This may be the author’s version of a work that was submitted/accepted

for publication in the following source:

Kesawan, Sivakumar & Mahendran, Mahen

(2019)
Buckling behaviour and design of complex-shaped aluminium mullion sec-
tions.
Advances in Structural Engineering, 22(2), pp. 547-564.

This file was downloaded from: https://eprints.qut.edu.au/200564/

c The Author(s)

This work is covered by copyright. Unless the document is being made available under a
Creative Commons Licence, you must assume that re-use is limited to personal use and
that permission from the copyright owner must be obtained for all other uses. If the docu-
ment is available under a Creative Commons License (or other specified license) then refer
to the Licence for details of permitted re-use. It is a condition of access that users recog-
nise and abide by the legal requirements associated with these rights. If you believe that
this work infringes copyright please provide details by email to [email protected]

License: Creative Commons: Attribution-Noncommercial 4.0

Notice: Please note that this document may not be the Version of Record
(i.e. published version) of the work. Author manuscript versions (as Sub-
mitted for peer review or as Accepted for publication after peer review) can
be identified by an absence of publisher branding and/or typeset appear-
ance. If there is any doubt, please refer to the published source.

https://doi.org/10.1177/1369433218795118
 Buckling Behaviour and Design of Complex-shaped Aluminium
Mullion Sections

Kesawan Sivakumar1 and Mahen Mahendran2

1
Post doctoral research fellow, Queensland University of Technology (QUT), Brisbane,
Australia
2
Professor, Queensland University of Technology (QUT), Brisbane, Australia

1
Abstract: Thin-walled aluminium mullions are the vertical framing members of the façade

systems used in buildings. This paper investigates the buckling behaviour of these complex-

shaped aluminium mullions. For this purpose, the aluminium mullion sections were

simplified into elements of varying thickness, and modelled using CUFSM finite strip

analysis program. Elastic buckling analyses were performed with and without considering the

availability of glass panel restraints for both negative and positive wind actions, and the

results are presented in this paper. The effect of providing return flanges to enable a good

connectivity between the male and female mullions was also evaluated. The lateral restraints

provided by glass panels were simulated using the spring stiffness option available in

CUFSM, and the analyses were performed for spring stiffness values in the range of 0-1

N/mm/mm. The applicability of the buckling analysis results to the design of aluminium

mullions was then evaluated using the Direct Strength Method (DSM). For this purpose the

section moment capacities of mullions were determined from finite element analyses and

compared with the DSM predictions using the CUFSM buckling analysis results. This

comparison showed that DSM based design can be adopted for the complex-shaped

aluminium mullions provided their elastic buckling capacities are available. Overall, this

study has provided good understanding of the buckling behaviour of mullion sections under

both positive and negative wind actions and has proposed the use of DSM for the design of

aluminium mullion sections.

Keywords: Aluminium mullions; Façade systems; Buckling; Finite strip analysis; Direct
Strength Method

2
Corresponding author’s email address: [email protected], Phone number: 61 73138 2543,
Fax: 61 73138 1170

2
3
1. Introduction
Façade systems are important components of both low- and high-rise buildings as they form

the building envelope that protects the buildings against the environmental hazards while

providing an aesthetic appearance to buildings (Fig. 1). Present-day façade systems are

commonly made of glass supported by extruded aluminium framing members. Aluminium

framing is used in these façade systems due to the capability of aluminium being extruded

into complex shapes, which provide the versatility required to build elegant and complex

building enclosures. Aluminium also has higher strength to weight ratio than steel, and is

about 65% lighter than steel. Generally, aluminium alloy 6063-T6 is used to fabricate these

extruded aluminium framing members. The mullions are the vertical structural members of

aluminium framing, which resists the wind forces acting on the building envelope through

bending actions.

Curtain wall system is one kind of façade system, popularly used in high rise buildings. There

are mainly two types of curtain wall installations, which are stick and unitised panel wall

systems. In the stick wall system (the first generation curtain walls), the vertical aluminium

framing is installed first, which is followed by the installation of horizontal members and

then the glass panels. This system allows for greater flexibility during construction, where

site modification is possible, and it is also economical if correctly designed and installed. But

this takes excessive time to assemble on site, and could be a barrier in present day fast track

projects. The unitised wall system (the second generation curtain walls) is assembled in

factories as a series of panels (Fig. 1), and are then brought to site and assembled through

interlocking mullions and transoms. The unitised panels allow the benefits of factory

fabrication in a controlled environment, and quicker construction at site, which is crucial in

modern high rise buildings (Russell, 2006 and Allana and Carter, 2012). To allow for easier

4
assembly, a pair of extrusion mullion sections is used. These mullions are a pair (male and

female) of aluminium extrusions, which fit together as shown in Fig. 2. The connectivity

between these pairs must allow for expansion and shortening of the panel caused by

temperature variations and wind actions. Therefore male and female mullions are not clipped

or screwed together in most cases. This current practice of Australian façade industry allows

the movement of panels during temperature variation and building deflections. It also allows

easier alteration during fabrication if any problems are found in the constructed façade

assembly as panels can be easily isolated or removed from the curtain wall. Provision of

interlocking clips would make the removal of panels from the constructed façade assembly

difficult. Therefore until there is a need to provide clips or screw connections, the male and

female mullions are not connected by any external mechanisms. The mullions are also

designed individually without considering their combined effects.

