You are on page 1of 8

Live-Load Distribution Factors for Concrete

Box-Girder Bridges
Shin-Tai Song1; Y. H. Chai2; and Susan E. Hida3
Downloaded from ascelibrary.org by WASHINGTON UNIV IN ST LOUIS on 05/20/13. Copyright ASCE. For personal use only; all rights reserved.

Abstract: The current American Association of State Highway and Transportation Officials 共AASHTO兲 Load and Resistance Factor
Design 共LRFD兲 Specifications impose fairly strict limits on the use of its live-load distribution factor for design of highway bridges. These
limits include requirements for a prismatic cross section, a large span-length-to-width ratio, and a small plan curvature. Refined analyses
using 3D models are required for bridges outside of these limits. These limits place severe restrictions on the routine design of bridges in
California, as box-girder bridges outside of these limits are frequently constructed. This paper presents the results of a study investigating
the live-load distribution characteristics of box-girder bridges and the limits imposed by the LRFD specifications. Distribution factors
determined from a set of bridges with parameters outside of the LRFD limits are compared with the distribution factors suggested by the
LRFD specifications. For the range of parameters investigated, results indicated that the current LRFD distribution factor formulas
generally provide a conservative estimate of the design bending moment and shear force.
DOI: 10.1061/共ASCE兲1084-0702共2003兲8:5共273兲
CE Database subject headings: Live loads; Load distribution; Bridges, girder; Bridges, concrete; Box girders; California.

Introduction commonly defined 共Barker and Puckett 1997; AASHTO 1998兲 as

F distributed
Concrete multicell box-girder bridges are one of the most com- g⬅ (1)
mon types of highway bridges built in California. Voids are intro- F beamline
duced in the superstructure of these structures to reduce its self-
where F distributed corresponds to the largest bending moment or
weight while maintaining a relatively large flexural and torsional
shear force distributed to the girder for all load combinations;
stiffness as well as strength. These structures are not only very
efficient in terms of their ability to resist external loads, they may while F beamline corresponds to the maximum bending moment or
also hide unsightly utilities in the interior and enhance the aes- shear force determined from a simple beam-line analysis of one
thetics of the surrounding environment. Under live-load condi- lane of traffic, assuming that the bridge superstructure can be
tions, a vehicle traveling on a bridge will first transfer its loads, idealized as a continuous beam. For routine design of bridges, the
including vehicular contents and any forces associated with dy- distribution factor g is used in conjunction with results from the
namic effects, to the bridge deck. The deck then acts like a con- simple beam-line analysis to estimate the design bending moment
nection element distributing the load to different girders. The dis- or shear force in the girder; i.e.
tribution of traffic loads to different girders, as commonly
characterized by a live-load distribution factor, is conceptually an Bending Moment or Shear Force in the Girder⫽F beamline⫻g
estimate of the portion of the traffic load distributed to each (2)
girder. The live-load distribution factor is extensively used to Guidelines for estimating the live-load distribution to different
streamline the design of bridges in the United States. girders are currently available in various bridge design codes. For
For most bridges, including multicell box-girder bridges, dis- example, the 1998 AASHTO LRFD Bridge Design Specifications
tribution of traffic loads to the different girders is not uniform.
共AASHTO 1998兲, referred to herein as the LRFD specifications,
The girder closest to the traffic load is expected to resist the
provide a set of distribution factor formulas for estimating the
largest portion of the load. The live-load distribution factor g is
bending moment and shear force in the interior and exterior gird-
1
ers. These formulas, which are primarily based on a study by
Graduate Research Assistant, Dept. of Civil and Environmental
Zokaie et al. 共1991兲, express the distribution factor in terms of a
Engineering, Univ. of California, Davis, CA 95616. E-mail:
ssong@ucdavis.edu set of simple geometric parameters such as span length, girder
2
Associate Professor, Dept. of Civil and Environmental Engineering, spacing, overall depth of the girder, number of cells, etc. The
Univ. of California, Davis, CA 95616. E-mail: yhchai@ucdavis.edu current set of distribution factor formulas in the LRFD specifica-
3
Senior Bridge Engineer, California Dept. of Transportation, tions are intended for lane loads instead of wheel loads, and dif-
Sacramento, CA 95816. E-mail: Susan – Hida@dot.ca.gov ferent formulas are provided for single-lane loading and multiple-
Note. Discussion open until February 1, 2004. Separate discussions lanes loading. Note that a multiple presence factor m, as
must be submitted for individual papers. To extend the closing date by
reproduced in Table 1 共AASHTO 1998兲, is also embedded in the
one month, a written request must be filed with the ASCE Managing
Editor. The manuscript for this paper was submitted for review and pos- formulas to reflect the reduced likelihood of all traffic lanes being
sible publication on November 6, 2001; approved on September 23, 2002. loaded simultaneously. Thus, the distribution factor formulas rec-
This paper is part of the Journal of Bridge Engineering, Vol. 8, No. 5, ommended by the LFRD specifications strictly correspond to
September 1, 2003. ©ASCE, ISSN 1084-0702/2003/5-273–280/$18.00. mg formula .

