J. Construct.

Steel Research 7 (1987) 317-334

Inelastic Local Buckling of Fabricated I-Beams

M. A . B r a d f o r d
.School of Civil Engineering, The Universityof New South Wales, Kensington,NSW,
2033, Australia
(Received 3 February 1987;revisedversionreceived 12 March 1987;accepted 8 April 1987)

SYNOPSIS
A finite strip method of analysis is presented for the inelastic local buckling
o f 1-beams fabricated by welding. Stiffness and stability matrices for the
section are developed at a monotonically increasing load factor, and the
critical moment is that for which the buckling determinant vanishes. Critical
moments determined in this way are shown to agree well with test results. The
limiting depth-to-thickness ratios for the web which correspond to compact
and semi-compact sections are investigated, and it is shown that the values
given in BS 5950 : Part I are unconservative for an extensive range of section
geometries. Based on a parametric study, alternative and more accurate
formulations for the critical web slendernesses are proposed.

NOTATION
The section geometry is defined in Fig. 2 and the material properties are
defined in Fig. 3. Other principal notation is as below.
k, g
kw
q
n
A
B,B*
D
Et

Stiffness and stability matrices respectively.

Web local buckling coefficient.
Nodal displacement vector in eqn 7.
Displacement vector in eqn 6.
Area of section or area of strip.
Strain and slope matrices respectively.
Property matrix.
Tangent modulus.
317
J. Construct. Steel Research 0143-974X/87/$03.50 Elsevier Applied Science Publishers
Ltd, England, 1987. Printed in Great Britain

318

M. A. Bradford

Elastic shear modulus.

Buckling half-wavelength.
Bending moment.
M
Interpolation matrix.
Inelastic local buckling moment.
MI,
Elastic local buckling moment.
Mo,
Mx, M. Mxy Infinitesimal buckling moments.
My, Mp
Nominal first yield and plastic moments respectively.
Axial force.
P
V
Volume of strip.
Applied strain.
Ca
~, y, ~/~y
Membrane buckling strains in eqn 8.
Buckling generalised strains in eqn 10.
E
Coordinate of neutral axis.
0, Op
Rotation and rotation at Me respectively.
h
'Load' factor.
Flange and web generalised slendernesses respectively.
hs, hw
Web slenderness constants for compact and semi-compact
hwc, h ,
sections respectively.
//
Elastic Poisson's ratio.
Infinitesimal buckling curvatures in eqn 8.
P~,Py,P~y
or
Generalised stress vector.
Or a
Applied stress.
o'~ , o'y , ~'~
Infinitesimal buckling stresses.
Curvature
G
L
M

1 INTRODUCTION

It is weU-known that slender elastic steel sections can carry bending

moments greater than the elastic local buckling moment Mo, because of the
postbuckling reserve of strength.' This increase in moment is finally limited
by plastification o f the component plates of the section, at which point
unloading usually occurs. However, stocky sections or sections with significant residual stresses may buckle inelasticaUy at a moment Mn which is
lower than Mol. This inelastic local buckling, which is considered in this
paper, will always occur when the elastic critical.moment Mo, is greater than
the nominal yield moment Mr. An illustration of the strain hardening
buckling, inelastic local buckling and post-local buckling of sections is
shown in Fig. 1.
A significant number of experimental studies on the inelastic local
buckling of I-sections have been made, mainly with the intention of

Inelastic local buckling of fabricated 1-beams

319

I
My

MI~ for section

with residual stresses ~

%
u
I
u~

1.0
Section slenderness

/(My/Mo)

Fig. 1. Local buckling of welded I-section.

