ApPROXIMATE ANALYSIS OF IRREGULAR SLAB-ON-GIRDER BRIDGES

By Sami W. Tabsh, l Associate Member, ASCE, and Kalpana Sahajwane

ABSTRACT: A simple, approximate method for determining the live-load girder distribution factors of irregular
slab on I-beam bridge layouts is presented. The method can also be used for the live-load analysis of bridges.
subjected to one truck, as in the case of fatigue, or two (or more) different truck configurations, as in the case
of permit loading. The method takes into account the longitudinal as well as transverse effects of the truck loads.
It is based on isolating strips of the deck slab in the transverse direction directly under the wheel loads and
treating them as beams on elastic supports. The stiffness of the supports at the strip locations is computed based
on the geometric properties of the beams and location of the trucks in the longitudinal direction. The accuracy
of the method for shear and flexure of I-beams in simple and continuous span bridges is verified using a detailed
three-dimensional finite-element analysis.
Downloaded from ascelibrary.org by New York University on 05/16/15. Copyright ASCE. For personal use only; all rights reserved.

INTRODUCTION (overly conservative or unconservative) results. Therefore,

there is a need to study the behavior of irregular bridges and
Live-load girder distribution factors (GDF) are important for develop a simple, rational method for determining live-load
highway bridge design because they are needed to determine GDFs for bridges with unequal girder spacings.
the required strength and serviceability of bridge elements. In this study, a simple, approximate method is developed
The American Association of State Highway and Transporta- for the live-load analysis of highway girder bridges with ir-
tion Officials (AASHTO) Standard Specifications for Highway regular geometry. It can also be used for the live-load analysis
Bridges (1996) provide approximate wheel load distribution of bridges subjected to one truck, as in the case of fatigue
factors that are limited to bridges with equal girder spacings, loading, or two (or more) different truck configurations, as in
constant bridge width, AASHTO truck axle configuration, and the case of permit loading. The method is simple enough to
no skewness between the superstructure and substructure ele- be used with either hand calculations or a continuous beam
ments. Further, AASHTO's distribution factors do not consider computer program. The method was originally proposed in the
the effects of the span length, number of girders, deck slab American Institute of Steel Construction (AISC) Highway
thickness, and stiffness of the girders. In recent years, more Structures Design Handbook (1982) for checking fatigue
accurate distribution factors than the ones in the AASHTO's stresses under a single truck. It is expanded to include the
standard specifications have been developed based on the fi- torsional effect of the girders and the interaction with dia-
nite-element procedure (Khaleel and Itani 1990; Zokaie et al. phragms on the wheel load distribution characteristics. It con-
1991; Marx et al. 1991; Tarhini and Frederick 1992; LRFD
1994). However, these live-load distribution factors are mainly
Widened deck
applicable to either bridges with regular geometry or to segment
skewed bridges.
In general, the use of irregular girder layouts on highway
bridges is not desirable because the resulting complex structure
will need advanced methods of analysis. Often, irregular
bridges with unequal beam/girder spacings are generated when
Existing Bridge j----;r~j~
staged construction is used during bridge replacement and
when widening an existing bridge, as shown in Fig. 1. Addi-
tional examples include the presence of under-bridge utilities
and the use of optimized designs.
The use of three-dimensional finite-element analysis to de- added new
termine the live-load distribution in irregular bridges is an ex- girder
pensive and time-consuming process. Therefore, traditionally,
simple methods for determining girder distribution factors for (a)
bridges with unequal spacings have been employed in practice.
One simple method is based on evaluating the average girder STAGE2 ~STAGE 1
,-,
spacing on each side of the girder under consideration at the ," "
location of the truck axle and then using it with AASHTO's t__ ~~::::::;;~;::::::.~- ;I;;:::::;;;;;;=;;~
GDF expressions. Another method assumes the deck slab to "
..
II
II

be simply supported at the location of the girders and uses ,C'.!!:.: L.!.P.J' .. ......- .

statics to determine the GDF, as in the case of finding GDFs ,

I
I
'.
for shear for truck axles located near an end support. The
traditional approaches to determine GDFs for irregular bridges
......,_._-----_._----~------~
,
have not been verified in the past and may lead to erroneous
,,
I

,
I,
'Asst. Prof., Dept. of Civ. and Envir. Engrg., Univ. of Houston, 4700 I
I
pier
I
Calhoun St., Houston, TX 77204-4791. ,,
I

