Sunteți pe pagina 1din 8

Transactions on the Built Environment vol 13, 1995 WIT Press, www.witpress.

com, ISSN 1743-3509

Optimum design of welded plate girders


subjected to highway bridge loading
M. Asghar Bhatti,* A.S. Al-Gahtani^
"Department of Civil and Environmental Engineering, University of
7owa, /owa CzYy, A4 J2242, L%4
^Department of Civil Engineering, University of Petroleum and
Minerals, Dhahran, Saudi Arabia

Abstract
Optimum design of plate girders subjected to highway bridge loading is
presented in this paper. The formulation is capable of handling composite or
noncomposite designs, shored or unshored construction, stiffened or
unstiffened design, symmetric or unsymmetric cross-section, simple or
continuous spans, and prismatic or nonprismatic girders. The bridge loading is
applied through influence line functions at selected points along the girder. The
design problem is constrained to satisfy the AASHTO specifications.
1 Introduction
The use of welded plate girders for short to medium span highway bridges
offers greatest flexibility in choosing the most economical section. They offer
the following advantages over the standard rolled sections [1].
Welded plate girders can be designed as long and as deep as required by the
design and fabrication method.
Rolled beams are of constant cross section along their length but in plate
girders dimensions can be easily varied if desired.
For composite girders, an unsymmetrical section offers considerable savings
in weight since typically the top flanges need only be large enough to
accommodate the shear connectors.
With advancement of welding technology the welded girders can be
fabricated on fully automatic welding lines with reasonable cost and uniform
quality control. Thus the delivery time for welded girders may be more
predictable than that for rolled shapes.
Use of optimization techniques is necessary to take advantage of this flexibility
in achieving the most economical design. Razani and Goble [2] studied the
optimum design of constant-depth plate girders based on 1961 AASHTO
specifications. The work was extended by Goble and DeSantis [3] to include
the composite sections. Holt and Heithecker [4] presented formulas for
optimum proportions for laterally supported symmetrical plate girders. The
problem was formulated on the basis of the 1967 American Institute of Steel

Transactions on the Built Environment vol 13, 1995 WIT Press, www.witpress.com, ISSN 1743-3509

226

Structural Optimization

Construction (AISC) specifications. Schilling [5] presented some numerical


ratios for optimum plate girder dimensions based on 1969 AISC specifications.
Vachajitpan and Rockey [6] developed a nonlinear programming method to
obtain minimum weight design of unstiffened constant depth symmetric plate
girders. Burns and Ramamurthy [7] presented a formulation to optimize a
uniform plate girder design for composite or noncomposite situation. This
study shows the efficiency of using a general formulation which can be used
for more than one design. Fleisher [8] presented a mathematical formulation
and nomographic design for symmetric plate girders based on the web buckling
theory and on the assumption of flange configuration that do not necessitate
bending stress reduction in the compressive flange stress under the AISC
specifications.
Except for few early references, none of the other papers have dealt with
optimum design under moving highway loads. In this paper a method for
optimum weight/cost design of welded plate girders under AASHTO
specifications [9] is presented. The formulation is very general and includes a
variety of design options such as:
single or continuous span highway bridge girders
stiffened or unstiffened girders
shored or unshored construction
fixed or variable depth along the span
symmetrical or unsymmetrical design
variable flange thicknesses along span
Section 2 briefly reviews analysis and design of composite highway bridge
girders. Optimal design formulation is presented in section 3. Section 4
presents few numerical results.
2 Analysis and Design of Highway Bridge Girders
From analysis point of view the noncomposite girders and shored composite
girders are special cases of the unshored composite case. The unshored
composite case involves consideration of three different cross sectional
properties under three different loading conditions. Only one of these section
properties is needed in the case of noncomposite design and two of them are
used in the shored composite design.
As shown in Figure 1, a composite welded plate girder consists of four parts;
(a) reinforced concrete slab, (b) steel section, (c) shear connectors, and (d) web
stiffeners. The purpose of the slab deck is to distribute wheel loads to the
girders. The shear connectors are designed to transfer longitudinal shear
between the slab and the steel section. They are needed in the positive moment
regions only. If the calculated web shear stress exceeds the allowable shear
stress of the web, stiffeners may be used to increase the web shear capacity
without increasing the web thickness.
Analysis of a composite girder
A composite beam finite element is used for analysis of the girder. The element
consists of unsymmetrical steel section with a concrete slab at the top acting as
a cover plate.. The transformed cross-sectional area is calculated by dividing
the effective slab area by the modular ratio m = ES/EC, where ES is the modulus