ascertaining limiting width-to-thickness ratios of the flanges and webs for

which the section is suitable for plastic design. Among the recent studies
have been those of Climenhaga and Johnson, 2 Holtz and Kulak, 3'4Johnson 5
and Kemp. 6 However, relatively little research has been attempted to study
the inelastic buckling of sections theoretically.
In principle, the finite element method can be used to investigate the
inelastic local buckling of plates and sections. 7Such a method can lead to the
iterative solution of equations with large numbers of degrees of freedom,
and is therefore computationally inefficient for many applications. On the
other hand, the finite strip method is suitable for analysing prismatic
members of arbitrary section such as I-beams, and has been used by
Yoshida," and more recently by M a h m o u d 9 and Plank ~ for investigating
inelastic local buckling. A pseudo-strip method has also been employed by
Dawe and Kulak" for this study. The advantage of the finite strip method is
that it is computationally more efficient than the finite element method.
This paper presents a finite strip method of analysis of the elastic and
inelastic local buckling of I-beams fabricated by welding along the flangeweb junction. The residual stresses induced by such welding are included in

320

M. A. Bradford

the analysis. Although restricted to I-sections herein, it is a general method

that may be applied to prismatic sections of arbitrary geometry.
Following development of the finite strip model, its accuracy is demonstrated by a comparison with independent test results. The scope of the
analysis is then demonstrated by a study of the limiting depth-to-thickness
ratios of the web for compact and semi-compact fabricated sections, and
these are compared with the existing provisions of the new BS 5950: Part 1
limit states code.~2

2 FINITE STRIP BUCKLING ANALYSIS

2.1 General

For the analysis in this paper, the I-beam is assumed to be partitioned into
finite strips which are connected to one or more other strips along one or
both of their longitudinal edges in the normal way.~3The x, y, z axis system of
a typical strip is shown in Fig. 2a, which also indicates the displacements
u, v, w in the direction of these axes.
The stress-strain curve assumed for the structural steel is shown in Fig. 3.
It is a trilinear idealisation, with a plastic plateau and a constant strain
hardening modulus E,t = E/h'.
The finite strip procedure requires a calculation of the distribution of
strains applied to the member prior to invoking the bifurcation analysis.
This involves:
(a) selection of an appropriate residual stress model for the fabricated
1-beam; and
(b) application of an initial axial strain and curvature which would occur
if the member was bent and compressed between rigid plattens.
The above procedure is outlined in Section 2.2. The basic steps in the
inelastic bifurcation analysis, presented in the remainder of this section,
then involve:
(a) a definition of the displacement functions to describe the membrane
and flexural deformations of the plate strips during buckling;
(b) a statement of the strain--displacement relations;
(c) selection of an appropriate elastic-plastic plate theory to describe the
membrane and flexural behaviour of the plate strips;
(d) application of the principle of virtual displacements to determine the
stiffness and stability matrices; and
(e) solution of the buckling equation to determine the lowest load factor
for inelastic buckling.

321

Inelastic local buckling of fabricated 1-beams

Z~W

y,v

ILX~U

(a)
I T

bf
-I

1
n

(b)

Fig. 2. Dimensions and displacements. (a) Strip displacements; (b) cross-section

dimensions.

2.2 Applied strains and resulting moment

The residual strains er assumed for the fabricated sections are based on the
so-called 'tendon force concept' developed by the Cambridge group as
summarised by Kitipornchai and Wong-Chung.14 Tension blocks stressed to
o-r are assumed to occur at the welds, accompanied by adjacent compression
blocks such that the plate is in longitudinal equilibrium (Fig. 4). Empirical
expressions for the size of these blocks depend on the plate thickness, weld
size and process efficiency. These expressions, which are set out clearly in
Reference 14, are not reproduced in this paper.
Initially, a curvature ~ is applied to the section, and this is increased

322

M. A. Bradford

Est'E/h'

(st
Strltn

Fig. 3. Stress-strain curve for structural steel.

!
////f~

av

Fig. 4. Residual stresses in fabricated I-section.

Inelastic local buckling of fabricated 1-beams

323

monotonically by a 'load' factor k until buckling occurs, so that the applied

strain , at the point (~, 7?) in the cross-section of the beam may be written as
o(~:,r/) = ~.&(~ - , ) + ~,(~:,~)

(1)

where ~ is the 77 coordinate of the neutral axis, which in general will be

nonzero because of the monosymmetry caused by the presence of the
residual strains. The stress or, at (~:,~) can be found from ~, as
o'o =

I?

