.5 0 48 12 48 × 42 NG Beam: 10 NG similar to HS-20 differential NG
(2000) foil strain
Truck transformer
gages;
35
(LVDT)
Badwan and 29 kips dump
Strain
Liang 1 2 29/29 30 44 6 87 × 15-18 NG NG NG truck (9+20 No NG
transducer
(2007b) kips)
Attanayake
and Aktan 1 2 79/79 0 93.5 22 48 × 33 NG Deck: 6.4 Early summer No No No Visually
(2008)
Linear
77.2 kips dump
Kim et al. variable Strain
1 2 43/43 5 39 14 30 × 31.5 4.4 ksi Beam: 7.3 NG truck (16.7+60.5 Visually
(2008) differential gauges
kips)
transducers
35 kips dump
Grace et al. Beam: 7.0; Strain
1 1 35 ft 0 33 11 36 × 15 NG NG truck (10.8+24.2 No NG
(2012) Deck: 4 sensor
kips)
5. Updated Work Plan
5.1 Experimental and Analytical Evaluations
5.1.1 Introduction
To meet the overall project goal which results in an adjacent box beam bridge that is crack and
leak-free while at the same time meeting individual project goals (e.g., investigating cracking,
etc.), a modified work plan has been developed and is described here. Although the
fundamentals of the original project RFP, proposal, and work plan remain, the plan described
below, in short, proposes to complete the analytical and experimental programs by alternating
between them in an attempt to leverage information learned before proceeding to other
interdependent steps. The experimental and analytical evaluation program activities have been
developed with the specific goals of investigating cracking in adjacent box beam bridges,
improving the cracking resistance and behavior of adjacent concrete box beams connection
details, and eliminating cracking and leakage in the keyways between beams. Note that the
cracks of primary concern in this work consist of longitudinal cracks in the keyways between
adjacent box beams, which can cause serviceability issues (but tend to not degrade lateral load
distribution characteristics). Based upon Table 4, the design and construction attributes
influencing the long-term performance of these bridges include: (1) keyway geometry and
preparation (including the use of shear keys), (2) joint material characteristics (including the
influence of curing conditions), (3) bearing details, (4) use (or lack of use) of a cast-in-place
concrete deck, (5) amount and location of transverse post-tensioning and diaphragms, and (6)
construction sequence. Based upon the results of the literature search, the survey of states, and
our own experiences, it seems clear that the primary loads causing cracks in the joints between
adjacent box beams are the result of early age shrinkage effects, temperature effects, and live
loads. Obviously the influence of these loading conditions must be considered such that the most
critical cases are considered in the evaluation programs.
In previous research, four commercially available software packages (i.e., SAP2000, ANSYS,
ABAQUS, and DIANA), were used to investigate keyway performance and cracking in adjacent
box beam bridges (see Table 5). Although it is hard to determine if one software package
outperforms the others, the key to deriving reasonable analytical results is to use appropriate FE
modeling techniques. Due to previous extensive experience, the software ANSYS will be used in
this project to develop the nonlinear FE models for the analytical evaluation program. ANSYS is
a computationally advanced analytical software package that allows for the simulation of most
any structural engineering problems through proper application of FE modeling techniques. In
constructing the models, proper and appropriate material properties and element types within
ANSYS will be selected and utilized for establishing different bridge components (including
nonlinear effects).
The experimental and analytical evaluation programs will be conducted through nine, linear
steps with the objectives as follows:
Step A: The objective is to collect time-dependent material properties, including properties
resulting from variable curing methods, and select the most viable joint material and
keyway preparation options based upon: (1) tensile strength, (2) interfacial bond strength,
and (3) shrinkage characteristics.
36
Step B: The objective is to experimentally investigate cracking (and collect information for
validation of FE modeling approaches) in longitudinal joints by conducting small scale
tests of: (1) early age crack development and (2) effectiveness of post-tensioning for
crack closure prior to live loading.
Step C: The objective is to validate nonlinear, time-dependent FE modeling approaches for
predicting cracking in keyways and crack closure due to post-tensioning.
Step D: The object is to use the validated FE model to analytically investigate cracking in
bridges with variable: lengths, widths, skews, post-tensioning details, diaphragm designs,
differential cambers, and keyway geometry subjected to early age shrinkage.
Step E: The objective is to experimentally study the influence of shear key geometry on joint
behavior under early age shrinkage and live loads (e.g., AASHTO HL-93 loading).
Step F: The objective is to validate FE modeling approaches for predicting the behavior of joints
with different shear key geometries under shrinkage and live loads.
Step G: The objective is to design a crack free bridge based upon experimental results and
validated FE modeling techniques resulting from Steps A-F that will be tested and
evaluated in Step H.
Step H: The objective of this step is to fabricate and test two full scale specimens and validate
the FE models against experimental results.
Step I: The objective of this step is to determine the required post-tensioning and other details
that result in crack free bridges by considering different span lengths, bridge widths, and
skews via a parametric study using the validated FE models.
A typical/common box beam cross-section used by multiple State Departments of Transportation

is illustrated in Figure 10. The basic dimensions of the box beam section and details for
prestressing strands, mild longitudinal reinforcement, stirrups, and ducts are all conceptually
shown in Figure 10. The basic box beam configuration shown in Figure 10 will be used
throughout the analytical and experimental work described in the following pages. The test
specimens described in the subsequent sections will have the basic cross-section shown in Figure
10, with the specific number of presstressing strands as designed for a 45 ft span adjacent box
beam bridge (based on AASHTO LRFD bridge design specifications). In the analytical
evaluation, the number of strands and dimensions of box beams of the bridges with other span up
to 100 ft will be designed and updated based upon the AASHTO LRFD bridge design
specifications. Note, that for clarity details and information regarding the shear key and keyway
configuration have been omitted in Figure 10 as those details will be established in the various
steps described below.
37
Figure 10 Typical Box Beam Cross-section
5.1.2 Collection of Material Properties and Selection of the Joint Material and Keyway
Preparation
Step A – Material Testing and Selection
The objective of Step A is to collect time-dependent, nonlinear material properties, including
properties resulting from variable curing methods, and select the most viable joint material and
keyway preparation options based upon: (1) tensile strength, (2) interfacial bond strength, and (3)
shrinkage.
Initial basic material testing and characterization will be conducted to measure and quantify: (1)
shrinkage and (2) tensile strength of various joint materials at 28 days as shown in Table 8.
During initial material characterization, the following joint materials will be tested (note these
materials have been selected based upon the results of the literature review and consultation with
bridge owners and material specialists): epoxy grout and non-shrink grout for use in the Type IV
keyway; fiber reinforced concrete, ultra-high performance concrete (UHPC) and cast-in-place
(CIP) concrete formulated with Type K cement for use in the Type V key (Type IV and Type V
keyway are as generically shown in Figure 1). For materials placed in joints between adjacent
box beams, the principal environmental factor impacting curing is the ambient temperature
during curing (especially during early age curing). Given the fact that joint materials are
typically only placed during certain temperature ranges, initial testing will evaluate the basic
properties after the specimens have been cured at two temperatures: room temperature and at
approximately 55 ºF. After conducting the initial joint material characterization tests, the best
joint material for use in a Type IV keyway and a Type V keyway will be selected.
The interfacial bond strength between the box beam concrete and the two joint materials selected
above will be evaluated considering two types of keyway preparations (i.e., powerwashed and
mechanical roughened) as shown in Table 9. Curing of the specimens will occur at room
temperature due to the fact that temperature has little influence on interfacial bond strength.
38
After conducting the interfacial bond strength testing, the best keyway preparation approach will
be selected.
Based upon the initial basic material test results, the most promising joint materials and keyway
preparation will be selected. Then, time-dependent material testing will be conducted to
characterize the nonlinear changes in shrinkage, bond strength, and tensile strength with time
(e.g., at 1 day, 7 days, 14 days, and 28 days) as shown in Table 10. These measured time-
dependent, nonlinear material characteristics will be extensively utilized in the nonlinear FE
simulations conducted in subsequent steps.
Table 8 Initial Joint Material Property Testing
Materials Curing Properties
Room
Epoxy grout; non-shrink grout (a) Tensile
temperature
(Joint materials for Type IV strength;
keyways) ~55ºF (b) Shrinkage
Fiber reinforced concrete, Room

UHPC, cast-in-place concrete temperature (a) Tensile
with Type K cement
strength;
(Joint material for Type V
~55ºF (b) Shrinkage
keyways)
Table 9 Keyway Interface Preparation Testing

Keyway
Material Curing Property
Preparations
Most viable Powerwashed
Type IV and Room Interfacial bond
Type V joint Mechanical temperature strength
materials roughened
Table 10 Time Dependent Material Property Testing

