Drexel University Senior Design: Group 17
March 16, 2012Dr. Franklin Moon3141 Chestnut StreetAlumni
Engineering Labs Room 280-GPhiladelphia, PA 19104
Dr. Franklin Moon:We are submitting a progress report that you
requested for the Drexel Senior Design course. The report contains
the work completed thus far in the winter term for the Burlington
Bridge located in Burlington County, New Jersey. Please feel free
to contact us with any questions.Sincerely,James DrogalisBrandon
GlencrossNeil PatelBrandon WeaverCharlie Young
Progress Report of:Feasibility Study to Investigate the Need for
a Public Transportation Connection across the Delaware River within
Burlington County, New Jersey
Design Team:
James [email protected]
Brandon [email protected]
Neil [email protected]
Brandon [email protected]
Charlie [email protected]
Advisor:
Dr. Franklin [email protected]
Technical Reviewer:
Dr. Joseph [email protected]
Owner:Burlington County Bridge Commission
Table of ContentsExecutive Summary1Project Overview2Advantages
of Modeling2Modeling of Structural Systems3Construction Plans and
Documents5Construction of 3-D CAD Model6Modeling of Floor
System7Modeling of Secondary Elements9Member Cross Sections10Error
Screening Process13Future Work15Bibliography17Appendix A: Project
Schedule18Appendix B: Tower Span Model19Appendix C: Floor System
Photographs21
Table of Figures
Figure 1: Macro-Level Model of a Structure.3Figure 2: Element
Level Modeling of a Structure.4Figure 3: Micro-level Finite Element
Model.5Figure 4: Burlington Bristol Bridge Elevation
View....5Figure 5: CAD Wireframe Model of Lift Span.6Figure 6: Plan
View of Lift Span Roadway Steel Grate.7Figure 7: Lift Span Steel
Grate Roadway Deck.7Figure 8: Typical Cross Section of Lift Span
Deck. (Keystone Structure Steel Co., 1993)8Figure 9: Cross Section
View of Lift Span Steel Grate for Roadway Deck.8Figure 10: Cross
Section View of Lift Span Steel Grate for Sidewalk.8Figure 11:
Tower and Lift Span Floor Systems.9 Figure 12: Lift Span Floor
System.9Figure 13: Modeling of Cross-sectional Properties.10Figure
14: Modeling Using SAPs Section Builder.10Figure 15: Typical Single
Lacing in Truss Member.11Figure 16: Approximate Analysis of a
Simply Supported Beam.13Figure 17: Moment Diagram of Simply
Supported Beam.13Figure 18: Moment Arm Calculation for Approximate
Analysis.14Figure 19: Maximum Deflection of Simply Supported
Beam.14Figure 20: Chord Member Cross Sectional Area
Diagram.15Figure 21: 2D CAD Model of Tower Span.18Figure 22: 3D CAD
Model of Tower Span with Lateral Members.18Figure 23: 3D CAD Model
of Tower Span with Floor System.19Figure 24: SAP Model of Tower
Span.19Figure 25: Lift Span during Bridge Opening20 Figure 26:
Underside of Lift Span.20Figure 27: Lift Span Guiderail.20Figure
28: Lift Span in Open Position.20
Executive Summary
Burlington County, New Jerseys public transportation system is
currently inaccessible and inefficient for many of the countys
residents. As a result, public transportation ridership within the
county is low in comparison to surrounding counties. Commuters
wishing to cross the Delaware River into and out of Burlington
County by rail currently have two options. These options are to
cross the Delaware River at the Morrisville Trenton Bridge to the
north or the Ben Franklin Bridge to the south. Both bridges are
approximately 20 miles in either direction of Burlington County. In
order to eliminate the inconvenience to the countys residents, a
feasibility study is being conducted to investigate the need and
practicality of a rail crossing within Burlington County. During a
previously conducted investigation, the Burlington Bristol Bridge
was selected as a potential candidate for a light rail crossing.The
project scope over the past ten weeks intended to develop a tool to
evaluate the structural capacity of the Burlington Bristol Bridge.
