MD 15 SP309B4M - Serviceability Related Issues Bridge Live Load Deflection - FinalReport
MD 15 SP309B4M - Serviceability Related Issues Bridge Live Load Deflection - FinalReport
MD 15 SP309B4M - Serviceability Related Issues Bridge Live Load Deflection - FinalReport
RESEARCH REPORT
DR. CHUNG C. FU
GENGWEN ZHAO
YUNCHAO YE
FAN ZHANG
University of Maryland
SP3098B4M
FINAL REPORT
June 2015
The contents of this report reflect the views of the author who is responsible for the facts
and the accuracy of the data presented herein. The contents do not necessarily reflect
the official views or policies of the Maryland State Highway Administration. This report
does not constitute a standard, specification, or regulation.
Technical Report Documentation Page
1. Report No. 2. Government Accession No. 3. Recipient's Catalog No.
MD-15-SP309B4M
4. Title and Subtitle 5. Report Date
June 2015
Serviceability-related Issues for Bridge Live Load Deflection and 6. Performing Organization Code
Construction Closure Pours
7. Author/s 8. Performing Organization Report No.
Dr. Chung C. Fu; Gengwen Zhao; Yunchao Ye; Fan Zhang MD-15-SHA-UM-3-12
12. Sponsoring Organization Name and Address 13. Type of Report and Period Covered
Final Report
Maryland State Highway Administration 14. Sponsoring Agency Code
Office of Policy & Research
707 North Calvert Street
Baltimore MD 21202
This study investigated the design criteria and practices in an effort to improve the quality of bridge designs in the
State of Maryland and beyond. This first criterion investigated was the live load deflection for steel bridges. The
second design/construction criterion investigated was designing and detailing bridge deck closure pours.
Previous and current practices and future planning on the serviceability of bridges have been documented.
State-of-the-practice methods from federal and other state agencies were collected. Three bridges were chosen for
refined analyses to investigate the live load deflections. Field measurements for these three bridges were collected
from the research team to facilitate this study. Thirty steel girder bridges from the Maryland State Highway
Administration’s (SHA) inventory were selected for statistical analyses. Steel bridges designed with the live load
deflection limit have been evaluated. Closure-pour analyses were conducted by line-girder models,
two-dimensional grid models or three-dimensional finite element models. All three methods generate accurate
enough camber diagrams to predict differential deflections between stages for straight girder systems, if creep is
not considered. Creep effect could be alleviated by proper camber and scheduling on pouring.
17. Key Words 18. Distribution Statement: No restrictions
Live Load Deflection, Construction This document is available from the Research Division upon request.
Closure Pours
19. Security Classification (of this report) 20. Security Classification (of this page) 21. No. Of Pages 22. Price
None None 77
Form DOT F 1700.7 (8-72) Reproduction of form and completed page is authorized.
Table of Contents
ii
List of Tables
Table 2.1 Historic Depth-to-Span, D/L, Ratio for Highway Bridges ........................................ 4
Table 3.1 Multiple Presence Factors in ASD Method ............................................................. 17
Table 3.2 Multiple Presence Factors in the LRFD Method ..................................................... 18
Table 3.3 Loads and Factors considered in ASD and LRFD Methods .................................... 18
Table 4.1 Beam Sections.......................................................................................................... 23
Table 4.2 Dead Load Information of Route1 Bridge ............................................................... 27
Table 4.3 Dead Load Information of the I-95 over Patuxent River Bridge ............................. 30
Table 4.4 Beam Sections.......................................................................................................... 32
Table 4.5 Live Load Deflection of Three Representative Bridges .......................................... 34
Table 4.6 Live Load Deflection of Three Bridges with Adjustment ....................................... 35
Table 5.1 Single-span Bridges Live Load Deflection .............................................................. 39
Table 5.2 Two-span Bridges Live Load Deflection................................................................. 40
Table 5.3 Three-span Bridges Live Load Deflection............................................................... 42
Table 5.4 Single-span Bridges Live Load Deflection with Modified Distribution ................. 44
Table 5.5 Two-span Bridges Live Load Deflection with Modified Distribution .................... 45
Table 5.6 Three-span Bridges Live Load Deflection with Modified Distribution .................. 47
Table 6.1 Maximum Displacements Comparison for Normal and Staging Models by DESCUS
.................................................................................................................................................. 57
Table 6.2 Comparison Table between Different Models with Diaphragms ............................ 57
Table 6.3 Comparison Table between Different Models without Diaphragms ....................... 57
iii
List of Figures
iv
Figure 6.3 Bridge Pouring Stages and Sequences ................................................................... 52
Figure 6.4 DESCUS Model ..................................................................................................... 53
Figure 6.5 CSiBridge Model for MD140 over MD27 Bridge Isometric View ....................... 54
Figure 6.6(a) Time Dependent Concrete Strength in CSiBridge Model (kip/in2) ................... 55
Figure 6.6(b) Time Dependent Concrete Stiffness in CSiBridge Model (kip/in2) .................. 55
Figure 6.6(c) Time Dependent Creep Coefficient in CSiBridge Model .................................. 56
Figure 6.7 Vertical Displacement Results from CSiBridge (with Diaphragm) (inch). ........... 58
Figure 6.8 Vertical Displacement Results from CSiBridge (without Diaphragm) (inch). ...... 58
Figure 6.9 The Creep Growth in Girder 11(Orange) and 12(Green) for with Diaphragms Model
(in). ........................................................................................................................................... 59
Figure 6.10 The Creep Growth in Girder 11(Orange) and 12(Green) without Diaphragms
Model (in). ............................................................................................................................... 60
Figure 6.11 The Creep Growth in Girder 11(Orange) and 12(Green) without Diaphragms
Model (in). ............................................................................................................................... 60
v
SERVICEABILITY-RELATED ISSUES FOR BRIDGE LIVE LOAD
DEFLECTION AND CONSTRUCTION CLOSURE POURS
EXECUTIVE SUMMARY
This study investigated the design criteria and practices in an effort to improve the quality of
bridge designs in the State of Maryland and beyond. This first criterion investigated was the
live load deflection for steel bridges. Since the live load deflection criterion is optional in the
AASHTO LRFD Bridge Design Specifications (2014), the Maryland State Highway
Administration (SHA) establishes no maximum limit on deflection and leaves the burden on
the designers to establish limits. This study developed a menu of criteria that designers can
choose from in their bridge designs.
The second design/construction criterion investigated was designing and detailing bridge deck
closure pours. A closure pour is a small area of concrete bridge deck that connects two portions
of a bridge deck placed in different stages of construction. For staged construction, the designer
should consider the deflections of the bridge on either side of the closure pour to ensure proper
transverse fitting.
In order to achieve these two objectives, the following tasks were completed:
1) Previous and current practices and future planning on the serviceability of bridges were
documented. This study looked at bridges within the short and median span range and
selected 30 samples from SHA’s inventory; all are steel girder bridges, where the highest
live load deflection occurs. Steel bridges designed with the live load deflection limit were
evaluated and summarized in this study.
2) The next step was to collect and study state-of-the-practice methods from federal and
other state agencies. All available current state-of-the-practice methods from the Federal
Highway Administration’s regulations, research and testing findings in the past and also
the practices from other states were located, collected and listed for study. Three bridges,
the I-270 over Middlebrook Road (bridge no. 1504200), Route 1 over Paint Branch
(bridge no. 1600400) and I-95 over Patuxent River (bridge no. 1619701) were chosen for
refined analyses to investigate the live load deflections. Field measurements for these
three bridges were collected from the research team to facilitate this study.
3) Several finite element models, with different software, were developed for the entire
bridge to compare the differences in deflection for the bridge model versus the simple
single girder analysis traditionally performed by SHA. The two-dimensional grid models
and three-dimensional finite element models can be used for the live load deflection
analysis as well as the staged construction analysis. In addition to the I-270 over
Middlebrook Road (bridge no. 1504200), Route 1 over Paint Branch (bridge no. 1600400)
vi
and I-95 over Patuxent River (bridge no. 1619701) bridges, the MD 140 bridge (bridge no.
6032) was modeled and studied to identify the impacts resulting from different
construction methods.
4) A summary of all the work listed above is included. Recommendations associated with
precast concrete beam or steel girder construction, complemented with current Maryland
practices on live load deflection limit and closure pours, are listed below:
This study investigated the design criteria and practices in an effort to improve the quality of
bridge designs in the State of Maryland and beyond. This first criterion investigated was the
live load deflection for steel bridges. Since the live load deflection criteria in the AASHTO
LRFD Bridge Design Specifications (2014) is optional, the Maryland State Highway
Administration (SHA) establishes no maximum limit on deflection and leaves the burden on
the designers to establish limits. This study developed a menu of criteria that designers could
choose from in their bridge designs. The second design/construction criterion investigated was
designing and detailing bridge deck closure pours. A closure pour is a small area of concrete
bridge deck that connects two portions of a bridge deck placed in different stages of
construction. For staged construction, the designer should consider the deflections of the
bridge on either side of the closure pour to ensure proper transverse fitting.
In summary, the objectives of this study were to assess and set the criteria on the “optional”
live load deflection evaluation specified in the AASHTO LRFD Bridge Design Specifications
(2014) and provide recommendations on bridge deck closure pours practices.
1
The focus of this study was about serviceability-related issues of median and short span steel
bridges, which mainly includes live load deflection. Chapter Two summarizes the literature
and the exploration of serviceability issues for steel bridges. The development of bridge live
load deflection limit or criteria in different states and countries were reviewed and discussed.
In Chapter Three, live load deflections of steel bridges were investigated. The methods of both
Allowable Stress Design (ASD) from the AASHTO Standard Specifications for Highway
Bridge (2002), and Load and Resistance Factor Design (LRFD) from the AASHTO LRFD
Bridge Design Specifications (2014), were compared. The comparison includes the types of
live load, which are mainly truck loads and lane loads; as well as some important factors such
as multiple presence factor and dynamic load allowance (i.e. impact factor).
In Chapter Four, three representative bridge models were analyzed. Each bridge was analyzed
with three different programs: MERLIN-DASH, DESCUS-I and CSiBridge. The
MERLIN-DASH analyzes the bridge by a line-girder model; DESCUS-I conducts the analysis
with two-dimensional grid method and the CSiBridge uses the three-dimensional (3D) finite
element method. After an overall comparison of these three representative bridges, about thirty
sample bridges from Maryland’s Bridge Inventory were collected, modeled, analyzed and
tabulated in Chapter Five. Several plots and charts are used to analyze the data and summarize
results of the sample bridges.
