You are on page 1of 13

Thin-WalledStructures22 (1995)203-215

© 1995ElsevierScienceLimited
Printed in Great Britain. All rights reserved
0263-8231/95/$9.50
ELSEVIER 0263-8231(94)00037-9

Optimum Design of Load-Bearing Box Girder Diaphragms


Having a Central Support

T. H. G. Megson a & G. Hallak b


aDepartment of Civil Engineering, University of Leeds, Leeds LS2 9JT, UK
bDepartment of Civil Engineering, Teshreen University, Lattakia, Syria

(Received 23 October 1993; accepted 11 April 1994)

ABSTRACT

A parametric study is carried out to determine the optimum design of load-


bearing box girder diaphragms which are supported on a single bearing
positioned at the diaphragm centre line. The parameters varied include the
depth~breadth ratio, the depth~thickness ratio, the size and location of stif-
feners and the eccentricity of loading. The criterion used for comparison
purposes is the collapse load for each configuration, this being determined
using a finite element package. Finally, the effects on the collapse load of
the different configurations are discussed and recommendations made.

NOTATION

B Breadth of diaphragm
d~ Depth of stiffener
D Depth of diaphragm
E Young's modulus
J Width of bearing
L Position of bearing
tD Thickness of diaphragm
ts Thickness of stiffener
Wd Initial geometric imperfection mode of diaphragm
Ws Initial geometric imperfection mode of stiffener
X,Y,,Z Global axes of box girder

203
204 7". H. G. Megson, G. Hallak

V Poisson's ratio
~Ty Yield stress of steel

1 INTRODUCTION

This paper describes part of a research programme I into the optimum


design of load-bearing box girder diaphragms. The diaphragms are
rectangular in shape and supported by a single bearing positioned on the
centre line of the diaphragm; other bearing positions are discussed in
related papers.
In the parametric study a three-dimensional finite element model 2 of the
diaphragm and its associated length of box girder was analysed using the
ABAQUS package 3 to determine the collapse load of each configuration.
The collapse load was then used to compare the merits of the different
configurations so that an optimum arrangement could be arrived at.

2 P A R A M E T R I C STUDY

The study covers a range of practical geometry parameters including the


breadth/depth ratio of the diaphragm, the depth/thickness ratio, the bear-
ing position, the stiffener arrangement and size and the initial geometric
imperfections of the diaphragm and stiffeners. During the study the depth
(D) of the diaphragm was maintained at a constant value of 1000mm.
Further, the diaphragm and its length of box girder were designed such that
collapse would occur in the diaphragm rather than elsewhere.

2.1 Material

Steel, grade 50B, was used for the model since it is the type of steel most
usually employed in box girder construction; the following properties were
assumed:
E = 210 000 N/mm 2, gy = 350 N/ram 2, v = 0-3

2.2 Bearing position

Box girders are generally supported by single or twin bearings. Single


bearings are more economical in cases where torsional loads are low or
where torsional supports are provided at each end of the box girder.
Where torsional loads are high twin bearings are employed and the width
Load-bearing box girder diaphragms 205

of these can be varied across the width of the box girder. In this research
programme three bearing positions were investigated, twin bearings
placed under the flange/web junctions, twin bearings in an intermediate
position and, finally, a single bearing positioned in the vertical plane of
symmetry of the box girder. This paper describes the results obtained for
the case of a single bearing placed in the vertical plane of symmetry of the
box girder; the width of the bearing was taken to be one fifth of the
breadth of the diaphragm, i.e. J = 0-2 B.

2.3 Breadth/depth ratio

Three values of B/D ratio were chosen, i.e.


s/D = 1,s/D = 2,s/D = 3

representing deep, moderate and shallow box girders, respectively.

2.4 Depth/thickness ratio

The influence of the thickness of unstiffened diaphragms was investigated


by determining the collapse loads of diaphragms for which
D/tD = 30, 55, 80, 100
these values representing a range of values encountered in practice. This
range of depth/thickness ratios was investigated for each of the BID ratios
specified in Section 2.3.

