School of Civil Engineering

Sydney NSW 2006

AUSTRALIA
http://www.civil.usyd.edu.au/
Centre for Advanced Structural Engineering

Buckling Analysis Design of Steel Frames

Research Report No R891

N S Trahair BSc BE MEngSc PhD DEng

June 2008
ISSN 1833-2781

School of Civil Engineering

Centre for Advanced Structural Engineering
http://www.civil.usyd.edu.au/

Buckling Analysis Design of Steel Frames

Research Report No R891
N S Trahair BSc BE MEngSc PhD DEng
June 2008
Abstract:
Steel design codes do not provide sufficient information for the efficient design of steel
structures against out-of-plane failure, and what is provided is often overly
conservative. The method of design by buckling analysis corrects this situation for
beams, but the extension of this method to columns is only suggested, while there is no
guidance on how to apply this method to the design of beam-columns and frames.
Beam design by buckling analysis uses the design code formulation for the member
nominal design strengths in terms of the section moment capacities and the maximum
moments at elastic buckling, accurate predictions of which may be determined by
available computer programs. Column design by buckling analysis is similar to beam
design, in that it uses the design code formulation for the column nominal design
strengths in terms of the section compression capacities and accurate predictions of the
elastic buckling loads which may also be obtained from computer programs.
However, design codes do not provide formulations for the direct buckling design of
beam-columns, but instead use the separate results of beam design and column design
in interaction equations. The further extension to frames is not directly possible,
because frames are not designed as a whole (except through the rarely used methods
of advanced analysis), but as a series of individual members. This paper shows how
the method of design by buckling analysis can be used to design beam-columns and
frames as well as beams and columns. Two example frames are designed and very
significant economies are demonstrated when the method of design by buckling
analysis is used.

Keywords: beams, beam-columns, bending, buckling, columns, compression, design,

frames, member strength, moments, steel.

Buckling Analysis Design of Steel Frames

June 2008

Copyright Notice
Buckling Analysis Design of Steel Frames
2008 N.S.Trahair
[email protected]
This publication may be redistributed freely in its entirety and in its original form without the
consent of the copyright owner.
Use of material contained in this publication in any other published works must be appropriately
referenced, and, if necessary, permission sought from the author.

Published by:
School of Civil Engineering
The University of Sydney
Sydney NSW 2006
AUSTRALIA
June 2008
http://www.civil.usyd.edu.au

School of Civil Engineering

Research Report No R891

Buckling Analysis Design of Steel Frames

June 2008

INTRODUCTION

Steel beam design by out-of-plane (flexural-torsional) buckling analysis allows greater

efficiency than when the buckling approximations embedded in some design codes
such as AS4100 [1] or BS5950 [2] are used. Design by buckling analysis is allowed in
AS4100 and implied in EC3 [3]. Steel column design by buckling analysis is suggested
in the Commentary to AS4100 [4] and implied in EC3. However, no methods are given
or suggested for designing beam-columns or frames against out-of-plane failure.
Beam design by buckling analysis uses the design code formulation for the member
nominal design strength Mbx in terms of the section capacity Msx and the elastic buckling
moment Moo of the basic simply supported beam in uniform bending. Column design by
buckling analysis is similar to beam design, in that it uses the design code formulation
for the column nominal design strength Ncy in terms of the section capacity Ns and the
elastic buckling load Noc of the basic simply supported column. For other than the basic
cases, the elastic out-of-plane buckling moment or load needs to be determined.
Elastic out-of plane buckling of beams and columns [5] depends on:
(a)
the elastic moduli, the section properties, and the length,
(b)
the distribution of the moments and compressions and the load heights,
(c)
the restraints,
(d)
any non-uniformity of the section, and
(e)
interactions between these.
The effects of the elastic moduli, the section properties, and the length are allowed for
in the basic cases, and formulations are given in most [1, 2, 6] but not all [3] codes.
The effect of the distribution of moments is allowed for approximately in most codes [1,
2, 3, 6]. The effect of load height is partially allowed for in some codes [1, 2], using
fairly inaccurate approximations.
Restraints may be concentrated or distributed, and may be elastic or rigid. Restraints
may be translational, rotational (out-of-plane), torsional, or warping. The effects of
translational and rotational restraints depend on their height. No codes allow for the full
range of restraint effects. Some [1, 2] allow approximately for partial torsional end
restraints, and some [1, 2, 6] allow for rotational elastic end restraints acting at the
centroid.
Some design codes [1, 2, 6] provide limited approximations for the effect of nonuniformity of the section.
Design codes assume that these effects are largely independent so that they can be
treated separately, but this is not the case [7]. Thus there is a lack of precision in the
often conservative approximations used in codes to determine the elastic buckling
moments and loads. This can be overcome to a certain extent by using some of the
wealth of published research information [7] but this is difficult to access. However, for
some time now computer programs [8, 9] have been available, and some of these such
as ABAQUS [10] are often used by designers. The use of these programs now
provides a viable method of carrying out the efficient and economical design of beams
and columns by buckling analysis.
School of Civil Engineering
Research Report No R891