Material Curing Properties
Box beam concrete Room Compressive strength;
temperature tensile strength
Joint material Compressive strength,
Room
(Most viable Type IV and tensile strength,
temperature
Type V joint materials) shrinkage
Keyway interface
(Most viable Type IV and Room Interfacial bond
Type V joint materials and the temperature strength
best keyway preparation)
39
5.1.3 Crack Development and Resistance Investigation using Small-Scale Testing and FE
Simulations
Step B – Small Scale Testing
If early age cracks form in the keyway between adjacent box beams they primarily form due to
shrinkage which is sometimes referred to as temperature shrinkage, autogeneous shrinkage,
and/or drying shrinkage. This shrinkage is due to many factors including temperature during
curing, loss of water, and chemical reactions occurring during the hydration process. The crack
resistance of the keyway is dependent upon both the material characteristics and several bridge
details.
In step B, the objective is to experimentally investigate cracking (and collect information for
validation of FE modeling approaches) in the longitudinal joints by conducting small scale tests
of: (1) early age crack development and (2) the effectiveness of post-tensioning at closing cracks
prior to live load application. Note that during Step B the most viable Type IV and Type V joint
materials and the most viable keyway preparation identified in Step A will be utilized in the
keyways. Further, to preliminarily investigate the effectiveness of post-tensioning at closing
cracks developed during early age curing, post-tensioning will be applied after early age crack
development has been investigated and documented.
To complete Step B, small scale specimens consisting of two box beam sections will be
fabricated, instrumented, and tested in the Iowa State University Structural Engineering
Laboratory. The specimens will consist of two, side-by-side 2 ft long box beam sections (recall
the basic geometry mentioned previously). Two types of keyway geometries (i.e., the Type IV
and Type V as shown in Figure 11) will be evaluated. Due to material placement limitations
associated with the geometric width of the keyways, the materials selected during Step A will in
some ways drive the specific material/geometry combination. For example, epoxy grout and
non-shrink grouts would be used with the Type IV keyway and fiber reinforced concretes,
UHPC, and CIP concrete with Type K cement would be used with the Type V keyway. Again,
the materials and keyway preparation utilized will be based upon the results from the material
testing and selection process described in Step A.
(a) Type IV (b) Type V
Figure 11 Small Scale Specimen Keyway Geometries (all units are inches)
40
Table 11 Small Scale Specimens
Keyway
Joint materials Keyway preparation
geometry
Either power washed or mechanical
Type IV Either epoxy grout or non-shrink grout based upon
roughened based upon Step A results
Step A results
Either fiber reinforced concrete, UHPC, or CIP Either power washed or mechanical
Type V concrete with Type K cement based upon Step A
roughened based upon Step A results
results
The specimens will be constructed in two stages which fundamentally replicate the way adjacent
box beam bridges are constructed: the first stage is to fabricate the box beam sections and the
second stage is to place the joint material in the keyway. In the first step, the basic box beam
sections will be formed (including the interior void) using typical concrete forming techniques.
Prestressing strands, mild longitudinal reinforcement, stirrups, and ducts will be placed as shown
in Figure 12 (using the basic box beam geometry shown in Figure 10). Note that the specific
number of strands will be based the basic design for a 45 ft span adjacent box beam bridge
(based upon AASHTO LRFD bridge design specifications). In the second step, the box beam
sections will be placed side-by-side as shown in Figure 12 and Figure 13. Note that during
placement and curing of the joint material, the individual box beam sections will be restrained
from moving by applying a vertical “tie-down” force over the webs which are away from the
joint of interest (described in greater detail subsequently). Before placing the joint material, the
post-tensioning ties will be inserted into the ducts as shown in Figure 12. However, no post-
tensioning will be applied at this stage.
Keyway Dished plate and chuck
Post-tensioning ties
Figure 12 Basic Configuration for Step B Testing
During the early age-curing period, vertical restraining forces (as shown in Figure 14 which
induce a lateral frictional force between the box beams and the laboratory floor) will be applied
to simulate the transverse restraint provided by the bearings and other box girders. In this
configuration, the interaction of the two materials will be able to be discretely observed.
41
Following placement of the joint material, the material will be cured for 7 days. During curing,
cracking will be investigated in the following ways: (1) displacement transducers will be
mounted at the top of the keyway joint to measure any horizontal separation at the keyway-box
beam interface; (2) the formation of cracks during curing will be documented every 12 hours
using visual crack mapping techniques with crack widths measured and recorded.
After 7 days of curing, the specimen will be released from the vertical tie down forces and post-
tensioning will be applied as shown in Figure 14. For comparison purposes, the horizontal
relative displacement at the keyway-box beam interface will again be measured using the
previously mentioned displacement transducers. In addition, to measure strain changes in the
keyway and over the keyway-box beam interface during post-tensioning, strain transducers will
be installed at the top, mid-depth and bottom of the box beam cross-section as shown in Figure
14. To further investigate crack closure after post-tensioning, a water dam will be placed on the
top of the specimen and any observed leakage will be documented. An additional crack map
survey will be conducted after post-tensioning to document any visually apparent cracks in the
joint material or at the keyway-box beam interface.
Tie down Tie down Displacement

rod frame transducer
(a) Side view
Displacement
transducer
Tie down frame
(b) Top view
Figure 13 Small Scale Specimen Test Setup During Curing
42
Displacement
transducer
Strain transducer
(a) Side view
Displacement
transducer
Strain transducer
(b) Top view
Figure 14 Small Scale Specimen Test Setup during Post-tensioning
Step C – FE Element Simulations of Shrinkage Crack Formation and Closure

The objective of Step C is to develop and validate nonlinear, time-dependent FE modeling
approaches for predicting cracking in keyways and crack closure due to post-tensioning. To
accomplish this, nonlinear FE models of the Step B small scale specimens will be established
and validated against the experimental results obtained in Step B. The nonlinear, time dependent
material properties obtained in Step A will be utilized in the creation of the FE models. Note that
cracking experimentally observed and/or measured during curing and the experimentally
observed and/or measured crack closure during post-tensioning will both be compared with those
from the FE models as part of the validation and investigation of cracking process.
Two types of FE models (Type A and Type B) will be used to investigate cracking and crack
closure in the keyway, respectively. The Type A model will take advantage of the crack
prediction capability of the ANSYS SOLID65 element. Once cracking in the keyway is well
predicted using the Type A model, the Type B model will be used to simulate crack closure after
post-tensioning. It should be noted that in the Type B model, contact elements will be assigned at
the crack interfaces identified from the results of Step B and the results from analyses with the
Type A models.
To investigate the cracking mechanism during curing, Type A models will be established. The
Type A model is a nonlinear, time-dependent FE model established using ANSYS in the
43
following ways. The concrete portions of box beams and keyways will be modeled using brick
elements – SOLID65. During the simulations, the keyways will be connected to the beams at
common nodes. All internal reinforcement will be modeled by defining the reinforcement ratios
in those elements by taking advantage of the reinforcing-bar simulation capability of the
SOLID65 element. The pre-stress induced by the prestressing strands will be ignored because the
behavior of interest is transverse (and the specimens have limited length); however, the stiffness
of strands will be considered as part of the reinforcement ratio capability of the SOLID65
element. The shrinkage strain will be simulated by applying an artificial temperature change on
the solid elements, which can be expressed as
ΔTsh = εsh/α (1)
where, ΔTsh = artificial temperature change; εsh = the desired shrinkage strain; and α = the
coefficient of thermal expansion.
The nonlinear, time-dependent shrinkage properties of the concrete and the joint material used in
the FE model will be as determined from the basic material property testing conducted in Step A.
As a first approximation, uniform shrinkage will be applied to the FE models to check if
cracking and other behaviors can be well predicted through comparison with the experimental
results. It may be necessary to apply non-uniform shrinkage distributions (i.e., non-uniform
ΔTsh) to better predict the crack behavior of the keyway. Other time dependent material
properties including the tensile strengths of the concrete and joint materials, and the bond at the
interface will be assigned as obtained from our basic material testing conducted in Step A.
When creating the Type A models, the crack capability of the SOLID65 element will be
activated. The assigned maximum tensile stress in the concrete, the joint material, and at the
keyway interfaces will be equal to the tensile strength of the joint material, the tensile strength of
the concrete, and the bond strength at the interfaces determined in Step A. The comparisons of
cracking observed experimentally and analytically will include, but will not necessarily be
limited to, the time when the cracks form and the location of the cracks. Also any horizontal
separation at the keyway-box beam interface observed experimentally will be compared to the
results of the FE analysis. If needed, the modeling methodology will be modified until good
agreement exists between the experimental and analytical results. Through these comparisons the
nonlinear FE models for predicting early age shrinkage cracking will be calibrated.
Once the primary cracking interfaces are identified from the Type A models as well as the
experimental testing, Type B models will be established to simulate crack closure occurring
during post-tensioning. In the Type B models the transverse post-tensioning ties will be modeled
using truss elements – LINK8. The post-tensioning ties will be modeled to share common nodes
with the box beam elements at the contact interfaces. To model the post-tensioning forces, the
concept of a temperature change will be used for the truss elements – this approach is known to
be effective at simulating prestressing/post-tensioning forces. All portions of box beams and
keyways will still be modeled using brick elements – SOLID65. At the non-cracking interfaces,
the keyways will be connected to the beams at common nodes. At the cracking interface(s), 3-D
contact elements (CONTA173 and CONTA174) will be used to model the interaction at the
interface. Parameters such as the normal penalty stiffness factor (FKN), the penetration tolerance
factor (FTOLN), and the coefficient of friction will be input to allow compressive stress and
friction to develop at the cracking interface A cohesive zone material model will be defined for
the contact elements to simulate debonding at the cracking interface. The maximum debonding
44
stress for the contact element is equal to the lesser of the tensile strength of the concrete, the
tensile strength of the joint material and the bond strength between the two. Shrinkage will be
then applied to the FE models to simulate the crack openings during curing. Then simulation of
the post-tensioning will then be conducted to simulate crack closure until compressive stresses
are generated at the interfaces. The compressive stress contours will be plotted to check their
distribution of the post-tensioning force. The predicted crack closure from the Type B model will
be compared to that observed during the small scale testing. Again, strains in the keyway and
horizontal separation at the keyway found in the specimens and FE models will also be
compared and modifications to the modeling methodology made, if necessary.
5.1.4 Investigation of Cracking in Bridges with Variable Attributes