'Struct. Analyst, Lockheed Martin, 2400 NASA Rd. I, Houston, TX ,

77258. I---~---------------~-------"--
~
I
J__ __J
Note. Discussion open until July I, 1997. To extend the closing date
one month, a written request must be filed with the ASCE Manager of
Journals. The manuscript for this paper was submitted for review and (b)
possible publication on January 22, 1996. This paper is part of the Jour-
nal of Bridge Engineering, Vol. 2, No.1, February, 1997. iClASCE, ISSN FIG. 1. Examples of IrregUlar Bridges: (a) Widened Bridge;
1084-0702197/0001-0011-0017/$4.00 + $.50 per page. Paper No. 12408. (b) Staged Construction

JOURNAL OF BRIDGE ENGINEERING / FEBRUARY 1997/11

J. Bridge Eng. 1997.2:11-17.

 siders the longitudinal as well as transverse effect of the wheel p
loads on the bridge. The generated GOF using this method is k=E
Ll
a function of the bridge length and width; number, spacing,
and stiffness of girders; deck slab thickness; and skew angle.
The method also accounts for the number, spacing, and weight
of axles. .",.-- ....
DEVELOPMENT OF METHOD
The proposed method is based on isolating transverse strips
of the deck slab directly under the wheel loads, as shown in
--\ filii'

~
Deflected Shape
Fig. 2. In general, the width of the strips is a function of
FIG. 5. Determination of Elastic Spring Constant for a Contin-
several parameters including the slab thickness, tire pressure, uous Bridge
tire contact area, and girder spacing. However, the analysis
can be greatly simplified by taking the width of the strip equal
Downloaded from ascelibrary.org by New York University on 05/16/15. Copyright ASCE. For personal use only; all rights reserved.

to the distance between the girders or beams, as shown in Fig. Diaphragm T Diaphragm

~t=====I···===========r~
3. If the bridge is composed of several girders at unequal spac-
ings, then the strip width may be taken equal to the average

Y d-y
•
(a) Girder segment between diaphragms

FIG. 2. illustration of Approximate Method

(b) At a diaphragm (c) Away from a diaphragm
p
FIG. 6. Girder Rotation between Diaphragms

spacing of all girders. The strips are then analyzed as contin-

Strip Width ~+- -+--(

uous beams on elastic supports, as shown in Fig. 4. The dis-
placement and rotational stiffnesses of the supports (Le., gird-
ers) in each strip are computed based on the geometric
properties of the girders, position of the truck wheels in the
longitudinal direction, and location of the diaphragms. Fig. 5
shows the procedure for computing the displacement spring
constant, k, for a continuous bridge. The rotational spring con-
stant, c, on the other hand, is determined assuming the slab
does not rotate near a diaphragm, as shown in Fig. 6. The
spring reactions at the location of the girder under considera-
1111( tion are then determined for all strips. These vertical reactions
are placed on the isolated girder under investigation and the
FIG. 3. Determination of Strip Width Based on a 450 Load Dis- maximum load effect (flexure or shear) is determined at a
tribution specified location. Finally, the wheel load girder distribution
factor is computed as the ratio of the beam moment (or shear)
Lane 1 Lane 2 to the effect of one-half the truck load.
3.6m 3.6m APPLICATION OF METHOD
Truck 1 Truck 2 The method is demonstrated using a simply supported com-
posite steel beam bridge with regular geometry. The consid-
ered bridge is 9.6 m (32 ft) wide and 30.5 m (100 ft) long
1.8 m 1.2 m 1.8 m between the bearings. It is composed of a 200 mm (8 in.) thick
slab supported on four W920 X 446 beams spaced at 2.4 m
(8 ft). An average concrete haunch above the top steel flange
of the beam equal to 25 mm (1 in.) is considered. The specified
28-day concrete compressive strength in the slab is 28 MPa
(4,000 psi). Diaphragms in the form of cross-frame bracing
are placed at 7.6 m (25 ft) intervals. The wheel load GDFs for
maximum flexure at midspan and maximum shear at the sup-
FIG. 4. Isolated Strip Subjected to Axle Loads port in an interior beam are considered.
12/ JOURNAL OF BRIDGE ENGINEERING / FEBRUARY 1997