Buckling Analysis Design of Steel Frames

June 2008

However, the extension of design by buckling analysis to beam-columns cannot be

carried out in the same way, even though accurate predictions of the elastic buckling
loads can be obtained. The reason for this is that design codes do not provide
formulations for the buckling design of beam-columns, but instead use the separate
results of beam design and column design in interaction equations. The further
extension to frames is also not possible, because frames are not designed as a whole
(except through the rarely used methods of advanced analysis [11]) but as a series of
individual members.
The purposes of this paper are to show how the methods of design of beams and
columns by buckling analysis can be used to design beam-columns and frames.

DESIGN BY BUCKLING ANALYSIS

2.1

Beams

The basic design code case for beams is the simply supported doubly symmetric beam
in uniform bending, for which the moment which causes elastic flexural-torsional
buckling [7] is
M oo =

2 EI y

GJ +

L2

2 EI w
L2

(1)

in which E and G are the Youngs and shear moduli of elasticity, Iy, J and Iw are the
minor axis second moment of area, torsion and warping section constants, and L is the
length.
Most design codes provide approximations which allow for the effect of non-uniform
bending in simply supported beams loaded through the shear centre through
formulations of the type
M os = m M oo

(2)

in which the moment modification factor m is approximated by [1]

m =

1.7 M m
( M 22 + M 32 + M 42 )

2.5

(3)

in which Mm is the maximum moment and M2, M3, and M4 are the moments at the
quarter-, mid-, and three-quarter points, or by similar expressions [2, 6]. Alternatively,
m may be determined from Equation 2 by using the value of Mos determined by an
elastic buckling analysis. The effects of load height and end restraints are allowed for
approximately in some codes [1, 2] by replacing the beam length L in Equation 1 with
an effective length Le.

School of Civil Engineering

Research Report No R891

Buckling Analysis Design of Steel Frames

June 2008

The method of design by buckling analysis of the Australian code AS4100 [1] allows the
direct use of the results of elastic buckling analyses. For this, the maximum moment
Mob at elastic buckling is used in the equation

M bx

= 0.6 m
M sx

M
m sx
M ob

2
M

+ 3 m sx
M ob

(4)

to determine the nominal major axis moment strength Mbx, in which Msx is the nominal
major axis section capacity (reduced below the full plastic moment Mpx if necessary to
allow for local buckling effects). The variations of the dimensionless nominal strength
Mbx / Msx with the modified beam slenderness b = (Msx / Mob) and the moment
modification factor m are shown in Fig. 1. It can be seen that as the value of m
increases, the nominal design strengths Mbx approach the elastic buckling moments
Mob, reflecting the additional influence of non-uniform bending on inelastic buckling
[5, 7, 12].

2.2

Columns

The basic design code case for columns is the simply supported column in uniform
compression, for which the load which causes elastic flexural buckling is
N oc =

2 EI y
L2

(5)

Many design codes [1, 2, 6] provide approximations which allow for the effect of flexural
end restraints on simply supported columns by replacing the column length L with an
effective length Le. Some codes for hot-rolled steel structures do not allow for column
failure by flexural-torsional buckling [7, 13].
The method of design by buckling analysis of the Australian code AS4100 [1, 4] allows
the direct use of the results of elastic buckling analyses. For this paper, the AS4100
nominal design strength Ncy can be approximated by using
N cy / N s = 1.003 + 0.095c 0.8322c + 0.4093c 0.0584c 1

(6)

in which Ns is the nominal section capacity (reduced below the full plastic load Afy if
necessary to allow for local buckling effects), c = (Ns / Nom) is the modified column
slenderness and Nom is the elastic buckling load, although a more complicated general
formulation is given in the code. The variations of this dimensionless nominal strength
Ncy / Ns with the modified slenderness c are shown in Fig. 2.