These individual aluminum mullions are thin-walled members with complex cross-sections.

The complex cross-sections are based on industry driven studies to allow adequate thermal

movements and to have easier connectivity between the vertical (mullions) and horizontal

(transoms) members. Fig. 3 shows the panel system of mullion sections and glass as used in

the Australian curtain walls subject to positive (pressure) and negative (suction) wind actions.

Importantly, they are asymmetric sections. The vertical mullion sections are subject to

bending stresses when resisting the wind forces applied to the building envelope, and no

other forces are carried by these mullions. In Australia, these mullions are designed based on

the design rules given in AS/NZS 1664.1 (1997) and AS/NZS 1664.2 (1997), which is similar

to the Aluminium Association Design Manual (ADM, 2010). These standards and manuals

do not cover the complete design of these complex aluminium cross-sections, thus design

calculations are made with various assumptions and conservative approaches.

5
The current method adapted in designing the aluminium mullion sections is based on the

allowable stress method given in the ADM (2010) or AS/NZS 1664.1 (1997). To determine

the bending capacity, it is checked first to determine whether any individual element buckles

and if it does, the critical stress is determined and then the section moment and flexural

torsional buckling capacity are found by using complex formulae (ADM, 2010). Currently

the popular trend to design these thin-walled members of complex cross-sections is based on

the Direct Strength Method (DSM) proposed by Schafer (2008). Zhu and Young (2009)

investigated the applicability of DSM to design the tubular aluminium members under

bending and compression actions. This method relates the elastic buckling capacity to the

section and member capacities. Therefore as the first step in this design method, it is

important to understand the buckling behaviour of complex-shaped mullion sections shown

in Fig. 2.

It is a cumbersome task to find the buckling loads of these complex cross-sections using hand

solutions, and adapting a finite element approach would also be time consuming. Finite strip

analysis is the most suitable approach to examine the instability/buckling behaviour of these

complex aluminium members subject to longitudinal stresses (bending and axial). Li and

Schafer (2010) developed an open source programme, CUFSM, based on finite strip analysis

method, and this application facilitates easy determination of the various buckling loads and

modes of thin-walled members of different lengths and sizes (Schafer and Adany, 2006).

Previously Wang (2006) has used the CUFSM software to determine the buckling load of E-

type mullion section. The mullion sections subject to bending action could undergo three

types of instabilities; local, distortional and flexural torsional buckling. The buckling modes

and load factor versus half wave length plots could be different for aluminium extrusion

6
sections compared to steel sections. Furthermore, due to the asymmetric nature of these

mullions, and the lower elastic modulus of aluminium, buckling could occur at lower stress

levels. This study uses CUFSM software to analyze the various buckling modes of complex-

shaped male and female mullion sections restrained by glass as used in curtain walls, and

then investigates the applicability of DSM based design approach for them.

2. Modelling and analyses of 650 series sections – Male mullions

2.1. Development of CUFSM model

The 650 series - 027 section (650-027 in Fig. 2), which is commonly used among leading

glass suppliers in Australia, was considered. It does not have a uniform thickness. In the

CUFSM software, the thickness of a particular element cannot be varied. Hence the male

mullion section shown in Fig. 2 was idealized into sections of different elements as shown in

Fig. 4(a). The AutoCAD drawing has more than 30 co-ordinates, and no facility is available

in CUFSM to directly import the AutoCAD drawings or the co-ordinates of those drawings.

Manually entering each of the node and element number in CUFSM for a particular mullion

section is a cumbersome task since there are about 40 different nodes and elements in each

mullion section. To facilitate this, a small AutoCAD application (PLIST) was loaded in

AutoCAD, which enabled the transfer of nodal co-ordinates of the drawn polyline to an excel

file. An excel spreadsheet was developed, which lists the node and element numbers based on

the imported AutoCAD co-ordinates. The list could be copied and pasted to the cells of

CUFSM. Supplementary data is provided with this paper for the idealized 650-027 mullion

section.

The section properties of the male mullion were determined from the software ShapeDesigner

(refer Supplementary Data). Similarly the section properties of the idealized section were

7
found from the CUFSM software and are given in Fig. 4(b). There are some differences,

especially in the torsional constant values, but they are small, and the area and the second

moment of areas about the X and Z axes are about the same.

The mechanical properties of aluminium sections (6063-T6 alloy) used in the modelling are

summarized as follows (AS/NZS 1664.1, 1997): Yield strength – 172 MPa, Elastic modulus

– 70000 MPa, Poisson’s ratio – 0.32 and Shear modulus – 26515 MPa. These are the

commonly used mechanical properties in Australia for aluminium design.

As seen in Fig. 3, the glass is connected to one side of the mullion along its length. It is

assumed that the glass provides a continuous lateral restraint along the section length when

designing the mullion sections. In this study, analyses of the mullion sections were performed

with and without glass restraints, and then also with varying spring stiffness. In asymmetric

sections, the principal axes are different to those axes where the forces are applied. However,

if the sections are restrained against movement in one direction (X-X axis), the bending

stresses (f) about that axis can be found from the basic equation (Eq. 1) without finding the

stress components in the principal axis. More details can be found in Ing and Trahair (1984).