A significant investment in time was dedicated to the creation of
an element level finite element model. The finite element model was
created in SAP2000 and is believed to have great value for
assessing the current behavior of the bridge as well under the
design train loading. The model, if constructed properly, has the
ability to predict local member actions much more realistically
than traditional methods. Sites visits, photographs, construction
drawings and inspection reports were gathered before the model was
begun. AutoCAD was then utilized to draw a single-line model of
three spans of the bridge. These spans included the two tower spans
and well as the vertical lift span as shown in Figure 4. The model
was imported into SAP, where member cross sections and material
properties were assigned. Further information about member geometry
may be found in Member Cross Sections.Simultaneously as the model
was being created, it was constantly being error screened. Refer to
Error Screening Process for the methodology behind the error
screening as well as some approximate techniques that were
leveraged.The SAP model is currently at 90% completion. The
remaining tasks for the model include accounting for additional
dead load of the structure and well as a mesh sensitivity study.
Final design including any necessary retrofits will begin after the
model has been validated. A detailed project schedule may be found
in Appendix A: Project Schedule.
Project OverviewResidents of Burlington County, New Jersey
currently lack an efficient public transportation system. As a
result, a significantly lower number of residents utilize public
transportation than in surrounding counties. A major reason why
people are not using public transportation is accessibility and
convenience. With about 15% of commuters traveling into
Pennsylvania daily, only a small number choose to use public
transportation. In fact, only 3% of Burlington County commuters
chose to ride public transportation as a mode to work. (DMJM
Harris/AECOM, 2006)Commuters from Burlington County who wish to
cross the Delaware River into Pennsylvania only have a choice of
crossing over the Morrisville-Trenton Bridge or the Ben Franklin
Bridge to the south. Travelers have a commute of about 20 miles in
either direction to each river crossing.A proposal was prepared in
the preliminary stages of this project to determine the feasibility
of a light rail transportation connection over the Delaware River
in Burlington County. The Burlington Bristol and Tacony Palmyra
Bridge were considered as possible alternatives for a light rail
crossing. It was determined after the study that the Burlington
County Bridge was the most suitable option. The current proposed
alignment is over what is now the roadway of the bridge. Temporary
traffic stoppages would be conducted as necessary while still
maintaining it as a vehicular bridge. Further details of the
objectives and findings of the study are contained in the project
proposal.Advantages of ModelingIn order to justify the importance
of the work explained in detail throughout this report, it is
important to explain the significance and meaning of what can be
accomplished through creating an analytical computer-based model of
a structure. With a structure as complex as the Burlington Bristol
Bridge, an inherent level of uncertainty in analyzing any
structural system only increases with the complexity of the system.
Every structure has its own unique level of uncertainty, which is
influenced by parameters including the amount of documentation
available on its construction, redundancy of its design, degrees of
indeterminacy, ease of visual inspection, access restrictions, etc.
Depending on the application, there can also be a level of
uncertainty associated with the demand required of a structure.
These forms of inherent uncertainty can oftentimes necessitate an
iterative process to achieve reliable and thereby meaningful
results. With a design such as the one proposed, it is of the
upmost importance to analyze the structure in a way that will yield
the most accurate results. This project involves both analyzing the
structural responses of the bridge due to its current demand and
ensuring structural adequacy that will result from increasing the
demand. For this reason, the required level of accuracy of the
analysis leaves little room for error. By leveraging the power of
computer based modeling, it is possible to change certain
parameters of the model with relative ease, thereby minimizing
uncertainty. It is also makes it possible to minimize inaccuracies
due to human error through various error screening techniques;
therefore, it is the concept of reproducibility that makes the
model so valuable. Although a model typically requires a
significant effort upfront, it allows the user to run an infinite
number of scenarios. These results are not only quicker than hand
calculations, but ultimately provide a more accurate representation
of actions on the local member level. Modeling of Structural
SystemsBecause the Burlington Bristol Bridge was originally
designed exclusively for vehicular traffic, a detailed analysis
must be conducted to determine the current capacity of the
structure. The overall cost of incorporating light rail
transportation onto the bridge hinges largely on the extent of
retrofits that need to be designed. For this part of the analysis,
only three spans of the bridge were considered. The vertical lift
span and two tower spans were considered to be the most costly and
difficult areas of the bridge to retrofit. These spans are shown in
Figure 4. If the bridge proves to be sufficient in these three
spans, appropriate modifications can be made to the remaining
approach spans.In order to assess the feasibility of incorporating
light rail, it is imperative that a model be created to monitor
these changes. One of the most useful types of models for analyzing
the response of the structure is a finite element model. Finite
element models are especially useful for existing structures for
several reasons. If the structure is modeled correctly, it can be
an extremely useful tool for performing load ratings, detecting
signs of distress and quantifying the effects of changing loads.
