STRUCTURAL BEHAVIOUR AND DESIGN OF

ALUMINIUM FACADE MULLIONS UNDER

WIND ACTIONS

Thananjayan Sivaprakasam
BSc Eng (Hons)

A thesis submitted to the School of Civil Engineering and Built Environment

Queensland University of Technology in fulfilment of the
requirements for the degree of
Doctor of Philosophy

School of Civil Engineering and Built Environment

Science and Engineering Faculty
Queensland University of Technology
2019
This Page is Intentionally Left Blank
 Keywords

KEYWORDS

Aluminium facade, Mullion couple, Glass facade, Wind action, Unitized facade,
Lateral-torsional buckling, Moment capacity, Facade testing, Ultimate strength test,
Finite element analysis, Tensile testing, Aluminium alloy, Finite element model,
Mullion analysis, Aluminium design manual, Facade design, Hollow mullion, Clip
system, Mullion design.

i
 Abstract

ABTRACT

Unitized glazed aluminium facades used commonly in buildings are pre-fabricated in

factory, and assembled on site. To facilitate the pre-fabrication and site assembly
process, the vertical members of the aluminium frame comprise two open sections,
called male and female mullions (or mullion couple). Since these complex-shaped
mullions generally have their shear centre outside the section, the wind loads transferred
from glass panes to the mullions act eccentrically to their shear centre. Under negative
wind action, both male and female mullions rotate to lean on each other and give lateral
support to their unrestrained compression flange. This may cause them to act as a single
section to achieve a higher capacity. However, such behaviour is not well understood,
and thus the capacity of mullion couples is taken as the sum of the individual male and
female mullion capacities. For positive wind action, the capacity of mullions is taken as
their section capacity since their compression flanges are laterally restrained by glass
panes. However, glass panes may not provide adequate lateral restraints to the mullions,
in which case, it would result in unsafe designs. In addition, clip and hollow systems are
used as an interim solution to increase the ultimate and serviceability capacities of
mullions although there are no guidelines in the current aluminium design standards or
in the literature. Therefore, full-scale tests and advanced finite element analyses (FEA)
were conducted on unitized aluminium mullion couples with and without clip and
hollow systems to investigate their structural behaviour and capacity under negative and
positive wind actions.

Full-scale wind pressure tests were first conducted on unitized facade assemblies, made
of mullion couples without any clip or hollow systems, under both negative and positive
wind actions. It was found that the interaction between the mullions is not effective, and
the lateral restraint given by glass panes is inadequate to achieve section/yield moment
capacity under negative and positive wind actions, respectively. The same type of
mullion couple was then tested with a clip at its mid-span. These tests showed that the
clip did not increase the capacity significantly under negative wind action as expected
while there was no capacity increment under positive wind action. Finally the mullion
couple with a hollow stiffener and a clip at mid-span was tested under negative wind
action, which showed that the capacity of mullion couples could be improved by using
hollow systems.

ii
 Abstract

Finite element (FE) models of the mullion couple tested without any clip or hollow
systems were developed under both wind actions, where mechanical properties obtained
from tensile coupon and bending tests were used. Two types of FE models were
developed. In the first model (“Simple model”), only the mullion couple was modelled,
and the other components were simulated by using suitable boundary conditions. In the
second model (“Full model”), mullion couple, glass panes and gaskets were modelled.
Comparisons of the results from FEA and tests showed that the developed simple and
full FE models predicted the structural behaviour and capacity of the mullion couples
reasonably well. However, simple FE models were used in the detailed parametric
studies since full FE models are expensive in terms of computing time and resources.

In the parametric study of mullion couples without any clip or hollow systems under
negative wind action, simple and full FE models of selected mullion couples were first
developed and analysed, and their results showed that predictions from simple and full
FE models agreed well. Then, the effects of gap between mullions, lateral restraint, load
distribution profile, ratio of the loads acting on male and female mullions (load
distribution ratio) and intermediate transoms were investigated. This investigation
showed that lateral restraint and load distribution ratio significantly influenced the
mullion capacity than other parameters. Twenty mullion couples that included captive
and structural glazing mullions were then analysed with varying spans from 1800 to
4200 mm. Comparison of the results with current design predictions using ADM (AA,
2015) showed that the current design approach could result either in conservative or
unsafe designs. A similar study was conducted for the mullion couples under positive
wind action. It showed that glass restraint is inadequate for the mullions to achieve their
section/yield moment capacity, and thus the current design assumption is not
acceptable. Furthermore, it was shown how FEA based predictions can be used to
develop design capacity charts for mullion couples, which can be used in routine
mullion designs.

Simple and full FE models of the mullion couple tested with a clip at mid-span were
also modelled under both wind actions, and the results were validated using test results.
A parametric study was then conducted on different mullion couples with a clip at their
mid-span under both wind actions. It was found that the use of clips, in general, did not
significantly increase the ultimate capacity under both wind actions. This was the same
for mullion couples with continuous clip. In addition, ADM (AA, 2015) predictions,

iii
 Abstract

considering the currently used assumption for clips, were unsafe, and are thus not
acceptable. Thereafter, mullion couples with hollow systems were analysed under wind
actions, which showed that the capacity of mullion couples comprising open mullions,
in general, can be significantly increased by using hollow systems. This was the same
for the mullion couples with clip and hollow systems. These parametric studies showed
that when developing new mullion sections both ultimate and serviceability limit state
criteria should be considered to achieve efficient and economical sections.

Finally, based on the outcomes obtained from full-scale tests and FEA based parametric
studies, suitable design guidelines and recommendations are provided for economical
and safe design of mullion couples used in unitized facades subject to wind actions.
This included design moment capacity predictions for unitized facade mullion couples
without any clip and hollow systems, with clip systems, with hollow system and with
clip and hollow systems. It is recommended that the moment capacity of a mullion
couple can be determined using three methods, design rules, finite element analysis or
experimental testing.

iv
 Publications

PUBLICATIONS

Refereed International Conference Papers

Sivaprakasam, T., Kesawan, S., Mahendran, M., Stringfellow, J., Baleshan, B. and
Gunalan, S. (2018). Full-scale tests of aluminium mullion couples in facades systems
under wind actions. Proc. of Eighth International Conference on Thin-Walled
Structures, University of Lisbon, Portugal.

Sivaprakasam, T., Kesawan, S., Mahendran, M. (2019). Numerical study of split

mullions subject to negative wind action. First International Conference on Civil
Engineering and Applications, University of Moratuwa, Sri Lanka.

QUT Research Reports

Sivaprakasam, T., Kesawan, S. and Mahendran, M. (2017). Investigation of the

Mechanical Properties of 6063-T6 Aluminium Alloy. Research report No.1, Queensland
University of Technology, Brisbane, Australia.

Sivaprakasam, T., Kesawan, S., Stringfellow, J., Baleshan, B., Gunalan, S. and
Mahendran, M. (2017). Full-scale tests on a facade system made of coupled open
mullion sections. Research report No.2, Queensland University of Technology,
Brisbane, Australia.

Sivaprakasam, T., Kesawan, S. and Mahendran, M. (2017). Development and validation

of finite element models of mullion couples. Research report No.3, Queensland
University of Technology, Brisbane, Australia.

Sivaprakasam, T., Kesawan, S. and Mahendran, M. (2018). Design capacity charts for
open mullion sections. Research report No.4, Queensland University of Technology,
Brisbane, Australia.

Sivaprakasam, T., Kesawan, S. and Mahendran, M. (2018). Design capacity charts for
hollow mullion sections. Research report No.5, Queensland University of Technology,
Brisbane, Australia.

Sivaprakasam, T., Kesawan, S. and Mahendran, M. (2018). Design capacity charts for
centrally glazed mullion sections. Research report No.6, Queensland University of
Technology, Brisbane, Australia.

v
 Table of contents

TABLE OF CONTENTS
KEYWORDS ................................................................................................................................ i

ABTRACT ................................................................................................................................... ii

PUBLICATIONS ........................................................................................................................ v

TABLE OF CONTENTS.......................................................................................................... vii

LIST OF FIGURES .................................................................................................................. xii

LIST OF TABLES .................................................................................................................. xxii

STATEMENT OF ORIGINAL AUTHORSHIP ................................................................. xxiv

ACKNOWLEDGEMENTS .................................................................................................... xxv

ABBREVIATIONS ............................................................................................................... xxvii

1 INTRODUCTION

1.1 Background ................................................................................................................1-1

1.2 Research problem .....................................................................................................1-10
1.3 Research aim and tasks.............................................................................................1-10
1.4 Thesis Outline...........................................................................................................1-11
2 LITERATURE REVIEW

2.1 General .......................................................................................................................2-1

2.2 Mechanical properties of aluminium..........................................................................2-2
2.3 Experimental study.....................................................................................................2-8
2.4 Numerical analysis ...................................................................................................2-22
2.5 Design of mullion couples........................................................................................2-26
2.6 Literature review findings ........................................................................................2-31
3 MECHANICAL PROPERTIES OF 6063-T6 ALUMINIUM ALLOY MULLIONS

3.1 Tensile coupon tests ...................................................................................................3-1

3.2 Four-point bending tests .............................................................................................3-5
3.3 Test results and discussion .........................................................................................3-9
3.4 Conclusion................................................................................................................3-22
4 EXPERIMENTAL STUDY

4.1 General .......................................................................................................................4-1

4.2 Previous studies..........................................................................................................4-1
4.3 Full-scale wind pressure tests.....................................................................................4-3
4.3.1 Test panel details ................................................................................................4-3

vi
 Table of contents

4.3.2 Test set-up .......................................................................................................... 4-7

4.3.3 Loading and test procedures............................................................................. 4-23
4.4 Test results and observations ................................................................................... 4-23
4.4.1 Test 1 ................................................................................................................ 4-25
4.4.2 Test 2 ................................................................................................................ 4-28
4.4.3 Test 3 ................................................................................................................ 4-31
4.4.4 Test 4a .............................................................................................................. 4-34
4.4.5 Test 4b .............................................................................................................. 4-37
4.4.6 Test 4c .............................................................................................................. 4-40
4.4.7 Test 5 ................................................................................................................ 4-42
4.5 Discussion of test results .......................................................................................... 4-46
4.5.1 Test 1: Mullion couple without clip or hollow stiffener under negative wind
pressure loading ............................................................................................... 4-46
4.5.2 Test 2: Mullion couple without a clip or a hollow stiffener under positive wind
pressure loading ............................................................................................... 4-48
4.5.3 Test 3: Mullion couple with a clip at mid-span under negative wind pressure
loading .............................................................................................................. 4-51
4.5.4 Tests 4a, 4b and 4c: Mullion couple with a clip under positive wind pressure
loading .............................................................................................................. 4-53
4.5.5 Test 5: Mullion couple with a clip and a hollow stiffener under negative wind
pressure loading ............................................................................................... 4-57
4.5.6 Comparison between the different loading and the mullion couple systems ... 4-58
4.6 Conclusions .............................................................................................................. 4-62
4.7 Recommendations .................................................................................................... 4-63
4.7.1 Mullion couple without any clips or hollow system (or hollow stiffeners) ..... 4-63
4.7.2 Mullion couple with a clip at mid-span ............................................................ 4-64
4.7.3 Mullion couple with hollow stiffeners or mullion couple made of hollow
mullions with or without clips.......................................................................... 4-64
5 FINITE ELEMENT ANALYSES OF MULLION COUPLES WITHOUT CLIP OR
HOLLOW SYSTEMS

5.1 General ....................................................................................................................... 5-1

5.2 Development of finite element models ...................................................................... 5-1
5.3 Development of simple model ................................................................................... 5-6
5.3.1 General ............................................................................................................... 5-6
5.3.2 Idealization of the mullion couple ...................................................................... 5-6
5.3.3 Boundary conditions .......................................................................................... 5-8
5.3.4 Loading method ................................................................................................. 5-9

vii
 Table of contents

5.3.5 Finite element type ........................................................................................... 5-11

5.3.6 Mesh details ...................................................................................................... 5-15
5.3.7 Contacts and constraints ................................................................................... 5-18
5.3.8 Material properties of mullions ........................................................................ 5-20
5.3.9 Types of finite element analysis ....................................................................... 5-23
5.4 Development of full model....................................................................................... 5-31
5.4.1 General ............................................................................................................. 5-31
5.4.2 Idealization of the model .................................................................................. 5-31
5.4.3 Boundary conditions......................................................................................... 5-34
5.4.4 Loading method ................................................................................................ 5-35
5.4.5 Selection of Finite Elements ............................................................................. 5-36
5.4.6 Mesh ................................................................................................................. 5-38
5.4.7 Contact and constraints .................................................................................... 5-41
5.4.8 Material properties ........................................................................................... 5-42
5.4.9 Type of finite element analysis......................................................................... 5-45
5.5 Results and discussion .............................................................................................. 5-47
5.5.1 General ............................................................................................................. 5-47
5.5.2 Simple model .................................................................................................... 5-47
5.5.3 Full model......................................................................................................... 5-58
5.6 Conclusion…………………………………………………………………………....5-64

6 PARAMETRIC STUDY OF MULLION COUPLES WITHOUT CLIP OR

HOLLOW SYSTEMS UNDER NEGATIVE WIND ACTION

6.1 General ....................................................................................................................... 6-1

6.2 Mullion sections and influential parameters ............................................................... 6-1
6.3 Finite element modelling of mullion couples ............................................................. 6-5
6.4 Comparison of full and simple finite element models ................................................ 6-7
6.5 Effects of different parameters on the structural behaviour and capacity of mullion
couples ...................................................................................................................... 6-12
6.5.1 Gap between the mullions ................................................................................ 6-12
6.5.2 Effectiveness of lateral restraint provided by glass pane ................................. 6-14
6.5.3 Load distribution profile ................................................................................... 6-16
6.5.4 Ratio of the loads acting on male and female mullions .................................... 6-18
6.5.5 Intermediate transoms ...................................................................................... 6-21
6.6 Comparison of FEA and design rule predictions ..................................................... 6-29
6.6.1 General ............................................................................................................. 6-29

viii
 Table of contents

6.6.2 FEA and Design rule predictions ..................................................................... 6-29

6.6.3 Capacity charts for mullion couples under negative wind action .................... 6-36
6.7 Conclusion ............................................................................................................... 6-39
6.8 Recommendations .................................................................................................... 6-40
7 PARAMETRIC STUDY OF MULLION COUPLES WITHOUT CLIP OR
HOLLOW SYSTEMS UNDER POSITIVE WIND ACTION

7.1 General ....................................................................................................................... 7-1

7.2 Mullion sections and influential parameters .............................................................. 7-1
7.3 Effects of different parameters on the structural behaviour and capacity of mullion
couples ....................................................................................................................... 7-3
7.3.1 Gap between the mullions .................................................................................. 7-3
7.3.2 Effectiveness of lateral restraint provided by glass pane ................................... 7-6
7.3.3 Ratio of the loads acting on male and female mullions ..................................... 7-8
7.4 Comparison of FEA and current design predictions ................................................ 7-10
7.4.1 General ............................................................................................................. 7-10
7.4.2 FEA and current design predictions ................................................................. 7-11
7.4.3 Capacity charts for mullion couples under positive wind action ..................... 7-19
7.5 Conclusion ............................................................................................................... 7-21
7.6 Recommendations .................................................................................................... 7-22
8 NUMERICAL INVESTIGATION OF MULLION COUPLES WITH CLIP AND/OR
HOLLOW SYSTEMS

8.1 General ....................................................................................................................... 8-1

8.2 Validation of the FE models of mullion couples with a clip ...................................... 8-1
8.2.1 Development of FE models ................................................................................ 8-2
8.2.2 Results and discussion........................................................................................ 8-3
8.3 Mullion couples with clip system ............................................................................ 8-14
8.3.1 General ............................................................................................................. 8-14
8.3.2 Mullion couples with clip under negative wind action .................................... 8-15
8.3.3 Mullion couples with clip under positive wind action ..................................... 8-22
8.4 Mullion couples with hollow system ....................................................................... 8-27
8.4.1 General ............................................................................................................. 8-27
8.4.2 Mullion couples with hollow under negative wind action ............................... 8-29
8.4.3 Mullion couples with hollow under positive wind action ................................ 8-34
8.5 Mullion couples with clip and hollow systems ........................................................ 8-36
8.5.1 General ............................................................................................................. 8-36
8.5.2 Mullion couples with hollow and clip under negative wind action ................. 8-36

ix
 Table of contents

8.5.3 Mullion couples with hollow and clip under positive wind action ................... 8-37
8.6 Serviceability performance of mullion couples ........................................................ 8-39
8.7 Conclusion ................................................................................................................ 8-42
8.8 Recommendations .................................................................................................... 8-43
9 CONCLUSIONS AND RECOMMENDATIONS

9.1 Conclusions ................................................................................................................ 9-2

9.1.1 Mechanical property tests ................................................................................... 9-2
9.1.2 Full-scale experimental study ............................................................................. 9-2
9.1.3 Development and validation of FE models of mullion couples without any clip or
hollow systems ................................................................................................... 9-3
9.1.4 Parametric study of mullion couples without any clip/hollow systems under
negative wind action ........................................................................................... 9-4
9.1.5 Parametric study of mullion couples without any clip/hollow systems under
positive wind action............................................................................................ 9-4
9.1.6 Parametric study of mullion couples with clip and/or hollow systems .............. 9-5
9.2 Recommendations for the design of mullion couples ................................................ 9-6
9.2.1 Full scale testing ................................................................................................. 9-7
9.2.3 Design rules ........................................................................................................ 9-8
9.3 Future research ......................................................................................................... 9-12

APPENDIX A .......................................................................................................................... A-1

APPENDIX B........................................................................................................................... B-1

APPENDIX C .......................................................................................................................... C-1

REFERENCES ........................................................................................................................ R-1

x
 List of figures

LIST OF FIGURES
Figure 1.1: Glazed aluminium facades....................................................................................... 1-1
Figure 1.2: Typical load transfer mechanism of a building facade ............................................ 1-2
Figure 1.3: Unitized facade panel installation............................................................................ 1-3
Figure 1.4: Components of an aluminium facade system .......................................................... 1-3
Figure 1.5: Mullion couple......................................................................................................... 1-3
Figure 1.6: Structural and captive glazing systems .................................................................... 1-4
Figure 1.7: Behaviour of mullion couples.................................................................................. 1-5
Figure 1.8: Mullion couple with clip systems ............................................................................ 1-7
Figure 1.9: Mullion couple with hollow systems ....................................................................... 1-8
Figure 1.10: Custom shapes of mullion couple sections used in industry ................................. 1-9
Figure 2.1: Material models for aluminium (CEN, 2007) .......................................................... 2-5
Figure 2.2: Four point bending test model used by Wang (2006) .............................................. 2-9
Figure 2.3: Four point bending test set-up used by Wang (2006) ............................................ 2-10
Figure 2.4: Test conducted by Zyla (2010) .............................................................................. 2-11
Figure 2.5: Section geometry of mullions used by Yip and Carrick (1995) ............................ 2-12
Figure 2.6: General arrangement of test set-up used by Yip and Carrick (1995) .................... 2-13
Figure 2.7: Loading set-up and deflection measurement layout (Yip and Carrick, 1995) ...... 2-13
Figure 2.8: Simulation of lateral and end rotational restraints (Yip and Carrick, 1995) .......... 2-14
Figure 2.9: Summary of loads at serviceability failure (Yip and Carrick, 1995) ..................... 2-15
Figure 2.10: Mullion couple used by Hui et al. (2015) ............................................................ 2-16
Figure 2.11: Testing of a single beam comprises of male and female mullions (Hui et al., 2015)
.................................................................................................................................................. 2-17
Figure 2.12: Full-scale testing of unitized facade panels (Hui et al., 2015)............................. 2-18
Figure 2.13: Silicone joint shear test (Yip and Carrick, 1995) ................................................ 2-20
Figure 2.14: Shear load versus displacement (Yip and Carrick, 1995) .................................... 2-20
Figure 2.15: Seal resistance...................................................................................................... 2-21
Figure 2.16: Friction load versus displacement of neoprene seal (Yip and Carrick, 1995) ..... 2-21
Figure 2.17: Failure modes from experiment and finite element analysis (Wang, 2006) ........ 2-23
Figure 2.18: Discretized ABAQUS model used by Huang (2014) .......................................... 2-24
Figure 2.19: Idealization of the geometry of complex mullions (Wang, 2006) ....................... 2-25
Figure 2.20: Discretized shell finite element model used by Wang (2006) ............................. 2-25
Figure 2.21: Three regimes of local buckling of elements (Kissell and Ferry, 1995) .............. 2-28
Figure 3.1: Tensile coupon geometry and dimensions ............................................................... 3-2
Figure 3.2: Mullion sample after tensile coupon extraction....................................................... 3-2
Figure 3.3: Coating thickness measurements in μm .................................................................. 3-3

xi
 List of figures

Figure 3.4: Calibration details of Instron testing machine ......................................................... 3-4

Figure 3.5: Thickness measurement of a tensile coupon using digital Vernier caliper .............. 3-4
Figure 3.6: Tensile test set-up using Instron testing machine .................................................... 3-4
Figure 3.7: Bending test specimen ............................................................................................. 3-6
Figure 3.8: Mullion sample after cutting a bending test specimen ............................................. 3-6
Figure 3.9: Schematic diagram of a beam under four-point bending ......................................... 3-7
Figure 3.10: Calibration of weights ............................................................................................ 3-7
Figure 3.11: Four-point bending test .......................................................................................... 3-8
Figure 3.12: Measured stress versus strain curves of 650-027 and 650-028 mullions ............. 3-11
Figure 3.13: EN 1999-1-1 (CEN, 2007) model for elastic region and test results ................... 3-14
Figure 3.14: EN 1999-1-1 (CEN, 2007) model for plastic region and test results ................... 3-15
Figure 3.15: Ultimate fracture strain predictions ..................................................................... 3-17
Figure 3.16: Ultimate strain predictions ................................................................................... 3-17
Figure 3.17: Predictions for proof strength by means of 0.1% offset ...................................... 3-18
Figure 3.18: M/I versus 1/R relationship of various specimens ............................................... 3-22
Figure 4.1: Typical mullion couple in a captive glazed facade assembly .................................. 4-4
Figure 4.2: Components of a captive glazed unitized facade assembly ..................................... 4-5
Figure 4.3: Interlocking part removed from male mullion ......................................................... 4-6
Figure 4.4: Continuous assembly of unitized facade panels ...................................................... 4-8
Figure 4.5: Installation of test panels ......................................................................................... 4-9
Figure 4.6: Torsional rotation of the side mullion and preventive mechanism ........................ 4-11
Figure 4.7: Restraint to torsional rotation of the side mullions ................................................ 4-12
Figure 4.8: Sub-sill fixed to the test rig using TEKS screws ................................................... 4-13
Figure 4.9: Transoms to sub-head and sub-sill assembly details ............................................. 4-14
Figure 4.10: Sealing of openings between test rig and side mullions ...................................... 4-16
Figure 4.11: Application of silicone spray to the seal strips .................................................... 4-17
Figure 4.12: Deflection measurements using LVDTs .............................................................. 4-18
Figure 4.13: Calibration of a LVDT using a standard block .................................................... 4-19
Figure 4.14: X-axis displacement measurements of the mullions ............................................ 4-20
Figure 4.15: Final test set-up .................................................................................................... 4-22
Figure 4.16: Description of the direction of translation and rotation of the mullions .............. 4-24
Figure 4.17: Mullion couple behaviour in Test 1 ..................................................................... 4-27
Figure 4.18: Applied negative wind pressure versus displacement curves of the mullion couple
in Test 1 ............................................................................................................... 4-28
Figure 4.19: Mullion couple behaviour in Test 2 ..................................................................... 4-30
Figure 4.20: Applied positive wind pressure versus displacement curves of the mullion couple
in Test 2 ............................................................................................................... 4-31

xii
 List of figures

Figure 4.21: Mullion couple behaviour in Test 3 ..................................................................... 4-33

Figure 4.22: Applied negative wind pressure versus displacement curves of the mullion couple
in Test 3 ............................................................................................................... 4-34
Figure 4.23: Mullion couple behaviour in Test 4a ................................................................... 4-36
Figure 4.24: Applied positive wind pressure versus displacement curves of the mullion couple
in Test 4a ............................................................................................................. 4-37
Figure 4.25: Mullion couple behaviour in Test 4b ................................................................... 4-39
Figure 4.26: Applied positive wind pressure versus displacement curves of the mullion couple
in Test 4b ............................................................................................................. 4-40
Figure 4.27: Mullion couple behaviour in Test 4c ................................................................... 4-42
Figure 4.28: Mullion couple behaviour in Test 5 ..................................................................... 4-44
Figure 4.29: Applied negative wind pressure versus displacement curves of the mullion couple
in Test 5 ............................................................................................................... 4-45
Figure 4.30: Typical load transfer from glass to mullions ....................................................... 4-47
Figure 4.31: Schematic diagram of the behavior of mullions in Test 1 ................................... 4-48
Figure 4.32: Schematic diagram of the behavior of mullions in Test 2 ................................... 4-50
Figure 4.33: Schematic diagram of the behavior of mullions in Test 3 ................................... 4-53
Figure 4.34: Schematic diagram of the behavior of mullions in Test 4a ................................. 4-55
Figure 4.35: Schematic diagram of the behavior of mullions during Test 4b .......................... 4-56
Figure 4.36: Schematic diagram of the behavior of mullions during Test 4c .......................... 4-57
Figure 4.37: Schematic diagram of the behavior of mullions during Test 5 ............................ 4-58
Figure 4.38: Mullion couples with and without a clip under negative wind action ................. 4-59
Figure 4.39: Mullion couples with and without a clip under positive wind action .................. 4-60
Figure 4.40: Mullion couples under positive and negative wind actions 4-61
Figure 5.1: Typical test set-up.................................................................................................... 5-3
Figure 5.2: Idealized test panel based on symmetry .................................................................. 5-5
Figure 5.3: Mullion geometry .................................................................................................... 5-7
Figure 5.4: Idealized simple finite element model ..................................................................... 5-7
Figure 5.5: Boundary conditions applied to the simple model .................................................. 5-9
Figure 5.6: Load applied on the simple model ......................................................................... 5-10
Figure 5.7: Naming convention used in ABAQUS for the three-dimensional shell elements . 5-12
Figure 5.8: Comparison of pressure versus displacement curves from simple models with
different types of shell element ................................................................................................ 5-14
Figure 5.9: Comparison of von-Mises stress plots from simple models with different types of
shell element ......................................................................................................... 5-15
Figure 5.10: Discretized simple model .................................................................................... 5-16

xiii
 List of figures

Figure 5.11: Comparison of pressure versus displacement curves from simple models with
different mesh sizes ............................................................................................. 5-17
Figure 5.12: Comparison of von-Mises stress plots from models with different mesh sizes ... 5-18
Figure 5.13: Comparison of results from models with different types of contact stiffness ...... 5-19
Figure 5.14: Comparison of results from models with different values of friction coefficient 5-20
Figure 5.15: Stress versus strain curves of Test 1 and 2 mullions ............................................ 5-22
Figure 5.16: Back tracking in Riks method .............................................................................. 5-24
Figure 5.17: Comparison of results obtained from models with different mass scaling .......... 5-27
Figure 5.18: Comparison of results obtained from different analysis procedures .................... 5-29
Figure 5.19: Comparison of failure modes obtained from different analysis procedures ........ 5-30
Figure 5.20: Idealization of the gasket geometry ..................................................................... 5-32
Figure 5.21: Comparison of results from full model with different idealizations of gasket
geometry .............................................................................................................. 5-33
Figure 5.22: Idealized full finite element model ...................................................................... 5-34
Figure 5.23: Boundary conditions applied to the full model .................................................... 5-35
Figure 5.24: Application of a uniform negative wind pressure load ........................................ 5-36
Figure 5.25: Naming convention used in ABAQUS for three-dimensional solid elements ..... 5-37
Figure 5.26: Comparison of results from full models with different brick elements for gaskets ....
.................................................................................................................................................. 5-38
Figure 5.27: Discretized full model in ABAQUS/Standard ..................................................... 5-39
Figure 5.28: Comparison of results from full models with different mesh sizes under negative
wind action .......................................................................................................... 5-40
Figure 5.29: Comparison of results from full models with general and contact pair approaches ...
.................................................................................................................................................. 5-42
Figure 5.30: Stress versus strain curves of DC983 silicone rubber (Dow Corning, 2017) ...... 5-43
Figure 5.31: Comparison of results from models with different aluminium sheet density ...... 5-46
Figure 5.32: Mullion couple behaviour in FEA (simple model) and Test 1 ............................ 5-49
Figure 5.33: Comparison of results from FEA (simple model) and Test 1 .............................. 5-50
Figure 5.34: Energy ratio for FEA (simple model) - Test 1 ..................................................... 5-50
Figure 5.35: Mullion couple behaviour in FEA (simple model) and Test 2 ............................ 5-52
Figure 5.36: Comparison of results from FEA (simple model) and Test 2 .............................. 5-53
Figure 5.37: Comparison of results from FEA (simple model) with spring lateral restraint and
Test 2 ................................................................................................................... 5-54
Figure 5.38: Energy ratio for FEA (simple model) - Test 2 ..................................................... 5-55
Figure 5.39: Pressure versus displacement curves for single and coupled mullions under
negative wind action ............................................................................................ 5-57
Figure 5.40: Comparison of results from FEA (full model) and Test 1 ................................... 5-59

xiv
 List of figures

Figure 5.41: Energy ratio versus pressure for FEA (full model) - Test 1 ................................ 5-60
Figure 5.42: Predicted failure mode of the mullion couple by full model – Test 1 ................. 5-60
Figure 5.43: Load carried by the mullion couple in the full model - Test 1 ............................ 5-61
Figure 5.44: Comparison of results from FEA (full model) and Test 2 ................................... 5-62
Figure 5.45: Energy ratio versus pressure for FEA (full model) - Test 2 ................................ 5-63
Figure 5.46: Predicted failure modes of the mullion couple by full model and Test 2 ............ 5-63
Figure 5.47: Load sharing to the mullion couple in the full model - Test 2…………………….…….5-64
Figure 6.1: Shapes of mullion couples ....................................................................................... 6-2
Figure 6.2: Naming convention used to label mullion couples .................................................. 6-3
Figure 6.3: Captive and structural glazing mullions .................................................................. 6-4
Figure 6.4: Failure modes predicted by full and simple FE models of mullion couple MC5 .... 6-8
Figure 6.5: Moment versus displacement curves predicted by full and simple FE models of
mullion couple MC5 ............................................................................................... 6-8
Figure 6.6: Failure modes predicted by full and simple FE models of mullion couple MC12 .. 6-9
Figure 6.7: Moment versus displacement curves predicted by full and simple FE models of
mullion couple MC12 ............................................................................................. 6-9
Figure 6.8: Failure modes predicted by full FE model of mullion couple MC15 with single and
double glazing ........................................................................................................ 6-10
Figure 6.9: Failure modes predicted by full and simple FE models of mullion couple MC15 6-11
Figure 6.10: Moment versus displacement curves predicted by full and simple FE models of
mullion couple MC15 ......................................................................................... 6-11
Figure 6.11: Typical gap between male and female mullions.................................................. 6-12
Figure 6.12: Moment versus displacement curves of mullion couple MC5 with different gaps ....
.................................................................................................................................................. 6-13
Figure 6.13: Moment versus displacement curves of mullion couple MC15 with different gaps ..
.................................................................................................................................................. 6-13
Figure 6.14: Typical spring lateral restraint under negative wind action ................................. 6-14
Figure 6.15: Moment versus displacement curves of mullion couple MC5 with different lateral
restraint levels ..................................................................................................... 6-15
Figure 6.16: Moment versus displacement curves of mullion couple MC15 with different lateral
restraint levels ..................................................................................................... 6-15
Figure 6.17: Load distributions ................................................................................................ 6-16
Figure 6.18: Moment versus displacement curves of mullion couple MC5 with different load
distribution profiles ............................................................................................. 6-17
Figure 6.19: Moment versus displacement curves of mullion couple MC15 with different load
distribution profiles ............................................................................................. 6-17
Figure 6.20: Typical loads acting on male and female mullions of a mullion couple ............. 6-18