J. Bridge Eng. 1997.2:11-17.

 GDF for Flexure at Midspan 22.3 leN 78.3 leN 89.0 leN
The relevant longitudinal and transverse truck positions for
maximum flexure in an interior beam at midspan are shown I 10.95 m 4.3 m 4.3 m 10.95 m I

I , -,

-
in Figs. 7 and 8, respectively. ~

The elastic stiffness of a displacement spring, k, at a strip ...

location is the ratio of applied load, P, to the vertical displace- -'l ABC J,J
ment, d. For a simply supported bridge, the following expres-
sion can be used:
FIG. 9. Spring Reactions on Isolated Beam 2 for Case of Flex-
~
k_ !:. _ 3Eb l b L
(1)
ur. at Midspan
- A - x 2(L - xi
since the torsional resistance of the steel I-beam is negligibly
where E b and I b = modulus of elasticity and moment of inertia small. Alternatively, the combined effect of the concrete slab
of the transformed composite beam, respectively; L = length and steel beam can be considered based on the method sug-
of the bridge; and x = distance of the strip from the left sup- gested by Heins and Kuo (1972). The rotational constant at
Downloaded from ascelibrary.org by New York University on 05/16/15. Copyright ASCE. For personal use only; all rights reserved.

port. Substitution of E b = 200 GPa (29,000 ksi), I b = 1.91 X strip B is infinitely large since the strip is located at an inter-
106 mm4 (45,900 in. 4 ), L = 30.5 m (100 ft), and x = 11.0 m mediate diaphragm. The rotational constants at strips A and C
(36 ft) in (1) results in k at strip A (as specified in Fig. 7) using (2) are both equal to 39,200 leN· rnIrad (28,900 kip-ftl
equal 7.62 kN/mm (43.5 kip/in.). Similarly, the k-values at the rad) due to symmetry.
locations of strips B and C are equal to 6.48 and 7.62 kN/mm Structural analysis of the strips on elastic supports resulted
(37.0 and 43.5 kip/in.), respectively. in the spring reactions at beam 2 equal to 22.3, 78.3, and 89.0
The stiffness of a rotational spring, c, is the ratio of the kN (5.01, 17.6, and 20.0 kips) at the location of strips A, B,
applied torque, T, to the twist, e. For an isolated strip between and C, respectively. Beam 2 is then isolated with the spring
cross-frame diaphragms, c is given by reactions applied at the respective strip locations, as shown in
Fig. 9. The maximum moment in the beam at midspan is equal
T GJd
c=-=
6 y(d - y)
(2) to 1,230 kN . m (903 kip' ft). Finally, the wheel load GDF for
flexure at midspan, (GDF)M' can be computed as the ratio of
where G = shear modulus of the transformed section; d = the moment in the beam to one-half the moment due to an
HS20 truck, that is
diaphragm spacing; J = torsional moment of inertia of the
composite transformed section; and y = distance between the
M max 1,230 leN· m = 1.19
strip location and closest diaphragm from the left. For flexible (GDF)M = (1/2)M = (1/2)(2,090 leN' m)
HS20
diaphragms (e.g., those made from channels), it is suggested
that (2) be used with consideration of end diaphragms only. The foregoing value is about 17% smaller than the one ob-
The transformed torsional moment of inertia of the composite tained using the AASHTO standard specification (Standard
beam, J, is equal to 95,800 mm 4 (2,301 in. 4 ). The value of J 1996) and given as follows:
is based on only the concrete portion of the composite beam,
S 2.40 m
35 leN 145 leN 145 leN G D F = - = - - = 1.43 (3)
1.68 1.68
where S = beam spacing (m).

GDF for Maximum Shear at the Support

iiiiii;.iiiiiiii.Abutment The longitudinal truck position for maximum shear at the

