You are on page 1of 16

ANALYSIS OF SOME SKEW BRIDGES AS EIGHT BRIDGES

By Baidar Bakiit1
Downloaded from ascelibrary.org by Motilal Nehru National Institute of Technology on 01/25/19. Copyright ASCE. For personal use only; all rights reserved.

ABSTRACT: Those methods of bridge analysis that are developed basi-


cally for right bridges are also sometimes used for analyzing (skew
bridges provided that the angle of skew is less than 20°. A critical review
of this practice is presented in this paper, which also contains a
recommended procedure for obtaining longitudinal moments with good
accuracy in skew slab-on-girder bridges by this practice. It is shown that
the errors in analyzing skew slab-on-girder bridges as right are not
characterized by the angle of skew but by two dimensionless parame-
ters, which depend upon the angle of skew, the spacing and span of
girders, and their flexural rigidities relative to theflexuralrigidity of the
deck slab. Recommendations are given for the use of the simplified
methods of analysis for skew slab-on-girder bridges. It is proposed that
bridges having (S tan cj>/Z,) less than 0.05 can be analyzed as equivalent
right bridges, where S, L, and (j> are the girder spacing, bridge span, and
angle of skew, respectively.

INTRODUCTION

Many rigorous methods of bridge analysis have been developed basi-


cally for right bridges (i.e., for bridges of zero angle of skew). Adapting
them to the analysis of skew bridges is quite tedious and often difficult.
Examples of such rigorous methods are: the orthotropic plate method
(Cusens and Pama 1975); the finite strip method (Cheung 1975); and the
semicontinuum method (Jaeger and Bakht 1985a). When the skew angle of
a bridge is small (e.g., less than 20°) it is frequently considered safe to
ignore the angle of skew, and analyze the bridge as a right bridge whose
span is equal to the skew span, as shown in Fig. 1. ,
The procedure just outlined is considered permissible on the supposition
that it gives conservatively safe estimates of longitudinal moments and
shears in the skew bridge. Even when this supposition is valid, it becomes
difficult for the engineer to establish the degree of conservatism involved.
Some codes [e.g., Ontario Highway Bridge Design Code (OHBDC) (1983)]
do not permit the above procedure for large angles of skew. A limit on the
angle of skew implies that the differences between the responses of a skew
bridge and the equivalent right bridge depend solely upon the angle of
skew. It is shown in this paper that other factors also influence the values
of these differences.
This paper also contains a very simple procedure by using which even
bridges with large angles of skew can be analyzed to acceptable design
accuracy by means of methods originally developed for the analysis of
'Prin. Res. Engr., Ministry of Transp. and Communications, Downsview,
Ontario, Canada.
Note. Discussion open until March 1, 1989. 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 possible publication on
March 26, 1987. This paper is part of the Journal of Structural Engineering, Vol. 114,
No. 10, October, 1988. ©ASCE, ISSN 0733-9445/88/0010-2307/$1.00 + $.15 per
page. Paper No. 22861.
2307

J. Struct. Eng., 1988, 114(10): 2307-2322


Downloaded from ascelibrary.org by Motilal Nehru National Institute of Technology on 01/25/19. Copyright ASCE. For personal use only; all rights reserved.

FIG. 1. Usual Method of Analyzing Skew Bridges as Right Bridges: (a) Plan of
Skew Bridge; (b) Plan of Equivalent Right Bridge

right bridges. This simple procedure is intended mainly for the determina-
tion of longitudinal moments in slab-on-girder bridges.

MECHANICS OF LOAD DISTRIBUTION

Before developing the procedure, it is instructive to study the mechanics


of load distribution in skew bridges. A slab-on-girder bridge with negligible
torsional rigidity is chosen as a vehicle for the study. With reference to
Figs. 2(a) and 2(b), it can be readily shown that an increase in the skew
angle of the bridge can affect the pattern of load distribution due. to two
unrelated factors: (1) The change in the relative longitudinal positions of
loads; and (2) the change in the effective flexural rigidities of adjacent
beams. Separate discussions of these two factors follow.