Different levels of finite element models can be created depending
upon level of sophistication required, cost and computing power
available. Modeling can be classified into three categories:
Macro-level models This is the most general way of representing a
structure. This is the least sophisticated type of model and
simplifies the structure by incorporating rotational and
longitudinal springs. Complex systems are reduced to just a few
members that can simulate the behavior of the real system.For
example, the deck truss shown in Figure 1 may be represented as
just a few members, but given properties to simulate the behavior
of the entire truss.
Deck Truss Figure 1: Macro-Level Model of a Structure. Element
level Element level models are more sophisticated than macro- level
models. Beams are represented by frame elements which are drawn
along the neutral axis of the members. Because the frame elements
are represented by single lines, they must be connected to other
members through rigid links. The rigid links account for the
differences in geometry and are infinitely stiff. The links help to
enforce continuity. A diagram showing this type of modeling is
shown in Figure 2. The structure may be represented by smaller
areas of shells and more frame elements. This is often referred to
as the discretization. A model that is discretized more will take
longer for a computer to analyze, but may give more accurate
results up to a certain point where results will plateau.
Rigid LinkShellFrame ElementFigure 2: Element Level Modeling of
a Structure. Micro-level models Micro-level models are the most
sophisticated type of finite element models; therefore, they may
require the most computing power when modeling large systems. This
type of model actually represents each piece of material in the
structure. The discretized elements are referred to as bricks or
meshing. An example of a micro-level model is shown in Figure
3.
Figure 3: Micro-level Finite Element Model.Due to the nature of
this project, it was decided that an element-level finite element
model was most appropriate. The entire structure could be
represented using shell, link and frame elements. An element level
model can accurately model the behavior of such a large structure
and give useful results in a reasonable amount of time.Construction
Plans and DocumentsPlans were obtained from the original
construction of the bridge as well as several retrofits. To
construct the model, the following plan sets were utilized: 1930
Original contract and shop drawings 1976 - Deck Replacement of
Spans 4 & 6 1988 Replacement of Sidewalk on Spans 4 & 6
1993 Lift Span Deck Replacement (Span 5)
Span 4Span 5Span 6 NFigure 4: Burlington Bristol Bridge
Elevation View. (Howard-Needles-Tammen & Bergendoff, 1976)Shown
above in Figure 4 is an elevation view of the bridge. Note that
only Spans 4, 5 and 6 were constructed in the finite element model.
These spans were considered the most critical parts of the
analysis. If the analysis shows that the light rail loading is
feasible for these spans, then the remaining approach spans can be
retrofit accordingly. A 2010 in-depth fractural critical member
inspection report was also used as a reference for this project.
The inspection was conducted in December 2010 by Pennoni Associates
Inc. Construction of 3-D CAD ModelThe first step in creating a
finite element model was to convert the structures 2-D paper
drawings into a 3-D single line structure. AutoCAD was used to
create this 3-D drawing. This drawing is often referred to as a
wireframe model because it does not incorporate any of the
structures cross sectional properties. These will be assigned later
in the finite element analysis program. Members with the same cross
sectional properties were grouped on the same layer. This made
assigning cross-sectional properties easier when they were imported
into the finite element software. Shown below in Figure 5 is the
wireframe model created for the lift span. First, the frame
elements were drawn in 2-D, adding rigid links where necessary.