E, de, + E~,

(2)

where E, is the tangent modulus of elasticity. The value of C-/ineqn I can be

determined from
Atr, dA = P = 0

(3)

which is the condition for pure bending. The integration in eqn 3 is carried
out numerically over the area A, and the solution of eqns 1, 2 and 3 for ~ is
performed by a chordal Newton-Raphson technique. Finally, the resulting
moment M in the cross-section corresponding to h~b is determined, subject
to eqn 3, by
M = fA tro~dA

(4)

which again is integrated numerically over the area A of the cross-section.

2 . 3 B u c k l i n g d i s p l a c e m e n t s a n d strains for a strip

In accordance with the finite strip method of analysis, the infinitesimal

buckling displacements for a strip are combinations of series functions in the
longitudinal direction, and shape functions in the transverse direction.
Except for a differing axis convention, the displacements for an individual
strip are those employed by Hancock.15 These displacements consist of a
single harmonic in the longitudinal direction, and a cubic polynomial in the
transverse direction, and are appropriate for strips with simple supports for
which the moment gradient in the longitudinal direction is zero.
In matrix notation, the displacements may be written as
u = Mq

(5)

where
u = [u, v, w] r

(6)

324

M. A. Bradford

in which u and v are the membrane displacements, w is the flexural displacement (Fig. 2), and q is the vector of nodal line displacements given by

(7)

q = [Ul, P1, w1,01,/,/2, P2, w2, 02] T

In cqn 7, the subscripts refer to the nodal lines 1 and 2 of the strip,
0 = Ow/Oy, and M is an interpolation matrix of shape functions.
The generalised strains, which include normal and shear strains as well as
bending and twisting curvatures, are obtained by appropriate differentiation of the displacement functions with respect to the relevant coordinate
variables. The strains and curvatures applicable to a plane stress bending
strip are
~x ---- OU/OX, Ey ----- o v l O y , Yxy ---- c)ulOy + OV/OX
Px = -- 0 2 W / o x 2 , Py = -- C92w/OY 2, Pxy = 2 0 2 W / O X O Y

(8)

which may be written in matrix format as

= Bq

(9)

where

0o)
and B is the strain matrix obtained by the appropriate differentiation of eqn
5.
2.4 Constitutive relationship

The buckling constitutive equations relate the membrane normal and shear
stresses to the axial strains and shear strains in the strip, and the bending and
twisting moments per unit length of strip to the curvatures and twist of the
strip. They may be written as

try
l"xy

= |D21
L 0

Dz2
0

ay
Da3

(11)

"Y~

(12)

Mxy

Dss

p,,y

Inelastic local buckling of fabricated 1-beams

325

which may be combined into matrix format as

or= De

(13)

where D is the property or constitutive matrix. For isotropic elastic buckling

(i.e. in regions of the cross-section where c. is less than ev)
Dll = D22

E/(1- v2)

D12 = D21 =

(14)

vDu

D33 = G

where v is the elastic Poisson's ratio and

E
G - 2'l+v------~/t

(15)

For inelastic buckling (i.e. in regions where e, is greater than or equal to v),
rigidities appropriate for inelastic buckling must be used. In this paper, the
rigidities employed by Dawe and Kulak H for buckling of hot-rolled sections
have been used, which are based on a derivation from the flow theory of
plasticity presented by Haaijer.16 These approximate rigidities are given by
Dll = E~/(I - ulv2)
D22 = 4 E E J ( 3 E , t + E)(1 - u~v2)

D~2 = D21

[3Es,+E

(2v-1)E~I+E

/(1-v~v2)

(16)

4EEs,
0 3 3 ~---

4E~,(1 + v) + E
{(2v - 1)E,t + E} 2

i/lV 2 =

E(3E,~ + E)

The expression for the shear rigidity D33 differs from that of Haaijer, and
was derived by Lay 17 to be applicable for the trilinear stress--strain model
shown in Fig. 3. It has been used for tangent modulus lateral buckling, is as
well as for local buckling of hot-rolled beam-columns by Dawe and Kulak. u