Change in Relative Longitudinal Load Positions


Fig. 2(a) shows the plan of a skew bridge with four girders that carries
four concentrated loads. The loads, which represent a highway truck, have
an orthogonal pattern on the plan. For the sake of convenience it is

FIG. 2. Load Positions on Plan of Right and Skew Bridges: (a) Skew Bridge; (b)
Right Bridge with Orthogonal Pattern of Loads; (e) Right Bridge with Skew Pattern
of Loads

2308

J. Struct. Eng., 1988, 114(10): 2307-2322


assumed that each of the two longitudinal lines of wheels of the truck lies
directly above a a girder.
Downloaded from ascelibrary.org by Motilal Nehru National Institute of Technology on 01/25/19. Copyright ASCE. For personal use only; all rights reserved.

Keeping the girder span and spacings constant, the plan of the skew
bridge is conceptually deformed so that it becomes right, as shown in Fig.
2(b). The same four loads, as applied to the skew bridge of Fig. 2(a), are
applied to the equivalent right bridge by maintaining their longitudinal
positions on only the left-hand outer girder. An examination of Figs. 2(a)
and 2(b) will readily show that this procedure changes the relative
longitudinal load positions on the second girder.
By contrast to the procedure that led to Fig. 2(b), the bridge plan of the
skew bridge is now conceptually deformed along with the loads. As shown
in Fig. 2(c), the bridge is again made right but in this case the load pattern
on the plan becomes skew. It can be seen that the relative longitudinal
positions of the loads are the same in this equivalent right bridge of Fig.
2(c) as they were in the skew bridge of Fig. 2(a).
If the transverse load distribution characteristics of the bridge were so
poor that the loads would be retained almost entirely by the loaded girders,
the girder moments and shears in the equivalent right bridge of Fig. 2(c)
would be very nearly the same as those for the skew bridge. The same
cannot be said about the equivalent right bridge of Fig. 2(b), for which the
change of longitudinal load positions of one line of wheels would signifi-
cantly affect the responses of the second girder. It can therefore be
concluded that the approximate equivalence represented by Fig. 2(c) is far
superior to that of Fig. 2(b),

Change in Effective Flexural Rigidities


When the transverse load distribution characteristics of the bridge are
not very poor, even the better simplification of Fig. 2(c) may lead to
erroneous estimates of the girder responses. A discussion of this aspect of
skew bridge behavior follows with the help of a simplistic model.
Consider a four-girder skew bridge carrying a concentrated load at the
midspan of the left-hand outer girder. As shown in Fig. 3(a), the bridge is
idealized as a grid of four torsionless longitudinal beams and one trans-
verse beam. The transverse beam, which passes through the load and is at
right angles to the longitudinal beams, represents the deck slab area shown
shaded in this figure. The plan of the skew grid is now conceptually
deformed to become right, as shown in Fig. 3(b). Clearly, this transposition
does not alter the longitudinal position of the load on the directly loaded
beam. Therefore, any changes in the girder responses corresponding to the
two idealizations should result only from changes in the load distribution
patterns of the two bridges.
The transfer of loads from the loaded girder to adjacent girder is
influenced by the relative flexural rigidities of the adjacent girders. For the
grids under consideration, the load transfer between the longitudinal
beams can take place only through the transverse beam. As shown in Fig.
3(c), the transverse beam intersects the longitudinal beams of the right grid
at different locations than it does in the skew grid. The result of relocating
the points-of-load transfer in the equivalent right grid is to change the
deflections in the adjacent beams at the points-of-load transfer, and thus to
alter their effective flexural rigidities. Consequently, the fraction of the
applied load accepted by the directly loaded longitudinal beam of the skew

2309

J. Struct. Eng., 1988, 114(10): 2307-2322


Girder No.
*1
Downloaded from ascelibrary.org by Motilal Nehru National Institute of Technology on 01/25/19. Copyright ASCE. For personal use only; all rights reserved.

—»-e—
—«H s H*—

-» o
K—• 0 a4
(c)

® Point of intersection with the


transverse beam in the
, — A A A skew bridge
Girder No. 1
'Concentrated 1 2 3 4
© Point of intersection with
load
(a) (b) the transverse beam
in the right bridge

FIG. 3. Simplistic Models to Study Load Retention in Loaded Girder: (a) Plan of
Skew Grillage; (b) Plan of Equivalent Right Grillage; (c) Girders Considered
Separately

grid is not the same as that accepted by its counterpart in the equivalent
grid.
It can be readily appreciated that the relocation of the points of load
transfer, defined by the distance s shown in Fig. 3(c), depends not only on
the angle of skew but also on the spacing of the longitudinal beams. It can
thus be seen that the skew angle is not the only factor that must be taken
into account while analyzing a skew bridge as right.
Another factor which must be considered is the flexural rigidity of the
transverse beam relative to the flexural rigidities of the longitudinal beams.
In the context of real-life slab-on-girder bridges, the transverse flexural
rigidity of the bridge deck becomes this factor that influences the simpli-
fication.