WdI- -.; .10;. ; .9. . ; .5. ; :;m~r

support is shown in Fig. 10. The transverse position of two
side-by-side trucks is as in Fig. 8.
The stiffness of the displacement springs at strips Band C
10.95 m are computed using (1) and are equal to 28.0 and 9.98 kN/
mm (160 and 57 kip/in.), respectively. For strip A, it is ana-
FIG. 7. Longitudinal Truck Location for Flexur. at Midspan lyzed as a beam on rigid supports with no rotation being al-
lowed at the girder positions, since it is located at an end
3.6m 3.6m 1.8m support and close to a diaphragm. The use of (2) to evaluate
Lane 1 Lane 2 the rotational spring constants at strips Band C results in c

o o being equal to 39,200 and 91,600 kN· rnIrad (28,900 and

145 leN 145 leN 35 leN

--l!ii·Biiiic·iiiiiiiiiiiiir-~t
14-~f?-:.==:""="":::":"'':'''':':''::'--':'''':'::'''':'':'':'_~~ 1.2 m
21.9m
FIG. 8. Transverse Truck Position for Maximum Load Effect on
Beam 2 FIG. 10. Longitudinal Truck Location for Shear at Support

JOURNAL OF BRIDGE ENGINEERING / FEBRUARY 1997/13

J. Bridge Eng. 1997.2:11-17.

 120 kN 108 kN 24.9 kN TABLE 1. Comparison of GDF Results

~.. 21_.9_m • Load effect

Approximate
method
OHBDC
method
AASHTO's
LRFD method

~~ C
FIG. 11. Spring Reactlone on leolated Beam 2 for Caee of
Shear at Support
1 (1)
Moment
Shear

AASHTO's LRFD Code

(2)
1.19
1.56
(3)
1.06
1.41
1.22
1.63
(4)

The method included in the AASHTO LRFD code (1994)

67,500 kip-ft/rad), respectively. Structural analysis of the strips is based on the results of National Cooperative Highway Re-
subjected to the wheel loads results in spring reactions at beam search Program (NCHRP) 12-26 study (Zokaie et al. 1991). It
2 at the location of strips A, B, and C equal to 120, 108, and requires the computation of the longitudinal stiffness param-
24.9 kN (26.9, 24.3, and 5.6 kip), respectively. Beam 2 is then eter, K, = n(I + Ae;), where n is the modular ratio between
Downloaded from ascelibrary.org by New York University on 05/16/15. Copyright ASCE. For personal use only; all rights reserved.

isolated with the spring reactions applied at the respective strip the stringer and slab, I and A are the moment of inertia and
locations, as shown in Fig. 11. The resulting maximum shear area of the stringer. respectively, and e, is the distance between
in the beam at the support is equal to 230 kN (51.8 kips). The the centroid of the stringer and middepth of the concrete slab.
(GDF)y for shear can then be computed as the ratio of the
For the steel stringer under consideration, n = 8, 1= 8.45 X
beam shear to one-half the shear due to the HS20 truck, that 10- 7 m4 (20,300 in. 4 ), A = 0.0570 m 2 (88.3 in. 2), and e, =
is 0.594 m (23.4 in.). Therefore, the wheel load GDF for flexure
in interior I-beam designed for two or more lanes can be eval-
Vmax 230 kN uated from
(GDF)y = (1/2)VHs20 = (1/2)(295 kN) = 1.56
0.6 ( )0.2 (K )0.1]
Compared to the AASHTO specifications (Standard 1996), the
(GDF)M =2 [ 0.075 + ( 2~9 ) ~ Lt8~ (6)
foregoing GDF value is about 1.3% larger. The distribution
which upon substitution of S = 2.40 m (8 ft); L = 30.5 m (100
factor for shear in the standard specifications is computed as ft); K, = 0.228 m4 (54,800 in. 4 ); and ts = 0.20 m (8 in.) results
an equivalent GDF because AASHTO requires the distribution in a GDF value equal to 1.22. The wheel load GDF for shear
of axles at the end support to be based on the slab being simply
can be calculated in term of the girder spacing, S, as follows:
supported at the location of the beams, while the distribution