326

M. A. Bradford

2.5 Buckling solution

Invoking the principle of virtual displacements for a strip gives ~9

(17)

awl ax

aawl ax

where (au/ox, av/Ox, aw/Ox)r is the vector of slopes with respect to the x-axis
which may be written as
au/ax
av/ax

= B*q

(18)

aw/ax
and B* is the matrix obtained by appropriate differentiation of eqn 5.
Substituting eqns 9, 13 and 18 into eqn 17, performing the variation and
noting that 8q is arbitrary, leads to

Ik(k) - g(k) l -- 0

(19)

for a nontrivial q, where

k(k) =

nrD(k)ndV

(20)

and

g(k) = fv B*r%(h)B*dV

(21)

are the strip stiffness and stability matrices respectively. The strain matrix B
and the slope matrix B* have been given by Hancock 2with permuted axes.
In this paper, the integrations with respect to z in eqns 20 and 21 are
performed analytically, and are then performed numerically with respect to
the area A using four-point Gaussian quadrature.
For a particular 'load' factor h, k and g are determined for each strip, and
are then assembled into the global matrices K and G.13 These matrices are
calculated at monotonically increasing values of h, and the determinant of
their difference is inspected. The procedure is repeated until the determinant changes sign, when a suitable interpolation scheme is used to hasten
convergence to the load factor for which the determinant is zero.
The critical moment is then determined from the critical value of k using
eqn 4. An eigenvector routine ~1is also invoked at the critical value of h to

Inelastic local buckling of fabricated 1-beams

327

obtain the buckled shape. The increments in h need to be kept reasonably

small to ensure that the lowest buckling m o d e is not missed.

3 EXPERIMENTAL VERIFICATION
T h e accuracy of the finite strip m e t h o d for inelastic buckling was investigated by comparing the theoretical solutions with tests on fabricated Ibeams c o n d u c t e d by Holtz and Kulak. 3,4 These tests were chosen primarily
to verify the assumptions m a d e for the material properties and residual
stresses in the theoretical model.
T h e theoretical and experimental buckling moments M~, normalised with
respect to the plastic m o m e n t of resistance Me, are compared in Table 1. For
the finite strip analysis, residual stresses were assigned to the sections in
accordance with Kitipornchai and Wong-Chung, ~4while a strain hardening
m o d u l a r ratio h' of 50 was used. It can be seen that the agreement between
t h e o r y and experiment is good, with the mean ratio of theory to experiment
being 0.97 with a coefficient of variation of 0.06 for the twelve beams
considered. T h e inelastic buckling of beams with low values of Mn/Mp is
d o m i n a t e d by the effects of residual stresses, and the agreement between
test and theory for these beams illustrates the accuracy of the residual stress
modelling in the finite strip treatment.

TABLE 1
Comparison of Finite Strip Solutionswith Tests by Holtz and Kulak

Mn (theory)

Specimen
WS- 1
WS-2
WS-3
WS-4
WS-6
WS-7-P
WS-8-P
WS-9
WS-10
WS- 11
WS-12-N
WS-13-N
mean = 0.97, c.o.v. = 0.06