Step D: FE Simulations of Cracking in Bridges
The objective of Step D is to use the validated FE modelling approach to analytically investigate
cracking in adjacent box beam bridges with variable lengths, widths, skews, post-tensioning
details, diaphragm designs, differential cambers, and keyway geometry during bridge service
life.
Based upon the current AASHTO LRFD bridge design specifications, adjacent box beam bridges
need to be designed to satisfy both the ultimate strength design and service strength design
criteria. For these design considerations, the load combinations and load factors are well
described in the AASHTO LRFD bridge design specifications. However, as the objective of this
work ultimately results in the development of a crack-free bridge it seems clear that, although
typically thought of as a serviceability criteria, that the bridge should remain crack-free to the
ultimate load state. Thus, work in Step D will investigate cracking in bridges with variable
attributes when applied loads are multiplied by applicable load factors. Note also that although
all loads must globally be considered during design, here, only those loading conditions known
to induce cracking in the longitudinal joint need be considered. Thus, loads such as beam dead
load (note dead load due to the barrier will be considered as this results in an asymmetric loading
condition), which acts on the primary elements before the joint material is placed, will not be
considered and the primary focus will be on: (1) early age crack development resulting from
shrinkage, (2) cracking resulting from temperature changes, and (3) cracking resulting from live
loads.
To investigate cracking in Step D, FE simulations will be performed with the FE models of the
bridges established taking advantage of the validated FE modelling approach developed in Step
C. The concrete portions of the box beams, diaphragms, keyways, and barrier (when applicable)
will be modeled using brick elements – SOLID65. The beams will be connected to the diagrams
and the barriers (when applicable) by sharing common nodes. Initially, the keyways will be
connected to the beams using the Type A modelling techniques described above in Step C. Once
the primary cracking interface is identified from the preliminary model, contact elements will be
assigned at the cracking interface, (again, following the validated Type B modelling techniques
described in Step C). All internal reinforcement and presstressing strands will be modeled by
defining the reinforcement ratios in those elements by taking advantage of the reinforcing-bar
simulation capability of the SOLID65 element. When considered, differential cambers will be
simulated by assigning misalignment(s) of the shear keys in adjacent beams. Bearing conditions
will be modeled using horizontal and vertical springs, with the respective stiffnesses determined
45
based on typical material properties. The transverse post-tensioning ties will be modeled using
truss elements – LINK8. The post-tensioning ties will be modeled to share common nodes with
the box beam elements at the contact interfaces. And, as before, to model the post-tensioning
forces, the concept of a temperature change will be used for the truss elements.
The FE models will consider the loading conditions experienced at different stages of bridge life.
Based upon the AASHTO LRFD bridge design specifications there are three important stages
which should be considered:
 Stage 1 – Girder fabrication and erection: misalignment of shear keys can result from
differential cambers caused by prestressing effects (note this stage will be considered by
geometrically varying the location of the shear keys in adjacent beams);
 Stage 2 – Joint material early age creep and transverse post-tensioning application;
 Stage 3 – In-service loadings consisting of barrier loads, live loads, temperature gradient
effects, and shrinkage effects while in service.
During stage 1 differential camber can be induced. The primary impact of this differential
camber on adjacent box beam bridge performance is that it causes misalignment of the keyways.
The misalignment of keyways will be geometrically simulated to study the impact of differential
camber on crack formation and crack resistance.
At stage 2, the bridge system consists of the box beams, keyways, post-tensioning ties and
diaphragms. During stage 2 there are two important loading effects: shrinkage of the joint
material and the transverse post-tensioning. To study these, shrinkage effects will be considered
first by simulating the behavior of the system through the first 7 days of curing. Next the
transverse post-tensioning will be applied. Subsequently, additional shrinkage that occurs for the
following 21 days will be considered.
To evaluate early-age shrinkage, the material properties measured during Step A and the
validated modelling approach developed in step C will be utilized. Shrinkage strains after
grouting the keyways will then be assigned to the model elements to investigate cracking in the
keyways. The shrinkage strains will be estimated using the approaches validated with the small
scale test results and the small scale FE simulations completed in Step C. After 7 days of curing,
the post-tensioning will be analytically applied to the post-tensioning ties. If cracks were
observed to develop during the first 7 days, the effectiveness of the post-tensioning at closing the
cracks will be specifically observed. After the post-tensioning is applied, any additional
shrinkage of the joint material occurring will be simulated. As before, crack development during
this period will be of interest.
During stage 3 the barrier rail is added and the bridge is subjected to live loads and temperature
changes. First, the barriers will be added to the bridge models (both in terms of dead load and
stiffness) and the impact on cracking will be investigated. To apply live loads present during
stage 3, the AASHTO design load HL-93 will be applied on the bridge models considering
different longitudinal and transverse positions. This systematic live load placement will ensure
that worst case scenarios can be fully captured. The temperature gradients present during stage 3
will be assigned to the model elements using the temperature profiles specified by the AASHTO
LRFD bridge design specifications. Three different temperature induced conditions will be
analytically investigated. First, uniform temperature changes will be applied. Second, top-to-
46
bottom differential temperatures will be applied. Finally, side-to-side differential temperatures
will be applied. Considering all of the loadings, the individual contribution to internal stresses
and the superposition of multiple loading scenarios can be studied.
One obvious way to study the effectiveness of various post-tensioning schemes is by observing
the compressive stress contours in the keyways after post-tensioning (and after additional curing
of the joint material). These stress contours will be especially useful as the attributes affecting
the distribution of post-tensioning are studied as shown in Table 12. Many researchers [Ulku, et
al. 2010; Grace et al. 2012] have shown that post-tensioning effects are mainly localized at the
diaphragm regions due to shear lag effects. Further, Hansen et al. [2012] showed the potential
adequacy of a post-tensioned box beam system developed without diaphragms. Thus, to evaluate
the influence of the diaphragms on the post-tensioning distribution, two types of diaphragm
designs will be studied (i.e., Design I: diaphragms with ties and Design II: no diaphragm). To
determine the required number of ties per unit bridge length, three longitudinal distributions of
post-tensioning ties will be investigated as shown in Table 12 (i.e., 5 ft, 10 ft, and 25 ft). To
determine the required number of ties per cross-section, two types of vertical post-tensioning tie
configurations will be investigated as shown in Table 12 (i.e., 2 ties and 3 ties vertically).
Likewise, the required post-tensioning force, per tie, required to achieve a crack free system will
be determined. Two cases of differential camber will be considered by modelling bridges with
matched and misaligned shear keys as shown in Table 12. Note that required the number of
prestressing strands and dimensions of the box beams will be designed based upon the AASHTO
LRFD bridge design specifications.
Table 12 Step D Parametric Study Matrix

Post-tensioning ties
Model Diaphragm Span Width Skew
Alignment Number Magnitude (ft)
type design Spacing (ft) (degree)
vertically (kips)
(a) Design I:
Diaphragms
(a) Match (a) 5 ft
3D FE with ties (a) 2 To be (a) 50 (a) 30 (a) 0
(b) Misalignment (b) 10 ft
models (b) Design III: (b) 3 determined (b) 100 (b) 80 (b) 30
(1/2 in.) (c) 25 ft
No
diaphragm
5.1.5 Influence of Shear Key Geometry on Keyway Behavior

Step E: Medium Scale Testing
The objective of Step E is to experimentally study the influence of shear key geometry on
keyway behavior and crack development under early age shrinkage and live loads (e.g.,
AASHTO HL-93 loading).
To conduct Step E medium scale specimens will be fabricated, instrumented, and tested in the
Iowa State University Structural Engineering Laboratory. Two types of shear keys, based upon
the literature review, will be considered as shown in Figure 15. Note that the specific keyway
considered in this Step will be determined based upon the results of Steps B and C (either Type
IV or Type V). The test specimens will consist of two box beam sections placed side-by-side.
The length of the specimens will be determined based upon the FE simulations conducted in Step
D and will be approximately equal to the transverse spacing of the post-tensioning ties required
47
to eliminate cracking. The number of post-tensioning ties vertically and the presence of
diaphragms will, likewise, be determined based upon the results from Step D. Other details of the
medium scale specimens will be similar to those of the small scale specimens as shown in Figure
10 and Figure 12.
(a) Octagonal shear key (b) Triangular shear key

Figure 15 Shear Key Geometries for Medium Scale Specimens
The specimens tested in Step E will be constructed similar to that described in step B. Like the
evaluation completed in Step B, the keyway will be grouted and cured for 7 days. During curing,
crack development will be documented using visual crack mapping techniques. After the joint
materials reached the required strength, post-tensioning will be applied. To measure strain
changes in the keyway during post-tensioning, strain transducers will be installed at the top, mid-
depth and bottom of the box beam cross-section as shown in Figure 16. To investigate any crack
closure occurring after post-tensioning, a water dam will be placed on the top of the specimen
and leakage will be documented.
The medium scale specimens will be tested under loading conditions simulating that occurring
under the AASHTO HL-93 loading. The load applied to the specimen will be estimated based
upon the results of the FE simulation conducted in Step D (for the HL93 loading). The load will
applied to the top of one of the box beams in a transverse position that results in the moment and
shear combination estimated from step D as shown in Figure 16. A spreader plate will be placed
at the loading location to distribute the load and a load cell will be used to measure the applied
load as shown in Figure 16. During loading, displacement transducers and strain transducers will
be mounted at the top of the keyway joint to measure horizontal separation at the keyway top as
shown in Figure 16. Displacement transducers will be also installed at the bottom of the keyway
to measure the differential displacement between the boxes which would be indicative of a shear-
type failure of the joint (i.e., differential displacement). Strain transducers will be installed at the
top, mid-depth and bottom of the keyway to measure the strain change as shown in Figure 16.
Note that the strain transducers installed at the top of the specimen (see Figure 16(b)) will also be
used to investigate shear lag effects as the post-tensioning is applied.
48
Hydraulic jack Load cell
Displacement transducer
Spreader
plate
Strain transducer
Displacement transducer
(a) Side view
Strain transducer
Spreader plate
Displacement
transducer
(b) Top view

Figure 16 Medium Scale Specimen Test Setup
Step F: FE Simulations of Keyways with Different Shear Key Geometries