QUANTIFICATION OF LOAD DISTRIBUTION

The method of analysis of Jaeger et al. (1988) is used as a means for


quantifying differences in girder responses of skew and equivalent right
bridges. This method deals with the load distribution analysis of skew
slab-on-girder bridges with three girders of negligible torsional stiffness; it
is based on idealizing the bridge as an assembly of discrete longitudinal
beams and a continuous transverse medium, and the representation of
loads by harmonic series.
The general method (Jaeger and Bakht 1985a) is referred to as the
semicontinuum method of analysis. It is shown by Jaeger et al. (1988) that
load distribution in three-girder skew bridges depends upon two dimen-
sionless characterizing parameters, in and e, which are denned as follows:
12 L^ LD}
1T) =
Jrmsf ~EI (1)

2S tan $
(2)
' I
2310

J. Struct. Eng., 1988, 114(10): 2307-2322


where m = harmonic number under consideration; L = girder span; S =
girder spacing; Dy = transverse flexural rigidity per unit length; El =
flexural rigidity of a girder; and <f> = angle of skew.
Downloaded from ascelibrary.org by Motilal Nehru National Institute of Technology on 01/25/19. Copyright ASCE. For personal use only; all rights reserved.

It was shown earlier that the two factors responsible for differences in
girder responses of skew and the equivalent right bridges are the quantity
s and the flexural rigidity of the deck slab with relative to the flexural
rigidity of the girders. It is reassuring to note that these two factors are
implicit in the expressions for the two characterizing parameters.
The degree of error incurred by analyzing a skew bridge as right can
conveniently be considered in terms of a factor A, which is defined by:

(Pn,n)e = 0 ,-..
A = • / s. (3)
(Pn,«)e

where p„„ is the distribution coefficient for girder number n due to a load
also on girder number n. The subscript to p„ „ outside the parentheses
refers to the value of e. As can be seen from Eq. 2, a right bridge (i.e., a
bridge with zero angle of skew) has E equal to zero. Values of A less than

FIG. 4. Contours of A for Harmonic No. 1

2311

J. Struct. Eng., 1988, 114(10): 2307-2322


1.0 indicate that the girder responses in the skew bridge are smaller than
their counterparts in the equivalent right bridge.
Using the method of Jaeger et al. (1988), the values of A were calculated
Downloaded from ascelibrary.org by Motilal Nehru National Institute of Technology on 01/25/19. Copyright ASCE. For personal use only; all rights reserved.

for three-girder bridges covering the entire practical range of e and t). Two
load cases were considered. In one load case, a single point load was
applied at the midspan of an outer girder, and in the other, a single point
load was applied at the midspan of the middle girder.
It was found that the values of A corresponding to second and higher
harmonics were very nearly equal to 1.0. This is understandable since, as
shown by Jaeger and Bakht (1985b), the loads represented by higher
harmonics are increasingly retained by the externally loaded girders, in
which case changes in the effective flexural rigidities of the various girders
have little effect on load distribution.
Contours of A corresponding to the first harmonic effects of the two load
cases are plotted on the ~t]-e space in Fig. 4. The space covering r\ of less
than about five is not of particular interest because it corresponds to
bridges with very small spans. It can be seen that in the rest of the Tq-e
space, a particular value of A can be maintained by the simultaneous
increase of-n and decrease of e. It is interesting to note that the value of A
is mostly less than 1.0, indicating that the process of analyzing a skew
bridge ajs an equivalent right bridge mostly leads to safe side errors when
only one of the girders is directly loaded.

500

400

300

200

100

FIG. 5. Range of t] for First Harmonic Plotted against Span Length


i&O I &>

J. Struct. Eng., 1988, 114(10): 2307-2322


Relationship between Span and 'n
As a step in the development of a procedure proposed later in this paper,
it is necessary to develop a relationship between r\ and the span length, L.
Downloaded from ascelibrary.org by Motilal Nehru National Institute of Technology on 01/25/19. Copyright ASCE. For personal use only; all rights reserved.