School of Civil Engineering

Research Report No R891

Buckling Analysis Design of Steel Frames

2.3

June 2008

Beam-Columns and Frames

Design codes do not explicitly allow the use of a method of design by buckling analysis
for the out-of-plane design of beam-columns and frames. Instead, each member of a
frame is considered as a beam-column and designed independently by using out-ofplane interaction equations of the type

N max M max
+
=1
N cy
M bx

(7 )

in which Nmax and Mmax are the maximum nominal design actions (which are often
reduced by using capacity (or resistance) factors). Thus each beam-column is
considered first as a beam to determine Mbx and second as a column to determine Ncy,
before these are used in Equation 7.
When buckling analyses are used in the determination of Mbx and Ncy, then this
becomes the method of design by buckling analysis. This method is demonstrated in
the following sections for the two example frames shown in Figs 3 and 4, and compared
with the use of the normal code method of design without buckling analysis. For these
demonstrations and comparisons, the Australian design code AS4100 [1] is used, but
they could equally well be done by using other codes [2, 3, 6].

EXAMPLE FRAME 1

The members of the pin-based portal frame shown in Fig. 3a have the properties shown
in Fig. 5. The horizontal member has two equal transverse loads (initially equal to 1)
acting at the bottom flange. Warping is prevented at both ends and there are elastic
translational restraints of stiffness t = 10 N/mm acting at the load points at the bottom
flange. The in-plane reactions and moment distribution are shown in Fig. 3b.
For this frame, the design is controlled by the horizontal member. For this, this member
is first treated as the beam shown in Fig. 3c, and then as the column shown in Fig. 3d.
The results of these are then used in Equation 7.
The results (No DBA) of using the Australian code AS4100 [1] alone are summarized in
Table 1. AS4100 does not provide any guidance on warping restraint [7, 14, 15], while
the bottom flange loads and elastic translational restraints cannot be accounted for.
Because of this, the values determined for the maximum nominal design actions Nmax
and Mmax are quite conservative.
Also summarized in Table 1 are the results (DBA) of using the method of design by
buckling analysis. For this the elastic buckling moments Moo, Mos, Mob were determined
using the computer program PRFELB [8, 9] which is able to account for the warping
and translational restraints and the bottom flange loading. It can be seen that the
values determined for the maximum nominal design actions Nmax and Mmax are
significantly higher than those determined without using design by buckling analysis.
School of Civil Engineering
Research Report No R891

Buckling Analysis Design of Steel Frames

Quantity
Mob
Mos
Moo
m
Mob/m
Msx
s
Mbx
N/M
Ns
Nom
Ncy
Mmax
Nmax

June 2008

Table 1 Elastic Buckling and Design

Frame 1
Frame 2
Units
DBA
No DBA
DBA
No DBA
kNm
61.38
107.12

kNm
45.79
72.12

kNm
25.78
25.78
40.00
40.00
1.776
1.719
1.803
1.817

kNm
34.56
25.78
59.42
40.00
kNm
155.52
155.52
155.52
155.52
0.1931
0.1463
0.3127
0.2210

kNm
53.34
39.10
87.68
62.45
0.2000
0.2000
1.867
1.867

kN
1520
1520
1520
1520
kN
93.32
49.66
270.63
111.72
kN
89.03
48.34
243.59
105.82
kNm
42.87
30.29
47.20
26.74
kN
8.57
6.06
88.11
49.92

The program PRFELB is able to determine the elastic buckling load factors o (i.e. the
ratios of values of the actions at elastic buckling to the initial values) of beam-columns
and frames, as well as those of beams and columns. While these are not required for
design by buckling analysis, they are shown in Table 2. The value of o = 20940 for the
beam-column having the combination of the actions shown in Fig. 3c and d is slightly
different to that of o = 20990 for the frame. This is because the beam-column is
assumed to be prevented from twisting but free to rotate horizontally at both ends,
although in the frame end twisting and rotation are elastically restrained by the vertical
members.

Beam
Column
Beam-Column
Frame

Table 2 Buckling Load Factors o

Frame 1
22520
171200
20940
20990

Frame 2
200000
270600
128100
128400

EXAMPLE FRAME 2

The members of the pin-based portal frame shown in Fig. 4a have the properties shown
in Fig. 5. The horizontal member has a central transverse load (initially equal to 2)
acting at the centroid. Warping is prevented at both ends and there are rigid
translational and torsional restraints acting at the load point. Each vertical member has
an elastic translational restraint of stiffness t = 100 N/mm acting at the outer flange.
The in-plane reactions and moment distribution are shown in Fig. 4b. For this frame, the
design is controlled by the vertical members.
School of Civil Engineering
Research Report No R891

