Notice Changes introduced as a result of publishing ... · published in Thin-Walled Structures,...
Transcript of Notice Changes introduced as a result of publishing ... · published in Thin-Walled Structures,...
This is the author’s version of a work that was submitted/accepted for pub-lication in the following source:
Seo, Jung Kwan, Mahendran, Mahen, & Paik, Jeom Kee (2011) Numericalmethod for predicting the elastic lateral distortional buckling moment of amono-symmetric beam with web openings. Thin-Walled Structures, 49(6),pp. 713-723.
This file was downloaded from: http://eprints.qut.edu.au/49375/
c© Copyright 2011 Elsevier Ltd.
NOTICE: this is the author’s version of a work that was accepted forpublication in Thin-Walled Structures. Changes resulting from the pub-lishing process, such as peer review, editing, corrections, structural for-matting, and other quality control mechanisms may not be reflected inthis document. Changes may have been made to this work since itwas submitted for publication. A definitive version was subsequentlypublished in Thin-Walled Structures, [VOL 49, ISSUE 6, 2011 DOI10.1016/j.tws.2011.01.003
Notice: Changes introduced as a result of publishing processes such ascopy-editing and formatting may not be reflected in this document. For adefinitive version of this work, please refer to the published source:
http://dx.doi.org/10.1016/j.tws.2011.01.003
1
Numerical Method for Predicting the Elastic Lateral Distortional Buckling Moment of a Mono-symmetric Beam with Web Openings
Jung Kwan Seo,a,b,* Mahen Mahendrana and Jeom Kee Paikb
a) Faculty of Built Environment and Engineering, Queensland University of Technology, Brisbane, QLD 4000, Australia
b) The Lloyd’s Register Educational Trust Research Centre of Excellence, Pusan National University, Busan 609-735, Republic of Korea * Corresponding author. Tel.: +82-(0)51-510-2415. Email address: [email protected] (J.K. Seo)
Abstract: When used as floor joists, the new mono-symmetric LiteSteel beam (LSB)
sections require web openings to provide access for inspections and various services.
The LSBs consist of two rectangular hollow flanges connected by a slender web, and
are subjected to lateral distortional buckling effects in the intermediate span range.
Their member capacity design formulae developed to date are based on their elastic
lateral buckling moments, and only limited research has been undertaken to predict
the elastic lateral buckling moments of LSBs with web openings. This paper
addresses this research gap by reporting the development of web opening modelling
techniques based on an equivalent reduced web thickness concept and a numerical
method for predicting the elastic buckling moments of LSBs with circular web
openings. The proposed numerical method was based on a formulation of the total
potential energy of LSBs with circular web openings. The accuracy of the proposed
method’s use with the aforementioned modelling techniques was verified through
comparison of its results with those of finite strip and finite element analyses of
various LSBs.
Keywords: Cold-formed steel structures, Mono-symmetric beams, LiteSteel beams,
Web openings, Equivalent reduced web thickness, Elastic lateral distortional
buckling, Numerical method.
2
Nomenclature
30 aa − Coefficients defining web deflections
WD Plate rigidity ( xzzx DDD ,, )
1d Centreline dimension of web height E Young’s modulus of elasticity G Shear modulus of elasticity h Distance between flange shear centres
FI Polar second moment of flange area
WI Warping section constant
zy II , Second moments of area about y and z axes
FJ torsion section constant of hollow flange
[ ]K Total stiffness matrix
TB kk , Curvatures at the bottom and top of the web L Length M Moment
yM First yield moment
odM Elastic lateral distortional buckling moment
Fr Polar radius of flange gyration
WWWF SSSS,
, 21 θθ Stiffness parameters
t Wall thickness equt Equivalent reduced wall thickness
U Strain energy stored during in-plane buckling TB vv , Bottom and top flange deflections in y direction
Wv Web deflection in y direction V Work done during buckling
zyx ,, Principal axes
0z Defined in Fig. 8 δ Vector of web end deformations ν Poisson’s ratio σ Stress
BxxWxxTxx σσσ ,, Normal stresses of top flange, web and bottom flange
BzzWzzTzz σσσ ,, Transverse normal stresses of top flange, web and bottom flange
BxxWxxTxx τττ ,, Shear stresses of top flange, web and bottom flange
TB φφ , Mid-span values of FTFB φφ ,
FTFB φφ , Bottom and top flange twist rotations
3
1. Introduction
OneSteel Australian Tube Mills (OATM) was the first to develop the so-called ‘dog-
bone’ sections or hollow flange beams (HFBs) shown in Fig. 1(a). The efficiency of a
HFB lies in the combination of its torsionally rigid closed hollow flanges and an
economical manufacturing method based on simultaneous dual electric resistance
welding and roll-forming process [1,2]. OATM recently introduced the mono-
symmetric LiteSteel Beam (LSB) shown in Fig. 1(b) [3], which is the first of its
innovative cold-formed steel hollow flange sections suitable for large-scale
production. The new LSBs consist of two rectangular hollow flanges connected by a
slender web. There are 13 LSB sections whose depth (d) varies from 125 mm to 300
mm while their hollow flange width (bf) varies from 45 mm to 75 mm. The thickness
of high strength steel (t) used in LSBs varies from 1.6 mm to 3.0 mm. The LSB
section designation is based on d x bf x t, for example, 300x60x2.0 LSB. Due to their
light weight and cost-effectiveness, LSBs have become increasingly popular in
residential, industrial and commercial buildings. The LSB sections are used in these
building systems as floor joists, truss members and the like.