xv
 List of figures

Figure 6.21: Failure modes of mullion couple MC5 with load distribution ratios of 1:2 and 2:1 ...
.................................................................................................................................................. 6-19
Figure 6.22: Failure modes of mullion couple MC15 with load distribution ratios of 1:2 and 2:1
.................................................................................................................................................. 6-20
Figure 6.23: Moment versus displacement curves of mullion couple MC5 with different load
distribution ratios (a:b) ........................................................................................ 6-20
Figure 6.24: Moment versus displacement curves of mullion couple MC15 with different load
distribution ratios (a:b) ........................................................................................ 6-21
Figure 6.25: Load and moment profiles of mullion couples with one transom........................ 6-22
Figure 6.26: Failure mode of mullion couple MC5 with an intermediate transom at 900 mm 6-23
Figure 6.27: Mullion couple MC5 with and without an intermediate transom ........................ 6-23
Figure 6.28: SPACE GASS (SG, 2014) analysis result for the unit pressure loading ............. 6-24
Figure 6.29: Load and moment distribution profiles of mullion couples with two intermediate
transoms............................................................................................................... 6-26
Figure 6.30: Failure modes of mullion couples MC5 and MC15 with two intermediate transoms
.................................................................................................................................................. 6-27
Figure 6.31: Moment versus displacement curves of mullion couples MC5 and MC15 with two
intermediate transoms .......................................................................................... 6-27
Figure 6.32: Moment versus displacement curves of mullion couple MC5 with one and two
intermediate transoms .......................................................................................... 6-28
Figure 6.33: CUFSM model of mullion couple MC10 ............................................................ 6-30
Figure 6.34: CUFSM analysis results of mullion couple MC10 .............................................. 6-31
Figure 6.35: CUFSM analysis results of mullion couple MC12 .............................................. 6-31
Figure 6.36: Combined and individual behaviour of male and female mullions ..................... 6-35
Figure 6.37: Moment versus displacement curves of mullion couple MC3 ............................. 6-36
Figure 6.38: Moment capacity predictions of FEA versus span of mullion couples ................ 6-36
Figure 6.39: Capacity chart for mullion couple MC5 under negative wind action .................. 6-38
Figure 7.1: Removing the leg of male mullion.............................................................................. 2
Figure 7.2: Moment versus displacement curves of mullion couple MC5 with different gaps ..... 4
Figure 7.3: Moment versus displacement curves of mullion couple MC15 with different gaps ... 4
Figure 7.4: Failure modes of mullion couple MC5 with different gaps ........................................ 5
Figure 7.5: Failure modes of mullion couple MC15 with different gaps (4200 mm span) ........... 5
Figure 7.6: Typical spring lateral restraint under positive wind action ......................................... 6
Figure 7.7: Moment versus displacement curves of mullion couple MC5 with different lateral
restraint levels............................................................................................................. 7
Figure 7.8: Moment versus displacement curves of mullion couple MC15 with different lateral
restraint levels............................................................................................................. 7

xvi
 List of figures

Figure 7.9: Typical loads acting on male and female mullions of a mullion couple .................... 8
Figure 7.10: Failure modes of mullion couple MC5 with load distribution ratios of 1:2 and 2:1 9
Figure 7.11: Failure modes of mullion couple MC15 with load distribution ratios of 1:2 and 2:1
....................................................................................................................................................... 9
Figure 7.12: Moment versus displacement curves of mullion couple MC5 with different load
distribution ratios (a:b) ........................................................................................... 10
Figure 7.13: Moment versus displacement curves of mullion couple MC15 with different load
distribution ratios (a:b) ........................................................................................... 10
Figure 7.14: CUFSM analysis results of mullion couple MC10 ................................................. 12
Figure 7.15: CUFSM analysis results of mullion couple MC12 ................................................. 13
Figure 7.16: Combined and individual behaviour of male and female mullions ........................ 18
Figure 7.17: Moment versus displacement curves of the mullion couple MC3 .......................... 18
Figure 7.18: Capacity chart for mullion couple MC5 under positive wind action ...................... 20
Figure 8.1: Mullion couple with a clip at mid-span ................................................................... 8-3
Figure 8.2: Mullion couple behaviour in FEA (simple model) and Test 3 ................................ 8-5
Figure 8.3: Comparison of results from FEA (simple model) and Test 3 .................................. 8-6
Figure 8.4: Energy ratio for FEA (simple model) - Test 3 ......................................................... 8-6
Figure 8.5: Mullion couple behaviour in FEA (simple model) and Test 4a .............................. 8-8
Figure 8.6: Comparison of results from FEA of simple model and Test 4a .............................. 8-9
Figure 8.7: Energy ratio for FEA (simple model) - Test 4a ....................................................... 8-9
Figure 8.8: Mullion couple behaviour in FEA (full model) and Test 3 ................................... 8-10
Figure 8.9: Comparison of results from FEA (full model) and Test 3 ..................................... 8-11
Figure 8.10: Energy ratio for FEA (full model) - Test 3 .......................................................... 8-11
Figure 8.11: Mullion couple behaviour in FEA (full model) and Test 4a................................ 8-12
Figure 8.12: Comparison of results from FEA (full model) and Test 4a ................................. 8-13
Figure 8.13: Energy ratio for FEA (full model) - Test 4a ........................................................ 8-13
Figure 8.14: Mullion couples with a clip at mid-span.............................................................. 8-14
Figure 8.15: Mullion couple with and without continuous clip system ................................... 8-15
Figure 8.16: Predicted failure modes of mullion couple MC5 with and without a clip under
negative wind action ........................................................................................... 8-16
Figure 8.17: Predicted failure modes of mullion couple MC15 with and without a clip under
negative wind action ........................................................................................... 8-16
Figure 8.18: Predicted failure modes of mullion couple MC2 with and without a clip under
negative wind action ........................................................................................... 8-17
Figure 8.19: Load and shear centre details of mullion couple MC2 ........................................ 8-17
Figure 8.20: Moment capacity ratio of mullion couples with and without a clip versus span
based on FEA under negative wind action .......................................................... 8-19

xvii
 List of figures

Figure 8.21: Reduction in unbraced length due to the use of clip under negative wind action 8-19
Figure 8.22: Predicted failure modes of mullion couple MC5 with and without continuous clip
system under negative wind action...................................................................... 8-22
Figure 8.23: Moment versus displacement curves of mullion couple MC5 with and without
continuous clip system under negative wind action ............................................ 8-22
Figure 8.24: Predicted failure modes of mullion couple MC5 with and without a clip under
positive wind action............................................................................................. 8-23
Figure 8.25: Predicted failure modes of mullion couple MC15 with and without a clip under
positive wind action............................................................................................. 8-23
Figure 8.26: Predicted failure modes of mullion couple MC2 with and without a clip under
positive wind action............................................................................................. 8-24
Figure 8.27: Moment capacity ratio of mullion couples with and without a clip versus span
based on FEA under positive wind action ........................................................... 8-25
Figure 8.28: Predicted failure modes of mullion couple MC5 with and without continuous clip
system under positive wind action ...................................................................... 8-26
Figure 8.29: Moment versus displacement curves of mullion couple MC5 with and without
continuous clip system under positive wind action ............................................. 8-26
Figure 8.30: Hollow stiffener and hollow mullions ................................................................. 8-27
Figure 8.31: Screw connections in mullion couples ................................................................. 8-28
Figure 8.32: Mullion couples with hollow systems used in analyses ....................................... 8-28
Figure 8.33: Moment versus displacement curves of mullion couples MC5 and MC2 with
different hollow sizes under negative wind action .............................................. 8-29
Figure 8.34: Moment versus displacement curves of hollow mullion couples with full and spring
lateral restraints under negative wind action ....................................................... 8-32
Figure 8.35: Predicted failure modes of mullion couples MC5 and MC2 with a hollow mullion
under negative wind action .................................................................................. 8-33
Figure 8.36: Moment versus displacement curves of mullion couples MC5 and MC2 with a
hollow mullion under negative wind action ........................................................ 8-33
Figure 8.37: Moment versus displacement curves of hollow mullion couples with full and spring
lateral restraints under positive wind action ........................................................ 8-34
Figure 8.38: Predicted failure modes of mullion couples MC5 and MC2 with a hollow mullion
under positive wind action................................................................................... 8-35
Figure 8.39: Moment versus displacement curves of mullion couples MC5 and MC2 with a
hollow mullion under positive wind action ......................................................... 8-35
Figure 8.40: Mullion couples with a hollow and a clip ............................................................ 8-36
Figure 8.41: Predicted failure modes of mullion couples MC5 and MC2 with hollow and clip
systems under negative wind action .................................................................... 8-37

xviii
 List of figures

Figure 8.42: Moment versus displacement curves of mullion couples MC5 and MC2 with
hollow and clip systems under negative wind action .......................................... 8-37
Figure 8.43: Predicted failure modes of mullion couples MC5 and MC2 with hollow and clip
systems under positive wind action ..................................................................... 8-38
Figure 8.44: Moment versus displacement curves of mullion couples MC5 and MC2 with
hollow and clip systems under positive wind action ........................................... 8-38
Figure 8.45: Serviceability performance of mullion couple MC5 with and without hollow
systems under negative wind action (4200 mm span)…………………………8-39
Figure 9.1: Combined and individual behaviour of male and female mullions. ........................ 9-7
Figure 9.2: Typical mullion couple without any clip or hollow systems ................................... 9-9
Figure 9.3: Typical mullion couple with clip system ................................................................. 9-9
Figure 9.4: Typical mullion couple with different hollow systems ......................................... 9-10
Figure 9.5: Typical mullion couple with clip and hollow systems .......................................... 9-11
Figure A. 1: Measured stress versus strain curves of 6063-T6 aluminium alloy ...................... A-2
Figure A. 2: EN 1999-1-1 (CEN, 2007) model for plastic region and test results .................. A-11
Figure A. 3: Specimen under four-point bending ................................................................... A-22
Figure B. 1: Measured thicknesses of sample taken from Test 1 .............................................. B-1
Figure B. 2: Measured thicknesses of sample taken from Test 2 .............................................. B-2
Figure B. 3: Measured thicknesses of sample taken from Tests 3 and 4a ................................. B-4
Figure B. 4: Stress versus strain curves of Test 3 mullions ...................................................... B-5
Figure B. 5: Stress versus strain curves of Test 4a mullions..................................................... B-6
Figure B. 6: Thicknesses of the clip parts used in FE modelling .............................................. B-7
Figure B. 7: Load transfer from aluminum sheets to center mullion couple ............................. B-7
Figure B. 8: Shear deformation ................................................................................................. B-9
Figure B. 9: Shear stress versus strain curve of DC983 silicone rubber (Dow Corning, 2017) B-9
Figure B. 10: Different load and lateral restraint locations under negative wind action......... B-11
Figure B. 11: Pressure versus displacement curves of mullion couple with different stiffness
values under negative wind action .................................................................... B-12
Figure B. 12: Different load and lateral restraint locations under positive wind action ......... B-13
Figure B. 13: Pressure versus displacement curves of mullion couple with different load and
lateral restraint cases under positive wind action .............................................. B-14
Figure B. 14: Typical mullion to transom connection ............................................................ B-15
Figure B. 15: Pressure versus displacement curves of mullion couple with different support
conditions .......................................................................................................... B-16
Figure C. 1: Representation of the section properties ............................................................... C-3
Figure C. 2: Representation of the section properties ............................................................... C-5
Figure C. 3: Capacity charts for mullion couples under negative wind action ....................... C-11

xix
 List of figures

Figure C. 4: Reduction factors and conditions of mullion couple MC1 under negative wind
action ................................................................................................................... C-12
Figure C. 5: Capacity charts for mullion couples under positive wind action ........................ C-17
Figure C. 6: Reduction factors and conditions of mullion couple MC1 under positive wind
action ................................................................................................................... C-18
Figure C. 7: Simply supported beam with UDL...................................................................... C-19

xx
 List of tables

LIST OF TABLES
Table 2.1: Experimental results (De Martino et al., 1990) ......................................................... 2-3
Table 2.2: Alternative parameters for the exponent n (De Martino et al., 1990) ....................... 2-3
Table 2.3: Mechanical properties of I-sections made of 6063-T6 aluminium alloy (Kim, 2003) ..
.................................................................................................................................................... 2-7
Table 2.4: Mechanical properties of 6061-T6 and 6063-T5 aluminium alloys (Su et al., 2014) ....
.................................................................................................................................................... 2-7
Table 2.5: Member moment capacity formulae ....................................................................... 2-31
Table 3.1: Details of mullion samples and tensile coupons ....................................................... 3-1
Table 3.2: Details of mullion samples and bending test specimens ........................................... 3-5
Table 3.3: Summary of test results (average values) ............................................................... 3-15
Table 3.4: Flexural modulus of elasticity of 6063-T6 aluminium alloy .................................. 3-22
Table 4.1: Full-scale test details ................................................................................................. 4-7
Table 4.2: Summary of full-scale test results ........................................................................... 4-45
Table 4.3: Typical section properties of the mullions used ...................................................... 4-47
Table 5.1: Conventional stress/displacement shell elements (DS, 2014a) ............................... 5-12
Table 5.2: Conventional stress/displacement solid elements (DS, 2014a) ............................... 5-38
Table 5.3: Ultimate bending moment capacity predictions under negative wind action (kNm) .....
.................................................................................................................................................. 5-56
Table 5.4: Ultimate bending moment capacity predictions under positive wind action (kNm)
………………………………………………………………………………………………...5-58
Table 6.1: Details of mullion couples ........................................................................................ 6-3
Table 6.2: Moment capacity predictions for mullion couple MC5 with and without an
intermediate transom ............................................................................................... 6-24
Table 6.3: Moment gradient factor Cb of mullion couple MC5 based on FEA and ADM (AA,
2015) ....................................................................................................................... 6-25
Table 6.4: Moment gradient of mullion couples with two intermediate transoms ................... 6-28
Table 6.5: Moment capacity predictions of FEA and ADM (AA, 2015) ................................. 6-32
Table 6.6: Comparison of FEA and ADM (AA, 2015) predictions of mullion couples that show
interaction............................................................................................................... 6-33
Table 6.7: Comparison of FEA and ADM (AA, 2015) predictions of mullion couples in which
male and female mullions behave individually ....................................................... 6-34
Table 7.1: Details of mullion couples ........................................................................................ 7-2
Table 7.2: Moment capacity predictions of FEA and Yield moment capacity ........................ 7-13
Table 7.3: Comparison of FEA predictions and Yield moment capacity ................................ 7-14
Table 7.4: Moment capacity predictions of FEA and ADM (AA, 2015) ................................. 7-15

xxi
 List of tables

Table 7.5: Comparison of FEA and ADM (AA, 2015) predictions of mullion couples with shear
centre outside the line of action of loading ............................................................. 7-16
Table 7.6: Comparison of FEA and ADM (AA, 2015) predictions of mullion couples with shear
centre inside the line of action of loading ............................................................... 7-17
Table 8.1: Moment capacities of the mullion couples with and without a clip at mid-span based
on FEA under negative wind action ........................................................................ 8-18
Table 8.2: Moment capacities of the mullion couples with and without a clip based on ADM
(AA, 2015) under negative wind action .................................................................. 8-20
Table 8.3: Moment capacities of the mullion couples with a clip based on FEA and ADM (AA,
2015) under negative wind action ........................................................................... 8-20
Table 8.4: Moment capacities of the mullion couples with and without a clip based on FEA
under positive wind action....................................................................................... 8-24
Table 8.5: Moment capacities of the mullion couples with different hollow sizes based on FEA
under negative wind action...................................................................................... 8-30
Table 8.6: Moment capacities of the mullion couples with different hollow sizes based on FEA
and ADM (AA, 2015) under negative wind action ……………………………….8-31
Table A.1: Sample details ......................................................................................................... A-1
Table A. 2: Summary of test results ........................................................................................ A-20
Table B. 1: Measured section properties of sample taken from Test 1 ..................................... B-1
Table B. 2: Measured section properties of sample taken from Test 2 ..................................... B-2
Table B. 3: Section properties of male mullion - Test 1 ........................................................... B-3
Table B. 4: Section properties of female mullion - Test 2 ........................................................ B-3
Table B. 5: Mullion couple under positive wind action with different Young’s modulus of
gasket and gasket-aluminium sheet frictional coefficient .................................... B-16
Table C. 1: Section properties of the male and female mullions of mullion couples MC1 to
MC20 ...................................................................................................................... C-1
Table C. 2: Section properties required for the design calculations .......................................... C-3
Table C. 3: Moment capacity calculations based on ADM (AA, 2015) ................................... C-4
Table C. 4: Section properties required for the design calculations .......................................... C-5
Table C. 5: Moment capacity calculations based on ADM (AA, 2015) ................................... C-6

xxii
 Statement of original authorship

STATEMENT OF ORIGINAL AUTHORSHIP

The work contained in this thesis has not been previously submitted to meet
requirements for an award at this or any other higher education institution. To the best
of my knowledge and belief the thesis contains no material previously published or
written by another person except where due reference is made.

QUT Verified Signature

Thananjayan Sivaprakasam
24 November 2019

xxiii
 Acknowledgements

ACKNOWLEDGEMENTS

First of all, I would like to express my gratitude to my principal supervisor Prof. Mahen
Mahendran for his great support, guidance and integrity throughout this journey. It has
been a pleasure to work under you. I would also like to thank my associate supervisor
Dr. Kesawan Sivakumar for his valuable guidance and tremendous support throughout
each step of this journey. I am lucky to have him as one of my supervisors. Then I
would like to thank Mr. Jim Stringfellow, Dr. Balachandren Baleshan, Mr. Jason
Boldery and Mr. Mitchell McPherson, structural design engineers of G.James Glass and
Aluminium Pty. Ltd., for their valuable technical guidance. I would also like to
acknowledge Dr. Shanmuganathan Gunalan, now who is a lecturer at Griffith
University, for his valuable support during the early phase of this research study. I
would like to express my gratitude to the staff members of G.James Glass and
Aluminium Pty. Ltd., Eng. Allan Hickey, Mr. Glen Thomas, Mr. Glen Middlebrook,
Mr. Brian lee and other staff members, for providing test specimens and valuable
support to conduct full-scale wind pressure tests using their laboratory facility. I would
like to thank Mr. Gregory Paterson, Mr. Benjamin Brownlee and other staff members of
‘O’ block structural laboratory at QUT for their valuable support to conduct laboratory
tests.

I would like to acknowledge the staff members of high performance computing (HPC)
and research support services for providing high performance computing facilities,
analysis software packages and other necessary services. I would also like to thank
Australian Research Council (ARC), Queensland University of Technology (QUT) and
G.James Glass and Aluminium Pty. Ltd. for providing financial support to my research
study.

I would like to express my sincere gratitude to all my teachers who taught me since my
preschool and the lecturers at the University of Moratuwa, Sri Lanka, for laying the
foundation of my early career. I am also lucky to have Prof. M.T.R Jayasinghe and Prof.
Chintha Jayasinghe throughout my formative years, and would like to extend my thanks
for their valuable support. I would also like to acknowledge Kasun D. Kariyawasam,
Kasun Nandapala and other friends at the University of Moratuwa for their support.

xxiv
 Acknowledgements

As a new postgraduate student to QUT and Brisbane, I was guided by Janarthanan

Balasubramaniam and Myuran Kathekeyan, and would like to extend my thanks to
them. Then, it is my pleasure to thank my fellow post-graduate students, Hanwellage
Yomal Viduranga Dias, Harikrishnan Magarabooshanam, Mohamed Imran Ahamed and
others, for their support. I would also like to thank my housemates Myuran Kathekeyan
and Jasnav Pancheti, and Mayooran Sivapathasundaram for their support and
friendship.

Finally, I would like to express my sincere appreciation to my parents and brothers for
their blessings, providing endless support and encouragement. I would like to thank my
uncle, Rakulan, and aunt, Rajani, for their support. I would also like to thank my wife,
Sugarniya Subramaniam, for giving moral support throughout this journey. My
apologies to anyone who were not acknowledged here.

xxv
 Abbreviations

ABBREVIATIONS

ADM - Aluminium Design Manual

AR - Aspect Ratio

AS - Australian Standard

DOF - Degrees of freedom

FE - Finite Element

FEA - Finite Element Analysis

FEM - Finite Element Modelling

IE - Internal Energy

KE - Kinetic Energy

SLS - Serviceability Limit State

TSR - Transom Spacing Ratio

UDL - Uniformly Distributed Load

ULS - Ultimate Limit State

xxvi
 Introduction

CHAPTER 1

1 INTRODUCTION

1.1 Background

Aluminium facade systems are commonly used in high-rise buildings as the external
envelope and are primarily designed to protect the occupants from wind, rain, noise and
other environmental hazards. These facade systems comprise aluminium frame and
infilled glass/granite/aluminium panels. Glass facades became the choice for many
designers due to the increasing aesthetic and natural lighting requirements (Figure 1.1).
The aluminium frame that supports the glass panels is connected to the main structure,
usually to the slab edges. In terms of load transfer mechanism of aluminium facade
systems, the aluminium frame transfers its own weight and the wind loads acting on the
glass panels to the main structure. Figure 1.2 shows the lateral load transfer mechanism
from the facade panels to the main structural frame.

Figure 1.1: Glazed aluminium facades

(Image from http://gjames.com [Accessed on May 14, 2016])

Steel was used in the supporting frame when glazed facades were introduced in the 18th
century, but with time aluminium has become the preferred material due to its excellent
features considering production, construction and maintenance. Aluminium is lighter
than steel giving an exceptional strength to weight ratio. It also has good ductility and
hence less susceptible to brittle fracture. Aluminium is alloyed with a variety of metals
to improve its strength. 6063-T6 aluminium alloy (solution heat treated and artificially
aged magnesium-silicone alloy) has good extrudability, corrosion resistance and surface
finish. Further, extrudability of aluminium makes it easier to manufacture sections of

1-1
 Introduction

complex geometries based on designers’ needs. Therefore, 6063-T6 aluminium alloy is

commonly used to manufacture the frames of unitized facades in Australia.

Glass panes
Transoms

Mullions

Wind load

Figure 1.2: Typical load transfer mechanism of a building facade

Types of facade system and its components

There are mainly two categories of facade systems: stick and unitized systems. Stick
systems, the first generation of facade systems, are installed piece by piece on site, thus
provide flexibility for installation, but they require more time and labour. Stick systems
are more attractive for low-rise building construction. Unitized systems are
prefabricated as modules in the factory and delivered as panels to the site. The panels
are then assembled together in sequence and connected to the main structural systems
(Figure 1.3). Since the panels of unitized systems are factory produced, they have many
advantages over construction time and quality control, and are also economically
beneficial in large projects (Allana and Carter, 2012).

In unitized facades, the frame is an assembly of mullions and transoms. Mullions are the
vertical members spanning two or more floors in height that transfer the lateral wind
loads acting on the glass to the supports through bending and shear actions. Transoms
are the horizontal members connected to the mullions (Figure 1.4). In unitized facades,
mullion couples are used, which are a combination of male and female mullions. This
facilitates the prefabrication and site assembly thus rapid construction, and allows for
expansion and shortening of the panels caused by temperature variations, and building
deflections due to wind and earthquake actions. Mullion couples are thin-walled

1-2
 Introduction

aluminium members with complex asymmetric open profiles, and are custom extruded
based on designers’ requirements to provide more practical and efficient solutions
considering all the aspects including structural, thermal, noise and water penetration
(Figure 1.5). It is difficult to accurately predict the structural behaviour of these mullion
couples under wind actions.

Transom
Mullion
Glass
Sealant
Covering cap
Glass sealant

Figure 1.3: Unitized facade Figure 1.4: Components of an aluminium

panel installation facade system
(Image from http://nrwindows.com (Image from http://profalis.lt
[Accessed on May 14, 2016]) [Accessed on April 1, 2016])

(a) Male mullion (b) Female mullion (c) Male and Female
mullions together
Figure 1.5: Mullion couple

Note: s and c are the shear centre and centroid of the mullion sections, respectively.

1-3
 Introduction

Captive and structural glazing mullions

Mullion geometry in unitized facades can vary based on the type of glazing systems
used, which can be either captive or structural. In captive glazed system, the glass is
fixed to the frame using gaskets (Figure 1.6(a)), whereas in structurally glazed system
glass is attached to the frame by means of structural silicone (Figure 1.6(b)). Architects
prefer structural glazing over captive glazing due to their thermal efficiency and
aesthetic appearance. Since the glass to aluminum connections in mullions used in
captive and structurally glazed systems are different, the structural behaviour and
capacity of the mullions used in such systems can also differ between them. Therefore,
the mullions used in both structural and captive glazing systems need to be investigated.

Glass to mullion Glass to mullion

connection using connection using
gasket sealant
Mullion
Mullion

Transom

Gasket Transom

Glass

Glass

(a) Captive glazing (b) Structural glazing

(Image from http://gjames.com [Accessed on (Image from http://reynaers.com [Accessed on
September 13, 2016]) September 13, 2016])

Figure 1.6: Structural and captive glazing systems

Mullion couples subject to wind actions

Facade system subject to wind actions can fail due to the failure of either glass panel or
its supporting mullions and associated connections. These failures can lead to broken
glass panels and associated components falling from the top floors of high-rise
buildings. This can lead to a severe disaster involving loss of lives and properties.
Furthermore, breaking of facade systems can directly expose the building occupants to
wind action, leading to catastrophic situation. Therefore it is critical to ensure the safe

1-4
 Introduction

function of the facade system during high wind events including both positive
(pressure) and negative (suction) wind actions.

During the positive wind action (wind acting towards the building), the bending
moment at the mid-span of the mullions induces compressive and tensile stresses on the
external and internal sides of the mullions, respectively, as shown in Figure 1.7(a). In
this case the mullion couples are laterally restrained in the compression zone (outside)
by the glass panes (Figure 1.7(a)). Mullions are open sections similar to channel
sections, thus the shear centre is located outside the sections as shown in Figure 1.5.
Since the loads are not acting at the shear centre, the mullion couple will open up on the
inner side (see Figure 1.7(a)) due to elastic torsional rotation.

Inside- Inside-Compression
Tension zone zone

Gap opens up Gap closes up

Load transfer from

glass to mullions Load transfer from
glass to mullions

Outside- Outside-
Compression Tension zone
zone
Positive wind action Negative wind action
(a) Gap opening under positive wind (b) Gap closing under negative wind
action action

Figure 1.7: Behaviour of mullion couples

During the negative wind action (wind acting away from the building), the bending
moment at mid-span of the mullion induces tensile and compressive stresses on the
external and internal sides of the mullions, respectively, as shown in Figure 1.7(b). In
this case, the lateral restraints provided by glass are on the tension side, thus no
restraints are available to the critical compression flange. However, due to the elastic
torsional deformations, the mullion couple will rotate to lean against each other. This
may result in male and female mullions providing lateral support to each other, but it is

1-5
 Introduction

not yet proven. Due to the aforementioned complexities in the behaviour of mullion
couples under different wind actions, the structural engineers have to make conservative
assumptions in the analysis and design of these members. That is, the mullion couples
are designed separately without considering any advantages offered by the male and
female mullion interaction as a couple.

Current knowledge on the structural design of mullions subject to wind actions

For positive wind action, design engineers assume that adequate lateral restraints are
provided by the glass to the critical flange of the mullion couple along its length. Hence,
they determine the structural bending capacity of the mullion couple by adding the
section capacity of the male and female mullions. Wang (2006) investigated the
behaviour of a mullion couple assuming the glass restraint to be adequate to resist the
lateral movements in accordance with Clift and Austin (1989). However, it is still
questionable whether the glass provides adequate rotational restraints. If the glass does
not provide adequate restraints, torsional deformations can occur in the mullions, where
the capacity of the mullion couple could be less than their section moment capacity due
to the combined bending and torsional actions. In this case, adding the section moment
capacities of the male and female mullion capacities in design might lead to unsafe
designs, leading to premature failure of the facade systems.

For negative wind action, it is generally assumed that no lateral restraints are available
to the critical flange of the mullion sections, and design engineers determine the
capacity of the mullion couple by adding together the lateral torsional buckling
capacities, which were found separately for the male and female mullions without
considering the combined couple behaviour. However, the actual capacity of the
mullion couple might be higher than the addition of the individual capacities due to the
leaning action of the mullion couple. Torsional forces are also generally neglected in the
capacity calculations. Hence the predicted structural capacity can be different to the true
capacity of the mullion couples, leading to unconservative or conservative designs.
Although some investigations have been conducted on mullions (Clift and Austin, 1989,
Kim, 2003, Wang, 2006, Yip and Carrick, 1995, Zyla, 2010, Hui et al., 2015, Kesawan
and Mahendran, 2017, Kesawan et al., 2017a, Kesawan et al., 2017b, Kesawan and
Mahendran, 2019), they were not comprehensive. Therefore, limited knowledge is
available on this subject, and these problems need to be addressed. The current

1-6
 Introduction

aluminium design standards including Australian (SA, 1997), American (AA, 2015) and
European standards (CEN, 2007) do not give a detailed design approach to determine
the structural capacity of mullion couples. Furthermore, serviceability considerations
(deflection in the direction of wind load, lateral movements and opening up of mullion
couples) are also important in design (Yip and Carrick, 1995).

In some instances, clip systems (either separate clip parts or continuous clip as shown in
Figures 1.8(a) and (b), respectively) are used to increase the ultimate and serviceability
capacities of mullion couples. The clips shown in Figure 1.8(a), which are separate
pieces of extrusion, are commonly used at the mid-span of the mullion couple whereas
continuous clips (Figure 1.8(b)) are introduced while the sections are extruded.
However, these clip systems are not commonly/frequently used in the industry as they
resist thermal movements and removal of the facade panels in the case of any changes
in the already installed facade systems. The use of clip systems is an interim solution,
and hardly any literature is available on their use. In addition, the effects of clip systems
on the bending moment capacity of the mullion couples are not addressed to date.
Design standards or manuals (SA, 1997, AA, 2015, CEN, 2007) also do not provide any
guidelines in relation to the use of clip systems. Therefore, the effects of clip systems on
the structural behaviour and capacity of mullion couples need to be investigated in
detail under both positive and negative wind actions.

Continuous
clip
Separate clip
parts

(a) Separate extruded clip parts (b) Continuous clip

Figure 1.8: Mullion couple with clip systems

Hollow systems are also sometimes used in order to increase the ultimate capacity of
mullion couples. Hollow system can be either a separate hollow stiffener (see Figure

1-7
 Introduction

1.9(a)) or extruded hollow mullions (see Figure 1.9(b)). In the hollow stiffener system, a
separate hollow stiffener section is inserted into the mullion couple as shown in Figure
1.9(a) so that the mullion couple and hollow section will act together giving a higher
capacity than the capacity of the open mullion couple. Unlike the hollow stiffener
system, mullions comprising hollow parts are extruded in the hollow mullion system
(Figure 1.9(b)). To date, the structural behaviour and capacity of the mullion couples
with hollow system has not been studied. Furthermore, design standards or manuals
(SA, 1997, AA, 2015, CEN, 2007) do not provide any guidelines in relation to the use
of hollow systems. Therefore, mullion couples with hollow systems needs to be
investigated in detail under both positive and negative wind actions.

Hollow stiffener Hollow mullion

(a) Hollow stiffener (b) Hollow mullions

Figure 1.9: Mullion couple with hollow systems

Complexity of mullion profiles and importance of economical facade design

Architectural designs of modern building facades are primarily driven by cutting edge
aesthetic appearance requirements. This, in general, forces the designers unequivocally
to use complex custom shaped mullion sections to reform their thoughts into reality.
These shapes could vary between projects and facade companies. For instance, Figures
1.10(a) and (b) show some complex shaped mullion couple sections used by different
facade companies. However, although their shape changes, glazing pocket, screw ports
and ports for gasket and seal strips (see Figure 1.10(a)) are the basic needs. Moreover,
such complexities in shape and the structural behaviour of mullion couples discussed
earlier make the structural design of facade mullions stands unique.

1-8
 Introduction

Gasket Seal strips

Glazing
pocket

Screw ports

(a) G.James Glass and Aluminium (http://gjames.com)

Figure 1.10: Custom shapes of mullion couple sections used in industry

Gasket slots

Glazing pocket

Screw ports

(b) Ullrich Aluminium (UA, 2005) (c) Northern Facades

(Image from http://www.northernfacades.com
[Accessed on April 15, 2018])
Figure 1.10: Custom shapes of mullion couple sections used in industry

The glazed facade system claims a significant portion (about 20%) of the total cost in a
building project (Mudie and Rawlinson, 2010). Furthermore, there is a possibility of
reducing the use of materials and thus associated costs if the structural behaviour of the
coupled mullions is well understood (Yip and Carrick, 1995, Hui et al., 2015, Lee et al.,
2017, Lee et al., 2018). Considering all the problems discussed above, it was proposed
to conduct a research study to investigate the structural behaviour and capacity of the
coupled mullion sections subject to both negative and positive wind actions.