The restrained bending option available in the CUFSM was used for these mullion sections.
𝑀𝑦
𝑓= ………………………………………………………………………………………(1)
𝐼𝑥

where,

f is the bending stress, M is the bending moment about the x axis, Ix is the second moment of

area about the x axis, and y is the perpendicular distance from the centroidal x axis to the

point considered.

8
2.2. Buckling behaviour of unrestrained male mullion sections

2.2.1. Unrestrained male mullions subject to positive wind action

The first series of analyses was performed for 650–027 male mullion sections subject to

positive action. In the first set, lateral restraint provided by the glass was omitted in the

analyses. In CUFSM analyses, a uniform moment can be applied. However, a member with

different moment distributions other than uniform moment cannot be modelled. The stress

distribution and the tension/compression sides are shown in Fig. 5(a) for positive wind action.

The yield moment about the X-X axis is 3.95 kNm. Beam lengths of 1 to 4000 mm were

considered in the analyses. The buckling analysis results for the positive wind action (Fig.

3(a)) are given in Figs. 6(a) and (b). Fig. 6(a) presents the load factor versus half wave

buckling length plot to show the variation of load ratio with increasing member length. Here,

load factor is the ratio of the buckling and yield moments. The minima buckling occurs at a

half wave length of 90 mm, where the load factor is 3.51 (buckling stress of 603.7 MPa in Fig.

6(b) and Table 1). Fig. 6(b) also shows that the buckling mode at this length is local. At a

length of 300 mm also, local buckling is the critical mode, and the corresponding load factor

is 5.0989. The buckling load factor is high, and hence most likely this member might not

undergo any component buckling before yielding failure. At a length of 900 mm, the

buckling mode is flexural torsional buckling, and the corresponding load factor is 0.989

(buckling stress of 170.1 MPa), and this shows that these sections can undergo flexural

torsional buckling at very low stress levels. For 3000 and 3600 mm lengths, global buckling –

flexural torsional buckling is the critical buckling mode, where the load factors are 0.125 and

0.0967, respectively.

2.2.2. Unrestrained male mullions subject to negative wind action

9
The second series of analyses was performed for 650-027 male mullion section subject to

negative wind action (Fig. 3(b)), and the lateral restraints were omitted. Fig. 5(b) shows the

stress distribution across the cross-section and the tension/compression sides, and the yield

moment is 3.95 kNm. The load factor versus half wave buckling length plot obtained from

CUFSM analysis is shown in Fig. 7(a).

The minimum local buckling stress occurs at a half wave length of 150 mm, where the load

factor is 1.25 (buckling stress of 216 MPa) – Table 1 and Fig. 7(b). But a closer look at Fig.

7(b) shows minor distortion of the flange element also, which is not common in thin-walled

members. Importantly the flange elements (Elements 1 and 2) buckle together rather than

buckling as two individual elements. At a half wave length of 450 mm, it is subjected to local

and distortional buckling as evident from Fig. 7(b). At a length of 950 mm, the buckling

mode is flexural torsional buckling, and the load factor is 0.991 (buckling stress of 170 MPa).

For 3000 and 3600 mm lengths, flexural torsional buckling is the critical mode, where the

load factors are about 0.140 and 0.107, respectively.

Overall, buckling analyses of unrestrained mullions show that these sections are more prone

to flexural torsional buckling under both positive and negative wind actions if they are not

restrained. Hence the potential lateral restraints provided by the glass panel should be

considered in the buckling analyses.

2.3. Buckling behaviour of restrained male mullion sections

2.3.1. Restrained male mullions subject to positive wind action

In curtain walls as shown in Fig. 3, the glass is connected to the flanges on one side of the

mullions through sealant or gasket connection. The glass will restrain the lateral movement of

these flanges during wind action. During positive wind action, the load will be applied or

10
concentrated on the inner flange (Fig. 3(a)), and thus the lateral restraint is expected to be

mobilized at the loading point through friction. In this analysis, the lateral restraint was

simulated at Point A as shown in Fig. 8(a). The analysis results are provided in Table 2 and

Figs. 9(a) and 10. The half wave length versus load factor plot of sections without lateral

restraints are also shown in Fig. 9(a). There are significant differences between both plots.

The positive wind action induces compressive stresses on the outer flange, which is restrained

by the glass. Therefore the compression flange is unable to move laterally, thus eliminating

the flexural torsional buckling of mullions. As evident from Fig. 9(a), the buckling plots of

both mullions (laterally restrained and unrestrained) almost coincide up to a length of 350

mm. Thereafter the plot of unrestrained mullion slopes down while that of restrained mullion

goes up due to the elimination of flexural torsional buckling. Buckling modes for varying

lengths are shown in Fig. 10. As evident from Fig. 10, flexural torsional buckling was not

observed in the 3000 and 3600 mm long mullions with lateral restraints, whereas Fig. 6(b)

showed severe flexural torsional buckling in the 3000 and 3600 mm long unrestrained

mullions subject to positive wind action.

