INNOVATIVE METHODS OF CONCRETE BRIDGE DESIGN
Design engineers use finite element analysis to optimize
the benefits of concrete materials on large bridge projects.
by: Greg Shafer, P.E., Senior Bridge Engineer
Jerry Pfuntner, P.E., Bridge Engineer
 | Operated by the Expressway and Rapid Transit Authority of Thailand,
the Bang Na-Bang Pli-Bang Pakong Expressway in Bangkok is an
elevated roadway built in the median of an existing highway. Currently,
12 kilometers of the 54 kilometer-long bridge are operational with
completion expected early in the year 2000. |
December 30, 1998, Pittsburgh, Pennsylvania - Civil engineers and bridge developers worldwide
must design bridges and elevated roadways to conform to both natural and urban landscapes while
ensuring sound construction, durability and aesthetic value at the lowest cost. The combined cost of
the design, materials, construction and maintenance is an important factor in many design decisions
made by today's engineers.
Engineers at J. Muller International (JMI) Bridge Engineering Consultants, based in San Diego, CA,
meet these challenges with innovative designs often using a precast segmental concrete construction
for bridges and viaducts throughout the United States, Europe and Asia. When designing bridges of
3,000 meters or more, special care must be taken to minimize the quantity of materials used and to
accurately determine maximum loads and post-tensioning requirements. Therefore, the engineers use
computer-aided drawing and finite element analysis (FEA) tools from Pittsburgh-based ALGOR, Inc.
to ensure their designs are refined before construction begins.
The following discussion chronicles two ways in which engineers from JMI use ALGOR's design and
analysis tools to produce concrete bridge and viaduct designs. The first instance combines the
traditional civil engineering practice of strut-and-tie analysis with FEA to study the local effects of
external post-tensioning in segmental concrete box girders.
The second instance uses FEA to study the load effects of trucks on the box girder superstructure of
the Bang Na-Bang Pli-Bang Pakong Expressway in Bangkok, Thailand. The engineers used FEA to
more accurately predict structural behavior, reducing the structure cost while maintaining the
required strength and serviceability.
The firm also has applied similar analysis techniques to various other bridge designs, including the
Northumberland Strait Crossing on Prince Edward Island, Canada and the H-3 Windward Viaduct in
Hawaii.
Working with Concrete
JMI uses a precast segmental match casting method of bridge and viaduct construction, in which box
girder spans, column supports and other bridge components are built using a casting machine prior to
assembly. Invented in 1962 by Jean Muller, who founded JMI in 1986, this technique allows for
rapid segment production, transportation and assembly at the construction site. Because the bridge
components are cast in a separate casting yard, the construction site can be prepared concurrently
with the segment casting, enabling construction to proceed at a rapid pace.
Engineers at JMI have found that prestressed concrete structures exhibit excellent durability with
virtually no cracking and minimal maintenance requirements. Box girder bridges are aesthetically
pleasing in high profile and urban environments, and they add to rather than detract from the
landscape. They have significant fatigue resistance and overstrength capacity. These aspects
combine to produce a bridge that is both durable and adds significant value to a community.
The use of prestressing allows precast segmental bridges to be designed so that concrete stresses are
maintained well below the tensile cracking stress of concrete. This provides the durability found in
concrete segmental bridges because cracking in the deck slab is virtually eliminated. This is a
significant advantage over reinforced concrete bridge decks, which must crack in order for the
embedded reinforcement to resist the loads caused by vehicles. Cracking causes water and road salts
to permeate the bridge deck, which can result in further cracking, increase maintenance costs and
jeopardize the integrity of the bridge.
 | These post-tensioning tendons induce compression along a box
girder span. The tendons are threaded through deviation saddles
located in the void of the box girder segment, which transfer large
forces created by the deviation to the surrounding areas. |
There are many complex design aspects in precast segmental bridge design; most of them relate to
the application of post-tensioning. JMI engineers use transverse post-tensioning to pre-compress the
top slab against the local forces induced from vehicular traffic. Determining the longitudinal
post-tensioning specifications requires significant effort. The tendon configuration, sizes and stressing
forces are computed using a time-dependent analysis. Special segments are required to anchor and
route the tendons through the span. A concrete diaphragm at the end of each span is designed and
detailed to resist the anchorage forces from the longitudinal post-tensioning tendons. Segmental
bridges with external tendons also require "deviation saddles," which act as harping points for the
longitudinal tendons. These complex segments also require a special analysis.