The objective of Step F is to validate FE modeling approaches for predicting the behavior of
joints with different shear key geometries under live loads.
FE models of the medium scale specimens will be established using the FE techniques developed
and validated in Step C. The models will be further validated (for live load response prediction)
through comparisons with the results of the medium scale testing. Specifically, comparisons of
the horizontal separation at the keyway top, differential displacements, and the strains at various
49
locations throughout the specimen will be made between the results from FE simulations and
testing.
5.1.6 Cracking and Resistance Investigation using Full-Scale Testing and FE Simulations
Step G: Design of Crack Free Bridges using Full–Scale Bridge Models
The objective of Step G is to design crack free bridges, that will be tested and evaluated in Step
H, based upon the experimental results and validated FE modeling techniques resulting from
Steps A-F. From the results of the experimental testing and FE simulations of the small scale and
medium scale specimens, the most viable keyway geometry, the most viable joint material, the
most viable shear key geometry, and most viable keyway preparation will have been identified.
These characteristics will be used to establish and design the bridge which will be evaluated in
the laboratory in Step H. However, prior to testing, FE models of a complete bridge will be
established to investigate the influences of transverse post-tensioning details and diaphragm
designs on the crack resistance of the keyways. The FE techniques developed and validated in
Steps C through F will be utilized in this FE model. The three stages of bridge life considered in
Step D will also be considered here (and in the same way as described previously). A crack free
bridge will be designed with the keyways well pre-compressed and possessing the required crack
resistance by completing a preliminary parametric study (principally on factors related to
transverse post-tensioning). Note that the typical cross-section of the box beam shown in Figure
10 will be utilized for the bridge design. Further note the results of Step G will be compared to
the results of Step H as described below as a further validation of the model predictions.
Step H: Full-Scale Testing and FE Model Validations

The objective of this step is to fabricate and test two full scale specimens and validate the FE
models against additional experimental results. The full-scale test program will consist of the
investigation of shrinkage during curing and the application of service loads and thermal loading
conditions to the two specimens. During curing, any crack formation will be investigated and any
subsequent crack closure during applying post-tensioning will be documented (as described
previously).
In both cases the specimens will be tested under simulated live load and thermal loads. The
specimens will be configured using the configuration of the crack free bridge designed and
analytically investigated in Step G using the most viable properties established in the previous
steps. The specimens will be 45 ft in length and consist of 3 adjacent beams and 2 keyways. The
principal difference between the two full-scale specimens will be the skew: one with 0 degree
skew and one with 30 degrees skew. The specimens will be fabricated following the two-stage
construction approach described in Step B. The 0 degree skew specimen testing will be
conducted first and then compared to the predictions from Step G. If necessary, based upon the
comparison between test and analysis results, modifications will be made to the design of the
specimen with 30 degrees skew to improve the design and performance.
As before, the keyways will be grouted and cured for 7 days. Cracking in the keyways during
curing will be documented using visual crack mapping techniques. Post-tensioning will then be
applied. As before, closure of any previously observed cracks will be documented. After post-
tensioning, a water dam will be placed on the top of the specimen and leakage will be checked.
50
Service load testing will be conducted with the basic test setup shown in Figure 17(a). Given the
bridge length, the critical live load design condition is when two axles are spaced 14 ft apart
(simulating two axles of the HL93 loading). Live loads will be applied at various longitudinal
and transverse locations and in combinations of one wheel line and two wheel lines. A variety of
live load placements will allow for the study of the most critical loading conditions.
Instrumentation, not unlike that utilized in the small and medium scale testing, will be deployed
throughout the test specimens to monitor for crack development and closure. Additionally, strain
and deflection transducers measuring the global response (i.e., at midspan, quarterspan, etc.) will
be installed and monitored during loading. The formation of cracks before, during, and after each
test will be documented using visual crack map techniques. Should any cracks develop, a repair
will be designed and implemented. The loading conditions causing initial cracking will be
replicated to, again, investigate cracking. The results of the first round of live load testing will be
compared to the predictions from Step G.
Next, the specimen will be subjected to thermal load testing. To conduct this testing, a thermal
enclosure will be constructed on top of the specimens as shown in Figure 17 (b). The enclosure
consists of foam insulation board and reflective foil which prevents heat from escaping from the
enclosure. The air in the enclosure will be continuously heated and circulated to maintain the
desired (and uniform) temperature. Note that thermistors will be installed during the thermal load
testing to record the temperatures at several locations in the beams which will provide data to
validate the model developed in Step G. Several different loading conditions will be considered,
including: (1) heating one cell of the enclosure, (2) heating two cells of the enclosure, and (3)
heating three cells of the enclosure. The formation of cracks before, during and after each test
will be documented using visual crack map techniques. The measurement and results during the
thermal testing will, again, be used to validate the FE models.
After the thermal load testing, the service load testing will be again conducted with the same test
setup previously used. The same load levels and positions will once again be considered.
Similarly, the same instrumentation plan will also be used. The formation of cracks before,
during and after each test will be documented using visual crack map techniques. The detected
cracks will be compared with those found in the previous service load test. The measurements
and other results during this period will be used to further validate the FE models.
51
Hydraulic jack Load cell
Spreader beam
Bearings
(a) Basic Service Live Load Test Configuration
Foam insulation board Heated air
(b) Thermal Load Test

Figure 17 Setup for Large Scale Testing
5.1.7 Parametric study

Step I: Determination of Required Details for Crack Free Bridges
The objective of Step I is to determine the required post-tensioning and other details that result in
crack free bridges for bridges with different span lengths, bridge widths, and skews via a
parametric study using the validated FE models. To accomplish this, a parametric study will be
performed using the validated FE models to determine the required number of post-tensioning
ties, the required magnitude of post-tensioning for each tie, and other details. It is anticipated that
at the conclusion of Step I that design guidance charts will be developed to determine the
required details for different bridge geometries. At this point in the experiment and analytical
investigations several factors have been extensively considered. These include:
 Differential camber
 Keyway shape
 Keyway configuration
 Shear key geometry
 Keyway preparation
 Keyway joint material
52
 Keyway curing methods
What remains is to establish design guidance on other details (e.g., post-tensioning information,
etc.) that are impacted by gross bridge geometrics. It is the establishment of this design guidance
that will be the focus of Step I. The basic modeling approach described above in Step D and
further refined as a result of subsequent Steps (e.g., Step G) will be utilized in the completion of
Step I. At this point in the work a highly validated modeling approach will have been developed
such that it predicts early age cracking, response to post-tensioning force application, and in-
service behavior under live loads and temperature variations. Thus, using the approach
described in Step D (including consideration of the AASHTO design specifications, etc.), the
required details will be determined for bridges with the following attribute ranges:
 Span: 0 ft to 100 ft
 Width: 30 ft to 80 ft
 Skew: 0 degree to 30 degree
5.2 Specification Development