JOURNAL OF BRIDGE ENGINEERING © ASCE / SEPTEMBER/OCTOBER 2003 / 273

J. Bridge Eng. 2003.8:273-280.


Table 1. Number of Loaded Lanes and Corresponding Multiple
Presence Factor
Number of loaded lanes Multiple presence factor
1 1.20
2 1.00
3 0.85
4 or greater 0.65

Objectives
Downloaded from ascelibrary.org by WASHINGTON UNIV IN ST LOUIS on 05/20/13. Copyright ASCE. For personal use only; all rights reserved.

Although the objective of the distribution factor formulas is to


simplify the design process, fairly strict restrictions are imposed
by the LRFD specifications on the use of these formulas. Limits
are placed on the bridge curvature, the span-length-to-width ratio,
and the number of bridge girders, and requirements for prismatic
cross sections with parallel girders are imposed. More specifi-
cally, the LRFD specifications distribution factor formulas may
only be used for design of bridges with 共1兲 constant deck width;
共2兲 number of girders not less than four; 共3兲 parallel beams with
approximately equal stiffness; 共4兲 roadway overhand less than
910 mm; 共5兲 superstructure span length exceeding 2.5 times its
width; and 共6兲 angular change of less than 12° in plan for a
torsionally stiff closed section 共AASHTO 1998兲.
Box-girder bridges outside of these limits are frequently con-
structed in California; consequently, such constraints place severe
restrictions on the routine design of bridges. Design outside of
these limits requires refined 3D models such as grillage or finite-
element models, which are currently not a part of the standard
bridge design process and often require special skills and analysis
tools. Among the aforementioned restrictions, restrictions 1, 3, 5,
and 6 are particularly restrictive. A study was recently conducted
on the limits imposed by constraint numbers 1, 3, 5 and 6. The
study compared the distribution of live-load bending moments
and shear forces with those suggested by the LRFD specifications
for a typical range of parameters encountered in concrete box-
girder bridges in California.
Fig. 1. Comparison between grillage and finite-element models: 共a兲
two-span continuous bridge structure; 共b兲 bending moment distribu-
Methodology tion 共girder 2兲; 共c兲 shear force distribution 共girder 2兲

Bridges outside the limits of the current LRFD specifications


were analyzed using refined and simple beam-line models so that the grillage model for box-girder bridges follows the general
distribution factors for bending moment and shear force could be guidance available in the literature 共e.g., Hambly 1991; Barker
determined. Distribution factors thus obtained are denoted as and Puckett 1997; O’Brien and Keogh 1999兲. Before proceeding
g analysis to signify that the distribution factor is obtained from the with the actual usage of the grillage model for distribution factors,
analysis instead of the formula: the grillage model was first calibrated against a finite-element
F refined model in order to gain confidence in the model.
g analysis⫽ (3) The basic dimensions of a continuous box-girder bridge as
F beamline well as one of the loading conditions used for the calibration of
where F refined corresponds to the largest bending moment or shear the grillage model are shown in Fig. 1共a兲. The box-girder has two
force in the girder from the refined analysis; while F beamline cor- equal spans, each with a length of 30.3 m, and has four cells with
responds to the maximum bending moment or shear force from equal spacing between girders. The overall depth of the box-
the simple beam-line model 共Barker and Puckett 1997兲 subjected girder is 1.263 m, and the edge-to-edge width is 12.21 m. The
to one lane of traffic. The distribution factor g analysis is then mul- loading condition shown in Fig. 1共a兲 consists of two point-loads
tiplied by the appropriate multiple presence factor m and com- applied off-center to induce significant twisting of the box-girder
pared with the distribution factor obtained using the LRFD speci- in addition to bending moment and shear force.
fications formulas, which is denoted as mg formula , in order to Figs. 1共b and c兲 show the distribution of bending moment and
allow an assessment of the level of conservatism in the formulas. shear force in one of the longitudinal girders 共girder 2兲. The con-
tinuous line corresponds to the grillage model, whereas the ‘‘dot’’
symbol corresponds to the finite-element model. As can be seen in
Refined Model
Fig. 1共b兲, a sharp increase in bending moment occurs under the
The commonly used grillage model is used as the refined analysis point load, resulting in the largest bending moment occurring at
tool in this study. Selection of properties and implementation of that location. In general, the bending moment from

