Anda di halaman 1dari 6

EUROSTEEL 2014, September 10-12, 2014, Naples, Italy

BUCKLING STRENGTH OF HIGH-STRENGTH STEEL BEAMS


Mark A. Bradford, Huiyong Ban
Centre for Infrastructure Engineering and Safety, The University of New South Wales, Sydney, Australia
m.bradford@unsw.edu.au, h.ban@unsw.edu.au

INTRODUCTION
High-strength steel (HSS) flexural members are becoming increasingly popular as design solutions,
because of their higher strength to weight ratios compared with mild steel beams and of the more
robust fabrication techniques that have evolved in recent years [1]. HSS is also finding its way into
many national standards, such as in the Australian AS4100 Steel Structures code [2], which now
allows for steels with a yield stress up to 690 N/mm2. One such application is as a half-through
girder arrangement, which is typical of many railway bridges. In this structural configuration, the
top (nominally unbraced) flange of the plate girder is subjected to compressive normal stresses,
which may precipitate its overall or lateral buckling. Usually, the thickness of the web of the plate
girder is used to prevent or to delay this overall buckling. To optimise the strength to weight ratio of
a HSS plate girder in a half-through girder bridge, the issue of overall buckling requires careful
consideration, particularly as the elastic modulus of the steel (~ 200 kN/mm2) does not increase
with the yield strength.
Cross-girders provide restraint at the location of the bottom flange against out-of-plane or overall
buckling. These elastic restraints are depicted in Fig. 1 as having a translational stiffness ka (units of
force / length) and a twist rotation restraint kr (units of force length). Also shown in this figure is
the lateral-distortional mode shape [3], for which the lateral restraint to the top compression flange
is provided by the stiffness of the web alone as a cantilever, and whose cross-section changes shape
or distorts during the buckling. The buckling mode shown in Fig. 1 is sometimes considered as a
flange strut under uniform compression and restrained by an elastic spring (the web), and in this
form the buckling mode treated as a so-called inverted U-frame. For this, an elastic buckling
solution exists in closed form [4], but it has been shown that a correct representation of this
buckling mode including the restraint stiffnesses ka and kr, as well as the variation of the
longitudinal stress along the flange, necessitates are more complex formulation.
In buckling strength calculations, both the elastic buckling moment Mo and the moment to cause
plastic failure of the cross-section Mp need to be known, and the buckling strength Mb is related to
these in the form shown in Fig. 2 [5] as

Mb
= F ( ) ,
Mp

(1)

in which
=

Mo

(2)

is the generalised member slenderness ratio and F () the empirical design relationship. The shape
of this curve depends on many factors, including the residual stresses, geometric imperfections and
the possibility of local buckling, and is given empirically in design codes of practice. Because HSS
is fabricated from plate elements, further optimisation of the plate girder in a half-through girder
can be achieved by tapering the web depth of the member, for which elastic buckling solutions are
also possible [6].
In order to develop a suitable design model for a HSS through-girder, the correct representation of
the function F () is needed. To this end, the present paper presents a study of tapered plate girders
of HSS, using ABAQUS modelling of the steel. The results incorporate the interaction of elastic

buckling, cross-sectional distortion, yielding, residual stresses and tapering, they form a basis for
developing design guidance for HSS half-through girders.

Dimensionless buckling load Mb/Mp

Fig. 1. Buckling model


2

Full plasticity

1.5

Elastic buckling
1

0.5

0
0

0.5

1.5

2.5

Generalised member slenderness


Fig. 2. Representative buckling curve F ()

ANALYSIS

1.1 Finite element model


A finite element analysis was undertaken on a tapered I-section beam of length L = 30 m having a
flange breadth bf = 200 mm, flange thickness tf = 20 mm and web thickness t = 20 mm. The midspan web depth was 3000 mm and at the ends its depth was 3000 mm, with the taper ratio being
05, 075 and 10. The membrane stiffness of the deck is such that, sensibly, the elastic lateral
stiffness in Fig. 1 is ka = , whilst for the buckling analysis the elastic torsional stiffness kr was
taken as kr [0, ].
The 4-node shell element S4R with reduced integration in the ABAQUS library was employed to
mesh both the flanges and the web, and in addition to the simply supported ends of the beam, the
lateral displacement of the bottom flange was fixed and the twist restraint (kr) was simulated using
the spring element SPRING1 in the ABQUS library. The uniformly distributed load was applied at
the junction of the bottom flange and web as a series of concentrated forces at the nodes. Fig. 3
shows a typical three-dimensional meshing of half of the girder, in which the flange had four
elements across its width, the web 30 elements through its depth, and in which the length of each
element was 100 mm (300 elements lengthwise). Because lateral-distortional buckling modes are
the focus of the present paper, local bucking was not considered a relevant design issue and so no
stiffeners were included.
The HSS was represented as having an elastic modulus E = 210 kN/mm2, shear modulus G = 81
kN/mm2 and Poissons ratio = 03. The material properties in the plastic domain for 690 kN/mm2
steel were adopted from the authors study [1], while for the non-linear buckling analysis, a

geometric imperfection was applied in accordance with the first-order eigenmode with the out of
straightness being 1/1,000 of the span [7].