At the conclusion of Step I the research team will prepare a detailed outline for proposed
AASHTO LRFD Specifications language, commentary, and guidelines that may need changed or
created. In order to facilitate a comparison of existing vs. proposed specifications, we anticipate
following the same order as the sections and subsections of the current AASHTO LRFD and
provide a side by side comparison of current vs. proposed code language wherever applicable.
An annotated outline format will be used to summarize the guidelines for design and
construction as well as those aspects of the design process to be included in the proposed
AASHTO Specification language. This task, and the subsequent development of guidelines and
specifications in the original RFP Task 11 are identified as the ultimate objective of the project
and as such will be given due consideration for completeness and usefulness in their final format.
The development of flowcharts for use in the design process are seen as very beneficial to
assisting the end user in finding their way through the necessary AASHTO Specification sections
that apply to this design. The research team anticipates developing a draft flow chart as part of
this task for consideration by the advisory panel.
In Phase IV, specific criteria for the design of connection details will be developed but is
anticipated (some of this depends heavily upon the results of the analytical and experimental test
results) to consider such things as:
 Material properties
 Loads
 Section Properties
 Prestressing requirements
 Mild steel reinforcing limits
 Limit states covering both temporary and service conditions
The development of detailed design examples for use by future designers will not likely present
many significant changes in the overall structural design of the major bridge components. It is
anticipated that three design examples will be developed as part of the proposed work. The
53
design examples will focus on the connection details between adjacent precast concrete box
beams. In order to accommodate a variety of girder configurations, we anticipate the
development of an automated design process using Mathcad and/or Excel software (depending
upon complexity of the design process either may be used; however, preference will be given to
using Mathcad). In order to facilitate acceptance of these design examples, we propose to follow
the format used by the PCI Bridge Design Manual, Chapter 9.7. These design examples are quite
familiar to experienced users within the industry and serve as a suitable format not only for
design examples but for actual user calculations as well.
Additional features of the design examples could include demonstration of any special LRFD
loadings (including thermal as necessary); load combinations; stress, fatigue, and strength
checks; demonstrations of the levels of analysis required to ensure long-life connection details
and other features unique to this structural system.
We anticipate developing a minimum of three examples to illustrate the design of adjacent box
beam bridges and more specifically, the design of the longitudinal joints and connections
between them. Our initial thought is that one example could include a lateral post-tensioning
system to maintain sufficient compressive force on the joints to minimize cracking in the joint
and surrounding concrete. A second example could include a non-post-tensioned system if
testing and analysis indicates that a reliable system can be standardized. A third example could
include the design of a multi-span continuous bridge with special attention given to continuity
and creep effects and subsequent moment redistribution.
In Phase V following review by the project advisory panel, the research team is prepared to
resolve the comments and update the previously drafted guidelines and AASHTO specification
language. The use of comment resolution forms will be used to document the individual
comments, research team concurrence with same, and any needed supporting information or
discussion.
54
6. Data Archiving and Sharing Plan
The data archiving and sharing plan consists of seven sections detailing how the data generated
in this effort and the effort that it has taken to gather data are available in easy to follow and
understand formats. The details of the seven sections are described as follows.
6.1 Background and Significance
The purpose of any data archiving and sharing plan is to ensure that data generated during a
project are maintained in perpetuity and that the procedures followed (which are frequently not
fully documented in archival reports) can be repeated by others in future. With the importance of
the work to be conducted under NCHRP Project 12-95, it will be important that the results and
conclusions could be validated by others – should the need arise. The archiving and sharing plan
proposed for NCHRP Project 12-95 ensures that the principal types of data will be accessible and
easily understood.
6.2 Expected Data Formats
The work to be conducted under this project will come from three primary sources. First,
technical literature and other information will be collected from published and unpublished
sources. These pieces of information will be stored in Portable Document Format (.pdf) with all
required reference information captured in the electronic files. Second, information on the
analytical modeling portion of the project will result in two types of information: (1) information
on the developed models (i.e., the inputs) and (2) results of the analytical models (i.e., the
outputs). Because commercially available software will be used for creating the models, the
input information will be stored in software’s native format. Results of the analysis will be made
available in two forms: (1) the software’s native format and (2) Microsoft Excel files. Third, data
and information from the experimental testing will be generated in multiple formats. For
example, the data from electronic sensors will be collected in electronic format. The raw data
files (in .dat or similar format) and those used for processing the data (in Excel or similar format)
will be archived. Further, photographs of all testing procedures will be stored in standard
photography format. Finally, the step-by-step testing procedures will be documented in
Microsoft Word (.doc) files. When testing follows standardized procedures (e.g., ASTM test
protocols) these procedures will be retained in pdf format.
6.3 Description of Data Archiving and Quality Assurance Plan
The data from technical sources, analytical modeling, and experimental testing will all be
recorded in electronic format. The data will come directly from the commercial software or from
test apparatus. These raw formats will be archived to ensure that future users can recreate actions
taken by the research team.
To insure data quality, it will be imperative that all test equipment is in good working order and
has been calibrated following manufacturer recommendations. Where possible, multiple “tests”
will be conducted to ensure repeatability of the results. No data will be excluded from the
archiving plan to ensure unbiased conclusions can be reached.
55
6.4 Description of Data Sharing Plan
Data from this project will not be released to the public before approval by the project panel.
Further, scholarly publications will only be submitted to journals and conference proceedings
upon approval from the NCHRP.
6.5 Schedule for Data Archiving and Public Release of Data
At the conclusion of each task where data are collected, the data will be archived to CD and
provided to the NCHRP. At the conclusion of the project (e.g., submission of the final report), a
CD (or DVD) with all project data will be provided to the NCHRP.
6.6 Milestones for the Implementation of the Plan
The Milestones pertinent to the Data Archiving and Sharing Plan will match the Milestones set
for the individual project Tasks. In other words, when a project task that includes the collection
of data is complete, there will be an archiving process completed in parallel.
6.7 Resources and Budget
Costs of collecting and archiving all project data are included in the budgets associated with the
respective tasks. No additional resources will be required to implement the plan.
56
7. Bibliography
Attanayake, U., and Aktan, H. (2008). “Issues with Reflective Deck Cracks in Side-by-Side Box
Beam Bridges.” Proceedings of the 2008 Concrete Bridge Conference, Federal Highway
Administration, National Concrete Bridge Council, Missouri Department of Transportation,
American Concrete Institute, 18p.
Attanayake, U., and Aktan, H. M. (2009). “Side-by-side Box-beam Bridge Superstructure:
Rational Transverse Post-tension Design.” Transportation Research Board Annual Meeting,
Washington D.C.
Badwan, I. Z., and Liang, R. Y. (2007a). “Transverse Post-Tensioning Design of Precast
Concrete Multi-beam Deck.” PCI Journal, 52(4), 84-92.
Badwan, I. Z., and Liang, R. Y. (2007b). “Performance evaluation of precast post-tensioned
concrete multibeam deck.” Journal of Performance of Constructed Facilities, 21(5), 368-374.
Dong, H., Li Y., and Ahlborn T.M. (2007). “Performance of Joint Connections between Decked
Prestressed Concrete Bridge Girders,” PCI National Bridge Conference, Proceedings,
Phoenix, Arizona.
El-Remaily, A., Tadros, M. K., Yamane, T., and Krause, G. (1996). “Transverse design of
adjacent precast prestressed concrete box girder bridges.” PCI Journal, 41, 96-113.
Fu, C. C., Pan, Z., and Ahmed, M. S. (2010). “Transverse Post-tensioning Design of Adjacent
Precast Solid Multibeam Bridges.” Journal of Performance of Constructed Facilities, 25(3),
223-230.
Grace, N. F., Jensen, E. A.,and Bebawy, M. R. (2012). “Transverse post-tensioning arrangement
for side-by-side box-beam bridges.” PCI Journal, 57(2), 48-63.
Greuel, A., Baseheart, T. M., Rogers, B. T., Miller, R. A., and Shahrooz, B. M. (2000).
“Evaluation of a high performance concrete box girder bridge.” PCI journal, 45(6), 60-71.
Gulyas, R. J., Wirthlin, G. J., & Champa, J. T. (1995). “Evaluation of keyway grout test methods
for precast concrete bridges.” PCI Journal, 40(1), 44-57.
Hanna, K. E. (2008). “Behavior of Adjacent Precast Prestressed Concrete Box Girder Bridges.”
PhD diss., University of Nebraska, Lincoln, NE.
Hanna, K.E., G. Morcous, and M. K. Tadros (2007). “Transverse Design and Detailing of
Adjacent Box Beam Bridges,” PCI National Bridge Conference, Proceedings, Phoenix, Ariz.
Hanna, K. E., Morcous, G., & Tadros, M. K. (2009). Transverse post-tensioning design and
detailing of precast, prestressed concrete adjacent-box-girder bridges. PCI journal, 54(4),
160-174.
Hanna, K., Morcous, G., & Tadros, M. K. (2011). Adjacent box girders without internal
diaphragms or post-tensioned joints. PCI journal, 56(4), 51-64.
Hansen, J., Hanna, K., and Tadros, M. K. (2012). “Simplified transverse post-tensioning
construction and maintenance of adjacent box girders.” PCI journal, 57(2), 64-79.
Harries K.A. (2006). “Full-scale Testing Program on De-commissioned Girders from the Lake
View Drive Bridge,” Report No. FHWA-PA-2006-008-EMG001, Pennsylvania Department
of Transportation, Harrisburg.
Huckelbridge Jr, A. A., and El-Esnawi, H. H. (1997). “Evaluation of Improved Shear Key
Designs for Multi-beam Box Girder Bridges.” No. FHWA/OH-97/009, report to Ohio
Department of Transportation, Case Western Reserve University, Cleveland, OH.
Huckelbridge Jr, A. A., El-Esnawi, H., and Moses, F. (1995). “Shear key performance in
multibeam box girder bridges.” Journal of Performance of Constructed Facilities, 9(4), 271-
285.
57
Illinois DOT (2008). “Concrete Deck Beams,” Guide Bridge Special Provisions (GBSP), No. 62,
Illinois Department of Transportation, Springfield.
Issa, M. A., do Valle, C. L. R., Abdalla, H. A., Islam, S., and Issa, M. A. (2003). “Performance
of transverse joint grout materials in full-depth precast concrete bridge deck systems.” PCI
Journal, 48(4), 92-103.
Kim, J. H. J., Nam, J. W., Kim, H. J., Kim, J. H., & KEUN, J. B. (2008). “Overview and
applications of precast, prestressed concrete adjacent box-beam bridges in South Korea.” PCI
Journal, 53(4), 83-107.
Lall, J., E.F. DiCocco, and S. Alampalli (1997). “Full-Depth Shear-Key Performance in Adjacent
Prestressed-Beam Bridges.” Special Report No. 124, Transportation Research and
Development Bureau, New York State Department of Transportation, Albany.
Lall, J., S. Alampalli, and E. F. DiCocco. (1998). “Performance of Full-Depth Shear Keys in
Adjacent Prestressed Box Beam Bridges.” PCI Journal, 43(2), 72-79.
Macioce, T.P., H.C. Rogers, R. Anderson, and D.C. Puzey (2007). “Prestressed Concrete Box
Beam Bridges—Two DOTs’Experience,” PCI National Concrete Bridge Conference,
Proceedings, Phoenix, Ariz.
Miller, R. A., Hlavacs, G. M., Long, T., and Greuel, A. (1999). Full-scale testing of shear keys
for adjacent box girder bridges. PCI Journal, 44(6), 80-90.
Nottingham, D. (1995). Discussion of “Evaluation of Keyway Grout Test Methods for Precast
Concrete Bridges,” by R.J. Gulyas, G.J. Wirthlin, and J.T. Champa, PCI Journal, 40(4), 98–
103.
Precast/Prestressed Concrete Institute (PCI) (1997; 2004), PCI Bridge Design Manual,
Precast/Prestressed Concrete Institute, Chicago, IL, 1997 (updated July 2004).
Precast/Prestressed Concrete Institute (PCI). (2003). PCI bridge design manual, 2nd Ed.,
Precast/Prestressed Concrete Institute, Chicago, IL.
Russell, H. G. (2009). Adjacent Precast Concrete Box Beam Bridges: Connection Details (Vol.
393). Transportation Research Board.
Sang, Z. (2010). “A Numerical Analysis of the Shear Key Cracking Problem in Adjacent Box
Beam Bridges.” Doctoral dissertation, The Pennsylvania State University.
Sharpe, G. P. (2007). “Reflective cracking of shear keys in multi-beam bridges”, Doctoral
dissertation, Texas A&M University.
Ulku, E., Attanayake, U., and Aktan, H. M. (2010). “Rationally Designed Staged Post-tensioning
to Abate Reflective Cracking on Side-by-Side Box-Beam Bridge Decks.” Transportation
Research Record: Journal of the Transportation Research Board, 2172(-1), 87-95.
Yamane, T., M.K. Tadros, and P. Arumugasaamy (1994). “Short to Medium Span Precast
Prestressed Concrete Bridges in Japan,” PCI Journal, 39(2), 74–100.
58
Appendix A - Project Survey to State DOTs
Questionnaire
Background
Bridges constructed with adjacent precast prestressed concrete box beams have been in service
for many years and provide an economical solution for short and medium span bridges. A
recurring problem is cracking in the longitudinal grouted joints between adjacent beams,
resulting in reflective cracks forming in the asphalt wearing surface or concrete deck. The
cracking appears to be initiated by stresses caused by temperature gradients, live loads,
transverse post-tensioning, or a combination thereof. Once the cracking has occurred, chloride-
laden water can penetrate the cracks and cause corrosion of the reinforcement and prestressing
strands.
A wide variety of practices, used by state highway agencies for the connection details between
adjacent box beams, include partial depth or full depth grouted keyways, keyways grouted before
or after transverse post-tensioning, prepackaged or non-prepackaged grout materials, post-
tensioned or non-tensioned transverse ties, a wide range of applied transverse post-tensioning
forces, and cast-in-place concrete decks or no decks. These practices are extensively summarized
in NCHRP Synthesis 39: Adjacent Precast Concrete Box Beam Bridges: Connection Details. The
objective of a current NCHRP project is to evaluate those practices at the design and
construction phases to eliminate cracking and leakage in the longitudinal joints between adjacent
boxes. This survey is intended to gather information on owner experiences with adjacent box
beam bridges.
General Questions
1. If we may contact you should we have additional questions, please provide your contact
information below:
Agency: ___________________________________________________________________
Address: ___________________________________________________________________
City: ___________ State: ______ Zip: ______
Primary person completing Questionnaire: _______________________
Current Position/Title: __________________________
Email: ________________________ Phone: _________________________
Date: ___________________
2. Does your agency use adjacent precast prestressed concrete box beams for bridges?
☐ Yes ☐ No
If yes, please answer the following questions. If not, thank you for your input.
3. Five generic types of common keyway geometries are illustrated on page 2. Do you currently
use a keyway geometry similar to the five generic types?
59
☐ Yes ☐ No
If yes, please describe the keyway detail or provide an internet link where we may obtain
detail information.
4. Based upon your agency’s experience, please rank the geometries on page 2 in order of
Type Evaluation from 1 to 5
I
II
III
IV
V
☐ Type I ☐ Type II ☐ Type III
Narrow Longitudinal Wide and deep longitudinal joint

