You are on page 1of 12

Thin-Walled Structures 40 (2002) 853–864

www.elsevier.com/locate/tws

Buckling behavior of cold-formed zed-purlins


partially restrained by steel sheeting
Zhi-ming Ye a, Roger J. Kettle b, Long-yuan Li b,∗, Benjamin
W. Schafer c
a
Department of Civil Engineering, Shanghai University, Shanghai, 200072, PR China
b
School of Engineering and Applied Science, Aston University, Birmingham, B4 7ET, UK
c
Department of Civil Engineering, Johns Hopkins University, Baltimore, MD 21218, USA

Received 15 September 2001; received in revised form 26 April 2002; accepted 26 April 2002

Abstract

This paper presents a study on the buckling behaviour of purlin-sheeting systems under
wind uplift loading. The restraint provided by the sheeting to the purlin is modeled by using
two springs representing the translational and rotational restraints. The analysis is performed
using finite strip methods in which the pre-buckling stress is calculated based on the same
model used for the buckling analysis rather than taken as the ‘pure bending’ stress. The results
obtained from this study show that, for both local and distortional buckling, the restraints have
significant influence on the critical loads through their influence on the pre-buckling stress
rather than directly on the buckling modes. While for lateral-torsional buckling, the influence
of the restraints on the critical loads is mainly due to their influence on the buckling modes
rather than the pre-buckling stress.  2002 Elsevier Science Ltd. All rights reserved.

Keywords: Cold-formed; Purlin; Steel; Thin-walled; Restrained; Sheeting; Buckling; Instability

1. Introduction

Cold-formed steel members are widely used as purlins or rails in farming and
industrial buildings, the intermediate members between the main structural frame and
the corrugated roof or wall sheeting. The shape of the cross-section of the members is


Corresponding author. Tel.: +44 121 3593611; fax: +44 121 3333389.
E-mail address: l.y.li@aston.ac.uk (L.-y. Li).

0263-8231/02/$ - see front matter  2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 2 6 3 - 8 2 3 1 ( 0 2 ) 0 0 0 2 9 - 0
854 Z.-m. Ye et al. / Thin-Walled Structures 40 (2002) 853–864

mostly that of zed, channel or sigma. These types of cross-sections are inherently
sensitive to local, distortional and lateral–torsional buckling. As a result of current
trends, these sections are becoming more slender and more highly profiled and there-
fore in some cases more prone to local, and particularly distortional buckling. Histori-
cally, the load-bearing capacity of the purlin-sheeting system has been determined
by full-scale testing [1–6]. However, with the increasing use of various new cladding
systems, and the ever increasing cost of testing, there is now a real need for the
development of analytical design procedures.
Early research on the cold-formed steel section was focused on the buckling
behaviour of various individual sections [6,7], in which the restraint of the sheeting
on the member was assumed to be either ignored or fully provided. For example,
Hancock [8] developed a finite strip method which can be used to carry out local,
lateral–torsional and distortional buckling analyses of various thin-walled sections
subjected to pure bending or uniform compression or a combination of these two
stress states; Davies et al. [9,10] developed a generalized beam theory which can
also be used to carry out buckling analyses of various sections subjected to uniformly
distributed loads. More recently, Schafer [11] developed a general finite strip buck-
ling analysis program that includes the option of nodal spring restraints and so can
be used to carry out local, lateral–torsion and distortional buckling analyses of purlin-
sheeting systems for any prescribed pre-buckling stress.
In this paper, the buckling problem of purlin-sheeting systems subjected to wind
uplift loading is investigated. The focus of the study is on the influence of the
restraints provided by the sheeting on the buckling behaviour of the purlin. The
buckling analysis is performed using finite strip methods [8,11] in which the calcu-
lation of the pre-buckling stress is based on the same analysis model as the buckling
analysis [12].