(a) HFB (b) LSB
Fig. 1 HFBs and LSBs and their Lateral Distortional Buckling Modes
Both HFB and LSB flexural members are subject to lateral distortional buckling
effects in their intermediate span ranges as shown in Fig. 1 [1,2,4]. When employed as
floor joists, LSB sections with higher depths and web openings are commonly used to
provide access for inspection and various services. The moment capacities of floor
bf
df
d
t
d1
Ro
30°
bf
d
t
d1
4
joists made of slender LSB sections with larger spans very much depend on their
elastic lateral distortional buckling moments. Several researchers have investigated
and summarised the elastic lateral buckling behaviour of channel section beams [5-8]
while Hancock et al. [9] and Bradford [10] have explored the effects of lateral
distortional buckling on the strength behaviour of conventional I-sections.
Pi and Trahair [11] adopted energy methods to investigate the elastic lateral
distortional buckling behaviour of simply supported HFBs subject to a uniform
moment, and developed a simple closed-form elastic buckling solution for HFBs
without web openings. They modified the classical flexural torsional buckling formula
by determining the effective torsional rigidity of the sections. They then adopted a
nonlinear inelastic method to investigate the lateral distortional buckling behaviour of
HFBs, and developed suitable member capacity equations for HFBs within the
guidelines of AS 4100 [12].
Moen and Schafer [13] developed elastic buckling approximations for cold-formed
steel structural members with holes, which they then employed to predict the elastic
local and distortional buckling moments of these members using a finite strip analysis
(FSA) program. For distortional buckling, the method simulates the C-section’s loss
in bending stiffness from the presence of a web hole within a distortional buckling
half-wave by modifying the cross-section thickness in the finite strip method. The
thickness of the entire web is reduced on the basis of the relationship between the web
bending stiffness and bending stiffness matrix terms of a finite strip element. The
distortional half-wavelength of the cross-section without holes is first determined
using the cross section of the column in a FSA program to generate an elastic
buckling curve. Hence Moen and Schafer [13] defined a half-wavelength by the
location of the distortional minimum, with the web thickness modified in the finite
strip method to account for the stiffness loss resulting from the holes. This
approximate method has been implemented successfully for C-section beams with
web holes, although it may have limitations in predicting the elastic buckling
moments of LSBs with web holes when subjected to a uniform bending moment.
Wang and Salhab [14] reported their numerical results for compression tests of a
lightweight panel using cold-formed thin-walled channel sections with perforated
webs. They also investigated the local buckling behaviour of web plates with
perforations, and their numerical results were employed to develop an approximate
method for calculating the equivalent thickness of the perforated web, which they
5
defined as the reduced thickness of a solid web plate that gives the same local
buckling strength as the perforated web. In their study, the effective thickness was
obtained through regression analysis of a large number of numerical results using a
finite element analysis program, ABAQUS. However, their approximate method is
applicable only for the three-dimensional modelling of perforated channel sections
with reduced web thickness.
At present, there is no accepted design method for LSB flexural members with web
openings. Such existing design methods as AISI [15] deal only with channels with
web openings and do not give the required elastic buckling moments and member
moment capacities for LSBs. Seo and Mahendran [16] recently developed new design
rules for determining the ultimate member moment capacities of LSBs with web
openings. These design rules were developed as a function of a non-dimensional
member slenderness parameter ( dλ ) defined as y
od
MM
, and therefore require elastic
lateral distortional buckling moments (Mod) for a given LSB and web opening
configuration. Hence there is a need for a general prediction method for LSB sections
and web opening configurations. The aim of the study reported in this paper was to
develop web opening modelling techniques and a numerical method for the prediction
of the elastic lateral buckling moments of LSBs with web openings.
The paper first proposes an equivalent reduced web thickness modelling method
for the elastic buckling analysis of LSBs with web openings. Four methods employing
four different equations for the prediction of equivalent web thicknesses are proposed,
and their accuracy is investigated using finite element analysis (FEA). In the second
section of the paper, the numerical energy-based buckling solution for LSBs without
web openings is extended to LSBs with web openings by employing the equivalent
web thickness approach used in FEA. An explicit buckling solution is obtained from a
formulation of the total potential energy for LSBs with circular web openings. The
accuracy of the proposed numerical method’s use with equivalent reduced web
thicknesses is verified by comparing its results with those of the FSA and FEA of
LSBs.
6
2. Proposed equivalent reduced thickness method for LSBs
2.1 Equivalent reduced thickness methods for LSBs
It is proposed that the effects of web hole configurations on the lateral distortional
buckling of LSBs can be approximated by reducing the cross-section thickness in
FSA and FEA using the four following methods, which employ four different
equations. Doing so allows numerical analyses to be undertaken with solid webs of
equivalent reduced thicknesses.
In the first method (see Fig. 2(a)), an equivalent reduced thickness (tequ) is
determined for the full height solid web element along the full member length, based
on the following equation.
web
webholewebequ A
tAAt )( −= . (1)
In the second method (see Fig. 2(b)), an equivalent reduced thickness (tequ) is
determined for the web hole region alone, based on the following equation.
webequ tSDt )/1( −= . (2)
In the third method (see Fig. 2(c)), an equivalent reduced thickness (tequ) is
determined for both the web and flange elements, based on the following equation.