joint (Cast-in-place concrete)

☐ Type IV ☐ Type V
Keyway Geometries
5. Do you have knowledge of another keyway geometry which performs better than (or varies
significantly from) those shown in the figure? If yes, please describe the keyway detail or
provide an internet link where we may obtain detail information
6. What is your greatest performance problem related to cracking and seepage at these joints?
What documentation do you have with regard to this problem (data, specifications,
60
7. Based upon your experiences, which are the best practices for keyway construction as related
to the box beam performance:
 Keyway preparation:
☐ No preparation
☐ “Power-washed rough” (clean and roughen the keyway faces of beams using
compressed air or water)
☐ “Mechanical rough” (sandblast (or other similar) the keyways to roughen joint
surfaces)
☐ Other _________________________
 Grout material: ☐ Non-shrink grout ☐ Mortar, epoxy grout or resin ☐ Concrete
topping ☐ Other _________________________
 Transverse post-tensioning: ☐ Yes ☐ No
If yes, what is the best construction sequence: ☐ Single-stage post-tensioning ☐ Multi-
stage post-tensioning (Note: Single-stage post-tensioning means all the box sections are
post-tensioned together transversely in one stage; Multi-stage post-tensioning means all
the box sections are post-tensioned together transversely in more than one stage.)
 Cast-in-place Deck: ☐ Yes ☐ No
8. Thank you. Once you click 'Done' you will have successfully completed the survey. If you
would like to send the Bridge Engineering Center a question, comment, or additional
information please feel free to include that below or contact Yaohua Deng via the email
address or phone number provided.
Yaohua Deng (Jimmy), Ph.D.