2. Model for pre-buckling stress analysis

Consider a purlin-sheeting system subjected to wind uplift loading as shown in


Fig. 1a. The restraints provided by the sheeting to the purlin are assumed to take
two forms, a translational spring, ks, and a rotational spring, kr, as shown in Fig. 1b.
By setting the origin of the coordinate system (y, z) at the centroid of the cross-
section, with the x axis along the longitudinal direction of the beam, and the y and
z axes in the plane of the cross-section (see Fig. 1c), the displacement forms of the
torsion and bending equilibrium equations can be expressed as follows [12]:
d4wj 2d2wj ksg2 4krg2 qg3
g1L4 ⫺L ⫹ (w ⫹ w) ⫹ w ⫽⫺ (1a)
dx4 dx2 E j Ed2 j E
d4w kslz qlyz
L4 ⫹ (w ⫹ w) ⫽ ⫺ (1b)
dx4 E j E
d4v kslyz qly
L4 ⫽ (wj ⫹ w) ⫹ (1c)
dx4 E E
Z.-m. Ye et al. / Thin-Walled Structures 40 (2002) 853–864 855

Fig. 1. Analytical model of purlin-sheeting system. (a) purlin-sheeting system, (b) analytical model, (c)
coordinate system, (d) degrees of freedom of displacements.

in which,
jd E⌫ E(dL)2 EadL2
wj ⫽ , g1 ⫽ , g 2 ⫽ , g3 ⫽ (2a)
2 GJL2 4GJ 2GJ
IzL4 Iy L 4 IyzL4
lz ⫽ , l ⫽ , l ⫽ , (2b)
IyIz⫺I2yz y
IyIz⫺I2yz yz
IyIz⫺I2yz
where ks and kr are the translational and rotational spring constants, E is Young’s
modulus, Iy and Iz are the second moments of the cross-section area about y and z
axes, Iyz is the product moment of the cross-section area, E⌫ is the warping stiffness,
GJ is the torsion stiffness, a is the distance between the loading point and the web
centre line, L is the span length of the beam, q is the density of the uniformly
distributed load, j is the angle of twist, v and w are the deflections of the beam in
y and z directions, respectively (see Fig. 1d). The detailed derivation of Eq. 1a,b,c
can be found in [12] and thus is not given here.
The longitudinal stress at any coordinate point (x, y, z) in the pre-buckling state
can be calculated based on the bending and warping stress formulas as follows:
856 Z.-m. Ye et al. / Thin-Walled Structures 40 (2002) 853–864

sx ⫽ sxm ⫹ sxw (3)


in which,
MzIy⫺MyIyz MyIz⫺MzIyz d2j
sxm ⫽ y⫹ z, sxw ⫽ ⫺Ew 2 (4)
IyIz⫺Iyz
2
IyIz⫺Iyz
2
dx
where sxm is the bending stress, My and Mz are the bending moments about y and
z axes which can be calculated based on the formula of asymmetrical bending of
the beam in terms of the displacement functions, sxw is the warping stress, and w
is the sectorial co-ordinate and is calculated by


s
w ⫽ hds (5)
o
where h is the perpendicular distance from the shear centre to a tangent at the point
under consideration and s is the distance from some chosen origin to the same point
measured along the middle line of the section.
For given boundary conditions the set of differential Eq. 1a,b,c can be solved
analytically or numerically and thus the pre-buckling stress can be determined for a
given loading density q.

3. Model for buckling analysis

The spring model used for the buckling analysis is the same as that used for the
pre-buckling stress analysis. The buckling analysis is performed using the finite strip
method, in which the finite element mesh is only applied to the cross-section and
so the purlin is actually subdivided into a number of strips. The finite strip method
was originally developed by Cheung [13] and has been extended by Hancock [14]
and Schafer [15] to cold-formed steel members.
For the simply supported purlin the strip can be assumed to be free to deform both
in its plane (membrane displacements) and out of its plane (flexural displacements) in
a single half sine wave over the length of the section being analyzed. Hence, the
three-dimensional problem is reduced to a two-dimensional one. For the buckling
analysis, the assumed single half sine wave also satisfies the simply supported bound-
ary conditions, which together with the finite element discretization in the cross-
section plane, can include all types of buckling modes ranging from local to lat-
eral–torsion.
The finite strip equation for the linear elastic buckling analysis may be obtained
by means of energy methods [13] which, after a tedious derivation, leads to the
following algebraic eigenvalue problem:
([Kp] ⫹ [Ks]⫺l[G]){X} ⫽ {0} (6)
where [Kp] and [Ks] are the stiffness matrices corresponding to the purlin and the
Z.-m. Ye et al. / Thin-Walled Structures 40 (2002) 853–864 857