Taking advantage of symmetry, the model was drawn in 3-D and
connected with lateral bracing members. The last step was to draw
the floor system where shells, link and frame elements were all
used. A similar approach was used for the tower span of the bridge
also taking advantage of symmetry. Figures of the CAD model for the
tower span may be found in Appendix B: Tower Span Model. Figure 5:
CAD Wireframe Model of Lift Span.Modeling of Floor SystemThe floor
system on the lift span portion of the bridge (Span 5) consists of
a steel grate deck which was last replaced in 1993. The steel grate
is ASTM A709 Grade 36 steel as indicated on the as-built drawings.
The steel grate on the roadway has a thickness of 5 3/16 built on
top of spacer tee beams running longitudinally to the roadway. A
typical plan view is shown in Figure 6 and a cross section view is
shown in Figure 8. Figure 8 is a typical view at the center of the
lift span where a 60 maintenance sidewalk exists. The maintenance
sidewalk also has a steel grate deck with a 2 thickness as shown in
the cross section in Figure 10. Additional site photographs of the
floor system can be found in Appendix C: Floor System. Figure 6:
Plan View of Lift Span Roadway Steel Grate.
Figure 7: Lift Span Steel Grate Roadway Deck.
Figure 8: Typical Cross Section of Lift Span Deck. (Keystone
Structure Steel Co., 1993)
Figure 9: Cross Section View of Lift Span Steel Grate for
Roadway Deck.
Typical Floor Beam Cover Plate2 GratingFigure 10: Cross Section
View of Lift Span Steel Grate for Sidewalk.The steel grate is
represented in the model as shell elements. These shells were
assigned the appropriate thickness and material properties.
Continuity was enforced through rigid links and frame elements.
Shell Elements Figure 11: Tower and Lift Span Floor Systems.
Figure 12: Lift Span Floor System.
Modeling of Secondary ElementsAttention was given to secondary
elements of the bridge including the concrete sidewalk, barriers
and pavement. Omitting some of these elements can lead to modeling
errors and caution must be exercised. For this model, it was
determined that the concrete sidewalk would be modeled with shell
elements. By choosing to model the sidewalk on the tower and lift
spans, the model became more realistic of the true behavior. The
contribution of the mass and stiffness of the sidewalk is a
critical part of structures resistance. Because of the extremely
low stiffness of the steel guiderail and pavement, these secondary
elements were chosen not to be modeled as shells. Instead, the
guiderail and pavement were simply modeled as distributed and line
loads. This means that their mass is accounted for, but no
stiffness is provided by them. (Goulet, Kripakaran, & Smith,
2009) Refer to Figure 27 for a photo of the lift span
guiderail.
Member Cross Sections
Each beam was imported from the 3-D CAD model as a frame
element. Cross sectional properties needed to be assigned to each
member. To do this, the member cross section needed to be created
using SAPs section builder. Cross sectional properties were
determined from original construction and rehabilitation drawings.
The drawings provided a standard section callout and a visual
representation of the sections configuration. The sections in the
drawing were compared to the real member to ensure that they were
being properly represented in the model. An example of this is
shown in Figure 13. Figure 13: Modeling of Cross-sectional
Properties.
Figure 14: Modeling Using SAPs Section Builder.
Most of the sections of the Burlington Bristol Bridge also
consisted of lacing. The lacing for each member pairs sections in
parallel and makes them continuous. An example of a truss member
with lacing is shown in Figure 13. During excessive compression in
the members, the main sections tend to move away from the central
axis. The lacing connects resists this action. Despite assisting
the full sections compatibility, the lacing will carry minimum
force compared to the overall section. Therefore, the sections in
the model were designed to ignore lacing. The cross sections were
modeled in the section designer, as shown in Figure 14, to ignore
the forces carried by the lacing. Although the lacing is being
ignored in the cross section designer, full compatibility was
assumed over the entire section. Unaccounted Dead LoadEstimating
the true dead weight of a structure can be a very difficult task.