2.3.2. Influence of lateral restraint location on the buckling behaviour of male mullion

subject to positive wind action

Buckling analyses were conducted to determine whether the location of the lateral restraint

along the flange width has any influence on the load factor versus half wave length plot. For

this purpose lateral restraint was applied at two different locations as shown in Fig. 8 (Points

A and B). The analyses were performed, and the results are plotted in Fig. 9(b). This figure

demonstrates that the lateral restraint location along the flange width did not affect the

buckling behaviour of the laterally restrained mullion sections – a useful outcome.

11
2.3.3. Restrained male mullions subject to negative wind action

During the negative wind action, the load is mainly transferred to the outer flange (Fig. 3(b))

through the glass panels, and the lateral restraint is expected to be acting on that flange.

Therefore for the negative wind action the restraint was defined in CUFSM at the outer

flange of the mullion as shown in Fig. 11(a). Fig. 11(b) compares the buckling plots of

mullions with and without lateral restraints. It depicts that there are no significant differences

in the plots since the compressive stresses occur at the inner flange, which is unrestrained.

Buckling modes for varying lengths are shown in Fig. 12, and at 3000 and 3600 mm lengths,

flexural torsional buckling modes are observed. They are the same as shown in Fig. 7(b) for

unrestrained male mullions.

2.4. Comparison of the buckling behaviour of mullions with and without return lips

This analysis compares the buckling behavior of the aluminium mullion Sections 1 and 2

(Figs. 13(a) and (b)). The difference between these sections is that Section 2 has a straight

flange which is similar to the flange of a channel section. This will depict the benefit of the

return lip of Section 1 (Fig. 13(a)). Analyses were conducted and the load factor versus

buckling half wave length plots of Sections 1 and 2 are shown in Fig. 13(c). Both plots

coincide after a length of 450 mm. After this length the flexural torsional buckling becomes

the dominant buckling mode, and the results show that return lip does not provide any

benefits for flexural torsional buckling as this is apparent since this return lip will not

significantly alter the second moment of area, torsion constant and warping constant.

However, for lengths less than 450 mm, the return lip provides significant benefit as evident

from Fig. 13(c). The critical load factor for Section 2 is 1.02, and hence the critical buckling

stress is 175.4 MPa, while that of Section 1 is about 23% higher (215 MPa) than Section 1.

This shows the benefit of the increased local buckling capacity due to the provision of the

12
return lip. Fig. 13(d) shows the local buckling mode of Section 2. In conclusion, although this

return lip is provided to increase the connectivity between the couple mullions, it also

increases the local buckling capacity.

2.5. Buckling behaviour of mullions with lateral restraints provided via spring elements

Although façade engineers assume that full lateral restraints are provided by the glass to the

mullions in design, in some instances (finite element modelling) a spring constant of

1N/mm/mm is assumed to simulate the lateral restraints provided by the glass. Glass restraint

is continuous along the mullion length, and continuous spring constraints can also be

simulated in CUFSM where a spring can be attached to any node (Fig. 14(a)). The spring

constant can be applied as a constant value or as varying with member length. In this study

different spring stiffness values (1, 0.1 and 0.01 N/mm/mm) were used in CUFSM, and the

analyses were performed. Negative wind action was not considered as it was shown in

Section 2.3.3 that even assuming full lateral restraint did not affect the buckling

behavior/plot. For a male mullion section subject to positive wind action with spring stiffness

(1N/mm/mm), the minimum load factor for flexural torsional buckling is about 2.67. Hence

most likely this mode of buckling cannot be expected in this 650-027 mullion section. The

half wave length versus load factor plots of unrestrained mullion section, and mullion section

with spring lateral restraint – 0.01 N/mm/mm are almost the same. Figure 14(b) provides a

good understanding of the buckling behaviour of mullion sections with lateral restraints

provided by spring elements of varying stiffness. Considering these results, an appropriate

spring stiffness in the range of 0.1 to 1.0 is most suitable to simulate glass panel restraints. As

a conservative approach, 0.1 N/mm/mm can be used. Full scale façade test results are needed

to confirm this.

13
3. Modelling and analyses of 650 series sections - Female mullions

In the unitised panel system, male and female mullions are locked together to form the

building envelope. The female mullion section (650-028) commonly used is shown in Fig. 2.

Buckling analyses were performed with and without considering the lateral restraints from

the glass. The methodology used was identical to that described for male mullions in Section

2. The buckling analysis results showed that the behavior of female mullions is similar to that

described for male mullions. Hence only the selected results of load factor versus the half

wave buckling length plots and buckling modes are presented without detailed explanations

(Figs. 15-17).

4. Applicability of the CUFSM buckling analysis results in the Mullion design

This section investigates the applicability of CUFSM buckling analysis results to determine

the structural capacity of mullions. To consider a larger range of section slenderness, a simple

mullion with geometry similar to 650-027 mullion, but with uniform thickness, was

considered (Fig. 18(a)). The thickness of this mullion was changed from 0.8 to 8 mm, which

resulted in cross-sectional slenderness in the range of 30 to 160 based on ADM (2015).