springs, [G] is the geometrical stiffness matrix resulting from the pre-buckling stress,
l is the loading proportional factor (the actual load is q=lqo, where qo is the refer-
enced loading density for which the fully restrained purlin has a maximum com-
pression stress equal to the yield stress), and {X} represents the vector of displace-
ments describing the mode of buckling (the eigenvector). Note that the spring
restraints have influence on both the value and distribution of the pre-buckling stress
[12] and so the geometrical stiffness matrix [G] in fact is an indirect function of the
two spring constants. The complete derivation of the stiffness matrices can be found
in Schafer [15]. For thin-walled sections the linear elastic buckling analysis has been
demonstrated to be very effective and able to provide reasonable results [16].

4. Results and discussion

As an example, a simply supported purlin, whose section dimensions are: web


depth d=120 mm, flange width b=50 mm, lip length c=15 mm, thickness t=1.5 mm,
has been investigated. Fig. 2 shows the normalized pre-buckling stress distribution
in the cross-section of the purlin when it is subjected to a uniformly distributed uplift
load for four different restraint cases and three different loading cases. All the stresses
are normalized with respect to the maximum compressive stress in the case of a=b/2
and ks=kr=⬀. It can be seen from the figure that the spring constants have a significant
influence not only on the value of the maximum compressive stress (see Table 1)
but also on the stress distribution.
For cases where kr=⬀ there is no torsion and so no warping stress. Therefore, the
stress is pure bending stress, although the bending itself is not symmetric because
the cross-section of the purlin is asymmetric. When both spring constants are infinite
the bending occurs only in the vertical plane and so the loading position does not
affect the stress results. For cases where kr=0, torsion is generated because both the
vertically applied load, q, and the horizontal reaction force of the translational spring,
ks(w+wf), do not pass through the shear centre. Consequently, warping stresses are
generated and so the stress distribution is not symmetric about the neutral axis. It
is interesting to note from the stress results shown in Fig. 2 that the translational
spring is more important than the rotational spring in reducing the maximum tensile
and compressive stresses. Furthermore, it is apparent that in most cases the warping
stress is compressive, except for the case of a=b/4, ks=⬀ and kr=0 where it is actually
tensile. It can, therefore, be concluded that as far as buckling is concerned the worst
case is a purlin with no restraints, whereas the best case is when the purlin is fully
restrained in both translational and rotational directions. For a purlin partially
restrained, the membrane stress result is in between these two extreme cases.
For any given spring constants ks and kr, the pre-buckling stress can be calculated
and then input into the finite strip analysis program to obtain a buckling curve. Fig.
3 shows a typical buckling curve for the purlin with spring constants ks=20.5 N/m
and kr=7380 kN-mm. It can be seen from the figure that there are two minimum
points that represent a local buckle, occurring at a critical load l=2.27 over a half-
wavelength of 65 mm and a distortional buckle, occurring at a critical load l=1.87
858 Z.-m. Ye et al. / Thin-Walled Structures 40 (2002) 853–864

Fig. 2. Distribution of pre-buckling stress for different loading and restraint cases (L=5 m, tensile stress
is plotted above the flange or on the right side of web or lips).