This follows.
Using the familiar orthotropic plate notation, for the first harmonic, Eq.
1 can be rewritten as:
12 (LV (Dy
"= 7 ^ ^ l (4)

where Dx , the longitudinal flexural rigidity per unit width, is obtained by


dividing the flexural rigidity, EI, of a girder by the girder spacing, S.
It has been shown by Bakht and Moses (1988) that for slab-on-girder
bridges in North America, the upper-bound values of Dx are related to the
span length, L, according to the following empirical equation:
D.v = 2,257L2 + 59,575L (5)
and the lower-bound values of Dx are given by:
Dx = 1.790L2 + 9250L (6)
where L is in meters and Dx in kilonewton meters. Using Eqs. 4, 5, and 6,
the upper- and lower-bound values of TI can be readily obtained for
different span lengths; these two bounds are plotted in Fig. 5 against the
span length for the first harmonic. The figure also shows the curve for the
median values of T\.
ERROR IN LONGITUDINAL MOMENTS AND SHEARS
Another factor 8 is now introduced to quantify errors incurred in the
estimation of longitudinal moments and shears in skew bridges by analyz-
ing them as equivalent right bridges; this factor is defined as:
(R)E-(R)E =0
8= x 100 (7)
00,-c
where R is the longitudinal response (i.e., bending moment or shear, and
the subscript outside the parentheses refers to the value of e used in the
analysis). Thus, 8 gives the percentage difference in the skew and right
bridge responses. A negative value of 8 indicates that the responses in the
equivalent right bridge are larger than their skew bridge counterparts.
The loading cases considered in the development of the charts given in
Fig. 4 are not representative of real-life situations in which the live loading
covers a considerable width of the bridge. To represent real-life situations
more closely, the previously mentioned three-girder bridge was loaded
with loads on both an outer and the middle girder. A large number of
bridges was analyzed again for this load case by the method of Jaeger et al.
(1988). As before, the bridges covered the practical range of T\ and s. From
the results of the analyses, values of 8 were calculated for maximum
moments and shears in the outer and middle girders. The following
observations can be made regarding these values of 8:
1. The values of 8 were negative for longitudinal moments in all cases,
indicating that the process of analyzing a skew bridge as equivalent right is
a safe one as far as longitudinal moments are concerned.
2313

J. Struct. Eng., 1988, 114(10): 2307-2322


Downloaded from ascelibrary.org by Motilal Nehru National Institute of Technology on 01/25/19. Copyright ASCE. For personal use only; all rights reserved.

FIG. 6. Contours of 8 for Moments in Interior Girder

2. For longitudinal shears, the value of 8 was nearly always positive,


with absolute values larger than the corresponding absolute values for
longitudinal moments. This reverse trend of errors shows that for obtaining
longitudinal shears in skew bridges the method of analyzing them as
equivalent bridges incurs unsafe errors.
3. The absolute values of 8 for the outer girder, for both moments and
shears, were always smaller than those for the middle girder.

These observations, as shown later, were also found to be valid for


bridges with more girders, which were analyzed by the grillage analogy
method.
The values of 8 for moments in the middle girder, as obtained by the
above analyses, were chosen as a vehicle for a qualitative assessment of
the degree of error in the determination of longitudinal moments obtained
by analyzing a skew bridge as right.
A value of 8 corresponding to a given set of values of iq and e, can be
readily transcribed on the L-S tan 4> space by using Eq. 2 and the median
curve of t] plotted against the span length in Fig. 5. The values of 8 for
moments in the middle girder are plotted on the L-S tan 4> space in Fig. 6
in the form of contours. The emerging trend indeed appeals to the
engineering judgement using which it could have been predicted—that the
influence of the angle of skew diminishes as the span length of the bridge
increases.
The value of 5 tan §, which is the same as the distance s shown in Fig.
2314

J. Struct. Eng., 1988, 114(10): 2307-2322


/ Values of S tan <t>
in meters
/
Downloaded from ascelibrary.org by Motilal Nehru National Institute of Technology on 01/25/19. Copyright ASCE. For personal use only; all rights reserved.