Postdoctoral Research Associate
Bridge Engineering Center, Institute for Transportation
Iowa State University Research Park
2711 South Loop Drive, Suite 4700
Ames, IA, 50010
Voice: (515)294-2882
Email: jimdeng@iastate.edu
Summary of Responses
61
Question 1
ID
Agency: Address: City, State, and Zip: Name: Current Position/Title: Email: Telephone: Fax: Date:
1035 PARKWAY EDDY GERMAIN, SUPERVISING eddy.germain2dot.s
1 NJDOT TRENTON, NJ 08625 609-530-2561 10/11/2013
AVENUE P.E. ENGINEER tate.nj.us
Maryland State
707 N. Calvert St. MS jrobert@sha.state.m
2 Highway Baltimore, MD 21202 Jeff Robert Senior Project Manager 410-545-8327 410-209-5002 10/10/2013
- C203 d.us
Administration
Michigan Dept. of Bridge Design Systems guerrazzis@michiga
3 425 W. Ottawa St. Lansing, MI., 48906 Sam Guerrazzi 517-335-3381
Transportation Engineer n.gov
1581 Mail Service
4 NCDOT Raleigh, NC 27699 Brian Hanks Project Engineer bhanks@ncdot.gov 919-707-6400 919-250-4082 10/02/2013
Center
Jefferson City, MO Senior Structural Darren.Kemna@mo
5 MoDOT 105 W. Capitol Ave. Darren J. Kemna 573-522-8725 573-526-5488 09/30/2013
65202 Designer dot.mo.gov
raymond.trujillo@st
6 NMDOT PO Box 1149 Santa Fe NM 87504 Ray Trujillo State Bridge Engineer 09/24/2013
ate.nm.us
7 NCDOT
Policies, Standards &
2300 South Dirksen gary.kowalski@illin
8 IL DOT Springfield IL 62764 Gary Kowalski Specifications Unit 217-785-2914 09/20/2013
Pkwy ois.gov
Chief
Vermont Agency Structures Project james.lacroix@state
9 1 National Life Drive Montpelier, VT 05633 James LaCroix 802-828-0047 09/20/2013
of Transportation Manager .vt.us
62
MnDOT-State Aid
3485 Hadley Ave State Aid Bridge dave.conkel@state.
10 for Local Oakdale, Mn, 55128 Dave Conkel 651-366-4493 651-366-4497 09/20/2013
North Engineer mn.us
Transportation
Bridge Design Section john.holt@txdot.go
11 TxDOT 125 E. 11th Street Austin, TX 78701 John M. Holt 512-416-2212 09/25/2013
Director v
Massachusetts
Alexander K. alexander.bardow@
12 Department of 10 Park Plaza Boston, MA 02116 State Bridge Engineer 857-368-9430 857-368-0636 09/20/2013
Bardow state.ma.us
Transportation
tim.keller@dot.state
13 Ohio DOT 1980 West Broad St. Columbus Ohio 43223 Tim Keller State Bridge Engineer 614.466.2463 09/19/2013
.oh.us
1900 Kanawha Blvd William.H.Varney@
14 WVDOT Charleston, WV 25302 Bill Varney Sr Bridge Engineer 3045589490 09/19/2013
E wv.gov
Question 1
ID
Agency: Address: City, State, and Zip: Name: Current Position/Title: Email: Telephone: Fax: Date:
4802 Sheboygan Madison, WI, 53707- Chief, Structures william.oliva@dot.w
15 Wisconsin DOT William Oliva 608-2666-0075 608-266-5166 09/18/2013
Ave, Room 601 7916 Development i.gov
ahmad.abu-
Ahmad Abu- Chief Structural
16 Iowa DOT 800 Lincoln Way Ames, Iowa 50010 hawash@dot.iowa. 515-239-1393 515-239-1978 09/18/2013
Hawash Engineer
gov
PO Box 778, 800 Bay jason.hastings@sta
17 Delaware DOT Dover, DE 19903 Jason Hastings Bridge Design Engineer 302-760-2310 302-739-2217 09/18/2013
Road te.de.us
NYSDOT Office of Manager, Concrete mtwiss@dot.state.n
18 50 Wolf Road Albany, NY 12232 Michael Twiss 518-457-4534 518-457-6010 09/18/2013
Structures Engineering y.us
Nebraska
Lincoln NE PO BOX Assistant State Bridge fouad.jaber@nebra
19 Departmnet of 1500 highway 2 FOUAD JABER 402-479-3967 402-479-3752 09/26/2013
94759 Engineer ska.gov
roads
Oklahoma
Oklahoma City, Assistant Bridge
20 Department of 200 NE 21st St Walter Peters wpeters@odot.ort (405) 521-2606 (405) 522-0134 09/18/2013
Oklahoma, 74105 Engineer - Maintenance
Transportation
South Carolina Structural Design bowersbw@scdot.o
21 P. O. Box 191 Columbia, SC 29202 Barry Bowers 803-737-4814 803-737-0608 09/18/2013
DOT Support Engineer rg
State Structures Design robert.robertson2@
22 Florida DOT 605 Suwannee St Tallahassee, Fl 32301 Robert Robertson 850-414-4267 09/18/2013
Engineer dot.state.fl.us
63
Tennessee Dept. Suite 1100 James K. Director, Division of wayne.seger@tn.go
23 Nashville, TN 37243 Wayne J. Seger 615-741-3351 615-741-7745 09/17/2013
Of Transportation Polk Building Structures v
Kansas
700 SW Harrison Chief, Bureau of
24 Department of Topeka KS 66603 Loren Risch loren@ksdot.org 785.296.423 09/17/2013
Street Structures and Geotech
Transportation
Minnesota
3485 Hadley Avenue State Bridge Design arielle.ehrlich@stat
25 Department of Oakdale, MN 55128 Arielle Ehrlich 651.366.4506 651.366.4497 09/17/2013
North Engineer e.mn.mus
Transportation
South Dakota
700 E Broadway kevin.goeden@stat
26 Department of Pierre, SD 57501 Kevin Goeden Chief Bridge Engineer (605)773-3285 (605)773-2614 09/17/2013
Avenue e.sd.us
Transportation
601 Kamokila Blvd., paul.santo@hawaii.
27 Hawaii DOT Kapolei, HI 96707 Paul Santo Bridge Design Engineer 808-692-7611 808-692-7617 09/17/2013
Rm. 611 gov
Assistant State
sam.fallaha@dot.st
28 FDOT 605 Suwannee St. Tallahassee, FL 32399 Sam Fallaha Structures Design 850-921-7111 09/17/2013
ate.fl.us
Engineer
ID Agency: Question 2 Question 3 Question 4 Question
5
Response Response Keyway Detail Type Type Type Type Type
I II III IV V
1 NJDOT Yes
2 Maryland State No
Highway
3 Michigan Dept. Yes Yes Type III 3 3 3 3 3
of
Transportation
4 NCDOT Yes Yes Keyway detail is similiar to the Type III shown below. 4 4 3 1 1
https://connect.ncdot.gov/resources/Structures/Structure%20Specs/pcbb2_12.pdf
5 MoDOT Yes Yes Type III used for most of our in-service adjacent box-beams (similar to Illinois and 4 4 4 4 4
South Carolina DOT Geometry). Illinois detail primarily used in-house and the
South carolina detail used in our Safe-n-Sound project bridges. Recently used a
Type IV configuration with 1 1/4" opening for top 4" and a 2 1/2" width for rest of
depth that ends at top of bottom slab. This bridge was opened to traffic recently.
6 NMDOT Yes Yes Typically use Type III. Have not used the other keyway types which is why I 4 4 3 4 4
answered #4 to question 4 on the other keyway types below.
7 NCDOT Yes
8 IL DOT Yes Yes we use the Type 3 version shown below, we have 2 versions one for shallow 5 5 4 2 2
beams and one for deep beams. seee the following link for base sheet details
http://www.dot.il.gov/cell/PPC_deckbeam.pdf
9 Vermont Yes Yes Our keyway geometry is as depicted in TYPE IV below. 3 3 3 2 3
Agency of
Transportation
10 MnDOT-State Yes Yes We use a 1-1/2" minimum full depth keyway, similar to a type IV 5 5 5 1 1
Aid for Local
Transportation
11 TxDOT Yes Yes Similar to Type V. 5 5 5 5 4
12 Massachusetts Yes Yes It is Type IV. The details can be found 5 5 4 1 1
Department of http://www.mhd.state.ma.us/default.asp?pgid=bridge/bridgemanual_04&sid=about
Transportation . The detail to download is 4.2.8. We started using this detail in 2005, so we don't
have as much track record. We also require a minimum 5 inch topping slab with
single layer of reinforcement. The details of this topping slab can also be found in
Chapter 4 at the link specified above.
13 Ohio DOT Yes Yes Type III 4 4 4 4 4
14 WVDOT Yes Yes 3/4" opening at top flares out to 1-1/2" similar to Type III. The keyway goes about 3 3 3 3 3
8" to 12" down from top of beam depending on the depth of beam.
15 Wisconsin DOT Yes Yes Our Keyway is simular to type II and Type III. Please see our Standard 19.51 & 5 4 4 2 2
19.52. http://on.dot.wi.gov/dtid_bos/extranet/structures/LRFD/standards.htm
16 Iowa DOT No
17 Delaware DOT Yes Yes Type III 1 1 1 1 1
18 NYSDOT Yes Yes Full depth keyway: https://www.dot.ny.gov/main/business- 5 5 5 2 3
Office of center/engineering/cadd-info/bridge-details-sheets-repostitory-usc/BD-PA7E.pdf
Structures
19 Nebraska Yes Yes our existing details is type III. We don't use this type of structure anymore 3 3 3 3 3
Departmnet of
roads
20 Oklahoma No
Department of
Transportation
21 South Carolina Yes Yes http://www.scdot.org/doing/technicalPDFs/bridgeDrawings/CSDetails.pdf 4 4 4 3 3
DOT
22 Florida DOT No
23 Tennessee Dept. Yes Yes Type III is the detail used. While this type of bridge superstructure is not largely 4 4 4 3 2
Of used in Tennessee, those that are in place have the type III grout keyway.
Transportation
24 Kansas No
Department of
Transportation
25 Minnesota No
Department of
Transportation
26 South Dakota No
Department of
Transportation
27 Hawaii DOT Yes Yes We actually have not used adjacent box sections but it is being proposed for a 1 1 1 1 1
bridge project. The difference is we are adding a reinforced concrete topping of at
least 6 inches. We don't anticipate to have the cracking issues experienced
without this topping. Remaining questions will be answered only so this
comment can be read but responses should be ignored.
28 FDOT No
64
ID Agency: Question 6 Question 7 Question 8
Open-Ended Response Keyway Grout Transverse Cast-in-place Other (please specify) Open-Ended Response
preparation material posttensioning Deck
1 NJDOT
2 Maryland State
Highway
3 Michigan Dept. Long term performance. No preparation Non-shrink Single-stage post- Yes
of grout tensioning
Transportation
4 NCDOT Greatest performance problem-Seepage leads to corrosion of P/T tendon No preparation Non-shrink Multi-stage post- No
connecting the units. Documentation is limited. We believe cracking is related grout tensioning
to shear key location and possibly the order of construction. Grouting the shear
keys occurs after the tensioning the P/T strands.
5 MoDOT We have gone through many stages that directly impact the performance of “Mechanical Non-shrink Multi-stage post- Yes We have found adjacent box beams having a CIP Deck
seepage and cracking that are independent of the keyway. We started our safe- rough” grout tensioning with two mats of reinforcing perform better than CIP
n-sound project with tie-rods and a tar/roofing felt topcoat (Colorado DOT spec) decks with a single layer. We are trying Multi-Stage post
with asphault. Leakage would be rated a 5. We then did some post-tensioning tensioning on the project that also utilizes the Type IV
and the leakage would be rated as a 4. We then swapped the tar topping with a keyway.
poly-urea coat and leakage was helped considerably (rating = 2). Reflective
(longitudinal) cracking was present with essentially all Safe-n-Sound adjacent
box beam bridges.
6 NMDOT The joint failures we have seen have been on bridges overlaid with asphalt and “Mechanical Mortar, Multi-stage post- Yes Typically have the fabricator roughen the keyway either I would like a copy of your
with the following conditions: 1. Boxes were not post-tensioned together 2. rough” epoxy grout tensioning through sandblasting. Have had success with non-shrink survey results to see how other
Boxes are post-tensioned however the ducts were not grouted. 3. Boxes were or resin grout as well. states are responding and what
post-tensioned together with one long strand, hence not enough friction between practices they have.
the interfaces of the boxes.
7 NCDOT
8 IL DOT the greatest performance problem would be a shorter design life. No readily “Mechanical Mortar, Multi-stage post- Yes
available data, but we have replaced a considerable amount of these type rough” epoxy grout tensioning
structures in the past 10 yrs, all originally built in the 1960s-70s the cracks are or resin
65
related to loading and bearing rocking or uneven bearing.
9 Vermont Typically we have cast-in-place concrete overlays over these structures so in “Power-washed Non-shrink Multi-stage post- Yes
Agency of general there are not performance issues. rough” grout tensioning
Transportation
10 MnDOT-State We have only built 6 adjacent box beam bridges on the local system to date with “Power-washed Non-shrink Single-stage post- Yes We transverse post-tension at the bridge ends and at 1/4
Aid for Local detail type IV . The earliest being constructed in 2008. To date, we have no rough” grout tensioning points with 3- (1/2" 270 ksi pt strands) in grouted 4-1/2"
Transportation record of leakage problems. diameter tubes.
11 TxDOT Amount and location of transverse post-tensioning has varied over the years. Other Concrete Single-stage post- Yes We treat it like any other formwork that we place
Current details appear to have adequately addressed cracking/leakage. topping tensioning concrete against—wet it down thoroughly
12 Massachusetts Previously we used the Type III detail without a topping slab. The problems we “Power-washed Mortar, Multi-stage post- Yes We post tension the transverse strands to an initial 5,000
Department of had associated with cracking and seeping included corrosion of the transverse rough” epoxy grout tensioning pound prior to grouting. After grouting and after the grout
Transportation reinforcing bars and of the corner prestressing strands along the joints which or resin has set, we tension the strands to the final specified force.
results in concrete spalling. In advanced deterioration conditions, bigger pieces
of the concrete beam have broken off, prestressing strands have corroded
through and are hanging down. These problems are documented in the bridge
inspection reports for these bridges. Road salt and high ADT seem to the joint
cracking problem and contribute to the corrosion and deterioration problems.
13 Ohio DOT ODOT research shows that these cracks accur prior to opening the bridge to No preparation Non-shrink Single-stage post- Yes Please allow an answer of N/A
traffic. When we use asphalt wearing surface on top of a waterproofing grout tensioning on your questionaire.
membrane, on many bridges the membrane fails and the joint has excessive
leaking. The concrete will spall and the prestressing wire is exposed. NOTE:
IN ORDER TO CONTINUE FILLING OUT THIS FORM, I HAD TO
ANSWER QUESTION 4 FOR ALL TYPES. WE DO NOT HAVE
EXPERIENCE WITH ALL TYPES. DISREGARD MY ANSWER FOR
TYPE I, II, IV, & V. YOU SHOULD HAVE GIVEN ME THE OPTION OF
N/A.
14 WVDOT truck loads and salt exposure. No preparation Non-shrink Single-stage post- No
grout tensioning
ID Agency: Question 6 Question 7 Question 8
Open-Ended Response Keyway Grout Transverse Cast-in-place Other (please specify) Open-Ended Response
preparation material posttensioning Deck
15 Wisconsin DOT We experience reflective cracking that allows salts to penitrate between box No preparation Non-shrink Single-stage post- No We have experimented with concrete decks with ingle Dr. Deng, we also have some
section and cause deterioration of sections grout tensioning layer of reinforcement with some success. simular research underway
through our Wisconsin Highway
Research Program (WHRP)
being conducted by Western
Michagin Univeristy by: Upul
Attanayake, Ph.D., P.E.
Assistant Professor Department
of Civil & Construction
Engineering Western Michigan
University Office: (269) 276-
3217 Fax: (269) 276-3211
URL:
http://homepages.wmich.edu/~ua
ttanayake/
16 Iowa DOT
17 Delaware DOT We use a type III keyway with post-tensioning and steel plate shear connectors “Power-washed Mortar, Single-stage post- Yes Steel plate shear connectors at 4' spacing along the joints.
at 4' spacing along the joints. We most commonly use a CIP composite deck, rough” epoxy grout tensioning
although we have a few installations with a membrane and HM overlay. We or resin
have recently used epoxy grout on a few installations and they worked out well.
Our biggest issue has been deck cracking, an issue that extends beyond adjacent
box beams.
18 NYSDOT Reflective cracking in deck above the keyways. Chlorides (road salt) penetrate “Mechanical Other Multi-stage post- Yes Shear key grout used for adjacent box beams must meet
Office of the cracks, eventually reaching the beams. If left unchecked, the result is deck rough” tensioning the requirements of Section 709-06 of NYSDOT material
Structures deterioration, and eventually beam deterioration. These issues are documented specifications. I am also providing a link to the Approved
in oue bi-anual inspection reports. List for shear key grouts showing all the grouts that have
been tested and found to be in compliance with 701-06:
66
https://www.dot.ny.gov/divisions/engineering/technical-
services/technical-services-repository/alme/pages/230-
1.html
19 Nebraska They don't have full moment connection. They will crack if they don't followed “Power-washed Concrete Multi-stage post- Yes
Departmnet of by posttensioning. They will leak without an overlay. hard to inspect from inside. rough” topping tensioning
roads
20 Oklahoma
Department of
Transportation
21 South Carolina Concerned about long-term performance of the superstructure. “Mechanical Non-shrink Single-stage post- No
DOT rough” grout tensioning
22 Florida DOT
23 Tennessee Dept. Greatest problem is seepage between beams promotes deterioration of the sides “Mechanical Mortar, None Yes
Of of the beams and prevents visual monitoring of damage. Bridge Inspection rough” epoxy grout
Transportation reports. I believe that cracking is due to differential deflection of the beams. or resin
Load distribution rods between beams are not always installed.
24 Kansas
Department of
Transportation
25 Minnesota
Department of
Transportation
26 South Dakota
Department of
Transportation
27 Hawaii DOT N.A. Other Non-shrink None Yes N.A.
grout
28 FDOT
Connection Details
NC Similar to the Type III
https://connect.ncdot.gov/resources/Structures/Structure%20Specs/pcbb2_12.pdf
0.375"
6"
0.375
0.75" 6"
0.75"
MO Type III, recently Type IV configuration with 1 1/4" opening for top 4" and a 2 1/2" width
for rest of depth
MN Similar to Type IV, 1-1/2" minimum full depth keyway
IL Type III, http://www.dot.il.gov/cell/PPC_deckbeam.pdf
67
Transverse tie rods just below shear key, parallel to skew, tighten to snug fit.
MA Type IV, http://www.mhd.state.ma.us/default.asp?pgid=bridge/bridgemanual_04&sid=about, Detail
4.2.8, topping slab details in Chapter 4
0.375" 4"
0.375"
0.75"
0.75" 4"
0.75" CHAMFER
Transverse post-tension ties at ends and midspan for spans less than 50 ft, additional tie at
quarter point if greater than 50 ft (at mid-depth). Ties tensioned to 5 kips, keyways filled with
mortar, then ties tensioned to 44 kips.
WV Type III, 3/4" opening at top flares out to 1-1/2", goes about 8" to 12" down from top
WI Similar to Type II and III, http://on.dot.wi.gov/dtid_bos/extranet/structures/LRFD/standards.htm,
Standard 19.51 & 19.52.
3"
4"
0.375"
68
Mid-depth, parallel to skew post-tensioning, joints grouted before post-tensioning to 86.7 kips,
grouted post-tensioning ducts,
NY Full depth keyway, https://www.dot.ny.gov/main/business-center/engineering/cadd-info/bridge-
details-sheets-repostitory-usc/BD-PA7E.pdf
MIN. 1.5" MAX.2"
SHEAR KEY TO BE FILLED WITH MATERIAL