Table 1
Normalized maximum compressive stress (L=5 m)

Case ks=kr=0 ks=⬀, kr=0 ks=0, kr=⬀ ks=kr=⬀

a=b/4 2.2756 1.1732 2.2756 1.0000


a=b/2 2.5563 1.0978 2.2756 –
a=3b/4 3.3229 1.7689 2.2756 1.0000

over a half-wavelength of about 400 mm. Lateral–torsional buckling is found to occur


at relatively long half-wavelengths (⬎1500 mm) with lower critical loads (l⬍1.87).
The influence of the two spring constants on the buckling behaviour of the purlin
is shown in Fig. 4. The results show that, for local buckling ks is the significant
influence with kr has almost no influence. However, for lateral-torsional buckling kr
seems to be more influential than ks while for distortional buckling, ks and kr have
a mixed influence as they interfere with each other.
Z.-m. Ye et al. / Thin-Walled Structures 40 (2002) 853–864 859

Fig. 3. Relationship between critical stress and the half-wave length (L=5 m, a=b/2, ks=20.5 N/m,
kr=7380 kN-mm).

Fig. 4. Buckling curves for different restraint cases (L=5 m, a=b/2).


860 Z.-m. Ye et al. / Thin-Walled Structures 40 (2002) 853–864

The influence of the spring restraints on the buckling behaviour of the purlin can
be illustrated from two aspects. The first is their influence on the pre-buckling stress
and the second is their influence on the buckling modes. In order to demonstrate
these two influences, Fig. 5 shows the buckling curves in which all four cases have
the same pre-buckling stress corresponding to the case ks=kr=⬀. It is surprising that
for the same pre-buckling stress, the four radically different restraint cases produce
almost no differences in the local and distortional buckling loads (the half-wave-
length is changed by fixing kr, but leaving ks free), the only difference between them
is in lateral–torsional buckling. This implies that the increase in critical loads related
to the local and distortional buckling shown in Fig. 4 is mainly due to the change
in the pre-buckling stress (see Table 1) attributable to the particular restraints.
As is shown in Fig. 2, the restraints affect not only the value of the maximum
compressive stress, but also the distribution of stress within the cross-section. The
influence of the change of stress distribution on the buckling behaviour of the purlin
is shown in Fig. 6, in which all four cases have the same maximum compressive
stress but the stress distributions are actually obtained from their individual cases.
It is evident from this figure that the change in the stress distribution has a significant
influence on the local and distortional buckling. The comparison of the four buckling
curves shown in Fig. 6 indicates that, as far as the shape of the stress distribution
is concerned, the worst stress distribution in terms of the critical buckling loads is
the “ pure bending” stress distribution corresponding to the case ks=kr=⬀, which is
the reverse of the earlier situation with respect to warping stresses.

Fig. 5. Normalized buckling curves for different restraint cases (L=5 m, a=b/2).
Z.-m. Ye et al. / Thin-Walled Structures 40 (2002) 853–864 861

Fig. 6. Normalized buckling curves for different restraint cases (L=5 m, a=b/2).

Fig. 7 shows the buckling behaviour of the purlin with spring constants ks=20.5
N/m and kr=73.8 N-mm with respect to three different loading positions. It can be
seen from the figure that the main influence of the loading position is on the distor-
tional buckling, where the critical load decreases with increase in the distance
between the loading point and web. For local and lateral–torsional buckling, the
influence of the loading position is almost negligible.
Note, that when the spring restraint is taken into account in the pre-buckling stress
analysis, the pre-buckling stress is actually influenced by the span length of the beam,
both in terms of its value and the corresponding stress distribution. Thus, the results
shown in Figs. 2–7 are only for the specified single span length. Fig. 8 shows the
influence of the span length on the critical loads for local and distortional buckling.
It is observed that, for the local buckling the critical load increases with the length,
while for the distortional buckling the critical load is found initially to increase and
subsequently to decrease with the further increase in span length. Note, that the
influence of the span length on the buckling behaviour is purely due to its influence
on the pre-buckling stress. For a given pre-buckling stress the change of span length
in the buckling analysis does not affect the results of the buckling curve.

5. Conclusions

The buckling behaviour of zed purlins partially restrained by steel sheeting under
wind uplift loading has been investigated. The analysis was performed using finite
862 Z.-m. Ye et al. / Thin-Walled Structures 40 (2002) 853–864

Fig. 7. Buckling curves for different loading points (L=5 m, ks=20.5 N/m, kr=73.8 N-mm).