)()(
flangeweb
webholewebfflangeequ AA
tAAtAt
+
−+= . (3)
Finally, in the fourth method (see Fig. 2(d)), an equivalent reduced thickness (tequ)
is determined for the web hole region alone, based on the following equation.
DLtADLt webhole
equ
)( −= . (4)
In the above equations, Aweb = web area = L× d1; Aflange = flange area = 2(L×bf ) +
2(L×df ); Ahole = the area of web holes = 2( D / 4) (L / S)π × ; d1 = the centreline
dimension of web height; D and S = the web hole diameter and spacing, respectively;
and L = the member span.
7
All four methods allow the effects of web hole size and spacing to be determined
through the use of a reduced thickness, and are based on the web hole area and
spacing, and the selected areas of web and/or flange elements, except for Method 2.
(a) Method 1 (b) Method 2 (c) Method 3 (d) Method 4
Fig. 2 Equivalent reduced web thicknesses in Methods 1 to 4
2.2 Verification of the proposed methods
In order to determine the accuracy of the proposed equivalent thickness methods,
the ideal finite element model of LSB without web openings developed by Kurniawan
and Mahendran [17,18] was used in this study for three LSB sections, 300x60x2.0
LSB (slender section), 250x75x3.0 LSB (compact section) and 300x75x3.0 LSB
(non-compact section). This shell finite element model incorporated ideal simply
supported end conditions and a uniform bending moment within the span. The model
was used with the equivalent plate thickness calculated from Methods 1 to 4 instead
of explicitly modelling the web holes. Hence it was considered the approximate FEA
model of LSBs with web openings used to determine the elastic lateral buckling
moments. Following web hole diameter (D) and Spacing (S) were chosen in this
study: D = 50, 100, 150 mm and S = 200, 250, 500 mm. Seo and Mahendran [19]
used similar ideal finite element models that included the actual web openings and
their spacing. These accurate FEA models were also used in this study to obtain the
elastic lateral buckling moments of the chosen three LSB sections and web opening
configurations.
Both accurate and approximate FEA models were developed using the S4R5 three
dimensional (3D) thin isoparametric quadrilateral shell element with four nodes and
five degrees of freedom per node, available in ABAQUS [20]. An element size of
approx. 5 mm×10 mm (width by length) was used to provide accurate results. The
8
accuracy of such models of LSB flexural members with and without web openings
has been verified by Kurniawan and Mahendran [17] and Pokharel and Mahendran
[21]. Fig.3 (a) shows a typical half-length finite element model of LSB with web
openings used in this study. It includes the details of the ideal loading and boundary
conditions used. The required uniform bending moment distribution within the span
was achieved by applying equal end moments using linear forces at the end support.
Figure 3 (b) shows the typical lateral distortional buckling mode obtained from the
elastic buckling analyses of LSB with web openings. Further details of FEA models
are given in [19].
(a) Finite Element Model
(b) Lateral Distortional Buckling Mode
Fig. 3 Finite Element Modelling of LSBs with Web Openings
9
Figs. 4 to 6 compare the elastic lateral distortional buckling moments obtained
from these approximate and accurate FEA models. In these figures, the horizontal axis
is the ratio of the web hole diameter to the centreline dimension of web height (D/d1)
and the vertical axis is the elastic lateral distortional buckling moment (Mod). The
results are also compared in tabular form in [19]. The ratio of Mod predicted by the
approximate and accurate finite element models was calculated in each case from
these results. The mean and COV of this ratio for Methods 1 to 4 are as follows:
0.9605 and 0.0375 (Method 1); 0.9265 and 0.1002 (Method 2); 0.6696 and 0.2911
(Method 3); 0.9831 and 0.0483 (Method 4). These results and Figures 3 to 5 confirm
that Methods 1 and 4 are accurate in predicting the elastic lateral distortional buckling
moments of the chosen slender, compact and non-compact LSB sections with circular
web openings using the approximate finite element models. As expected, Method 3
that includes both web and flange elements under-predicts Mod significantly and is
inaccurate.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7D/d1
0
20
40
60
80
100
Mod
(kN
m)
FE model no holes FE Model with holes FE model without holes (Method 1)FE model without holes (Method 2)FE model without holes (Method 3)FE model without holes (Method 4)
Mod=89.720kNm
M1
M3
M2
M4
250x75x3.0LSB (L=2000mm)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1D/S
0
20
40
60
80
100
Mod
(kN
m)
FE Model no holesFE Model with holes FE model without holes (Method 1)FE model without holes (Method 2)FE model without holes (Method 3)FE model without holes (Method 4)
M1
M3
M2
M4
250x75x3.0LSB (L=2000mm, D=150mm)Mod=89.72 kNm
(a) Mod vs. D/d1 (b) Mod vs. D/S
Fig. 4 Comparison of Mod from approximate and accurate Finite Element Models of
250 x 75 x 3.0 LSBs with web holes (Span = 2000 mm)
10
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7D/d1
0
40
80
120
Mod
(kN
m)