0 15 30 45 60
<j), degress

FIG. 7. Contours of S tan 4>

3(c), can be easily computed from known values of S and §; alternatively,


it can be read directly from the chart provided in Fig. 7.
Strictly speaking, the contours of 8 given in Fig. 6 apply to bridges with
three girders carrying loads on the middle and an outer girder. In a general
way, however, they are also applicable to other bridges (as verified later in
this paper).
Fig. 6 provides a ready means of assessing the degree of error in
obtaining maximum longitudinal moments by analyzing a skew bridge as
right, it being understood that:
1. A value of 8 obtained from this figure applies to an interior girder.
2. The value of 8 for an outer girder is also negative but with a smaller
absolute value.
3. These values of 8 do not apply to the maximum longitudinal shear, for
which 8 is nearly always positive with a larger absolute value than the
absolute values of maximum longitudinal moments.

NUMERICAL EXAMPLES

To assess quantitatively the effect of analyzing real-life skew bridges as


right, several skew bridges were analyzed both as they are and as
equivalent right bridges, by the grillage analogy method [e.g., Jaeger and
Bakht (1982)].
For the right bridges, the pattern of loads on the plan was made skew
according to the scheme shown in Fig. 2(c). All bridges had five girders
spaced at 1.8 m, and a skew span of 16.0 m. Each bridge was analyzed
under the American Association of State Highway and Transportation
Officials (AASHTO) HS 20 design vehicle (Standard specifications for
highway bridges 1984) placed in two different transverse positions. The
skew angle of these bridges varied from 30° to 60°. Girder rigidities and the
thickness of the concrete deck slab were kept the same in both cases, and
were as shown in Fig. 8.
Fig. 8 also contains the relevant details of analyses, and some of their
results, for the 30° skew bridge; it shows the positions of loads on the plan
2315

J. Struct. Eng., 1988, 114(10): 2307-2322


' LOAD CASE 1 (HS20)
I LOAD CASE 2 (HS 20)
Downloaded from ascelibrary.org by Motilal Nehru National Institute of Technology on 01/25/19. Copyright ASCE. For personal use only; all rights reserved.

O SKEW BRIDGE
D — RIGHT BRIDGE

= 0.1295 m
4

» 0.0003 m

S U B THICKNESS = 180 MM

LOAD CASE 2

LOAD CASE 1
80
* /*"'"""" """'"' hf
_/
1000 L-T 100 L

FIG. 8. Girder Moments at Midspan and Support Reactions in 30° Skew Bridge

of actual and equivalent right bridges; the comparisons of longitudinal


moments at midspan; and girder reactions at a support. It is noted that the
response longitudinal shear becomes the same as the support reaction at
simple supports.
In Fig. 8 and subsequent figures, the points representing the girder
responses are joined by continuous curves to facilitate visual comparison,
rather than to suggest that their distribution is continuous in the transverse
direction.
It can be seen in Fig. 8 that for the 30° skew bridge, the analysis with the
equivalent right bridge gives longitudinal moments that are very close to
the actual ones, the maximum difference, 8, being equal to about -2%.
The value of S tan 4> for the bridge under consideration is 1.05 m; for this
2316

J. Struct. Eng., 1988, 114(10): 2307-2322


value of 5 tan <$> and for L equal to 16.0 m, Fig. 6 gives 8 also equal to about
-2%. Such a close agreement between the two values of 8, although
fortuitous, is reassuring.
Downloaded from ascelibrary.org by Motilal Nehru National Institute of Technology on 01/25/19. Copyright ASCE. For personal use only; all rights reserved.

Also, as shown in Fig. 8, the maximum values of support reactions


obtained for the right bridge are smaller than the corresponding responses
in the skew bridge, the maximum difference being equal to about 13%.
These results corroborate the previous observation that the process of
determining longitudinal shear by treating the skew bridge as right leads to
unsafe errors.

• LOAD CASE 1 (HS 20)


• LOAD CASE 2 (HS 20)

GIRDER PROPERTIES AND SLAB


THICKNESESS, SAME AS SHOWN IN FIG. 8
GIRDER <£ (Typ)

LOAD CASE 2

1000L 100

FIG. 9. Girder Moments at Midspan and Support Reactions in 45° Skew Bridge
2317

J. Struct. Eng., 1988, 114(10): 2307-2322


Downloaded from ascelibrary.org by Motilal Nehru National Institute of Technology on 01/25/19. Copyright ASCE. For personal use only; all rights reserved.

16.0 m
p GIRDER MOMENTS AT MID-SPAN, kN-m
a
Q.