1.2 Research problem

Mullion couples made of asymmetric male and female sections are commonly used in
unitized facade systems of high-rise buildings, and are subjected to out of plane loading
1-9
 Introduction

under wind actions. However, the structural behaviour of coupled mullion sections
made of asymmetric male and female mullions is not well understood. Generally, these
mullions are designed based on the individual behavior of the male and female
mullions, which does not represent the couple’s true behaviour. When exposed to wind
actions, male and female mullions interact, and thus there is a possibility of mullion
couples reaching a higher capacity than the addition of the individual mullion
capacities. Since mullion couples are asymmetric and lateral restraint is provided only
on one side by the glass, i.e., the restrained flange is under compressive stress under
positive wind action while it is under tensile stress under negative wind action, the
mullion couple behaviour could be different under positive and negative wind actions.
The available design standards and guidelines do not have any provisions to address
these problems. In addition to this, structural engineers use clip systems and/or hollow
systems as an interim solution to increase the ultimate and serviceability limit state
capacities of mullions. However, the effects of these clip and hollow systems on the
behavior and capacity of mullions are not known, and the currently available design
standards do not address them. More details were provided in the last section.
Therefore, it is important to conduct a comprehensive investigation of mullion couples
with and without clips and/or hollow systems under wind actions, to address these
issues and to develop suitable design guidelines.

1.3 Research aim and tasks

The aim of this research is to investigate the structural behaviour and capacity of
aluminium mullion couples made of asymmetric male and female mullion sections used
in unitized facades of high-rise buildings subject to both positive and negative wind
actions in order to develop a suitable design method to determine their moment
capacities. For this purpose, the following tasks are proposed:

a. An extensive literature review based on previous investigations conducted on the

structural behaviour of mullion couples.
b. Determine the accurate mechanical properties of aluminium mullions through
tensile coupon and bending tests.
c. Conduct full-scale tests of unitized facades made of selected mullion couples
without clip or hollow systems, with clip system, and with hollow system to

1-10
 Introduction

develop reliable structural behavioural data of mullion couples under both

positive and negative wind actions.
d. Develop finite element models using ABAQUS finite element program to
predict the structural behaviour and capacity of mullion couples and validate
them using full-scale test results.
e. Conduct an extensive parametric study of different mullion couples without any
clip or hollow systems under negative wind action using the validated finite
element models.
f. Conduct an extensive parametric study of different mullion couples without any
clip or hollow systems under positive wind action using the validated finite
element models.
g. Conduct an extensive numerical investigation of different mullion couples with
clip and/or hollow systems using the validated finite element models.
h. Evaluate the applicability of the currently used design assumptions and the
available design rules to predict the moment capacities of mullion couples under
wind actions, and propose suitable design methods and recommendations.

1.4 Thesis Outline

The outline of this thesis is as follows:

x Chapter 2: A comprehensive literature review of mechanical properties of

mullions, experimental study, numerical analysis of mullions and
design rule predictions are given. Finally, it presents the main
literature review findings.
x Chapter 3: Mechanical properties of 6063-T6 aluminium alloy mullions were
determined using tensile coupon and bending tests. It presents the
details, results and discussion of these tests.
x Chapter 4: It presents the details, results and discussions of the full-scale wind
pressure tests conducted on a unitized facade mullion couple
without clip, with clip and with hollow system exposed to negative
and positive wind actions.
x Chapter 5: Full and simple finite element models of the test mullion couple
without any clip or hollow system were developed, and their results
were validated using the full-scale test results. Full details are given

1-11
 Introduction

for the finite element model development and analysis procedures

in this chapter.
x Chapter 6: This chapter presents the details of a finite element analysis based
parametric study conducted under negative wind action on 20
mullion couples without clip or hollow systems, and the results
and discussion. It also compares the predictions of design rules
with the finite element analysis predictions, and provides suitable
recommendations.
x Chapter 7: This chapter presents the details, results and discussion of a finite
element analysis based parametric study conducted under positive
wind action on the same mullion couples investigated in Chapter 6.
It also compares the predictions of current design approaches with
the finite element analysis predictions, and provides suitable
recommendations.
x Chapter 8: It presents the details, results and discussion of the extensive
numerical investigation conducted on mullion couples with clip
and/or hollow systems under wind actions. It also compares the
predictions of current design approaches with the finite element
analysis predictions, and provides suitable recommendations.
x Chapter 9: This chapter presents the overall conclusions and recommendations
of this research investigation.

1-12
 Literature review

CHAPTER 2

2 LITERATURE REVIEW

2.1 General

In present day construction, glazed facade systems became the choice of structural
design engineers and are now widely used in single to more than 200 storey buildings.
In the modern architecture, aesthetic requirements seem to be a primary fact in the
selection of building profiles and materials apart from structural engineering concerns.
Hence, glazed facade systems are highly preferred as natural light can penetrate deep
into the buildings, and they also provide a nice vision to the outdoor nature. Unitized
facade system, which is one kind of façade system, is preferred in high-rise buildings
since it is economically beneficial due to significant reduction in construction time,
labour force, storage requirement, and on-site quality control requirements.

Unitized facade panels are supported by aluminium frames. Generally, the vertical
members of these frames are made of two separate mullion sections, called as male and
female mullions or mullion couple, which are connected together. This facilitates faster
and economical construction (Figure 1.3). Furthermore, the geometry of the mullions
are often asymmetric. The structural response of the mullion couple subject to wind
actions is difficult to predict through routine design methods. At present, although many
standards are available for the design of aluminium structures, they do not address the
design of mullion couples in detail. It is not possible to conduct extensive large-scale
laboratory tests to identify these issues as it requires a lot of time and resources. Due to
the advancements in computer technology, many finite element analysis packages are
available for the analysis and design of complex structures. Hence, finite element
models can be used to study this problem. However, the developed finite element
models should be validated using the experimental results for reliability, and hence
some full-scale tests are also needed.

In order to understand the problems and the available literature on the couple behaviour
of mullion sections in depth, a literature review has been undertaken in the area of
experimental studies, numerical analyses, and design methods to provide a robust
background to the proposed investigation. This section contains four main parts. The
first part of the review discusses about the mechanical properties of aluminium

2-1
 Literature review

mullions, while the second part gives the background knowledge on the physical testing
of aluminium mullions. The third part gives details about the advanced finite element
modeling and analysis techniques used in the previous studies, and highlights their
limitations. The fourth part gives an overview of the currently available design
approaches for the design of mullion sections in various design standards and
literatures.

2.2 Mechanical properties of aluminium

Aluminium shows a non-linear stress-strain behaviour with considerable strain

hardening beyond the yield point. The most widely used isotropic material model
(Equation 2.1) for non-linear materials such as aluminium was initially proposed by
Ramberg and Osgood (1943), and then modified by Hill (1944). Equation 2.1 is based
on conventional limit of elasticity (fe) calculated using 0.2% offset method and its
corresponding residual strain (εo,e) of 0.002.

ߪ ߪ ௡
ߝൌ ൅ ͲǤͲͲʹ ൬ ൰
‫ܧ‬ ݂଴Ǥଶ ---------------- (2.1)

where:
σ = stress
H = strain corresponding to the stress σ
E = modulus of elasticity
f0.2 = proof strength by means of 0.2% offset method
n = strain hardening exponent

The strain hardening exponent (n) in the Ramberg-Osgood equation has a significant
influence on the characteristic behaviour of stress-strain curves. The evaluation of
exponent (n) requires a second reference stress (fx) and conventional limit of elasticity
(fe) (Equation 2.2). De Martino et al. (1990) conducted an investigation on the stress-
strain behaviour of four different aluminium alloys listed in Table 2.1 where they
considered six different cases to evaluate the strain hardening exponent (n) as shown in
Table 2.2.

ߝ௢ǡ௘ ݂௘
݊ ൌ Ž ቆ ቇ Ȁ Ž ൬ ൰ ---------------- (2.2)
ߝ௢ǡ௫ ݂௫

2-2
 Literature review

where:
Ho,x = residual strain corresponding to the stress fx
Ho,e = residual strain corresponding to the stress fe

Table 2.1: Experimental results (De Martino et al., 1990)

1050-H14 5056-H11 6060-T6 6082-T4

E (MPa) 63920 71980 63410 68850

f0.1 (MPa) 81.9 228.2 214.7 254.6
f0.2 (MPa) 85.4 233.3 217.3 256.7
fp2 (MPa) 83.3 240.9 221.2 257.6
fc (MPa) 86.5 305.4 232.5 265.5
fmax (MPa) 89.9 387.1 244.4 273.7
Hmax (%) 1.79 9.91 7.06 5.97

Table 2.2: Alternative parameters for the exponent n (De Martino et al., 1990)

fe Ho,e fx n
݂଴Ǥଵ
݊ ൌ ŽሺͲǤͷሻ Ȁ Ž ൬ ൰
Curve 1 f0.2 0.002 f0.1 ݂଴Ǥଶ
ߝ௢ǡ௠௔௫ ݂௠௔௫
݊ ൌ Ž ቀ ቁ Ȁ Ž ൬ ൰
Curve 2 f0.2 0.002 fmax ͲǤͲͲʹ ݂଴Ǥଶ
ߝ௢ǡ௖ ݂௖
݊ ൌ Ž ቀ ቁ Ȁ Ž ൬ ൰
Curve 3 f0.2 0.002 fc ͲǤͲͲʹ ݂଴Ǥଶ
ͲǤͲͲͳ ݂଴Ǥଵ
݊ ൌ Ž ቆ ቇ Ȁ Ž ቆ ቇ
Curve 4 fp2 Hp2/2 f0.1 ߝ௢ǡ௣ଶ ݂௣ଶ
ߝ௢ǡ௠௔௫ ݂௠௔௫
݊ ൌ Ž ቆ ቇ Ȁ Ž ቆ ቇ
Curve 5 fp2 Hp2/2 fmax ߝ௢ǡ௣ଶ ݂௣ଶ
ߝ௢ǡ௖ ݂௖
݊ ൌ Ž ቆ ቇ Ȁ Ž ቆ ቇ
Curve 6 fp2 Hp2/2 fc ߝ௢ǡ௣ଶ ݂௣ଶ

Note:
f0.1 = proof strength by means of 0.1% offset method
fp2 = proof strength by means of half elastic slope method
fc = ඥ݂௠௔௫ ݂௣ଶ
fmax = ultimate tensile strength

2-3
 Literature review

Hmax = strain corresponding to stress fmax

Hp2 = strain corresponding to stress fp2
Ho,max = residual strain corresponding to stress fmax
Ho,c = residual strain corresponding to stress fc

Based on the error percentages evaluated by comparing the experimental and simulated
curves for different cases, it was suggested that the choice of conventional limit of
elasticity does not have a significant influence on the approximation of the curve (De
Martino et al., 1990). Furthermore, De Martino et al. (1990) suggested that the second
reference stress (fx) can be chosen depending on the criteria that needs to be considered,
i.e., if the analysis involves small deformations, f0.1 can be chosen as second reference
stress, and if the analysis involves large deformations, the use of fmax as second
reference stress will give a good approximation.

EN 1999-1-1 (CEN, 2007) provides various stress-strain models, such as piecewise

model which can be either bi-linear or tri-linear, and continuous model for aluminium
alloys as shown in Figure 2.1. They have varying levels of complexity, and therefore, an
appropriate model has to be adopted depending on the level of accuracy required.
Continuous model is represented in the form of σ = σ (ε) or ε=ε (σ). The Ramberg-
Osgood model in the form of ε=ε(σ) (Equation 2.1) is considered due to its simplicity in
terms of computation.

(a) Bi-linear model (b) Tri-linear model

Figure 2.1: Material models for aluminium (CEN, 2007)

2-4
 Literature review

(c) Continuous model in the form of (d) Continuous model in the form of
σ = σ (ε) ε = ε (σ)
Figure 2.1: Material models for aluminium (CEN, 2007)
Note:
Definitions to symbols which are not given here are given where they first appear.

fp = conventional elastic limit of proportionality

εe = strain corresponding to the stress fe
εp = strain corresponding to the stress fp
E1 = first hardening modulus
E2 = second hardening modulus

Determination of strain hardening exponent (n) in EN 1999-1-1 (CEN, 2007) is based

on different methods depending on the scope of analysis to be performed, that is,
whether the analysis involves elastic or plastic deformations. The general form of
equation for the determination of n is given by Equation 2.2. Equation 2.2 can be
expressed as Equation 2.3 when 0.2% proof strength (f0.2) and the corresponding
residual strain (0.002) are taken as the conventional limit of elasticity (fe) and its
corresponding residual strain (εo,e), respectively.

Ž൫ͲǤͲͲʹȀߝ௢ǡ௫ ൯
݊ൌ --------------- (2.3)
Žሺ݂଴Ǥଶ Τ݂௫ ሻ

When the analysis is limited to elastic deformations, Equation 2.4 can be used to
determine n taking f0.1 as the second reference stress (fx). As an alternative to Equation
2.4, Equation 2.5 is suggested based on a wide range of test results in EN 1999-1-1
(CEN, 2007). A very small strain of 0.000001 is used as the residual strain
corresponding to the proportional limit stress of fp.

2-5
 Literature review

Žሺʹሻ
݊ൌ --------------- (2.4)
Žሺ݂଴Ǥଶ Τ݂଴Ǥଵ ሻ
ŽሺͲǤͲͲͲͲͲͳȀͲǤͲͲʹሻ
݊ൌ --------------- (2.5)
Ž൫݂௣ Τ݂଴Ǥଶ ൯

where the proportional limit (fp) is expressed in terms of f0.2:

݂௣ ൌ ݂଴Ǥଶ െ ʹඥͳͲ݂଴Ǥଶ ሺܰȀ݉݉ଶ ሻ if f0.2 > 160 N/mm2 --------------- (2.6)

݂௣ ൌ ݂଴Ǥଶ Τʹ if f0.2 ≤ 160 N/mm2 --------------- (2.7)

When the analysis is concerned about plastic deformations, Equation 2.8 can be used to
determine n, taking fmax as the second reference stress (fx). As an alternative to Equation
2.8, Equation 2.9 is suggested based on a wide range of test results (CEN, 2007).

Ž൫ͲǤͲͲʹȀߝ௢ǡ௠௔௫ ൯
݊ൌ ----------------- (2.8)
Žሺ݂଴Ǥଶ Τ݂௠௔௫ ሻ

ŽሺͲǤͲͲʹȀߝ௨ ሻ
݊ൌ ---------------- (2.9)
Žሺ݂଴Ǥଶ Τ݂௠௔௫ ሻ

where:
εu = nominal value of ultimate fracture strain

In the absence of actual data, Hu can be determined using Equation 2.10, which is a
function of 0.2% proof strength (f0.2), and Equation 2.11 (CEN, 2007).

݂଴Ǥଶ ሺܰȀ݉݉ଶ ሻ
ߝ௨ ൌ ͲǤ͵Ͳ െ ͲǤʹʹ ݂଴Ǥଶ ൏ ͶͲͲܰȀ݉݉ଶ ----------------(2.10)
ͶͲͲ

ߝ௨ ൌ ͲǤͲͺ ݂଴Ǥଶ ൒ ͶͲͲܰȀ݉݉ଶ ----------------(2.11)

Other than this, the equation for the calculation of ultimate strain (Hmax) corresponding
to ultimate tensile strength (fmax) proposed by Su et al. (2014) is given in Equation 2.12,
where fmax/f0.2 must be greater than 1.01.

ߝ௠௔௫ ൌ ͲǤͳ͵ሺͳ െ ݂଴Ǥଶ Τ݂௠௔௫ ሻ ൅ ͲǤͲ͸ ----------------(2.12)

2-6
 Literature review

Determination of Young’s modulus (E) is a challenging task for materials such as

aluminium with non-linear stress-strain behaviour. Aluminium design standards
specifies typical values of tensile modulus of elasticity in the range of 68250-73500
MPa, which is about 2% less than the values of compressive modulus of elasticity given
in the standards, and it varies depending on the type of aluminium alloys (Kissel and
Ferry, 1995). In some cases, the actual test results obtained from laboratory tensile tests
were found to be generally lower than that in design standards (Kim, 2003, Su et al.
2014, Huang and Young, 2014). This variation can be due to the errors in strain
measurements using clip-on extensometer caused by eccentricity introduced during the
tensile tests. Mechanical properties obtained by Kim (2003) and Su et al., (2014) are
summarized in Tables 2.3 and 2.4, respectively.

Table 2.3: Mechanical properties of I-sections made of 6063-T6 aluminium alloy

(Kim, 2003)

Specimen name fe (f0.2) fmax E E

He Hmax
(ksi) (ksi) (ksi) (MPa)
Sample 1 (flange) 29.91 0.00497 34.24 0.06250 10068 69416
Sample 2 (web) 29.82 0.00514 33.78 0.05400 9509 65562
Sample 3 (web) 29.92 0.00506 33.89 0.06580 9775 67396
Sample 4 (flange) 28.84 0.00508 33.41 0.06270 9359 64528
Sample 5 (flange) 29.57 0.00598 33.98 0.06700 7430 51228
Mean 29.61 0.00525 33.86 0.06240 9228 63624
COV 0.015 0.079 0.009 0.081 0.113 0.113

Table 2.4: Mechanical properties of 6061-T6 and 6063-T5 aluminium alloys

(Su et al., 2014)

E fe fmax
Specimen n
(MPa) (MPa) (MPa)
66000 234 248 12
65000 193 207 22
71000 229 242 11
6061-T6
69000 226 238 10
(high-strength alloy)
65000 225 234 13
67000 228 240 12
69000 215 229 8
69000 179 220 10
6063-T5 71000 139 194 9
(normal strength alloy) 69000 181 228 9
72000 151 181 11

2-7
 Literature review

Wang (2006) used the mechanical properties of 6063-T6 aluminium alloy given in
ADM (AA, 2000) for the investigation of bending capacity of two custom extruded
aluminium sections that are used in unitized facade systems using finite element
analyses, and found that the finite element analysis results are less than the experimental
results. Therefore, he conducted tensile tests to determine the actual mechanical
properties. He used the measured mechanical properties in the finite element models
and showed that the finite element analysis results agreed well with the experimental
results. Furthermore, Gardner and Ashraf (2006) investigated the aluminium hollow
sections under compression and bending actions and found that the use of bi-linear
material model for the non-linear materials leads to inaccurate and conservative
predictions. Ramberg-Osgood continuous material model was preferred for round house
type materials such as aluminium for the accurate prediction of material behaviour (De
Martino et al., 1990, Gardener and Ashraf, 2006).

It is important to use the accurate mechanical properties and material models in finite
element modelling and design equations to obtain accurate results. The most suitable
material model for 6063-T6 aluminium alloy used in Australia needs to be identified.
This study, therefore, aims to perform tensile tests to determine the accurate mechanical
properties of 6063-T6 aluminium alloy and compare the results with the models given
in EN 1999-1-1 (CEN, 2007) in order to find a suitable material model. In this process,
various equations given in EN 1999-1-1 (CEN, 2007) to determine the strain hardening
exponent (n) will be evaluated to find the most suitable one to be used in Ramberg-
Osgood model for the 6063-T6 aluminium alloy.

2.3 Experimental study of facade mullions

2.3.1 Bending capacity tests of mullions

Laboratory experiments are more reliable than numerical or analytical methods as they
reflect the real scenario, given that a proper test set-up is used and the procedures
followed comply with relevant standards. However, laboratory experiments acquire
more time and resources. AS 2047 (SA, 2014) provides standardized test methods and
procedures to determine the ultimate load capacity of windows.

Few experimental studies have been undertaken in the past on asymmetric mullion
couples to investigate their structural behaviour and capacity under wind actions (Wang,

2-8
 Literature review

2006, Zyla, 2010, Yip and Carrick, 1995, Hui et al., 2015). Wang (2006) conducted four
point bending tests to determine the structural behaviour and bending capacity of two
custom extruded E-type aluminium mullions (Figure 2.2(a)). A schematic diagram of
the four point bending test used by Wang (2006) is shown in Figure 2.2. This set-up
simulates the situation where the bending moment is constant in the region between the
two point loads while the shear force is zero in the same region as shown in Figure
2.2(b).
(Load) (Load) E-type mullions

(a) Schematic diagram of the test set-up

P
P
Shear force diagram

P*a
Bending moment diagram

(b) Bending moment and shear force diagrams

Figure 2.2: Four point bending test model used by Wang (2006)

Wang (2006) initially used a set-up where timber stiffening blocks were used at the
supports and the loading points as shown in Figure 2.3(a). However, in his final test set-
up, aluminium plates were used instead of timber stiffening blocks to constrain the ends
of the two mullions, and no stiffeners were used at the loading points (Figure 2.3 (b)).

2-9
 Literature review

Wooden stiffening block at loading point

Wooden stiffening
block at the end

Load transfer beam

(a) Test set-up 1

Load transfer beam

Aluminium end plates

(b) Test set-up 2

Figure 2.3: Four point bending test set-up used by Wang (2006)

Wang (2006) made two basic assumptions in his study. The first assumption is that the
mullions in unitized facade systems are subjected to only bending moment and low
level of torsion. The second assumption is that the connections between the mullions
and the transoms and the glass give adequate lateral restraint to prevent the lateral
torsional buckling of the mullions as per Clift and Austin (1989). Therefore, his study
mainly focused on investigating the section moment capacity under pure bending.

Location of the loading points is also important for mullions with asymmetric open
sections as the load position from the shear centre of the mullion sections has a
significant influence on their structural capacity. In addition, the loading plate attached
to the mullions may give restraint to the lateral movement of the mullions. Furthermore,
the four point bending set-up fails to simulate the bending moment and shear force
combination of real facade systems which are subject to uniformly distributed loading.
Therefore, the experimental set-up used by Wang (2006) cannot be used to accurately
simulate the structural behaviour of a mullion couple.
2-10
 Literature review

Full-scale tests on a mullion couple comprised of a 451-016 (male) and a 451-017

(female) mullion, were conducted under both positive and negative wind actions by
Contract Glass & Aluminium Pty. Ltd (Zyla, 2010). The used test set-up is shown in
Figure 2.4. The length and the width of each panel were 2450 and 1125 mm,
respectively. This test set-up closely simulated the glass and mullion interaction
mentioned earlier. However, the loading set-up used, i.e. the four point bending test
which results in a bending moment and shear force distribution in mullions as shown in
Figure 2.2(b), cannot predict the actual behaviour of the mullion couple as it does not
simulate the torsional forces that exist in real facade systems. Furthermore, as only
limited tests have been done and load-displacement relationships were not obtained, a
comprehensive decision on the behaviour and the capacity of the mullion couples could
not be made.

Loading jack
Centre mullion
couple

Loading points
Glass

(a) Full-scale test set-up (b) Loading set-up

Figure 2.4: Test conducted by Zyla (2010)

An additional test on the same mullion couple was conducted under negative wind
action, where the male and female mullions were connected together using a 1#10
PK/SS screw installed at the mid-height. The outputs showed that the presence of screw
connecting the male and female mullions significantly increased the bending capacity of
the mullion couple. Since the inner unrestrained flanges of mullion couple experience
compressive stresses under negative wind, the use of screw will improve their coupling

2-11
 Literature review

action. Various clip systems, which are currently used, can also give similar effect as
that of the screw. However, a solid conclusion cannot be drawn based on only one test.
Therefore, the use of clip systems and their effects on the behaviour and the bending
capacity of mullion couples need to be further investigated in detail.

Yip and Carrick (1995) conducted large-scale experiments to investigate the behaviour
of mullion couples. They considered different parameters as listed next:
x Positive and negative wind actions.
x Mullions with and without rotational end restraints.
x Mullions with and without rubber seal strips (Gaskets).
x Mullions with and without lateral restraint given by glass through structural
silicone.
x Male and female mullions assembled with and without a gap.

The mullion sections used in their investigation are shown in Figure 2.5 while the test
set-up is shown in Figure 2.6. The mullion couple was simply supported with a span of
2500 mm and a uniformly distributed loading was achieved through loading at eight
points except for the upper bound tests which were conducted up to the failure of the
mullion couple (Figures 2.6 and 2.7). The upper bound tests were done by applying the
loads at third points. It was stated that the restraint provided by the loading cable to the
mullions is not significant due to its length and flexibility. The behaviour and the
interaction between male and female mullions were investigated by measuring the
deflections in the vertical and horizontal directions using calibrated potentiometer
transducers (Figure 2.7(c)).

Female mullion Male mullion

Figure 2.5: Section geometry of mullions used by Yip and Carrick (1995)

2-12
 Literature review

Typical loading point

Figure 2.6: General arrangement of test set-up used by Yip and Carrick (1995)

Mullion pair

Potentiometer
transducers

(a) Positive loading (b) Negative loading (c) Deflection measurement layout
set-up set-up

Figure 2.7: Loading set-up and deflection measurement layout

(Yip and Carrick, 1995)

The lateral restraint given by the glass through the shearing action of silicone seals was
simulated by attaching aluminium strips to the face of mullions and to arms which are
fixed to the steel tubes. The joint between the aluminium strips and the mullion face
was provided by means of structural silicone representing the real scenario. The
rotational/torsional end restraint was modeled by fixing aluminium sections to the ends
of mullions (Figure 2.8).

2-13
 Literature review

Force applied here

Restraint here

Shear centres

Interlock here

Aluminium sections fixed to the

end of mullions

Figure 2.8: Simulation of lateral and end rotational restraints

(Yip and Carrick, 1995)

Load and deflection data were recorded for the serviceability failure criteria under both
the positive and negative wind actions. Figure 2.9 shows a summary of the total load
carried by the mullion couple at serviceability failure. Figure 2.9(b) clearly shows that
the presence of seal strips (gaskets) did not have much influence on the load carrying
capacity of the mullion couple under negative wind loading. However, a significant
increment has occurred in the case of positive wind loading (Figure 2.9(a)). The lateral
restraint did not have much impact on the capacity under negative wind load, because
the flanges, which are laterally restrained, are under tension. But, as the laterally
restrained flanges become compression during the positive wind loading, a significant
increment in the capacity can be seen (Figure 2.9(a)).

2-14
 Literature review

(a) Positive wind load tests

(b) Negative wind load tests

Figure 2.9: Summary of loads at serviceability failure (Yip and Carrick, 1995)

2-15
 Literature review

Hui et al. (2015) conducted experimental and theoretical investigations on the bending
capacity of 6063-T6 aluminium alloy mullions. They studied the structural behaviour
and capacity of the mullion couples under different loading and support conditions.
They compared their experimental results with the predicted values that were calculated
based on the currently available different aluminium design standards. The mullions
used in the testing are of structurally glazed type, where the glass is connected to the
mullions using structural silicone (Figure 2.10(a)). The tested mullion couple also
comprised pothook, which is generally used to improve the serviceability performance,
i.e., reducing the opening-up of mullion couples (Figure 2.10(b)).

(a) Torsional rotation due to eccentric (b) Typical mullion assembly

loading
Figure 2.10: Mullion couple used by Hui et al. (2015)

Hui et al. (2015) used three different types of experimental set-ups, and they are: 1. a
single 2730 mm long beam comprising male and female mullions subjected to
trapezoidal load distribution; 2. complete panel loaded in a pressurized chamber; 3.
complete panel loaded using sand bags. The experimental set-up used for the single
beam test is shown in Figure 2.11(b), where the ends were restrained using brackets and
bolts. The bolts were arranged in the span direction and close to the top flange of the
mullions (Figure 2.11(b)). The trapezoidal loading was simulated by applying point
loads at six location of the span (Figure 2.11(a)).

2-16
 Literature review

(a) Schematic diagram of simulating trapezoidal loading

Bolt arrangement used to

connect bracket and mullions

Loading points

Bracket

End rotational restraint

(b) Test set-up 1

Figure 2.11: Testing of a single beam comprises of male and female mullions
(Hui et al., 2015)

In Test set-up 2 unitized facade panels assembled together were subjected to a uniform
wind pressure (Figure 2.12(a)). The assembled panel is 2730 mm in length and 2370
mm in width. They applied a pressure of 5 kPa, and could not further load up to mullion
couple’s failure capacity due to limitations in air supply. Because of this, they tried the
3rd test set-up (only for positive wind case) where they simulated the wind pressure load
using sand bags as shown in Figure 2.12(b).

2-17
 Literature review

(a) Test set-up 2

(b) Test set-up 3

Figure 2.12: Full-scale testing of unitized facade panels (Hui et al., 2015)

Hui et al. (2015) concluded that the load bearing capacity of the mullions are similar
irrespective of the test set-up, and that the load bearing capacity predictions using
different standards are approximately similar, but significantly less than the test results.
However, since all three test set-ups were not tested under identical conditions, i.e., for
instance Test set-up 2 was loaded only up to 5 kPa, and Test set-up 3 was used only for
the positive wind load case, such a conclusion is questionable. Furthermore, brackets,

2-18
 Literature review

which were used to simulate the restraint to torsional rotation provided by transoms, in
the single beam test may give partial restraint to bending rotation of the mullions, thus
end moments could be developed. This will not be present in the simply supported full-
scale test set-up. Therefore, a complete understanding about the structural behaviour of
the mullion couple cannot be drawn from their study.

In summary, previous experimental studies conducted on the bending behaviour of

mullion couples demonstrate that their behaviour is complex when subject to wind
actions. It is difficult to simulate the real behaviour of mullion couples using simple test
models. Full-scale tests conducted in the past are not adequate, and sufficient data is not
available on the structural capacity of mullion couples in unitized facade systems
subject to wind actions. Also, the previous studies were inadequate to differentiate the
structural behaviour of mullion couples with and without clips and hollow stiffeners,
and these need to be studied in detail. Therefore, it is important to perform large number
of full-scale tests to completely understand the structural behaviour of mullion couples
subject to wind actions.

2.3.2 Interface shear/friction between glass/aluminium-polymer surfaces

The behaviour of mullions in unitized facades is greatly influenced by the contact

between glass/aluminium and structural silicone/gaskets (Clift and Austin, 1989).
Therefore, the interface shear resistance given by the glass to the mullions through the
shearing action of structural silicone or gaskets is important to know. Furthermore, there
is a need for proper investigation of these interaction properties and their effects on the
structural behaviour and capacity of mullion couples. Yip and Carrick (1995) conducted
small-scale tests to evaluate the restraint given by the glass to the mullions, in order to
provide inputs to their finite element modeling. The test set-up used to investigate the
shear resistance given by the silicone joint is shown in Figure 2.13. It comprises of two
cranked aluminium strips of 50 mm and a bead of silicone (width=10 mm, height=10
mm and length=50 mm). Load was applied to pull the strips apart, and the load and
displacement readings were recorded. It was concluded that a high force is required
even for a small shear displacement of the silicone joint (Figure 2.14). Therefore,
silicone joints can be expected to provide effective lateral restraint to mullions.

2-19
 Literature review

Structural silicone
Section view

Applied Applied
load load

Plan view

Figure 2.13: Silicone joint shear test (Yip and Carrick, 1995)

Figure 2.14: Shear load versus displacement (Yip and Carrick, 1995)

Similar to silicone sealants, neoprene seal strips also have some degree of resistance to
the relative movements of mullions in a mullion couple (Figure 2.15). Therefore, Yip
and Carrick (1995) conducted some small-scale laboratary tests on mullion couples
assembled with neoprene rubber seal strips, at different temperatures, to investigate the
importance of seal resistance on the mullion behaviour. The tests were conducted at 0,
25, and 50oC. Load was applied as shown in Figure 2.15, and this test set-up simulates a
parellel seperation of mullions. It was concluded that the seal resistance is not sufficient
for design and the temperature effect is not significant (Figure 2.16). However, in an
actual scenario a rotational seperation will occur in a mullion couple, and at this point

2-20
 Literature review

seal strips can give more frictional resistance than parallel seperation. This is because
when the rotational seperation occurs, the normal force induced between the seal strips
and the mullion surfaces gets increased. Although the shear resistance given by the seal
strips seemed to be insignificant, it is important for the composite action of mullions,
particularly under positive wind loading. The importance of seal strips on the behaviour
of mullion couples can be understood through full-scale testing of facade panels.