FE model no holes FE Model with holes FE model without holes (Method 1)FE model without holes (Method 2)FE model without holes (Method 3)FE model without holes (Method 4)
Mod=97.871kNm
M1
M3
M2
M4
300x75x3.0LSB (L=2000mm)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1D/S
0
40
80
120
Mod
(kN
m)
FE Model no holesFE Model with holes FE model without holes (Method 1)FE model without holes (Method 2)FE model without holes (Method 3)FE model without holes (Method 4)
M1
M3
M2
M4
300x75x3.0LSB (L=2000mm, D=150mm)
Mod=97.87 kNm
(a) Mod vs. D/d1 (b) Mod vs. D/S
Fig. 5 Comparison of Mod from approximate and accurate Finite Element Models of
300 x 75 x 3.0 LSBs with web holes (Span = 2000 mm)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7D/d1
0
10
20
30
40
Mod
(kN
m)
FE model no holes FE Model with holes FE model without holes (Method 1)FE model without holes (Method 2)FE model without holes (Method 3)FE model without holes (Method 4)
Mod=35.040kNm
M1
M3
M2
M4
300x60x2.0LSB (L=2000mm)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1D/S
0
10
20
30
40
Mod
(kN
m)
FE Model no holesFE Model with holes FE model without holes (Method 1)FE model without holes (Method 2)FE model without holes (Method 3)FE model without holes (Method 4)
M1
M3
M2
M4
300x60x2.0LSB (L=2000mm, D=150mm)
Mod=35.04 kNm
(a) Mod vs. D/d1 (b) Mod vs. D/S
Fig. 6 Comparison of Mod from approximate and accurate Finite Element Models of
300 x 60 x 2.0 LSBs with web holes (Span = 2000 mm)
The implication of these results is that the equivalent reduced web thicknesses
obtained based on Methods 1 and 4 can be used in the modelling of LSB flexural
members with web holes when employing available FSA and FEA programs, thereby
avoiding the complicated task of explicitly modelling the web hole configuration.
Method 4, which has a mean ratio of 0.9831, is considered to be the most accurate of
the four methods, and it can be employed in numerical modelling with a reduced
11
equivalent thickness in the web hole region alone. In applications in which a single
web element is to be used, Method 1 is recommended.
3. Elastic buckling moments of LSBs with proposed reduced thickness methods
and energy method
Also developed in the study reported in this paper was an accurate energy method-
based solution for the elastic lateral distortional buckling moments of mono-
symmetric LSB members with web openings subject to a uniform moment. The
numerical method employs the same theory and principles as those used by Trahair
[22] for HFBs without web openings. A similar approach is proposed here to derive
an explicit buckling solution for LSBs with web openings. For this purpose, an energy
method derivation was employed with the implementation of an equivalent web
thickness to represent the web opening configurations.
In the discussion of the proposed reduced thickness methods, it was shown that
approximate finite element models using equivalent reduced web thicknesses can be
used to predict elastic lateral distortional buckling moments and avoid the complex
modelling of web openings. Equation 1, which was based on Method 1, was
employed to include the equivalent web thickness in the total stiffness matrix terms.
This method was chosen instead of Method 4 as it allows a single web element of
constant thickness to be used in the derivations (Fig. 2). The method’s accuracy was
verified by comparing its results with those obtained from FSA and FEA.
12
3.1 Displacement field
Fig. 7 Cross-section of LSB deformations
The deformation of an LSB cross-section through buckling can be described
approximately by a displacement field with six degrees of freedom Bv , Bφ , Tv , Tφ ,
Bk , Tk T [5]. The cross-section deformations Bv , Tv , Bφ and Tφ are shown in Fig.
7, where Bv and Tv are the lateral displacements of the bottom and top flanges with
respect to the y axis, respectively, Bφ is the rotation at the bottom of the flange, and
Tφ is the rotation at the top of the flange. Bk and Tk are the curvatures in the z-y
plane at the bottom and top of the web, respectively. It is assumed that the rectangular
hollow flange cross-sections remain undistorted during deformation.
The displacements of the flanges and web in the x-direction (u) are obtained from
simple beam theory, and the displacements in the z and y directions (w, v) are
obtained from the kinematic conditions. These displacements may be written as
follows.
For the beam web:
0w)z,x(vv
0u
W
W
W
===
(5a)
For the bottom flange:
13
)x(vv)x(yw
zvu
BB
BB
x,BB
=φ=
−=
(5b)
For the top flange:
)x(vv)x(yw
zvu
TT
TT
x,TT
=φ−=
=
(5c)
The cross-sectional shape of the web is allowed to deform, and the web’s rotation
and lateral deflection are then coupled, with the displacements proceeding as follows.
With regard to the flanges, it can be assumed that the displacements and twists
vary sinusoidally lengthwise along the member and that the flanges do not change
shape. The buckling deformations of the flanges can be written as
L
xmsinL
xmsin
wvwv
wvwv
B
B
T
T
B
B
T
T
πδ=
π
=
. (6)
With regard to the web, it can be assumed that its out-of-plate buckling shape is a
cubic polynomial. A cubic buckling shape yields good results for beams that are
subject to equal end moments, but not for those subject to a uniformly distributed load.