Mp

0.962
0.899
0.920
0.898
0.932
0.963
0.944
1.002
0.998
0.959
0.916
0.919

Mn

theory

1"023
0-977
0"893
0.863
0-879
0.993
0-954
1-077
1.147
1.066
0-944
0.919

0-94
0-92
1-03
1"04
1-06
0.97
0.99
0.93
0.87
0.90
0.97
1.00

Mp (tesO

test

328

M. A. Bradford

A non-dimensional moment-rotation plot for specimen WS-6, obtained

from the first part of the finite strip analysis, is compared in Fig. 5 with that
given by Holtz and Kulak. a The agreement between test and theory is
reasonable, particularly for the lower beam rotations. The figure illustrates
that the flexural stiffness is reduced by 30% below the linear elastic response
due to the presence of residual stresses. It can also be seen that unloading
takes place after attainment of the inelastic local buckling moment, and this
behaviour was evident with the other tests. 3'4 Because of this, the critical
moment Mn, determined by the finite strip method, is a good prediction of
the ultimate strength of the section in bending.
Finally, a plot of the elastic and inelastic buckling moments Mol and M~, is
given in Fig. 6 as a function of the buckling half wavelength L for specimen
WS-13-N. 4 The elastic curve was obtained from the finite strip analysis
presented herein by appropriate choice of the material properties. While the
elastic critical moment Mot (which is greater than the yield moment My) has
the familiar U-shape, the inelastic critical moment Mtt shows a very much
flatter curve, with the moment being nearly constant in the range 0.3 < L/
h < 1.0. Also shown in Fig. 6 are the elastic and inelastic eigenmodes at the
respective minima. While both plots display a combined flange-web buckle
with significant relative web deformations, there is a difference between the
elastic and inelastic buckled shapes, particularly with respect to the flange
deformations.

4 DEPTH-TO-THICKNESS LIMITS FOR WEB BUCKLING

4.1 General
The finite strip analysis has been used to determine the depth-to-thickness
ratios d/t for webs which correspond to a section classification of compact
and semi-compact. The limiting d/t value for a compact section is that for
which the inelastic buckling moment Mn equals the plastic moment My,
while the limiting d/t value for a semi-compact section is that for which Mn
equals the nominal yield moment My (Fig. 1).
For the analysis in this section, a strain hardening modular ratio h' of 50
was used, with ~t = 10~v. It was found that the solutions were not greatly
sensitive to the value of h'. Residual stresses were assigned in accordance
with Reference 14.
4.2 Compact sections
Figure 7 shows plots of the dimensionless buckling moment MII/Mp against
the modified web slenderness Xw = (d/t)X/(trv/275) near Mn/Mp = 1,

Inelastic local buckling of fabricated I-beams

329

1.0

Linear

Local

/ ~ . , ~ e - - -

~,-e.,,.

buckling

0.8

~/# / f

Rp

/pC~

Thisstudy

0.6

0.4

0.2

B/Bp

Fig. 5. Moment-rotation curveforspecimenWS-6.

\
M0~

\
\

//
\

//

\
\

z o

z
z

etgen~des

0.1

0.2

0.5

1,0

2,0

Dimensionless half wavelength L/h (log scale)

Fig. 6. Local buckling behaviour of specimen WS-13-N.

M. A. Bradford

330
1.02

MI/MP
1.01

1.00

70

0.99

80 \

"<'-,,

"~xloo

O.98
kf=8

Calculated

p0~nts

Fig. 7. Parametric curves for local buckling near Mp.

w h e r e o'v is in units of N/ram 2. The curves are for bt/h = 0.12 and 0.24, and
for flange slendernesses he = (b/73 ~/(o-v/275) of 6.0, 7.0 and 8.0. These
flange slendernesses represent a compact classification in BS 5950 : Part 1.12
I n t e r p o l a t e d values of the limiting web slenderness (hw),m, for which
Mn = Mp, are given in Table 2. T h e residual stresses were found to have
little effect on the values of (h,,)~. It can be seen that the prediction of
h,, = 98 given in BS 5950: Part 1 is an unconservative estimate of the critical
slenderness (k,,)m, and does not account for the variation of the critical
slenderness with he and bf/h. This variation was investigated by writing the
limiting d/t ratio for the web in the form
(Xw),im = ~/(23"9/kw)

(22)

w h e r e k,, is the elastic local buckling coefficient for the web (given in Table
2) which can be d e t e r m i n e d from the charts in Reference 22, while 23.9 is the
theoretical value for k,, for a web in bending with simple supports.l It was
f o u n d f r o m the results in Table 2 that the m e a n value of h,,c in eqn 22 for the
c o m p a c t classification was equal to 73, with a coefficient of variation of 2%
for the sections considered. This low coefficient of variation means that h,,c
m a y be considered as a near constant, so that eqn 22 can be used to
d e t e r m i n e the limiting web depth-to-thickness ratios for a compact section
w h e n the web buckling coefficient is known. T h e use of this value of h~,~
implies that the limiting value of k,, = 98 in BS 5950 : Part 1 will be correct
for a section with an elastic buckling coefficient k,, = 43.