Force

Apply load to 12mm rods

attached to mullion halves

Seal strips Mullions joined

with seal strips
Glass

Force

(a) Typical assembly of mullion couple (b) Seal resistance test (Yip and Carrick,
1995)
Figure 2.15: Seal resistance

Figure 2.16: Friction load versus displacement of neoprene seal

(Yip and Carrick, 1995)

2-21
 Literature review

2.4 Numerical analysis of facade mullions

2.4.1 General

Advanced finite element analyses are becoming a vital part in research studies since
laboratory tests are expensive and time consuming. There are finite element analysis
packages which can handle complex problems with reduced time and resources. Several
types of analysis methods are available, each having a unique characteristic for a
particular problem. In general, a problem can be static, quasi-static or dynamic in nature
in which linear or non-linear behaviour can be present. Depending on the nature of the
problem, an analysis type can be used to solve it, but it is possible to have various
methods in which the best efficient, accurate and economical method can be adopted.
ABAQUS analysis program version 6.14 will be used throughout this study and this
section discusses about the analysis techniques used in the previous studies and their
relevance and limitations to the proposed study.

2.4.2 Type of analysis

Aluminium mullion sections are generally complex in geometry as well as asymmetric.

They show a non-linear response under bending actions. Hence static linear analysis is
not suitable to study the actual behaviour of a mullion couple system. Therefore, non-
linear analysis has to be adopted in the static analysis or quasi-static analysis with
implicit/explicit integration scheme. Static non-linear analysis using static general
analysis procedure or modified Riks method is widely used in the computational
modeling of structural engineering problems. However, when large strain problems with
instabilities are involved, static non-linear analysis using static general analysis
procedure alone will not work. Therefore, static general analysis procedure with
damping factor or modified Riks method is generally used in ABAQUS/Standard (DS,
2014a).

Wang (2006) employed static non-linear analysis with modified Riks method for the
analysis, initially with tri-linear material model and later with the actual material model,
and obtained converged solutions. It can be seen from the failure mode obtained from
the experiment and the finite element analysis that the failure prediction from finite
element analysis matched well with that of experiment (Figure 2.17). His study mainly
focused on predicting the section moment capacity of a mullion couple, and his finite

2-22
 Literature review

element model did not involve any significant nonlinearities such as contact or global
instabilities. It is argued by a previous researcher that when a large number of degrees
of freedoms and many contact definitions are present, static non-linear analysis
produces significant convergence problems (Kaitila, 2007). Therefore, a solid
conclusion on the adaptability of static non-linear finite element analysis cannot be
made without a proper investigation based on models with instability and contact
interactions.

(a) Experimental model (b) Finite element model

Figure 2.17: Failure modes from experiment and finite element analysis
(Wang, 2006)

An alternative method to static non-linear analysis in ABAQUS program is to perform

quasi-static analysis with implicit/explicit integration methods. Sun et al. (2000)
conducted an investigation on this, and concluded that the quasi-static analysis with
implicit integration scheme is preferable for slow contact problems since the explicit
integration scheme creates high frequency “noise” and requires large number of
increments.

Natário et al. (2014) conducted a study on flange crushing in cold-formed steel sections.
Finite element analyses were performed using both static non-linear analysis and quasi-
static analysis with explicit integration techniques. They compared the finite element
analysis results with the experimental results. It was concluded that the results of quasi-
static analysis with explicit integration scheme showed good agreements in terms of the
ultimate load prediction, post-collapse load-deflection curve, and failure mechanism.

2-23
 Literature review

However, since the problem did not have any global instability, applicability of the
analysis procedure to analyse the coupled mullions’ behaviour need to be investigated.

2.4.3 Model idealization

Mullions in unitized facades have a very complex geometry. Therefore, idealizing the
geometry for the development of numerical models has to be done carefully to minimize
the inaccuracy in results. The development in Computer Aided Design (CAD) programs
offers numerous features to draft models with complex geometry. Furthermore, the
communication of these programs with other structural analysis packages such as
ABAQUS is also possible (DS, 2014b). Therefore, importing three-dimensional CAD
models directly into ABAQUS/CAE is a preferred approach.

Huang (2014) used three-dimensional solid discretization for the numerical analysis of
mullions in thermal break facade system (Figure 2.18). However, as aluminium
mullions in unitized facades are generally thin-walled (usually 1-4 mm thickness) and
their lengths are large, the use of three-dimensional solid discretization is questionable
in terms of computational effort. This is because three-dimensional discretization will
create a large number of degrees of freedom as modeling of thin-walled sections require
refined meshing in the thickness direction to get accurate results.

Three-dimensional
solid elements

Figure 2.18: Discretized ABAQUS model used by Huang (2014)

2-24
 Literature review

An efficient method as an alternative to three-dimensional solid modelling is to use

shell elements, and this technique was widely used by many researchers, especially to
model thin-walled sections (Wang, 2006, Jeyaragan, 2010, Natario et al., 2014, Kaitila,
2007, Skejic et al., 2016). Wang (2006) used shell element modelling in ABAQUS
program to investigate the behaviour of a mullion couple. The actual and the idealized
geometries of his mullion sections are shown in Figure 2.19. As the shell element
formulation offers through thickness integration points, greater accuracy with
comparatively larger element sizes are possible. Furthermore, ABAQUS element library
offers various types of shell element, and a suitable element type has to be adopted
depending on the type of analysis and the level of accuracy required. The discretized
shell finite element model used by Wang (2006) is shown in Figure 2.20.

(a) Actual mullion geometry (b) Idealized mullion geometry

Figure 2.19: Idealization of the geometry of complex mullions (Wang, 2006)

Figure 2.20: Discretized shell finite element model used by Wang (2006)

2-25
 Literature review

In summary, modelling of complex shaped mullion sections using three-dimensional

solid elements is not a preferable approach since it will result in a large number of
degrees of freedom and increased computational cost. Shell element discretization was
successfully used by many researchers in the past to model thin-walled members
(Wang, 2006, Jeyaragan, 2010, Natario et al., 2014, Kaitila, 2007, Skejic et al., 2016).
Therefore, shell elements can be used to model the mullion couples in this investigation.

2.4.4 Interaction

Male and female mullions in a mullion couple interact with each other during loading.
When a negative load is applied they may lean against each other whereas they may
slide on each other and separate during positive loading. Skejic et al. (2016) conducted a
study on a mullion couple with and without a clip. He modeled the interaction of the
mullions using ‘No separation’, and the clip using ‘Bonded’ type contacts in ANSYS.
These contact interactions restrict contacting bodies moving away from each other thus
can produce unrealistic behaviour of the mullion couple. Furthermore, Skejic et al.
(2016) did not validate their finite element models against any reliable test results, and
hardly no other literature is available on the modelling of the structural behaviour of
mullion couples. Therefore, further investigation is required.

2.5 Design of mullion couples

2.5.1 General

Currently, many standards are available for the design of aluminium structures.
Aluminium Design Manual (ADM) (AA, 2015), AS/NZS 1664 (SA, 1997), and EN
1999-1-1 (CEN, 2007) are some of them. ADM and AS/NZS 1664 provide guidelines
for Load Resistance Factor Design (LRFD) and permissible stress design. AS/NZS
1664.1 for LRFD and AS/NZS 1664.2 for permissible stress methods are technically
equivalent to the outdated Aluminium Design Manual (AA, 2010). EN 1999-1-1 (CEN,
2007) gives effective width based design method to determine the section moment
capacity of aluminium sections. In this method, effective thickness of each element of a
cross-section needs to be calculated using their slenderness, and then effective section
modulus is calculated. This is an iterative process since the neutral axis changes due to
the consideration of effective width. Thereafter, the section moment capacity can be
determined by multiplying the effective section modulus of the section and its yield

2-26
 Literature review

strength. This approach seems to be time consuming and complex to deal with complex
shaped mullion sections, and thus would not be considered in this study. This study
mainly focuses on the design guidelines provided by ADM (AA, 2015).

Failure of an aluminium structural member under flexure can occur through section
yielding with or without local element buckling, and flexural-torsional buckling.
Section moment capacity including local buckling effect, and moment capacity of the
member accounting for global member buckling effect need to be determined in order to
find the flexural capacity of a member. Design of the aluminium mullions using
Aluminium Design Manual (AA, 2015) is described in the next section.

2.5.2 Section moment capacity

The Aluminium Design Manual (AA, 2015) provides three design approaches to
calculate the flexural capacity of a section. They are: limiting stress method, weighted
average method, and direct strength method. In the limiting stress method, the allowable
stresses of all the elements in a cross-section are calculated. Then the minimum of those
stresses is used to determine the ultimate section moment capacity by multiplying that
stress with the elastic section modulus. Similarly, the allowable stresses of all the
elements are calculated in the weighted average method. But the stresses are averaged
based on the contributory cross-sectional area. The average stress is then multiplied by
the elastic section modulus of the cross-section to obtain the ultimate section moment
capacity. In both the limiting stress and weighted average stress methods, the allowable
stresses of the elements of a cross-section are calculated using the slenderness of each
element. In direct strength method, an equivalent slenderness for the entire cross-section
is calculated based on the elastic local buckling stress of the entire cross-section to
determine the allowable stress (Kim, 2003, AA, 2015). CUFSM software can be used
to determine the elastic local buckling stress of the cross-section (Li and Schafer, 2010).

Kesawan et al. (2017a) conducted experimental and theoretical investigations on

aluminium mullion sections to determine the section moment capacity. They calculated
the section moment capacity of the mullion sections using the limiting stress and direct
strength methods given in AA (2015) and the total moment capacity approach proposed
by Kim (2003), and then compared with the experimental results. They concluded that
the direct strength approach is a very convenient and accurate method to determine the
section moment capacity of the complex mullion sections. Therefore, the direct strength
2-27
 Literature review

method is considered in this investigation. The procedure used to determine the section
moment capacity of a cross-section in accordance with ADM (AA, 2015) is described
next.

Equivalent slenderness of an entire cross-section is calculated using Equation 2.13.

Then the allowable stress for the cross-section can be determined using the Section
F.5.5.5 of the ADM (AA, 2015), where the slenderness limits λ1 and λ2 define the three
local buckling regimes which are: yielding, inelastic buckling and post buckling (Figure
2.21). The allowable stresses for the three buckling regimes are given by Equations 2.14
to 2.16.

λ1 λ2

Figure 2.21: Three regimes of local buckling of elements (Kissell and Ferry, 1995)

‫ܧ‬
ߣ௘௤ ൌ ߨඨ ----------------(2.13)
‫ܨ‬௘

For λeq ≤ λ1,

‫ܯ‬௡௣
‫ܨ‬௕ ൌ ----------------(2.14)
ܵ௫௖

2-28
 Literature review

For λ1  λeq  λ2,

‫ܯ‬௡௣ ‫ܯ‬௡௣ ߨ ଶ ‫ߣ ܧ‬௘௤ െ ߣଵ

‫ܨ‬௕ ൌ െ ቆ െ ଶ ቇቆ ቇ ----------------(2.15)
ܵ௫௖ ܵ௫௖ ‫ܥ‬௣ ‫ܥ‬௣ െ ߣଵ

For λeq ≥ λ2,

݇ଶ ඥ‫ܤ‬௣ ‫ܧ‬
‫ܨ‬௕ ൌ ----------------(2.16)
ߣ௘௤

where:

B p  Fcy
O1 : O2 C p , Bp, Dp and Cp are buckling co-efficient which can be found
Dp

from Table B.4.1 of ADM (AA, 2015). Mnp is the nominal flexural strength of the
section and Sxc is the section modulus about the compression side of the x axis.

Finally, the section moment capacity of the cross-section can determined as shown by
the next equation.

‫ܯ‬௡௟௕ ൌ ‫ܨ‬௕ ܵ௫௖ ----------------(2.17)

2.5.3 Member moment capacity

To ensure that a member is safe under flexure, member moment capacity needs to be
determined. Chapter F of ADM (AA, 2015) provides guidelines to design an aluminium
member for flexure that is loaded in a plane parallel to a principal axis that passes
through the shear centre, or restrained about its longitudinal axis at the loading points
and the supports. Section F.4.2.5 is considered in this study as it covers any shapes. The
procedure to calculate the nominal flexural strength for the lateral-torsional buckling
strength Mnmb is explained next.

First, the elastic lateral-torsional buckling moment Me is determined using Equation

2.18, where y-axis is the centroidal symmetry or principal axis such that the tension
flange has a positive y coordinate and bending is about x-axis. Furthermore, the
intersection of the principal axes is the origin of the coordinate system. The coefficients
U, Ex and Cb can be calculated using Equations 2.19 to 2.21.

2-29
 Literature review

‫ܥ‬௕ ߨ ଶ ‫ܫܧ‬௬ ଶ
ͲǤͲ͵ͺ‫ܮܬ‬ଶ௕ ‫ܥ‬௪ ----------------(2.18)
‫ܯ‬௘ ൌ ቌܷ ൅ ඨܷ ൅ ൅ ቍ
‫ܮ‬ଶ௕ ‫ܫ‬௬ ‫ܫ‬௬

ܷ ൌ ‫ܥ‬ଵ ݃௢ ൅ ‫ܥ‬ଶ ߚ௫ Ȁʹ ----------------(2.19)

ͳ ----------------(2.20)
ߚ௫ ൌ ቆන ‫ ݕ‬ଷ ݀‫ ܣ‬൅ න ‫ ݔݕ‬ଶ ݀‫ܣ‬ቇ െ ʹ‫ݕ‬଴
‫ܫ‬௫ ஺ ஺

ͳʹǤͷ‫ܯ‬௠௔௫ ----------------(2.21)
‫ܥ‬௕ ൌ
ʹǤͷ‫ܯ‬௠௔௫ ൅ ͵‫ܯ‬஺ ൅ Ͷ‫ܯ‬஻ ൅ ͵‫ܯ‬஼

where:

go = the distance from shear centre to the point of application of the load. It is
positive when the load acts away from the shear centre and negative when it acts
towards the shear centre.
C1 = 0.5
C2 = 0.5
Lb = length of member
Iy = moment of inertia about y axis
Cw = warping constant
Mmax= absolute value of the maximum moment in the unbraced segment
MA = absolute value of the moment at the quarter point of the unbraced segment
MB = absolute value of the moment at the midpoint of the unbraced segment
MC = absolute value of the moment at the three-quarter point of the unbraced segment

After determining the elastic lateral-torsional buckling moment, the slenderness λ can
be calculated using Equation 2.22. The slenderness λ defines the limit state of a member
whether it is elastic or inelastic buckling. The nominal flexural strength for the limit
state of the lateral-torsional buckling of a member is calculated based on Table 2.5. The
coefficient Cc can be determined from Table B.4.2 of ADM (AA, 2015).

‫ܵܧ‬௫௖ ----------------(2.22)
ߣ ൌ ߨඨ
‫ܯ‬௘

2-30
 Literature review

Table 2.5: Member moment capacity formulae

Limit state Mnmb Slenderness limits

ߣ ߨ ଶ ‫ܵߣܧ‬௫௖
Inelastic Buckling ‫ܯ‬௡௠௕ ൌ ‫ܯ‬௡௣ ൈ ሺͳ െ ሻ൅ ߣ ൏ ‫ܥ‬௖
‫ܥ‬஼ ‫ܥ‬௖ଷ

Elastic Buckling ߨ ଶ ‫ܵܧ‬௫௖ ߣ ൒ ‫ܥ‬௖

ߣଶ

Note:
Sxc = elastic section modulus

E = elastic modulus

Mullions in the unitized facade systems are asymmetric and the load from glass is
transferred eccentrically to the mullions thus creating combined flexural and torsional
stresses in the mullions. Wang (2006) considered the glass restraint to be adequate to
provide effective lateral restraint to the compression flange of mullion couples in
unitized facade systems, and to resist torsion. His study is limited to the section moment
capacity of a mullion couple. But, there is still hardly any valid study available to verify
these assumptions, where such assumptions can sometimes lead to unconservative
design if the torsional deformations of the mullions are sufficient to initiate failure. This
is because the torsional rigidity of open mullion sections is generally low. The
provisions given in ADM (AA, 2015) to determine the member moment capacity need
to be assessed for their applicability to mullion design.

2.6 Literature review findings

The following key outcomes have been found through the accompanied literature
review:

x It is imperative to understand the material behaviour of aluminium mullions to

predict their structural response during wind actions. Aluminium shows a non-
linear stress-strain behaviour with considerable strain hardening. The use of
multi-linear material models for aluminium leads to inaccurate predictions
(Gardener and Ashraf, 2006). Therefore, good knowledge of its accurate stress-
strain characteristics is important. Experiments are needed to obtain the accurate

2-31
 Literature review

mechanical properties including yield strength, elastic modulus and stress-strain

behaviour of aluminium alloy mullions.
x Simple and full-scale tests were conducted on mullion couples in the past (Yip
and Carrick, 1995, Hui et al., 2015). Unitized facade components including
glass, gasket, silicone joint, and transom could affect the behavior and capacity
of mullion couples, thus simple tests conducted on single mullion couples are
not adequate to simulate the real behaviour of mullion couples in unitized façade
system. Therefore, full-scale tests are essential. However, the full-scale tests
conducted on mullion couples are not comprehensive to understand the
structural behaviour of mullion couples subject to wind actions in detail (Yip
and Carrick, 1995, Hui et al., 2015).
x Mullion couples with clip systems are also used in unitized façade systems to
increase their ultimate and serviceability capacities. Skejic et al. (2016)
investigated the structural behaviour and capacity of a mullion couple with a clip
system based on finite element analysis. They showed that the capacity of the
mullion couple increased when an anti-buckling clip (or separate clip parts) was
used. But making any suggestions based on their study is questionable since
their finite element model was not validated against reliable test results. This
demonstrates the need to perform full-scale tests of unitized facades made of
mullion couples with a clip.
x To date, hardly any literature is available regarding the use of hollow systems in
mullion couples to improve their structural behaviour and capacity. Therefore,
full-scale tests are required to investigate the effects of using hollow stiffeners
on the structural behaviour and capacity of mullion couples.
x Silicone sealant, which is used to fix the glass with mullions, provides effective
lateral restraint to the mullions (Yip and Carrick, 1995). However, the
effectiveness of gaskets, which are used to hold the glass panes in captive glazed
system, to provide the required lateral restraint to the mullions is still
questionable, and further investigation is needed by means of full-scale tests.
x Previous studies (Wang, 2006, Huang, 2014) showed that the numerical models
using ABAQUS finite element program can be used to investigate the structural
behaviour of mullions. Male and female mullions interact with each other during
loading. During the negative wind loading they may slide on each other and lean

2-32
 Literature review

against themselves while during the positive wind loading they may slide on
each other and separate. However, none of the past research studies simulated
this real behaviour of mullion couples. Also the developed models by past
research studies were not validated against reliable test results. Therefore,
reliable finite element models which can accurately predict the structural
response of mullion couples subject to wind actions need to be developed and
validated against test results. Finite element models of mullion couples with and
without clip systems and hollow stiffeners need to be developed.
x Many standards are available for the strength design of aluminium members
(AA, 2015, SA, 1997, CEN, 2007). EN 1999-1-1 (CEN, 2007) provides a
comparatively different approach which is difficult to adopt in our day-to-day
practice since it requires iterative calculations, and will be a cumbersome task to
design asymmetric mullions using this approach. Aluminium Design Manual
(ADM) (AA, 2015) provides guidelines for the design of aluminium sections,
particularly, symmetric sections are explicitly covered. The provisions given for
the design of members for flexure do not incorporate torsional forces. Structural
engineers design the mullions using the guidelines given in ADM (AA, 2015)
assuming that the effects of torsional forces are minimal. This may sometimes
lead to unconservative designs. Furthermore, interactions between the male and
female mullions is not covered in this manual. ADM (AA, 2015) neither
provides guidelines on required clip spacing nor provide a design approach to
determine the structural capacity of mullion couple with clip systems. Therefore,
further investigations are needed to evaluate the applicability of the currently
available design rules to predict the bending capacity of mullions with and
without clip systems and hollow stiffeners. If required, necessary modifications
to the available design rules need to be proposed to improve their accuracy.

2-33
This Page is Intentionally Left Blank
 Mechanical properties

Chapter 3

3 Mechanical properties of 6063-T6 aluminium alloy mullions

3.1 Tensile coupon tests

3.1.1 Tensile coupon details

Tensile coupon tests were conducted to determine the mechanical properties of 6063-T6
aluminium alloy mullions and to develop their material model (stress-strain curve).
Tensile coupons were cut from the flange and web elements of 19 different mullion
samples (Table 3.1) provided by G.James Glass and Aluminium. Preparation of the
tensile coupons was in accordance with AS1391 (SA, 2007) specification. Figure 3.1
shows the tensile coupon geometry and the dimensions in millimeters (mm) while
Figure 3.2 shows the typical locations of tensile coupons in the mullion samples. Table
3.1 includes the location detail for each mullion sample, i.e., web or flange.

Table 3.1: Details of mullion samples and tensile coupons

Mullion sample No. of specimens Nominal thickness (mm) Location

450-044 2 1.80 Web
450-045 2 1.80 Web
650-027 3 2.10 Web
650-028 3 2.10 Web
651-012 2 2.50 Web
651-013 2 2.50 Web
651-014 2 2.30 Web
1 2.00 Flange
651-015 2 2.30 Web
651-537 2 2.00 Web
1 2.80 Flange
651-599 1 2.50 Web
1 2.80 Flange
851-512 2 2.50 Web
546-035 2 3.00 Web
1 3.50 Flange
546-036 2 3.00 Web
570-001 2 3.00 Web
1 4.00 Flange
570-053 2 3.00 Web
579-001 2 3.00 Web
579-002 2 3.00 Web

3-1
 Mechanical properties

Table 3.1: Details of mullion samples and tensile coupons (Continued)

Mullion sample No. of specimens Nominal thickness (mm) Location

579-003 2 3.00 Web
1 3.50 Flange
579-004 2 3.00 Web
1 3.50 Flange

Figure 3.1: Tensile coupon geometry and dimensions

Tensile coupon locations

Mullion

Figure 3.2: Mullion sample after tensile coupon extraction

3.1.2 Preparation of tensile coupons

Aluminium is naturally protected by the formation of its oxide-Al2O3, product of the

reaction of bare metal with the atmospheric oxygen, which is a hard layer and highly
resistive against weathering. This layer prevents further deterioration of the bare metal
due to environmental hazards. However, aluminium is treated whenever additional
protection and/or enhanced appearance is required. Three different types of coating are
commonly used for aluminium and they are anodizing, painting and powder coating.

The aluminium samples given by G.James Glass and Aluminium were protected with
either anodizing or powder coating. The coating thicknesses were measured using the
3-2
 Mechanical properties

coating thickness measuring instrument (Figure 3.3(a)), and the maximum coating
thickness was found to be in the range of 100 microns in total considering both inner
and outer surfaces. The coatings create a smooth surface, and hence fixing the clip-on
extensometer on the coupon may not give accurate strain measurements due to potential
slip of the extensometer. Therefore, the coatings on both sides were removed using two
different methods depending on the type of coating.

Aluminium oxide (Al2O3) is an amphoteric compound, and therefore it can be etched

using an acid or a base. Therefore, the coupons with anodized finish were treated with
caustic soda (NaOH) of 0.5 mol/dm3 while powder coating was removed using paint
stripper (methylene chloride/CH2Cl2). Coupons were cleaned using soap and water after
treating with the above chemicals to remove any residuals present on the surface.
Special attention was taken while removing the coating to ensure that the coupon
surface was not damaged (particularly localized damages) as it can affect the tensile test
results. Figure 3.3(b) shows the coating thickness measurement after the removal of
coating, which is negligible.

(a) Sample with coating (b) Coupon after the removal of

coating
Figure 3.3: Coating thickness measurements in μm

3.1.3 Test details

Tensile tests were conducted using an Instron tensile testing machine of 50 kN capacity,
which was calibrated in January, 2015 (Figure 3.4). The thickness and width of each
tensile coupon were measured using a digital Vernier caliper of 0.01 mm accuracy
(Figure 3.5). As the actual thickness of the specimens can vary, the average width and
thickness obtained from the measurements at various locations were used in the
calculations. The coupon was mounted within the grips and the extensometer was

3-3
 Mechanical properties

attached to it. An appropriate clip was used depending on the thickness of the coupon.
The load was applied by means of constant cross-head movement of 1 mm/min
according to the specification given in AS1391 (SA, 2007). Figure 3.6 shows the tensile
test set-up used in this investigation.

Figure 3.4: Calibration details of Instron testing machine

Figure 3.5: Thickness measurement of a tensile coupon using digital Vernier

caliper

6063-T6 aluminium
tensile coupon
Upper and lower grips

Clip-on extensometer

Figure 3.6: Tensile test set-up using Instron testing machine

3-4
 Mechanical properties

3.2 Four-point bending tests

3.2.1 General

Modulus of elasticity of aluminium alloys in tension and compression differs by about

2% (AA, 2015, Kissel and Ferry, 1995), where the compressive modulus of elasticity is
higher than the tensile modulus of elasticity (Compressive and tensile moduli of
elasticity in accordance with SA (1997a,b) are 70000 and 68600 MPa, respectively).
Modulus of elasticity that governs the flexural deflection of a member under bending is
taken as the average of the tensile and compressive moduli of elasticity, i.e. about 1%
less (69300 MPa) than the value of the compressive modulus of elasticity (SA, 1997a,b,
AA, 2015, Kissel and Ferry, 1995). The average value of the tensile modulus of
elasticity of aluminium alloys using tensile coupon test is generally found to be lower
than the values specified in the standards (Kim, 2003, Su et al., 2014, Huang and
Young, 2014). The average modulus of elasticity from the conducted tensile coupon
tests was also found to be slightly lower (refer Section 3.3.1) than the value specified in
AS 1664 (SA, 1997a,b). Therefore, it was planned to conduct four-point bending tests.

Suansuwan and Swain (2001) and Haines et al. (1996) used a four-point bending test
method in order to determine the flexural rigidity of some metallic alloys and other
types of materials. Therefore, six four-point bending tests were conducted on
rectangular aluminium specimens taken from four different aluminium mullions in
order to determine the flexural modulus of elasticity of the 6063-T6 aluminium alloy.
The details of the mullion samples and the bending test specimens are given in Table
3.2. The length of the bending test specimens used was 450 mm. Figure 3.7 shows the
geometry of a bending test specimen, and Figure 3.8 shows a mullion sample after the
extraction of a bending test specimen.

Table 3.2: Details of mullion samples and bending test specimens

Mullion sample No. of

Thickness (mm) Width (mm)
specimens
2.03 19.39
650-027 2
1.93 19.85
1.97 19.71
650-028 2
1.94 19.91
651-012 1 2.00 15.86
651-013 1 2.01 15.83

3-5
 Mechanical properties

A B C D

450 mm

Figure 3.7: Bending test specimen

Bending test specimen

Figure 3.8: Mullion sample after cutting a bending test specimen

3.2.2 Test details

A schematic diagram of a beam under four-point bending, and the variation of the
bending moment due to the externally applied loads (W) are shown in Figure 3.9. The
loads are applied at Points A and D, and the supports are at Points B and C, thus the
region BC will be subjected to a uniform bending moment (Figure 3.9). The geometric
dimensions (width and thickness) of each rectangular specimen were measured using a
digital Vernier caliper, and the locations of the mid-span and Points A, B, C and D were
marked on the specimens as shown in Figure 3.7. The measured dimensions of the test
specimens are given in Table 3.2.

Four-point bending tests were conducted using the bending test frame available in the
structural laboratory of Queensland University of Technology. This apparatus has a
digital dial gauge that can measure the deflection of a beam at an accuracy of 0.01 mm.
In this test, specimens are loaded at Points A and D with equal weights (W), which are
hung on the specimen using a clip (Figure 3.11). Since the clip also has considerable
weight, the weight W at a point is the total weight of the clip and the weights that are to

3-6
 Mechanical properties

be hung. Therefore, digital electronic balance, which can measure a mass with an
accuracy of 0.01 g, was used to calibrate the weights (Figure 3.10). These calibrated
weights were used to load the specimen at loading Points A and D in order to simulate
the four-point bending as shown in Figure 3.9.

Figure 3.9: Schematic diagram of a beam under four-point bending

Weights
Digital electronic
balance
Clip

Figure 3.10: Calibration of weights

The specimen was placed in the frame appropriately and the dial gauge reading was
reset to zero. The calibrated weights were then hung on the specimen at Points A and D,
and the upward mid-span deflection reading was recorded. Thereafter, the specimen was

3-7
 Mechanical properties

loaded with an approximate load increment of 100 g to a maximum of 1000 g at each

loading point, and the deflection at mid-span was recorded for each load increment.
Figures 3.11(a) and (b) show a specimen before and after loading, respectively.

Dial gauge
Specimen

Bending test apparatus

(a) Before loading

Clip Dial gauge

Specimen

Weights

(b) After loading

Figure 3.11: Four-point bending test

3-8
 Mechanical properties

3.3 Test results and discussion

3.3.1 Tensile coupon tests

The load and strain data recorded in the Instron Bluehill software application was used
in the analysis. A spreadsheet was prepared for the evaluation of mechanical properties
such as Young’s modulus (E), proof strength (f0.2) based on 0.2% offset, proof strength
(f0.1) based on 0.1% offset, ultimate strength and strain (fmax and Hmax, respectively),
strain at rupture (Hu), and strain hardening exponent (n) considering different cases as
specified in EN 1999-1-1 (CEN, 2007). The stress-strain curves obtained from the
tensile tests for the 650-027 and 650-028 mullions are shown in Figure 3.12 whereas the
results of other mullions are given in Appendix A.2. Table 3.3 gives the summary of
averaged test results. Refer Appendix A.1 and A.4 in relation to tensile coupon labels
and summary of test results for each tested coupon, respectively.

As seen in Table 3.3, the various parameters obtained from the tests (E, f0.1, f0.2, fmax,
Hmax and Hu) vary considerably depending on the mullion due to possible variations in
composition between different batches (Mondolfo, 1976, Kissel and Ferry, 1995). For
example, E, f0.2 and fmax varied from 60795 to 68476 MPa, 181 to 240 MPa, 211 to 263
MPa, respectively. Therefore, in research studies involving numerical modeling, the
measured mechanical properties for each mullion should be used instead of mean
values. The minimum values specified in AS/NZS 1664 (SA, 1997) can be used for
design purposes.

The average Young’s modulus (tensile modulus) measured was around 65158 MPa with
a coefficient of variation (COV) of 0.033. However, the value of modulus of elasticity
specified in AS/NZS 1664 (SA, 1997) will be used for the study since the values
obtained from test results are generally found to be lower than the values specified in
design standards. Typical value of compressive modulus of elasticity of 6063-T6
aluminium alloy specified in AS/NZS 1664 (SA, 1997) is 70000 MPa, where the tensile
modulus of elasticity can be taken as 2% (1400 MPa) less than the compressive
modulus. The effect of modulus of elasticity on strength can be related to compression
members (Kissel and Ferry, 1995), and therefore the typical value of compressive
modulus of elasticity (i.e. 70000 MPa) specified in AS/NZS 1664 (SA, 1997) will be
used in the study. It is because, this study focuses on finding the ultimate moment

3-9
 Mechanical properties

capacity of mullion couples, where local buckling and yielding of compression elements
can be expected.

Different methods given in EN 1999-1-1 (CEN, 2007) to determine the strain hardening
exponent (n) in Ramberg-Osgood model have been evaluated using the tensile test
results in Table 3.3. The tested samples (mullions) are used in the aluminium frame of
unitized facade systems. The mullions transfer the wind load acting on the glass panels
to the main structure. Their flexural capacity and the structural behaviour are important
in design. Mullions are subjected to bending stresses under applied wind actions. Since
they can fail by section yielding with or without inelastic local and distortional
buckling, stress-strain curves in the elastic and plastic regions are important in their
analysis and design.

The model given in EN 1999-1-1 (CEN, 2007) for the elastic region (Equations 2.1 and
2.4) was evaluated and compared with the test results. The strain hardening exponent
nElastic was calculated from Equation 2.4 using the measured test parameters (f0.1 and f0.2)
in Table 3.3 and then used with Equation 2.1 to predict the stress-strain curve. It is clear
from Figures 3.13(a) to (e) that the model suggested for elastic region does not give
consistent and accurate results within the strain hardening region. Therefore, the model
based on the provisions given for the plastic region in EN 1999-1-1 (CEN, 2007) is
considered (Equations 2.1 and 2.8), which also includes the elastic region. In this case,
the strain hardening exponent nPlastic was calculated from Equation 2.8 using the
measured test parameters (f0.2, fmax and Ho,max=Hmax-fmax/E) in Table 3.3 and then used
with Equation 2.1 to predict the stress-strain curve. The stress-strain curves predicted
for 650-027 and 650-028 mullions are shown in Figures 3.14(a) and (b), respectively,
which agreed well with the test results. Similar agreement can be seen in Figures A.2
(a.1) to (a.17) for other mullions (Appendix A.3). Furthermore, the measured average
mechanical properties of different mullions are given in Table 3.3 while the measured
mechanical properties of each tested coupons are given in Table A.2 (Appendix A.4).