A shape function for the cubic polynomial was employed in the current study to
represent the web buckling shape of LSBs subject to equal end moments. The flexural
displacement ( Wv ) of the web, of which the maximum value in Fig. 7 is Wv , is
obtained by assuming that the web buckles out-of-plane as a cubic polynomial curve,
and is given by
3
13
2
12
110
222
+
+
+=
dza
dza
dzaavw . (7)
It can be assumed that the displacements and twists vary sinusoidally lengthwise
along the member. The buckling deformations of the web can thus be written as
14
Lx
Lxv
vvv
B
T
B
T
B
T
B
T
πδπ
φφ
φφ
sinsin =
=
, (8)
where Tv , Bv , Tφ and Bφ are the maximum amplitudes of the buckling
displacements, and m is a positive integer representing the number of harmonics into
which the beam buckles.
The coefficients 0a , 1a , 2a and 3a are obtained by the compatibility identities
( )
( )
−−−−
=
∂∂
−
∂∂
−=
=
−=
=
−=
1
1
1
1
5.01
5.01
5.0
5.0
5.05.0
)(5.0)(5.0
5.0
5.0
)()(
1
1
1
1
dd
dhvdhv
zvd
zvd
vv
B
T
BB
TT
dz
w
dz
w
dzw
dzw
φφφφ
δ (9)
The coefficients 0a , 1a , 2a and 3a can be expressed as
−+
−−+=
221
21
221
21
21
21 1
1
1
10
ddhd
dhvva BTBT φφ (9a)
−+
−−+−=
221
43
221
43
43
43 1
1
1
11
ddhd
dhvva BTBT φφ (9b)
+
=
241
241 11
2dda BT φφ (9c)
+
−−=
241
241
41
41 1
1
1
13
ddhd
dhvva BTBT φφ (9d)
3.2 Solution for lateral buckling using the work resulting from buckling
If a beam is loaded under a constant moment, M, then the stress distribution will
vary linearly from the top to the bottom flange, as shown in Fig. 8.
15
Fig. 8 Stress distributions under bending moment
The bending moment is written as
∫σ=A
zdAM , (10)
and the values of the bending stress, σ , corresponding to the bending moment are
given by
−=σ
−=σ
)zz(IM
)dz(IM
0y
W
10y
F
. (11)
With regard to a beam subject to end moments, it is noted that for an assumed
shape function such as Eq. 7, the energy terms involving shear stress are zero; thus,
the external work (V) arising from lateral buckling (i.e., with the pre-buckling work
omitted) may be written as [22]
∫∫
∫∫
∫∫
σ+
+σ+
+σ=
L2
x,WWA
L2
x,B2
x,BFBA
L2
x,T2
x,TFTA
dxdA)v(21
dxdA))wv((21
dAdx))wv((21V
W
FB
FT
. (12)
3.3 Strain energy stored during buckling
The total potential energy of a LSB subjected to edge loading is the sum of strain
energy U and the potential energy of the applied load Ω :
Ω+=Π U . (13)
16
The strain energy of a three-dimensional isotropic medium referred to as arbitrary
orthogonal coordinates may be written as
dv)(21U yzyzxzxzxyxyzzyy
vxx γτ+γτ+γτ+εσ+εσ+εσ= ∫ . (14)
In accordance with the basic approximations of thin-plate theory, xzγ , yzγ and zσ
can be omitted. It is assumed that the strains and curvatures are much less than unity.
The finite-strain expressions for the in-plane strain components of the mid-surface are
given by
yxyxyxxyxy
2x
2x
2xyy
2x
2x
2xxx
wwvvuuvu
)wvu(21v
)wvu(21u
++++=γ
+++=ε
+++=ε
. (15)
In von Karman’s plate theory, only displacement gradients xw and yw are
expected to achieve significantly large amplitudes, and thus, among the nonlinear
terms in Eq. 15, only xw2 , yw2 and xw yw are retained. However, in the present
application to LSBs, the gradients of u and v, in addition to that of w, may become
large due to in-plane rotation, especially for the flange component. The terms xu and
yv are of a higher order than the other terms, and the second-order terms involving
xu and yv are thus ignored.
The utilisation of potential energy concepts in the establishment of relationships
for instability analysis presumes that pre-buckling deformations have taken place and
that an examination of the perturbed equilibrium state is being conducted. Thus, the
pre-buckling work ( Ω ), which is the product of the applied loads and their
corresponding deflection, is ignored in the buckling analysis. Hence,
U=Π . (16)
The strains in an arbitrary plate component and web location can be written as Eqs.
17 and 18.
For the flanges:
17
Txyy,Tx,Tx,Ty,TTxy
Ty2
x,T2
x,Ty,TTy
Tx2
x,T2
x,Tx,TTx
zw2)wwvu
zw)wu(21v
zw)wv(21u
−++=γ
−++=ε
−++=ε
,
xy,By,Bx,Bx,By,BBxy
y,B2
x,B2
x,By,BBy
x,B2
x,B2
x,Bx,BBx
zw2)wwvu
zw)wu(21v
zw)wv(21u
−++=γ
−++=ε
−++=ε
(17)
For the web:
xz,Wz,Wx,Wx,Wz,WWxy
zz,W2
x,W2
x,Wz,WWy
xx,W2
x,W2
x,Wx,WWx
yv2)vvwu
yw)vu(21w
yw)wv(21u
−++=γ
−++=ε
−++=ε
(18)
Drawing on the flange displacement assumption and ignoring the small terms, the
total potential energy of the top and bottom flanges and web can be obtained as
follows.