In addition to these detailed design aspects, one of the greatest challenges of precast segmental
bridge design is determining the most cost-effective erection scheme, box girder proportions and
post-tensioning configuration. Bridge design experience and superior analysis tools and methods can
significantly reduce bridge construction costs.
 | This diagram shows the placement of post-tensioning tendons within
the void of a box girder segment. Post-tensioning tendons are placed
transversely across each individual box girder segment and
longitudinally along the entire box girder cross section. |
Combining Strut-and-Tie Analysis with FEA
A combination of strut-and-tie analysis and FEA software, such as that from ALGOR, Inc., is
sometimes used to analyze discontinuous regions where local effects may not allow the use of a
plane section hypothesis.
As with any structural analysis, the area of interest must be divided into discontinuous and
beam-type regions. For combined strut-and-tie and finite element models, the beam-type regions
exist where plane sections may be assumed to remain planar and are modeled with plate/shell finite
elements. A strut-and-tie model is made of truss elements and assembled directly onto the finite
element plate/shell model. This allows modeling of a complete load path of the discontinuous forces
that spread into the surrounding sections. The combined model can show the internal forces of both
the strut-and-tie model and the surrounding plate/shell elements of the finite element model.
A combined strut-and-tie model and finite element model was used in the analysis of a trapezoidal,
precast segmental box girder section, containing a deviation saddle. Large forces generated by the
deviation of the tendons were transferred through the deviation saddle to the surrounding thin webs
and bottom slab. The forces in these elements had to be determined for the design.
The engineers developed the strut-and-tie model by determining the load path of the discontinuous
forces through the deviation rib. First, the compressive forces were outlined and tensile zones
identified. A reinforcement pattern that provides tension ties through the corresponding tensile zones
was chosen. Compressive struts and tension ties were then determined so that a complete load path
was formed.
The development of the strut-and-tie model can take several iterations to produce a model that
closely follows the load path through the discontinuous region. In the accompanying model, the finite
elements, which are continuous through the discontinuity region, represent the bending stiffness of
the deviation saddle. This particular truss model was constructed to ensure that the strut-and-tie
model did not contribute to the bending stiffness by acting compositely with shell elements.
 | x = Fixed Edge
c = Symmetrical Edge
Engineers at J. Muller International Bridge Engineering
Consultants, based in San Diego, CA, combined traditional
strut-and-tie analysis with FEA software, such as that from ALGOR,
Inc., Pittsburgh, PA, in the analysis of this trapezoidal, precast
segmental box girder deviation segment. This technique enabled
them to analyze both the discontinuous and beam-type regions of the
structure using a truss and plate/shell model.
|
A half-segment model was used because the segments and loads were symmetrical. The edges at the
centerline of the segment were fixed in rotation about the longitudinal axis and transverse
displacement. All other edges were fully fixed. The strut-and-tie model nodes were released in the
vertical and transverse directions and fixed against the longitudinal translation.
The required time for the analysis was reduced because a single model was used to determine the
effects of both the discontinuous and beam-type regions. Analyzing the two regions together also
gave a better sense of the overall behavior in the element. The engineers received force output for
effects in the slabs and webs. For example, the tensile force behind a post-tensioning anchorage
block and the bending effects it causes in the web could now be determined. This analysis method
enhances strut-and-tie model technology so that JMI engineers are not limited by their analysis
capabilities as new designs are developed.
JMI engineers use ALGOR's Superdraw III design program to create many components of their
designs because of the ease in which different finite elements can be combined into one model.
Engineers created a combined beam and plate/shell element model to study the box girder of the
Bang Na-Bang Pli-Bang Pakong Expressway in Bangkok.
Designing the High Road
The Bang Na-Bang Pli-Bang Pakong Expressway is a 54 kilometer-long elevated highway
connecting the First Expressway System with the Chonburi bypass in Bangkok. Joint
Venture-Bilfinger-Berger, Ch. Karnchang, Dwidag is building the expressway in stages under
contract for the Expressway and Rapid Transit Authority of Thailand to alleviate Bangkok's
notoriously congested roadways.
The expressway presented JMI engineers with several challenging design considerations and
limitations:
- The engineers needed an exceptionally wide box girder to support 6 lanes (27 meters) of traffic.
Several box girder designs had to be considered to compare overall use of materials, load-bearing
capabilities, which include significant live loading effects from traffic, and method of assembly.