10 a>
-* a. -L

J. Struct. Eng., 1988, 114(10): 2307-2322


m

f
1 ransverss girder Transversa gir !@r Transverse cjirder
position position position
Downloaded from ascelibrary.org by Motilal Nehru National Institute of Technology on 01/25/19. Copyright ASCE. For personal use only; all rights reserved.

n Mi
fp
¥
I -
/
0
-
f/i
i

ij 45° SKEW

1000
i
I
FIG. 11. Girder Moments at Midspan in Bridges with Low Values of Longitudinal
Flexural Rigidities

Fig. 9 presents the same data as given in Fig. 8, except that it relates to
the bridge with a skew angle of 45°. In this case again, the longitudinal
moments due to the two idealizations are fairly close, with a 8 of about
—5%. The charts of Fig. 6 give a 8 of —4%, again confirming their validity.
For this case, 8 for support reactions is about 20%.
Results for the bridge with 60° skew are given in Fig. 10. In this case, 8
for maximum longitudinal moment as obtained by the grillage analogy is
about —12%, with the corresponding value obtained from Fig. 6 being
equal to about - 1 5 % . The gap between the support reactions obtained by
the two idealizations widens even further with 8 being equal to 22%.
The girder flexural rigidities used in these analyses were somewhat on
the higher side. These analyses were repeated by reducing the girder
rigidities to conform to the lower-bound values of Dx given by Eq. 6. It is
noted that these lower-bound values of longitudinal flexural rigidity are
rarely realized for bridges with such short spans as those of the bridges
under consideration. The results of the analyses are summarized in Fig. 11
for midspan longitudinal moments, for which 8 was found to be equal to
- 2 , - 1 0 , and - 2 7 % for bridge with angles of skew of 30°, 45°, and 60°,
respectively. These values of 8 confirm that the charts of Fig. 6 provide
conservatively safe estimates of errors in determining longitudinal mo-
ments by treating skew bridges as right.
It was found that the reduction of the longitudinal flexural rigidity also
leads to increases in the value of 8 for longitudinal shear.

APPLICATION OF SIMPLIFIED METHODS TO SKEW BRIDGES

AASHTO (Standard specifications 1984) and Ontario (Ontario highway


bridge design code 1983) codes permit the use of simplified methods of
2319

J. Struct. Eng., 1988, 114(10): 2307-2322


Li SPAN-16.0 m
Downloaded from ascelibrary.org by Motilal Nehru National Institute of Technology on 01/25/19. Copyright ASCE. For personal use only; all rights reserved.

1 1 1 I 1 TTTTT
External girder, load case 1

1.0 2.0
Stan^m

FIG. 12. Effect of S tan 4> on Maximum Girder Responses

analysis by which the live load analysis of a slab-on-girder bridge is


reduced to beam analysis. In this analysis, a girder and its associated
portion of the deck slab are subjected to loads consisting of one longitu-
dinal line of loads multiplied by a fraction (S/D). S is the girder spacing and
D is a quantity with units of length whose value is either prespecified for a
given bridge type {Standard specifications 1984) or obtained by a semi-
graphical method {Ontario highway bridge design code 1983). Strictly
speaking, these simplified methods are applicable to only right bridges in
which the different longitudinal lines of loads are similar (i.e., the loads
along a transverse line have the same intensity). For practical consider-
ations, however, the simplified methods are allowed to be used for skew
angles of up to 20°.
The procedure represented by Fig. 2(c) is meant for those rigorous
methods of analysis which can deal with loads anywhere on the bridge; it
is not applicable to those methods which require similar loads on different
longitudinal lines. As can be seen in Figs. 8-10, the proposed procedure
leads to nonsimilar loads on different longitudinal lines; therefore, this
procedure cannot be used in conjunction with methods of the D-type. The
effect of changes in the value of S tan <jj on girder moments and shears,
however, can still be studied by using the bridge with S tan <> j equal to zero
(i.e., a right bridge as the datum, it being assumed that the simplified
methods can give reliable results for this bridge).
Maximum longitudinal moments and shears obtained by grillage analogy
for the cases reported in Figs. 8-10 are plotted against S tan <> j in Fig. 12.
It can be seen that as the value of S tan 4> increases, maximum moments
for both external and internal girders decrease. The same trend can be
observed in Fig. 12 for maximum longitudinal shears in the external
girders. The trend, however, reverses for maximum longitudinal shear in
internal girders, the value of which increases with increase of S tan <j>.
These two trends can be easily rationalized from the physical aspects of
load distribution. Hence, it can be confidently concluded that the trends of
the results of the limited study reported in Fig. 12 are also applicable to
other bridges and other load cases.
2320