( )( )
( )( )dxdAww2)y(v21dxJG
21dxvIE
21
dxdAww2)y(v21dxJG
21dxvIE
21U
Bf
Tf
A
L
0ByBxBxy
2Bx
2BxxT
2Bx
L
0fB
2Bx
L
0zB
A
L
0TyTxTxy
2Tx
2TxxT
2Tx
L
0fT
2Tx
L
0zTFlange
∫ ∫∫∫
∫ ∫∫∫
τ+φ+σ+φ++
τ+φ+σ+φ+=(19)
( )
dxdzv12
t4GvvvD2vDvD21
dxdzv)v1(2vvv2vvv1
Et21
dxdz)wu(2
v1wvu2wuv1
Et21
dv)(21U
2Wz
3
WzWxxz2Wzz
2Wxx
2WzWzWx
2Wz
2Wx2
2WxWzWzWx
2Wz
2Wx2
WxzWxzWzWzv
WxWxWeb
∫∫
∫∫
∫∫
∫
++++
−+++−
+
+
−+++
−=
γτ+εσ+εσ=
(20)
In the above,
)v1(2EG,
)1(12Et
D,)1(12
EtD,
)1(12Et
D 2
3equ
xz2
3equ
z2
3equ
x +=
ν−=
ν−=
ν−= .
The proposed reduced thickness (tequ) of the web can be expressed as (based on Eq.1)
)Sd4D1(tt
1
2
equπ
−= (21)
18
where t = the web thickness and is also the flange thickness of a LSB, D = the web
opening diameter, S = web opening spacing and d1= the centreline dimension of web
height.
This deformation web pattern involves both lateral and vertical deflections
( Wv , Ww ) and rotation. If the cross-sectional profile of the web at any transverse
cross section is prohibited from curving, then the web is regarded as rigid, and the
rotation and lateral deflection are coupled [6, 23]. The displacements in the flexible
web case are the same as those in the classical (rigid web) case except for the lateral
displacement, Wv . The vertical displacement ( xxWv , ), however, is quite small and can
thus be ignored.
3.4 Energy equation
The total potential energy is
VU −=Π . (22)
Transformation of the total potential energy given by Eq. 21 into the desired
stiffness expressions requires selection of the displacement functions to describe the
behaviour of the element. Each displacement component is assumed to take the
general form
iiN δ=δ , (23)
where iN is a function of the x coordinate. Substituting Eq. 23 into Eq. 22 yields the
following expression for the total potential energy.
δδ=Π ]K[21 T . (24)
The symmetric total stiffness matrix is as follows.
[ ]
=
44342414
34332313
24232212
14131211
ji
KKKKKKKKKKKKKKKK
K (25)
19
[ ]
−+
+
−+
−
−
+−
−
−
−
+−
+−
−−++−
−
=
θθθ
θθθ
θθθθ
θ
21
WW2
21
21
2WFF
Fo32
211
WW21
2
F
Fo32
211
WW2
21
2WF1
2
12
11WW21
1
ij
hd
1S21
h3d
hd
1S21S
AI
hM2
hd
hd
SS21
hd
1S0
AI
hM2
hd
hd
SS21
hd
S21S
hd
S0
hd
1Shd
SS2S2ShM2
00hM2S
K
(26)
In the above matrix, 2z
2
1 LEIS π
=θ , 3
1
2
231
2
2
2 Sd4Dtt
)1(12dEL24S
π−
ν−π=θ and
2F
WF hGJ2S = ,
3
1
2
1WW Sd4
Dttd3G2S
π−= .
This equation can be expressed as an explicit form of
[ ] 0Kdet = , (27)
in which case Eq. 26 can be expressed as follows.
0KKKKKKK)KKKKKK()KK( 44223114433221112
24433
2342211
23412 =−++− (28)
0KKKKKKK
MKKhr4
M
h4KKKK
hr2
hr2
Sd4Dtt
h3Gd4
Sd4Dtt
h3G4
Sd4Dtt
)1(12hdEL24
Sd4Dtt
)1(12hdEL24
442231144332211
22113
2F
2
444332211
2
3F
2
3
2F
2
1
2
31
3
1
2
2
3
1
2
231
2
23
1
2
2221
2
2
=−+
−
+
−
−
π−+
π−−
π−
ν−π−
π−
ν−π
(29)
3.5 Verification of the proposed numerical method
To demonstrate the accuracy of the equivalent web thickness terms’ inclusion in
the energy method solution, the elastic lateral buckling moments of a single
20
symmetric LSB were computed using the proposed energy method and the FSA
(THIN-WALL) [24] and FEA programs (ABAQUS) [20]. The influence of web
openings on the lateral distortional buckling of open thin-walled cross-sections can be
approximated by modifying the cross-section thickness in ABAQUS and THIN-
WALL using Method 4 (Eq. 4), which was found to be the most accurate of the four
methods introduced in the previous section. Unlike the energy method solution,
Method 4 can be used with FEA and FSA.
Fig. 9 presents a comparison of the lateral distortional buckling modes obtained
from FEA with and without web hole modelling (Figs. 9(a) and (b)) and from FSA
without web hole modelling (Fig. 9(c)) for the most slender section with larger web
holes. It can be seen that the lateral deformation of the LSB with web hole modelling
is slightly greater than that without web hole modelling.