In Chapter Six, for bridge deck closure pours of one sample bridge in Maryland,
two-dimensional grid models created in DESCUS-I and three-dimensional finite element
models created in CSiBridge were studied beyond the MERLIN_DASH used in the design
stage. The creep effect of concrete was considered during the staging analysis. Finally, in
Chapter Seven conclusions of the serviceability-related issue are summarized to complete the
study.
2
2. Literature Review
General design principles are detailed in Chapter 2 of the AASHTO LRFD Bridge Design
Specifications (2014). Article 2.5.2.6.2 gives the criteria for deflection only as optional; it
advises that the maximum deformation of a bridge should not exceed 1/800 of the span length
for general vehicular bridges and 1/1000 of the span length for vehicular bridges with
pedestrian traffic. The reason for the smaller allowable deflection for the pedestrian bridges is
that pedestrians are more sensitive to bridge vibrations than vehicular drivers or passengers.
In order to better understand the rationale behind the current AASHTO LRFD deflection limits,
identifying how they were developed was of interest in this study The 1/800 span length (L)
limit was initially mentioned in the first American Highway Bridge Design Specification in
1953 and has been incorporated in every revision thereafter. The ASCE Committee on
Deflection Limitations of Bridges of the Structural Division (1958) reported on their
examination of the live load deflection limits and depth-to-span ratio, D/L, which was shown in
the 1953 American Association of State Highway Officials (AASHO) Standard Specifications
for Highway Bridges. The earliest deflection limits were adopted by the Phoenix Bridge
Company in 1871, which limited deflection to 1/1200 of the span length for a train moving at a
velocity of 30 miles per hour. The American Railway Engineering Association (AREA) took
depth-to-span ratios, which are an indirect method of limiting deflection in the early 1900s;
however, the limits were without any basis at that time. The concept of depth-to-span ratios for
highway bridges was originally raised in 1913 and adopted by AASHO in 1924.
Vibrations first became an important issue in the 1930s and the Bureau of Public Roads tried to
provide a correlation between the vibration problems of bridges and bridge structural
properties. They conducted a study that attempted to link the objectionable vibration felt on a
sample of bridges built in that era. This study concluded that structures having unacceptable
vibrations determined by subjective human response had deflections that exceeded L/800, and
this conclusion resulted in the L/800 deflection design limit. Given how old these studies are,
information regarding the specifics was not available. However, the bridges included in this
early study had wood plank decks, and the superstructure samples were pony trusses, simple
beams or pin-connected through-trusses. The Bureau did not incorporate composite girder
bridges, which are more popular today. The ASCE Committee in 1958 reviewed the history of
the bridge deflection criteria, completed a survey to obtain data on the behavior of bridges and
the opinions of bridge designing experts, reviewed field measurements of bridges subjected to
moving loads, and gathered information on human perception to vibration. The survey
concluded: (1) maximum oscillations occur with passage of medium weight vehicles not heavy
3
vehicles, (2) reports of objectionable vibrations came from continuous span bridges more often
than simple span bridges, and (3) there was no defined level of vibration which constituted as
being undesirable. Many factors would affect the vibration of the bridge. Some of them are
listed here:
Flexibility and natural frequency of bridge
Flexibility and natural frequency of vehicle
Relative weight of vehicles and bridge
Vehicle speed
Frequency of load application
Motion caused by loads in adjacent spans of continuous span structures
Damping characteristics of bridge and vehicle
The use of depth-to-span ratio, D/L, began in the early 1900s with the AREA stating that pony
trusses and plate girders should have a depth no less than 1/10 of the span length. There has
been little change with these ratios over the years. The current depth-to-span limits are 1/10 for
simple span trusses and 1/30 for simple span rolled shapes and plate girders.
The early specifications for highway bridges used with some changes the depth-to-span ratios
from the American Railway Engineering and Maintenance-of-Way Association (AREMA),
which was abbreviated to AREA at that time. Table 2.1 shows the limiting D/L ratios that have
been incorporated in previous AREA and AASHTO specifications (Taly 1998).
Both AREMA and AASHTO Specifications included statements that required flanges to be
strengthened if section depths smaller than those required by the limiting depth-to-span ratio
are used.
The use of depth-to-span ratios was primarily to limit deflections but was also driven by
economics. The limiting values of depth-to-span ratios have decreased with time while
allowable stresses have increased. This would result in shallower sections being used, which
would result in larger deflections. This result confused the ASCE Committee on Deflection
4
Limitations of Bridges of the Structural Division, which was tasked with investigating the
origins of the deflection and depth-to-span limits. Furthermore, the committee quoted the 1905
AREA Committee’s explanation of their depth-to-span ratios:
“We established the rule because we could not agree on any. Some of us in
designing a girder that is very shallow in proportion to its length decrease the
unit stress or increase section according to some rule which we guess at. We
put it there so that a man would have a warrant for using whatever he pleased.’’
A conclusion was reached in the 1905 report that the two criteria, deflection limit and
depth-to-span ratio, are of different origin. The deflection limit is to limit undesired vibration
while the depth-to-span ratio is a result of economics. Also, the committee could not provide
any recommendations or methods for the best way to limit deflections or vibrations.
In the American practice, the deflection of bridges supporting vehicular traffic is generally
limited to the span length divided by 800 (L/800) for simple and continuous spans, and L/300
for cantilever arms. For bridges intended to also carry pedestrian and bicycle traffic the
AASHTO specifications have placed further limits. These deflection provisions are very
simple to use, but not directly related to the real issue of concern about the vibration response
under live load. The deflection and dynamic response both involve the stiffness of the bridge as
well as some other parameters, such as the mass, damping and so on.
In the 1970s, Wright and Walker performed a study reviewing the rationality of the deflection
limitation provisions and Roeder, et al. revisited the subject decades later in 2002 suggesting
that the current AASHTO live load deflection limits L/800 for vehicular traffic bridges and
L/1000 for pedestrian are not always sufficient in controlling excessive bridge vibration and
should ultimately be removed. Fountain and Thunman conducted a study, which examined live
load deflection criteria for steel bridges with concrete decks in 1987. They concluded that
AASHTO’s live load deflection criteria did not achieve the purported goal for strength,
durability, safety, or maintenance of steel bridges. They questioned the AASHTO deflection
criteria because the influencing 1930 Bureau of Public Roads study did not incorporate
composite girder bridges. In 2007, Barker and Barth compared the procedure in AASHTO
LRFD, which should have provided some uniformity in application, to the specific procedures
used in several states. They found that there is wide variation in the deflection limit employed
by the various states. Of the 47 states reporting deflection limits for bridges without pedestrian
access:
1 state employs a L/1600 limit,
1 state uses a L/1100 limit,
5 states employ a L/1000 limit,
1 state expresses a preference for L/1000 but requires L/800 limit, and
39 states employ a L/800 limit.
5
Of the states reporting deflection limits for bridges with pedestrian access:
1 state employs a L/1600 limit,
2 state use a L/1200 limit,
1 state employs a L/1100 limit,
39 states use a L/1000 limit,
3 states employ a L/800 limit.
There is very wide variation in these deflection limits, since the largest deflection limit is twice
as large as the smallest deflection limit. Two of the 47 states use the deflection limit as a
recommendation rather than a design requirement.
Another problem is the live load that bridge designers use in order to obtain the live load
deflections. The AASHTO Specification indicates that deflections due to live load plus impact
are to be limited by the deflection limit. Within this context, there is ambiguity in the loads and
load combinations that should be used for the deflection calculations, because design live loads
are expressed as both individual truck loads and uniform lane loads. The survey showed that
the loads used to compute these deflections have even greater variability than observed in the
deflection limits.
1 state employs the HS-20 truck load only,
16 states use the HS-20 truck load plus impact,
1 state uses the HS-20 lane load plus impact,
1 state uses the HS-20 truck load plus lane load without impact,
7 states use the larger deflection caused by either the HS-20 truck load plus
impact or the HS-20 lane load plus impact,
17 states use the HS-20 truck load plus lane load plus impact,
4 states consider deflections due to some form of military or special permit
vehicle,
8 states use the HS-25 truck load.
It is not easy to compare the variability of the load and the variability of the deflection limit in
different states because these are not mutually exclusive. Wisconsin DOT, for instance, uses
the smallest deflection limit, but it also employs smaller loads than most other states. However,
the relative importance of the lane load and design truck load are possible to be different for
long and short span bridges, and so the L/1600 limit used in Wisconsin may be more restrictive
for short span bridges. On the contrary, the Wisconsin limit may be a generous deflection limit
for very long span bridges, because the truck load becomes relatively smaller with longer
bridge spans despite the small deflection limit.
The actual methods used to calculate deflections are not defined in the AASHTO
Specifications. Deflection limits are based upon deflections caused by service loads under
actual service conditions in typical engineering practice. Load factors or other factors used to
modify design loads are not normally used in these deflection calculations, and the actual
6
expected stiffness of the whole structure is also needed for calculation. This is another reason
that explains the variability in the application of the deflection limits. Load factors and lane
load distribution factors are employed in some states while they are not adopted in others. Lane
load distribution factors can significantly affect the magnitude of the loads used to calculate the
deflections. The survey shows that 26 states use lane load distribution factors from the
AASHTO Standard Specifications in calculating these deflections. Three states report that they
use the LRFD lane load distribution factors. Thirteen states indicate that they effectively apply
the loads uniformly to the traffic lanes by the AASHTO multiple presence lane load rules. They
then compute the deflections of the bridge as a system without any modification for load
factors, girder spacing or lane load distribution. These states effectively use an equal
distribution of deflection principle. One state uses its own lane load distribution factor that is
comparable to system deflection calculations. Several states indicate some flexibility in the
calculation method, and a few states indicate a reluctance to permit the bridge deflection limit
to control the design. The effect of the lane load distribution factor can be very significant.
Depending on the spacing of bridge girders, the load used for bridge deflection calculations can
be getting, at most, 100% larger than the load used for states where deflections are computed
for the bridge as a system or where the loads are uniformly distributed to girders. Load factors
may also be an issue of concern. Five states report that they apply load factors to the load used
for the deflection calculation. These load factors also increase the loads used to compute bridge
deflections, and they increase the variability in the application of the deflection limit between
different states.