2.5 Stiffener arrangements

In the analysis of stiffened diaphragms a D/to ratio of 100 was selected


for all cases since, as can be seen later, the collapse loads of unstiffened
diaphragms having B/D = 1, 2, 3 were very nearly the same at this value.
The following arrangements of stiffeners were considered.
(1) Single load-bearing stiffeners located at the centre of the bearing
(Fig. 1).
(2) Twin load-bearing stiffeners placed at each edge of the bearing
(Fig. 2).
(3) Single load-bearing with stub stiffeners (Fig. 3).
(4) Single load-bearing with horizontal stiffeners (Fig. 4).
The flail depth and stub stiffeners were located in pairs, symmetrically
placed., on opposite faces of the diaphragm. The horizontal stiffeners were
attached to the concave face of each diaphragm, this having been
"saou~jj~s qnls pue ~u.ueoq-p~o I gl~U!S ,£q pouo3j!;s u~raqde!(I "£ "B.~I

B~.O j.8~.O: F,. 1

F'E
I - -

sI

4
0£-I=8

"saouoAI!ls Su.ueaq-peo I u!~l ,~q pou~J3!ls u ~ e a q d r ! o "~ "g!d

w~.o yvo=L
r ",.' "It

~-s~--~

0£-~--8

•aougJj!ls ~u.u~oq-p~o I ol~u!s e ,~q pouojj.tls u~eatld~!o "l "~bl

89.0 ±SZ.0=F± ]
r T T

T
I

0£-I =8

~l°llOH "D 'uosSalq "D " H ".I, 90~5


Load-bearing box girder diaphragms 207

i~ B=I "3D

t$
D

I* L ~'J=0"2B #" 0 "4B

Fig. 4. Diaphragm stiffened by single load-bearing and horizontal stiffeners.

previously determined from the analysis of the single load-bearing stif-


fener case.

2.6 Stiffener sizes

Throughout the parametric study flat stiffeners were used all having a
thickness of 10 mm (0.01 D). The depth/thickness ratios investigated were
ds/ts = 2-5, 5.0, 7.5, 10.0.

2.7 Imperfections

The modes of the initial geometric imperfections of the diaphragms and


stiffeners were assumed to be those obtained from an elastic buckling
(eigenvalue) analysis. Thus, the diaphragms were analysed initially with
zero imperfections and the elastic buckling modes obtained. These were of
sinusoidal form so that the initial imperfections of the diaphragm and
stiffeners were taken to be
B/D=I
~Y
W~ = 3.0sin rc(X + - sm--" (1)
B D
~z(X+ 0.5B) . ~zY
Ws = 3.0 sin ~/~ sm-~- + 2.0 s i n - ~ (2)

B/D=2
Wd. = 3.0 sin 2re(X+ 0.5B) sm
. uY (3)
B D
208 T. H. G. Megson,G. Hallak
~,, . 27~(X+0.5B) . 7~Y
Ws = J.o sin B sm-~- + 2 . 0 s i n - ~ (4)

8/D:3
3 n ( X + 0 - 5 B ) . nY
Wd = 3-0 sin stu- (5)
B D
+0.5B) . n Y . ~zY
Ws = 3.0 sin 37~(X B s m - f f ~ 2.0 s i n - f f (6)

The amplitudes of these initial imperfections were obtained by reference to


BS:5400:Part 6. 4 In addition the concentrated loads were applied to the
bearing plates with two values of eccentricity, i.e. e = 0 m m , e = 10mm
(0.01 D). In order to obtain the minimum collapse load the eccentricity
was always provided in a direction opposite to the initial imperfections.

2.8 Finite element model

A full description of the finite element model used in the parametric study
is given in Ref. 2. The ABAQUS package is described in Ref. 3.