Therefore, the investigated-mullion sections included compact sections, and sections subject

to elastic and inelastic local buckling. The CUFSM buckling analyses were performed and

the obtained critical local buckling stresses are summarized in Table 3. The load factor versus

half-wave buckling length of 1 mm thick section is shown in Fig. 18(b).

The use of these critical local buckling stress values in determining the ultimate section

moment capacity of mullions is evaluated in this study. These critical buckling stresses were

used in the direct strength method given in ADM (2015). The mullion section moment

capacities were determined from finite element analyses (FEA) using ABAQUS under

14
negative wind action. Details of the finite element modelling procedure are given next,

followed by the comparison of FEA and design rule predictions.

The finite element models developed were similar to the ones developed to determine the

section moment capacity of cold-formed steel beams in Anapayan et al. (2011) and Siahaan

et al. (2016). The developed models represented a four point bending setup. The aluminium

mullions were modelled using S4R shell elements with 4 mm x 4 mm mesh. Figure 19

presents the finite element models including the support and symmetric boundary conditions

and the loading points. Furthermore, in order to prevent any X-axis displacement, a

continuous lateral restraint was defined at the outer flange along the mullion length to

simulate the restraint provided by the glass panels. The geometric imperfection was applied

in the shape of the critical local buckling mode according to Schafer and Pekoz (1998). A

local geometric imperfection value of 0.004b was used, based on the allowable tolerance

given in Aluminium standards and data metric SI (2013). The limitation of the developed

models is that they need to be validated using experiments in the future.

The yield strength, ultimate strength and elastic modulus of mullion were taken as 172, 207

and 70000 MPa, respectively from AS/NZS 1664.1 (1997). Eurocode stress-strain model

(Eurocode 9 Part 1.1, 2007) was used to determine the engineering stress-strain model (Eqs. 2

and 3). The corresponding true stress and logarithmic plastic strain data were used as input in

ABAQUS.

𝜎 𝜎 𝑛
𝜀= + 0.002 ( ) …………………………………………………………………(2)
𝐸 𝑓0.2
𝜀𝑜,𝑥 𝑓𝑥
𝑛 = log ( ) / log ( ) ………………………………………………………………... (3)
𝜀𝑜,𝑒 𝑓𝑒

where: n is the strain hardening exponent, fx is the reference stress, fe is the conventional limit
of elasticity, ɛo,x is the reference strain and ɛo,e is the strain corresponding to the stress fe

15
Finite element analyses were performed under negative wind action using Modified Riks

analysis method, and the results are summarized in Table 3. The mullions failed by section

yielding where local buckling was observed in slender sections or in sections subjected to

inelastic buckling. The section moment capacities of mullions were predicted using the direct

strength method (DSM) given in ADM (2015), where the buckling stresses obtained from

CUFSM analyses were used to determine the cross-sectional slenderness of mullions. The

DSM equations (ADM, 2015) to calculate the section moment capacity (Mnl) are given next.

𝑀𝑛𝑙 = 𝐹𝑏 𝑆𝑥𝑐 ……………………………………………………………………..…………...(4)

The allowable stress (Fb) is found using the following equations and limits:

For 𝜆𝑒𝑞 ≤ 𝜆1

𝐹𝑏 =

𝑀𝑛𝑝
……………………………………………………………………………………...(5)
𝑆𝑥𝑐

𝜆𝑒𝑞 =

𝐸
𝜋√𝐹 …………………...………………..…………………………………………….(6)
𝑐𝑟

For 𝜆1 < 𝜆𝑒𝑞 < 𝜆2

𝑀𝑛𝑝 𝑀𝑛𝑝 𝜋2𝐸 𝜆𝑒𝑞 −𝜆1

𝐹𝑏 = −[ − ](𝐶 ) …………………………………………………………...(7)
𝑆𝑥𝑐 𝑆𝑥𝑐 𝐶𝑝2 𝑝 −𝜆1

For 𝜆𝑒𝑞 ≥ 𝜆2

𝑘2 √𝐵𝑝 𝐸
𝐹𝑏 = ………………………………………………………………………………….(8)
𝜆𝑒𝑞

where 𝜆𝑒𝑞 is the equivalent numerical slenderness, 𝐹𝑐𝑟 is the critical buckling stress, 𝐸 is the

elastic modulus, k2 is a post-buckling constant

𝐵𝑝 − 𝐹𝑐𝑦
𝜆1 = , 𝜆2 = 𝐶𝑝
𝐷𝑝

16
Cp, Bp, and Dp are buckling coefficients which are given in Table B.4.2 of the ADM (2015).

Sxc is the section modulus about the compression side of the X-axis.

Mnp is the nominal flexural strength, calculated by:

𝑀𝑛𝑝 = 𝑓𝑦 ∗ min(1.5𝑆𝑥𝑐 , 𝑍)………………………………………………………………….(9)

where 𝑍 is the plastic modulus of the section.

The ratios between the section moment capacities obtained from FEA and DSM design rules

are also given in Table 3 and Fig. 20. The DSM (ADM, 2015) predictions agree well with the

FEA results of compact mullion sections, and those subject to inelastic buckling, as reflected

by the mean FEA to design rule prediction ratios of 1.08 and 1.04, respectively. However, the

predictions were conservative for slender mullion sections with a mean FEA to DSM

prediction ratio of 1.40. The mean and COV for the FEA to design rule prediction ratios

considering all the mullion sections are 1.17, and 0.16, respectively. This shows that the

DSM in ADM (DSM, 2015) can be conservatively used to predict the section moment

capacity of aluminium mullion sections subject to local buckling.