274 / JOURNAL OF BRIDGE ENGINEERING © ASCE / SEPTEMBER/OCTOBER 2003

J. Bridge Eng. 2003.8:273-280.


the grillage model compares very well with the bending moment
from the finite-element model, even though the bending moment
from the finite-element model is slightly larger. However, the dif-
ference in bending moment between the two models is within 5%.
Similarly, the shear force distribution shown in Fig. 1共c兲 also
indicates very good agreement between the two models. The gril-
lage model is able to simulate very well the discontinuity in the
shear force under the point load and at the continuous support, as
can be seen by the good agreement between the grillage model
and the finite-element model. Although not shown in Fig. 1共c兲, the
comparison of vertical shear forces in the exterior girder between
Downloaded from ascelibrary.org by WASHINGTON UNIV IN ST LOUIS on 05/20/13. Copyright ASCE. For personal use only; all rights reserved.

the grillage and finite-element models is also very good, despite


the inclined web in the exterior girder.
Although not presented in this paper, four additional load
cases were used for the calibration of the grillage model. These
additional load cases were selected to simulate the large negative
bending moment over the support, the large positive bending mo-
ment in the loaded span and associated negative bending moment
in the adjacent span, and the large shear force near the support. In
general, the simulation of bending moment and shear force by the
grillage model compares very well with the finite-element model.
Details of the calibration of the grillage model can be found in
Song et al. 共2001兲.

Fig. 2. AASHTO design vehicular live loads: 共a兲 HL-93K, design


Traffic Loads truck with lane load; 共b兲 HL-93M, design tandem with lane load; 共c兲
Unlike the calibration of the grillage model, where point loads dual trucks; 共d兲 dual tandems
were used, the distribution factors were calculated using the
AASHTO HL-93 design vehicular loads 共AASHTO 1998兲 in this
study. The HL-93 loads consist of a single design truck combined and are separated into the following subsections: 共1兲 nonprismatic
with a design lane load, referred to as HL93K, or a single design cross section; 共2兲 small plan aspect ratio; and 共3兲 curved bridges.
tandem combined with a design lane load, referred to as HL93M.
Extreme load effects, as characterized by the largest positive and
negative bending moments and shear forces, are determined using Nonprismatic Cross Section
the HL-93 load combinations per LRFD specifications 共AASHTO One of the rather restrictive constraints imposed by the LRFD
1998兲. In addition, dual trucks and dual tandems are used for the specifications for the use of its distribution factor formulas in
maximum negative bending moment 共AASHTO 1998兲. Figs. 2共a conjunction with the simple beam-line analysis is the requirement
and b兲 show the elevation of the axle loads and lane load associ- for the girders of the box-girder to be parallel and for the deck
ated with HL93K and HL93M, while Figs. 2共c and d兲 show the width to be constant. Bridges with variable-width and nonparallel
axle loads associated with the dual trucks and the dual tandems. It girders, which violate constraints 1 and 3 discussed previously,
should be noted that a 33% dynamic load allowance is applied to are frequently used in bridge structures to accommodate the sepa-
the design truck or design tandem, but not to the design lane load. ration or merging of traffic lanes.
For dual trucks and dual tandems, on the other hand, their mag- For the study of possible application of the LRFD distribution
nitudes are reduced to 90% including that of the design lane load. factor formulas to bridges with a nonprismatic cross section, two
For the design truck and design tandem, the transverse spacing of continuous box-girder bridges with dimensions shown in Fig. 3
the wheels is 1.8 m. Further details of the vehicular loads are were analyzed. These bridges have two equal spans, each with a
given in Song et al. 共2001兲. span length of 30.3 m. The nonprismatic cross section bridge
shown on the left-hand side of Fig. 3 is typical of an on-off ramp
structure where widening of the bridge is often required. In this
Results case, the widening of the bridge is facilitated by a flare of 6.25%,
starting from 0.2L in one span and continuing full length in the
The accuracy or conservatism of the LRFD specifications distri- second span. It is worth noting that the rate of widening is close
bution factor formulas may be assessed by comparing the distri- to the practical geometric limit of 6.7% for widening of bridges in
bution factors from the refined analyses 关i.e., Eq. 共3兲兴 with the California 共California Department of Transportation 2001兲. An
distribution factors calculated from the LRFD formulas. To this additional girder is added to the widened portion of the bridge,
end, an acceptance ratio R A is defined as resulting in nonparallel girders. For comparison of results, a pris-
m⫻g analysis m⫻F refined matic bridge with parallel girders, shown on the right-hand side of
R A⬅ ⫽ (4) Fig. 3, was also analyzed. Except for the deck width, the pris-
mg formula F beamline⫻mg formula
matic bridge is similar to that of the bridge with the nonprismatic
where m⫽multiple presence factor, as given in Table 1. An ac- cross section. Note that a nominal deck width of 11.36 m, as
ceptance ratio of R A ⬍1 indicates that the distribution factor rec- adopted for these structures, can only accommodate two traffic
ommended by the LRFD specifications formula is conservative. lanes at most after allowing for the width of the barriers. In order
Discussions of the results are primarily based on this parameter to create the most severe load condition while maintaining a rea-

