Professional Documents
Culture Documents
GIRDER BRIDGES
By Khaled M. Sennah1 and John B. Kennedy,2 Fellow, ASCE
ABSTRACT: Composite steel-concrete multicell box girder bridges combine excellent torsional stiffness with
elegance. While the current design practices in North America recommend few analytical methods for the design
of composite multicell box girder bridges, practical requirements in the design process necessitate a need for a
simpler design method. This paper presents an extensive parametric study using the finite-element method
in which 120 bridges of various geometries were analyzed. The parameters considered are: number of cells,
number of lanes, span length, and cross bracings. Results from testing a simply supported three-cell bridge
model is used to substantiate the analytical modeling. Based on the parametric study, moment and shear distri-
Downloaded from ascelibrary.org by New York University on 05/13/15. Copyright ASCE. For personal use only; all rights reserved.
bution factors are deduced for such bridges subjected to AASHTO truck loadings as well as dead load. Saint-
Venant torsional stiffness for composite cellular cross sections used in this study is also investigated. Recom-
mendations to enhance the torsional stiffness are formulated. An illustrative design example is presented.
The single concentrated load was taken as 80 kN for the mo- tained for each bridge prototype from the finite-element anal-
ment distribution study and 116 kN for the shear distribution ysis, taking into account the modification factor for multilane
study. Modification factors of 1, 1, 0.9, and 0.75 for one-, loading. The shear distribution factor, DSS, was then deter-
two-, three-, and four-lane loadings, respectively (AASHTO mined as follows:
1996) were applied. The two types of live loads were first
Vmax
applied on a simply supported girder, with a span equal to that DSS = (2)
of the bridge prototype, to determine which case produced the V
maximum moment at midspan or the maximum shear force at
the support. Subsequently, three loading cases were considered RESULTS
for each bridge prototype: full and partial AASHTO truck
loadings (or equivalent lane loadings), shown in Fig. 4, and Table 2 shows a comparison between the experimental and
the bridge dead load. In the partial loading cases [Fig. 4(b)], theoretical results for the deflection, longitudinal strain at mid-
the wheel loads close to the curbs were applied at a distance span, and the reaction distribution under eccentric two con-
of 0.6 m from the inside edge of the curb. centrated loading applied to the model at the midspan section
over locations W1 and W2, shown in Fig. 1. Reasonable agree-
ment can be observed between the theoretical and experi-
LOAD DISTRIBUTION FACTORS
mental results. It should be noted that this model failed at 255
The cellular cross section was divided into I-beam shaped kN when statically loaded by four concentrated loads, one load
girders as shown in Fig. 3(b). Each idealized girder consisted over each web at the midspan section.
of the web, steel top flange, concrete deck slab, and steel bot- A sensitivity study was first undertaken to determine the
tom flange. In order to determine the moment distribution fac- different factors that may influence the lateral load distribu-
tor, DMS, carried by each girder of the bridge, the maximum tion. The sensitivity study revealed that changing span-to-
moment, M, was calculated in a simply supported girder sub- depth ratio, concrete deck slab thickness, or bottom flange
jected to a line of wheel loads of an AASHTO truck, or a line thickness has an insignificant effect on both moment and shear
of half the lane loading. The longitudinal moment carried by distribution. This was also confirmed for moment distribution
elsewhere (Nutt et al. 1988). Therefore, the concrete deck slab
thickness was taken 225 mm as a minimum requirement rec-
ommended by the Canadian highway bridge design code
(CHBDC 1997). It should be noted that the effect of torsional-
to-flexural rigidity is implied in studying the effect of span-
to-depth ratio, concrete slab thickness, and the presence of
cross bracings. In practice, X-type bracings as well as top-
chords (lateral ties to the steel top flanges) are made from
single or back-to-back angles. The sensitivity study showed
that replacing the angle cross section by rectangular one, or
t2 2.6
= (3)
t3 NC
TABLE 4. Effect of Cross Bracings on Shear Distribution Fac- improves the ability of the cross section to transfer loads from
tors for Eccentric AASHTO Lane Loadings one girder to the adjacent ones.
Shear Distribution Factor, DSS, at Table 4 shows the effect of cross bracings on the distribution
Number of
Girder Number of shear forces between webs at the support line under an
cross bracings
eccentric AASHTO loading. It is observed that adding cross
Bridge type between supports 1 2 3 4 5
bracings enhances the shear distribution between webs. As an
(1) (2) (3) (4) (5) (6) (7)
example, when using eight cross-bracing systems, the maxi-
2l-4c-60 0 1.11 0.79 0.29 0.04 0.24 mum shear force is reduced by 10% in the loaded outer web,
1 1.40 0.80 0.30 0.04 0.19
2 1.02 0.80 0.31 0.04 0.19
increased by 17% in the central web, and the uplift force in
3 1.01 0.80 0.32 0.05 0.19 the outer web away from load is reduced by 16%.