3 RESULTS

3.1 Unstiffened diaphragms

Figure 5 shows the variation of collapse load with D/tD ratio for the three
breadth/depth ratios B/D = 1, 2, 3 and for the loading eccentricities e = 0
and 10 mm. Also included in the analysis was the initial geometric imper-
fection mode Wd, eqns (1), (3) and (5).
For diaphragms having B/D = 1, D/tD = 30 and 55 failure was due,
primarily, to yielding in the diaphragm above the bearing. For values of
D/to = 80 and 100 failure was produced by buckling.
In diaphragms having B/D = 2 and D/tD = 30 failure was caused by
yielding in the diaphragm above the bearing while for diaphragms having
D/tD = 55 failure was due to a combination of yielding and buckling. In
diaphragms having D/tD = 80 and 100 failure was produced by buckling.
For diaphragms in which B/D = 3 and D/tD = 30 failure was due to
yielding in the diaphragm above the bearing; diaphragms having
D/tD = 55, 80, 100 all failed by buckling over the bearing.
In all cases the effect of eccentricity of loading was negligibly small with
a maximum reduction in collapse load of 2% for a diaphragm having
B/D = 1 and D/tD = 30.
Load-bearing box girder diaphragms 209

7"5 [ unstiffcn~l diaphragm ]

o olo=,.:I0
" ~ BID=3 e=0
6"5 ! BID = 2 e=10mm
,, "4'- B/D =2 e=0
' o B/D =1 ¢=10mm
• 0"

s.s
"\ \
x\

,,,
%
r~ B/D=1 ¢=0

oO
o 5"0

_9.
° ~'5

o 4'0

o 3"5
~o ~ ,,

3"0

2"5

2"0

1"5
30 40 50 60 70 80 90 100
D/t0 ratio

Fig. 5. Variation of collapse load with D/tv ratio for different B/D ratios and load
eccentricities.

3.2 Diaphragms stiffened by single load-bearing stiffeners

The variation of collapse load with stiffener depth for the three
breadth/depth ratios and two values of load eccentricity is shown in
Fig. 6.
For B/D = 1 the greatest increase in collapse load over the unstiffened
case was 81.3% and occurred when 100 x 100mm stiffeners were used,
but in the case of the larger stiffeners, failure was due to yielding and
buckling of the diaphragm in the vicinity of the bearing.
For diaphragms having B/D = 2 failure was caused by global buckling
for both 25 x 1 0 m m and 50 x 1 0 m m stiffeners, whereas for 80 x 10ram
and 100 x 1 0 m m stiffeners collapse occurred due to a combination of
210 T. H. G. Megson, G. Hallak

5"0
~ e d iietphragm
~'5
[--N-R/D=3.e=0 --e-~ 10
I --~-B/O=Z.e=0 --4-B/D=2.==10ramI
z ~-B/D=I, ¢=0 .-o- B/D=1,e= 10_rriml

3"5
"U
0
_.o3"0

0
=0 25
u

1.5I

1"0
0 25 50 75 100
Depth of the stiffeners (ram)

Fig. 6. Variation of collapse load with size of single load-bearing stiffeners for different
B/D ratios and load eccentricities.

yielding in the diaphragm and stiffeners in the vicinity of the bearing and
buckling of the diaphragm above the edges of the bearing.
In the case of diaphragms having B/D = 3 global buckling caused fail-
ure when 25 × 10mm and 50 x 10mm stiffeners were used. When larger
stiffeners were employed failure was due to a combination of shear buck-
ling in the overhanging parts of the diaphragm and local buckling in the
diaphragm above the edges of the bearing.
In all cases the effect of load eccentricity was negligible.

3.3 Diaphragms stiffened by twin load-bearing stiffeners

Figure 7 shows 'the variation of collapse load with stiffener depth for the
different B/D ratios and load eccentricities.
For B/D = 1 the greatest increase in collapse load over the unstiffened
case was 112.9% and occurred when 100 x 1 0 m m stiffeners were used.
Failure was due to global buckling when 25 x 10mm stiffeners were
employed, but in the case of larger stiffeners failure was caused by yielding
in the diaphragm and stiffeners in the vicinity of the bearing.
The greatest increase in collapse load for B/D = 2 was 142% and occur-
red when the diaphragm was stiffened by 100 x 10 m m stiffeners. The modes
of failure were similar to those described for diaphragms having B/D = 1.
Load-bearing box girder diaphragms 211

~0
J stiffened diaphragm [

l,'5

~-0
0 7
x

0
r -

1.s i BID:3. e:0 + B/D:3.e: 10ram!