Transactions on the Built Environment vol 13, 1995 WIT Press, www.witpress.com, ISSN 1743-3509

Structural Optimization

227

of elasticity of the steel beam and EC is modulus of elasticity of the concrete


slab. Based on concrete strength the modular ratio varies between 6 and 11 [9].
The modular ratio is multiplied by a factor KC to approximately account for
creep effects. This factor is equal to 3 for sustained dead loads and is equal to 1
for other loads. Corresponding to three different loading conditions, following
three different sections properties (moment of inertia and plastic section
modulii) must be computed for each element:
(a) Steel beam section properties for noncomposite dead loads
(b) Composite section properties with creep effects (K^ = 3) for dead loads
(c) Composite section properties with K^ = 1 for live loads
Each span can be divided into a number of finite elements. For a variable depth
girder enough elements must be defined to get a reasonable approximation of
the depth using constant depth elements. Since the loads are applied at nodal
points, even for a prismatic girder, enough elements must be used to get a
fairly smooth influence line. Based on numerical experiments it is suggested to
use 10 elements per span.
top slab

-sheer studs

top flange
plate

transverse
stiff ener

web plate-

bottom flange^
plate
I - section
Figure 1. Typical Highway Bridge Cross Section
Design forces and stresses
A composite girder must be analyzed under several different loadings
depending upon the construction method. An influence line analysis is required
for truck loads. Figures 2 shows typical bending moment diagrams. Bending
moments due to four different loadings are shown identified as MS
(noncomposite dead loads), Mrj (composite dead loads), M^ (positive live
loads), and M~ (negative live loads). The total design forces at any point are

Transactions on the Built Environment vol 13, 1995 WIT Press, www.witpress.com, ISSN 1743-3509

228

Structural Optimization

the algebraic sum of the forces from different loadings. For additional details
see reference [10].

HS20-44

(k-ft.)

20

260
140
180
220
Bridge length,ft.
Bending moments from different loads

60

TOO

300

( k-ft.)

100
140
180
Bridge length,ft.
b. Design bending moments

220

260

3CT0

Figure 2. Typical Design Bending Moments from Different Bridge Loads


3 Optimization Formulation
The design of welded plate girders requires determination of six geometrical
dimensions; depth and thickness of web, width and thickness of top flange, and
width and thickness of bottom flange. Changing the flange thickness along the
bridge is more practical and less costly than changing the flange width.

Transactions on the Built Environment vol 13, 1995 WIT Press, www.witpress.com, ISSN 1743-3509