The Burlington Bristol Bridge is a unique case because the
structure is a vertical lift span bridge. The mass of the
counterweights is known to be approximately equal to the mass of
the lift span. Based upon field tests and plan sets, the totally
dead load of the lift span is known to be 2,536 kips. (Moon, Aktan,
& Lowdermilk, 2010) With this extremely valuable information,
the dead weight of the structure in the model was constantly
compared to the known information. The model currently has a dead
weight of 1,400 kips. Dead weight was only being accounted for in
the beam and deck areas. As discussed in the previous section, the
lacing was ignored in the cross section builder. The unaccounted
weight of the lacing can lead to significant underestimation in the
dead load. The correction for the error is discussed further.Other
significant sources of dead weight that are not currently modeled
are the following:
Figure 15: Typical Single Lacing in Truss Member. Control Tower
Crows Nests Paint Gusset Plates Rivets Machinery Cables Utility
Conduits
Assigning SectionsMembers of the same property were created on
similar layers in AutoCAD. When single layers were imported into
SAP, the layers were assigned to a group. Groups were created so
that all members contained in it have the same cross sectional
properties. This kept the model organized and will make it easier
to modify properties. Use of Mass Modifiers to Correct for
Unaccounted Dead LoadAs discussed in the previous section, the
presence of lacing was ignored in the creation of cross sectional
geometry. The additional contribution to the dead load must be
accounted for. Through the use of a mass multiplier, the members
mass can be increased without increasing the cross sectional area
of the member. This will allow the model to account for the lacing
mass, but not contribute any additional capacity to the member. A
simple calculation was performed to determine a reasonable mass
multiplier to apply to frame elements. An example calculation is
shown as follows:
Further work will be conducted in the future to determine other
sources of unaccounted dead load. Mass corrections will be modified
as necessary.
Error Screening ProcessThe most vital part of the development of
the finite element model involves verifying the results to ensure
that the structure is being accurately represented. This involves
checking the numerical results, interpreting deflected shapes for
compatibility as well as using structural engineering intuition to
ensure that the results make sense. Oftentimes, the results can be
compared to previously computed values if available. (Fisher,
1983)Approximation of Axial Force in the Lift Span After the lift
span portion and floor system of the bridge were imported into SAP,
approximations were made about the axial forces in the top and
bottom chord members. A simplifying assumption was made to reduce
the lift span into a simply supported beam as shown in Figure 16.
The bracing in the face of the truss was assumed to have infinite
stiffness thus reducing the structure into a 2D face.
w = 2.39 kips/ft
R = 636 kipsR = 636 kips = 533 ft
Figure 16: Approximate Analysis of a Simply Supported Beam.Using
the model, the support reactions were applied from the SAP model,
which are represented by R. To put the beam into equilibrium, a
uniformly distributed load, w, was spread over the entire structure
to put the beam into equilibrium. The moment diagram from this type
of loading is shown below in Figure 17. A constant height of 65 was
assumed over the entire length of the beam. This is the height from
top to bottom chord on the real structure. This 65 acts as a moment
arm between the two chords as shown in Figure 18. By summing the
forces about a point, the total axial compression or tension in the
chords could be solved.
Figure 17: Moment Diagram of Simply Supported Beam.
Figure 18: Moment Arm Calculation for Approximate Analysis.
Table 1: Comparison of Axial Force in Chord Members for the
Simply Supported Beam Approach.Hand CalculationSAP with Floor
SystemSAP without Floor System
Top Chord (kips)652 (c)693 (c)530 (c)
Bottom Chord (kips)652 (t)176 (t)457 (t)
A comparison of the approximate analysis values and SAP results
can be found in Table 1. These results are for axial force in the
top and bottom chords. The approximate technique assumed equal
compression and tension in the top and bottom chords. The results
were then compared to the SAP model including the entire floor
system. The results showed a major discrepancy in the bottom chord
member. This is due to the floor system acting to stiffen the
entire lower portion of the structure. The top chord member still
remained consistent with the approximate analysis. To further
validate this trend, the floor system was removed from the bridge
in the SAP model. The results were shown to agree within 30%. The
reduced values of the moment couple was due to the loss of dead
load from the floor system. At this stage in the model development,
this was considered to be acceptable given the amount of
simplifying assumptions that were being made.
Approximation of Deflection in the Lift SpanAnother form of
approximate analysis was performed for the lift span portion of the
bridge by determining if the deflection in the model was
reasonable. A simplification was made to reduce the structure to a
simply supported beam as shown in Figure 19. Figure 20 also shows a
cross section view of lift span of the bridge where the moment of
inertia was calculated for each of the chord members. The results
of this analysis are shown in Table 2.The hand approximate analysis
was then compared to the SAP model and AASHTO allowable deflection.
