Professional Documents
Culture Documents
Greg Callahan
University of Missouri-Columbia
Advisor:
Prof. Mertz
Submitted to:
NSF-REU, University of Delaware Civil Engineering
Dept.
2
3
Abstract
The goal of this study is to determine stress ranges felt on Delaware Bridge #771.
This bridge contains two fatigue-sensitive details experiencing different stress ranges.
The determination of these stress ranges will be done using the structural analysis
computer program Staad.Pro 2004. In order to apply Staad, a model of DE Bridge #771
is generated and analyzed within the computer program. Ultimately, this study
determines if the fatigue-sensitive details are exposed to high enough stress levels to
cause fatigue failure and if so what the remaining fatigue life of each detail would be.
The concepts and methods used in analysis of the bridge come from studies conducted by
Dr. John Fisher of Lehigh University, AASHTO, and the NCHRP. Hopefully this study
will set forth a methodology that is easily understood and reproducible so that others can
4
Introduction
Bridge fatigue can be a serious problem. If ignored, fatigue can cause sudden and
total failure of a bridge. Fatigue failure of a detail can occur at stress levels far below the
ultimate strengths of that material. Fatigue failure is dependent on time and frequency of
loading; moreover, failure occurs due to the repeated cyclic loading of a structure.
larger number of loads cycles and are more susceptible to failure at lower stress ranges
than newer bridges that have not been exposed to as many cycles. The time it takes for a
structure to fail due to fatigue is called the fatigue life. Fatigue life can be calculated
throughout the life of a structure. Remaining fatigue life is the time remaining until
fatigue failure is
Bridge plans provide the necessary details for modeling in the computer program
Staad. Pro 2004. This program can determine the maximum and minimum stresses each
girder experiences. From these values a stress range can be determined and this can be
used to determine the remaining fatigue life of that detail. There are approximately
Pennsylvania if both directions are considered. The majority of these bridges are
relatively new so there is no immediate risk of fatigue failure, but determining the stress
ranges of the bridges would allow the remaining fatigue life to be calculated providing
Due to the relatively short time I spent at the University of Delaware, stress-
range calculations for all thirty-six bridges were not possible. Instead, Delaware Bridge #
771 was chosen for analysis. Bridge # 771 is a simply supported three span bridge that
5
runs over Shipley Rd in the direction of Philadelphia from Wilmington. This bridge was
chosen because it has fatigue sensitive cover plates, and is very similar to a number of
highways overpass bridges. Hopefully, it will be able to act as a case study for future
analysis of similar bridges on I-95 and other highways as well as aid in further studies on
Background
In order to understand why fatigue analysis is important, the nature and behavior
of fatigue failure must first be addressed. Fatigue is the process of cumulative damage in
a benign environment that is caused by repeated fluctuating loads and, in the presence of
Fatigue is a property that can be affected by a number of factors. The state of stress
carried by the structure, the geometric properties of the design, and the environment all
can influence the fatigue resistance of a structure (Barson & Rolfe, 1999). For this study
6
of Delaware Bridge #771, the geometric properties of the bridge design proved to be the
largest factor for fatigue. Fatigue is an issue for fracture-critical details such as the
As set forth by the AASHTO Manual for Maintenance Inspection of Bridges, the
details on bridges like bridge #771 are inspected at two-year intervals for fatigue analysis.
However, neither the AASHTO Manual for Maintenance Inspection of Bridges, nor
Standard Specifications for Highway Bridges yielded accurate or realistic fatigue life
approximations (NCHRP 299, p 5). Due to the need to have a standard, accurate, and
reproducible method to determine the remaining fatigue life in bridges, the National
Cooperative Highway Research Program, NCHRP, produced report 299. This report
covers procedures for fatigue evaluation of existing steel bridges, and how to design for
A table of these details and classifications can be found in Appendix A. These details
create points of stress concentrations that can, and often do, result in large stress ranges
ultimately causing fatigue failure. The type of stress these details are subjected to also
influences the degree of fatigue failure. In order for a bridge detail to be considered a
fatigue detail, it must be exposed to applied tensile stresses (Mertz). This is an extremely
important concept to address before any fatigue analysis takes place. The fatigue details
on Delaware Bridge #771 consist of longitudinal welds and transverse welds that attach
cover plates to the bottom flange of the girders in the middle portion of each span.