MEETING THE REQUIREMENTS OF SECTION
4.5.2 AND PLACED IN ACCORDANCE WITH
SECTION 8.4.5.4 OF THE PRESTRESSED
CONCRETE CONSTRUCTION MANUAL (PCCM)
CENTER LINE OF
SPRAY WITH EXPANDABLE FOAM
3" DIA.PRECAST
HOLESEE SHEAR
KEYAT TRANSVERSE SELF ADHESIVE
TENDON DETAIL COMPRESSIBLE SEALER
3"
BOTTOM OF SHEAR KEY
Post-tensioning to 84 kips at ends and midspan for spans less than 50 ft, additional placement at
quarter points if greater than 50 ft, parallel to skew
SC http://www.scdot.org/doing/technicalPDFs/bridgeDrawings/CSDetails.pdf
0.375"
0.375" 3"
0.75" 4"
0.75"
Cored slab, transverse tie rods tightened before grouting, asphalt wearing surface
69
Appendix B - Focus Group
Notes from Focus Group for NCHRP 12-95
September 27, 2013
Attendees:
Brian Jacob – Cramer & Associates
Steve Kunz – Shuck Britson
Dennis Drews – Coreslab
Ahmad Abu-Hawash – Iowa DOT
Phil Rossback – HDR Engineering
Dan Timmons – Rasmussen Group
Fouad Jaber – Nebraska Department of Roads
Mike LaViolette – HDR Enginering
Brent Phares
Lowell Greimann
Yaohua Deng
The following brief notes were taken during the Focus Group meeting. They are listed in
chronological order. A topical summary is included in the body of the report.
 Need to address bearing conditions to eliminate beam racking/rolling/sweep.
 Use 3 point bearing. (Texas)
 Railroads are very critical of tolerances at the bearings to ensure good initial fit-up. Use
ballast and do not tie girders together.
 Not used much on highways in Midwest
 One type of keyway would be more efficient for everyone.
 Differential camber was an issue during construction of various bridges in MO
 Nebraska Department of Roads (NDOR) doesn’t use; NE counties do but without P/T
 Inverted T-system in France
 Virginia DOT information on details
 Is a web-to-web shear connector a good idea? Does it give you added strength? Bars
projecting through side walls not a good idea for precasters. Some sort of threaded insert
completed in the field might work.
 Structural topping—usually not bars on top of girder for composite action. Usually only one
mat of reinforcement in topping
 Lots of variability in force levels: 120k in MI; 30k in NC
 Several variations in locations, e.g., four locations for 50 feet and under
 Why not just let water flow through the joints via weep points? Need to make sure water
can’t wick back up.
 Acquisition of strand is an acquisition headache – so contractors prefer rods. But, strand
easier to deal with crown and/or misalignment
70
 Would it be beneficial to match-cast the keyway?
 How about embedding a plate in each girder to which a welded connection could be added in
the field?
 P/T should be placed after grout is placed
 Some P/T before placement to align, then grout, then final P/T
 Easier to construct with P/T parallel to skew
 Grout material— high strength grout, concrete, high strength concrete options. Grout is
expensive for high volume gaps and the material selection may depend on size of gap at top.
 If use a concrete mix in the joint the availability of HPC can be an issue.
 For bridges with heavy skew, should install a sacrificial shoe so that once released the end
doesn’t get damaged due to camber.
 Use grinding to apply texture and eliminate differential camber.
 Probably need topping on paved roads for salt protection.
 Thin epoxy overlay
 We should investigate what policies/procedures have changed over time and what the owner
experiences have been.
 Question asked: what is the shear lag on the applied transverse P/T?
71

Connection Details for Adjacent Precast Box Beam Bridges

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Connection Details for Adjacent Precast Box Beam Bridges

Uploaded by

Copyright:

Available Formats

NCHRP

Connection Details of Adjacent

TRANSPORTATION RESEARCH BOARD

This Work Plan is furnished only for review by members of the

Type I Type II Type III

Wide and deep longitudinal joint

2.2 Best Practices

(a) Total number of adjacent box beam bridges

4.2 Publications Before 2008

Take Away Points:

(a) Joint A (b) Joint B (c) Joint C

(c) Full depth keyway (d) Full depth keyway

4.3 Publications After 2008

Take Away Points:

Figure 5 Typical Michigan Keyway Geometry and Post-tensioning

(a) No shear key (b) Top shear key

(c) Mid-depth shear key (d) Bottom shear key

(a) Isolated bearing pad (b) shared bearing pad

(b) Narrow joint

Figure 9 Connection Details Proposed by Hansen et al. [2012]

Take Away Points:

Partial depth keyway and Post-tensioning

Two post- Sandblast,

Badwan and Full depth keyway and Bonded post- Post-tensioning

Full depth and mid-depth

Transverse Tie Keyway Bearing Construction Concrete FE Laboratory Field

Partial depth keyway and

Type R-2, which is

FEA: Full depth keyway

Badwan and Solid Solid Common Link Simply

Dong et al. Solid Solid Common Concentrated

Attanayake Integrate Integrate Integrate

Beam Common Common Beam Simply

Fu et al. Solid Solid Contact Link Simply

Direct Current Foil-

Set 45: 5.8 ksi;

NG - Not Given; NA - Not Applicable

No. 1 1 32.5 20 44 11 48 × 17 NG Relative

A typical/common box beam cross-section used by multiple State Departments of Transportation

Fiber reinforced concrete, Room

Table 9 Keyway Interface Preparation Testing

Table 10 Time Dependent Material Property Testing

(a) Type IV (b) Type V

Keyway Dished plate and chuck

Tie down Tie down Displacement

(a) Side view

Tie down frame

(b) Top view

Figure 13 Small Scale Specimen Test Setup During Curing

(a) Side view

Figure 14 Small Scale Specimen Test Setup during Post-tensioning

Step C – FE Element Simulations of Shrinkage Crack Formation and Closure

5.1.4 Investigation of Cracking in Bridges with Variable Attributes

Table 12 Step D Parametric Study Matrix

5.1.5 Influence of Shear Key Geometry on Keyway Behavior

(a) Octagonal shear key (b) Triangular shear key

(a) Side view

(b) Top view

Step F: FE Simulations of Keyways with Different Shear Key Geometries

Step H: Full-Scale Testing and FE Model Validations

(a) Basic Service Live Load Test Configuration

Foam insulation board Heated air

(b) Thermal Load Test