Inelastic local buckling of fabricated/-beams

331

TABLE 2
Parametric Results for Compact Sections

bf/ h

hf

(hwh~

k,~

hwc

0-12
0.12
0.12

6-0
7-0
8.0

88
81
76

35
30
27

73
72
72

0-24
0.24
0.24

6.0
7.0
8.0

91
90
98

37
36
41

74
73
75

mean hwc = 73, c.o.v. = 0.02

4.3 Semi-compact sections

T h e m e t h o d of the previous sub-section was used to investigate the effect of
the p a r a m e t e r s bf/h and hf on the web slendernesses h~, for which Mn = My.
T h e dimensionless buckling moments Mn/Mv are shown in Fig. 8 in the
vicinity of Mn/My = 1 for the same parameter values as those used in the
investigation of compact sections.
I n t e r p o l a t e d values of the limiting web slenderness (hw)~ for a semic o m p a c t section are given in Table 3. These values of (hw)t~ were found to
d e p e n d significantly on the welding residual stresses, with the model of
R e f e r e n c e 14 being a conservative representation of these stresses and
resulting in reasonably low values of (hw)~. The value of hw = 120 in BS
5950 : Part 1 is clearly an unconservative prediction of the limiting web
d e p t h - t o - t h i c k n e s s ratio for semi-compact sections when the residual stress
p a t t e r n of Reference 14 is used.

TABLE 3
Parametric Results for Semi-CompactSections

bf/ h

hf

(Xw)tim

kw

Xw,

0-12
0.12
0.12

6.0
7.0
8.0

107
106
102

37
36
34

86
86
85

0.24
0-24
0.24

6.0
7-0
8-0

104
104
108

37
37
38

84
84
85

mean hws = 86, c.o.v. = 0-01

M. A. Bradford

332
1.06

\
1.04

\
\

MIL/M Y
1.02

I.~

0.~

U
\

~'~'

%
%

8
"~,L 6

0.96

\
).f=7

bf/h - 0.12
0.24 . . . .

0.94

Calculated points

Fig. 8. Parametric curves for local buckling near My.

The variation of (hw),imwith the chosen parameters was again investigated

by writing
(hw).m = X/(23.9/kw)

(23)

where kw is the elastic buckling coefficient for the web which may be
determined from Reference 22 and is given in Table 3 for the sections considered herein. The mean value of h,, was found to equal 86, with a
coefficient of variation of 1%, and so this value of h,, may be considered as a
constant for semi-compact welded sections. The formulation of eqn 23
enables the effects of the geometrical parameters on the limiting depth-tothickness ratio for a semi-compact section of the web to be included through
the presence of the elastic buckling coefficient kw.
The prediction of the limiting modified web slenderness for semi-compact
sections of 120 in BS 5950 : Part 1 thus implies an elastic buckling coefficient
kw = 46. This is reasonably consistent with the value of kw = 43 implied in
BS 5950 : Part 1 for compact sections.

5 CONCLUSIONS
A finite strip method of analysis for the inelastic local buckling of fabricated
I-beams in bending has been presented. The method includes nonlinear

Inelastic localbuckling of fabricated l-bearra"

333

material behaviour and the residual stress patterns present in welded

sections.
A m e t h o d for the nonlinear analysis of the cross-section was given, and
this enabled the m o m e n t of resistance and curvature to be obtained at
increasing values of a load factor. The stiffness and stability matrices of the
b e a m were then assembled at each value of this load factor. The critical load
factor was that for which the determinant of the difference of the stiffness
and stability matrices vanished.
Local buckling moments were calculated in this way for twelve beams
tested independently. The good agreement obtained validated the assumptions made in the finite strip method of analysis.
The limiting web slendernesses corresponding to compact and semicompact welded I-sections were then obtained parametrically. It was shown
that these slenderness limits were proportional to the square root of the
elastic web local buckling coefficient for the section for both classifications.
Values of the constant of proportionality were proposed for both compact
and semi-compact welded I-sections, and it was shown that the limiting web
slendernesses given in BS 5950 : Part 1 are unconservative for an extensive
range of section geometries.