JOURNAL OF BRIDGE ENGINEERING © ASCE / SEPTEMBER/OCTOBER 2003 / 275

J. Bridge Eng. 2003.8:273-280.


Downloaded from ascelibrary.org by WASHINGTON UNIV IN ST LOUIS on 05/20/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Sections and plan views of two-span continuous box-girder bridge with prismatic and nonprismatic cross section

sonable number of analyses, traffic lanes are positioned as close LRFD specifications are applicable to the two-span bridge in this
to the bridge girder as possible and follow the closest flare girder study, which has nonparallel girders or a nonprismatic cross sec-
when the bridge widens. tion.
Acceptance ratios for bridges with prismatic and nonprismatic A similar level of conservatism in the LRFD distribution factor
cross sections are presented in Table 2. In all cases, acceptance formulas is also observed for the prismatic bridge, even though
ratios are less than unity, irrespective of whether one or two traf- the acceptance ratio varies somewhat between the exterior and
fic lanes are loaded. For interior girders of the nonprismatic interior girders. For the interior girder, the acceptance ratio R A
bridge, acceptance ratios are 0.72 and 0.77 for positive bending varies between 0.8 and 1.0, as can be seen in Table 2 for both
moment, 0.90 and 0.88 for negative bending moment, and 0.95 positive and negative bending moments and for shear force. An
and 0.96 for shear force, for one and two lanes loaded, respec- acceptance ratio of 1.0 is calculated for the shear force of the
tively. For exterior girders of the nonprismatic bridge, acceptance
interior girder when the structure is subjected to a one-lane load-
ratios are 0.52 and 0.74 for positive bending moment, 0.66 and
ing. For the exterior girder, the acceptance ratio R A varies be-
0.86 for negative bending moment, and 0.61 and 0.78 for shear
tween 0.54 and 0.83 for both bending moment and shear force,
force, for one and two lanes loaded, respectively. The LRFD dis-
tribution factor formula is thus more conservative for the exterior and it is therefore smaller than that of the interior girder. Although
girder than for the interior girder. The acceptance ratios presented not presented in this paper, actual values of the distribution factor
in Table 2 indicate that the distribution factor formulas in the are available in Song et al. 共2001兲.

Table 2. Comparison of Acceptance Ratios between Bridges with Prismatic and Nonprismatic Cross Section
One lane loaded, m⫽1.2 Two lanes loaded, m⫽1
Bridge Positive moment Negative moment Shear force Positive moment Negative moment Shear force
Interior girders
Nonprismatic cross section 0.72 0.90 0.95 0.77 0.88 0.96
Prismatic cross section 0.85 0.93 1.00 0.88 0.89 0.83
Exterior girders
Nonprismatic cross section 0.52 0.66 0.61 0.74 0.86 0.78
Prismatic cross section 0.54 0.66 0.58 0.76 0.83 0.72
Note: m⫽multiple presence factor.