Since the 1930s, vibration becoming an important issue for bridge structure, the natural
frequency of the bridge was attracting people’s attention and becoming more and more of a
concern. Some researchers state that bridge vibration is better controlled by a limit based on a
dynamic property of the bridge, such as natural frequency rather than deflection limit criteria.
Other studies show that the presence of excess vibrations is caused more by the natural
frequency of the bridge, vehicle speed, and surface roughness than it is correlated to the
deflection. Deflection limits not considering these factors are insufficient in preventing excess
vibrations. Different methods and solutions were devised. Even fatigue was treated as a key
factor that would generate deflection of bridges, thus the topic of fatigue load has been
researched for many years. However, no uniform criterion has been established.
7
where fobs and fcal are the observed frequency and calculated frequency, respectively. Eb is the
modulus of elasticity of steel, Ib is the moment of inertial of the beam of cross-section, g is the
acceleration due to gravity, and w is the weight per unit length of the beam. Consistent units
must be employed for all variables. This equation was validated for structures with 2 Hz < fcal <
7 Hz.
More recent studies by Billings conducted over a wide range of bridge types and vehicle loads,
loads ranging from 22.5 kip to 135 kips (100 KN to 600 KN), confirm the results of the initial
study (Ontario Ministry of Transportation, 1991).
For both the Canadian Standards and the Ontario Code, only one truck is placed at the center of
a single traveled lane and the lane load is not considered. The maximum deflection is computed
due to factored highway live-load including the dynamic load allowance, and the gross
moment of inertia of the cross-sectional area is used (i.e. for composite members, use the actual
slab width). For slab-and-girder construction, deflection due to flexure is computed at the
closest girder to the specified location if the girder is within 1.5m of that location.
8
other limits specified in the codes, the vibration behavior of the bridge does not need to be
specifically investigated. Where these deflection limits are exceeded, the vibration behavior of
the bridge shall be assessed by a rational method, using acceptance criteria appropriate to the
structure and its intended use.
Meanwhile, the deflection of highway bridge girders under live load plus dynamic load
allowance shall not exceed 1/800 of the span length (AUSTRALIAN, 1996). However, the
work (Sergeev and Pressley, 1999) showed that the origin of this live load deflection limit is
uncertain. It was originally adopted in earlier versions of the Code, apparently taken from
contemporary AASHTO Specifications. In this study, the live load deflection limits for three
exiting bridges were investigated and alternative serviceability criteria were proposed. As a
result of the combination of both the proposed design live loadings (A160, S1600 and M1600),
which are heavier than the original design T44 truck, and the utilization of higher strength
steels 50 ksi (350 Mpa and higher), composite bridges were found to be particularly vulnerable
to the deflection limits. So, the validity of a live load deflection control criterion was
questioned. The Lotus Street Duplication Bridge is a slab on steel I-girder bridge with spans of
153 + 117 feet and the actual L/D equals 23.7, less than the recommended value of 25 for
composite girders. However, the live-load deflection limit was exceeded by 12% for 101t
Double Bottom Road Train (DBRT) loading and by 45% for M1600 loading. The Mortlock
River Bridge is a 6-span continuous composite steel bridge. The deflection under 101T DBRT
loading controlled the design and resulted in low L/D ratio of 13.7. The deflection limit L/800
is exceeded under M1600 by 7%. Bridge 1470 is a simply supported composite steel bridge
with a span length of 87.4 feet, the deflection under 101T DBRT loading also controlled this
design and the M1600 live-load deflection is 44% greater than the L/800 limit. Thus, it is
recommended that the Serviceability Limit State Criteria in the Australian Design Code should
be optimized by eliminating the artificial live load and placing more emphasis on the elastic
response of structures to serviceability loads, namely preventing rapid structure deterioration
by controlling crack widths under short term loads and controlling vibration as appropriate to
the situation.
10
2.2 Construction Closure Pours
2.2.1 Federal Highway Administration’s (FHWA) Regulation
Staged construction is defined as building a parallel portion of a bridge at a different time in
Federal Publication No. FHWA-NHI-08-048 (2007). Closure pours usually are employed
when connecting two parts of a bridge in a closure bay. Some owners are unwilling to use
closure pours due to unfortunate experiences with them. It is suggested to consider pouring
sequence, shrinkage, and proper computation of cambers in order to get very compatible
deflections. Another issue that should need attention is a case where some girders are
composite when adjacent girders are being decked. When the wet concrete deck is placed on
the non-composite steel, a disproportionate portion of this new load is drawn to the stiffer
composite girders. To avoid this effect, the cross-frames/diaphragms connected to the two
stages might be disconnected, and then a closure pour is used.
Similarly, the Nevada Department of Transportation NDOT Structures Manual (2008) also
emphasizes the two purposes of a closure: it defers final connection of the stages until after the
deflection from deck slab weight has occurred, and it provides the width needed to make a
smooth transition between differences in final grades that result from construction tolerances.
A minimum closure width of three (3) feet is recommended. Greater closure widths may be
required when larger relative dead-load deflections are anticipated. The required width can be
estimated by considering the closure pour to be a fixed-fixed beam and by limiting the stresses
in the concrete to the cracking stress. The manual also includes the following specifications:
Stay-in-place forms shall not be used under the closure pour.
11
Diaphragms/cross frames in the staging bay of structural steel girders shall not be
rigidly connected until after the adjacent stages of the deck have been poured.
Construct concrete diaphragms in the staging bay of prestressed concrete girders after
adjacent portions of the bridge are complete. The diaphragms may be poured as part
of the closure.
Reinforcing steel between different stages shall not be tied or coupled until after the
adjacent stages of the deck have been poured.
Support the finishing machine on an overhang jack that is connected to the girder
loaded by the deck pour. Do not place the finishing machine on a previously poured
deck. The bridge designer must indicate in the contract documents that this method of
constructing the closure pour is not allowed.
In 2010, Michael Sprinkel, Chris Blevins, Richard E. Weyers, and their group conducted a
study that examined the failure and repair of a deck closure pour on Interstate 81in Virginia.
They reported that several reinforced concrete decks on I-81 were replaced using the 3-ft wide
center closure pour with epoxy coated reinforcement extending from each of the decks in 1992.
However, three bridges were observed to be in a near failure or failure state after 17 years. The
reason for the failure is mainly due to reinforcement corrosion in the vicinity of the leaking
construction joints and transverse cracks. Therefore, this study recommended that expansive
deck concrete should be used for closure pours to minimize or prevent the opening of closure
pour construction joints and the formation of cracks due to shrinkage of the concrete. Also,
placing closure pour construction joints over beams is suggested so that the closure pour is
supported by the adjacent beams rather than the transverse reinforcement.
The findings of NCHRP Synthesis 345 (2005) are based on survey questionnaires and
interviews sent to steel bridge fabricators, steel bridge erectors and contractors from U. S.
states and Canadian provinces. This study reported that problems develop in staged
construction as the result of the difference in elevation between the deflected position of
members after pouring and the non-deflected position of the members before pouring.
12
Furthermore, deck alignment, cross-frame connection and girders between two stages all
require special considerations. For a successful stage construction, bridge owners developed
several different strategies. The first one is using a closure or construction pour between stages.
In Ohio, bridge owners use an 800 mm construction closure pour between stages to control
delta deflection between stages. In Montana, they use a 600 mm closure pour to deal with the
deflection between stages. In Tennessee, they require a closure pour for staged construction.
Differential displacements across a staged construction field-cast connection can have a
detrimental impact on the performance of the connection. Ben Graybeal (2012) investigated
the bond of #4 (#13M) reinforcing bars and found that differential displacements equal to or
greater than 0.05 inches (1.27 mm) caused a reduction in the bond strength. Not surprisingly,
differential displacements seemed to ream a hole in the embedment material around the
reinforcing bar, thus reducing the bond capacity.
Bridge Superstructure Design-MM No. 10 (2001) from the Iowa Department of Transportation
placed guidelines when considering closure pours for bridge decks with longitudinal
construction joints:
a. If there is more than 2 inches (50 mm) of dead load deflection in the bridge deck,
then closure pours should be used.
b. If the staged construction is on a highway system with a high volume of truck
traffic (approximately 500 or more trucks per day), then a closure pour should be
considered. This will be addressed on a case by case basis.
The closure pour should be wide enough to allow for splicing of the transverse reinforcing steel
along with 2 inches (50 mm) of clearance for the end of the bars from the construction joint.
The minimum closure pour width should be three feet (900 mm).
Closure pours should be placed in areas with constant cross-slope in the bridge deck. In
addition, closure pours over beams should be avoided.
When closure pours are used, follow these guidelines for the different types of bridges:
14
3. Live Load Deflection Validation of Steel Bridges
In the first part of this chapter, live load deflection of steel bridges is studied. Live loads are
assumed to consist of gravity loads (vehicular live loads, rail transit loads and pedestrian loads),
the dynamic load allowance, centrifugal forces, braking forces and vehicular collision forces;
they are considered to be transient loads that are assumed to be applied to the short-term
composite section. However, the live loads of interest in this study are only the design truck
and lane loads since the primary live loads on bridge spans are due to traffic. The methods of
both Allowable Stress Design (ASD) from the AASHTO Standard Specifications for Highway
Bridge (2002) and Load and Resistance Factor Design (LRFD) from the AASTHO LRFD
Bridge Design Specification (2014) are compared. The comparison includes the types of live
load as well as important factors such as multiple presence factor and dynamic load allowance
(i.e. impact factor). In the second part, three different programs, MERLIN-DASH, DESCUS-I
and CSiBridge, are briefly introduced including analysis principle and procedure. The
MERLIN-DASH program analyzes the bridge by a line-girder bridge model and the
DESCUS-I program conducts the analysis with two-dimensional grid method. Both are used as
design tools by SHA. The CSiBridge uses the three-dimensional finite element method. Three
bridges from SHA’s bridge inventory were selected to be representative bridges. The full
investigation and analyses including the field measured data as well as the model analysis of
these three bridges are shown in the next chapter.