(3~6) - C~.7) 2]
of other axles is treated in the same way as for flexure.
(GDF)y = 2 [0.2 + (7)
COMPARISON WITH OHBDC AND AASHTO'S
LRFDCODES Substituting S = 2.40 m (8 ft) in (7) results in the GDF for
shear equal to 1.63. Multiple presence factors are incorporated
The GDFs obtained using the approximate procedure for in the distribution factors shown in (6) and (7).
the composite steel bridge are now compared with the corre- A summary of the results of the approximate method,
sponding values from the OHBDC code (Bridge 1992) and OHBDC code, and AASHTO's LRFD specifications is pre-
AASHTO's load and resistance factor design (LRFD) (1994) sented in Table 1. The results indicate that the proposed ap-
code. The former method is based on the semicontinuum pro- proximate method yields slightly higher values than the
cedure (Bakht and Jaeger 1985; Jaeger and Bakht 1989), OHBDC code, about 11 % maximum difference. Very close
whereas the latter is based on the results of finite-element anal- results are obtained when the approximate method is compared
yses (Zokaie et al. 1991). Both are believed to yield more to the AASHTO LRFD code, with a maximum difference be-
accurate results than the standard AASHTO approximate GDF ing less than 5%.
values (Standard 1996).
VERIFICATION OF METHOD
OHBDCCode In this section, the analysis of several composite steel beam
The wheel load GDF for flexure using the Ontario Highway bridges with unequal girder spacings using the approximate
Bridge Design Code (OHBDC)s method can be computed method is verified by the use of a detailed finite-element study.
from The finite-element analysis is based on three-dimensional
modeling of the bridge superstructure using the computer pro-
S S gram ANSYS (ANSYS 1994).
(GDF)M = - = (4) The finite-element method involves subdividing the bridge
D
d
D (1 + ~~) continuum into a number of elements, interconnected at the
nodes. These discrete finite elements, representing various
where D = 2.2 - 4.01L = 2.07; Cf = 10 - 251L = 9.18 for components of the bridge, are idealized so as to correctly rep-
class C highway and two-lane bridge; fJ.. = 1.0; and S = 2.40 resent the behavior of the structure. Displacement functions
m. Substitution of D, Cf , and fJ.. in (4) results in D d = 2.26 m. are selected to define the state of displacement within each
Thus, the GDF = SIDd = 2.4012.26 = 1.06. For the case of finite element, as a function of the nodal displacements. Stress-
shear in interior girders subjected to two lanes of loading, the strain relationships are then defined for each material type and
GDF for an interior girder in the OHBDC code is based on the stiffness matrices for the elements are then formulated. The
global stiffness matrix for the structure is assembled and the
S equilibrium equations are solved for the nodal displacements,
(GDF)y = 1.70 (5) strains, and stresses. Approximate values of the displacements,
strains, and stresses within the elements can also be obtained.
Substitution of S = 2.40 m in (5) yields a GDF value equal to Three elements are used to model the geometry of each steel
1.41. stringer. The top and bottom steel flanges are modeled using
14/ JOURNAL OF BRIDGE ENGINEERING / FEBRUARY 1997

J. Bridge Eng. 1997.2:11-17.

 two-node, three-dimensional beam elements with six degrees
of freedom at each node. The geometric and stiffness prop- F M(%)
erties of the beam elements are lumped at the centroid of the !=-+-- (9)
flanges. The steel web of the bridge stringers is modeled by Af If
four-node rectangular shell elements. Each node has six de- where F = axial load in the bottom flange; M = bending mo-
grees of freedom. The shell elements considered have b~th ment in the bottom flange about a transverse axis in the hor-
membrane and bending stiffnesses and they account for 10- izontal plane; Af = cross-sectional area of the bottom flange;
plane and out-of-plane bending. The deck slab is also.m~eled If = moment of inertia of the bottom flange; and tf = thickn~ss
using thin four-node rectangular shell elements, sIDular to of the bottom flange. Further simplification for the expressIOn
those used to model the web of the steel stringers. The dia- of the stress in the bottom flange can be obtained if the second
phragms consist of cross frames made from an.gl~s and m:e term in (9) is neglected, with less than 3% loss of accuracy in
modeled by three-dimensional beam elements. RIgId three-dI- the stress.
mensional beam elements are used to connect the centroids of The GDF for an interior beam j in shear at an exterior sup-
the top flange steel beam elements to the centroids of the deck port can be determined in a similar fashion to (8)
slab elements directly above them. These elements are used in
Downloaded from ascelibrary.org by New York University on 05/16/15. Copyright ASCE. For personal use only; all rights reserved.