3-10
 Mechanical properties

250

Stress (MPa) 200

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (mm/mm)
650-027-A 650-027-B 650-027-1

(a)

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (mm/mm)
650-028-A 650-028-4 650-028-5

(b)
Figure 3.12: Measured stress versus strain curves of 650-027 and 650-028 mullions

3-11
 Mechanical properties

300

250
Stress (MPa)

200

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (mm/mm)
Test results Ramberg-Osgood model

(a) 450-044-A

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (mm/mm)
Test results Ramberg-Osgood model

(b) 650-027-1
Figure 3.13: EN 1999-1-1 (CEN, 2007) model for elastic region and test results

3-12
 Mechanical properties

300

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (mm/mm)
Test results Ramberg-Osgood model

(c) 546-035-A

300

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
Strain (mm/mm)
Test results Ramberg-Osgood model

(d) 570-001-A
Figure 3.13: EN 1999-1-1 (CEN, 2007) model for elastic region and test results

3-13
 Mechanical properties

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
Strain (mm/mm)
Test results Ramberg-Osgood model

(e) 851-512-A
Figure 3.13: EN 1999-1-1 (CEN, 2007) model for elastic region and test results

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (mm/mm)
Test results Ramberg-Osgood model
(a) 650-027-A
Figure 3.14: EN 1999-1-1 (CEN, 2007) model for plastic region and test results

3-14
 Mechanical properties

250

200
Stress (MPa)
150

100

50

0
0 0.02 0.04 0.06 0.08 0.1
Strain (mm/mm)
Test results Ramberg-Osgood model

(b) 650-028-A
Figure 3.14: EN 1999-1-1 (CEN, 2007) model for plastic region and test results
Table 3.3: Summary of test results (average values)

E f0.2 f0.1 fmax

Sample Label
(MPa) (MPa) (MPa) (MPa)
Hmax Hu nElastic nPlastic

650-027
A,B 65079 210 208 228 0.0724 0.1001 53 43
1 65824 192 189 211 0.0626 0.0935 47 37
650-028
A 64013 201 199 218 0.0625 0.0854 67 42
4,5 64367 216 213 234 0.0729 0.1020 42 44
450-044 A,B 60795 238 233 259 0.0931 0.1243 32 45
450-045 A,B 60921 217 213 233 0.0758 0.1038 38 48
651-013 A,B 63863 211 208 231 0.0723 0.1070 53 39
651-014 A,B,C 65308 233 230 256 0.0895 0.1150 47 39
651-012 A,B 66414 204 201 234 0.0929 0.1240 46 28
579-003 A,B,C 66250 218 214 236 0.0762 0.1053 37 43
579-004 A,B,C 66899 209 205 228 0.0730 0.1084 40 41
570-001 A,B,C 65874 236 230 257 0.0839 0.1219 27 42
546-035 A,B,C 63565 231 223 255 0.0824 0.1069 21 37
546-036 A,B 66147 240 234 263 0.0910 0.1272 28 41
651-599 A,C 68476 219 215 238 0.0790 0.1085 34 44

3-15
 Mechanical properties

Table 3.3: Summary of test results (average values)

E f0.2 f0.1 fmax

Sample Label
(MPa) (MPa) (MPa) (MPa)
Hmax Hu nElastic nPlastic

579-001 A,B 64771 181 178 219 0.0974 0.1221 44 20

579-002 A,B 63396 213 210 231 0.0740 0.1079 47 43
570-053 A,B 68064 215 211 234 0.0715 0.1044 38 42
651-015 A,B 65398 203 200 222 0.0660 0.0902 50 37
651-537 A,B,C 66066 221 216 244 0.0874 0.1107 35 37
851-512 A,B 66643 212 208 236 0.0896 0.1304 33 35
Minimum 60795 181 178 211 0.0625 0.0854 21 20
Maximum 68476 240 234 263 0.0974 0.1304 67 48
Mean 65158 N/A N/A N/A 0.0799 0.1105 40 40
COV 0.033 N/A N/A N/A 0.122 0.115 0.267 0.15

Ultimate strains at fracture as obtained from tensile tests (εu,test) were compared with the
predicted values (εu,pred) using the model given in EN 1999-1-1 (CEN, 2007) (Equation
2.10). The ultimate fracture strain values predicted by EN 1999-1-1 (CEN, 2007) are
considerably higher than the experimental results with significant scatter. In this study,
a new predictive model to determine the ultimate fracture strain (εu) value is proposed.
This new model (Equation 3.1) gives more accurate predictions than those found from
Equation 2.10 (see Figure 3.15) for 6063-T6 aluminium alloy. The coefficients of the
proposed model were obtained using the tensile test results by means of least-squares
regression method. However, it should be noted that the applicability of the proposed
model for the ultimate fracture strain prediction of the other types of aluminium alloy
cannot be concluded as this model is based solely on the tensile test results of 6063-T6
aluminium alloy.

ߝ௨ ൌ ͲǤͲͷʹ ൅ ͲǤͳͳሺ݂଴Ǥଶ ΤͶͲͲሻ ----------------- (3.1)

The strain at ultimate tensile stress εmax is one of the important parameters in
determining the strain hardening exponent for the plastic analysis case (Equation 2.8).
However, this value is not generally available in the aluminium design standards.
Therefore, a predictive equation proposed by Su et al. (2014) was used to determine the
ultimate strain (εmax,pred) corresponding to ultimate tensile stress (fmax) and then
compared with the values obtained from the test (εmax,test). Figure 3.16 shows that the

3-16
 Mechanical properties

predicted values of ultimate strain using the equation proposed by Su et al. (2014)
(Equation 2.12) gives a better prediction.

Figure 3.15: Ultimate fracture strain predictions

Figure 3.16: Ultimate strain predictions

3-17
 Mechanical properties

A new model for the prediction of 0.1% proof strength (f0.1) based on 0.2% proof
strength (f0.2) is proposed (Equation 3.2). The proposed model gives better predictions as
shown in Figure 3.17. The coefficients were calibrated using the tensile test results by
means of least-squares regression method. However, similar to the proposed ultimate
fracture strain prediction model, the applicability of the proposed model for the
evaluation of 0.1% proof strength to the other types of aluminium alloy cannot be
concluded as this model is based solely on the tensile test results of 6063-T6 aluminium
alloy.

݂଴Ǥଵ ൌ ݂଴Ǥଶ െ ͲǤʹͺඥ݂଴Ǥଶ ---------------- (3.2)

Figure 3.17: Predictions for proof strength by means of 0.1% offset

Improved predictive equations for ultimate fracture strain (εu) and proof strength by
means of 0.1% offset method (f0.1) are developed in this study since they can be used
with other predictive models given in EN 1999-1-1 (CEN, 2007), such as Equations 2.9
and 2.4, respectively. The values of ultimate strength (fmax) and 0.2% proof strength
(f0.2) can be generally found in design standards and literature. These values can be used
to find the strain (εmax) corresponding to ultimate strength (fmax) using Equation 2.12.
Then, Equation 2.1 can be used to predict the required stress-strain curves of 6063-T6
aluminium alloy samples tested in this study, where the strain hardening exponent n for
the plastic deformation analysis can be calculated using Equation 2.8.

3-18
 Mechanical properties

3.3.2 Four-point bending tests

The load and the corresponding mid-span deflection recorded for each specimen were
used to derive the M/I versus 1/R relationship, where the “M”, “I” and “R” are the mid-
span moment, second moment area of the cross-section of the specimen and the radius
of curvature of the specimen under the applied load. The derivation of the relationship
between M/I and 1/R is given in Appendix A.5. The test data of M/I and 1/R obtained
for each specimen were plotted in order to obtain the flexural modulus of elasticity of
the specimens, and are shown in Figures 3.18(a) to (f). These figures reveal that the M/I
versus 1/R curves show a linear relationship with less scatter. Table 3.4 gives a
summary of the test results, where the average flexural modulus of elasticity of the
6063-T6 aluminium alloy is 69703 MPa. Flexural modulus of elasticity of the 6063-T6
aluminium alloy based on AS 1664 (SA, 1997a,b) is 69300 MPa. Therefore, the results
from the four-point bending tests verified the value of the flexural modulus of elasticity
specified in the standard. Hence, the value of the flexural modulus of elasticity for the
6063-T6 aluminium alloy based on AS 1664 (SA, 1997a,b) can be used in the finite
element analysis.

80
y = 69304x
70
R² = 0.9998
60
M/I (N/mm3)

50

40

30

20

10

0
0.0000 0.0002 0.0004 0.0006 0.0008 0.0010 0.0012
1/R (mm-1)
(a) Specimen 1 (650-027 mullion)

Figure 3.18: M/I versus 1/R relationship of various specimens

3-19
 Mechanical properties

90
80 y = 69668x
R² = 0.9998
70
60
M/I (N/mm3)

50
40
30
20
10
0
0.0000 0.0002 0.0004 0.0006 0.0008 0.0010 0.0012
1/R (mm-1)
(b) Specimen 2 (650-028 mullion)

90
y = 69793x
80
R² = 1
70
60
M/I (N/mm3)

50
40
30
20
10
0
0.0000 0.0002 0.0004 0.0006 0.0008 0.0010 0.0012
1/R (mm-1)

(c) Specimen 3 (650-027 mullion)

Figure 3.18: M/I versus 1/R relationship of various specimens

3-20
 Mechanical properties

80

70 y = 69695x
R² = 0.9999
60
M/I (N/mm3)

50

40

30

20

10

0
0.0000 0.0002 0.0004 0.0006 0.0008 0.0010 0.0012
1/R (mm-1)

(d) Specimen 4 (650-028 mullion)

100
y = 70168x
90
R² = 0.9999
80
M/I (N/mm3)

70
60
50
40
30
20
10
0
0.0000 0.0002 0.0004 0.0006 0.0008 0.0010 0.0012 0.0014
1/R (mm-1)

(e) Specimen 5 (651-012 mullion)

Figure 3.18: M/I versus 1/R relationship of various specimens

3-21
 Mechanical properties

90
y = 69552x
80 R² = 1
70
60
M/I (N/mm3)

50
40
30
20
10
0
0.0000 0.0002 0.0004 0.0006 0.0008 0.0010 0.0012 0.0014
1/R (mm-1)

(f) Specimen 6 (651-013 mullion)

Figure 3.18: M/I versus 1/R relationship of various specimens

Table 3.4: Flexural modulus of elasticity of 6063-T6 aluminium alloy

Sample Average flexural modulus

Mullion
No of elasticity (MPa)
1 650-027 69304
2 650-028 69668
3 650-027 69829
4 650-028 69699
5 651-012 69552
6 651-013 70168
Average 69703

3.4 Conclusion

Tensile tests were conducted on 6063-T6 aluminium tensile coupons taken from various
aluminium mullion sections of different shapes and sizes. The test results were then
compared with the predictive equations given in the currently available design standards
and literature. The proof strength by means of 0.2% offset (f0.2), proof strength by
means of 0.1% offset (f0.1) and ultimate strength (fmax) values obtained from tensile tests
have significant variations among the samples taken from different aluminium mullion
sections due to possible variations in composition between different batches (Mondolfo,

3-22
 Mechanical properties

1976, Kissel and Ferry, 1995). However, the yield and ultimate strengths of 6063-T6
alloy were higher for all samples than the minimum values specified in AS/NZS 1664
(SA, 1997), which are 172 and 207 MPa, respectively.

The average tensile modulus of elasticity of aluminium obtained from tensile coupon
tests was 65158 MPa. This value is about 5% less than the value given in AS/NZS 1664
(SA, 1997a,b). The value of flexural modulus of elasticity based on AS/NZS 1664 (SA,
1997a,b) is about 69300 MPa, which is 1% higher than the tensile modulus of elasticity.
In order to verify the value of the flexural modulus of elasticity based on the standard,
four-point bending tests were performed. The average value of the flexural modulus of
elasticity obtained from the tests was 69703 MPa, which is slightly higher than the
value based on AS/NZS 1664 (SA, 1997a,b). Therefore, the value of flexural modulus
of elasticity (69300 MPa) based on AS/NZS 1664 (SA, 1997a,b) can be used in finite
element simulations to analyse the mullions under flexural action.

Ramberg-Osgood model with the strain-hardening exponent (n) determined based on

the model given for the plastic range analysis case given in EN 1999-1-1 (CEN, 2007)
was found to be the most suitable model to define the stress versus strain characteristics
of the 6063-T6 aluminium alloy mullions. Predictions from this model matched well
with the test results of a range of mullion sections. Therefore, this model is
recommended to be used in simulating the stress versus strain behaviour of the
aluminium alloy mullions in finite element analyses.

A new ultimate fracture strain (εu) model (Equation 3.1) was proposed in this study. Its
predictions were more accurate than that of CEN (2007) model prediction (Figure 3.15).
The model proposed by Su et al. (2014) can be used to predict the strain at ultimate
tensile stress (εmax). A new equation to predict the 0.1% proof strength (f0.1) based on the
0.2% proof strength was also proposed. This new model accurately predicted the 0.1%
proof strength. However, it should be noted that the testing was conducted on tensile
coupons obtained from 6063-T6 aluminium alloy mullions. Therefore, the applicability
of the proposed models for other types of aluminium alloy cannot be concluded based
on this study.

3-23
This Page is Intentionally Left Blank
 Experimental study

Chapter 4

4 EXPERIMENTAL STUDY

4.1 General

This research study investigates the structural behaviour and capacity of aluminium
mullion couples that are being used in unitized facade panels of buildings. To obtain
realistic structural performance data of such mullion couples under wind loading an
experimental study was performed. Full-scale tests were conducted in accordance with
AS 2047 (SA, 2014) and AS/NZS 4284 (SA, 2008), where facade systems of 4000 mm
x 2900 mm (length x width) made by assembling two captive glazed unitized facade
panels were subjected to incremental uniform air pressure loading until failure. The
tested facade systems comprised three different types of mullion couple: Mullion couple
without clip; Mullion couple with a clip; Mullion couple with a clip and a hollow
stiffener. Tests were conducted for mullion couples under both positive and negative
wind actions. The applied wind pressures and the displacements at various locations of
the facade assembly were measured to evaluate the structural behaviour and capacity of
the mullion couples, and to validate the developed finite element models. This chapter
presents the details of the full-scale tests and the evaluation of the test results.

4.2 Previous studies

In the past, some experimental studies were conducted on symmetric and asymmetric
aluminium mullions in order to investigate their structural behaviour and capacity under
wind actions (Wang, 2006, Huang, 2014, Zyla, 2010, Yip and Carrick, 1995, Hui et al.,
2015). Wang (2006) investigated the moment capacity of an E type mullion couple. His
study focused only on the section moment capacity of that mullion couple, where he
used a four point bending test set-up to simulate uniform bending. The mullions in
unitized facade systems are subject to uniform loading, but this loading acts on them at
a location away from their shear centre, thus torsional stresses are present in the
mullions. Therefore, Wang’s (2006) test set-up cannot be used to simulate the real
behaviour of mullions. Similarly, Zyla (2010) also conducted four point bending tests
on a captive glazed facade system comprising two unitized facade panels. Even though
the test was conducted on an assembled full-scale panel, the four point loading set-up

4-1
 Experimental study

used is not suitable to investigate the real behaviour of mullion couples as it does not
represent the real wind load acting on the facade panels.

Yip and Carrick (1995) conducted full-scale tests on a mullion couple used in
structurally glazed facades. They simulated the lateral restraint given by the glass to the
mullions by using aluminium strips. The end restraint given by transom was also
simulated in their experimental set-up. This study gave a good understanding about the
mullion couple’s behaviour under both positive and negative wind actions. However,
the tests were conducted only on a structurally glazed facade system. Although they
almost simulated the uniformly distributed loading on the mullion couple, their
investigation was limited to serviceability failure criteria for such loading case (refer
Section 2 for more details). They performed a few tests to determine the ultimate
capacity of the mullion couple, but the load was applied at third points which leads to
similar arguments as given for the tests conducted by Zyla (2010).

Hui et al. (2015) used three different test set-ups to investigate the structural
performance of a mullion couple used in structurally glazed facades. The mullion
couple included a pothook (or clip) to prevent the opening of mullion couple during
loading. In their test set-ups, the single beam with eight point trapezoidal loading set-up
does not simulate the lateral restraint provided by the glass (refer Section 2 for more
details). Full-scale tests were also conducted where loading was applied by means of air
pressure and sand bags. But the air pressure tests were not conducted until failure (only
up to 5 kPa), and the tests using sand bags were conducted only under positive wind
action. More details about this investigation can be found in Section 2.

It is evident from previous studies that the structural behaviour of mullion couples that
are being used in unitized facade systems is more complex, and therefore full-scale tests
are necessary to investigate their structural behaviour and capacity under wind actions.
In addition, a uniformly distributed load should also be applied to simulate the real
behaviour of mullion couples subject to wind action. Since mullion couples with and
without clip systems may behave differently due to possible influence of clip systems
on the relative movements of the male and female mullions, it is important to
investigate the facade systems with and without clips. Furthermore, none of the
investigations focused on mullions couples with a hollow stiffener. Therefore, it is also
necessary to conduct tests to determine the behaviour of hollow mullion couples. The

4-2
 Experimental study

different cases considered in this investigation under both positive and negative wind
loadings are:
x Mullion couple without a clip or a hollow stiffener
x Mullion couple with a clip
x Mullion couple with a hollow stiffener

4.3 Full-scale wind pressure tests

4.3.1 Test panel details

Unitized facade systems are made of aluminium frame infilled glass/granite/ aluminium
panels. Mullions are the vertical members of the aluminium frame that transfer the wind
load acting on the facades to the main structural system. Unitized facade panels are
factory fabricated and assembled at site. To enable easier installation at site, the vertical
members in unitized facade systems comprise of two individual sections, called as
mullion couple or male and female mullions (see Figure 4.1). These mullions are
custom extruded, and are of asymmetric open profiles. Generally the design capacity of
a mullion couple is found by adding the individual design capacities of male and female
mullions. However, because of the possible interaction between the male and female
mullions during loading, mullion couple could have a higher structural capacity, which
is investigated in this study.

Mullion couple

Panel 1 Panel 2

(a) Front view (exterior face of a facade)

Figure 4.1: Typical mullion couple in a captive glazed facade assembly

4-3
 Experimental study

Male
mullion
Mullion
couple
Female
mullion

Connecting
together

Panel 1 Panel 2

Front side Front side

(b) Before assembling (Section A-A) (c) After assembling (Section A-A)
Figure 4.1: Typical mullion couple in a captive glazed facade assembly

Other components in a unitized facade system are transom, glass, and gasket (Figure
4.2). Furthermore, clip systems are used in some instances to increase the serviceability
and ultimate capacities (Figure 4.2(c)). The presence of all these components in a
unitized facade system influences the structural behaviour and capacity of the mullion
couple. Their influence makes the behaviour of mullion couples more complex under
positive and negative wind actions. Therefore, it is important to simulate their effects in
the tests, and hence the test set-up was designed to include all these components.

(a) Typical mullion-transom connection

Figure 4.2: Components of a captive glazed unitized facade assembly

4-4
 Experimental study

Glass
Gaskets

Mullion

(b) Typical gasket connection

Seal strip

Clip

Mullions

Glass
Gaskets

(c) Typical clip assembly

Figure 4.2: Components of a captive glazed unitized facade assembly

Many different shapes of mullions are used in unitized facade systems based on
designers’ requirements. 650-027 and 650-028 mullion sections which are used in the
captive glazed system of G.James Glass and Aluminium were used in this study. These
types of mullions are commonly used in facade systems. Details of these mullions are
given in Table 4.1. Furthermore, tensile coupon tests of the test mullions revealed that
the yield and the ultimate strengths of the test mullion sections varied about 10% (refer
Appendix A).

650-027 male mullion has a leg at the bottom that interlocks into the 650-028 female
mullion when assembled (Figure 4.3). Structural behaviour of the mullion couple can be
influenced by their interlocking feature. However, this type of feature (interlocking leg)

4-5
 Experimental study

is not present in most of the other mullion sections used in G.James Glass and
Aluminium. This study aims to extend the investigation to the other types of mullion
couples after the validation of the developed finite element models of the tested mullion
couples. Therefore, this leg was removed prior to testing as shown in Figure 4.3, so the
mullions resembled all other commonly used sections in unitized facades.

Male mullion Female mullion

Part removed from

male mullion

Figure 4.3: Interlocking part removed from male mullion

In total, seven tests were conducted on three types of mullion couple under positive and
negative wind actions. The test details are given in Table 4.1. In Tests 1 and 2, a
mullion couple without any clip system was investigated, where the effect of interaction
between the male and female mullions on their structural behaviour and capacity was
assessed under both positive and negative wind actions. The effect of using a clip at
mid-height on the structural behaviour and capacity of the same type of mullion couple
was then investigated in Tests 3, 4a, 4b and 4c. Finally in Test 5, a mullion couple,
which comprised of a male mullion with a hollow stiffener and an open female mullion,
was investigated in order to evaluate its structural behaviour and capacity. As discussed,
650-027 and 650-028 mullions were used in all the tests. This enabled better
comparisons between mullion couples with and without clips and mullion couple with a
hollow stiffener. Furthermore, this is also advantageous in terms of cost and time since
the same sub-head and sub-sill could be used in all the tests. Changing the sub-head and
sub-sill is a time consuming task since the whole panel has to be removed to perform

4-6
 Experimental study

this task. The details of the transom, sub-head and sub-sill used in this investigation are
shown in Figure 4.9.

Table 4.1: Full-scale test details

Type of
Test Mullion couple Description
loading

Test 1 Mullion couple without Negative

clip
650-027 – Male
650-028 – Female
Test 2 Positive

Mullion couple with a

Test 3 clip at mid-height Negative
650-027 – Male
650-028 – Female
Note: In Test 3, the
Test 4a, interlocking part
shown in Figure 4.3 Positive
4b and 4c
was not removed.

Mullion couple with a

clip and a hollow
stiffener
Test 5 650-027 – Male Negative
650-028 – Female

4.3.2 Test set-up

Since facade panels are made of male and female mullions, it is necessary to have at
least two panels in the test set-up to investigate the real behaviour of the mullion couple
(see Figures 4.1 and 4.4). To include a male mullion of one panel and a female mullion
4-7
 Experimental study

of another panel form the mullion couple, a single panel is not adequate to simulate the
mullion couple’s behaviour. Therefore, two unitized facade panels including transoms
were considered for testing to represent the continuous assembly of unitized facade
panels (Figure 4.4).

The test facility available at G.James Glass and Aluminium was used. The test frame
has opening dimensions of 4063 mm x 3166 mm (height x width). The dimensions of
the test panel were decided based on the dimensions of the test rig. Therefore, the sizes
of a single facade panel were selected as 4000 mm x 1450 mm (height x width). This
dimension (4000 mm) will in fact help to overcome the problem related to laboratory
limit on pressure loading faced by Hui et al. (2015) as longer mullions can be expected
to fail at smaller pressures. Furthermore, the mullion spans used in unitized facades are
generally less than 4000 mm. Therefore, 4000 mm span can be considered as the worst
case. Hence, a decision on the structural behaviour of shorter mullions can be
confidently made later using the validated finite element models. The dimensions of the
test panels and test rig are shown in Figure 4.5.

Typical mullion couple

to be investigated

Panel 1 Panel 2

Transoms (horizontal members)

Figure 4.4: Continuous assembly of unitized facade panels

4-8
 Experimental study

Mullions Transoms

Panel 1

Panel 2

(a) Pre-assembled facade panels

4063 mm

(b) Test rig

Figure 4.5: Installation of test panels

4-9
 Experimental study

Selection of side mullions

It is not possible to use mullion couples for the side supports of the panels as there is no
continuity of the glass panels. If we use a single male mullion on one side and a female
mullion on the other side, it may create asymmetric behaviour at the ends of the panels.
If female mullions are used as side supports, the deflection of the middle mullion couple
will be less than the side mullions as the flexural rigidity of the mullion couple is higher
than that of two female mullions together. Therefore, male mullions were used as the
side supports.

Restraint to torsional rotation of the side mullions

The single male mullion has an asymmetric open profile. It will easily undergo torsional
rotation (about the longitudinal axis of the mullion) since the wind load transferred to it
from the glass is acting away from its shear centre (Figure 4.6(a)). This may lead to a
premature failure of the side mullions. However, the focus of this study is to determine
the structural behaviour and capacity of the mullion couples, and not the behaviour of
the single side mullions. Therefore, it is important to simulate the failure in the middle
mullion couple. In order to achieve this, the torsional rotation of the side mullions needs
to be prevented.

Any mechanism introduced to resist the torsional rotation should not affect the Y-axis
displacement of the side mullions (refer Section 4.4 for more details about the Y-axis
displacement). For this purpose, roller sets that allow the Y-axis translation of the side
mullions were used to prevent the torsional rotation. The schematic diagram of the
roller mechanism used is shown in Figure 4.6(b). Furthermore, since the span of the
ଵ ଶ
mullions is relatively high, it was decided to have two roller supports at L and L
ଷ ଷ

from the bottom, where L is the length of the mullions. This will ensure that the failure
does not occur in the side mullions. Installation of roller sets and the details of the
rollers and roller guides in the test assembly are shown in Figure 4.7.

4-10
 Experimental study

Load

X Shear centre

Torsional rotation
of the mullion

(a) Typical mullion rotation due to eccentric loading

Torsional rotation

Side mullion

Restraining
torsion given
by rollers Load

Direction of unrestrained horizontal translation of

the mullion

(b) Schematic diagram of the roller mechanism

Figure 4.6: Torsional rotation of the side mullion and preventive mechanism

4-11
 Experimental study

Rollers used to restrain

torsional rotation

Side mullion

(a) Installation of rollers at 1/3rd and 2/3rd lengths of the side mullions

Roller set-up used to simulate

Guide to the rollers to
the torsional restraint about
allow horizontal movement
longitudinal axis of the side
of the mullion
mullions
Side mullion

(b) Roller restraints assembled into the guide

Figure 4.7: Restraint to torsional rotation of the side mullions

4-12
 Experimental study

Installation of sub-head and sub-sill

In order to assemble the facade panels on the test rig, sub-head and sub-sill were fixed
to the top and bottom of the test rig (Figure 4.8). SS14g type 17 TEKS screws were
used to fix the sub-head and sub-sill to the test rig. After the sub-head and sub-sill were
fixed to the test rig, the edges of the sub-head and sub-sill and the screw ports were
sealed using silicone sealant as shown in the same figure. Thereafter the panels were
assembled where the mullions and transoms transfer the load to the sub-head and sub-
sill through bearing. The typical details of sub-head and sub-sill fixing, and the
transoms to sub-head and sub-sill assembly are shown in Figure 4.9.

Screw ports sealed

using sealant after
fixing the sub-sill to
the frame

Sub-sill

Bottom of the test rig

Edges fully sealed
using sealant

Figure 4.8: Sub-sill fixed to the test rig using TEKS screws

Sub-head (675-028)

Bearing locations – simple

support
Simply supported

Transom (650-003)
span

(a) Head level transom and sub-head assembly detail

Figure 4.9: Transoms to sub-head and sub-sill assembly details

4-13
 Experimental study

Simply supported
Transom (650-004)

span

Bearing locations – simple

support Sub-sill (675-021)

(b) Sill level transom and sub-sill assemble detail

Figure 4.9: Transoms to sub-head and sub-sill assembly details

Selection of infilling panel

One of the functions of the glass panes in facade panels is to transfer the wind load
acting on them to mullions and transoms. This study was aimed at conducting multiple
full-scale tests to investigate the structural behaviour and ultimate bending capacity of
mullion couples. It is not economically feasible to use glass panes in the test specimen
since there is a possibility for glass breakage when mullions fail. The glass breakage
could damage the test facility and as a result increase the cost per test. Also, more
significantly, safety precautions need to be undertaken in order to ensure the safety of
people involved in testing. Glass and aluminium have similar Young’s modulus, which
is about 70000 MPa, as specified in AS 1288 and AS/NZS 1664 (SA, 2006 and SA,
1997). Since the deflection of a glass pane is governed by its Young’s modulus, and
other properties such as density and strength of the glass do not affect the capacity of
mullion couple, aluminium sheets of the same thickness (6 mm) were used instead of
glass panes.

Prevention of air leakage

The load is applied on the test panels by means of uniform air pressure. The assembled
test set-up has significant openings around its perimeter, particularly between the side
mullions of the assembled panel and the test rig as shown in Figure 4.10(a). This will

4-14
 Experimental study

result in increased air leakage, thus requiring a higher rate of air supply. Since the air
supply may have limitations depending on the blower equipment capacity, there may be
difficulties in reaching the ultimate failure pressure of the mullion couples if the air
leakage is significant. Therefore, 1 mm thick EPDM (Ethylene Propylene Diene
Monomer) flexible seal was used to seal the opening between the side mullions and the
test rig (Figure 4.10(b)) with sufficient length of fold to ensure that the movements of
the side mullions are not restricted by the seal (SA, 1996a). In addition, the edges of the
seal were attached to the test rig and the flange of the side mullions using flat strip,
where SS10g pan screws were used at 450 mm spacing to connect them. The possible
air leakages through the interface of aluminium sheet to mullion, aluminium sheet to
transom and transom to sub-head and sub-sill connections were prevented by the
gaskets and seal strips (Figures 4.2 and 4.9). Furthermore, the ends of the sub-head and
sub-sill which intersect the frame, and the screw ports used to fix the sub-head and sub-
sill to the frame were also sealed using silicone sealant as shown in Figure 4.8.

Face of the test rig

Side mullion

Opening between the side

mullion and test rig
Outside of the
chamber

(a) Opening between side mullion and test rig

Figure 4.10: Sealing of openings between test rig and side mullions

4-15
 Experimental study

1 mm thick EPDM seal

25 x 3mm flat fixed using pan screws

to clamp the seal edges with the
aluminium angle and side mullion

Edges sealed using sealant

Aluminium angle fixed to

the test rig

(b) Sealing of openings using EPDM flexible seal

Figure 4.10: Sealing of openings between test rig and side mullions

Installation of panels
Unitized facade panels are generally assembled such that a gap of about 3 mm is present
between the male and female mullions. This practice is to allow movements in the
aluminium frame caused by the thermal expansion due to temperature variations and
building deflections. Otherwise, additional stresses could develop in the mullions
leading to their premature failure. This is a common practice in the industry. Therefore,
the panels were assembled in the test rig with a gap of 3 mm between the male and
female mullions. Furthermore, silicone spray was applied to the seal strips of the
mullions and transoms in order to eliminate friction that could affect the relative
movement between the male and female mullions (Figure 4.11).

4-16
 Experimental study

Figure 4.11: Application of silicone spray to the seal strips

Installation of LVDTs
Displacements at various locations were measured using LVDTs. In total, 10 LVDTs
that are numbered from 1 to 10 were installed as shown in Figure 4.12(b). These
LVDTs have an accuracy of r0.1 mm, according to the requirement of AS 4420.2 (SA,
1996b). As our main focus was the centre mullion couple, 8 LVDTs were installed on
the mullion couple in order to measure the X-axis and Y-axis displacements (refer
Section 4.4 for more details about the X-axis and Y-axis deflections). LVDTs 4, 6, 7
and 9 measured the Y-axis displacements of the centre mullions at their ends, LVDTs 2
and 3 measured the mid-span Y-axis displacements, and LVDTs 5 and 8 measured the
X-axis displacements of the male and female mullions at their mid-span (Figure
4.12(a)).

The mullions transfer the load to the sub-head and sub-sill through bearing and thus
there is a possibility for the sub-head and sub-sill to deflect in the Y-axis direction at
mullions’ bearing locations. The mullions can also move with the sub-head and sub-sill.
Therefore, the average of the top and bottom end deflections of the mullions were
deducted from the measured mid-span deflections to determine the actual mid-span
deflections. Side mullions were selected in order to simulate the symmetrical behaviour
that exists in a continuous facade assembly. In order to evaluate and compare the
deformation characteristics of the individual side mullions and the centre mullion
4-17
 Experimental study

couple it is necessary to measure the mid-span deflections of the side mullions.