ACKNOWLEDGEMENT
The c o m p u t e r program used in this paper was developed while the author
was a Visiting Fellow in the Department of Engineering at the University of
Warwick, UK. The support provided by the British Steel Corporation for
this work is gratefully acknowledged.

REFERENCES
1. Allen, H. G. and Bulson, P. S., Background to buckling, Mc.Graw Hill (UK),
1980.
2. Climenhaga, J. J. and Johnson, R. P., Local buckling in continuous composite
beams, The Structural Engineer, 50 (9) (1972) 367-74.
3. Holtz, N. M. and Kulak, G. L., Web slenderness limits for compact beams,
Struct. Engng. Report No. 43, Dept. of Civil Engng., Univ. of Alberta,
Canada, March 1973.
4. Holtz, N. M. and Kulak, G. L., Web slenderness limits for non-compact beams,
Struct. Engng. Report No. 51, Dept. of Civil Engng., Univ. of Alberta,
Canada, August 1975.
5. Johnson, D. L., Buckling of beam compression flanges, Research Report,
Butler Manufacturing Co., Grandview, Missouri, USA, November 1976.

334

M. A. Bradford

6. Kemp, A. R., Local buckling of webs and flanges in plastic design to SABS 0162
: 1984, Die Siviele Ingenieur in Suid-Afrika, December 1985, 605-11.
7. Owen, D. R. J. and Hinton, E., Finite elements in plasticity theory and practice,
Pineridge Press, 1980.
8. Yoshida, H., Coupled strength of local and whole bucklings of H-columns,
Proc. Japan Soc. of Civil Engrs, 243 (November 1975) 19-32 (in Japanese).
9. Mahmoud, N. S., Inelastic stability of plate structures using the finite strip
method, PhD Thesis, Univ. of Sheffield, UK, 1981.
10. Plank, R. J., 'A study of the elastic-plastic instability of stiffened panels using
the finite strip method', Int. Conf. of Instability and Plastic Collapse of Steel
Structures, Manchester, 1983, 399--408.
11. Dawe, J. L. and Kulak, G. L., Plate instability of W shapes, J. Struct. Engng.,
ASCE, 110 (6) (June 1984) 1278-91.
12. British Standards Institution, BS 5950 : Part 1. Structural use of steelwork in
buildings, London, BSI, 1985.
13. Cheung, Y. K., Finite strip method in structural analysis, Oxford, UK,
Pergamon Press, 1976.
14. Kitipornchai, S. and Wong-Chung, A. D., Inelastic buckling of welded monosymmetric I-beams, J. Struct. Engng., ASCE 113 (1987) 740-56.
15. Hancock, G. J., Local, distortional and lateral buckling of I-beams, J. Struct.
Div., ASCE, 1114(ST11) (1978) 1787-98.
16. Haaijer, G., Plate buckling in the strain hardening range, J. Engng Mechs
Div., ASCE, 81 (EM2) (1957) 1-47.
17. Lay, M. G., Flange local buckling in wide-flange shapes, J. Struct. Div., ASCE,
91 (ST6) (1965) 95-116.
18. Bradford, M. A., Cuk, P. E., Gizejowski, M. A. and Trahair, N. S., Inelastic
lateral buckling of beam-columns, J. Struct. Engng, ASCE (1987) (in press).
19. Bradford, M. A., Buckling of beams with flexible cross-sections, PhD Thesis,
Univ. of Sydney, Australia, 1983.
20. Hancock, G. J., Appendices in Local, distortional and lateral buckling of
1-beams, Research Report R312, School of Civil Engng., Univ. of Sydney,
Australia, 1977.
21. Hancock, G. J., Structural buckling and vibration analysis on microcomputers,
Civil Engng Trans., lnstn. Engrs Australia, CE26 (4) (1984) 327-32.
22. Hancock, G. J., Bradford, M. A. and Trahair, N. S., Web distortion and
flexural-torsional buckling, J. Struct. Div., ASCE, 106 (ST7) (1980) 1557-71.