3.1 Loading considered in the Load and Resistance Factor Design (LRFD)
and the Allowable Stress Design (ASD)
3.1.1 Loading considered in the Allowable Stress Design
In the AASHTO Standard Specifications for Highway Bridge (2002), the Allowable Stress
Design (ASD) method has been adopted. The live loads in highways are defined as standard
truck and lane loads in ASD.
For the standard truck load, there are four standard classes of highway loading, which are H-20,
H-15, HS-20 and HS-15. Loading H-15 of two axles is 75 percent of loading H-20 as well as
HS-15 of three axles is 75 percent of loading HS-20. The HS-20 design truck, which is
currently used in most states for live load analysis and design, indicates a vehicle with a front
tractor axle weighing 8 kips (two sets of wheels, 4 kips on the left and 4 kips on the right), a
rear tractor axle weighing 32 kips, and a semitrailer axle weighing 32 kips. The distance
between the front wheels and the rear tractor wheels is 14 feet and the distance from this axle to
the semitrailer axle is varying from 14 to 30 feet as shown in Figure 3.1.
15
Figure 3.1 HS-20 Design Truck
In addition to the standard truck loadings, the AASHTO specifications also allow the
representation of the truck as a single concentrated load and a uniform load. For the HS-20
truck loading, the concentrated load is defined as 18 kips for the moment, and 26 kips for the
shear; the uniform load is 0.64 kips per linear foot as shown in Figure 3.2. In this chapter, 18
kips is used because the deflection is related to the moment. Both the concentrated load and
uniform load should be considered as uniformly distributed over a 10-foot width on a line
normal to the centerline of the lane from the AASHTO specifications.
In Maryland, the new design truck loading called ‘HS-25’ was adopted due to some concern
that the HS-20 truck load did not adequately reflect actual conditions. The HS-25 truck load is
25 percent higher than the HS-20 truck load resulting in 10 kips in the front and 40 kips in the
back instead of 8 and 32 kips. Also, the concentrated load for moment and shear and the
uniform lane load are higher by 25 percent making them 22.5 kips, 32.5 kips and 0.8 kips per
linear foot, respectively. The live load deflection result from the worst case scenario of the both
truck loading and lane load plus concentrated load would be used during the deflection
analysis.
For the live load deflection analysis with ASD method, several factors are used in the
calculation of live load deflection. The live loads should be increased for steel bridges to allow
16
for dynamic, vibratory and impact effects. The impact allowance should be determined by the
following formula:
in which,
I = impact fraction (maximum 30 percent);
L = length in feet of the portion of the span that is loaded to produce the
maximum stress in the member.
Another important factor, the multiple-presence-factor, should be used for reduction in load
intensity. Based on the AASHTO Standard Specifications (2002), the multiple presence factors
are shown in Table 3.1.
The dynamic allowance would be different in this design method from the one in ASD method.
The dynamic load allowance in LRFD is a constant of 33 percent only applied to the truck
loading. The multiple presence factors are also slightly different from ASD, which are shown
in Table 3.2.
17
Table 3.2 Multiple Presence Factors in the LRFD Method
Number of loaded lanes Multiple presence factors
1 1.20
2 1.00
3 0.85
>3 0.65
Table 3.3 Loads and Factors considered in ASD and LRFD Methods
ASD LRFD
Loading
Dynamic
Allowance
(Truck only) (truck only)
Multiple
Presence
Factors
MERLIN-DASH
Design, Analysis and Rating of StraigHt Girder Bridge Systems (MERLIN-DASH), was
developed for bridge design engineers who function in a software production environment. In
order to provide a program which would be applicable nationally, MERLIN-DASH was
developed to offer the wide range of features and options necessary to meet the demands of
universal usage in the analysis, design, and rating of steel and reinforced concrete bridges.
In this portion of the study, the rating functions of AASHTO WSD (or called ASD) and LRFD
methods have been used for the three representative bridges and the rest of the sample bridges
as well. In MERLIN-DASH, the girders or beams of one bridge model are all assigned as the
same girder or beam as the interior one in the real bridge. The shoulders of the bridge models
are treated as curb to which no design load is assigned. Bracing between girders or beams is not
considered in the analysis of this software. The analysis was based on the influence line since
this is only in two dimensions. The HS-25 design truck and HL-93 design truck are used as
design load for ASD and LRFD methods, respectively.
DESCUS-I
The computer program Design and Analysis of Curved I-Girder Bridge Systems (DESCUS-I),
performs the complete analysis and partial design of a straight or horizontally curved bridge
composed of flanged steel sections which act either compositely or non-compositely with a
concrete deck. The program can be run using either WSD (or called ASD), LFD, or LRFD
method. The bridge may be of arbitrary plan configuration and can be continuous and skewed
over supports. The girders may have high degree of curvature, may be nonconcentric,
bifurcated, and may contain hinges.
The program models the bridge structure as a two-dimensional grid in a stiffness format with
three degrees-of-freedom at each nodal point (corresponding torsion, shear, and bending
moment). All nodal locations, member connectivity, and section properties are generated
internally from basic input. All dead load (DL) computations are performed automatically
within the program to satisfy the construction conditions specified by AASHTO. Additional
constant dead load is allowed per girder as a special program input option. All live load (LL)
computations also are performed automatically, where the AASHTO truck and lane loadings
are applied to an influence surface previously generated for the entire bridge. Impact (I)
effects also are included per AASHTO recommendations. Up to nine arbitrary trucks can also
be specified and analyzed concurrently.
Output contains the positive and negative maximum moments, shears, and torsion along with
the corresponding primary and warping stresses for each girder and beam or truss diaphragm
element. These maxima are given along with all AASHTO group combinations for DL + LL +
I. Also, outputs are deformations along each girder for DL and maximum LL + I, along with
the allowable deflections recommended by AASHTO. Finally, various tables are output which
19
yield information on the design of the sections including maximum stresses, allowable stresses,
shear ranges, shear connector, and stiffener spacing.
In DESCUS-I the analysis is based on the influence surface instead of the influence line in
MERLIN-DASH. Therefore the results from the model built in this program are expected to be
more accurate than those in MERLIN-DASH, since the transverse action would affect the
whole bridge behavior.
CSiBridge
CSiBridge is a new software for bridge analyses and it provides a way to model the whole
bridge in three-dimensions instead of 2-dimensions as in MERLIN-DASH and DESCUS-I.
CSiBridge is a module separated from the previous version of SAP2000. Modeling, analysis
and design of bridge structures have been integrated into CSiBridge to create the ultimate in
computerized engineering tools. Using CSiBridge, engineers can easily define complex bridge
geometries, boundary conditions and load cases. The bridge models are defined parametrically,
using terms that are familiar to bridge engineers such as layout lines, spans, bearings,
abutments, bents, hinges, and post-tensioning. The software creates spine, shell or solid object
models that update automatically as the bridge definition parameters are changed. CSiBridge
design allows for quick and easy design and retrofitting of steel and concrete bridges. The
parametric modeler allows the user to build simple or complex bridge models and to make
changes efficiently while maintaining total control over the design process. Lanes and vehicles
can be defined quickly and include width effects. Simple and practical charts are available to
simulate modeling of construction sequences and scheduling. In addition, AASHTO LRFD
design is included so that the user could easily obtain the results that are needed. In this study,
four load cases, which are the HS-25 truck and lane load plus concentrated load for the ASD
method, as well as the HS-20 truck along and 25 percent of the HS-20 truck load plus lane load
for the LRFD method, were applied for a certain bridge. The worst situation in each method
was found and analyzed. The analysis type used in this study is ‘moving load’ in CSiBridge
since it should be consistent with MERLIN-DASH and DESCUS-I.
Among these three programs, CSiBridge should be able to obtain the most precise results while
the MERLIN-DASH is the most convenient program for users to use to model the bridge and
do analysis. It is impossible to model all the sample bridges with the accurate but complicated
finite element method. Therefore, a few representative bridges need to be selected and
analyzed with different programs to examine whether the results from MERLIN-DASH are
acceptable.
21
4. Refined Analysis of Three Representative Maryland Bridges
In this chapter, all the three representative bridges were analyzed using MERLIN-DASH,
DESCUS-I and CSI Bridge computer programs. In the Allowable Stress Design method, the
HS-25 design truck was adopted for the live load deflection although the bridges are designed
with the HS-20 loading, because Maryland used HS-25 loading instead of HS-20 loading from
the 1990s until 2007 when LRFD was adopted. The HL-93 design truck was used for deflection
analysis in the Load and Resistance Factor Design method.
22
Figure 4.1 Typical Cross Section of the I-270 over Middlebrook Road Bridge S.B.R
For the I-270 over Middlebrook Road (bridge no. 1504200) bridge analysis in
MERLIN-DASH, each beam is assigned as the plate girder sections detailed in Table 4.1 below.
Section 1 is assigned from 0 to 35 feet and 105 to 140 feet, respectively, and section 2 is from
35 to 105 feet, which is in the middle 70 feet of the bridge. The bridge roadway is adjusted to
36 feet by removing the shoulders, and the edge of slab to curb is input as 12.5 feet by
averaging the remaining cross section of the bridge once the 36 feet of driving roadway width
is removed.
The arbitrary uniform load per beam in this program consists of dead load including ‘haunch’
and ‘stay-in-place’, and superimposed dead load that usually are ‘future wearing surface’ and
‘barrier’. In the I-270 over Middlebrook Road bridge (bridge no. 1504200), they were
determined to be 0.048 kips/ft., 0.022 kips/ft., 0.216 kips/ft. and 0.152 kips/ft., respectively.
Diaphragms and stiffeners are neglected in this study since only the moment strength was taken
into consideration. Shear is not a concern here. The model built in MERLIN-DASH is shown
in Figure 4.3 below.
The rating function of MERLIN-DASH has been used for both the ASD and LRFD methods.
In ASD, the design truck is selected as HS-25 and in LRFD, it is HL-93. The results obtained
for live load deflections for this bridge from DASH are 0.821 inches for ASD HS-25 loading
with lane load governing and 0.651 inches for LRFD HL-93 loading with lane plus 25% truck
governing.
The model of the I-270 over Middlebrook Road bridge (bridge no. 1504200) in DESCUS-I
could be more accurate than the model in MERLIN-DASH because the real bridge has a
24
14-degree skew angle that can only be treated as a straight girder bridge in MERLIN-DASH
due to the limitation of the line-girder program. In DESCUS-I, this skew issue can be solved. A
V-bracing diaphragm is applied for the bridge and the graphic generated from DESCUS-I is
shown in Figure 4.4. Also, the distribution factors can be automatically calculated based on the
cross section information pre-input into the program. In the ASD method the design load is still
the HS-25 truck and in LRFD it remains the HL-93 truck.