Therefore, LVDTs 1 and 10 were installed to measure the mid-span deflections of the
side mullions.

5
8

2 3

(a) LVDTs installed at the mid-span of the mullion couple

(b) LVDT layout for the whole panel

Figure 4.12: Deflection measurements using LVDTs

4-18
 Experimental study

Calibration of LVDTs
It is necessary to calibrate all the LVDTs before commencing the test to ensure that the
deflection measurements are accurate. Standard 3, 10 and 15 mm calibration blocks
were used to calibrate the LVDT readings. Calibration of an LVDT reading was carried
out by placing a calibration block of a particular dimension and then comparing the
LVDT reading with the dimension of the block. Figure 4.13 shows the calibration of a
LVDT using a 10 mm standard block.

Standard 10 mm
calibration block

LVDT

Figure 4.13: Calibration of a LVDT using a standard block

Preliminary tests conducted on the panels made of 650-027/650-028 mullion couple

revealed that the mullions could undergo significant torsional rotation under negative
wind loading, whereas the couple opens up under positive wind loading. Deflection
measurements at mid-span of the mullion couple could be difficult as the mullions
rotate in addition to the Y-axis displacements. To measure the actual X-axis
displacements, a parallelogram shaped system was created, where all the vertices were
hinged so that the opposite sides can move relative to each other but being always
parallel. This ensured the measurement of true X-axis displacement of the mullions
(Figure 4.14(a)). Furthermore, silicone spray was applied to the hinges and the contact
interfaces of the parallelogram shaped system and side face of the mullions to minimize
the effect of friction. The LVDT to measure the X-axis displacement of the mullion was
fixed and positioned in the parallelogram shaped system (see Figure 4.14(b)) such that it
would measure the change in the distance between the parallel sides of the
parallelogram. But the LVDT cannot record any reading if the mullion moves away
from the parallelogram shaped system.

4-19
 Experimental study

Side on which mullion push

When
mullions
push

Parallel sides Parallel sides

(a) Hinged parallelogram shaped system

Hinges
Interface between the side of
the system and mullion

LVDT positioned on the

Parallelogram shaped side of the system
hinged system

(b) Parallelogram shaped system fixed in the test set-up

Figure 4.14: X-axis displacement measurements of the mullions

4-20
 Experimental study

The final test set-up of the unitized facade assembly is shown in Figure 4.15. In the test
set-up the exterior face of the facade assembly is inside the chamber while the interior
face of the facade assembly is outside the chamber (Figures 4.15(a) and (b)). In testing,
it is a common practice to describe the facade assembly by looking at their exterior face
(face exposed to environment), and hence the “Left” and “Right” terms used for panels
as shown in Figure 4.15 are based on the view from inside the chamber.

LVDTs to measure the horizontal

displacements at mid-span

Left panel Right panel

Inside of the chamber (i.e., exterior

face of the facade system)

(a) View from inside the chamber

Figure 4.15: Final test set-up

4-21
 Experimental study

Side mullion

LVDTs at the top

Transom - head level

6 mm thick aluminium sheets

Side mullion

LVDTs to measure the X-axis

displacements at mid-span
Mullion couple

Outside of the chamber (i.e., interior

face of the facade system)

Right panel Left panel

Transoms - sill level

LVDTs at the bottom

(b) View from outside the chamber

Figure 4.15: Final test set-up

4-22
 Experimental study

4.3.3 Loading and test procedures

Once the test set-up has been completed, pre-loading was applied to ensure that the test
set-up was ready to load the mullion couple up to failure. AS 4420.1 (SA, 1996a)
specifies to increase the pressure until it reaches 50% of the design wind pressure
during pre-loading. In this study, -100 Pa was used as the preliminary test pressure for
the negative wind load whereas +600 Pa was used for the positive wind load test to
ensure that the mullion couple did not fail or undergo large deformations during pre-
loading. The preliminary test pressures were applied and maintained for 1 minute in
accordance with AS 4420.1 (SA, 1996a).

First test was conducted on panels made of a mullion couple without clip or hollow
stiffener. This test was conducted under negative wind pressure. The test panel was
loaded in steps of 100 Pa, where each load increment was kept for 1 min in accordance
with AS 4420.2 (SA, 1996b). Since the buckling of mullions can occur suddenly, the
load increment was reduced to 50 Pa per step closer to the failure load of the mullion
couple. The second test was conducted on the same type of mullion couple but under
positive wind pressure. In this test, the load was applied at an increment rate of 100 Pa
per step, where each load increment was kept for 1 min in accordance with AS 4420.2
(SA, 1996b). Similar to the negative pressure test, the load increment was reduced to 50
Pa closer to the failure of the mullion couple.

It is necessary to terminate the loading prior to catastrophic failure of the mullions and
subsequent damage to the test facility. Also, safety is an important concern to protect
the personnel who are present in the testing area. Therefore, when large rotations and
significant deformations were observed, the tests were immediately terminated, and the
differential pressure was returned to zero. Thereafter, the failed specimen was
investigated to further understand the mullions’ failure behaviour. All other tests were
also conducted following a similar procedure as discussed above.

4.4 Test results and observations

In total, seven full-scale tests (Table 4.1) were conducted under both positive and
negative wind pressures. The observed structural response of the test specimens, the
measured pressure versus displacement characteristics and the observed failure modes
of the mullion couples are discussed next. It should be noted that the deflection

4-23
 Experimental study

behaviour of the mullions is described in terms of “X-axis” and “Y-axis” displacements

of the mullions. The “Y-axis displacement” of mullions refers to the deflection of
mullions in the direction perpendicular to the plane of the panel/ in the direction of the
load, whereas the “X-axis displacement” is the movement perpendicular to the direction
of the load (Figure 4.16). Furthermore, the rotational displacement of the mullions is
referred in the clockwise and anti-clock wise directions as shown in Figure 4.16. The Y-
axis displacements were measured at the middle of the mullions’ flanges (i.e. Points A
and B in Figure 4.16) while the X-axis displacements were measured at Points C and D
as shown in Figure 4.16.

Y-axis displacement

Clockwise rotation

Point C Point D

X-axis displacement

Male Female
mullion mullion
Aluminium
sheet

Panel 1 (Left) Panel 2 (Right)

Point A Point B

Inside of the chamber

Figure 4.16: Description of the direction of translation and rotation of the mullions

4-24
 Experimental study

4.4.1 Test 1

A facade panel made of 650-027 male and 650-028 female mullions without a clip or a
hollow stiffener was tested under negative wind pressure loading in this test. During the
early stage of the loading (0 to 100 Pa), the male and female mullions started to rotate
inward to lean on each other Figure 4.17(a). The gap between the male and female
mullions reduced and finally the unrestrained compression flanges of the mullions were
in contact with each other. Thereafter, both mullions started to rotate together in the
clockwise direction without showing any separation between them as the pressure was
increased. This deformation of the mullion couple took place until the pressure reached
about 750 Pa, where the pressure increment was kept at a rate of 50 Pa per step. When
the next increment was applied the female mullion started to separate from the male
mullion, but both mullions continued to rotate in the same direction (Figure 4.17(b)).
Finally, after showing a significant separation and rotation, both mullions failed by
buckling (see Figures 4.17(c) to (e)), where the applied pressure at failure was about
850 Pa. The test was terminated at this point and the applied differential pressure was
brought back to zero. Figures 4.17(a) to (f) show the deformation behaviour of the
mullion couple during the loading and after the failure. Figure 4.17(f) shows a
significant permanent deformation in the mullion couple with a noticeable gap between
the mullions at the mid-span after being unloaded. The measured negative pressure
versus X-axis and Y-axis displacement curves of the male and female mullions are
shown in Figure 4.18.

Since the test was terminated immediately at the failure point, displacements at the
failure pressure were not recorded. For this purpose, displacements scaled from the
captured pictures and videos during the test were used. This was the same for Test 2.
However, an improved method to record the displacements at failure was developed and
used in the remaining tests.

4-25
 Experimental study

Female mullion Male mullion Unrestrained compression

flanges

(a) Mullions rotated to lean on each other (b) Clockwise rotation of the mullion
(~ 0 to 100 Pa) couple (~ 100 to 750 Pa)

Gap increased

(c) Separation of the female mullion from (d) Mullion couple just before failure
the male mullion (~ 800 Pa) (850 Pa)

Figure 4.17: Mullion couple behaviour in Test 1

4-26
 Experimental study

Female mullion

Permanent gap
Male mullion between mullions

(e) Mullion couple at failure (f) Mullion couple after failure

(unloading)

Figure 4.17: Mullion couple behaviour in Test 1

-1000
-900
-800
-700
Pressure (Pa)

When pressure was increased from 800 to

-600
850 Pa displacement of the mullion couple
-500 increased suddenly leading to failure
-400
-300 Female mullion
-200 Male mullion
-100
0
0 -10 -20 -30 -40
Displacement (mm)

(a) Pressure versus Y-axis displacement

Figure 4.18: Applied negative wind pressure versus displacement curves of the
mullion couple in Test 1

4-27
 Experimental study

-1000
-900
-800
-700
Pressure (Pa)

-600
-500
-400
-300 Female mullion
-200 Male mullion
-100
0
0 20 40 60 80
Displacement (mm)

(b) Pressure versus X-axis displacement

Figure 4.18: Applied negative wind pressure versus displacement curves of the
mullion couple in Test 1

4.4.2 Test 2

A facade panel made of 650-027 and 650-028 mullion couple without a clip or a hollow
stiffener was tested under positive wind pressure loading. During the initial stage of
loading, the mullion couple deflected in the Y-axis direction without showing any
noticeable opening between the male and female mullions. This behaviour continued up
to an applied pressure of about 1500 Pa. Thereafter, the mullion couple started to open
up as shown in Figure 4.19(b). The couple continued to open, and deflect in the Y-axis
direction (see Figure 4.19(c)) as the pressure was increased to 2600 Pa. When the next
increment was applied the mullion couple failed by buckling, and therefore the test was
immediately terminated and the applied pressure was brought back to zero. The failure
pressure was about 2700 Pa. The mullion couple showed a significant permanent
deformation when unloaded (Figures 4.19(e) and (f)). The measured positive wind
pressure versus displacement curves of the mullion couple are shown in Figure 4.20.

4-28
 Experimental study

Rotation

Female mullion Male mullion Aluminium beam used to fix the

LVDTs

(a) At zero pressure (b) Anti-clockwise rotation of the mullion

couple (~ 0 to 1500 Pa)

Sliding of mullion
on aluminium sheet
Unrestrained tension flanges

(c) Opening of the mullion couple (d) Mullion couple at failure

(~ 1500 Pa) (2700 Pa)

Figure 4.19: Mullion couple behaviour in Test 2

4-29
 Experimental study

Male mullion Female mullion

(e) Mullion couple after failure (f) Compression flange of the failed
(unloading) female mullion

Figure 4.19: Mullion couple behaviour in Test 2

3000

2500

When pressure was increased from 2400 to

2000
Pressure (Pa)

2700 Pa displacement of the mullion couple

increased suddenly leading to failure
1500
Female mullion
1000
Male mullion

500

0
0 50 100 150 200
Displacement (mm)

(a) Pressure versus Y-axis displacement

Figure 4.20: Applied positive wind pressure versus displacement curves of the
mullion couple in Test 2

4-30
 Experimental study

3000

2500

2000
Pressure (Pa)

1500

1000 Female mullion

Male mullion
500

0
20 0 -20 -40 -60 -80 -100
Displacement (mm)

(b) Pressure versus X-axis displacement

Figure 4.20: Applied positive wind pressure versus displacement curves of the
mullion couple in Test 2

4.4.3 Test 3

A facade panel made of 650-027 and 650-028 mullion couple with a clip at its mid-span
was tested under negative wind pressure loading. The length of the clip used was 100
mm (refer Figure 4.2(c) or Table 4.1 for a schematic diagram of the assembly of the
mullion couple with a clip).

During the early stage of the loading, the mullion couple started to rotate together in the
clockwise direction. The gap between the male and female mullions remained
unchanged (Figures 4.21(a) and (b)). When the applied pressure reached 600 Pa, the
male and female mullions became in contact with each other at their compression
flanges (Figure 4.21(c)). Thereafter, with the increasing load both mullions started to
rotate together in the clockwise direction without showing any separation between
them. This deformation of the mullion couple took place until the pressure reached
about 1100 Pa, where the pressure increment was kept at a rate of 25 Pa per step (Figure
4.21(d)). When the next load increment was applied the female mullion started to
separate from the male mullion, and subsequently both mullions failed in buckling. The
test was terminated at this point and the applied differential pressure was brought back
to zero. Figures 4.21(a) to (f) show the deformation behaviour of the mullion couple

4-31
 Experimental study

during the loading and after the failure. The measured negative pressure versus X-axis
and Y-axis displacement curves of the male and female mullions are shown in Figure
4.22.

Gap closed

Female mullion Male mullion Aluminium beam used to fix the

LVDTs

(a) At zero pressure (b) Clockwise rotation of the mullion

couple (~ 0 to 600 Pa)

Unrestrained compression
flanges with clip

Initial gap between the male

Separation of the female
and female mullions closed
mullion from the male mullion

(c) Mullions rotate to lean on each other (d) Separation of the female mullion from
(~ 600 Pa) the male mullion (~ 1100 Pa)

Figure 4.21: Mullion couple behaviour in Test 3

4-32
 Experimental study

Significant separation of the female

mullion at 450 mm from the mid-span

450 mm

Female mullion Male mullion

No significant separation at mid-span

(e) Mullion couple at failure (1125 Pa) (f) Mullion couple after failure (unloading)

Figure 4.21: Mullion couple behaviour in Test 3

-1400

-1200

-1000
Pressure (Pa)

-800

-600
Female mullion
-400
Male mullion
-200

0
0 -10 -20 -30 -40 -50 -60 -70
Displacement (mm)

(a) Pressure versus Y-axis displacement

Figure 4.22: Applied negative wind pressure versus displacement curves of the
mullion couple in Test 3

4-33
 Experimental study

-1400

-1200

-1000
Pressure (Pa)

-800

-600
Female mullion
-400
Male mullion
-200

0
0 20 40 60 80
Displacement (mm)
(b) Pressure versus X-axis displacement
Figure 4.22: Applied negative wind pressure versus displacement curves of the
mullion couple in Test 3

4.4.4 Test 4a

A facade panel similar to that in Test 3 was tested under positive wind pressure loading.
The difference between Tests 2 and 4a is that a mullion couple without a clip was tested
in Test 2 whereas a mullion couple with a 100 mm clip at mid-span was tested in Test
4a (refer Figure 4.2 or Table 4.1 for a schematic diagram of the assembly of the mullion
couple with a clip).

Initially, with the increasing load, the male and female mullions rotated together in the
anti-clockwise direction, where the gap between the male and female mullions remained
the same (Figures 4.23(a) and (b)). This behaviour continued up to 1800 Pa. But
thereafter, with the increasing load the gap between the mullions increased as shown in
Figures 4.23(b) and (c). However, the gap increment at mid-span was limited since the
clip was activated (Figures 4.23 (c) and (d)). The mullion couple continued to rotate
together until the pressure reached about 2600 Pa, where the pressure increment was
kept at a rate of 50 Pa. Thereafter, the mullion couple started to slide on the aluminium
sheet as it continued to rotate in the anti-clockwise direction as shown in Figure 4.23(d).
Finally, the aluminium sheet (right side in Figure 4.23(e)) popped out of the male
mullion, when the pressure was about 2650 Pa. The test was terminated at this point and

4-34
 Experimental study

the applied differential pressure was brought back to zero. Figures 4.23(a) to (f) show
the deformation behaviour of the mullion couple during the loading and after the failure.
The measured positive pressure versus X-axis and Y-axis displacement curves of the
male and female mullions are shown in Figure 4.24.

Rotation

Aluminium beam used to fix the

Female mullion Male mullion LVDTs

(a) At zero pressure (b) Anti-clockwise rotation of the mullion

couple (~ 0 to 1800 Pa)

Gap increased Sliding of mullion on Gap further

the aluminium sheet increased

Unrestrained tension flanges with clip Gap did not further increase

(c) Opening of the mullion couple (d) Closer to 2650 Pa

(~ 1800 Pa)

Figure 4.23: Mullion couple behaviour in Test 4a

4-35
 Experimental study

Mullion slided significantly

on the aluminium sheet

Aluminium sheet popped out

from the male mullion Female mullion Male mullion

(e) Mullion couple at failure (2650 Pa) (f) Mullion couple after failure (unloading)

Figure 4.23: Mullion couple behaviour in Test 4a

3000

2500

2000
Pressure (Pa)

1500

Female mullion
1000
Male mullion
500

0
0 10 20 30 40 50 60
Displacement (mm)

(a) Pressure versus Y-axis displacement

Figure 4.24: Applied positive wind pressure versus displacement curves of the
mullion couple in Test 4a

4-36
 Experimental study

3000

2500

2000
Pressure (Pa)

1500

Female mullion
1000
Male mullion

500

0
0 -5 -10 -15 -20 -25 -30
Displacement (mm)
(b) Pressure versus X-axis displacement
Figure 4.24: Applied positive wind pressure versus displacement curves of the
mullion couple in Test 4a

4.4.5 Test 4b

A test was conducted similar to Test 4a, but with an addition of restraints to prevent the
sliding of the mullions as shown in Figure 4.25(a). This test was performed in order to
investigate the effect of sliding restraint on the capacity of the mullions.

During the early stage of the loading, the mullion couple started to rotate in the anti-
clockwise direction. The gap between the male and female mullions remained
unchanged (Figures 4.25(a) and (b)). When the applied pressure reached about 1800 Pa,
the gap between the mullions started to increase. As the load was increased, the mullion
couple continued to rotate together, and the gap between the mullions further increased
(Figures 4.25(c) and (d)). But the gap at mid-span did not increase beyond a certain
limit (Figures 4.26(b), 4.25(c) and 4.25(d)). When the pressure reached about 3400 Pa,
the test was terminated due to the concern on the possibility of catastrophic failure of
the panel and the consequence damage to the LVDTs. It is clear from Figure 4.25(d)
that the mullions did not show any visible sliding on the aluminium sheet. Figures
4.25(a) to (e) show the deformation behaviour of the mullion couple during the loading.
The measured positive pressure versus X-axis and Y-axis displacement curves of the

4-37
 Experimental study

male and female mullions are shown in Figure 4.26. Furthermore, Figure 4.26(b)
reveals that the mullion couple is close to its failure pressure since the slope of the
pressure-displacement curve is reducing rapidly.

Sliding restraints provided at 660 mm

c/c throughout the span for both the
Female mullion Male mullion male and female mullions

(a) At zero pressure

Rotation

Gap increased

Aluminium beam used to fix the

LVDTs Unrestrained tension flanges with clip

(b) Anti-clockwise rotation of the mullion (c) Opening of the mullion couple
couple (~ 0 to 1800 Pa) (~ 1800 Pa)

Figure 4.25: Mullion couple behaviour in Test 4b

4-38
 Experimental study

No visible sliding Gap further

of mullion increased

Female mullion Male mullion

Gap did not further increase

(d) Mullion couple at 3400 Pa (e) Mullion couple after failure (unloading)

Figure 4.25: Mullion couple behaviour in Test 4b

4000

3500

3000
Pressure (Pa)

2500

2000

1500 Female mullion

1000 Male mullion

500

0
0.0 20.0 40.0 60.0 80.0
Displacement (mm)
(a) Pressure versus Y-axis displacement
Figure 4.26: Applied positive wind pressure versus displacement curves of the
mullion couple in Test 4b

4-39
 Experimental study

4000

3500

3000
Pressure (Pa)

2500

2000

1500 Female mullion

1000 Male mullion

500

0
0 -5 -10 -15 -20 -25 -30
Displacement (mm)
(b) Pressure versus X-axis displacement
Figure 4.26: Applied positive wind pressure versus displacement curves of the
mullion couple in Test 4b

4.4.6 Test 4c

The method adopted in Test 4b to prevent the sliding of the mullions seemed to be
overly restraining their torsional rotation of the mullions based on the visual
observations and test results. Also, this type of method is not suitable for use in practical
applications. Therefore, silicone sealant was used instead of the method used in Test 4b
to prevent the sliding of the mullions by connecting the edge of the aluminium sheets
and the leg of the mullions (see Figure 4.27(f)). This represents the use of setting blocks
in real applications where the mullions are very slender. The length and the thickness of
the sealant bite used was 100 mm, and the thickness of the aluminium sheet,
respectively.

During the early stage of the loading, the mullion couple started to rotate in the anti-
clockwise direction. The gap between the male and female mullions remained
unchanged (Figures 4.27(a) and (b)). When the applied pressure reached about 1500 Pa,
the gap between the mullions started to increase. As the load was increased, the mullion
couple continued to rotate together, and the gap between the mullions increased. But the
gap at the mid-span did not increase beyond a certain limit (Figure 4.27(c)). When the
pressure reached about 3500 Pa, the mullion couple failed in buckling. It is evident from

4-40
 Experimental study

Figure 4.27(c) that the mullions did not show any visible sliding on the aluminium sheet
just before failure. Figures 4.27(a) to (e) show the deformation behaviour of the mullion
couple during the loading, while Figure 4.27(f) shows the silicone sealant bite that was
found to have torn after the failure.

Rotation

Unrestrained tension
flanges with clip
Female mullion Male mullion

(a) At zero pressure (b) Clockwise rotation of the mullion

couple (~ 0 to 1500 Pa)

Rotation

Gap increased

No visible sliding Gap did not Buckling failure

of the mullion further increase at mid-span

(c) Just before failure (~ 3500 Pa) (d) After failure – view from outside the
chamber
Figure 4.27: Mullion couple behaviour in Test 4c

4-41
 Experimental study

100 mm

Silicone sealant bite

Male mullion Female mullion

(e) After failure - view from inside the (f) Silicone sealant bite
chamber

Figure 4.27: Mullion couple behaviour in Test 4c

4.4.7 Test 5

A facade panel made of 650-027 and 650-028 mullion couple, with a clip at its mid-
span, and a 650-042 hollow stiffener (refer Table 4.1) was tested under negative wind
pressure loading. The length of the clip used was 100 mm.

During the first few increments (from 0 to 200 Pa), the male and the female mullions
rotated to lean on each other (Figure 4.28(b)). Thereafter, the mullions did not show any
noticeable rotation as the pressure was increased to about 1950 Pa. When the next load
increment was applied the aluminium sheet popped out of the transom at the head level
(Figure 4.28(e)). The test was terminated at this point and the applied differential
pressure was brought back to zero. Figures 4.28(a) to (c) show the deformation
behaviour of the mullion couple during the loading. The measured negative pressure
versus X-axis and Y-axis displacement curves of the male and female mullions are
shown in Figure 4.29.

4-42
 Experimental study

Unrestrained compression
flanges with clip

Female mullion Male mullion

(a) At zero pressure (b) Clockwise rotation of the mullion

couple

(c) Just before failure (d) Mullion couple after unloading

Figure 4.28: Mullion couple behaviour in Test 5

4-43
 Experimental study

Aluminium sheet
popped out of the
transom here

(e) Failure at the aluminium sheet to transom connection

Figure 4.28: Mullion couple behaviour in Test 5

-2500

-2000
Pressure (Pa)

-1500

-1000

Female mullion
-500
Male mullion

0
0 -10 -20 -30 -40 -50
Displacement (mm)

(a) Pressure versus Y-axis displacement

Figure 4.29: Applied negative wind pressure versus displacement curves of the
mullion couple in Test 5

4-44
 Experimental study

-2500

-2000
Pressure (Pa)
-1500

-1000

Female mullion
-500
Male mullion

0
-10 -5 0 5 10
Displacement (mm)
(b) Pressure versus X-axis displacement
Figure 4.29: Applied negative wind pressure versus displacement curves of the
mullion couple in Test 5

In the previous sections, the results of the conducted full scale tests were given in detail.
The observed structural behaviour of the mullion couples and their pressure versus
displacement characteristics were also given and discussed. A summary of all the
conducted full scale tests is given in Table 4.2.

Table 4.2: Summary of full-scale test results

Test Type of loading Failure pressure Remark

(Pa)

Test 1 Negative 850 Mullion couple failed

Test 2 Positive 2700 Mullion couple failed
Test 3 Negative 1125 Mullion couple failed
Test 4a Positive 2650 Aluminium sheet popped out
Test 4b Positive 3400 Test was terminated early
Test 4c Positive 3500 Mullion couple failed
Test 5 Negative 1950 Aluminium sheet popped out

4-45
 Experimental study

4.5 Discussion of test results

The applied wind pressure and the displacements of mullions were measured in all the
full-scale tests, and are presented in Figures 4.17 to 4.29 and Table 4.2. The observed
structural behavioural characteristics of the mullions during loading were also
presented. This section explains the reasons for the displayed behaviour of the mullion
couples in the full-scale tests conducted.

4.5.1 Test 1: Mullion couple without clip or hollow stiffener under negative wind
pressure loading

In this test, the load acting on the aluminium sheets was transferred to the mullions
through the gaskets (Figure 4.30). Therefore, the loading point on the mullions was the
connection between the mullions and gaskets. However, the shear centre of the mullions
is located away from the line of action of the loading for both mullions. This resulted in
the torsional rotation of the individual mullions. During the early stage of loading, the
male and female mullions rotated towards each other due to the torsional action (Figure
4.31(b)). Once both mullions became in contact at their compression flanges, they
started to rotate together in the clockwise direction (Figures 4.18(b) and 4.31(c)). This
means the male mullion pushed the female mullion over as there was an imbalance in
the horizontal forces applied by the male and female mullions on each other.

The male and female mullions have approximately similar torsional rigidity (see Table
4.3), but the eccentricity from the shear centre to the line of action of the load is higher
for the male mullion. The larger eccentricity in the male mullion could have resulted in
higher torsional forces in the male mullion compared to that in the female mullion. Thus
the male mullion pushed the female mullion and rotated in the clockwise direction. The
female mullion began to separate from the male mullion after certain loading since the
lateral torsional buckling of the female mullion was initiated (Figure 4.31(d)). Finally,
this separation of the female mullion together with the rotation of both mullions led to
the failure of the mullion couple in buckling. This failure mode of the mullion couple is
lateral-torsional buckling. It should be noted that the applied negative pressure
developed compressive stresses on the unrestrained flanges of the mullion couple.
Therefore, the aluminium sheets did not provide any lateral restraint to prevent the
lateral-torsional buckling of mullions. Figures 4.18(a) to (f) show the observed

4-46
 Experimental study

structural behaviour and the failure mode of the mullion couple tested under negative
wind pressure loading.

Furthermore, in these systems it was expected that the leaning action of the male and
female mullions will let the mullions act as a hollow section, thus achieving a higher
capacity. However, this did not happen in the test, instead the female mullion separated
from the male mullion after a certain load. This separation of the female mullion
induced the premature failure of the mullion couple. But the capacity of the mullion
couple can be still higher than the addition of the capacities of the individual male and
female mullions.

10.3 6.5

Line of action of the load

from glass

Figure 4.30: Typical load transfer from glass to mullions

Table 4.3: Typical section properties of the mullions used

Mullion A (mm2) J (mm4) Ixx (mm4) eL (mm)

Male - 650-027 585 867 16.756 x 105 10.3

Female - 650-028 558 1069 16.603 x 105 6.5
A – area
J – torsional constant
Ixx – second moment of area about the x axis
eL – eccentricity of loading position from the shear centre

4-47
 Experimental study

Compression
Male Female zone
mullion mullion

Tension
zone

(a) Initial state (b) Rotating to lean on each other

(c) Rotating together (d) Separation of female mullion

Figure 4.31: Schematic diagram of the behavior of mullions in Test 1

4.5.2 Test 2: Mullion couple without a clip or a hollow stiffener under positive
wind pressure loading

When a positive wind pressure loading was applied on the mullion couple without a
clip, the mullion couple did not open up, instead rotated in the anti-clockwise direction
in the early stage of loading (Figure 4.32(b)). It is expected that the male mullion must
rotate in the anti-clockwise direction while the female mullion must rotate in the
clockwise direction because of the direction of torsional forces developed in the
respective mullions. However, as described during the early stage of loading, female
4-48
 Experimental study

mullion also rotated in the anti-clockwise direction. The reason for this response is the
higher torsional forces in the male mullion and the frictional resistance given by the seal
strip (Figure 4.32(b)). The male mullion has a higher loading eccentricity compared to
the female mullion as shown in Figure 4.30. There is not much difference between the
torsional rigidity of the male and female mullions (Table 4.3). Therefore, the male
mullion started to rotate in the anti-clockwise direction. The movement of the male
mullion pulled the female mullion to rotate with it. The frictional resistance given by the
seal strip facilitated this, which is justifiable from the point that the contact between the
male and female mullions is only through the seals, as otherwise female mullion should
have rotated in the other direction.

However, once the seal friction was exceeded, the mullion couple opened up, where
although the male mullion continued to rotate in the same direction, the female mullion
rotated in the opposite (clockwise) direction (Figures 4.32(c)). This is because torsional
force on the individual mullions causes rotation in those respective directions under
positive pressure loading. As the pressure was increased the male mullion continued to
rotate and failed in buckling (Figure 4.32(d)). The female mullion also rotated in the
same direction as the male mullion (anti-clockwise) and failed.

Tension
Male Female zone
mullion mullion

Seal strip

Gaskets Aluminium sheets Compression

zone

(a) Initial state (b) Rotating together

Figure 4.32: Schematic diagram of the behavior of mullions in Test 2

4-49
 Experimental study

(c) Gap opening (d) Separation of male mullion

Figure 4.32: Schematic diagram of the behavior of mullions in Test 2

The failure pressure of the mullion couple under positive wind pressure was about three
times higher than that under negative wind pressure. During the negative wind pressure
loading, the compression flange of the mullions is unrestrained whereas the presence of
aluminium sheets provide lateral restraint to the compression flanges of the mullions
under positive wind pressure loading (Figure 4.32). This lateral restraint led the
mullions to achieve a greater capacity under positive wind pressure loading in
comparison to that achieved under negative wind pressure loading. However, the
mullion couple did not reach its yield moment capacity (9.55 kNm which is equivalent
to 3.53 kPa pressure). Here it should be noted that the section moment capacity of the
mullion couple could be higher than the yield moment capacity since these
commercially used mullions are generally non-slender sections. The lateral restraint for
the mullions was available only to their inner compression flange as shown in Figure
4.32. This along with the increasing torsional force and the consequent torsional
deformations of the mullions caused the mullions to fail prematurely by lateral-torsional
buckling. This is also a significant finding in this research study, because it is generally
assumed that the mullion couple could achieve its yield moment capacity since they are
not prone to elastic local buckling while its compression flange is laterally restrained by
the glass and the torsional effects are minimal.

4-50
 Experimental study

4.5.3 Test 3: Mullion couple with a clip at mid-span under negative wind
pressure loading

In this test, a mullion couple made of 650-027 and 650-028 mullions was tested similar
to Test 1, but a 100 mm long clip was installed at the mid-span of the mullions as shown
in Figure 4.33(a). Similar to Test 1, the negative wind pressure load is transferred to the
mullions at a point away from their shear centre as shown in Figure 4.30. As a result,
torsional forces are induced in the mullions.

During the early stage of loading, the male and female mullions rotated together in the
clockwise direction (Figure 4.33(b)). This is because of the imbalance in the torsional
forces that are induced on the mullions together with the possible frictional resistance of
the clips. That is, the frictional resistance between the clips resisted the individual
rotation of the mullions that was observed in Test 1 (see Figure 4.31(b)) thus both
mullions rotated together. The reason for the imbalance in the induced torsional force is
due to the difference in the distance between the shear centre and the load position of
the male and female mullions, where the male mullion has a higher eccentricity thus
having higher developed torsional moment than that on the female mullion. Once the
frictional resistance given by the clip is exceeded, the male and female mullions rotated
to lean on each other (Figure 4.33(c)). Thereafter, with the increasing load, both
mullions rotated together in the clockwise direction as the male mullion pushed the
female mullion (Figure 4.33(d)). Here the male mullion pushed the female mullion due
to the imbalance in the forces exerted by them on each other.