Structural Optimization

229

therefore, the flange widths are kept constant along the girder and are not
considered as design variables. Thus for each beam element or a group of
elements the design variables are defined as:
1)
Top flange area (At) 2)
bottom flange area (Ay)
3)
web depth (d\J
4)
web thickness (tw)
5)
stiffener spacing (a) in case of stiffened design
The volume of the steel beam and the stiffeners is the total weight to be
minimized. In most practical bridge designs, the cost is related to the weight of
the structure. Cost optimization requires estimation of construction cost that
depends on many variables such as, the contractor, construction equipment,
method of erection, fabrication, transportation, labor, etc. These variables
change continuously with time and differ from one location to another. The
volume (or weight) as an objective function is more universal and requires no
input or change in input parameters, which allows the designer to try different
designs in short time with less difficulty especially at the preliminary design
stage. The minimized volume/weight can then be transformed to a cost design
if the construction cost is determined.
The objective function is minimized subject to the bending stress, shear stress,
slenderness ratios, deflection, and depth limitations. It is necessary to set
reasonable and practical limits for the design variables to avoid calculations for
impractical values. A 5/16 inch plate thickness is the minimum allowed by the
AASHTO specifications. The constraints are applied to welded plate beams
with welding metal having an equal or greater ultimate strength than the base
metal. Flange plates must be joined end to end by full penetration butt welds.
The weld metals must conform to AASHTO standard specifications for
welding of structural steel for highway bridges.
4 Numerical Results
The interior girder of a 100 feet long highway bridge subjected to HS20-44
loading is designed with uniform section over the span. The design information
is shown in Figure 3. Different design options under the formulated AASHTO
constraints are used. The optimum sections are summarized in Table 1. It can
be seen that the unshored composite design reduced the beam cross section
area by 20% compared to the noncomposite design. The shored composite
design reduced the beam cross section even more by almost 30% of the
noncomposite design. More savings could be obtained with shored composite
design if narrower top flange is chosen because the flange width to thickness
constraint governed the design in this case. The savings will be more with the
composite design if the noncomposite design is obtained with the compression
flange assumed to be partially supported since the allowable compression
stress of the flange will be reduced and a larger flange area would be required
for moment resistance.
The beam depth also shows wide variation. The composite design as expected
offers shallower depth. Shored composite design reduces the depth by more
than 25% of the noncomposite design. The unstiffened design in this problem
has deeper beam than the stiffened design. This is because the unstiffened
design is optimized under both bending and shear stresses while the stiffened
design has abundant shear strength and the beam depth is mainly decided by
the moment action and is limited by depth to thickness ratio. The unstiffened
design requires larger web area for shear resistance and thus thicker and deeper

Transactions on the Built Environment vol 13, 1995 WIT Press, www.witpress.com, ISSN 1743-3509

230

Structural Optimization

web plates. The stiffened design has smaller web area and larger flange areas
than the unstiffened design.
22ft.

I
8ft.

-*-

8ft.

. bt j
4
H

Span length,100. ft.


Number of lanes, 2
Beam spacing, 8. ft.
Steel type,A36
Slab thickness, 7.5 in.
Concrete corepressive srength, 4 ksi.
Modular ratio, 8
Design live load, HS20-44
Super imposed dead load , 45 Ib/sq. ft.
Top flange width, 1 2 i nches
Bottom flange width, 20 inches
Allowable live load deflection, 1.5 inch
Allowable noncomposite dead load deflection, 2. in.
Total allowable deflection, 5. in
Unsupported flange length, 0.0 in.

Figure 3. Data for Numerical Example


Table 1
Optimum Design Dimensions of Example 1
DESIGN OPTION

dw
(in)

tw
(in)

tt
(in)

Unstiffened
noncomposite
85.23 0.568 1..186
composite unshored 70-19 0.533 0.438
composite shored
59.25 0.503 0.438
Stiffened
noncomposite
77.27 0.473 1,.544
composite unshored 72.50 0.441 0.,502
composite shored
57.95 0.355 0.438

Stiffeners No. of
Spacing
Iters.
ratio
(a/dw)

tb
(in)

AS
(in 2)

0.694

76.52
62.58
57.72

11

73..16 1.5*
57.96 1.5*
51.43 1.5*

9
12
9

0.996
1.133
0.904

1.008
1.28

10
11

Bottom flange width = 20" Top flange width = 12"


The economical design can be selected from these designs based on minimum
beam cross section area or minimum depth or other design criteria such as its
suitability and applicability to the site conditions and material and labor
availability. The most important fact is the availability of these designs in a