-4~-B/D:2, e:0 --+-EYD:2,e: 10mm
3/D=~ e=O +B/D=I, ¢= 10rami
1"0 I
0 25 50 75 100
Depth of the stiffeners (ram)
Fig. 7. Variation of collapse load with size of twin load-bearing stiffeners for different
B/D ratios and load eccentricities.

In diaphragms having B/D = 3 there was only a small increase in collapse


load (2%) when the stiffener size was increased from 75 x 10mm to
100 x 10 mm. For the former size of stiffener the increase in collapse load
over the unstiffened case was 127%. Global buckling caused failure in
diaphragms having 25 x 10mm stiffeners, while in the case of 50 x 1 0 m m
stiff~mers, failure was due to a combination of shear buckling in the outer
portions of the diaphragm and yielding of the diaphragm and stiffeners in
the vicinity of the bearing. For 75 x 10mm and 100 x 10mm stiffeners
failure was caused by shear buckling in the outer portions of the diaphragm.
In, all cases the effect of load eccentricity was negligibly small.

3.4 Diaphragms stiffened by single load-bearing and stub stiffeners

The variation of collapse load with stiffener size for the different B/D
ratios and load eccentricities is shown in Fig. 8.
In diaphragms having BID = 1 an increase in collapse load of 110.3%
over the unstiffened case resulted when 100 x 10 m m stiffeners were used.
For all stiffener sizes failure occurred when a central strip, approximately
0.5 B wide, buckled.
For diaphragms having BID = 2 the maximum increase in collapse load
212 T. H, G. Megson, G. Hallak

5.0
stiffened diaphragm
I I
45- -'>*-B/D=3,e=0 --4.-B/D=3.e=10mm
-a- B/O=2,e=0 -I-B/D=2.e=10mm
-la- B/D--I, ¢=0 -e- B/Dffil, e=10mm
49
0
0
3.5
x

o 3.O

0
25
0
u

0
,_o 10

1.5 !
Y
1.17
0 25 50 75 100
Depth of the stiffeners (ram)

Fig. 8. Variation of collapse load with size of single load-bearing and stub stiffeners for
different B / D ratios and load eccentricities.

over the unstiffened case was 120% and occurred when 100 x 10 m m stif-
feners were used. In diaphragms having 25 x 10 m m stiffeners failure was
due to global buckling, whereas for all other stiffener sizes failure was
caused by buckling of the diaphragm above and to the outside of the tip of
each stub stiffener.
In diaphragms having B/D = 3 an increase in collapse load of 113%
over the unstiffened case was achieved when 100 x 10 m m stiffeners were
used. The failure modes for all stiffener sizes were similar to those of
diaphragms stiffened by twin load-bearing stiffeners and having B / D = 3.
In all cases the effect of load eccentricity was negligibly small.

3.5 Diaphragms stiffened by single load-bearing and horizontal stiffeners

Figure 9 shows the variation of collapse load with stiffener size for the
different B / D ratios and load eccentricities.
For diaphragms having BID = 1 the greatest increase in collapse load
occurred when 100 x 1 0 m m stiffeners were used and was 81.3%. The
failure modes were similar to those of diaphragms stiffened by twin load-
bearing stiffeners for which BID = 1.
For diaphragms in which B / D = 2 the greatest increase in collapse load
Load-bearing box girder diaphragms 213

5.0
stiffened diaphr0cjm
I I
t,.5 "~- B/D=3.e=0 -O" B/D=3.1= 10ram
ql.-B/O=2,e=0 -I-- B/D=2.e=10mm
-B-B/D=I.e=0 -o- B/D=1. e=10mm
~.0
8
3.5
x