The results agreed closely, but further investigation must be
completed to verify these results.
Figure 19: Maximum Deflection of Simply Supported Beam.
Figure 20: Chord Member Cross Sectional Area Diagram.
Table 2: Approximate Analysis of Lift Span Deflection.Hand
CalculationSAP ModelAASHTO Allowable Deflection (l/800)
2.82.98
Knowing there is 44% of the actual load missing in the model,
the deflections are expected to increase. Since deflection is
directly related to the distributed load, it is expected that the
deflection will increase by a factor of 1.81. This will still allow
the deflection to satisfy the AASHTO limit.Future WorkThe future
work of the project can be simplified into two categories. These
are completion of the model and the final design phase of the
project. Further information about project scheduling may be found
in Appendix A: Project Schedule. Model DevelopmentCurrently, the
SAP model is at 90% completion. Two vital tasks must be completed
to further validate the model. The first is to perform an in-depth
analysis of unaccounted dead weight of the structure. The current
dead weight of the lift span is 1400 kips, while the true dead
weight of the span is known to be 2,536 kips. (Moon, Aktan, &
Lowdermilk, 2010) Mass multipliers will be determined with more
confidence and assigned as necessary to the appropriate frame
sections. A mesh sensitivity study must be completed particularly
for the shell elements of the deck. This entails reaching a careful
balance between discretization and computing time. The mesh will
reach an acceptable level of discretization at the point when
further increase in shell elements yields no further change in
results.Final Design PhaseAfter the performance of the model is
validated, the final design stage can begin. During this stage, a
load rating as per AASHTO will be performed. This will ensure that
the bridge currently rates as it stands today. A load rating will
then be performed per the American Railway Engineering and
Maintenance-of-Way Association (AREMA) loading configuration. At
this point, the behavior of the bridge under the train loading will
be known and retrofits can be performed as necessary. The final
alignment of the tracks will be selected and construction costs
will be determined.
BibliographyAsh-Howard-Needles & Tammen. (1930). Burlington
Bristol Bridge: Original Construction Drawings. New York, New
York.DMJM Harris/AECOM. (2006, March). New Jersey Long-Range
Transportation Plan 2030. Philadelphia, Pennsylvania. Retrieved
Novemeber 2011, 25Fisher, T. A. (1983). Long-Span Bridge Computer
Modeling. Journal Of Structural Engineering, 1402.Goulet, J. A.,
Kripakaran, P., & Smith, I. F. (2009). Estimation of Modelling
Errors in Structural System Identification. Lausanna, Switzerland:
cole Polytechnique Fdrale de Lausanne.Howard-Needles-Tammen &
Bergendoff. (1976). Burlington Bristol Bridge: Reconstruction of
the Bridge Deck. Fairfield, New Jersey.Industrial Engineering
Works. (1988). Deck Replacement at Beam and Truss Spans. Trenton,
New Jersey.Keystone Structure Steel Co. (1993). Burlington Bristol
Bridge: Lift Span Deck Replacement. Newtown, PA.Moon, F. L., Aktan,
A. E., & Lowdermilk, D. S. (2010). Model Experiment Correlation
I. Philadelphia, PA: Intelligent Infrastructure Systems.Pennoni
Associates. (2010). Burlington Bristol Bridge: In-Depth and
Fracture Critical Member Bridge Inspection Report. Haddon Heights,
New Jersey.
17
Appendix A: Project Schedule
Phase 3Phase 2Phase 1Appendix B: Tower Span Model
Figure 21: 2D CAD Model of Tower Span.
Figure 22: 3D CAD Model of Tower Span with Lateral Members.
Figure 23: 3D CAD Model of Tower Span with Floor System.
Figure 24: SAP Model of Tower Span.
Appendix C: Floor System Photographs
Figure 25: Lift Span during Bridge Opening. Figure 26: Underside
of Lift Span.
Figure 27: Lift Span Guiderail.
Figure 28: Lift Span in Open Position.