Because these details are located on the bottom flange of a simply supported beam, they
experience tensile stress. This can be assumed because of how stresses are distributed in
7
an I-Beam girder. The top flange will support compressive stresses while the bottom
Fig 2 - Cross-Section of a sample W36X135 beam found in Delaware Bridge #771. These beams have cover plates,
As seen in Fig 2, the cover plate clearly falls below the neutral axis of the I – Beam. The
added mass to the beam increases the moment of inertia of the beam and allows the girder
8
Fig 3- Moment diagram generated from the structural analysis program Staad.Pro 2004.
Figure two illustrates the moment diagram for one W36X135 girder with a bottom cover
plate. The load on the girder is a HS-15 truck, a standard fatigue truck as set forth by the
AASHTO Guide Specifications for Fatigue Evaluation of Existing Steel Bridges (1990).
The model was created in the structural analysis program Staad. Pro 2004. The sign
analysis sign convention the maximum moment for figure two is a positive 3906 kip-in.
Another interesting property of Staad is that while the moment diagrams produced are off
by a factor of negative one, the diagrams do accurately reflect the deformed shape of the
beam under the set loading conditions. So, from figure 2 it can be concluded the bottom
flange of the beam is indeed subjected to tensile stress and therefore, the bottom flange
cover plate welds are fatigue - sensitive details. This means that Delaware Bridge #771 is
9
Fatigue failure was first seen around 120 years ago by Wohler who noticed stress
concentrations and sharp angles in axle configuration of railway rolling stock that led to
failure well below the ultimate strength of the material (Barsom 16). Studies for weld
examinations in the 1930’s and later in WWII laid the initial groundwork for fatigue
design in the North America and led to the fatigue test program sponsored by the
Before the establishment of the NCHRP fatigue analysis was very limited based only on
small specimens and on a limited quantity of test data, this is why many existing bridges
built before 1970 have large amounts of fatigue cracking (Barsom 16). Since the
establishment of a fatigue analysis program by the NCHRP, fatigue failure has become
less and less of a problem due to standardized testing methods and continual research
This study deals with the stress ranges experienced by girders due to loading. The
stress range a girder is subjected to is a key factor in the remaining fatigue life of that
girder. The equation below is the equation used to determine remaining fatigue life as
The variables above are given in the table on the next page.
10
Variable Represents
Table 1 – explanation of the variables found in NCHRP 299 remaining fatigue life equation
In order to determine if the equation above is needed to find the remaining fatigue
life, it is it must be determined if the details are exposed to stresses larger than their stress
threshold values. These detail classifications and threshold stresses were determined by
Dr. John Fisher of Lehigh University in the late 1960’s and early 1970’s. Fisher
construction would fail at different stresses for the same amount of cyclic loading. In
details. These details were each developed using a 95 % confidence limit (Fisher). This
means that many trials were conducted, and the failing points were analyzed. After the
analysis of the data was conducted, the line that 95 % of the failures occurred at or above
was found to be the stress threshold. The dashed line in Fig 4 illustrates this limit
11
Fig 4 - Example of the 95 % confidence limit method used by Dr. John Fisher of Lehigh.
If the details were subjected to stresses below this confidence limit it can be assumed that
they have an infinite theoretical fatigue life, meaning they will not fail due to fatigue. If
they are exposed to stresses above the threshold limit, the detail will have a finite fatigue
life and once the appropriate number of cycles has been applied fatigue failure will occur.
The details Dr. Fisher established ranged from the least fracture - critical detail (A), to the
most fracture-critical (F) (Fisher). An interesting aspect of Fisher’s study is that even
though his work was conducted over thirty year ago, his detail classification system is
12
still used today. AASHTO adopted the system and made one rather sizeable change.
The category F was not easy to understand and some problems occurred with accurately
recognizing and analyzing the details. To remedy this, AASHTO renamed the category
E’ and set forth parameters that were easier to understand and apply. These details along
with the corresponding stress ranges can be seen in table 2 below. The table clearly
shows that class A details can be exposed to much more stress on a consistent basis than
A 24
B 16
B’ 12
C 10
C’ 12
D 7
E 4.5
E’ 2.6
The particular classifications of details this study will examine are Class B and Class E
The reason that this study is interested in details B and E is they are the fatigue
sensitive details found on DE Bridge #771. The Bridge is composed steel girders with
cover plates on the bottom flange. On top of these girders sits an 8 in. concrete deck.