Six LSB members subjected to equal end moments were analysed using the three
aforementioned methods. They included two non-compact sections (300x75x3.0 LSB
and 200x60x2.0 LSB), two slender sections (300x60x2.0 LSB and 250x60x2.0 LSB)
and two compact sections (250x75x3.0 LSB and 150x45x2.0 LSB). Numerous spans
in the range of 500 to 10,000 mm were considered. The results of the comparison are
presented in Fig. 10.
(a) (b) (c)
Fig. 9 Lateral distortional buckling modes for 300 x 60 x 2.0 LSB (D = 150 mm, S =
250 mm): (a) Finite element model with web holes; (b) Finite element model without
web holes; (c) Finite strip model without web holes
21
An explicit buckling solution was obtained from a formulation of the total potential
energy for LSBs with circular web openings (from Eq. 28). The accuracy of the
proposed numerical method’s use with an equivalent reduced web thickness was
verified by comparing its results with those obtained from FSA and FEA of LSBs.
The results demonstrate that the proposed explicit solutions based on the energy
method are able to predict with accuracy the elastic lateral distortional buckling
moments of LSBs with web holes as shown in Fig. 10. The mean and corresponding
CoV of the ratio of elastic lateral distortional buckling moments obtained with the
proposed method and in FSA (THIN-WALL) are 1.022 and 0.040, respectively, thus
demonstrating good agreement.
0 2000 4000 6000 8000 10000Span L (mm)
0
100
200
300
Elas
tic B
uckl
ing
Mom
ent,
Mod
(kN
m)
300x75x3.0 LSBFEA (ABAQUS)FSA (THIN-WALL)Present Study
0 2000 4000 6000 8000 10000Span L (mm)
0
20
40
60
80
100
Elas
tic B
uckl
ing
Mom
ent,
Mod
(kN
m)
250x60x2.0 LSBFEA (ABAQUS)FSA (THIN-WALL)Present Study
(a) 300 x 75 x 3.0LSB (D = 150 mm, S = 250 mm) (b) 250 x 60 x 2.0LSB (D = 150 mm, S =
250 mm)
0 2000 4000 6000 8000 10000Span L (mm)
0
20
40
60
80
Elas
tic B
uckl
ing
Mom
ent,
Mod
(kN
m)
200x60x2.0 LSBFEA (ABAQUS)FSA (THIN-WALL)Present Study
0 2000 4000 6000 8000 10000Span L (mm)
0
40
80
120
160
Elas
tic B
uckl
ing
Mom
ent,
Mod
(kN
m)
300x60x2.0 LSBFEA (ABAQUS)FSA (THIN-WALL)Present Study
(c) 200 x 60 x 2.0LSB (D = 100mm, S = 250mm) (d) 300 x 60 x 2.0LSB (D = 150 mm, S =
250 mm)
22
0 2000 4000 6000 8000 10000Span L (mm)
0
50
100
150
200
250
Elas
tic B
uckl
ing
Mom
ent,
Mod
(kN
m)
250x75x3.0 LSBFEA (ABAQUS)FSA (THIN-WALL)Present Study
0 2000 4000 6000 8000 10000Span L (mm)
0
20
40
60
80
100
Elas
tic B
uckl
ing
Mom
ent,
Mod
(kN
m)
150x45x2.0 LSBFEA (ABAQUS)FSA (THIN-WALL)Present Study
(e) 250 x 75 x 3.0LSB (D = 150 mm, S = 250 mm) (f) 150 x 45 x 2.0 LSB (D = 50 mm, S =
250 mm)
Fig. 10 Comparison of elastic lateral buckling moments obtained from FEA, FSA and the proposed numerical method
All of the comparisons are based on the equivalent web thickness method.
However, as discussed in Section 2, approximate finite element models based on solid
webs with an equivalent reduced web thickness provide accurate buckling solutions
compared with accurate finite element models that include actual web hole
configurations. Hence, we can also conclude that employing the same equivalent web
thickness approach with energy method buckling solutions is also accurate.
In summary, the proposed numerical method, which employs an equivalent web
thickness approach, provides accurate elastic buckling solutions for LSBs with web
openings, except for those with a very short span. For such short beams, yielding and
local buckling become more important, and the current solution’s loss of accuracy
will have only a small effect on the prediction of beam strength with varying web hole
configurations. It is therefore recommended that the numerical solution be used to
determine the elastic lateral buckling moments of LSBs with web openings without
the need for more complex FEA or FSA programs.
5. Conclusion
This paper has presented the details of an investigation into the elastic lateral
distortional buckling behaviour of LSBs with circular web openings that are subject to
equal and opposite end moments using FEA, FSA and energy methods.
An equivalent thickness method for the elastic buckling analysis of LSBs with web
openings was proposed. Four equations for predicting equivalent plate thicknesses
were proposed, and their accuracy was investigated using FEA. Elastic lateral
23
distortional buckling moments from approximate FEA based on solid web elements
with an equivalent reduced thickness were compared with those from accurate FEA
with actual web hole and spacing configurations, and the results show that two of the
proposed methods can be successfully employed with approximate finite element
models to predict these moments.