Figure 4.4 DESCUS-I Graphic of the I-270 over Middlebrook Road Bridge
The live load deflections for both design methods were calculated as 1.151 inches with lane
load governing in ASD and 1.017 inches with HL-93 governing in LRFD.
In CSiBridge, a three-dimensional finite element model (Figure 4.5) was built and moving load
analysis was used for both the ASD and LRFD methods.
The load factors mentioned previously have been manually applied in for both design methods.
For the ASD method, the lane load governs and the live load deflection under the HS-25 design
truck load is 1.107 inches; for the LRFD method, the truck load alone governs and the live load
deflection is 0.984 inches.
25
Figure 4.5 I-270 over Middlebrook Road Bridge Model in CSiBridge
26
Figure 4.6 Typical Cross Section of the Route 1 over Paint Branch Bridge
The Route 1 over Paint Branch bridge (bridge no. 1600400) is a skewed bridge. In
MERLIN-DASH, however, this bridge is built to be straight due to line girder limitation.
Figure 4.6 only shows half of the bridge roadway so that it is assumed that there is an extra
length on the right beyond the overhang of 3.79 feet, same as the left overhang, and 4.15 feet as
the sidewalk in average. The dead load information is shown in Table 4.2. The model view of
MERLIN-DASH is shown in Figure 4.8.
27
Figure 4.8 Route 1 over Paint Branch Bridge Model in MERLIN-DASH
The live load deflections for this bridge are calculated as 0.85 inches for both spans within the
ASD method and 0.687 inches for both spans within the LRFD method. The lane load governs
in ASD and HL-93 truck load governs in LRFD.
In DESCUS-I, the 30-degree skew angle is considered. The lateral bracing between two
adjacent girders used a steel channel. Analyses were performed for both the ASD and LRFD
design options and the live load deflection results were 1.079 inches for the ASD method with
the HS-25 truck load governing and 0.918 inches for the LRFD method with HL-93 truck load
governing. The bridge framing plan from DESCUS-I is shown in Figure 4.9.
Figure 4.9 DESCUS-I Graphic of the Route 1 over Paint Branch Bridge
In CSiBridge, this bridge is modeled with shell elements as shown in Figure 4.10. For the ASD
method, the HS-25 truck load governs and the live load deflections are 1.016 and 0.973 inches
28
for the two respective spans; for the LRFD method, the truck load alone governs and the live
load deflections are 0.821 and 0.786 inches for two respective spans.
The I-95 over Patuxent River bridge (bridge no. 1619701) has two vehicle travel directions
N.B.R. and S.B.R., whose cross sections are the same but in the opposite directions. A typical
cross-section at the mid-span of N.B.R. shown in Figure 4.11 was taken for analysis. The
four-lane roadway width is 48 feet and the edge of slab to curb is 11.5 feet including a 10 feet
wide shoulder. There are eight girders under 8.5 inches of deep concrete slab in this bridge; the
two exterior girders are slightly different from the six interior girders. The girders are at the
same spacing of 9 feet 2 inches and the overhang is 3 feet 5 inches. Figure 4.12 only shows half
of the girder elevation; the other half is symmetric with this half. In the area of the pier, the web
depth of the girder varies in a parabolic shape; therefore, more work needs to be done during
modeling of the bridge in each program analysis. The I-95 over Patuxent River bridge was built
in the 1970s; hence, the structural steel has the yield stress of 36 ksi.
29
Figure 4.11 Typical Cross Section at Mid-span of the I-95 over Patuxent River Bridge
In MERLIN-DASH, the interior girder sections were used for all eight of the girders and in the
area of the parabolic shape defined as concave down. The dead load information is shown in
detail in Table 4.3 below. The spacing of the lateral bracing is 23.57 feet for the end spans and
22.5 feet for the middle span. The model built in MERLIN-DASH is shown in Figure 4.13.
Table 4.3 Dead Load Information of the I-95 over Patuxent River Bridge
30
Figure 4.13: The I-95 over Patuxent River Bridge Model in MERLIN-DASH
The live load deflections of the I-95 over Patuxent River bridge (bridge no. 1619701) obtained
from MERLIN-DASH rating function are 0.750 inches for end spans and 0.808 inches for the
mid span within the ASD method. In the LRFD method, the end span deflection is 0.470 inches
and the mid span deflection is 0.527 inches. The lane load governs in ASD and HL-93 design
truck load governs in LRFD.
The I-95 over Patuxent River bridge (bridge no. 1619701) is a straight steel girder bridge so
skew needs to be considered in the DESCUS-I model. An assumption for parabolic shape in
girder elevation was made so that the average depth of the web in the changing area was
selected to be the constant; therefore, seven different plate girder sections, according to Figure
4.12, were used for the bridge model in DESCUS-I. As a result, the girder elevation could be
modeled, as shown Figure 4.14, and detailed, as listed in Table 4.4. Figure 4.15, is half of the
framing plan for the I-95 over Patuxent River bridge from DESCUS-I model graphics.
Figure 4.14 Girder Elevation of Half of the I-95 over Patuxent River Bridge in
DESCUS-I
31
Table 4.4 Beam Sections
P.G. Web P.G. Flange
Top
Unit: inch Depth Thick Top Thick Bot Width Bot Thick
Width
Section1 72 0.4375 18 1 18 1.5
Section2 72 0.4375 18 1.5 18 2.25
Section3 72 0.4375 18 1.25 18 1.25
Section4 120 0.4375 18 1.75 18 1.75
Section5 72 0.4375 18 0.875 18 1.25
Section6 108 0.4375 18 1.75 18 1.75
Section7 84 0.4375 18 1.25 18 1.25
Figure 4.15 Half Framing Plan of the I-95 over Patuxent River Bridge in DESCUS-I
The live load deflections of the I-95 over Patuxent River bridge (bridge no. 1619701) obtained
from DESCUS for the ASD Design Method are 1.012 inches for all the spans. In the LRFD
method, the end spans’ deflection is 0.684 inches and the mid span deflection is 0.732 inches.
The lane load governs in ASD and HL-93 design truck load governs in LRFD.
In CSiBridge, this bridge is modeled as the actual one (Figure 4.16). For the ASD method, the
lane load governs and the live load deflections are 0.670 inches for end spans and 0.720 inches
for the mid span; for the LRFD method, 25 percent truck load plus lane load governs and the
live load deflections are 0.930 and 0.933 inches for two end spans and mid span, respectively.
A field bridge test was conducted for the I-95 over Patuxent River bridge (bridge no. 1619701)
and a few sensors were used to acquire the live load deflection data. A long-term measurement
provided the maximum deflection on the location of 48 feet and 10 inches from the south
bridge abutment to be 5.96 mm (0.23 in.) and 4.98 mm (0.20 in.) for the third girder and the
second girder, respectively. From the software results, it was found that the measurement data
are slightly smaller. This may be due to the fact that the actual vehicles’ effect on the bridge did
not reach the same high level as the fully loaded assumption used in the program analysis.
32
Figure 4.16 The I-95 over Patuxent River Bridge Model in CSiBridge
33
Table 4.5 Live Load Deflection of Three Representative Bridges
LRFD ASD
I-270 over Multiple lane factor MF=0.85 MF=0.9
Middlebrook Road Impact factor IF=33% IF=50/(L+125)=18.9%
1 span CSI CSI
DASH DESCUS-I DASH DESCUS-I
L=140 ft. Bridge Bridge
3 lanes -0.651 -0.984 -1.017 -0.821 -1.107 -1.151
MF=0.75
MF=0.65
I-95 over Patuxent IF=50/(L+125)=17.2% (End span)
IF=33%
River IF=50/(L+125)=16.4% (Mid span)
3 spans CSI CSI
DASH DESCUS-I DASH DESCUS-I
L=165-180-165 ft. Bridge Bridge
4 lanes -0.470(end) -0.670 -0.684 -0.763 -0.93 -1.012
-0.527(mid) -0.720 -0.732 -0.808 -0.933 -1.012
MF=0.9
MF=0.85
Route 1 over Paint IF=50/(L+125)=24.4% (span 1)
IF=33%
Branch IF=50/(L+125)=24.4% (span 2)
2 spans CSI CSI
DASH DESCUS-I DASH DESCUS-I
L=80-80 ft. Bridge Bridge
3 lanes -0.687(span1) -0.821 -0.918 -0.825(span1) -0.973 -1.079
-0.687(span2) -0.821 -0.918 -0.825(span2) -0.973 -1.079
According to the AASHTO Standard Specifications for Highway Bridges (2002), Article
10.6.4
”When spans have cross-bracing or diaphragms sufficient in depth or strength to ensure
lateral distribution of loads, the deflection may be computed for the standard H or HS
loading (M or MS) considering all beams or stringers as acting together and having equal
deflection.”
“When investigating the maximum absolute deflection for straight girder systems, all
design lanes should be loaded, and all supporting components should be assumed to
deflect equally.”
It could be found that the distribution of live load is used as an average value (number of
lanes/number of girders) since all the girders are treated as equal. This may cause a discrepancy
that the live load deflection is relatively small using line-girder program, which can be
explained in Figure 4.17.
34
Figure 4.17 Distribution of Live Load along the Roadway
In DASH, the live load distribution factors are using this average value and multiple-presence
factor together. Taking I-270 over Middlebrook Road bridge (bridge no. 1504200) as an
example, there are three lanes and eight girders so that the distribution factor for ASD is 0.338
and for LRFD is 0.319. This could cause the live load deflection to be very small and then
affect the results of analysis. In order to avoid this situation, an adjustment was developed in
DASH. The live load distribution factor is defined as the number of the traffic lanes divided by
the effective number of girders, which means the girders directly support the lanes but not the
curb. For instance, for the I-270 over Middlebrook Road bridge (bridge no. 1504200), the
distribution factor for both ASD and LRFD is changed to 0.5. With this idea in mind, the live
load deflection for the three representative bridges from DASH becomes higher, which is
closer to the actual case and more conservative. With this adjustment made to all of the test
bridges, the live load deflections are provided in Table 4.6.