Buckling Analysis Design of Steel Frames

June 2008

The results (No DBA) of using the Australian code AS4100 [1] alone are summarized in
Table 1. AS4100 does not provide any guidance on warping restraint [7, 14, 15], while
the outer flange elastic translational restraints cannot be accounted for. Because of
this, the values determined for the maximum nominal design actions Nmax and Mmax are
quite conservative.
Also summarized in Table 1 are the results (DBA) of using the method of design by
buckling analysis. It can be seen that the values determined for the maximum nominal
design actions Nmax and Mmax are significantly higher than those determined without
using design by buckling analysis.
The elastic buckling load factors o are shown in Table 2. The value of o = 12810 for
the beam-columns having the combination of the actions shown in Fig. 4c and d is
slightly different to that of o = 12840 for the frame. This is because the beam-columns
are assumed to be prevented from twisting but free to rotate vertically at the top,
although in the frame end twisting and rotation are elastically restrained by the
horizontal member.

5.

CONCLUSIONS

Steel design codes [1, 2, 3, 6] do not provide sufficient information for the efficient
design of steel structures against out-of-plane failure, and what is provided is often
overly conservative. The method of design by buckling analysis explicitly permitted by
the Australian code AS4100 [1] corrects this situation for beams, but the extension of
this method to columns is only suggested [1, 4], while there is no guidance on how to
apply this method to the design of beam-columns and frames.
Beam design by buckling analysis uses the design code formulation for the member
nominal design strengths Mbx in terms of the section moment capacities Msx and the
maximum moment Mob at elastic buckling, accurate predictions of which may be
determined by available computer programs [8, 9, 10]. Column design by buckling
analysis is similar to beam design, in that it uses the design code formulation for the
column nominal design strengths Ncy in terms of the section compression capacities Ns
and accurate predictions of the elastic buckling load Nom which may also be obtained
from computer programs.
However, design codes do not provide formulations for the direct buckling design of
beam-columns, but instead use the separate results of beam design and column design
in interaction equations. The further direct extension to frames is also not possible,
because frames are not designed as a whole (except through the rarely used methods
of advanced analysis [11]) but as a series of individual members. This paper shows
how the method of design by buckling analysis can be used to design beam-columns
and frames as well as beams and columns. Two example frames are designed and
very significant economies are demonstrated when the method of design by buckling
analysis is used.

School of Civil Engineering

Research Report No R891

Buckling Analysis Design of Steel Frames

APPENDIX 1
[1]

June 2008

REFERENCES

SA. AS 4100-1998 Steel structures. Sydney: Standards Australia; 1998.

[2]
BSI. BS5950 Structural Use of Steelwork in Building. Part 1:2000. Code of practice
for design in simple and continuous construction: Hot rolled sections. London: British
Standards Institution; 2000.
[3]
BSI. Eurocode 3: Design of steel structures: Part 1.1 General rules and rules for
buildings, BS EN 1993-1-1. London: British Standards Institution; 2005.
[4]
SA. AS 4100-1998 Steel structures Commentary. Sydney: Standards Australia;
1998.
[5] Trahair, NS, Bradford, MA, Nethercot, DA, and Gardner, L. The behaviour and
design of steel structures to EC3, 4th edition. London: Taylor and Francis; 2008.
[6] AISC. Specification for structural steel buildings. Chicago: American Institute of Steel
Construction; 2005.
[7]

Trahair, NS. Flexural-torsional buckling of structures. London: E & FN Spon; 1993.

[8]
Papangelis, JP, Trahair, NS, and Hancock, GJ. PRFELB Finite element flexuraltorsional buckling analysis of plane frames, Sydney: Centre for Advanced Structural
Engineering, University of Sydney; 1997.
[9]
Papangelis, JP, Trahair, NS, and Hancock, GJ. Elastic flexural-torsional buckling
of structures by computer. Computers and Structures, 1998; 68: 125 - 37.
[10] HKS. Abaqus user manual. Pawtucket, RI, USA: Hibbitt, Karlsson and Sorensen;
2005
[11] Trahair, NS and Chan, S-L. Out-of-plane advanced analysis of steel structures.
Engineering Structures, 2003; 25: 1627-37.
[12] Nethercot, DA and Trahair, NS. Inelastic lateral buckling of determinate beams.
Journal of the Structural Division, ASCE, 1976; 102 (ST4): 701-17.
[13] Trahair, NS and Rasmussen, KJR. Flexural-torsional buckling of columns with
oblique eccentric restraints. Journal of Structural Engineering, ASCE, 2005; 131 (11):
1731-7.
[14] Vacharajittiphan P and Trahair NS. Warping and distortion at I-section joints.
Journal of the Structural Division, ASCE, 1974; 100 (ST3): 547-64.
[15] Pi, Y-L and Trahair, NS. Distortion and warping at beam supports. Journal of
Structural Engineering, ASCE, 2000; 126 (11): 1279-87.