This section has demonstrated the direct applicability of buckling analysis results from

CUFSM to determine the section moment capacity of aluminium mullion sections. Similarly,

the member moment capacity can also be determined using the same approach. The use of

DSM simplifies the moment capacity calculations of complex-extruded aluminium mullion

sections, compared to the other approaches such as the limiting stress method and the total

moment capacity approach (ADM, 2015 and Kim and Pekoz, 2003). The CUFSM buckling

analysis software can be easily used to find the buckling stresses of complex-shaped mullion

sections, which form the major input to the DSM calculations.

5. Conclusion

17
This paper has presented an investigation into the buckling behaviour of complex-shaped

aluminium mullion sections, where the finite strip analysis software “CUFSM” was used. The

aluminium mullion sections were simplified into elements of different thicknesses, and were

modelled using CUFSM. Two commonly used sections, 650-027 male and 650-028 female

mullions, were considered. Analyses were performed with and without considering the

availability of glass panel restraints for both negative and positive wind actions. The results

showed that the potential lateral restraints provided by the glass panels can significantly

improve the flexural buckling behaviour of mullion sections subject to positive wind action

as compression flanges are restrained. The mullions with a spring constant of 1 N/mm/mm

appeared to simulate fully restrained conditions. An appropriate spring stiffness in the range

of 0.1 to 1 N/mm/mm can be used in CUFSM to simulate the glass panel restraint to mullions.

Furthermore, it was found that the provision of return flanges to enable a good connectivity

between the male and female mullions contributed to the increase in local buckling capacity

of mullion sections.

Due to the increasing usage of complex-shaped thin-walled steel and aluminium members,

the use of direct strength method is becoming popular, which requires a good understanding

of the buckling behaviour of mullion sections. This study has provided good understanding of

the buckling behaviour of male and female mullion sections under both positive and negative

wind actions including buckling plots and modes. Importantly, it has demonstrated the direct

applicability of buckling analysis results from CUFSM to determine the moment capacities of

aluminium mullion sections.

Acknowledgements

The authors would like to thank QUT for providing all the necessary support with the

computing facilities.

18
References

1. ADM (2010), Aluminium Design Manual, Specification for Aluminium Structures, The

Aluminium Association, Washington, D.C., USA.

2. ADM (2015), Aluminium Design Manual, Specification for Aluminium Structures, The

Aluminium Association. Washington, D.C., USA.

3. Aluminium Association (2013). Aluminium standards and data Metric SI, The

Aluminium Association, Washington, D.C., USA.

4. Allana, P. K. and Carter, D. (2012). Curtain Walls Issues, Problems and Solutions, Proc.

of the Building Envelope Technology Symposium, RCI Inc., pp. 97-111.

5. Anapayan, T., Mahendran M. and Mahaarachchi, D. (2011) Section moment capacity

tests of LiteSteel Beams. Thin-Walled Structures, 49(4), pp. 502-512.

6. European Committee for Standardization (CEN) (2007). Eurocode 9: Design of

aluminium structures - Part 1–1: General structural rules. BS EN 1999-1-1:2007,

Brussels, Belgium.

7. Ings, N.L. and Trahair, N.S. (1984), Lateral buckling of Restrained roof Purlins, Thin-

walled Structures, Vol. 2(4), pp. 285-306.

8. Kim, Y. and Pekoz, T. (2003), Ultimate flexural strength of aluminium sections, Thin-

walled Structures, Vol. 47(10-11), pp. 857-65.

9. Li, Z. and Schafer, B.W. (2010), Buckling analysis of cold-formed steel members with

general boundary conditions using CUFSM: conventional and constrained finite strip

methods, Proc. of the 20th International Specialty Conference on Cold-Formed Steel

Structures, St. Louis, MO. November, 2010.

19
10. Russell, M.S. (2006), Curtain Walls: Not Just another Pretty Façade, Journal of

Architectural Technology, Vol. 23(1), pp. 1-18.

11. Schafer, B. and Pekoz, T. (1998). Computational modeling of cold-formed steel:

Characterizing geometric imperfections and residual stresses. Journal of Construction

Steel Research, Vol. 47(3), pp. 193-210.

12. Schafer, B.W. and Adany, S. (2006), Buckling analysis of cold-formed steel members

using CUFSM: conventional and constrained finite strip methods, Proc. of the 18th

International Specialty Conference on Cold-Formed Steel Structures, Orlando, Florida,

USA.

13. Schafer, B.W. (2008), The Direct Strength Method of Cold-formed Steel Member

Design, Journal of Constructional Steel Research, Vol. 64(7-8), pp. 766–778.

14. Siahaan, R., Keerthan, P. and Mahendran, M. (2016). Finite element modeling of rivet

fastened rectangular hollow flange channel beams subject to local buckling. Engineering

Structures, 126, pp. 311-327.