276 / JOURNAL OF BRIDGE ENGINEERING © ASCE / SEPTEMBER/OCTOBER 2003

J. Bridge Eng. 2003.8:273-280.


Table 3. Parameters for Study of Plan Aspect Ratios
Number Deck width, Plan aspect ratio, Maximum number
of cells W 共m兲 L/W of traffic lanes
Girder spacing S⫽1.72 m and S/D⫽1.25
4 9.31 3.28 2
6 12.75 2.39 3
8 16.18 1.89 4
10 19.61 1.56 5
Girder spacing S⫽2.52 m and S/D⫽1.83
4 12.28 2.48 3
Downloaded from ascelibrary.org by WASHINGTON UNIV IN ST LOUIS on 05/20/13. Copyright ASCE. For personal use only; all rights reserved.

6 17.32 1.76 4
8 22.35 1.37 5
10 27.38 1.11 7
Girder spacing S⫽3.09 m and S/D⫽2.25
4 14.40 2.12 3
6 20.58 1.48 5
8 26.79 1.14 7
10 32.94 0.93 8
Note: Bridge span length, L⫽30.5 m; overall depth, D⫽1.37 m.

Small Plan Aspect Ratio


Bridge crossings in high traffic volume areas often result in plan
dimensions that are wide relative to its span. The current LRFD
specifications, however, permit the use of its live-load distribution
factors only in bridges with a plan aspect ratio, i.e., span-length-
to-width ratio, that is greater than 2.5. The large number of traffic
lanes needed for bridges in these areas often results in structures
that violate the limit on plan aspect ratio. Refined analysis is
required for bridges with a span aspect ratio smaller than this
limit, even though these structures often have rather simple ge-
ometry.
Fig. 4. Distribution factors in interior girders of different plan aspect
Twelve simple-span box-girder bridges of different plan aspect
ratio bridges with two lanes loaded: 共a兲 bending moment from LRFD
ratios are selected for this study. These bridges have a span length
formula; 共b兲 bending moment from refined analysis; 共c兲 shear force
of 30.5 m and a depth of 1.37 m. Their girder spacing varies from
from LRFD formula; 共d兲 shear force from refined analysis
1.72 to 3.09 m and their number of cells varies from 4 to 10 to
produce 12 different plan aspect ratios, ranging from 3.28 to 0.93.
The overall deck width and number of cells, as well as their
resulting plan aspect ratio, are given in Table 3. Note that the in the LRFD formula. Even though the acceptance ratio is less
majority of the cases are outside of the LRFD specifications plan than unity in all cases, the acceptance ratio for bridges with a
aspect ratio limit of 2.5. The maximum number of traffic lanes narrow deck and small girder spacing tends to be relatively large,
that the respective bridge structure can accommodate is also with values fairly close to unity, indicating that the distribution
shown in Table 3. factor from the refined analysis is not much smaller than the
Because of the large number of structures and load cases in- distribution factor calculated from the LRFD formula. On the
volved in this study, only selected results are presented in this other hand, bridges with a wide deck or large girder spacing result
paper. Fig. 4 shows histogram plots of the distribution factor from in a smaller acceptance ratio. Even though results in Fig. 5 show
the LRFD formula and refined analysis for bending moment and only the acceptance ratio for bending moment and shear force in
shear force in the interior girder when subjected to a two-lane the interior girder with two traffic lanes loaded, similar trends
loading. The distribution factor, denoted as mg, is plotted on the were observed for the bending moment and shear force in the
vertical axis, while the span-length-to-width ratio (L/W) and exterior girder for all load cases. The general trend in Fig. 5 also
girder-spacing-to-depth ratio (S/D) are plotted on the horizontal indicates that the LRFD specifications formula becomes increas-
axes. Cross comparison between Figs. 4共a and b兲, for bending ingly conservative as the plan aspect ratio becomes smaller than
moment, and between Figs. 4共c and d兲, for shear force, indicate the limit of 2.5, which seems to suggest that the limit on the plan
that the distribution factor from the refined analysis is smaller aspect ratio may be unwarranted. Further results of the study can
than the distribution factor calculated using the LRFD specifica- be found in Song et al. 共2001兲.
tions formula. Even though the distribution factor from the re-
fined analysis is not uniform with respect to the span-length-to-
Curved Bridges
width ratio or girder-spacing-to-depth ratios, its variation
generally follows the trend given in the LRFD specifications for- Curved structures are often required for highway bridges, espe-
mula. cially when separations or on-off ramps are involved. In the cur-
The same set of data is plotted in terms of the acceptance ratio rent LRFD specifications 共AASHTO 1998兲, however, a limit is
in Fig. 5 for a qualitative assessment of the level of conservatism placed on the distribution factor formula for curved bridges. More