- The expressway has 36 elevated ramps, including 10 ramps that lead into major interchanges,
and two elevated toll plazas, one at each end. The 80-meter wide toll plazas support up to 12
tollbooths plus toll surveillance buildings. This extreme increase in width at the toll plazas, as
well as the increased width at the ramps, required the engineers to use a support system different
from the one chosen for the lane traffic.
- Because of the location of the expressway in the median of an existing highway, the engineers
were severely limited in the positioning of the main support columns. The columns could only be
placed in a narrow 5-meter strip in the center and side medians of the project. Furthermore,
storm water runoff from the existing highway is channeled into the medians and could not be
blocked by the columns.
- The long length of the structure and visibility from the highway below meant creating a dramatic,
aesthetically pleasing column that minimizes the quantity of materials used. Each column
supports a 27-meter wide box girder with an average span of 42 meters of roadway between
each column.
- Because of the congested existing roadway and limited construction space, JMI engineers
needed to create a design that reduced the amount of work required on-site.
JMI engineers specifically designed box girders, columns and portal frames to meet all of the design
considerations. ALGOR's Superdraw III and linear static stress analysis software was important to
the optimization of the box girder design.
Box Girders
Beginning work on the project in September 1995, the engineers chose a single-cell box girder 27
meters wide with slender flanges for the main portion of the project. Typically, spans of this width
would be supported by multiple box girders; however, the engineers determined that twice the
number of precast segments would be needed as well as a special column cap to transfer the vertical
loads to the columns using a multiple box girder design.
A multi-cell box girder also was considered. This design was simpler, the post-tensioning more easily
arranged and the casting of concrete was more easily performed. However, expensive temporary
measures would be needed to mitigate stresses and deformations in the cross-section during the
span-by-span assembly. Eventually, the engineers determined that the single-cell box girder was the
most economically feasible.
Each precast segmental box girder is comprised of 16 to 18 individual box girder segments, weighing
85 tons each. Each 2.55 meter-long segment has two inclined struts that support the center of the
roadway and is transversely post-tensioned with strand tendons anchored at the edges of the slabs.
The segments are post-tensioned together to form a span using tendons anchored in the end
diaphragms and deviated at discrete points within the span.
 | A typical box girder segment of the Bang Na-Bang Pli-Bang Pakong
Expressway is lifted into place by a swivel crane attached to a large
erection girder. |
Once the type of box girder was determined, JMI engineers created a finite element model of the box
girder cross section. The purpose of the model was to optimize the design and reduce the quantity of
materials used and to ensure the cross section could withstand the loading effects of heavy truck
traffic on the deck of the viaduct.
In other cases, JMI engineers have used plate influence surfaces, which are published in standard
textbooks, to determine the effects of live loading on bridge decks. By creating a finite element
model, the engineers created a surface of influence specifically for the bridge deck of concern. This
eliminated conservative assumptions imposed by using textbook values, which are more generalized,
and enabled the engineers to further optimize the design.
JMI engineers used plate/shell elements to represent the top and bottom slabs of the box girder.
Beam elements represented the struts of the box girder segments while boundary elements
represented the column supports for the box girders. Boundary elements were used instead of nodal
boundary conditions because the supports were not oriented along the global axes. Furthermore, the
columns were not modeled because they were not of concern in this analysis. Additional boundary
conditions were added to fully constrain the box girder cross section.
The engineers then evaluated the truck loading on the bridge deck. They used three types of loading:
the HS25-44 standard set by the AASHTO code, alternating military loading and an overload vehicle
specified by the owner. The second standard is in place to ensure the expressway could withstand
loading from military use.
To determine the critical placement of the loads for acquiring maximum stresses on the deck, the
engineers began by placing the loads in two areas of concern: in areas directly above a strut support
and in areas between two struts. Influence surfaces for the plate/shell elements were developed by
placing unit loads on the bridge deck at many different locations in separate load cases. The effects
of the unit loads on a particular section were combined to evaluate the effect of a truck at the critical
loading location. The most significant local effect was the slab stresses perpendicular to the line of
traffic. The effects of global bending of the box girder were considered in a separate analysis, which
included the longitudinal post-tensioning and time-dependent effects.
The engineers also were concerned about the local deflection of the bridge deck. The AASHTO
requirements limit local deflections of a cantilever slab to the slab length divided by a factor of 300.