This paper has also presented an explicit elastic buckling numerical solution that
was obtained from a formulation of the total potential energy of LSBs with circular
web openings. The accuracy of the proposed numerical method’s use with an
equivalent reduced web thickness was verified through comparison of its results with
those of the FSA and FEA of LSBs. The numerical solution derived herein can be
adopted as the basis for an advanced theoretical method of predicting the nonlinear
lateral distortional buckling strength of LSBs with varying web opening
configurations. It is recommended that this solution be employed to determine the
elastic lateral buckling moments of LSBs with circular web openings, with no need
for the use of more complex FEA or FSA programs.
Acknowledgments
The authors would like to thank the Australian Research Council and OATM for
their financial support and the Queensland University of Technology for providing the
necessary research facilities and support for this research project. They would also
like to thank Mr Ross Dempsey, Manager of Research and Testing at OATM, for his
technical contributions and his overall support during the different phases of the
project.
References
[1] Dempsey, R.I. (1990), Structural Behaviour and Design of Hollow Flange Beams,
Proc. of the Second National Structural Engineering Conference, Institution of
Engineers, Adelaide, Australia, pp. 327-335.
[2] Dempsey, R.I. (2001), Hollow Flange Channel Development of the Product Range,
Investigation Report, Warren Brown Associates, Brisbane, Australia.
[3] Onesteel Australian Tube Mills (OATM) (2008), LiteSteel Beam Publication,
Brisbane, Australia.
24
[4] Anapayan, T., Mahendran, M. Mahaarachchi, D. (2011) Lateral Distortional
Buckling Tests of a New Hollow Flange Channel Beam, Thin-Walled Structures,
Vol. 49, pp. 13-25.
[5] Galambos, T.V. (1988), Guide to Stability Design Criteria for Metal Structures,
4th edition, John Wiley and Sons, New York, USA.
[6] Timoshenko, S.P and Gere, J.M. (1961), Theory of Elastic Stability, 2nd edition,
Engineering Societies Monographs, McGraw-Hill, New York, USA.
[7] Trahair, N. S. (1993), Flexural-Torsional Buckling of Structures, E & FN Spon,
London.
[8] Vlasov, V.Z. (1961), Thin-Walled Elastic Beams, 2nd edition. Israel Program for
Scientific Translation, Jerusalem, Israel.
[9] Hancock, G.J., Bradford, M.A. and Trahair, N.S. (1980), Web Distortion and
Flexural-Torsional Buckling, Journal of the Structural Division, ASCE, Vol.
106(ST7), pp. 1557-1571.
[10] Bradford, M.A. (1992), Lateral-Distortional Buckling of Steel I-Section
Members, Journal of Constructional Steel Research, Vol. 23, pp. 97-116.
[11] Pi, Y.L. and Trahair, N.S. (1997), Lateral-Distortional Buckling of Hollow
Flange Beams, Journal of Structural Engineering, ASCE, Vol. 123(6), pp. 695-702.
[12] Standards Australia (SA) (1998), Steel Structures, AS 4100, Sydney,
Australia.
[13] Moen, C.D. and Schafer, B.W. (2008), Simplified Methods for Predicting
Elastic Buckling of Cold-Formed Steel Structural Members with Holes, Proc. of
9th International Conference on Cold-Formed Steel Structures, St. Louis, MO,
USA, pp. 17-31.
[14] Wang, Y.C. and Salhab, B. (2009), Structural Behaviour and Design of
Lightweight Structural Panels Using Perforated Cold-Formed Thin-Walled
Section under Compression, Steel Structures, Vol. 9, pp. 57-67.
[15] American Iron and Steel Institute (AISI) (2006), Specification for the Design
of Cold-formed Steel Structural Members, AISI, Washington, DC, USA.
25
[16] Seo, J.K. and Mahendran, M. (2009), Member Moment Capacities and
Design of LSB Floor Joists Containing Web Openings, Research Report,
Queensland University of Technology, Brisbane, Australia.
[17] Kurniawan, C. and Mahendran, M. (2009) Lateral Buckling Strength of
Simply Supported LiteSteel Beams Subject to Moment Gradient Effects, ASCE J.
Structural Engineering, Vol.135, No.9, pp.1058-1067.
[18] Kurniawan, C. and Mahendran, M. (2009) Elastic Lateral Buckling of Simply
Supported LiteSteel Beams Subject to Transverse Loading, Thin-Walled
Structures, 47, pp.109-119
[19] Seo, J.K. and Mahendran, M. (2009), Development of an Equivalent Web
Thickness Method for the Design of LSB Floor Joists with Web Openings,
Research Report, Queensland University of Technology, Brisbane, Australia.
[20] Hibbitt, Karlsson and Sorensen (HKS) (2007). ABAQUS User’s Manual,
Pawtucket, RI, USA.
[21] Pokharel, N. and Mahendran, M. (2009) Finite Element Analysis of LSB
Floor Joists Containing Web Openings, Research Report, Queensland University
of Technology, Brisbane, Australia.
[22] Trahair, N. S. (1995), Elastic Lateral-distortional Buckling of Hollow Flange
Beams, Investigation Report S1009, School of Civil and Mining Engineering,
University of Sydney, Sydney, Australia.
[23] Pi, Y. L., Trahair, N. S. and Rajasekaran, R.(1992), Energy Equation for
Beam Lateral Buckling, Journal of Structural Engineering, Vol. 118(6), pp. 1463-
79.
[24] Hancock, G.J. and Papangelis, J.P. (1994), THIN-WALL User’s Manual,
Centre for Advanced Structural Engineering, University of Sydney, Sydney,
Australia.