36
5. Results of 30 Sample Bridges Using the Line-girder Method
In this chapter, 30 sample bridges from the Maryland Bridge Inventory were modeled in
DASH and analyzed for both the ASD and LRFD methods. The Inventory, maintained by SHA,
lists the majority of the bridges from the 23 counties in Maryland, including the important
information on the bridges, such as location, design load, bridge types and span length, as well
as the year when they were built. The design load is typically HS-20. H-20 and HS-25 loads
also occupy a certain proportion. Most bridges in this inventory have one span to three spans,
and the span lengths range from 30 feet to 300 feet. The sample bridges consist of 10
single-span bridges, 10 double-span bridges and 10 three-span bridges. The following chart
shows the length for each span of these bridges. The distribution of the span length of these 30
bridges covers most of the range of the bridge span length in the Inventory; therefore, these 30
bridges could reasonably serve as the sample bridges in this study.
37
bridges is shown Figure 5.2. It can easily be observed that the live load deflections from HS-25
are larger than ones from HL-93 through the two trend lines since the HS-25 truck is heavier.
However, all the bridges with these two different load types have acceptable deflections under
the L/800 deflection limit.
L/800 Limits
Figure 5.2 Deflection vs. Span Length for All Sample Bridges
Figures 5.3 to 5.6 show the comparison for bridges which have different numbers of spans
(single span, two spans and three spans for mid-span and side-span) with different vehicular
loads HS-20 for the ASD (or WSD) method and HL-93 for the LRFD method. Another value
added to this comparison is the deflection under the HL-93 design truck load multiplied by a
factor of 1.25. The trend lines for all these three values are shown in each figure and the tables
for live load deflection data are also displayed afterwards (Tables 5.1 to 5.3). In the ASD
method, the governing load types for each bridge are also listed in the tables. The capital letter
‘L’ stands for ‘lane load’ and ‘HS’ means the ‘HS-25 design truck’.
38
Figure 5.3 Deflection vs. Span Length for Single-span Bridges
39
Figure 5.4 Deflection vs. Span Length for Two-span Bridges
40
Figure 5.5 Deflection vs. Span Length for Three-span Bridges (Side Span)
Figure 5.6 Deflection vs. Span Length for Three-span Bridges (Mid Span)
41
Table 5.3 Three-span Bridges Live Load Deflection
WSD(HS-25) LRFD(HL-93)
Span L/800
Br. # Load HL-93x1.25
Length Deflection Deflection (in.)
Type
104 -0.376 -0.309 -0.386 1.560
107703 120 -0.641 HS -0.533 -0.666 1.800
112 -0.716 -0.591 -0.739 1.680
65 -0.259 -0.206 -0.258 0.975
216901 82 -0.383 HS -0.310 -0.388 1.230
65 -0.259 -0.206 -0.258 0.975
61 -0.259 -0.188 -0.235 0.915
308300 79 -0.393 HS -0.291 -0.364 1.185
61 -0.259 -0.188 -0.235 0.915
32 -0.062 -0.043 -0.054 0.480
318100 61 -0.264 HS -0.188 -0.235 0.915
32 -0.062 -0.043 -0.054 0.480
63.125 -0.176 -0.148 -0.185 0.947
803700 63.5 -0.138 HS -0.116 -0.145 0.953
63.125 -0.176 -0.148 -0.185 0.947
55 -0.173 -0.144 -0.180 0.825
1400501 55 -0.134 HS -0.111 -0.139 0.825
50 -0.145 -0.120 -0.150 0.750
165 -0.763 -0.470 -0.588 2.475
1619701 180 -0.908 L -0.587 -0.734 2.700
165 -0.763 -0.470 -0.588 2.475
33.167 -0.128 -0.099 -0.124 0.498
1703200 33.75 -0.099 HS -0.077 -0.096 0.506
33.167 -0.128 -0.099 -0.124 0.498
45.75 -0.207 -0.170 -0.213 0.686
2105700 51.667 -0.227 HS -0.188 -0.235 0.775
51.75 -0.304 -0.252 -0.315 0.776
80 -0.415 -0.352 -0.440 1.200
BHT80-99-80 99 -0.543 HS -0.469 -0.586 1.485
80 -0.415 -0.352 -0.440 1.200
42
With the tables and figures shown above, it is easy to see that the lane load governs for the
longer spans in length, and the design truck load governs for shorter spans. The live load
deflections under the HS-25 truck load are slightly larger than the ones under HL-93 with a
factor of 1.25; however, there is no obvious relationship between the live load deflection and
the various span lengths. The properties of the bridges themselves, such as different choices of
sections for the girders or different design of lateral bracing, would affect the behavior of the
bridges under vehicular loads. If the bridges are poorly designed, even a lighter load such as
HS-15 generates large deformation.
All the data above are based on the application of live load distribution factor with multiple
lane presence together, which were mentioned in Chapter Four. Using this method leads to a
relatively small effect on the deflection results, which is not conservative for analyzing bridge
behavior given the aforementioned. Therefore, in this section, the modified method of live load
distribution was applied, which is ‘effective-girder without live load distribution factors’ as
mentioned in the last part of Chapter Four. The live load deflections of 30 sample bridges under
the HS-25 design truck with the ASD method and the HL-93 design truck with the LRFD
method as well as the 1.25 times of deflections from HL-93 have been listed and analyzed as
the same as the previous analysis. In this case, however, the deflections of the HL-93 truck are
getting close to HS-25 and even slightly larger than HS-25 for double-spans bridges according
to the results shown in the Figures 5.6 to 5.10 and Tables 5.4 to 5.6, but still, all the deflections
are under the L/800 limit. This is because the multiple lane presence factors used in LRFD are
smaller than those in the ASD method (see Table 3.3 in Chapter Three).
Figure 5.7 Deflection vs. Span Length for Single-span Bridges with Modified
Distribution
43
Table 5.4 Single-span Bridges Live Load Deflection with Modified Distribution
WSD(HS-25) LRFD(HL-93)
Span L/800
Br. # Load HL-93x1.25
Length Deflection Deflection (in.)
Type
EB 100' 100 -0.876 HS -0.693 -0.866 1.500
201200 57 -0.508 HS -0.401 -0.501 0.855
300700 65 -0.431 HS -0.363 -0.454 0.975
502703 112 -0.716 HS -0.619 -0.774 1.680
1000400 63 -0.669 HS -0.516 -0.646 0.945
1504200 140 -0.912 L -0.766 -0.957 2.100
1610500 47 -0.160 HS -0.132 -0.165 0.705
1629400 140 -0.871 L -0.729 -0.912 2.100
EB 60' 60 -0.443 HS -0.359 -0.449 0.900
EB 200' 200 -1.505 L -0.981 -1.226 3.000
Figure 5.8 Deflection vs. Span Length for Two-span Bridges with Modified Distribution
44
Table 5.5 Two-span Bridges Live Load Deflection with Modified Distribution
WSD(HS-25) LRFD(HL-93)
Span L/800
Br. # Load HL-93x1.25
Length Deflection Deflection (in.)
Type
55 -0.244 -0.203 -0.254 0.825
101900 HS
55 -0.244 -0.203 -0.254 0.825
124 -0.937 -0.828 -1.035 1.860
206502 HS
124 -0.937 -0.828 -1.035 1.860
85 -0.899 -0.772 -0.965 1.275
319100 HS
97 -1.080 -0.937 -1.171 1.455
148 -0.375 -0.310 -0.388 2.220
603200 L
152 -0.400 -0.325 -0.406 2.280
100 -0.084 -0.073 -0.091 1.500
700300 HS
100 -0.084 -0.073 -0.091 1.500
53.5 -0.160 -0.133 -0.166 0.803
1202600 HS
53.5 -0.160 -0.133 -0.166 0.803
142 -1.647 -1.417 -1.771 2.130
1303003 L
142 -1.668 -1.431 -1.788 2.130
80 -0.917 -0.808 -1.010 1.200
1600400 HS
80 -0.917 -0.808 -1.010 1.200
87 -0.487 -0.419 -0.524 1.305
BHT87-87 HS
87 -0.466 -0.401 -0.501 1.305
137 -0.754 -0.564 -0.705 2.055
BHT137-137 L
137 -0.754 -0.564 -0.705 2.055
45
Figure 5.9 Deflection vs. Span Length for Three-span Bridges (Side) with Modified
Distribution
Figure 5.10 Deflection vs. Span Length for Three-span Bridges (Mid) with Modified
Distribution
46
Table 5.6 Three-span Bridges Live Load Deflection with Modified Distribution
WSD(HS-25) LRFD(HL-93)
Span L/800
Br. # Load HL-93x1.25
Length Deflection Deflection (in.)
Type
104 -0.418 -0.364 -0.454 1.560
107703 120 -0.712 HS -0.627 -0.784 1.800
112 -0.796 -0.695 -0.869 1.680
65 -0.288 -0.242 -0.303 0.975
216901 82 -0.426 HS -0.365 -0.456 1.230
65 -0.288 -0.242 -0.303 0.975
61 -0.345 -0.289 -0.362 0.915
308300 79 -0.524 HS -0.448 -0.560 1.185
61 -0.345 -0.289 -0.362 0.915
32 -0.083 -0.066 -0.083 0.480
318100 61 -0.352 HS -0.289 -0.362 0.915
32 -0.083 -0.066 -0.083 0.480
63.125 -0.176 -0.148 -0.185 0.947
803700 63.5 -0.138 HS -0.116 -0.145 0.953
63.125 -0.176 -0.148 -0.185 0.947
55 -0.173 -0.144 -0.180 0.825
1400501 55 -0.134 HS -0.111 -0.139 0.825
50 -0.145 -0.120 -0.150 0.750
165 -1.017 -0.723 -0.904 2.475
1619701 180 -1.211 L -0.903 -1.129 2.700
165 -1.017 -0.723 -0.904 2.475
33.167 -0.142 -0.116 -0.146 0.498
1703200 33.75 -0.110 HS -0.091 -0.113 0.506
33.167 -0.142 -0.116 -0.146 0.498
45.75 -0.207 -0.170 -0.213 0.686
2105700 51.667 -0.227 HS -0.188 -0.235 0.775
51.75 -0.304 -0.252 -0.315 0.776
80 -0.415 -0.352 -0.440 1.200
BHT80-99-80 99 -0.543 HS -0.469 -0.586 1.485
80 -0.415 -0.352 -0.440 1.200
Based on the analyses described above, a conclusion could be reached that lane load governs
those bridges having larger span length. The live load deflection under the HS-25 design truck
is larger than those under the HL-93 truck. A factor of 1.25 applied for HL-93 would increase
the deflection to become the same as HS-25 with or without distribution modification.