order to satisfy composite behavior and prevent relative de- GDF = 2NyrJ
formation between the top flange of the stringers and the con- (10)
•
crete slab. Fig. 12 shows a sketch of the finite-element model
of a single bridge girder. Verification and other details of the
2:'-I r,
finite-element bridge model are included in the work of Sa-
where rJ = support reaction of beam under consideration; and
hajwani (1995).
The reason behind modeling the steel beam top and bottom r, = support reaction of beam i.
flanges by three-dimensional beam elements, instead of shell Finite-element analysis of the 30.5-m-long (100 ft) com-
elements, is to simplify the computation of the GDF for flex- posite steel bridge in Fig. 8 showed that the GDF for flexure
at midspan and shear at the support are equal to 1.18 and 1.50,
ure. This modeling approach is based on the work of Bishara
respectively. A summary of the bottom flange stresses at
et al. (1993). Under linearly elastic conditions, stresses and
midspan and support reactions from ANSYS are shown in Ta-
bending moments are proportional to each other in the beams.
ble 2.
Therefore, instead of working with moments in the composite
Thirty-one different composite steel beam bridges are ana-
I-beams, maximum stresses at the extreme fiber of the bottom lyzed by both the approximate method and the finite-element
flanges can easily be used to compute the GDF, as follows: procedure to verify the accuracy of the proposed method. Sev-
enteen of the analyzed bridges are simply supported with span
2NYh length equal to either 15.3 or 30.5 m (50 or 100ft). The con-
GDF=-.- (8)
tinuous bridges have two equal spans of 22.9 m (75 ft). Dia-
2: It
/-1
phragms, in the form of cross bracing made from L89 X 89
X 9.5 angles, are located at 7.60-m (25 ft) intervals. The im-
portant parameters addressed in the analysis include the span
where !J = extreme bottom flange stress of the beam under length, continuity, number and spacing of the beams, bridge
consideration; It = extreme bottom flange stress of beam i; n width, and beam stiffness. The cross sections of the considered
= number of beams; 'Y = AASHTO's reduction factor for mul- simply supported bridges are defined in Table 3, based on the
tiple presence (Standard 1996), equal to 1.0 for one or two layouts shown in Fig. 13. The geometry of the continuous
trucks, 0.9 for three trucks, and 0.75 for four or more trucks; bridges is presented in Table 4. Note that cases 16, 17, 30,
and N = number of trucks on the bridge. and 31 involve steel beams with different wide-flange sections;
The extreme fiber stresses in the beams can be obtained larger sections are used on heavily loaded beams.
from internal axial and shear forces, bending moments about The applied live load on the considered bridges consists of
two perpendicular axes, and torque. If the effects of shear, one or more side-by-side HS20-44 trucks. For the case of mul-
torsion, >and bending about a vertical axis on the maximum tiple lanes of loading, appropriate reduction factors are con-
bottom flange stress in the beams is neglected, then the stress, sidered, in accordance with the AASHTO standard specifica-
/. may be computed from tions (Standard 1996). The GDF values are computed for
interior beams 2 or 3, or both. The critical longitudinal truck
positions for flexure and shear in the simply supported bridges
Yt! x
are at the midspan and end support, respectively. For the con-
tinuous bridges, the maximum positive moment effect is de-
termined at a distance equal to 10.1 (33 ft) from the left sup-
port, whereas the maximum negative moment is determined at
the location of the intermediate support. Shear in the contin-
uous bridges is computed at the exterior support.
Examination of the GDF values obtained using the approx-
Web
(Shell Element) TABLE 2. Finite-Element Results
Stress In bottom
flange at midspan Support reactIon
~~==p./"'---Bottom Flange Beam number (MPa) (kN)
... (Beam Element) (1 ) (2) (3)
1 51.6 153
Cross-Frame Bracing 2 46.9 206
(Beam Elements) 3 37.3 159
4 23.8 30.8
FIG. 12. Summary of Finite-Element Modeling