This limits the dynamic effects on the deck under live loading. The engineers consider the dynamic
effects of vibration from the trucks using an amplification factor of 30 percent according to the
standard specifications.
Once the model was drawn and boundary conditions and loading applied, the engineers ran the
model through ALGOR's linear static stress processor to determine the maximum stresses and
deflections across the bridge deck. While both the stresses and deflection were within the allowable
limits, the areas between two struts furthest from the supports exhibited the highest stress. Higher
levels of local deflection were noted at the edges of the bridge deck.
Several iterations of this process were required to optimize the final design for both the amount of
materials needed and the loading effects. Nonetheless, using this process, the engineers were able to
verify the design quickly and use approximately 10% less material compared to traditional design
methods. Based on the large scale of the structure and the $1 billion net value of the project, the
gains from the analysis of the finite element model were significant.
 
| JMI engineers used ALGOR's linear static stress analysis software to
determine areas of maximum stress and local deflection resulting
from the movement of large trucks across the expressway bridge
deck. The model at top shows the stresses resulting when a load is
placed near the center of the bridge deck. The model at bottom
shows the deflection resulting from loading at one edge of the deck. |
Columns
Because the expressway is extremely long and has foundations that are limited in width, the
engineers designed the lower portion of the column with two slender "legs" spaced as far apart as
possible to resist the transverse bending moment and to allow the flow of storm water. These legs are
joined by a beam at the mid-height of the column with the "arms" of the column extending upward at
a 35-degree angle. This open y-shape provides a wider base for stability of the box girder on its
bearings and provides an open area for the support of the erection girder, used to assemble the
precast segmental box girder sections. It also gives the impression that the girder box is virtually
floating for aesthetic appeal.
JMI engineers explored the possibility of using the slender, inclined arms for the upper portion of the
column without a tie across the top of the column; however, they determined that the gravity loads
on the arms far exceeded the capacity of the section. Therefore, inclined elastomeric bearings are
positioned on the top of the column normal to the axis of the arm so vertical concentric loads are
resisted by axial forces along the column arms without bending.
Non-concentric vertical loads are broken down into a concentric load and transverse moment. The
transverse moment is resisted by equal and opposite axial forces in the arms plus a lateral shear
force, which causes shear distortion of the elastomeric bearings. The shear distortion is a critical
design parameter and was limited to ensure sufficient fatigue resistance.
Buffer blocks between the box girder and column arms restrain lateral movement. The buffer blocks
engage after the structure has moved laterally by 25 mm. The distribution of loads to the column
arms is different before and after the buffer blocks engage; thus engineers conducted a detailed
analysis to capture the true effects of this system under concentric loading and transverse moments.
Segmental Portal Frames
In locations where the expressway width is increased to accommodate ramps and toll plazas, a single
column in the center of the roadway is no longer sufficient to support the superstructure. In these
areas, multiple box girders are supported on portal frames. These frames span the highway below and
are typically comprised of two bays. Because over 100 of these portal frames will be used in the
project, the precast segmental method was chosen to speed construction and limit the amount of
work that would be performed over the existing roadway below.
 | These large portal frames are designed to support multiple box
girders at locations where the expressway width is increased to
accommodate 36 elevated ramps and two elevated toll plazas. |
Before the first span was assembled, a full-scale load test was performed on a box girder in the
casting yard. The theoretical maximum ultimate load was placed on the girder. The span exhibited
behavior that was consistent with the designers' expectations. Work began on the construction of the
expressway in March 1996. The Expressway and Rapid Transit Authority is opening the expressway
in stages as it is completed. Currently, 12 kilometers of the 54 kilometer-long bridge are operational
with completion expected early in the year 2000.
Benefiting from FEA Software
Several major benefits are derived from using FEA software from ALGOR, Inc. in concrete bridge
design. Finite element models enable engineers to use multiple element types to get more refined
results and, therefore, avoid over-engineering a design. Older, more traditional methods of analysis
involve conservative assumptions and can unnecessarily require additional materials. ALGOR's
viewing capabilities enabled engineers to see on-screen whether the design was sensibly reinforced
and determine the feasibility of the design while reducing the number of calculations performed by
the engineer.
With FEA, engineers can model the behavior of complex segments and geometry to determine if the
overall design is a practical solution for the project at hand. In addition, faster computing times and
improved computer hardware enable engineers to model more complicated designs with challenging
design parameters. This means civil engineers and bridge designers are not limited to existing designs.
|