Fig. 3. Finite element meshing of half of the girder

1.2 Elastic buckling


For beams with elastic restraint against buckling, the buckling mode is known to be sinusoidal for
the case of uniform bending or similar to sinusoidal for other cases [5]. An elastic eigenvalue
analysis was undertaken to quantify the effects of the taper constant and dimensionless twist
restraint z on the buckling, in which
z =

k r L2
,
2GJ

(3)

and in which J is Saint Venants torsional constant for the mid-span cross-section.
Fig. 4 shows two typical first-order buckling eigenmodes for a tapered girder ( = 05), the curve in
Fig. 4(a) being a single sine curve over the buckling half-wavelength L = 30 m for low torsional
restraint stiffnesses but that with a high torsional stiffness having local waves in opposite directions.
Fig. 5 shows the out-of-plane (normalised) lengthwise buckling deformation of the top flange for
various taper ratios, in which it can be seen that the effect of the taper ratio is fairly inconsequential
but that the shape of the elastic buckle is strongly dependent on the stiffness z. The (normalised)
cross-sectional buckling deformations at mid-span are shown in Fig. 6, in which it can be seen that
the profile of the web departs from being perfectly straight (when z = 0) with an increase in z.
Again, it can be seen that the taper ratio has little effect on the distortion of the cross-section.
The elastic critical loads qcr of tapered girders with varying values of and with z [0, ],
determined by eigenvalue analysis, are plotted in Fig. 7. It can be seen that the elastic buckling load
is insensitive to z below a threshold value of z ~ 105, and insensitive to it above another threshold
value of z ~ 107. The form of the curve and the sharp rise in the elastic buckling load within z
[105, 107] is consistent with earlier theoretical research [8]. In addition, it can be seen from Fig. 7
that the elastic buckling load is insensitive to the taper ratio .
1.3 Non-linear buckling
In order to determine the inelastic buckling load using ABAQUS, a 30 mm out-of-straightness of
the top flange at mid-span was applied (30,000 / 1000). For an elastic analysis, the loaddeformation response was determined and the peak value (que) recorded as the elastic non-linear
buckling load, while for an inelastic analysis, material imperfections in the form of residual stresses
was applied and the peak value of the load-deformation response (qu) was recorded as the nonlinear buckling load for 690 MPa HSS. At mid-span, the load to cause full plasticity of the crosssection is qp = 342 kN/m, whilst that to cause first yield is 326 kN/m. The non-linear buckling loads
are also shown in Fig. 8. It can be see that both the non-linear buckling loads are above the
eigenvalue buckling load, because there is a stable post-buckling path for a geometrically imperfect

member past the theoretical bifurcation buckling load. These loads are considerably below the yield
and full plastic loads because of the effect of the member taper and of elastic buckling. Typical nonlinear equilibrium paths are shown in Fig. 8, and the results are given in graphical form in Table 1.

a)

b)

Fig. 4. Typical elastic buckling eigenmodes with = 05, a) z < 10 ; b) z > 10

Deformation

Normalised buckling deflection

1.0

= 0.5
= 0.75
= 1.0

z=0

0.5
6

z=10
7

z=10
z=+

0.0

-0.5
0.0

0.2

0.4
0.6
0.8
Along the length of span (/L)

1.0

x/L

Fig. 5. Lengthwise buckling deformations from first-order eigenmode analysis


(a)

(b)
Undeformed shape

(c)
Undeformed shape

z = +
8
z = 10
7
z = 10
6
z = 10
z = 0

Undeformed shape

z = +
8
z = 10
7
z = 10
6
z = 10
z = 0

z = +
8
z = 10
7
z = 10
6
z = 10
z = 0

Fig. 6. Cross-sectional buckling modes at mid-span: a) = 05; b) = 075; c) = 10

Buckling load (kN/m)

50

= 0.5
= 0.75
= 1.0

40

Nonliner
buckling
Elastic non-linear
(Elastic)
buckling

30

Nonliner buckling
Non-linear
(690 HS(690
steel)
buckling
HSS)

20
Eigen-value buckling

10
0

-2

0 10

10

10

10

10

10

10

12

10

Non-dimensional twist restraint stiffness z


Fig. 7. Buckling load as a function of twist restraint parameter z

(b)

Potential deformation beyond qcr

15 qu
que

Distributed load (kN/m)

Distributed load (kN/m)

(a)

10
qcr
5

Nonlinear buckling (Elastic)


Nonlinear buckling (690 HS steel)

0
0

500 1000 1500 2000 2500 3000


Out-of-plane deformation (mm)

50

Potential deformation beyond qcr

qu

40

que

30
20
qcr
10

Nonlinear buckling (Elastic)