J. Struct. Eng., 1988, 114(10): 2307-2322


The reduction of the maximum girder response with the increase of S tan
4> indicates that the relevant D-type method will give increasingly conserv-
atively safe results as the skew angle increases. This is true for the use of
Downloaded from ascelibrary.org by Motilal Nehru National Institute of Technology on 01/25/19. Copyright ASCE. For personal use only; all rights reserved.

the D-type methods in the calculation of longitudinal moments in both


external and internal girders, and for the calculation of longitudinal shear
in only external girders. These methods are not recommended for calcu-
lating longitudinal shear in the internal girders of those bridges for which
the ratio (S tan cjj/L) exceeds about 0.056. For this ratio, the calculated
shear value is expected to be in unsafe error by about 5%. It is noted that
the ratio (S tan §IL), while not being an exact measure of the effects of the
skew angle, is closely so for the purpose under consideration.

CONCLUSIONS

The conclusions and recommended practice regarding the analysis of


skew bridges as right can be summarized as follows:

1. It is clear from the results presented that the ratio (S tan cb/Z.), rather
than <j>, should be taken as an appropriate measure of skewness.
2. It has been shown that when rigorous methods for right bridges are
used for analyzing skew bridges, the preferred rearrangement of loads on
the plan is as illustrated in Fig. 2(c).
3. When the simplification identified in the second conclusion is fol-
lowed, the resulting maximum moments are always smaller than actual
with the minimum degree of error, as shown in Fig. 6.
4. The simplification identified in the second conclusion is not recom-
mended for use in obtaining longitudinal shears because the error thus
involved is always on the unsafe side.
5. D values, corresponding to the simplified methods of analysis, can be
safely used for bridges of any skew angle to calculate longitudinal moments
in both external and internal girders and longitudinal shear in only external
girders. The values thus calculated will always be smaller than actual, with
the difference between the two increasing with increase in (S tan 4>/L).
6. The D-type method for calculating longitudinal shear in internal
girders of skew bridges can lead to unsafe errors; its use is recommended
for bridges with (S tan §IL) up to only 0.05.

ACKNOWLEDGMENTS

Contributions of L. G. Jaeger towards the philosophical development of


the work reported in this paper, and of J. Maheu in performing the grillage
analyses on which Figs. 8-11 are based, are gratefully acknowledged.

APPENDIX. REFERENCES

Bakht, B., and Moses, F. (1988). "Lateral distribution factors in highway bridges."
J. Struct. Engrg., ASCE, 114(8).
Cheung, Y. K. (1975). Finite strip methods in structural analysis. John Wiley and
Sons, Inc., London, England.
Cusens, A. R., and Pama, R. P. (1975). Bridge deck analysis. John Wiley and Sons,
Inc., London, England.
Jaeger, L. G., and Bakht, B. (1982). "The grillage analogy in bridge analysis." Can.
2321

J. Struct. Eng., 1988, 114(10): 2307-2322


J. Civ. Eng., 9(2), 224-235.
Jaeger, L. G., and Bakht, B. (1985a). "Bridge analysis by the semi-continuum
method." Can. J. Civ. Eng., 12(3), 573-582.
Downloaded from ascelibrary.org by Motilal Nehru National Institute of Technology on 01/25/19. Copyright ASCE. For personal use only; all rights reserved.

Jaeger, L. G., and Bakht, B. (1985b). "The use of harmonics in the semi-continuum
analysis of bridges." Proc. Annual Conference of the Canadian Society for Civ.
Engrg., Canadian Society for Civil Engineering, 83-97.
Jaeger, L. G., Bakht, B., and Surana, C. S. (1988). "Application of analysis of
three-girder skew bridges." Second International Colloquium on Concrete in
Developing Countries, Bombay, India.
Ontario highway bridge design code. (1983). Ontario Ministry of Transportation
and Communications, Downsview, Ontario, Canada.
Standard specifications for highway bridges. (1984). American Association of State
Highway and Transportation Officials, Washington, D.C.

2322

J. Struct. Eng., 1988, 114(10): 2307-2322

You might also like