13
The deck and the girders act compositely. For a more detailed description of DE Bridge
#771 look in Appendix C. Class B and E details can be found on the welds which attach
the cover plate to the bottom flange. The transverse welds are class E details and the
longitudinal welds are class B details. Figure 5 shows the actual welds on bridge #771.
Fig 5 – Shows Class E and Class B fatigue sensitive details on DE Bridge #771
14
Methods and Procedures
The main task of this study was to correctly model DE Bridge # 771 in the Staad.
Pro 2004 computer program, Staad, so the stress ranges found would be correct.
Initially, the entire bridge was modeled in Staad. However, this did not seem to be a
practical process to determine the stresses felt by the girders. Each span of the bridge is
simply supported so, essentially the bridge can be considered as three individual spans.
It was assumed each span acts independently of the other two spans. In accordance with
this observation, a single span was chosen for modeling and analysis. This single span
was the longest span on the bridge. It was selected because it will carry the largest
moment, and thus be exposed to the highest levels of stresses. The modeled span had a
length of sixty – six feet and was composed of six W36X150 rolled girders, ten
MC18X42 steel channel diaphragms, an 8-inch concrete deck, and W40X655 steel
members. The 40X655 members acted as rigid connectors between the steel girders and
15
Fig 8 – Axle load and spacing of an HS-15 truck
Once the span was selected and modeled the loading needed to be applied. In
order to do this AASHTO procedures were followed. AASHTO calls for an HS-15 truck
to be used in fatigue design and evaluation. Figure 7 shows the initial position of the
truck loading (the red outline) on the span model. It should be noted that the truck is
placed on the right side of the bridge. The placement represents the most likely truck
traffic. On actual highway bridges, the outer lanes undergo a much higher frequency of
truck loading than the inner lanes. This placement acts to simulate real – world traffic
conditions. Figure eight shows the axle loads and spacing of the HS-15 truck used. The
actual test. This could be done in Staad using the load generation options. For a detailed
Appendix D.
Simulating a moving load across the modeled span was not the only problem
encountered during the modeling process. Instability issues became quite frustrating as
16
analysis of the model was conducted. Originally, the girders were modeled in three
pieces. Two smaller sections of ten feet and one forty-six foot section. The girders were
modeled this way because Staad was not able to model one member with a cover plate
covering only a portion of the member. Figure nine shows how the members were
modeled.
However, once the diaphragms were attached instabilities were recorded at the
joints where the diaphragms met the girders. While investigating the instability issue, a
discovery took place. The problem was not one of stability, rather a nodal problem with
the connections of the members. Staad uses defined points called nodes to serve as data
points for defining members, plates, different structures, etc. Members that composed the
girder sections of the span did not terminate and begin at the nodes the where the
diaphragms were connected. Because of this, Staad was not able to register the two
components were even connected. Once the problem was realized the original forty-six
girder section was broken down into five smaller sections shown in figure ten.
17
Fig 10 – Final modeling of girders with cover plates
This new method of producing the same girders fixes the instability problems seen in
The girder – diaphragm connections were not the only place instability issues
arose during modeling. Once the steel configuration was modeled and stable under
loading, the concrete deck had to be added to the model. This was done using plate
geometry in Staad. The first attempt to model the deck simply consisted of making one
large plate to sit on top of the steel sections. This method did not work and instability
problems as well as a non-composite action resulted under loading. The next attempt
data points in Staad. The deck was modeled as a synthesis of twenty smaller plates that
act together to simulate on large plate that covers the entire modeled span. The plate
elements were then given the thickness of eight inches and assigned concrete properties.
18
Figure eleven shows how the twenty smaller plates were configured to produce the
concrete deck.
Once the deck and the steel components of the bridge were modeled correctly, the
two sections needed to be connected in a manner that would result in composite action of
the girders and deck. The real DE Bridge #771 acts compositely so the generated model
needed to illustrate the trait as well. Many attempts were made to create a composite
model. The first was simply placing the deck on top of the steel without node
connections. This resulted in non-composite action and in the deck passing through the
webs of the steel girders. To remedy this mistake, the concrete deck was offset 1.495 ft,
which translates to the distance from the centroid of the girder’s cross section to the
centroid of the deck’s cross section. This attempt saw the deck placed correctly on the
beam, but composite action still did not occur when the model was loaded. Finally,
robust steel members were placed at 16.5 ft along each girder to act as rigid connectors
between the steel and the deck. The members used were W40X655 beams running in the
vertical direction. These beams benefited the model in two ways; first, they proved a
19
means to connect the deck to the steel and second, they created many nodes along the
girders to serve as more data points for analysis. Figure twelve shows a three-
As the figure above shows the connectors were very large beams. This was extremely
important because deformation of the connectors would throw off the stress ranges
determined from the model. Stiff members were selected due to the large values of
stiffness they posses. Once the rigid connectors were added, the model reflected
composite action. For in depth illustrations of stress distribution experienced by the deck
during each of the ten different load cases please refer to Appendix B.