JOURNAL OF BRIDGE ENGINEERING © ASCE / SEPTEMBER/OCTOBER 2003 / 277

J. Bridge Eng. 2003.8:273-280.


Downloaded from ascelibrary.org by WASHINGTON UNIV IN ST LOUIS on 05/20/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 6. Geometry of curved bridges: 共a兲 plan view; 共b兲 cross section

moment and shear force in the interior girder for two lanes
loaded. Note that the angular change is not a variable in the
LRFD distribution factor formula, even though a limit is placed
on the curvature of the bridge. Consequently, the distribution fac-
tor is plotted against the span length for different radii of curva-
ture in Fig. 7 instead of the angular change. Results in Fig. 7
show that distribution factors from the refined analysis are smaller
than those calculated from the LRFD formula. The distribution
factor from the refined analysis also does not vary significantly
Fig. 5. Acceptance ratios in interior girders of different plan aspect with the different radii of curvature or angular change between
ratio bridges with two lanes loaded: 共a兲 bending moment; 共b兲 shear bents. Note that the results from the refined analyses are the same
force
for a given span length, which means that the data points for the
four different radii of curvature collapse into the same point when
specifically, refined analyses are required for bridges with angular plotted in Fig. 7. The distribution factor from the refined analysis
change greater than 12° in any one span. This limit is rather generally follows the same trend as the distribution factor from
restrictive, as geometric design often necessitates the construction the LRFD specifications formula.
of highly curved structures, which are in frequent violation of the Acceptance ratios for the same set of data are shown in Fig. 8.
12° limit. All acceptance ratios are smaller than unity, including those out-
Four multispan box-girder bridges with different radii of cur- side of the LRFD limit of 12°. The variation of acceptance ratio
vature are selected for the study of the 12° limit imposed by the with angular change follows a similar trend for all bridges. The
current LRFD specifications. The four radii are R⫽76.3, 122.0, trend noted for positive bending moment and shear force in Figs.
183.0, and 228.8 m, as shown in Fig. 6共a兲. Even though the radius 7 and 8 is also observed for the negative bending moment and
of curvature is constant for a given bridge, the angular change
between two ends of a span is different due to the different span
length in the structure. The angular change between bents varies Table 4. Angular Change in Curved Bridges
from 5.7 to 34.4° in these four bridges. Details of the angular Angular change between bents 共°兲
change in curved bridges are given in Table 4. All curved bridges Bridge 1, Bridge 2, Bridge 3, Bridge 4,
are prismatic in cross section and continuous over the interior Bridge span R⫽76.25 m R⫽122.0 m R⫽183.0 m R⫽228.75 m
support. The cross section of the bridge is shown in Fig. 6共b兲. The
width of the bridge deck can accommodate a maximum of two Span 1, 17.19 10.74 7.16 5.73
traffic lanes. It should be noted that, unlike the pin-support con- L⫽22.88 m
dition that was assumed for straight bridges in the previous two Span 2, 22.92 14.32 9.55 7.64
parameters, circular columns with a diameter of 1.83 m and a L⫽30.50 m
height of 7.32 m are used for the interior support. These column Span 3, 34.38 21.49 14.32 11.46
supports, instead of pin-support, are intended to include any in- L⫽45.75 m
fluence of torsional effects on live-load distribution in curved Span 4, 25.78 16.11 10.74 8.59
bridges. L⫽34.31 m
Distribution factors from the refined analysis and LRFD speci- Note: L⫽span length 共along centerline of bridge兲; R⫽radius of curva-
fications formula are plotted in Fig. 7 for the positive bending ture.