The female mullion then began to separate from the male mullion after certain loading
since the lateral torsional buckling of the female mullion was initiated. However, the
separation away from the mid-span was more significant than the separation at the mid-
span region (Figures 4.33(e) and (f)). This is because the presence of the 100 mm long
clip limited the separation of the mullions at the mid-span region. That is, when the
female mullion started to separate from the male mullion, the clip is activated, and any
further separation of the female mullion is resisted (Figure 4.33(d)). Finally, both
mullions failed by lateral-torsional buckling (Figure 4.33(f)). It should be noted that the
applied negative pressure developed compressive stresses on the unrestrained flanges of
the mullion couple, thus the aluminium sheets did not provide any lateral restraint to

4-51
 Experimental study

prevent the lateral-torsional buckling of the mullions. Figures 4.22(a) to (f) show the
observed structural behaviour and the failure mode of the mullion couple.

Furthermore, in these systems it is expected that the leaning action of the male and
female mullions will let the mullions act as a hollow section, thus achieving a higher
capacity. Also, it is expected that the use of a clip system will significantly increase the
lateral-torsional buckling capacity of the mullions. However, this did not happen in the
test, instead the female mullion separated from the male mullion after certain loading.
This separation of the female mullion induced the premature failure of the mullion
couple. The capacity of the mullion couple under negative wind pressure loading
increased only by about 30%, which is not significant compared to the expected
increase with the addition of a clip.

Compression
Male Female zone
mullion mullion

Tension
zone

(a) Initial state (b) Rotating together

(c) Rotating to lean on each other (d) Rotating together

Figure 4.33: Schematic diagram of the behavior of mullions in Test 3

4-52
 Experimental study

(e) Separation of female mullion at mid- (f) Failure at 450 mm away from the mid-
span span
Figure 4.33: Schematic diagram of the behavior of mullions in Test 3

4.5.4 Tests 4a, 4b and 4c: Mullion couple with a clip under positive wind pressure
loading

When a positive wind pressure loading was applied to the mullion couple with a clip,
the mullion couple did not open up, instead it rotated in the anti-clockwise direction in
the early stage of loading (Figure 4.34(b)). This behavior of the mullion couple is
similar to the behaviour of the mullion couple in Test 2, i.e. the mullion couple without
a clip tested under positive wind action. Therefore, the reason for this behaviour is also
similar to that given for Test 2 (refer the explanation given for Test 2 behaviour).

Once the seal friction was exceeded, the mullion couple then started to open up.
However, the amount of opening at mid-span region was limited due to the presence of
the clip. Because, once the clip is activated it resists further separation between the male
and female mullions (Figure 4.34(c)). Thereafter, both mullions continued to rotate
together in the anti-clockwise direction. Closer to the failure the mullions slided on the
aluminium sheets as shown in Figure 4.34(d). This is because, the mullions are laterally
restrained by the aluminium sheets at their compression flange through frictional
resistance that is developed between the aluminium sheet and the gasket contact
interfaces, and when this frictional resistance was exceeded the sliding occurred. The
force that caused the mullions’ compression flanges to move laterally was initiated by
the torsional deformations of the mullions. Finally, this sliding of the mullions on the

4-53
 Experimental study

aluminium sheet led to the disengagement of the aluminium sheet from the male
mullion to occur prior to the buckling failure of the mullions (Figure 4.34(e)). The
mullions would also have failed closer to this load as is evident from the torsional
rotation of the mullion couple (Figures 4.19(d) and 4.23(d)).

Tension
Male Female zone
mullion mullion

Clip

Compression
zone

(a) Initial state (b) Rotating together

Clip activated

(c) Gap opening (d) Sliding of mullions closer to failure

Figure 4.34: Schematic diagram of the behavior of mullions in Test 4a

4-54
 Experimental study

(e) Aluminium sheet popping out

Figure 4.34: Schematic diagram of the behavior of mullions in Test 4a

The failure pressure of the mullion couple in this test was nearly equal to the failure
pressure of the mullion couple in Test 2. This means that although the presence of clip
at mid-span resisted the separation of male and female mullions, it did not enhance the
ultimate structural performance of the mullion couple. However, this was higher than
the ultimate failure pressure of the mullion couples tested under negative wind loading
(Tests 1 and 3). This is because, under the positive wind action the mullions are laterally
restrained at their compression flanges by the glass panes, whereas their compression
flanges are unrestrained during the negative wind action.

In order to assess the effectiveness of the lateral restraint given by the aluminium sheet
to the mullions’ compression flange, two additional tests (Test 4b and 4c) were
performed on the facade panel made of a mullion couple with a clip at its mid-span. In
Test 4b, aluminium pieces were installed as shown in Figure 4.35(a) to prevent the
sliding of mullions on the aluminium sheet. In Test 4c, a 100 mm long silicone sealant
bite was used to connect the mullions with the edge of the aluminium sheets as shown
in Figure 4.36(a).

4-55
 Experimental study

Male Female Tensionn

mullion mullion zone

Clip

Lateral sliding
Aluminium restraint
pieces
Compression
zone

(a) Initial state (b) At 3400 Pa

Figure 4.35: Schematic diagram of the behavior of mullions during Test 4b

In Test 4b, the mullion couple showed similar behaviour as that observed in Test 4a
(Figures 4.34(a) to (c)) during the early stages of loading. Therefore, the reasons for the
displayed behaviour is similar to that explained for Test 4a. Thereafter, the mullions
rotated together in the anti-clockwise direction, but no noticeable sliding of the mullions
on the aluminium sheet was observed (Figure 4.25). This is because of the lateral
restraint given by the installed aluminium pieces to the mullions’ compression flanges.
Since the mullions are laterally restrained at two locations, i.e. by the aluminium sheet
and the installed aluminium pieces, the effectiveness of the restraint in minimizing the
torsional rotation and the sliding of the mullions was higher in this test than that in Tests
2 and 4a. Therefore, the mullion couple could achieve a pressure of about 3400 Pa
without failure. In Test 4b, the test was terminated at the pressure of about 3400 Pa due
to the concern of catastrophic failure of the facade assembly and consequence damage
to the LVDTs. However, it should be noted that the mullion couple was almost close to
its failure pressure (refer Section 4.4).

In Test 4c, the mullion couple was loaded until it failed. Displacement readings were
not recorded due to the fear of damage to the LVDTs at failure. During the early stage
of loading, the mullion couple showed similar behaviour to that observed in Test 4
(Figures 4.34(a) to (c)). Therefore, the reason for such behaviour is the same as that
given for Test 4a. Thereafter, both mullions continued to rotate, but the torsional
rotation of the mullion couple was less than that observed during Test 4b (Figures

4-56
 Experimental study

4.25(d) and 4.27(c)). This is because, unlike in Test 4b, the male mullion is pulled by
the silicone sealant when it tries to move laterally (Figure 4.36(b)), whereas in Test 4b,
the male mullion can move laterally until it come in contact with the female mullion.
When the pressure reached about 3500 Pa, the mullion couple started to rotate
significantly and then failed spontaneously. It is because the lateral restraint provided by
the silicone sealant was not sufficient to prevent the torsional deformation of the
mullions at that high pressure. Furthermore, the silicone sealant bite was found to have
torn from the edge of the aluminium sheet possibly due to the significant deflection and
rotation of the mullion couple at its failure.

Tension
Male Female zone
mullion mullion

Clip

Silicone
sealant bite
Pushing
Compression
Pulling zone

(a) Initial state (b) Before failure

Figure 4.36: Schematic diagram of the behavior of mullions during Test 4c

4.5.5 Test 5: Mullion couple with a clip and a hollow stiffener under negative
wind pressure loading

The presence of the hollow stiffener inside the male mullion (see Figure 4.37) could
prevent the torsional rotation of the male mullion since it has considerably higher
torsional rigidity than the open male mullion section. That is, the male mullion and the
hollow stiffener will act as a single system. The female mullion was not restrained
against its torsional rotation, thus it rotated in the anti-clockwise direction during the
negative wind action (see Figure 4.37(b)). Once the female mullion became in contact
with the male mullion, whole system acted as a single system without showing any
further torsional deformations. In Tests 1 and 3, the mullion couple rotated together
after the male and female mullions rotated to lean on each other. But such behaviour

4-57
 Experimental study

was not observed when a hollow stiffener was used with the male mullion as the male
mullion could not independently rotate to push the female mullion. Therefore, the
mullion couple could achieve a much higher capacity than that of Test 1 (more than
twice) without showing any noticeable torsional rotations that were observed during
Test 1. However, since the aluminium sheet disengaged at the transom level the test was
terminated. It should be noted that the additional bending stiffness due to the inclusion
of hollow the stiffener is also one of the reasons for such increase in the capacity.

Compression
Male Female zone
mullion mullion

Hollow
stiffener

Tension
zone

(a) Initial state (b) Rotation of the female mullion to lean

on male mullion

Figure 4.37: Schematic diagram of the behavior of mullions during Test 5

4.5.6 Comparison between the different loading and the mullion couple systems

The observed behaviour of the mullion couples with and without a clip, and the reasons
for the displayed behaviour are given in the previous section. In order to draw a good
conclusion from the conducted full-scale tests, further comparisons are made between
the obtained results of the mullion couple with and without a clip under both positive
and negative wind actions. Comparisons were also made between the Y- and X-axis
deformation aspects of the mullion couple under positive and negative wind actions.

Figure 4.38 shows that the capacity of the mullion couple under negative wind action
increased when a clip was used at its mid-span. However, the increment is not
significant as expected. Furthermore, the Y- and X-axis displacement behaviour of the
mullion couple seems to be similar in both cases.

4-58
 Experimental study

-1200
with clip
-1000

-800
Pressure (Pa)

-600
Test 1-Female mullion
-400 without clip
Test 1-Male mullion
Test 3-Female mullion
-200
Test 3-Male mullion
0
0 -10 -20 -30 -40 -50 -60 -70
Displacement (mm)

(a) Pressure versus Y-axis displacement

-1200
with clip
-1000

-800
Pressure (Pa)

-600
without clip
Test 1-Female mullion
-400
Test 1-Male mullion
Test 3-Female mullion
-200
Test 3-Male mullion
0
0 20 40 60 80
Displacement (mm)

(b) Pressure versus X-axis displacement

Figure 4.38: Mullion couples with and without a clip under negative wind action

The capacities of the mullion couples with and without a clip under positive wind action
are approximately equal. However, a significant improvement in the lateral
displacement behaviour of the mullion couple is observed. That is, the male and female

4-59
 Experimental study

mullions separated significantly when a clip was not used, whereas both mullions
deformed together when a clip was used (Figure 4.39(b)). Therefore, the clip can
improve the serviceability performance of the mullion couple. The stiffness of the
mullion couple in the Y-axis direction is nearly the same for both cases (Figure 4.39(a)),
where the mullion couple without a clip has reduced stiffness due to the separation of
the mullion couple.

3000
with clip
2500
without clip
2000
Pressure (Pa)

1500
Test 2-Female mullion
1000
Test 2-Male mullion
Test 4a-Female mullion
500
Test 4a-Male mullion
0
0 50 100 150 200
Displacement (mm)
(a) Pressure versus Y-axis displacement
3500
without clip
3000

2500
Pressure (Pa)

2000

1500 with clip Test 2-Female mullion

Test 2-Male mullion
1000
Test 4a-Female mullion
500 Test 4a-Male mullion

0
20 0 -20 -40 -60 -80 -100
Displacement (mm)
(b) Pressure versus X-axis displacement
Figure 4.39: Mullion couples with and without a clip under positive wind action

4-60
 Experimental study

Figures 4.40(a) and (b) show that the stiffness of the mullion couple under both negative
and positive wind actions is similar for the mullion couple without a clip and with a clip
at its mid-span. This means that the elastic deflection of the mullion couple can be
related to the applied pressure and the section property of the mullion couple despite
whether a clip system is used or not.

3000

2500
Absolute pressure (Pa)

Positive wind action

2000

1500
Test 1-Female mullion
1000
Test 1-Male mullion

500 Test 2-Female mullion

Negative wind action
Test 2-Male mullion
0
0 50 100 150
Y-axis displacement (mm)
(a) Mullion couple without any clip

3000

2500
Positive wind action
Absolute pressure (Pa)

2000

Negative wind action

1500

Test 3-Female mullion

1000
Test 3-Male mullion

500 Test 4a-Female mullion

Test 4a-Male mullion

0
0 20 40 60 80 100 120
Y-axis displacement (mm)
(b) Mullion couple with a clip at its mid-span
Figure 4.40: Mullion couples under positive and negative wind actions
4-61
 Experimental study

4.6 Conclusions

Six full-scale tests were conducted under both positive and negative wind actions to
understand the structural behaviour and capacity of mullion couples with and without a
clip used in unitized facade systems. Although the male and female mullions with and
without a clip showed interaction (leaning action) between them under negative wind
action, the female mullion separated from the male mullion after certain loading giving
a lower capacity than expected. Therefore, contribution of the leaning action of the male
and the female mullions is minimal and cannot be considered as an effective interaction
to treat both mullions as a hollow section.

It was believed that under negative wind action the use of a clip in a mullion couple
could provide lateral support to each other at their unrestrained compression flanges,
thus providing a significant capacity improvement. However, the use of a clip at the
mid-span did not significantly increase the lateral-torsional buckling capacity of the
mullion couple as expected. That is, the capacity of the mullion under negative wind
action increased only by 30% when a clip was used at mid-span. Therefore, the
assumption that is currently made by designers, such as that the effective length for
lateral-torsional buckling of a mullion is half the span of the mullion with a clip at its
mid-span, is not applicable.

It is evident from the test conducted under positive wind action that the ultimate
capacity increased due to the lateral restraint given by the aluminium sheets to the
compression flanges of the mullions. However, although the mullion couple was
laterally restrained, it could not achieve its section moment capacity because of the
developed torsional stresses and that the lateral restraint was only available to the
internal compression flanges of the mullions. In addition, the lateral restraint given by
the aluminium sheets was not entirely reliable since it depended on the frictional
capacity of the aluminium sheet-gasket contact interface, and a significant slip of the
mullion on aluminium sheet was noticed before the failure. Therefore, the capacity of
the mullion couple under positive wind action cannot be taken as its section moment
capacity in design.

When a clip was used the capacity of the mullion under positive wind action did not
increase. Therefore, the ultimate capacity of the mullion couple cannot be increased by
using clip systems. However, the separation between the male and female mullions, and
4-62
 Experimental study

the lateral deflection of the mullion couple were reduced when a clip was used at mid-
span. Thus, the serviceability performance of a mullion couple can be improved by
using clip systems.

The seventh full-scale test showed that when a hollow stiffener was used the
performance of the mullion couple increased significantly. However, since the test was
not conducted until failure due to the disengagement of the aluminium sheets, further
tests or numerical studies are recommended.

Finally, the stiffness of the mullion couple under both positive and negative wind
actions was found to be similar despite their different lateral deflection behaviour, and
whether a clip was used or not. Therefore, the elastic deflection of the mullion couple
under both positive and negative wind actions can be calculated based on the applied
load and its section properties.

Full-scale tests presented in this chapter were all conducted on a mullion couple made
of 650-027 and 650-028 mullions used in captive glazing system. In general, many
different shapes of mullion couples are used in both captive and structural glazing
systems. However, they have similar geometric and section characteristics. Therefore,
the conclusions given in this chapter are in general applicable to all of them.

4.7 Recommendations

Based on the observed structural response of the mullion couples in the conducted full-
scale experiments, the following recommendations are made for the structural design of
the mullion couples.

4.7.1 Mullion couple without any clips or hollow system (or hollow stiffeners)

a. Negative wind action: Since the leaning action of the male and female mullions is
not capable of providing sufficient lateral restraint to the unrestrained compression
flange of the mullions, the capacity of the mullion couple can be considered as the
addition of the individual mullion capacities. Alternatively, finite element analysis
based solutions shall be used following validation.
b. Positive wind action: The lateral restraint provided by the glass to the compression
flange of the mullions has a significant effect on the ultimate capacity of the
mullion couple. However, sliding between the glass and the gaskets could lead to

4-63
 Experimental study

premature failure of the mullions. Therefore, the capacity of the mullion couple
cannot be considered as the section moment capacity. If the advantage of the lateral
restraint given by the glass is to be considered, a valid finite element analysis
approach could be used with appropriate reduction factors to minimize risk.

4.7.2 Mullion couple with a clip at mid-span

a. Negative wind action: Although the use of clip systems could prevent the
separation of male and female mullions, it cannot prevent the overall rolling of the
mullion couple. Therefore, conservatively the capacity of the mullion couple could
be considered as the capacity of the mullion couple without any clip systems.
b. Positive wind action: When a clip was used, it did not increase the ultimate capacity
of the mullion couple, rather it improved only the serviceability performance.
Therefore, the recommendations given for the mullion couples under positive wind
action in Section 4.7.1 are applicable here.

4.7.3 Mullion couple with hollow stiffeners or mullion couple made of hollow
mullions with or without clips

a. Negative wind action: If open mullions are used with hollow stiffeners or hollow
mullions are used, the capacity of the mullion couple could be considered as the
section moment capacity. This is because when a hollow mullion and an open
mullion are used in a mullion couple the hollow mullion would prevent the
torsional rotation of the open mullion since hollow mullion is less likely to undergo
torsional rotation. Therefore, using at least one hollow mullion in a mullion couple
could increase the performance significantly. Alternatively, finite element analysis
based solutions shall be used following validation. However, the possibilities of the
failures of other components in the system, such as the disengagement of the glass
panes, should be considered to prevent unsafe designs.
b. Positive wind action: There are no full-scale test results for this case. However, by
using the knowledge gained from other test results, the following recommendations
are made. If both mullions are of hollow sections or hollow stiffeners are provided
within both male and female mullions, the capacity of the mullion couple could be
considered as the section moment capacity. When only one mullion is hollow or
hollow stiffener is provided to only one mullion without the use of any clips,
considering the capacity of the mullion couple as the section moment capacity
4-64
 Experimental study

could result in unsafe design, since the mullion without hollow stiffener is prone to
torsional deformation. But if a clip is used to prevent its torsional deformation, then
the capacity could be considered as the section moment capacity. However, it is
advisable to verify these recommendations through valid experimental and/or finite
element analysis based studies.

4-65
This Page is Intentionally Left Blank
 Development and validation of finite element models

Chapter 5

5 FINITE ELEMENT ANALYSES OF MULLION COUPLES

WITHOUT CLIP OR HOLLOW SYSTEMS

5.1 General

Advanced finite element analyses are becoming a vital part in research studies since
laboratory tests are expensive and time consuming. There are finite element analysis
packages such as ABAQUS capable of handling complex structural engineering
problems with less time and resources. ABAQUS Version 6.14-2 (DS, 2014a, b, c) was
used in this study to simulate the structural behaviour of unitized facade mullion
couples subject to wind actions in order to determine their capacities.

Two different types of finite element models were developed to simulate the behaviour
of mullion couples subject to wind actions. In the first model (“Simple model”), only
the mullion couple was modelled, and the influences of the other components were
simulated by using suitable boundary conditions in the models. In the second model
(“Full model”), the individual components including mullion couple, glass and gaskets
were modelled. Measured mechanical properties of the mullion couples were used for
improved accuracy. This chapter presents the detailed procedures used in developing the
finite element models of the mullion couples without clip systems or hollow stiffeners,
and the results. The finite element analysis results were then compared with the test
results obtained from full-scale wind pressure tests in order to validate the developed
finite element models. Based on this comparison, reliable and efficient techniques to
model the behaviour of mullion couples are recommended.

5.2 Development of finite element models

Modelling the whole test set-up is a cumbersome task in finite element simulations as
the accuracy and the ability of a simulation is often governed by the size and the amount
of information present in the model, and also the availability of computing resources.
The test set-up used in the investigation is relatively large (4000 mm in height and 2900
mm in width). Furthermore, the tested mullions and transoms are thin-walled (Figure
5.1(b)). Direct importation of CAD models or modelling of the full test set-up may
result in increased resource requirements and uncertainties in analysis results.

5-1
 Development and validation of finite element models

Therefore, it is necessary to use an idealized test set-up, without violating the

underlying principles, to develop efficient and valid finite element models.

The functions of the components that are present in the facade assembly have to be
understood prior to developing the finite element models. Tested facade panels
comprised of aluminium mullions, transoms, gaskets, seal strips and aluminium sheets
(as a replacement to glass panels). The main function of the aluminium sheets is to
transfer the lateral wind load acting on them to the mullions and transoms. They also
provide lateral restraint to the mullions. Especially this restraint is important when the
mullions are subject to positive wind action since the compression flange of the
mullions is restrained. Gaskets and seal strips are generally used to provide a watertight
seal in the facades. Other than this the gaskets also act as a cushion to glass panels to
prevent glass to metal contact, which can cause spontaneous glass breakage (Sanders
and Hargrove, 2012, McCowan and Kivela, 2010). The structural behaviour and
capacity of the mullion couples is the key focus of this study. The other components in
the facade panels are expected to influence the structural behaviour of the mullion
couples. It is not always required to model the full test set-up. Because of the
symmetries present in the system, it may be adequate to develop only a portion of the
test set-up with suitable boundary conditions. Next section discusses the symmetries
present in the system, and the suitable idealization that can be adopted in the modelling.

Three types of symmetry can be present in a model, which are: geometry, boundary and
load. The size of a finite element model can be reduced significantly by considering
these symmetries. Typical experimental set-up of the facade panel used in the full-scale
wind pressure test is shown in Figure 5.1(a). The test panels comprised similar
components about the vertical centre mullion couple (V-V axis) as well as about the
horizontal axis (H-H axis). Since the system consists similar components about both the
V-V and H-H axes, it can be considered as geometrically symmetric about these two
axes.

5-2
 Development and validation of finite element models

(a) Typical test panel (transoms are not shown)

(b) Mullion to transom connection

Figure 5.1: Typical test set-up

5-3
 Development and validation of finite element models

The mullions and transoms transfer the load to the sub-head and sub-sill through
bearing. This implies that the test panels are simply supported at the head and sill levels.
The deflections of the sub-head and sub-sill at the mullion bearing locations are not
significant. Therefore, the boundary conditions at the top and bottom ends of the
mullions are the same. Furthermore, the load was applied to the system by means of
uniform air pressure, where it was applied to the facade assembly by creating a pressure
difference between the interior and exterior sides of the chamber. This creates a uniform
air pressure on the facade system at any instance and hence the loading is also
symmetrical about the H-H axis. Therefore, it is adequate to model only half span, i.e.,
1980 mm, of the full test panel. It should be noted that even though the length of the
panels was 4000 mm, the span is considered as 3960 mm, which is the distance
measured between the free edges of the sub-head and sub-sill. This is because when the
mullions bend about H-H axis, bearing at the top and bottom ends of the mullions
occurs at the free edges of the sub-head and sub-sill (refer Chapter 4 for more details).

Male mullions were used as the side mullions (see Figure 5.1(a)). Restraints against the
torsional rotation of the two side mullions were provided in the test set-up (refer Section
4.3.2 of Chapter 4 for more details), thus they were allowed to deflect only in the Y-axis
direction. The male and female mullions have approximately closer second moments of
area about the X axis, where the second moments of area of male and female mullions
are 17.94 E+5 and 16.96 E+5 mm4 (nominal section property), respectively, and hence
the Y-axis displacement due to flexure can be expected to be about the same for male
and female mullions. Since the second moment of areas are about the same, loading can
be taken as equally distributed between the centre and side mullions. Therefore, it is
adequate to consider the quarter portion (see Figure 5.2) in the modelling based on
geometry, boundary and load symmetries.

5-4
 Development and validation of finite element models

Figure 5.2: Idealized test panel based on symmetry

The load acting on the transom is transferred to the sub-head and sub-sill through
bearing as discussed earlier in this section. Unlike mullions, the transoms are supported
by the sub-head and sub-sill throughout their length (b) (Figure 5.2). Because of this the
Y-axis deflection of the transoms is negligible. Furthermore, the support provided by
the transom to the glass at sub-head and sub-sill levels can be taken as simply
supported. Based on these, the transoms were not considered in the developed finite
element models.

Seal strips (Figure 5.1(b), which are generally used to provide weather sealing in the
mullion couples, give frictional resistance to couple’s relative lateral movement and
rotation. In the conducted tests it was observed that the seal strip affected the movement
of the mullions, particularly under positive wind action, but its influence became
ineffective with the increasing loading. Also, there are practical concerns in simulating
the real effects of the seal strip in FEA. Therefore, the seal strips and their effects were

5-5
 Development and validation of finite element models

not included in the developed finite element models. These techniques further simplify
the finite element models with huge reduction in resource requirements. This chapter
presents the details of two different models developed. Here the first one is the simple
model and the second one is the full model.

ABAQUS Version 6.14-2 (DS, 2014a, b, c) was used to develop and analyse the finite
element models. The ABAQUS Version 6.14-2 is hereafter referred to as “ABAQUS”.
Furthermore, ABAQUS/CAE Version 6.14-2 was used for pre- and post-processing of
the finite element models. The finite element analysis results were then validated using
the test results. The procedures adopted to develop the full and simple finite element
models are detailed in the following sections.

5.3 Development of simple model

5.3.1 General

A simple model that comprises only the mullion couple was developed. But the effects
of other components on the mullion couple were simulated by defining suitable
boundary conditions. In comparison to the full model, the simple model will be more
computationally efficient due to huge reduction in model size and contact interactions.
This section describes the development of the simple model.

5.3.2 Idealization of the mullion couple

The mullion couple used in facade panels are thin-walled members with complex
geometries (Figure 5.3(a)). The shell idealization of thin-walled flexural members was
successfully used in the past to analyse problems involving flexure, and suggested as an
effective and economical solution (Sadowski and Rotter, 2013, Jeyaragan, 2010 and
Wang, 2006). Therefore, the elements of the mullion cross-sections were idealized
based on equivalent area (Figure 5.3(b) and 5.4). The geometric properties of the actual
and idealized cross sections reasonably agree with each other (refer Appendix B.1 for
more details). Modelling of these thin-walled mullion sections using three-dimensional
solid elements is not preferable as it requires at least three elements throughout its
thickness to accurately model the bending behaviour (Sadowski and Rotter, 2013). A
detailed discussion is given in Section 5.3.5.

5-6
 Development and validation of finite element models

(a) Actual mullion geometry (b) Idealized mullion geometry

Figure 5.3: Mullion geometry

Male mullion

Female mullion

Figure 5.4: Idealized simple finite element model

5-7
 Development and validation of finite element models

5.3.3 Boundary conditions

By default, ABAQUS uses a right handed, rectangular Cartesian coordinate system. The
letters “X”, “Y” and “Z” are used to label the three coordinate axes of the rectangular
Cartesian system. The term “Degrees of Freedom (DOF)” is used to define the
independent variables, such as translation and rotation, associated with the finite
element nodes. ABAQUS uses numerical designation, from “1” to “6”, to label the
DOFs of the nodes in those three global Cartesian axes (DS, 2014a). The boundary
conditions associated with translational DOFs of “1”, “2” and “3”, which are the
translations along “X”, “Y” and “Z” axes are named as U1, U2 and U3, respectively.
The rotational DOFs “4”, “5” and “6”, which are the rotations about “X”, “Y” and “Z”
axes are named as UR1, UR2 and UR3, respectively. Furthermore, three-dimensional
shell elements in ABAQUS have 6 DOFs per node that are three translations and three
rotations (U1, U2, U3, UR1, UR2 and UR3), whereas solid elements have only three
DOFs per node that are translations (U1, U2 and U3).

The transoms provide restraint to twisting and translations in “X” and “Y” directions at
the ends of mullions. Hence the simply supported end of the mullion couple was
simulated by restraining the DOFs “126” (refer Appendix B.5 for more details). The
DOFs “345” were used at the mid-span of the mullion couple based on symmetry (refer
Section 5.2). The aluminium sheets provide restraint to lateral movement of the
mullions at the gasket to aluminium sheet contact interface. Since the aluminium sheets
were not modelled in the simple model, appropriate boundary conditions were provided
on the mullions to simulate its effects. Under positive wind pressure loading, the
aluminium sheets bear on the upper level gaskets, and hence the DOF “1” of the top leg
(see Figure 5.5) of mullions was restrained along the mullion length. In contrast, under
negative wind pressure loading, the aluminium sheets bear on the lower gaskets, and
therefore the DOF “1” of the bottom edge of the mullions was restrained along the
mullion length (refer Appendix B.4 for more details on lateral restraint). Figure 5.5
shows the simple model with the applied boundary conditions.

5-8
 Development and validation of finite element models

ܷͳ ൌ Ͳ
൝ ܷʹ ൌ Ͳ
ܷܴ͵ ൌ Ͳ

ܷ͵ ൌ Ͳ
൝ܷܴͳ ൌ Ͳ
ܷܴʹ ൌ Ͳ
Support

Top legs:
Mid-span ܷͳ ൌ Ͳ for positive wind action

Bottom legs:
ܷͳ ൌ Ͳ for negative wind action

Figure 5.5: Boundary conditions applied to the simple model

5.3.4 Loading method

The load transferred from the aluminium sheets to the mullion couple can be
approximated by a trapezoidal load pattern, where the trapezoid has a 45o skew angle as
shown in Figure 5.6. However, this assumption was validated using the full finite
element model results, and the details are given in Section 5.5.3. Although the load is
nearly a line load, in the finite element model it was simulated as concentrated nodal
forces along the mullion length since a line load can be represented by many fractions
of concentrated nodal forces defined on element nodes. Therefore, the assumed
trapezoidal distributed loading was applied on both the male and female mullions at
different locations depending on the type of loading (see Figure 5.6), i.e., positive or
negative wind pressure loading (refer Appendix B.4 for more details). Furthermore, the
varying part of the trapezoidal distributed loading acts over a length of b/2, where the
width of the aluminium sheet (b) is 1450 mm. The length (h/2) of the finite element
model is 1980 mm (i.e., half the length between the mullion bearing locations on the
sub-head and sub-sill based on symmetry as explained in Section 5.2). The derivation of

5-9
 Development and validation of finite element models

the relationship between the support reaction and the applied equivalent pressure
loading is given in Appendix B.2.

Linearly
Mullion couple varying load

Uniformly
distributed load

Half model

(a) Trapezoidal load distribution for a mullion couple

Mullion couple

b/2 (h-b)/2

Linearly varying Uniformly distributed

load load

Positive loading points

Negative loading points

(b) Load applied on the finite element model

Figure 5.6: Load applied on the simple model

5-10
 Development and validation of finite element models

5.3.5 Finite element type

The selection of an appropriate finite element type and mesh size is vital in finite
element simulations as it governs the accuracy of the results, and the computational
time and memory. Depending on the size and nature of the problem, the selection of
different types of element can vary. When the problem size becomes larger with many
nonlinearities, such as many parts with contact in an assembly like unitized facades
(Figure 5.1), selection of appropriate elements is a challenging task. Therefore, a study
was conducted to determine the suitable finite element types available in ABAQUS to
model the facade system. This section describes the selection procedure to determine
the most suitable finite element type in terms of accuracy and economy.

Sadowski and Rotter (2013) investigated the use of different finite elements available in
ABAQUS, to simulate the bending behaviour of thin and thick tubes made of 6063-T6
aluminium alloy. The analysis was performed considering nonlinear geometry,
buckling, ductility and strain hardening. They stated that the continuum solid elements
become uneconomical as several layers of such elements are required to model thin
sections, and concluded that shell elements can be successfully used instead without any
noticeable variation in results. In addition, the shell elements were used by various
authors in the past to model thin-walled sections in flexure (Jeyaragan, 2009, Wang,
2006, Natario et al., 2014, Kaitila, 2007, Skejic et al., 2016). Therefore, in this study the
mullions and aluminium sheets were discretized using shell elements.

In general, three-dimensional shell elements (particularly stress/displacement elements

are of our interest) are used to model structures in which the thickness is relatively
smaller than the other two dimensions. ABAQUS offers various types of
stress/displacement shell element such as conventional and continuum shell elements.
Conventional stress/displacement shell elements, which can be either thin, thick or
general-purpose, are widely used in many structural engineering problems. The
conventional shell elements discretize a body with a reference surface, and their
thickness is specified in the section property definition (DS, 2014a and DS, 2014b). The
conventional stress/displacement shell elements have both the translational and
rotational DOFs. The convention used in ABAQUS to name the shell elements depends
on the element dimensionality, i.e., three-dimensional or axisymmetric shell elements.
The naming convention used for the three-dimensional shell elements is shown in

5-11
 Development and validation of finite element models

Figure 5.7. It should be noted that only three-dimensional shell elements were taken into
consideration as the axisymmetric shell elements are outside the scope of this
investigation.