Nonlinear buckling (690 HS steel)

0
0

500 1000 1500 2000 2500


Out-of-plane deformation (mm)

Fig. 8. Non-linear buckling solutions

Fig. 9 plots the relationship between the dimensionless buckling load qu/qp, which is the counterpart
to Mb/Mp in Eq. (1), against the modified slenderness (qp/qcr), which is the counterpart to in Eq.
(2). The solid curves in this figure are those determined using ABAQUS, while the dashed curves
are determined using F() = 032/ in the terminology of Eq. (1). It can be seen that buckling
solutions are only valid in the approximate domain [4, 7] and that the hyperbolic approximation
for F provides a lower bound suitable for design, when the taper ratio is such that [05, 10].
0.1

0.1

qu/qp

0.08

qu/qp

0.08

0.06

numerical
result

0.06

0.04

032/

0.04

0.02

= 05

0.02

numerical
result
032/

= 10

0
0

Modified slenderness = (qp/qcr)


Fig. 9. Buckling strength curves

DEISGN PROPOSAL

Based on the numerical results in this paper, a tentative design procedure is as follows, and
applicable for a practical half-through girder fabricated from HSS with a taper ratio greater than 05.
Based on the cross-section at mid-span, the fully plastic moment Mp is determined, as is the elastic
buckling moment Mo from appropriate curves that can be found in the literature that incorporate the
modified torsional restraint z. The buckling strength from the previous section and using Eq. (1) is
therefore Mb = 032(MoMp) Mp.

SUMMARY

This paper has investigated the buckling of half-through girders made from HSS, and whose webs
may be tapered. Hitherto, design codes have not considered this with any accuracy, because the
interaction of elastic buckling, tapering, yielding and residual stresses does not appear in design
standards. The problem was analysed using ABAQUS software.
The numerical approach can be computationally inefficient, because ABAQUS requires a nonlinear analysis to be deployed when yielding and residual stresses are included, and the entire load

versus deflection response (with the input of a small geometric imperfection) needs to be traced to
determine the peak load for the buckling strength of the member. Using elastic solutions for the
lateral-distortional buckling of torsionally-restrained I-section members reported elsewhere, a
formulation for the buckling strength using the same methodology as in Eurocode 3 has been
proposed, and which was shown to lead to safe lower bound solutions within the range of
parameters for which it is applicable.
Table 1. Buckling results
z
0
10-2
10-1
1
101
102
103
104
105
106
107
108
109
1010
1011
1012
1013
+

Taper ratio = 05
qcr
que
qu
5.5
5.5
5.5
5.5
5.5
5.5
5.5
55
60
87
130
183
199
201
202
202
202
202

144
144
144
144
144
144
144
144
146
166
274
431
469
483
489
490
490
490

142
142
142
142
142
142
142
142
144
165
230
300
323
327
327
327
327
327

Taper ratio = 075


qcr
que
qu
54
54
54
54
54
54
54
55
58
85
136
181
196
197
198
198
198
198

176
176
176
176
176
176
176
176
178
190
278
439
474
493
495
496
496
496

161
161
161
161
161
161
161
161
163
176
232
302
326
329
329
329
329
329

Taper ratio = 10
qcr
que
qu
56
56
56
56
56
56
56
56
59
83
141
183
195
197
197
197
197
197

200
200
200
200
200
200
200
200
201
212
280
440
476
495
497
498
498
498

171
171
171
171
171
171
171
171
172
183
234
303
326
328
329
329
329
329

REFERENCES
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]

Ban, HY, Bradford, MA, Flexural Behaviour of Composite Beams with High Strength Steel,
Engineering Structures, Vol. 56, pp. 1130-1141, 2013.
Standards Australia, AS4100 Steel Structures, SA, Sydney, 1998.
Bradford, MA, Lateral-Distortional Buckling of Steel I-Section Members, Journal of Constructional
Steel Research, Vol. 23, pp. 97-116, 1992.
Oehlers, DJ, Bradford, MA, Composite Steel and Concrete Structural Members: Fundamental
Behaviour, Pergamon Press, Oxford, 1995.
Trahair, NS, Bradford, MA, Nethercot, DA, Gardner, L, The Behaviour and Design of Steel Structures
to EC3, Fourth Edition, Taylor & Francis, London, 2008.
Bradford, MA, Cuk, PE, Lateral Buckling of Tapered Monosymmetric I-Beams, Journal of
Structural Engineering, ASCE, Vol. 114, pp. 977-996, 1988.
British Standards Institution, Eurocode 3: Design of Steel Structures Part 1.1: General Rules and
Rules for Buildings, BS EN 1993-1-1, BSI, London, 2005.
Bradford, MA, Ronagh, HR, Generalized Elastic Buckling of Restrained I-Beams by FEM, Journal
of Structural Engineering, ASCE, Vol. 123, pp. 1631-1637, 1997.