15. Standards Australia (1997), Aluminium Structures, Part 1: Limit State Design, AS/NZS

1664.1, Sydney, Australia.

16. Standards Australia (1997), Aluminium Structures, Part 2: Allowable Stress Design,

AS/NZS 1664.2, Sydney, Australia.

17. Wang, Y. (2006). Structural Behavior and Design of Two Custom Aluminum Extruded

Shapes in Custom Unitized Curtain Wall Systems, MSc thesis, School of Civil

Engineering, University of Cincinnati, Ohio, USA.

18. Zhu, J. and Young, B. (2009), Design of Aluminium Flexural Members using Direct

Strength Method, Journal of Structural Engineering ASCE, Vol. 135(5), pp. 558-566.

20
 Installation

Figure 1: Curtain wall system

Male Female
mullion mullion

Figure 2: 650-027 and 028 mullion couples

21
 Mullion
Restraint to the inner
Z flange during positive wind
action
X
Gasket

Inner flange
Positive wind
Outer flange

(a) Positive wind action

Restraint to the outer

flange during negative
wind action

Negative wind actions

(b) Negative wind action

Figure 3: Glass panel system subject to wind actions

Note: The lateral restraint is available at the gasket to mullion connection point.

22
 (a) Idealised model

(b) Section properties

Figure 4: Modelling of 650-027 male mullion in CUFSM

Note: In this figure, A is the area (mm2), J is the torsion constant (mm4), Xcg and Zcg are the
distances along the X and Z axes to the centroid of the cross-section (mm) from the origin, Ixx
and Izz are the second moments of area about X- and Z- axes (mm4), Ixz is the polar moment
of area (mm4), θ is the angle of the major and minor principal axes from the X and Y-axes, I11
and I22 are the second moments of area about the major and minor principal axes, Xs and Zs
are the distances to the shear centre from the centroid (mm) along the X- and Z-axis
directions, Cw is the warping constant (mm4), and β is the mono-symmetry constant about X-
and Y-axes.

23
 Tension side

Compression side

(a) Positive wind action – 650-027

Compression side

Tension side

(a) Negative wind action – 650-027

Figure 5: Bending stress distribution – Male mullion section

24
 Lo
ad
fac
tor

Half-wave buckling length (mm)

(a) Load factor versus half wave buckling length plot

LF – 9.117 LF– 3.506 LF – 5.099

at 18 mm at 90 mm at 300 mm

LF – 0.989 LF – 0.125 LF – 0.097

at 900 mm at 3000 mm at 3600 mm

(b) Buckling modes

Figure 6: Buckling analysis results of unrestrained male mullion (650-027) subject to positive
wind action
Note: *LF – Load factor

25
Load factor

Half-wave buckling length (mm)

(a) Load factor versus half wave buckling length plot

Element 1 Element 2

LF – 1.254 at 150 mm

LF – 2.834 LF – 0.991 LF – 0.140 LF – 0.107

at 450 mm at 950 mm at 3000 mm at 3600 mm

(b) Buckling modes

Figure 7: Buckling analysis results of unrestrained male mullion (650-027) subject to

negative wind action

26
 Lateral restraint –
Point B

Lateral restraint –
Point A

(a) Location A (b) Location B

Figure 8: Lateral restraints provided to the male mullion section during positive wind action

Restrained
mullion
Load factor

Unrestrained
mullion

Half-wave buckling length (mm)

(a) Mullion with and without lateral restraints during positive wind action

27
 Plots of mullions with lateral restraint
at Points A and B coincide
Load factor

Half-wave buckling length (mm)

(b) Mullion with lateral restraints at different locations during positive wind action

Figure 9: Load factor versus half wave buckling length plot of restrained male mullion (650-
027) subject to positive wind action

LF – 9.117 LF – 3.522 LF – 5.056 LF – 18.787 LF – 19.036

at 18 mm at 90 mm at 250 mm at 3000 mm at 3600 mm

Figure 10: Buckling modes of laterally restrained 650-027 male mullion subject to positive
wind action

28
 Lateral restraint

(a) Lateral restraint location during the negative wind loading

Plots of unrestrained and restrained

mullions coincide
Load factor

Half-wave buckling length (mm)

(b) Load factor versus half wave buckling length plots

Figure 11: Buckling analysis results of unrestrained and restrained male mullion (650-027)
subject to negative wind action

29
 LF – 1.254 LF – 2.848 LF – 0.143 LF – 0.112
at 150 mm at 450 mm at 3000 mm at 3600 mm

Figure 12: Buckling modes laterally restrained 650-027 male mullion subject to negative
wind action

Return lips

(a) Section 1 (b) Section 2

30
 Section 2
Load factor

Section 1

Half-wave buckling length (mm)

(c) Load factor versus half wave length plots

(d) Buckling mode of Section 2 at local buckling half wave length of 100 mm

Figure 13: Buckling analysis results of Sections 1 and 2

(a) Spring element usage for 650-027 mullion

31
 1 N/mm/mm
Load factor

Full lateral restraint

0.1 N/mm/mm

0.01 N/mm/mm

No lateral restraint
Half-wave buckling length (mm)