JOURNAL OF BRIDGE ENGINEERING / FEBRUARY 1997/15

J. Bridge Eng. 1997.2:11-17.

 TABLE 3. Cro.. sections for Simply Supported Bridges
Span S:' S1 ~ S, S4 SF!
Case (m) W-section (m) (m) (m) (m) (m) (m)
(1 ) (2) (3) (4) (5) (6) (7) (8) (9)
1 30.5 W920 x 446 1.2 2.4 2.4 2.4 - 1.2
2 30.5 W920 x 446 0.9 1.8 3.0 3.0 - 0.9
3 30.5 W920 x 446 0.9 3.0 1.8 3.0 - 0.9
4 15.3 W920 x 223 1.2 2.4 2.4 2.4 - 1.2
5 15.3 W920 x 223 0.9 1.8 3.0 3.0 - 0.9
6 15.3 W920 x 223 0.9 3.0 1.8 3.0 - 0.9
7 30.5 W920 x 446 1.2 2.4 2.4 2.4 2.4 1.2
8 30.5 W920 x 653 0.9 1.8 1.8 3.0 3.0 1.5
9 30.5 W920 x 653 0.6 1.2 1.2 3.6 3.6 1.8
10 30.5 W920 x 488 1.2 3.0 1.8 1.8 3.0 1.2
11 30.5 W920 x 488 1.2 3.6 1.2 1.2 3.6 1.2
12 30.5 W920 x 488 0.9 1.8 3.0 1.8 3.0 1.5
Downloaded from ascelibrary.org by New York University on 05/16/15. Copyright ASCE. For personal use only; all rights reserved.

13 30.5 W920 x 653 0.6 1.2 3.6 1.2 3.6 1.8

14 30.5 W920 x 653 1.2 1.8 3.0 3.0 1.8 1.2
15 30.5 W920 x 653 1.2 1.2 3.6 3.6 1.2 1.2
16 30.5 Variable 0.9 1.8 1.8 3.0 3.0 1.5
17 30.5 Variable 0.6 1.2 1.2 3.6 3.6 1.8

300mm 1.2 -r-----------------,.

200mm 1~
• • • •
• • • •• • •••• •••
•.......J~4-:--
•
•
••
• • •

0.4

0.2 • Sit_lit Support

(a) O+-+-+-+-+-HHI-I-+--+--+-+-+-+-+-+-f
300mm o 1 2 J 4 5 6 7 8 9 10111213 14 15 16 17
200mm 1~

~ 0~eawn
BrldgeCase
FIG. 14. Difference between Approximate and ANSYS Results
for Simple Bridges

imate method and ANSYS indicates that the approximate

method gives results comparable to the finite-element pro-
gram. In most cases, the approximate method results in slightly
conservative values of the GDF for both shear and moment.
(b) For the 17 simply supported bridges, the differences between
the GDF values obtained from the approximate and ANSYS
FIG. 13. Cross Sections of Bridges Considered In the Study analyses are in the range of +18 to -3.7%, positive means
with: <a) Four Beams; (b) Five Beams that the approximate method is conservative. The average dif-
ference between the two methods for flexure and shear is equal
TABLE 4. Cross Sections for Continuous Bridges to +5.7 and +3.1%, respectively. The corresponding coeffi-
Case Same layout as for case cients of variation for flexure and shear are equal to 0.058 and
(1 ) (2) 0.030, respectively. Fig. 14 shows a graphical illustration of
18 1 the comparison of the results.
19 2 As for the simply supported bridge cases, the results indi-
20 3 cate that the approximate method can be conservatively used
21 7 to determine the GDF for irregular continuous bridges. For
22 8 flexure, the approximate method results in GDF values that
23 9 are about +2.8 to +25% more conservative than ANSYS. The
24 10
25 11 results for shear show that the difference in the GDF between
26 12 the approximate and finite-element method varies between
27 13 -3.6 and +11 %. On average, the approximate method yielded
28 14 results that were +13 and +7.6% more conservative than
29 15 ANSYS for positive and negative bending moments, respec-
30 16 tively. The corresponding coefficients of variation for positive
31 17
and negative moments were 0.051 and 0.046%, respectively.
16/ JOURNAL OF BRIDGE ENGINEERING / FEBRUARY 1997

J. Bridge Eng. 1997.2:11-17.

 1.4 finite-element results for flexure and shear, respectively. For
continuous bridges, the approximate method resulted in an av-
1.2 • •• • •
•• . I · •.• • , . . • I•
erage of 13 and 7.6% more conservative GDF values than the

~i~~
finite-element analysis for positive and negative flexure, re-
1 -~--~~--~-~~~-~~- spectively. For shear at the end support of continuous bridges,
the approximate method yielded an average difference from
0.8
the finite-element analysis of only 1.9%.
Based on the foregoing, it can be concluded that the ap-
0.6
proximate method of analysis discussed in this study can be
0.4
safely used to analyze irregular I-beam bridges with unequal
• PrMItI!¥M.-t
girder spacings if a more detailed finite-element analysis is not
·N.....M.-t
0.2 available.
• S1t_

o+-+--+-+---I-+--+--+--+--+--+---l--+--l ACKNOWLEDGMENTS
Downloaded from ascelibrary.org by New York University on 05/16/15. Copyright ASCE. For personal use only; all rights reserved.