278 / JOURNAL OF BRIDGE ENGINEERING © ASCE / SEPTEMBER/OCTOBER 2003

J. Bridge Eng. 2003.8:273-280.


Downloaded from ascelibrary.org by WASHINGTON UNIV IN ST LOUIS on 05/20/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 7. Live load distribution factors in interior girders of curved Fig. 8. Acceptance ratios for interior girders of curved bridges with
bridges with two lanes loaded: 共a兲 bending moment; 共b兲 shear force two lanes loaded: 共a兲 bending moment; 共b兲 shear force

shear force in the interior girder or exterior girder as well as for


one lane loaded. For the four curved bridges in this study, results factors are generally more conservative for exterior girders than
do not indicate a larger acceptance ratio with an increased angular for interior girders. For the range of parameters investigated in
change between bents, which means that the LRFD formulas do this study, the current set of distribution factors appears appli-
not necessarily produce a less conservative estimate of the live- cable to box-girder bridges with nonparallel girders, to bridges
load distribution for curved bridges outside the limit of 12°. with a length-to-width ratio as small as unity, and to curved
bridges with an angular change as large as 34°. However, because
of the small set of bridges used in the study, results presented
Conclusions should not be construed to imply an overall conservatism of the
LRFD formulas; further study of the limits with a more extensive
The current AASHTO Design Specifications provide a set of dis- parameter range is warranted.
tribution factor formulas for estimating the distribution of bend-
ing moment and shear force effects in the interior and exterior
girders of highway bridges. These distribution factors, when com- Acknowledgments
bined with the critical bending moment and shear force from the
simple beam-line analysis, are routinely used for the design of The research described in this paper was funded by the California
highway bridges. The current LRFD specifications, however, im- Department of Transportation under Contract No. 59A0148. Their
pose fairly strict limits on the use of its live-load distribution financial support is gratefully acknowledged. The assistance pro-
factor formulas, including requirements for a prismatic cross sec- vided by Richard Brice of the Washington State Department of
tion, a large span-length-to-width ratio, and a small plan curva- Transportation and Toorak Zokaie of LEAP Software, Inc., is
ture. Refined analyses using 3D models are required for design of greatly appreciated. Any opinions, findings, and conclusions ex-
bridges outside of these limits. pressed in the paper are those of the writers and do not necessar-
Results pertaining to a study of the restrictions imposed by the ily reflect the views of the sponsor.
LRFD specifications are presented in this paper. Box-girder
bridges outside the limits of the LRFD specifications are analyzed
using the standard truck loading from the LRFD specifications. References
Results indicate that the current LRFD specifications distribution
factor formulas for box-girder bridges generally provide a conser- American Association of State Highway and Transportation Officials
vative estimate of the design bending moment and shear force, 共AASHTO兲. 共1998兲. AASHTO bridge design specifications, 2nd Ed.,
even for bridges outside of the specifications limits. Distribution Washington, D.C.

JOURNAL OF BRIDGE ENGINEERING © ASCE / SEPTEMBER/OCTOBER 2003 / 279

J. Bridge Eng. 2003.8:273-280.


Barker, R. H., and Puckett, J. A. 共1997兲. Design of highway bridges: Song, S., Chai, Y. H., and Hida, S. E. 共2001兲. Live load distribution in
Based on AASHTO LRFD bridge design specifications, Wiley, New multi-cell box-girder bridge and its comparison with the AASHTO
York. LRFD Bridge Design Specifications, California Department of Trans-
California Department of Transportation 共Caltrans兲. 共2001兲. Highway de- portation, Sacramento, Calif.
sign manual, Sacramento, Calif. Zokaie, T., Osterkamp, T. A., and Imbsen, R. A. 共1991兲. ‘‘Distribution
Hambly, E. C. 共1991兲. Bridge deck behavior, E&FN Spon, London. of wheel loads on highway bridges.’’ Final Rep. for Project
O’Brien, E. J., and Keogh, D. L. 共1999兲. Bridge deck analysis, E&FN 12-26/1, National Cooperative Highway Research Program, Washing-
Spon, London. ton, D.C.
Downloaded from ascelibrary.org by WASHINGTON UNIV IN ST LOUIS on 05/20/13. Copyright ASCE. For personal use only; all rights reserved.

280 / JOURNAL OF BRIDGE ENGINEERING © ASCE / SEPTEMBER/OCTOBER 2003

J. Bridge Eng. 2003.8:273-280.

You might also like