School of Civil Engineering

Research Report No R891

Buckling Analysis Design of Steel Frames

APPENDIX 2

A
E
fy
G
Ix, Iy
Iw
J
L
Le
M
Mbx
Mm
Mmax
Mob
Moo
Mos
Mpxm
Msx
M 2, M 3, M 4
N
Ncy
Nmax
Nom
Noy
x, y
m
s
t

b
c
o

June 2008

NOTATION
area of cross-section
Youngs modulus of elasticity
yield stress
shear modulus of elasticity
second moments of area about the x, y principal axes
warping section constant
torsion section constant
length
effective length
bending moment
beam moment capacity
maximum moment in member
maximum nominal design moment
maximum moment at elastic buckling
Mob for a simply supported beam in uniform bending
Mob for a simply supported beam with shear centre loading
fully plastic moment about the x axis
section moment capacity
moments at quarter-, mid-, and three quarter-points
axial compression
column compression capacity
maximum nominal design compression
N at elastic buckling
Nom for a simply supported column
principal axes
moment modification factor
slenderness reduction factor
stiffness of translational restraint
modified beam slenderness
modified column slenderness
buckling load factor

School of Civil Engineering

Research Report No R891

10

Buckling Analysis Design of Steel Frames

June 2008

Dimensionless moment resistance Mbx /Msx

1.4
1.2

m = 1.0

1.0

1.5 2.0 3.0

Mob /Msx
0.8
0.6
0.4
0.2
0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

Dimensionless slenderness b = (Msx /Mob)

Fig. 1 Beam lateral buckling moment resistances of AS4100

School of Civil Engineering

Research Report No R891

11

2.0

Buckling Analysis Design of Steel Frames

June 2008

Dimensionless compression resistance Ncy/Ns

1.2

1.0

0.8

0.6

0.4

0.2

0
0

0.5

1.0

1.5

Dimensionless slenderness c = (Ns/Nom)

Fig. 2 Column buckling compression resistance of AS4100

School of Civil Engineering

Research Report No R891

12

2.0

Buckling Analysis Design of Steel Frames

June 2008

3
x

4
x
1

(a) Frame o = 20990

5000

1
5000

5000

( )

5000

(+)

Deflection restrained
at bottom flange
t = 10

2
x
2726

2274

5
( )

2726

3
x

5
x

4
x
1

(c) Beam o = 22520

2726

0.5453 2
x

3
x

4
x

5
0.5453
x

2
x

3
x

4
x

5
x

0.5453
2726

Fig. 3 Example Frame 1

School of Civil Engineering

Research Report No R891

(b) Moment distribution

2726

Deflection and
warping prevented

0.5453

0.5453

0.5453

13

(d) Column o = 171200

(e) Beam-column
o = 20940

Buckling Analysis Design of Steel Frames

June 2008

2
x
3

x
4

1964

(+)
x

x
4

535.7

5000
x

6 x

5000
0.0536

0.0536

2500 2500
(a) Frame o = 128400

0.0536

Deflection and warping prevented

Deflection restrained at outer flange t = 100

Deflection and twist rotation prevented

0.0536

1
x

535.7

3
x

(b) Moment distribution

0.0536

x
3
x

0.0536

1
1

(c) Beam o = 200000

535.7

(d) Column o = 270600 (e) Beam-column o = 128100

Fig. 4 Example Frame 2

School of Civil Engineering

Research Report No R891

x
3

14

Buckling Analysis Design of Steel Frames

June 2008

146 mm

10.9 mm

(a)

Beam and Dimensions

250 UB 37.3 Grade 300

256 mm
6.4 mm
(b)

Material properties

E = 2E5 N/mm2
G = 8E4 N/mm2
fy = 320 N/mm2
(c)

Section Properties

A = 4750 mm2
Iy = 5.66E6 mm4
Ns = 1520 kN

Ix = 55.7E6 mm4
J = 158E3 mm4
Msx = 155.52 kNm

Fig. 5 Beam Section and Properties

School of Civil Engineering

Research Report No R891

15

Iw = 85.2E9 mm6