Therefore, the HL-93 truck load could be adopted as the design vehicular load whether they are
47
old bridges analysis or new bridges design to meet the L/800 limit of live load deflection. A
factor of 1.25 is suggested to make the current designs consistent with past practice, safer and
to stay on the conservative side.
where:
The results of the rating factor for both methods are shown in the following figures (Figures
5.11 to 5.13) and a ratio between LRFR and ASR are also shown in a third figure. The different
symbols stand for the design vehicular load for those bridges.
48
Figure 5.12 Load and Resistance Factor Rating
From Figures 5.11 to 5.13, the rating factors for almost all the bridges based on both methods
could be found to be larger than 1.0, which means the designs are appropriate. It could also
noted that the rating factors for the bridges designed in the period of 1990 to 2007 whose
design load are HS-25 would have larger values. This is because the lighter truck (HS-20 or
HL-93) has been applied in rating for the heavy-truck-design bridges. According to the results,
it could be concluded that using the HS-20 or HL-93 truck load is acceptable in design.
49
6. Construction Closure Pours Case Study
In the closure pour case study, two bridges, MD140 over MD27 (bridge no. 6032) and I-695
over Ingleside Avenue (bridge no. 0312300), were studied. Since the I-695 bridge is a straight
bridge with equally spaced girders and no skew and can be easily done within two stages, no
serious consequences with one closure pour were found. Therefore, only the MD 140 bridge
was studied. During the staged construction of the MD140 over MD27 bridge, the second
stage started when the first stage was completed and opened to travel from vehicle traffic. Then,
the third stage started afterwards when both stages 1 and 2 completed and opened to travel
from vehicle traffic. On the deck pouring sheet prepared by the SHA, the following statements
were made about the staged construction as part of the Maryland practice:
1. Closure pours placed after all adjacent pours are in place for a minimum of forty (40) hours.
2. Prior to placing closure pours the contractor shall verify that the girder dead load
deflections at girder nos. 5 and 12 (adjacent girders of the new stage) meet the required
elevations for the deck construction.
3. If the newly poured section of deck is higher than the required elevation, the contractor
shall place a load at midspan of the specified girder to deflect the girder to the required
elevation.
4. Place uniform loads over a length of 12 feet and placed symmetrically about the centerline
midspan of each girder.
5. The contractor shall not apply these loads until the concrete in the deck slab has reached a
minimum of 76% of the specified 28 days compressive strength and as approved by the
engineer.
6. The contractor shall submit calculations and details for the loading to the engineer for
approval.
During the construction of stage two, it was found that the new deck was higher than stage
one. The same phenomenon was also found during the construction of stage 3. Even though
some difference was expected, the differential displacements were higher than predicted.
Extra effort was then required to try to bring the elevation between stages to the same level.
The closure pour was then completed.
There were many factors that could have contributed to the differential displacements between
stages. Listed below are just several that the SHA project team brought up:
- Increased deflections resulting from a lower moment of inertia of the girders due to the use
of HPS
- Variable girder spacing on either side of the closure pour
- Different girder designs between stages
- Shrinkage of the concrete deck in the first stage.
50
More detailed analyses were made in the following section to determine the cause of the
differential displacements. Instead of shrinkage as stated above, creep effect was considered
more influential and was considered in the analysis process. Camber diagrams were not built
into the model, but were considered afterwards in order to find the actual differential
displacements.
The construction of the bridge was divided into three stages, as shown in Figure 6.3. The first
stage covers the middle strip from Girder 6 to Girder 11, the second stage consists of Girder 1
to Girder 5, and the last stage consists of Girder 12 to 15. For each stage, the construction was
arranged into three pouring sequences. The first sequence covers the 102 feet in the positive
moment area of longer span from the north abutment, the second sequence covers the 99 feet in
the positive moment area of short span from the south abutment, and the final sequence is the
remainder of the bridge. Between each construction stage, closure pour was adopted. The
displacements at the adjacent girders between stages 2 and 3 were studied to evaluate the
serviceability of the bridge during construction.
51
Figure 6.2 Girder Elevations (Mid Strip)
The bridge was selected for this study because of reported problems associated with closing
gaps between stages. The original design used the line-girder program Merlin-DASH to
establish the camber diagrams. They were grouped into eight camber diagrams, which are (1)
Girder 1, (2) Girders 2-5, (3) Girders 6 & 7, (4) Girder 8, (5) Girder 9, (6) Girder 10 & 11, (7)
Girders 12-14, and (8) Girder 15. The girders adjacent to the closure pours are Girders 5 and 6
between stages one and two as well as Girders 11 and 12 between stages two and three. The
case study is on the sixth camber diagram group for Girder 11 and the seventh camber diagram
group for Girder 12 where the second closure pour is located. Because of the similarity of
cambers due to weight of the parapets and sidewalk (ΔS.D.L.) and vertical curvature of the
52
roadway (ΔV.C.) among all girders, only the maximum cambers on the longer span (152′) due
to the self-weight of girder (ΔGIRDER) and concrete slab and deck forms (ΔD.L.) of Girders
11 and 12 are compared. From the camber diagram the sums of dead load cambers (ΔGIRDER
+ ΔD.L.) are 4 3/8″ and 5 9/16″, respectively, with a difference of 1 3/16″ between two girder.
However, by viewing the framing plan, it was found that the camber diagrams for Girders 10
and 11 should not be grouped together where the tributary width for Girder 10 is 7′-3″ and that
for Girder 11 is 8′-6″ (= 0.5[7′-3″ + 9′-9″]). With the average tributary area considered for
Girder 11, the sum of dead load cambers (ΔGIRDER + ΔD.L.) is now 5 1/8″ and the camber
difference between Girder 11 and 12 is reduced to 7/16″, instead of the original 1 3/16.″
Therefore, the first priority of matching the elevation between stages is to construct correct
camber diagrams with proper tributary area considered.
In order to obtain the global displacement profile under staged construction, the whole bridge
was modeled and re-analyzed using both DESCUS-I and CSiBridge computer programs. The
MD140 over MD27 Bridge was first constructed as a two-dimensional grid model in DESCUS.
In DESCUS, the parabolic sections were simulated using equivalent web depths varying from
45 inches to 81 inches, shown in Figure 6.4. Girders 11 and 12 are the edge girders of each
construction stage, and their reactions were monitored in this study.
The girder displacements of the non-composite stage were analyzed at first. Then, the bridge
model simulating the construction stages was introduced. To simplify the analysis, in this
staging construction model, Girders 1 to 11 were modeled as stage one with fully matured
concrete slab, while Girders 12 to 15 were modeled as stage two with concrete slab of one-day
age to simulate the stage before closure pours. To discuss the influence of bridge diaphragm
during construction, two sets of models were built for comparison. One set has diaphragms
between Girders 11 and 12, another set does not.
Closure pour
Figure 6.5 CSiBridge Model for MD140 over MD27 Bridge Isometric View
1. The first CSiBridge model was used to simulate non-composite behavior for the early
bridge construction. Since in the non-composite period, both bridge girders and slab
contribute to the stiffness property of the composite section, the constraint between girder
and deck was defined that corresponding joints only share the same vertical displacement.
2. For the second CSiBridge model, Young’s module of the bridge slab was set to zero,
similar to the assumption of the DESCUS model. Although such a method is not
recommended by CSi knowledge base, it can be used to validate the CSiBridge
non-composite model and compare later with the staging models.
3. In the third CSiBridge model, time dependent properties of materials and the construction
staging were introduced into the bridge model. Concrete compressive strength, stiffness, as
well as creep were considered for materials properties (Figure 6.6a-c). The deck
construction of Girders 1 to 11 was defined as stage one, construction of Girders 12 to 15
was defined as stage two. A 28-day period was defined between these two stages to allow
the concrete from the stage one construction to develop creep. A 3′7″ gap between Girder
11 and 12 was created to represent the closure pour gap during staged construction. The
maximum displacement of each girder at 60 feet from the north abutment was monitored
and the difference between Girders 11 and 12 (the edge girders of each stage) was also
observed and discussed.
54
Each model also has two sub-sets, with and without diaphragms between Girders 11 and 12,
therefore the effect of diaphragms in staged construction was modeled.
55
Figure 6.8(c) Time Dependent Creep Coefficient in CSiBridge Model
56
Table 6.1 Maximum Displacements Comparison for Normal and Staging Models by
DESCUS
Tables 6.2 and 6.3 show the comparison of results between various CSiBridge models and the
DESCUS model. By comparing the displacements from the non-composite model of
CSiBridge with that from the non-composite model of DESCUS, it can be observed that these
two results are very close. More detailed figures (Figures 6.7 and 6.8) show the displacement
along the entirety of Girder 11 and Girder 12. The conclusion can be made that the bridge
response during non-composite period can be well simulated by both DESCUS and CSiBridge
program.
57
Figure 6.9 Vertical Displacement Results from CSiBridge (with Diaphragm) (inch).
Figure 6.10 Vertical Displacement Results from CSiBridge (without Diaphragm) (inch).
The results from the second CSiBridge model with the Young’s module of bridge slab set to
zero are compared here with the results from the DESCUS models. The results from the
DESCUS and CSiBridge non-composite model are close, with small margins of differences.
Such differences exist because in CSiBridge setting concrete Young’s module to zero does not
necessarily make the bridge non-composite, though in theory these two assumptions should
yield the same results.
The results from staged construction models were also compared. The results shown in Tables
6.2 and 6.3 are the final displacements, where the bridge stage one concrete has an age of 31
58
days and the stage two concrete has the age of 3 days. Since the time dependent properties of
material are introduced in this model, the stage one concrete displacement is increased due to
the creep effect. It can be observed that the displacement from Girder 11 has a noticeable
increase, while the displacement from Girder 12 stays roughly the same (changed slightly due
to diaphragm dragging). A detailed investigation about how the maximum displacement
developed with time was conducted. The following figures (Figure 6.9 and 6.10) show the
creep development for both “with diaphragm” and “without diaphragm” models. The
displacement of Girder 11 (shown as a red line) between day 3 to day 30 is slowly increased
about 0.2 inches. This displacement increment is the effect from concrete creep behavior. Also,
it is noticed that for the “with diaphragm” case, Girder 11 deflects further after the second
construction phrase is completed. This is due to the diaphragms connecting the neighbor girder
of the next stage helping to distribute the load from the second stage to the first stage. Such
behavior is not shown in the “without diaphragm” model.