Transactions on the Built Environment vol 13, 1995 WIT Press, www.witpress.com, ISSN 1743-3509

Structural Optimization

231

very short time. The CPU time for these design ranged from 7 to 12 seconds.
The optimization process took about 9 to 12 iterations.
Effect of Stiffeners Cost on Beam Design
The effect of stiffeners cost on girder design is shown here by designing the
girder with different values of stiffener cost coefficient (%) while keeping cost
coefficient of plate girder (Ci) constant. Only the noncomposite stiffened
design case is used to show this effect. The effect of C2 value on web depth is
shown in Figure 4. It can be seen that as the stiffeners cost increases the web
depth decrease. This is because the stiffeners are mainly a function of the web
depth. Good estimate for stiffeners cost and beam cost will results in an
economical beam design accounting for full construction cost.

y = beam depth without sitteners cost (

.2

C2/C,
Figure 4. Change in girder depth with increase in stiffeners cost
Effect of Flange Width on the Optimum Beam Cross Section Area
Since the beam flange widths are not considered design variables in this paper
for the reasons mentioned in reference [10], it is important to study their effect
on the optimum girder cross section area. Figure 5 shows changes in the
optimum corss-section area with flange width.
Concluding Remarks
The ability to create and compare different designs for the same structure is the
real benefit from a general optimization based formulation. This approach
enables the designer to investigate different designs in relatively short time,
especially at the preliminary design stage.
The efficiency of the formulation presented in this study for optimum design of
bridge girders subjected to AASHTO loads is very encouraging. Similar
formulations can be applied to other type of bridges, especially long span
bridges, where the dead loads are dominant forces in the optimum design. A
similar formulation can also be applied to hybrid plate girders, box girders and
truss bridges.

Transactions on the Built Environment vol 13, 1995 WIT Press, www.witpress.com, ISSN 1743-3509

232

Structural Optimization

1 - Lateral supports spaced at 6 ft.


2- Lateral supports spaced at 12 ft.
3- Lateral supports spaced at 24ft.

18
22
26
Flange width Jnches

Figure 5. Changes in optimum cross-section area with flange width


References
1. Douwen, A.A., "Optimum Design of Plate Girders", Proceedings of the
International Symposium on Steel Plated Structures, London, 1976.
2. Razani, R. and Goble, G.G., "Optimum Design of Constant-Depth Plate
Girders", Journal of Structural Division, ASCE, ST2, Vol. 92, April 1966.
3. Goble, G.G. and DeSantis, P.V., "Automated Optimum Design of
Unstiffened Girder Cross Sections", Journal of the Structural Division,
ASCE, ST6, Vol. 92, December 1966.
4. Holt, E.C., and Heithecker, G.L., "Minimum Weight Proportions for Steel
Girders", Journal of Structural Division, ASCE, ST10, Vol. 95, 1969.
5. Schilling, G.C., "Optimum Properties for I-Shaped Beams", Journal of the
Structural Division, ASCE, ST12, Vol. 100, December 1974.
6. Vachajitapan, P. and Rockey, K.C., "Design Method for Optimum
Unstiffened Girders", Journal of the Structural Division, ASCE, ST1, Vol.
104, January 1978.
7. Burns, S.A., and Ramamurthy, S., "Mathematical Programming in
Structural Design", Journal of the Structural Division, ASCE, ST7, Vol.
109, July 1983.
8. Fleisher, W., "Design and Optimization of Plate Girders and Welded
Fabricated Beams for Building Construction", Engineering Journal, AISC,
No. 1, 1st Quarter, 1985.
9. American Association of State Highway and Transportation Officials
(AASHTO) Standard Specifications for Highway Bridges.
10. Al-Gahtani, Ahmad Saad, "Optimum Design of Welded I-Beams Subjected
to Highway Bridge Loads", Ph.D. thesis, University of Iowa, Iowa City, IA
52242, May 1986.

S-ar putea să vă placă și