5 1.01 0.80 0.33 0.07 0.19 The effect of cross-bracing systems on the deflection distri-
8 1.00 0.80 0.34 0.06 0.20 bution at midspan for fully loaded and partially loaded lanes
was also studied. It was observed that, when all lanes are
loaded, the uniformity of the transverse distribution of deflec-
8Z 2t4 tion is not significantly improved by adding cross bracings,
J= (5)
XY but is significantly improved when the bridge is partially
loaded, as expected. Furthermore, results not shown herein re-
For a three-cell cross section: vealed that adding cross-bracing systems with a spacing less
4Z 2t4(3X 10Y) than the maximum spacing of 7.5 m specified by the AASHTO
J= (6) (1996) has an insignificant effect on the distribution factors.
X 2 4XY 2Y 2
Therefore, it was decided to conduct the parametric study with
For a four-cell cross section: number of cross-bracing and top-chord systems of 3, 5, 8, 11,
and 17 for bridge spans of 20, 40, 60, 80, and 100 m, respec-
8Z 2t4(2X 5Y ) tively.
J= (7)
X 2 3XY Y 2
For a five-cell cross section: Bending Moment Distribution
4Z t4(5X 28XY 35Y )
2 2 2
The critical loading case for the maximum moment distri-
J= (8)
(X 2Y )(X 2 4XY Y 2) bution factor is affected by the number of lanes. The maximum
moment distribution factor for a fully loaded one-lane bridge
For a six-cell cross section: is approximately the same as that for a partially loaded one,
8Z 2t4(3X 4 70XY 3 62X 2Y 2 23X 3Y 28Y 4) while for two-, three-, and four-lane bridges the fully loaded
J= (9) bridge is the governing case for design. This is clearly evident
(X Y )(X 2Y )(X 3 Y 3 6Y 2X 5X2Y ) from the results in Table 3.
where X = B[1 (t4/t3)]; Y = d[(t4/t2)]; and Z = Bd. Fig. 7 shows the effect of the number of cells and span
length on the moment distribution factor for the central and
Effects of Cross-Bracing Systems outer girders of four-lane bridges due to full truck loading. It
is observed that for longer span lengths, the moment distri-
The torsional stiffness of a box girder results from three bution factor becomes smaller for the central girder by an av-
components: the Saint-Venant rigidity, the warping rigidity, erage of 8%, and larger for the outer girder by an average of
and the distortional rigidity. Increasing the flexibility of any 15%, thus exhibiting better moment distribution between gird-
of these components reduces the rigidity of the box girder. ers. Moreover, increasing the number of cells decreases the
Adding bracings between support lines is generally required girder moment, as expected.
for stability purposes at the construction phase. Table 3 shows Fig. 8 shows the change in the moment distribution factor
the effect of bracings on the moment distribution between ide- with different numbers of lanes for the outer girder of 60 m
alized girders. It can be observed that, in the case of bridges span bridge. It is observed that for a certain number of cells,
with cross bracings and with fully loaded lanes, the bending the moment distribution factor increases with increase in the
moment increases by a maximum of 3% in the outer girder number of lanes. It is also interesting to note that the moment
and decreases by a maximum of 3% in the central girder, while distribution factors for the intermediate girders are very close
in the case of bridges with partially loaded lanes the maximum to those of the central girder, as shown in Table 3. This is due
bending moment carried by the loaded outer girder is consid- to the excellent torsional stiffness of this type of bridge. There-
erably reduced by more than 30%. Thus, adding cross bracings fore, in practice, it is unnecessary to distinguish between cen-
JOURNAL OF BRIDGE ENGINEERING / FEBRUARY 1999 / 75
Shear forces in the webs at support lines are distributed in (3.4NL 2.15)(L)(1.25NL1.24) (15)
a manner quite different from that for bending moment, since
the governing loading case occurs when the truck loading is For the outer web due to fully loaded lanes:
located close to web and to the support line. Results shown in DSS = (0.8NL 0.15)(NC)(0.01NL0.96)(L)(0.02NL0.13) (16)
Table 4 reveal that, for the partial loading case, the reaction
at some outer bearings may be downward as a result of twist- For the central web due to fully loaded lanes:
ing. However, this effect is more than neutralized by the dead
DSS = (3NL 1.2)(NC)(0.14NL0.58)(L)(0.07NL0.04) (17)
load effects. The variation in the number of cells for the same
bridge width was studied for different numbers of lanes and It should be mentioned that the above expressions for shear
under dead load. Results not shown herein, for brevity, showed distribution factors do not provide only the maximum shear
that the shear force carried by the outer web decreases with distribution factors for the outer and central webs but also the
an increase in the number of cells. It was also observed that shear factors corresponding to each loading case. Thus, the
for the same number of cells, the shear distribution factor for use of these expressions for shear would lead to the design of
outer web increased with an increase in the number of lanes. shear connectors, the bridge bearings, and the supporting
frames or abutments. When necessary, the shear forces in the
Expressions for Moment and Shear Distribution intermediate webs can be found by linear interpolation.