With the model completed and simulating the actual DE Bridge #771, analysis of
the model needed to take place. However, the bridge was composed of six longitudinal
girders only one of which was loaded directly by the simulated HS- 15 truck. Originally,
it was thought that distribution factors would have to be calculated for analysis of all the
girders in the deck. On the first analysis run of the completed model, an interesting
20
discovery took place. Staad distributed the truckload over the girders for each generated
load case. At each of the ten different positions of the truckload green arrows signified
Fig 13- The load distribution of the HS-15 truck by Staad for the first and fifth load case
As figure thirteen illustrates the initial position of the truck stays consistently outlined as
red axle loads while the actual position and distribution of the generated load is shown by
the green arrows. For detailed positioning distribution of each generated load please
consult appendix B.
At this point of the project the loads were correctly defined and placed on the
structure, and the model correctly simulated DE Bridge #771 Stress analysis of the
Fig 14 – Three-dimensional renderings of the model used for stress range analysis
21
Data Analysis and Discussion
With the model completed the stresses felt in the steel girders that contained the
completing this task. After running an analysis, the “beams” tab was selected and from
that menu, the “stresses” option was chosen. In the “stresses” analysis screen any point
along any girder in the model could be selected and the stress felt by that girder during
any of the ten load cases could be accessed. This option made data collection very easy.
For the transverse welds along the ends of the cover plates the stresses were
recorded at the point the plate and bottom flange are welded together. This point was the
obvious choice since the weld itself is the fatigue-sensitive detail. Staad allowed exact
positioning of this point on the cross-section of the girders. Figure 15 shows form where
Stresses were taken from both ends of all the girders that had cover plates so twelve
readings in all.
The stresses at each end of the six girders that had cover plates were collected
from the model and placed into the Microsoft Excel program. From there graphs were
generated to find the stress ranges and see if the stresses felt exceed the threshold stress
of the detail. These graphs can be seen in Appendix E. The stresses felt by the Class E
22
details in the bridge simulation did not exceed the threshold stress of 4.5 Ksi. All other
trends displayed by the graphs were to be expected. The details located on the right side
of the bridge experience much greater stresses than those on the left. This was due to the
placement of the load on the right hand side of the bridge to simulate truck traffic.
In addition to the stresses felt by the class E transverse weld, the stresses in the
longitudinal welds were also determined. The longitudinal welds on the girders are a
class B fatigue sensitive detail so it has a higher stress threshold of 16 Ksi. Stress
readings were taken along the six beams containing the cover plates at 0, 11.5, 23, 43.5,
Fig 17- Girders that were analyzed for stresses felt in longitudinal welds
23
Once the data was colleted it was again exported to Excel and graphs were generated to
determine the stress ranges. These graphs can be seen in Appendix E. According to the
graphs, girders one and two experience the largest stress from the generated loads.
Because of the truck placement, this is expected. The maximum stress felt by any girder
does not come close to the threshold stress of 16 Ksi. Therefore, the results of the study
show that the threshold stress of both class B and class E details is not exceeded.
Conclusion
This study showed that DE Bridge #771 has a theoretical infinite fatigue life.
Both fatigue-sensitive details are not exposed to stresses exceeding their respective stress
thresholds. The equation outlined in NCHRP 299 does not need to be applied to either of
the details in this study because the fatigue life of the bridge is in theory infinite. All
calculations and analysis was done using the structural analysis program Staad.Pro 2004.
In this report the methods of modeling and analysis have been laid out so hopefully, they
will be useful for individuals who wish to study stress ranges in other bridges. This study
has proven insightful and very helpful as that it covered an area of immense personal
interest and taught the method and application of computer simulation using the Staad
program.