Comparing the results from the CSiBridge and DESCUS models, some displacement
differences are found. The reason is in CSiBridge, a 3′7″ closure pour gap between Girders 11
and 12 was created to model the bridge during construction precisely. The difference due to
loading can be narrowed down if the dead loads due to closure pour gap concrete are removed
with closure pour gap simulated. However, if the creep effect is considered, the CSiBridge
model could generate more realistic results.
Figure 6.11 The Creep Growth in Girder 11(Orange) and 12(Green) for with
Diaphragms Model (in).
59
Figure 6.12 The Creep Growth in Girder 11(Orange) and 12(Green) without
Diaphragms Model (in).
Figure 6.11 shows the displacement increment for Girder 11 over time from the without
diaphragm model. To better represent the real scenario, concrete deck was modeled to carry
load start at the age of 7 days. It can be observed that the creep develop relatively fast in the
first 30 days (22.63%), after 30 days the displacement growth become steady and gradually
reduced. The conclusion can be made that for bridge construction, creep effect has strong
impact on bridge during first 30 days.
Figure 6.131 Displacement Increment for Girder 11 without Diaphragms Model (in).
60
7. Summary and Conclusion
As a very important issue in the serviceability of steel bridges, live load deflection still
attracts great attention because ensuring safety is always the number one priority in structure
design. Bridge structures should be designed with sufficient strength but designers also need
to ensure the deflection is within an acceptable range so that drivers or passengers in the
vehicles would not believe the bridge is unsafe.
61
B: Findings associated with bridge construction closure pours -
5. The general practice in Maryland is to use a line-girder program to establish the camber
diagrams. The result is generally accurate enough and acceptable in practice.
Multiple girders with varied girder spacing are grouped into several camber diagrams.
It is noticed that camber diagrams are not grouped based on their tributary widths, but
rather, on the narrower girder spacing. In this case some girders may be
under-cambered. This does not usually cause problems for one-stage construction, but
may cause trouble for staged construction if one side of the closure pour is
under-cambered.
6. Multiple camber diagrams can be calculated by the line-girder models, two-dimensional
grid model or three-dimensional finite element model. All three methods generate
results accurate enough for straight girder systems, if the creep effect is not considered.
7. Maryland adopted the generally recommended practice of a minimum closure width of
three (3) feet and diaphragms/cross frames in the staging bay of structural steel girders
not rigidly connected until later. Not connected until final pour is a general practice by
many states, and is suggested by this study. There is no ill effect for non-connected
practice.
8. To investigate the staging effect of staged-construction, two-dimensional grid models and
three-dimensional finite element models are highly recommended. The differential
displacement between stages could not be considered in a one-dimensional line-girder
model.
9. When comparing the results of the two-dimensional grid model with those of the
three-dimensional finite element model, these two methods produced results accurate
enough for straight girder systems, if the creep effect is not considered. However, since
the two-dimensional grid model program has to simulate the closure pour by subtracting
the deck loads during staged construction, the three-dimensional finite element model
can be more closely simulated with a gap on the deck. Also, in the case where the
closure pour is significant, the three-dimensional finite element model would provide
more accurate results.
10. For further staging analysis, the creep effect of concrete was considered in this study.
Two controlled sets of models were studied. One model assumes the diaphragms in the
closure pours always connect with girders during the whole phrase construction. The
other model assumes the diaphragms are disconnected. To investigate the creep effect
with concrete slab, two different models analyses from DESCUS and CSiBridge were
performed. For the staging analysis of DESCUS, only the time-dependent property of
Young's Modulus was considered. However, for the CSiBridge, the staging analysis not
only simulated the time-dependent property of Young's Modulus, but also considered the
creep effect in concrete. If creep effect would be considered, computer programs
equipped with creep analysis feature should be adopted.
11. For general bridge with constant girder spacing, due to creep effect, the old stage built in
the early stage would deflect more and the displacement gap between stages would
increase. However, correct cambers would alleviate creep effect.
62
12. Creep effect on concrete occurs at an early stage. Due to improper camber, excess
loading due to superimposed dead load and live load in the early stage, and the creep
effect, the new deck can be expected to be higher than the existing deck. Based on the
analysis, the MD 140 bridge (bridge no. 6032) is expected to have a two (2) to three (3)
inch difference in elevation, which is also reported from the field and this differential
displacement between stages could be alleviated by proper camber and scheduling on
pouring.
13. To achieve better result during stage construction, 30-days waiting period is
recommended between finishing the new deck pour and starting the closure pour. In this
way the creep effect from both the old and new construction stages would enter a steady
growth stage and the displacement gap between these two stages would be narrowed.
63
References
’96 Australian Bridge Design Code (1996). Section Six-Code Steel and Composite
Construction, Austroads, Haymarket, Nsw, Australia.
AASHTO (2014), Load Resistance and Factor Design: Bridge Design Specifications, Seventh
Edition, American Association of State Highway and Transportation Officials, Washington
D.C.
American Institute of Steel Construction. (2012), Steel Construction Manual, 14th Edition,
Chicago, IL.
Anderson, E. (2005), Deflection Serviceability for Steel Girder Bridges, Master’s Thesis,
University of Wyoming, Laramie, WY, December.
Aramraks, T. (1975), Highway Bridge Vibration Studies, Joint Highway Research Project,
Indiana State Highway Commission, Purdue University, West Lafayette, IN, February.
Barker, M.G. and Barth, K.E. (2007), Live Load Deflection Serviceability of HPS Composite
Steel Girder Bridges, 2007 World Steel Bridge Symposium Papers, New Orleans, LA, Dec.
Barth, K., Bergman, A. and Roeder, C. (2002), Improved Live Load Deflection Criteria for
Steel Bridges, Report to the National Cooperative Highway Research Program. University of
Washington, Seattle, WA, May.
Barth, K.E., Roeder, C.W., Christopher, R.A., and Wu, H. (2003), Evaluation of Live Load
Deflection Criteria for I-Shaped Steel Bridge Design Girders, ASCE, Journal of Structural
Engineering Special Publication – High Performance Materials in Bridges, 193 – 208.
Barth, K., Bergman, A, and Roeder, C. (2004), Effect of Live-Load Deflections on Steel
Bridge Performance, Journal of Bridge Engineering ASCE, University of Washington, Seattle,
WA, May
Christopher, R.A. (2001), Live Load Deflection Criteria for Steel I-Girder Bridges, Masters
Thesis. West Virginia University, Morgantown, WV, February.
DeWolf, J.T., and Kou, J-W. (1997), Vibrational Behavior of Continuous Span Highway
Bridge-Influence Variables, Journal of Structural Engineering, March.
64
Fountain, R.S., and Thunman, C.E. (1987), Deflection Criteria for Steel Highway Bridges,
Presented at National Engineering Conference, New Orleans, LA, April.
Gandiaga, Lorehana (2009), Serviceability Limits and Economical Bridge Design, MS,
Laramie WY, May
Goodpasture, D.W., and Goodwin, W.A. (1971), Final Report on the Evaluation of Bridge
Vibration as Related to Bridge Deck Performance, The Tennessee Department of
Transportation, The University of Tennessee, Knoxville, TN.
Nevels, J.B., and Hixon, D.C. (1973), A Study to Determine the Causes of Bridge Deck
Deterioration, Final Report to the State of Oklahoma Department of Highways. Oklahoma
City, OK.
Ontario Highway Bridge Design Code, Second Edition (1983), Ontario Ministry of
Transportation and Communications Highway Engineering Division, Toronto, Ontario.
Taly, N., “Design of Modern Highway Bridges,” The McGraw-Hill Companies, Inc., New
York, 1998
Walker, W.H., and Wright, R.N. (1971), Criteria for the Deflection of Steel Bridges, Bulletin
for the American Iron and Steel Institute, November.
Forrest Gregory Ehmke (2006), Analysis of a bridge deck built on interstate highway 39/90
with full-depth, precast, prestressed concrete deck panels, Master Thesis, University of
Wisconsin-Madison.
Failure and repair of a deck closure pour on interstate 81, Virginia concrete conference,
65
Richmond, Virginia, March 4th and 5th, 2010.
H.J.Hung and Y.H.Chai (2011), Final Report No. CA10-0928, Shortening Closure Pour
Waiting Time for Bridge Construction, Department of Civil and Environmental Engineering,
University of California at Davis, June.
Hung-Jung Hung (2012), Evaluation of Waiting Time for Closure Pour in Concrete Bridge
Widening, PhD dissertation, University of California, Davis.
NCHRP Synthesis 345, Steel Bridge Erection Practices (2005), Transportation Research
Board.
Synthesis into the Cause of Concrete Bridge Deck Cracking and Observations on the Initial
Use of High Performance Concrete in the US95 Bridge over the South Fork of the Palouse
River (2008) National Institute for Advanced Transportation Technology, University of Idaho.
Gary Novey, Bridges and Structures, Bridge Superstructure Design-MM No. 10 (Closure
Pours)
Seunghwan Kim (2014). Creep and Shrinkage Effects on Steel-Concrete Composite Beams,
Masters Thesis, Virginia Polytechnic Institute and State University, May.
Jiansheng Fan, Jianguo Nie, Quanwang Li, and Hao Wang (2010). Long-Term Behavior of
Composite Beams under Positive and Negative Bending. I: Experimental Study, Journal of
Structural Engineering, July.
Weichen Xue, Min Ding, Chi He, and Jie Li (2008). Long-Term Behavior of Prestressed
Composite Beams at Service Loads for One Year, Journal of Structural Engineering, June.
X. Liu, R. E. Erkmen and M.A. Bradford (2012). Creep and Shrinkage Analysis of Curved
Composite Beams Including the Effects of Partial Interaction, Proceedings of the Eleventh
International Conference on Computational Structures Technology, Civil-Comp Press,
Stirlingshire, Scotland.
66