Factors Different methods found in the literature for load distribu-
tion of multicell box girder bridges were limited to reinforced
From the results of the parametric study, it became evident and prestressed concrete sections. Empirical equations for mo-
that both the moment and shear distribution factors for com- ment and shear distribution factors for such bridges can be
posite multicell box girder bridges are governed by the fol- found in (1) AASHTO specification (1996); (2) The National
lowing parameters: (1) number of cells, NC; (2) number of Cooperative Highway Research Program (NCHRP 12-26
lanes, NL; and (3) bridge span length, L. Using a statistical 1991); and (3) The National Cooperative Highway Research
package for best fit, the following empirical expressions were Program (Nutt et al. 1988). Table 5 shows the load distribution
generated for moment and shear distribution factors for simply factors for shear and moment predicted by the different meth-
supported straight composite multicell box girder bridges. It ods in the literature for the outer and central girders of rein-
should be noted that the modification factors for multilane forced or prestressed multicell box girder bridges. Results
loading (AASHTO 1996) were included in the proposed ex- based on the finite-element modeling adopted herein as well
76 / JOURNAL OF BRIDGE ENGINEERING / FEBRUARY 1999
as results from the above proposed expressions, (10) (17) in- ing on the outer lane. For the central web, the shear distribu-
clusive, for composite multicell box girder bridges are also tion factor can be calculated from (17) as 0.98 due to full truck
presented in Table 5. Comparing the developed moment and loading. The shear distribution factor for the central web due
shear distribution factors with the AASHTO and NCHRP for- to partial truck loading is 0.43, by interpolation. By multiply-
mulas is not beneficial because they represent different cross ing each moment distribution factor by MDL for dead load and
sections. However, it can be observed that the results based by MLL (1 DLA) for live load, the moments carried by each
on the methods in the literature are quite limited when com- girder are found. Thus, the resulting design moments are 6010
pared with those from the present analysis. Furthermore, good kN.m and 8077 kN.m for the outer and intermediate girders,
comparison is shown between the results from the finite-ele- respectively. By multiplying each shear distribution factor by
ment analysis and the proposed expressions. VLL (1 DLA) for live load, the design shear forces carried
by each web due to different loading cases can be obtained.
Thus, the resulting shear forces carried by outer and central
ILLUSTRATIVE DESIGN EXAMPLE webs under partial truck loading are 238 kN, and 99 kN, re-
Consider a two-lane, four-cell, simply supported bridge with spectively. The shear carried by the far end web becomes 42
a concrete deck slab composite with a steel section to be de- kN. Under full truck loading, the resulting shear forces carried
signed according to AASHTO specifications. The bridge de- by the outer and central webs are 130 kN and 227 kN, re-
tails are as follows: span length = 54 m, deck width = 9.3 m, spectively. By linear interpolation, the shears carried by the
intermediate webs can be calculated. These results are used to
deck slab thickness = 225 mm, all steel plate thicknesses = 10
design the girders, the webs, the shear connectors, the bridge
mm, top steel flange width = 500 mm, steel section depth =
bearings, and the supporting frames or abutments.
2.2 m, modulus of elasticity of steel = 200 GPa, and modular
ratio = 7. The superimposed dead load due to asphaltite ma-
CONCLUSIONS
terial = 10 N/m2. Calculate the design moment and shear
forces assuming a dynamic load allowance, DLA, of 25%. An extensive theoretical and experimental investigation was
Taking the concrete and steel densities as 2,400 and 7,800 conducted to determine the effect of several variables on the
N/m3, respectively, the total dead load per meter length for the moment and shear distributions in simply supported straight
bridge is 66.55 kN/m. Based on this value, the maximum dead composite concrete deck-steel multicell box girder bridges.
load moment at midspan of the bridge per girder, MDL, is 359 Expressions were deduced for computing the moment distri-
kN.m, and the dead load reaction at the end of the bridge per bution factors for each girder and the shear distribution factor
girder, VDL, is 359 kN. Applying a line of AASHTO wheel at each web. The results from this investigation provide val-
loads and half the lane loadings on a simply supported girder uable design information currently unavailable in bridge codes
of span 54 m, the maximum moment at midspan, MLL, is 2270 for the design of composite cellular bridges. Based on this
kN.m and the maximum reaction, VLL, is 185 kN. Considering study, the following conclusions can be drawn:
the dead load to be distributed to the girders according to their 1. An empirical formula for the ratio between web thick-
relative stiffness and using (11), the moment distribution fac- ness to bottom flange thickness was deduced to produce
tors for the outer girder of the bridge can be calculated as 0.16 maximum torsional resistance for the cellular section.
due to dead load and 0.75 due to truck loading. Using (13), 2. The presence of at least three cross-bracing systems be-
the moment distribution factors for the intermediate girders tween support lines, with a maximum spacing of 7.5 m,
can be calculated as 0.23 due to dead load and 0.88 due to significantly enhances the load transverse distribution of
truck loading. The shear distribution factors for the outer web moments, shear forces as well as deflections.
can be obtained, from (14) and (16), as 1.03 due to partial 3. The use of the proposed expressions for shear would lead
truck loading on the outer lane, and 0.56 due to full truck to a more economical and reliable design of shear con-
loading. Using (15), the shear distribution factor for the far nectors, bridge bearings, and supporting frames or abut-
end web can be determined as 0.18 due to partial truck load- ments.
JOURNAL OF BRIDGE ENGINEERING / FEBRUARY 1999 / 77