18 19 20 21 22 29 U 2S 26 27 28 29 90 91 This research project was partially supported by the University of

Houston under the Research Initiation Grant (RIG) program. This finan-
cial support is greatly appreciated by the writers.
BrldgeCose
FIG. 15. Difference between Approximate and ANSYS Results APPENDIX. REFERENCES
for Continuous Bridges ANSYS user's manual. (1994). Swanson Analysis Systems, Rev 5.0.
Houston, Pa.
For shear, the approximate method resulted in an average dif- Bakht. B., and Jaeger, L. (1985). Bridge analysis simplified. McGraw-
ference in the GDF value from ANSYS of about +1.9%, with Hill Book Co., Inc.• New York, N.Y.
a coefficient of variation equal to 0.036. A graphical presen- Bishara, A. G., Liu, M. C.• and EI Ali, N. D. (1993). "Wheel load dis-
tribution on simply supported skew I-beam composite bridges." J.
tation of the results for the continuous bridges is shown in Struct. Engrg., ASCE. 119(2),399-419.
Fig. 15. Bridge code, 3rd Ed. (1992). Ministry of Transp.• Quality and Standards
Div.• Ont., Canada.
SUMMARY AND CONCLUSIONS Heins, C. P., and Kuo, J. (1972). "Torsional properties of composite
girders." Engrg. J., 9(2), 79-85.
Thirty-one different composite steel I-beam bridges having Highway structures design handbook. Vol. I. (1982). AISC Marketing.
irregular beam spacings are analyzed in this study under the Inc.• Chicago, Ill.
effect of live load. The bridges are analyzed by two methods, Jaeger, L.. and Bakht. B. (1989). Bridge analysis by microcomputer. Mc-
an approximate procedure and the finite-element method. Graw-Hill Book Co.• Inc., New York. N.Y.
Khaleel, M. A., and Itani, R. Y. (1990). "Live load moments for contin-
Wheel load GDFs are computed for both flexure and shear in uous skew bridges." J. Struct. Engrg., ASCE, 116(9), 2361-2373.
interior girders. The applied loading consisted of two or more LRFD bridge design specifications-Sf units, 1st Ed. (1994). Washing-
side-by-side HS20-44 trucks. The approximate method is ton, D.C.
based on isolating strips of the deck slab in the transverse Marx, H. J., Khachaturian, N., and Gamble, W. L. (1991). "Design cri-
direction directly under the wheel loads and treating them as teria for right and skew-slab and girder bridges." Transp. Res. Rec.
beams on elastic supports. The finite-element analysis was No. 1319, Transp. Res. Board, Nat. Res. Council, Washington, D.C.,
72-75.
used to verify the accuracy of the approximate method. The Sahajwani, K. (1995). "Analysis of composite steel bridges with unequal
parameters investigated included the span length, continuity, girder spacings," MS thesis, Dept. of Civ. and Envir. Engrg., Univ. of
number and spacing of girders, bridge width, girder stiffness, Houston, Tex.
and number of loaded lanes. Standard specifications for highway bridges, 16th Ed. (1996). Am. Assn.
The results of this study indicate that GDFs obtained by the of State Hwy. and Transp. Officials (AASHTO), Washington, D.C.
approximate method for flexure and shear in simply supported Tarhini, K. M., and Frederick, G. R. (1992). "Wheel load distribution in
and continuous I-beams are in agreement with results obtained I-girder highway bridges." J. Struct. Engrg., ASCE, 118(5), 1285-
1294.
using a detailed three-dimensional finite-element analysis. On Zokaie, T., Imbsen, R. A., and Osterkemp, T. A. (1991). "Distribution
average, the approximate method yielded GDF values for sim- of wheel loads on highway bridges." NCHRP Rep. No. 89, TRB Publ.
ple bridges that were 5.7 and 3.1 % more conservative than the No. 1290, Vol. I, Nat. Res. Council, Washington, D.C., 119-126.

JOURNAL OF BRIDGE ENGINEERING / FEBRUARY 1997/17

J. Bridge Eng. 1997.2:11-17.