24
Acknowledgements
• Elliot Fink
• Mark Guzda
• Tim Stuffle
• Michelle Bensi
25
References
• Fisher, John W. "Bridge Fatigue Guide Design and Details." American Institute
of Steel Construction, New York, N.Y., 1977 p. 17-20.
• http://www.oregon.gov/ODOT/TD/TP_RES/docs/Reports/FatigueCrack.pdf
• Moses, F., C.G. Schilling, and K.S. Raju. "Fatigue Evaluation Procedures for
Steel Bridges." National Cooperative Highway Research Program Report
November 1987: 11-16, 70-74, 74-77, 59-67.
26
Appendix A
Fatigue Detail Pictures and Ilustrations
27
28
Appendix B
29
Load Case 1
30
Load Case 2
31
Load Case 3
32
Load Case 4
33
Load Case 5
34
Load Case 6
35
Load Case 7
36
Load Case 8
37
Load Case 9
38
Load Case 10
39
Appendix C
40
Overpass at Shipley Road Plan and Elevation
41
Overpass at Shipley Road N. B. L. Framing Plan
42
Overpass at Shipley Road N. B. L. Deck and Approach
Slabs Plan and Bar Schedule
43
Overpass at Shipley Road N. B. L. and S. B. L. Sections
and Details
44
Appendix D
Procedure for Defining and Generating Moving Truck Loads
Using Staad. Pro 2004
45
Load Definition
The first step in applying a moving truck load to a model is to define the load.
Figure1- D shows how to define the loadings option by selecting the command tab on the
main toolbar. From there, highlight the definitions option and ultimately select the
vehicle option.
46
After the load is defined as a vehicle, the actual weight of the load needs to be
entered into the program. This can be done in one of two ways; the axle spacing and weight
of the vehicle can be entered manually, or the vehicle can be selected from a number of pre-
defined trucks in Staad. Fortunately, the HS – 15 truck is already defined and can be selected
47
Moving Load Generation
Defining the load as a vehicle and assigning the properties of an HS-15 truck are
necessary steps, but in order to simulate a moving truck the load must be placed at a number
of positions on the bridge. In order to do this a moving load must be generated using Staad.
From the Commands tab select the loading option and highlight the moving load generation
option. From there a pop-up box will appear and want to know the number of load
generations desired. For the study a 10 generated loads were used. After the number of
truck loads generated has been determined a box will show on the right side of the screen.
Within this box highlight the load case with the generated loads. Select the add button at the
bottom of the box and this will cause a pop-up menu to appear. Inside the initial position of
the vehicle can be determined in the x, y, and z directions. In addition to the initial position,
the spacing interval can be entered for the distance between each generated load in the x, y,
and z directions. Finally, an outline of the load will become visible on the applied structure.
48
Fig 4-D Selection of number of generated loads
49
Fig 5-D Determination of initial position and intervals of generated loads
50
Final Generated load
51
Appendix E
Stress Range Graphs for Transverse
and Longitudinal Welds
52
Transverse Welds
Northern Side
53
Southern Side
54
Longitudinal Welds
Girder One
2500
Load Case 1
2000
Load Case 2
Load Case 3
1500
Load Case 4
S tr e s s (p s i )
Load Case 5
1000
Load Case 6
Load Case 7
500
Load Case 8
Load Case 9
0
Load Case 10
-500
0 10 20 30 40 50
Span (ft)
55
Girder 2
3500
-1000
0 10 20 30 40 50
Span (ft)
56
Girder 3
1600
1400
Load Case 1
1200 Load Case 2
1000 Load Case 3
Load Case 4
S tr e s s (p s i )
800
Load Case 5
600
Load Case 6
400 Load Case 7
200 Load Case 8
0 Load Case 9
Load Case 10
-200
-400
0 10 20 30 40 50
Span (ft)
57
Girder 4
400
Load Case 5
100
Load Case 6
Load Case 7
0
Load Case 8
0 10 20 30 40 50
Load Case 9
-100 Load Case 10
-200
Span (ft)
58
Girder 5
700
-200
0 10 20 30 40 50
Span (ft)
59
Girder 6
150
Load Case 1
100 Load Case 2
Load Case 3
Load Case 4
S tr e s s (p s i )
50
Load Case 5
Load Case 6
0 Load Case 7
Load Case 8
-50 Load Case 9
Load Case 10
-100
0 10 20 30 40 50
Span (ft)
60