0
o 3.0
/ f

I1
e-,

_o 2.5
g
//S
lS l f
1.0 0
25 50 75 100
Depth of the stiffeners (ram)

Fig. 9. Variation of collapse load with size of single load-bearing and horizontal stiffeners
for different BID ratios and load eccentricities.

was achieved when 100 x 10mm stiffeners were used and was 106% over
the unstiffened case. Again the modes of failure were similar to those of
diaphragms stiffened by twin load-bearing stiffeners for which BID = 1.
In the case of diaphragms having B/D = 3 the maximum increase in
collapse load occurred when 100 x 1 0 m m stiffeners were employed and
was 112.7%. Three types of failure occurred. First, when 25 x 1 0 m m
stiffeners were used, failure was caused by global buckling. Secondly, in
the case of 50 x 10mm stiffeners, failure was due to a combination of
yielding in the vicinity of the bearing and buckling in the central region of
the diaphragm. Finally, for diaphragms stiffened by 75 x 1 0 m m and
100 :× 10 m m stiffeners, failure was caused by yielding over the bearing.
In all cases the effect of load eccentricity was negligible.

4 DISCUSSION

4.1 Unstiffem~l diaphragms

In diaphragms having D/tD = 30 failure is dominated by yielding of the


diaphragm above the bearing. Thus, as the BID ratio increases the width
of the bearing increases (J = 0.2B) so that the ultimate load increases.
214 T. H. G. Megson, G. Hallak

For diaphragms in which D/to = 55 the failure mode depends upon


B/D ratio. Thus, for B/D = 1 failure is caused by yielding in the vicinity
of the bearing, while for B/D = 2 failure is due to a combination of
yielding around the beating and buckling. For B/D = 3 collapse is due to
global buckling.
When D/to = 80 and 100 the dominant mode of collapse is global
buckling, irrespective of B/D ratio.
In all cases the effect of load eccentricity is negligibly small.

4.2 Stiffened diaphragms

For diaphragms in which B/D = 1 approximately the same collapse load


increase over the unstiffened case is produced by providing either twin
load-bearing stiffeners or single load-bearing plus stub stiffeners. The
effect of a horizontal stiffener on diaphragm strength is small. In
diaphragms having B/D = 2 the greatest increase in collapse load occurs
when twin load-bearing stiffeners are provided; again the effect of a hori-
zontal stiffener on diaphragm strength is small.
In the case of B/D = 3 similar collapse loads are obtained when either
single load-bearing plus stub stiffeners or single load-bearing plus a hori-
zontal stiffener are used. However, both systems produce a relatively small
increase (13%) in collapse load over diaphragms stiffened only by a single
load-bearing stiffener. The greatest increase in collapse load from the
unstiffened case is achieved by the provision of twin load-bearing stiffeners.
The effect of additional vertical stiffeners in the outer portions of a
diaphragm for which B/D -- 3 was investigated but the collapse load was
very similar to those obtained when a single load-beating stiffener was
used; the results are therefore not presented.
In all cases the effect of load eccentricity was negligibly small.

5 CONCLUSION

F r o m the above it may be concluded that the most efficient method of


increasing the collapse load of diaphragms supported by a single centrally
positioned bearing is to provide twin load-bearing stiffeners placed at the
edges of the bearing, irrespective of B/D ratio.

REFERENCES

1. Hallak, G., Optimum design of box girder diaphragms. Ph.D Thesis, Univer-
sity of Leeds, Leeds, 1991.
Load-bearing box girder diaphragms 215

2. Megson, T. H. G. & Hallak, G., Finite element modelling of box girder


diaphragms at supports. Thin-Walled Structures, (To appear).
3. Hibbit, Karlsson and Sorenson Inc., ABAQUS Theory Manual and User's
Manual. Manchester Computing Centre, September, 1989.
4. BS5400:Part 6: 1980. Code of Practice for the design of steel bridges.

You might also like