(b) Load factor versus buckling wave length plots

Figure 14: Buckling analysis results of 650-027 mullion under positive wind action, and with
spring restraints

32
Load factor

Half-wave buckling length (mm)

(a) Load factor versus buckling wave length plot

LF – 2.790 LF – 2.837 LF – 0.361 LF – 0.078 LF – 0.063

at 130 mm at 200 mm at 1000 mm at 3000 mm at 3600 mm

(b) Buckling modes

Figure 15: Buckling analysis results of unrestrained female mullions (650-028) subject to
negative wind action

33
Load factor

Half-wave buckling length (mm)

(a) Load factor versus buckling wave length plot

LF – 7.746 LF – 3.749 LF – 4.011 LF – 0.079 LF – 0.069

at 22 mm at 200 mm at 280 mm at 3000 mm at 3600 mm

(b) Buckling modes

Figure 16: Buckling analysis results of unrestrained female mullions (650-028) subject to
positive wind action

34
 Laterally restrained
mullion
Load factor

Unrestrained
mullion

Half-wave buckling length (mm)

(a) Load factor versus buckling wave length plots

LF – 7.746 LF – 3.957 LF – 26.059 LF – 26.497

at 22 mm at 230 mm at 3000 mm at 3600 mm

(b) Buckling modes of restrained mullion

Figure 17: Buckling analysis results of laterally restrained and unrestrained female mullion
(650-028) subject to positive wind action

35
 40 mm

4.2 mm long stiffener

150
mm

17 mm

26 mm 10 mm

23 mm
Note: centre line dimensions are shown
in this figure
(a) Mullion geometry considered in the parametric study

Local
buckling
Load factor

Half-wave buckling length (mm)

(b) CUFSM Buckling analysis plot and critical buckling mode of 1 mm thick section

Figure 18: CUFSM Buckling analysis plot and critical buckling mode of 1 mm thick section
used in the parametric study

36
 Boundary of symmetry at mid-span
150 mm
(U3=UR1=UR2=0)
Inner flange Mullion-S4R
400 mm elements

50 mm

Loading point
(U1=UR3=0)

Stiffeners-
R3D4 elements
Support
(U1=U2=UR3=0)
Outer flange

Figure 19: Finite element model

Slender section
(λ≥73.8)
Full Inelastic
compact (73.8>λ>36.6)
(λ≤36.6)

Figure 20: FEA / DSM ratio versus slenderness plot

37
Table 1: CUFSM buckling analysis results for unrestrained male mullions (650-027) subject
to positive and negative wind action

Positive wind loading

Length (mm) Buckling mode Load factor Buckling stress (MPa)
18 Local 9.116 1568
90 Local 3.506 603
300 Local 5.099 877
300 Flexural torsional 0.989 170
3000 Flexural torsional 0.125 22
3600 Flexural torsional 0.096 17
Negative wind loading
150 Local/distortional 1.2537 216
450 Local/distortional 2.8337 487
950 Flexural torsional 0.9909 170
3000 Flexural torsional 0.1400 24
3600 Flexural torsional 0.1068 18

Table 2: CUFSM buckling analysis results for restrained male mullions (650-027) subject to
positive wind action

Length (mm) Buckling mode Load factor Buckling stress (MPa)

18 Local 9.116 1568
90 Local 3.522 606
300 Local 5.107 878
3000 Flexural 18.787 3231
3600 Flexural 19.036 3274

38
1
Table 3: Comparisons of section moment capacities of mullions from FEA and DSM (ADM, 2015)

Section moment capacity

Thickness Slender- Compactness Buckling Yield moment Buckling
(mm) ness (λ) stress (MPa) (kNm) moment (kNm) FEA DSM FEA/
(kNm) (kNm) DSM
0.8 155.9 28 1.41 0.23 0.73 0.46 1.58
1.0 133.8 39 1.75 0.39 0.95 0.67 1.41
1.2 113.3 Slender 54 2.10 0.66 1.37 0.95 1.45
1.4 98.3 (λ≥73.8) 71 2.44 1.02 1.75 1.28 1.37
1.6 86.88 92 2.78 1.48 2.16 1.64 1.32
1.8 77.8 114 3.12 2.07 2.62 2.06 1.27
2.0 70.4 139 3.46 2.80 3.10 2.74 1.13
2.2 64.3 167 3.79 3.68 3.61 3.36 1.07
2.4 59.2 197 4.12 4.73 4.15 4.00 1.04
2.6 54.8 230 4.44 5.95 4.70 4.63 1.02
2.8 51.0 Inelastic 266 4.77 7.37 5.29 5.26 1.00
3.0 47.7 (73.8>λ>36.6) 304 5.09 8.98 5.94 5.89 1.01
3.3 43.5 365 5.58 11.86 6.96 6.85 1.02
3.6 40.0 433 6.06 15.24 7.97 7.79 1.02
3.9 37.0 506 6.54 19.23 8.94 8.75 1.02
4.2 34.4 584 7.01 23.80 9.91 9.42 1.05
Fully compact
4.5 32.1 669 7.47 29.04 10.85 10.06 1.08
(λ≤36.6)
4.8 30.2 758 7.94 35.02 11.96 10.72 1.12
Mean 1.17
COV 0.16