Fig. 8. Influence of span length on critical loads (a=b/2, ks=20.5 N/m, kr=73.8 N-mm).
Z.-m. Ye et al. / Thin-Walled Structures 40 (2002) 853–864 863

strip methods in which the calculation of the pre-buckling stress was calculated based
on the same spring model as that used in the buckling analysis, rather than using
the assumed ‘pure bending’ stress. From the numerical results obtained in this study,
the following conclusions can be drawn:

1. The spring restraints may have significant influence on the local and distortional
buckling of the purlin. This influence is mainly related to its influence on the pre-
buckling stress rather than directly on buckling modes.
2. The spring restraints also have significant influence on the lateral–torsional buck-
ling, but this influence is mainly due to their restraint on the buckling modes.
3. Since the pre-buckling stress is span-dependent, the buckling curve obtained from
the finite strip analysis is applicable only for the span length specified in the pre-
buckling stress analysis. Thus, for different span lengths, the pre-buckling stress
has to be re-calculated and the buckling curve has to be determined based on this
re-calculated pre-buckling stress.
4. In terms of the critical loads it appears that the best location for fixing (that is,
the loading point) is close to the web as this provided the highest critical loads.
5. The present results highlight the importance of the pre-buckling stress used in the
buckling analysis. In order to provide accurate results, consistent analysis models
should be used for both pre-buckling stress and buckling analyses.

Acknowledgements

The work was completed when the first author visited Aston University. The first
author would like to acknowledge the support received from the School of Engineer-
ing and Applied Science at Aston University.

References

[1] Moore DB, Sims PAC. Load tests on full-scale cold formed steel roofs, Building Research Establish-
ment Report, Part 1: Sigma purlin system, Part 2: Zed purlin system and Part 3: Zeta purlin system,
Department of the Environment. Building Research Station, Garston, Watford, UK, WD2 7JR, 1988.
[2] Toma T, Wittemann K. Design of cold-formed purlins and rails restrained by sheeting. Journal of
Constructional Steel Research 1994;31:149–68.
[3] Davies JM. Recent research advances in cold-formed steel structures. Journal of Constructional Steel
Research 2000;55:267–88.
[4] Trahair NS. Flexural-torsional buckling of structures. London: E & FN Spon, 1993.
[5] Leach P, Robinson P. The behavior of purlins subject to wind uplift (an assessment of EC3: Part
1.3). The Structural Engineer 1993;71(14):250–2.
[6] Rhodes J, Lawson RM. Design of structures using cold formed steel sections. SCI Publication, 089.
The Steel Construction Institute, 1993.
[7] Walker AC. Design and analysis of cold-formed sections. London: International Textbook Company
Ltd, 1975.
[8] Hancock G. The behavior and design of cold-formed purlins. Steel Construction 1997;15(3):2–16.
[9] Davies JM, Leach P. First-order generalised beam theory. J. Constructural Steel Research
1994;31:187–220.
864 Z.-m. Ye et al. / Thin-Walled Structures 40 (2002) 853–864

[10] Davies JM, Leach P, Heinz D. Second-order generalised beam theory. J. Constructural Steel Research
1994;31:221–41.
[11] Schafer BW. Elastic buckling analysis of thin-walled members using the classical finite strip method,
CUFSM Version 2C, Johns Hopkins University, 2001 (www.ce.jhu.edu/bschafer).
[12] Ye ZM, Kettle R, Li LY. The strength analysis of cold formed zed purlins partially restrained by
steel sheeting, Computers and Structures (in press).
[13] Cheung YK. Finite strip method in structural analysis. New York: Pergamon Press, 1976.
[14] Hancock GJ. Design of cold-formed steel structures (To Australia Standard AS 1538–1988), 2nd
ed. North Sydney, Australia: Australia Institute of Steel Construction, 1994.
[15] Schafer BW. Cold-formed steel behaviour and design: analytical and numerical modeling of elements
and members with longitudinal stiffeners, Ph.D. Dissertation, Cornell University, Ithaca, NY, 1997.
[16] Gotluru BP, Schafer BW, Pekoz T. Torsion in thin-walled cold-formed steel beams. Thin-Walled
Structures 2000;37:127–45.

You might also like