S 4 R
Reduced integration (optional)

Number of nodes

Conventional stress/displacement shell (S)

Figure 5.7: Naming convention used in ABAQUS for the three-dimensional shell
elements

General-purpose conventional shell elements in both ABAQUS/Standard and Explicit

can switch between thick and thin shell formulations depending on the shell thickness.
That is, when the shell thickness increases, they use thick shell formulation that
accounts for transverse shear deformations and when the shell thickness decreases, they
become discrete Kirchhoff thin shell elements (DS, 2014a). The S4 (4-node
quadrilateral stress/displacement shell element with full integration and large strain
formulation) and S4R (4-node quadrilateral stress/displacement shell element with
reduced integration and large strain formulation) are some of the most widely used
general-purpose conventional shell elements available in ABAQUS. Furthermore, they
consider the change in thickness of the elements. Therefore, these elements can be used
to analyze problems that include large strain, such as mullions in unitized facade
systems subject to wind actions. Table 5.1 gives the details of the conventional
stress/displacement shell elements considered for the comparison.

Table 5.1: Conventional stress/displacement shell elements (DS, 2014a)

Order of DOFs/
Element Product Description
interpolation node
Standard/ General-purpose shell with finite
S4 Explicit Linear 6 membrane strain
General-purpose shell, reduced
Standard/ integration with hourglass control
S4R Explicit Linear 6 and finite membrane strain
The S4 element does not have hourglass modes either in membrane or in-plane bending,
and hence the element does not require hourglass control. However, this element is

5-12
 Development and validation of finite element models

computationally more expensive than that of the reduced integration element (S4R). It is
because S4 element has four integration locations per element whereas S4R element has
only one integration location. The S4R element in ABAQUS uses hourglass control to
prevent hourglassing (DS, 2014a).

In order to find a suitable element type that can be adopted to model the mullion couple,
analyses were conducted in ABAQUS/Standard by considering the S4 and S4R shell
elements. The pressure versus displacement relationship of the mullion couple under
negative wind load is shown in Figure 5.8 whereas the von-Mises stress plots are shown
in Figure 5.9. It is evident from the results that both elements produced similar outputs.
Since the S4R element is computationally more efficient than the S4 element, and is
available in both ABAQUS/Standard and Explicit, S4R element was considered as the
most suitable element to model the mullion couple.

-1000
-900
-800
-700
Pressure (Pa)

-600
-500
-400
-300
-200 S4R
-100 S4
0
0 -5 -10 -15 -20 -25 -30 -35 -40

Displacement of male mullion (mm)

(a) Negative pressure versus Y-axis displacement

Figure 5.8: Comparison of pressure versus displacement curves from simple
models with different types of shell element

5-13
 Development and validation of finite element models

-1000
-900
-800
-700
Pressure (Pa)

-600
-500
-400
-300
-200 S4R
-100 S4
0
-5 15 35 55 75 95 115
Displacement of female mullion (mm)

(b) Negative pressure versus X-axis displacement

Figure 5.8: Comparison of pressure versus displacement curves from simple
models with different types of shell element

(a) S4 elements

Figure 5.9: Comparison of von-Mises stress plots from simple models with
different types of shell element

5-14
 Development and validation of finite element models

(b) S4R elements

Figure 5.9: Comparison of von-Mises stress plots from simple models with
different types of shell element

5.3.6 Mesh details

The element size is one of the key factors that governs the accuracy of the analysis
results. However, the use of large number of elements will result in increased
computational time. The aspect ratio, i.e., the ratio between the length and width, has an
effect on the solution. Elements with large or small aspect ratios can create problems in
relation to contact problems and in problems involving large deformations.

Convergence studies conducted by Jeyaragan (2009) on LSBs (LiteSteel Beam) have

proven that an element with a width of 5 mm and a length of 10 mm, which was
recommended by Mahaarachchi and Mahendran (2005), sufficiently predicted the local
buckling deformations and the spread of plasticity. Therefore, an element aspect ratio of
approximately 2, i.e., 4 mm width and 8 mm length, was adopted in this study.
However, a convergence study was conducted on a mullion couple with an element size
of 3 mm width and 4 mm length (an aspect ratio of approximately 1.3). It should be
noted that the length of an element is along the length of the mullions while the width of
the element is on the mullion cross-section (Figure 5.10). Figures 5.11 and 5.12 show
that the results obtained from both mesh sizes agreed well. In order to discretize the
model using 3 mm x 4 mm mesh size, 100980 elements were required, whereas 39184
elements were required for 4 mm x 8 mm mesh size. It reveals that the use 3 mm x 4

5-15
 Development and validation of finite element models

mm mesh size requires about 2.5 times more finite elements than the 4 mm x 8 mm
mesh size thus being computationally expensive with no noticeable benefits to the
accuracy of results.

S4R elements
(Mullions)

4 mm x 8 mm mesh

Figure 5.10: Discretized simple model

-1000
-900
-800
-700
Pressure (Pa)

-600
-500
-400
-300
-200 4 mm x 8 mm mesh (39184 elements)

-100 3 mm x 4 mm mesh (100980 elements)

0
0 -10 -20 -30 -40
Displacement of male mullion (mm)

(a) Negative pressure versus Y-axis displacement

Figure 5.11: Comparison of pressure versus displacement curves from simple
models with different mesh sizes

5-16
 Development and validation of finite element models

-1000
-900
-800
-700
Pressure (Pa)

-600
-500
-400
-300
-200 4 mm x 8 mm mesh (39184 elements)

-100 3 mm x 4 mm mesh (100980 elements)

0
-5 15 35 55 75 95 115
Displacement of female mullion (mm)

(b) Negative pressure versus X-axis displacement

Figure 5.11: Comparison of pressure versus displacement curves from simple
models with different mesh sizes

(a) 3 mm x 4 mm mesh (100980 elements)

Figure 5.12: Comparison of von-Mises stress plots from models with different
mesh sizes

5-17
 Development and validation of finite element models

(b) 4 mm x 8 mm mesh (39184 elements)

Figure 5.12: Comparison of von-Mises stress plots from models with different
mesh sizes

5.3.7 Contacts and constraints

Contact interactions between different bodies can be defined in ABAQUS using general
contact or contact pair approaches. The general contact algorithm is an automated
technique, that is, the contact interaction can be enforced over all regions of a model by
defining a single interaction. The contact pair approach requires careful definition of the
surfaces, which may potentially come into contact. Furthermore, this approach has
many restrictions on the types of surface involved. However, the general contact
approach generally requires more computational time than the contact pair approach.

Contact in the facade assembly can have two types of property, such as pressure
(normal) and friction (tangential). Normal behaviour of the contacting surfaces can be
defined as a hard or a softened contact relationship. The hard contact relationship
eliminates the penetration of the slave surface nodes into the master surface while the
softened contact relationship allows certain amount of penetration depending on the
specified contact stiffness. Tangential behaviour can be defined by specifying a friction
coefficient.

Unlike the full model, the simple model has relatively less contact interactions thus the
simple models could be analysed without much convergence difficulties than the full

5-18
 Development and validation of finite element models

model in ABAQUS/Standard. In order to assess the contact approaches available in the

ABAQUS/Standard, sensitivity analyses were performed by varying the friction
coefficient and contact stiffness. Analyses were performed on the simple model under
negative wind loading with varying contact stiffness. ‘Hard contact’, ‘Hard contact with
default contact stiffness’ and ‘Linear softened contact’ were considered, where the
contact stiffness in the softened contact considered are 500 N/mm and 5000 N/mm.
Figure 5.13 reveals that the ultimate capacity of the mullion couple was not affected by
the stiffness. The ‘Hard contact with default contact stiffness’ will be adopted in this
study since it could reduce the possible convergence problems than the default ‘Hard
contact’. The ‘Linear softened contact’ with a stiffness lower than the default stiffness
will be used if any numerical convergence difficulties are found. Similarly, the simple
model was analysed under negative wind loading by varying the friction coefficient
value from 0.1 to 0.7 in order to investigate its sensitivity on the results. Figure 5.14
shows that the effect of friction coefficient on the ultimate capacity is minimal although
there is a noticeable difference beyond failure.

-1000
-900
-800
-700
Pressure (Pa)

-600
-500
-400
Hard contact
-300 Hard contact, stiffness=default
-200 Linear contact, stiffness=5000
-100 Linear contact, sitffness=500
0
0 -10 -20 -30 -40
Y-axis displacement of male mullion (mm)

Figure 5.13: Comparison of results from models with different types of contact
stiffness

5-19
 Development and validation of finite element models

-1000
-900
-800
-700
Pressure (Pa)

-600
-500
-400
-300
Friction = 0.1
-200 Friction = 0.4
-100 Friction = 0.7
0
0 -5 -10 -15 -20 -25 -30 -35
Y-axis displacement of male mullion (mm)
Figure 5.14: Comparison of results from models with different values of friction
coefficient

5.3.8 Material properties of mullions

Accurate material properties and suitable material models are key factors in developing
reliable and accurate finite element models. This study focuses on simulating the tested
panels, and hence the structural behaviour of mullion couples in facade systems subject
to wind actions needs to be modelled. Mullions are the critical elements, and are loaded
until failure in the tests. It is important to use the accurate mechanical properties of
these mullions to obtain the accurate failure modes and loads. Therefore, the measured
mechanical properties were used in the finite element modelling. Millimeter unit was
selected, and therefore, the relevant units for the stress and density in accordance with
the ABAQUS unit convention are MPa (N/mm2) and ton/mm3, respectively (DS,
2014c).

Aluminium is a ductile material which implies that the appropriate stress and strain
measures are true stress and logarithmic strain. ABAQUS requires a stress-strain
relationship of material in terms of true stress and logarithmic plastic strain (DS,
2014a). Since the measured stress-strain curves of the aluminium mullions are
engineering stress-strain curves, these curves were converted to true stress-logarithmic
plastic strain using Equations (5.1) and (5.2). These true stress and logarithmic plastic
strain values were used as inputs to the isotropic hardening plasticity model in
ABAQUS. For an example, the measured stress-strain curve, predicted stress-strain
5-20
 Development and validation of finite element models

model using EN 1999-1-1 (CEN, 2007) and the true stress-strain curves of the 650-027
and 650-028 mullions used in Test 1 and 2 are shown in Figures 5.15(a) to (d).

ߪ௧௥௨௘ ൌ ߪ௘௡௚ ሺͳ ൅ ߝ௘௡௚ ሻ ---------------- (5.1)

௣௟ ߪ௧௥௨௘
ߝ௧௥௨௘ ൌ Žሺͳ ൅ ߝ௘௡௚ ሻ െ  ---------------- (5.2)
‫ܧ‬

300

250

200
Stress (MPa)

150

100 Test results

50 EN 1999-1-1 (CEN, 2007)

True stress-strain
0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (mm/mm)
(a) 650-027 mullion – Test 1

300

250

200
Stress (MPa)

150

100
Test results
50 EN 1999-1-1 (CEN, 2007)
True stress-strain
0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (mm/mm)
(b) 650-028 mullion – Test 1
Figure 5.15: Stress versus strain curves of Test 1 and 2 mullions

5-21
 Development and validation of finite element models

300

250

200
Stress (MPa)

150

100
Test results
50 EN 1999-1-1 (CEN, 2007)
True stress-strain
0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (mm/mm)

(c) 650-027 mullion – Test 2

300

250

200
Stress (MPa)

150

100
Test results
50 EN 1999-1-1 (CEN, 2007)
True stress-strain
0
0 0.02 0.04 0.06 0.08 0.1
Strain (mm/mm)

(d) 650-027 mullion – Test 2

Figure 5.15: Stress versus strain curves of Test 1 and 2 mullions

5-22
 Development and validation of finite element models

Deflection calculations are performed using the average modulus of elasticity, that is,
the average of tensile and compressive moduli. Since the pressure versus displacement
characteristics of the mullion couples from finite element analyses are to be compared
with the experimental results, the average modulus of elasticity of 69300 MPa was
considered for mullions (refer Chapter 3). The density and the Poisson’s ratio were
taken as 2700 kg/m3 and 0.33, respectively (AA, 2015).

5.3.9 Types of finite element analysis

Problems in structural engineering can be either static or dynamic in nature, in which

linear or nonlinear response, can be present. There are three types of nonlinearities
which can be present in a model; geometric nonlinearity due to large displacements,
material nonlinearity and boundary nonlinearities such as contact and friction. Static
problems are those in which time dimension is not considered, and therefore, the inertia
effects are eliminated from its equilibrium equation (Equation 5.3) whereas in dynamic
problems the inertia effects are significant (Equation 5.4). In ABAQUS a nonlinear
static stress analysis can be performed using ABAQUS/Standard.

‫ —ܭ‬ൌ ݂ ----------------- (5.3)

‫ܯ‬ò ൌ ܲ െ ‫ܫ‬ ----------------- (5.4)

where:

K = global stiffness matrix

u = desired nodal displacement
f = load vector
M = lumped mass matrix
ü = desired nodal acceleration
P = applied external forces
I = internal forces

Static general and Riks analysis procedures, which are available in ABAQUS/Standard,
can be used to solve problems involving true static equilibrium, where the Newton’s
method is the basis (DS, 2014a). In general, nonlinear analysis problems can be solved
using static general analysis procedures unless the stiffness matrix of the system is
singular. Buckling or geometrically nonlinear collapse behaviour of a structure creates
negative tangent stiffness, and when such a phenomenon is found, the static general

5-23
 Development and validation of finite element models

analysis method is terminated. Therefore, static general analysis with artificial damping
or Riks (or arc length) method can be used.

The Riks method provides solution for stable or unstable responses. The Riks analysis
procedure is widely used to analyse problems involving thin-walled members subject to
buckling deformations. However, it is difficult to handle problems involving post-
buckling behaviour with loss of contacts using Riks method (DS, 2014a). In order to
assess this, an analysis was performed on a mullion couple under negative wind loading
using Riks method. Since the mullion couple involved separation of contact prior to
buckling under negative wind loading, the Riks analysis procedure back tracked (the
dotted line with red color in Figure 5.16). Therefore, Riks analysis procedure will not be
considered in this study.

-1000
-900
-800
Pressure (Pa)

-700
-600
-500
-400
-300
-200
-100
0
0 -2 -4 -6 -8 -10 -12 -14
Y-axis displacement of the male mullion (mm)

Figure 5.16: Back tracking in Riks method

Quasi-static analysis in ABAQUS/Standard can be used to solve static problems where

the inertia effects can be neglected. ABAQUS/Standard uses the dynamic equilibrium
equation (Equation 5.4) with backward Euler operator (Implicit scheme) for the quasi-
static analysis. When unstable behavior is encountered, inertia effects are introduced to
stabilize the models. Quasi-static analysis using ABAQUS/Standard, i.e., the implicit
integration scheme, follows similar steps used in the static general analysis procedure to
solve nonlinear dynamic equilibrium equations, where the Newton’s method is the
basis.

5-24
 Development and validation of finite element models

When large number of nonlinearities are present in a static analysis problem, such as
many contacts and post-buckling behaviour, ABAQUS/Standard, which uses Newton’s
method for iterations to keep the error/residual under specified tolerance, often struggles
to converge. Furthermore, when the problem size becomes very large with many
nonlinearities such as contact and friction, ABAQUS/Standard is out-performed by
ABAQUS/Explicit with comparatively less computational cost. Therefore, quasi-static
analysis using ABAQUS/Explicit can be employed. However, the kinetic energy to
internal energy ratio of the system should be kept to a minimum (less than 5-10%) so
that the analysis results are not affected by the inertia effects (DS, 2014c).

ABAQUS/Explicit uses forward Euler/central difference algorithm to integrate the

dynamic equilibrium equation of motion explicitly through time. The values of the
unknown variables, such as acceleration, velocity and nodal displacement at the current
increment are found using the values of the previous increment. The acceleration within
an increment is considered to be constant, and therefore, the time increment should be
small so that the acceleration can be treated as a constant, thus it is conditionally stable.
As the time increment required in the ABAQUS/Explicit is very small compared to that
in ABAQUS/Standard, large numbers of increments are required. However, as there are
no simultaneous equations to solve and no iterations or tolerance checks are required
except the inversion of diagonal or lumped mass matrix, which is an inexpensive
process, ABAQUS/Explicit becomes an efficient option. The algorithmic summary of
ABAQUS/Explicit analysis procedure is explained through Equations (5.5) to (5.7).

calculation of accelerations using dynamic equilibrium equation,

òሺ௧ሻ ൌ  ‫ିܯ‬ଵ ሺܲሺ௧ሻ െ ‫ܫ‬ሺ௧ሻ ሻ ----------------- (5.5)

integrate explicitly through time to obtain nodal velocities,

൫ο‫ݐ‬ሺ௧ାο௧ሻ ൅ ο‫ݐ‬ሺ௧ሻ ൯
—ሶ ο௧ ൌ —ሶ ο௧ ൅ òሺ௧ሻ  ----------------- (5.6)
ሺ௧ା ଶ ሻ ሺ௧ି ଶ ሻ ʹ

integrate explicitly through time to obtain nodal displacements,

—ሺ௧ାο௧ሻ ൌ —ሺ௧ሻ ൅ ο‫ݐ‬ሺ௧ାο௧ሻ —ሶ ο௧ ----------------- (5.7)

ሺ௧ା ଶ ሻ

where:
t = time

5-25
 Development and validation of finite element models

't = time increment

‫ݑ‬ሶ = desired nodal velocity

The time increment ('t) in ABAQUS/Explicit is relatively small compared to the

increment in ABAQUS/Standard. That is, the time increment should not exceed the
time required by an elastic stress wave to pass through an element, which is of the order
of 1 x E-6 in magnitude. This time limit depends on the modulus of elasticity and
density of the material and the length of the smallest element in the system (Equation
5.8).

ߩ
ο‫ ݐ‬ൌ ‫ܮ‬௘ ൈ ට ----------------- (5.8)
‫ܧ‬
where:
Le = smallest length of an element
U = mass density
E = Young’s modulus of the material

In ABAQUS/Explicit, the time increment can be increased using the mass scaling
option so that the analysis time can be reduced. This is because the stable time
increment ('t) in Equation (5.8) increases when the mass density of the structure is
increased. However, scaling the mass of the structure can affect the solution accuracy
since the kinetic energy of the system increases as the mass of the structure is increased.
In order to ensure that the solution is not affected by the inertia effect, the ratio of
kinetic energy (KE) to internal energy (IE) should be less than 5-10% (DS, 2014c).
Figure 5.17(a) shows a comparison made between different mass scaling, where the
ultimate pressure prediction increases as the mass scaling is increased. Figure 5.15(b)
shows that the kinetic energy in the system increases with increasing mass scaling.
Therefore, the mass scaling feature cannot confidently be used in the mullion analysis
thus not considered.

5-26
 Development and validation of finite element models

-1400

-1200

Pressure (Pa) -1000

-800

-600

-400 MS=1
MS=10
-200
MS=100
0
0 -5 -10 -15 -20 -25 -30
Displacement of male mullion (mm)
(a) Pressure versus Y-axis displacement

100
90 MS=1
MS=10
80
MS=100
70
100*KE/IE

60
50
40
30
20
10
0
0 -200 -400 -600 -800 -1000 -1200 -1400
Pressure (Pa)
(b) Kinetic energy to internal energy ratio
Figure 5.17: Comparison of results obtained from models with different mass
scaling

Static general analysis procedure with artificial damping and quasi-static analysis using
ABAQUS/Standard and Explicit were considered to analyse the developed simple
model of a mullion couple. When a structure becomes unstable the static general
analysis procedure uses specified damping factor to damp the structure. The default
value of the damping factor (0.0002) specified in ABAQUS was considered in the static
general analysis procedure. Similarly, quasi-static analysis using implicit dynamics uses
mass damping to damp the structure. The quasi-static analysis in ABAQUS/Explicit

5-27
 Development and validation of finite element models

was performed without scaling the mass of the structure since mass scaling significantly
affects the solution accuracy (see Figures 5.17(a) and (b)).

A comparison between the static general analysis procedure and the quasi-static analysis
procedure using implicit and explicit integration schemes was performed on a mullion
couple subjected to negative wind action. Figures 5.18(a) and 5.19 show that the
pressure versus displacement behaviour and the predicted failure mode of the mullion
couple matched well for all the analysis procedures considered. There is only a 5%
maximum difference in the predicted ultimate capacities. When artificial damping is
used in static general analysis procedure, the ratio of static dissipation (SD) to internal
energy (IE) should be less than about 5-10% (DS, 2014c). Similarly the ratio of kinetic
energy (KE) to internal energy (IE) in quasi-static analyses should be less than the
allowable tolerance (5-10%). Figure 5.18(b) shows that the energy ratios for all three
types of analysis are within the allowable tolerance except at the initial stage and after
failure. When the mullion couple becomes unstable, kinetic energy (in quasi-static
analyses using ABAQUS/Standard and Explicit) or the static dissipated energy (static
analysis in ABAQUS/Standard) increases in the model. This is because a structure
should dissipate its strain energy as it becomes unstable, thus it cannot be fully
eliminated if a converged numerical solution is needed.

-1000
-900
-800
-700
Pressure (Pa)

-600
-500
-400
-300 Static General
-200 Quasi-Static Implicit
-100 Quasi-Static Explicit
0
0 -5 -10 -15 -20 -25 -30 -35
Displacement of male mullion (mm)
(a) Negative pressure versus Y-axis displacement

Figure 5.18: Comparison of results obtained from different analysis procedures

5-28
 Development and validation of finite element models

100
90 Static General

100*KE/IE or 100*SD/IE
80 Quasi-Static Implicit
70 Quasi-Static Explicit
60
50
40
30
20
10
0
0 -200 -400 -600 -800 -1000
Pressure (Pa)
(b) Kinetic or static dissipation energy to internal energy ratio

Figure 5.18: Comparison of results obtained from different analysis procedures

(a) Static general

(b) Quasi-static implicit

Figure 5.19: Comparison of failure modes obtained from different analysis

procedures

5-29
 Development and validation of finite element models

Irregular stress variations

(c) Quasi-static explicit

Figure 5.19: Comparison of failure modes obtained from different analysis

procedures

The use of quasi-static analysis procedure using ABAQUS/Explicit is advantageous

when numerical convergence problems exist in an analysis. However, it was found that
(see Figure 5.17) the accuracy of the solution is affected when using this analysis
procedure to investigate the mullion couple problem involving buckling. Also it is
evident from Figure 5.19(c) that the variations in the stress contours from
ABAQUS/Explicit is irregular. Therefore, the ABAQUS/Explicit analysis procedure
will not be considered in this study. Furthermore, since the quasi-static analysis using
ABAQUS/Implicit gives reasonably good results with less convergence difficulties than
the Static general analysis procedure, quasi-static analysis approach using the
ABAQUS/Implicit is adopted in this study.

The final finite element analysis results and the comparison with the test results are
presented in Section 5.5. Next section details the development of full models.

5-30
 Development and validation of finite element models

5.4 Development of full model

5.4.1 General

In the simple model, only the mullion couple was modelled, where the interactions of
the other components with the mullion couple were simulated by applying suitable
boundary conditions. Full models that comprise of a mullion couple, gaskets and
aluminium sheets were also developed. This section presents the techniques used to
develop the full models and the corresponding analysis results. The concepts of
symmetries used in developing the full model have already been presented in Section
5.2.

5.4.2 Idealization of the model

Mullions

The concept used to idealize the mullion couple in the full model is the same as that
used for the simple model. Section 5.3.2 provides more details about the idealization of
the mullion couple.

Gaskets

The gasket elements used in the facade panel have a complex geometry (Figure
5.20(a)). Gaskets act as a cushion to the aluminium sheets thus facilitate smooth
transferring of the wind load acting on the aluminium sheets to the mullions through
compression. Gaskets are made of neoprene or silicone rubber. They show flexible
behaviour under bending, but stiffen under compressive loading. The stiffening
behaviour of gaskets under compression allows it to effectively transfer the wind load
from the aluminium sheets to the mullion couple. Since the gaskets can undergo large
deformations due to their flexibility, it is necessary to control the element distortion to
avoid numerical instabilities in finite element simulations. Complex geometry with
curvatures may demand very fine meshes to avoid possible element distortions.
Therefore, the gasket geometry was idealized into a simple form (“Idealization 1”
shown in 5.20(b)) such that the intended function of the gasket was maintained without
affecting the structural behaviour of the mullion couple. However, in order to assess the
effect of idealization, the original gasket geometry was idealized also into a more close
approximation (“Idealization 2”) as shown in Figure 5.20(c).

5-31
 Development and validation of finite element models

The idealized geometry of the gasket is shown by the hatched region in Figures 5.20(b)
and (c). The locking feature of the gasket, which keeps the gaskets in position, was
replaced by tying the surface of the gasket to the surface of the mullion (see Figure
5.20), which in fact helps to get a converged solution. Full finite element models with
these two different idealizations of gasket geometry were analysed under positive wind
action, and the pressure versus displacement curves are compared in Figure 5.21. The
comparison reveals that the simplest idealization of the gasket geometry could be used
in full finite element modelling, i.e. Idealization 1.

In addition, when a facade panel is under negative wind action, the lower level gaskets
become effective, whereas the upper level gaskets are inactive. Therefore, the lower
level gaskets were only considered in modelling when the mullions were analysed under
negative wind action. Similarly, during the positive wind action, the upper level gaskets
were modelled (Figure 5.22). This reduced the model size significantly.

Locking feature
Tied surface

Idealized
geometry

(a) Actual gasket geometry (b) Idealization 1

Tied part of the surface

Idealized
geometry

(c) Idealization 2

Figure 5.20: Idealization of the gasket geometry

5-32
 Development and validation of finite element models

3500

3000

2500
Pressure (Pa)

2000

1500

1000

500 Idealization 1
Idealization 2
0
0 10 20 30 40 50 60
Y-axis displacement of female mullion (mm)

Figure 5.21: Comparison of results from full model with different idealizations of
gasket geometry

Aluminium sheets

The function of the aluminium sheets is to transfer the lateral wind load acting on them
to the mullions through bending and shear actions. These aluminium sheets were used
in the experiments as a replacement to glass panels due to safety concerns. The shell
elements can be economically used to model the plate bending behaviour of aluminium
sheets than using the three-dimensional solid elements. Furthermore, the shell thickness
is accounted for contact interactions in ABAQUS. Therefore, the aluminium sheets
were modelled using three-dimensional shell elements. Figure 5.22 shows the
developed full model that includes mullions, aluminium sheets and gaskets.

5-33
 Development and validation of finite element models

Aluminium sheets

Mullion couple

Surfaces tied
together

Surface tied to the OR

leg of mullion

Under negative wind Under positive wind

action action

Figure 5.22: Idealized full finite element model

5.4.3 Boundary conditions

The mullions were simply supported but restrained against torsional rotation at the
head/sill level, and hence the DOFs “126” were restrained at the ends of mullions.
Aluminium sheets were simply supported at the head/sill level, thus the DOF “2” was
restrained. The gaskets do not have any restraint at the head/sill level. The DOFs “156”
were restrained at the mid-span of the aluminium sheets (in the short span) based on
symmetry. At the mid-span symmetric plane of the mullions and aluminium sheets (in
the long span), DOFs “345” were restrained, whereas the DOF 3 was restrained for the
gaskets. Figure 5.23 shows the full model with the applied boundary conditions.

5-34
 Development and validation of finite element models

Simply supported end

ܷͳ ൌ Ͳ
ܷͳ ൌ Ͳ ܷʹ ൌ Ͳ ൝ ܷʹ ൌ Ͳ
൝ܷܴʹ ൌ Ͳ
ܷܴ͵ ൌ Ͳ
ܷܴ͵ ൌ Ͳ
Symmetric plane ܷʹ ൌ Ͳ
(mid-span of the
aluminium sheet)

ܷͳ ൌ Ͳ
൝ܷܴʹ ൌ Ͳ
ܷܴ͵ ൌ Ͳ
Symmetric plane
(mid-span)
Symmetric plane (mid-span
of the aluminium sheet)
ܷ͵ ൌ Ͳ
൝ܷܴͳ ൌ Ͳ
ܷܴʹ ൌ Ͳ
ܷ͵ ൌ Ͳ

OR
Under negative wind action Under positive wind action

Figure 5.23: Boundary conditions applied to the full model

5.4.4 Loading method

The load on the facade assembly was applied by creating an air pressure difference
between the interior and exterior surfaces of the pressure chamber. This applied load
was uniformly distributed over the entire panel during testing. Since shell finite
elements have two surfaces, it is possible to apply a uniform pressure load on any of
those surfaces depending on the type of loading, i.e., positive or negative wind loading.
Figure 5.24 shows the typical negative wind pressure load applied on the full finite
element model.

5-35
 Development and validation of finite element models

Aluminium sheets
Mullion couple

Uniform pressure load

Figure 5.24: Application of a uniform negative wind pressure load

5.4.5 Selection of Finite Elements

Mullion couple and aluminium sheets

Various shell elements available in ABAQUS library and their characteristics were
discussed in Section 5.3.5. It was found that S4 and S4R elements produced similar
outputs, and therefore, the mullion couple in the full model was also modelled using
S4R elements as they are computationally more efficient than S4 elements. Similarly,
the aluminium sheets were also modelled using S4R elements.

Gaskets

Idealized gasket geometry was modelled using hexahedra solid elements. Since gasket
components are made of rubber type materials, they may undergo severe element
distortion in finite element analyses. Therefore, different types of solid element
available in ABAQUS/Standard library were taken into consideration to model the
idealized gaskets.

Solid/continuum elements discretize an entire three-dimensional body. These solid

elements (particularly stress/displacement elements are of our interest) can be used in
nonlinear analyses that include contact, plasticity and large deformations. ABAQUS
offers various types of solid element, which can be found in ABAQUS/Standard solid
element libraries (DS, 2014a). In general, the solid elements have only translational

5-36
 Development and validation of finite element models

DOFs, but the solid elements with hybrid formulation have additional variable(s)
relating to pressure. The naming convention used for the three-dimensional solid
elements is shown in Figure 5.25.

C 3D 8 R H
Hybrid (optional)
Reduced integration (optional)
Number of nodes
Three-dimensional

Continuum stress/displacement

Figure 5.25: Naming convention used in ABAQUS for three-dimensional solid

elements

ABAQUS/Standard includes hexahedra first-order (linear) and second-order (quadratic)

interpolation elements. In addition, reduced integration and hybrid elements are also
available in ABAQUS/Standard library (DS, 2014a). The elements with reduced-
integration use less integration points than fully integrated elements. This results in
reduced computational time since the number of calculations required are reduced for
the elements with reduced integration. However, hourglass mode can be a problem in
the first-order reduced integration elements as they have only one integration point.
Therefore, hourglass control is introduced for the first-order elements with reduced
integration (DS, 2014a).

Neoprene (CR) and Silicone rubbers are generally used to produce the gaskets. These
materials can be considered as incompressible materials (Poisson’s ratio approaching
0.5). Since the incompressible behaviour of the materials cannot be formulated by using
only the displacement field, hybrid elements that use additional DOFs to calculate the
pressure are provided in ABAQUS/Standard. Second-order elements with full
integration are not preferred to be used if the material response is nearly incompressible
(DS, 2014a). Also the second order elements with reduced integration have more DOFs
than the first order elements with reduced integration. Therefore, second order elements
were not considered. The three-dimensional solid elements taken into account are
summarized in Table 5.2.

5-37
 Development and validation of finite element models

Table 5.2: Conventional stress/displacement solid elements (DS, 2014a)

Order of
Element Product DOFs/node Description
interpolation
Brick, hybrid with constant
C3D8H Standard Linear 3 pressure
Brick, reduced integration
with hourglass control, hybrid
C3D8RH Standard Linear 3 with constant pressure

A comparison was made between the analysis results obtained from the models that
considered the two different element types to model the gaskets. Figure 5.26 shows that
the results from both models agreed well. Therefore, the first order brick elements with
reduced integration and hybrid formulation (C3D8RH) was used to model the gaskets.

-900
-800
-700
Pressure (Pa)

-600
-500
-400
-300
-200 Gasket with C3D8RH

-100 Gasket with C3D8H

0
0 -5 -10 -15 -20 -25 -30
Displacement (mm)
Figure 5.26: Comparison of results from full models with different brick elements
for gaskets

5.4.6 Mesh

Selection of appropriate element sizes to discretize the mullion couple in the simple
model was discussed in Section 5.3.6. A mesh sensitivity study was performed on the
simple model by considering two different mesh sizes (4 mm x 8 mm and 3 mm x 4
mm). The analysis results showed that the model with a 4 mm x 8 mm mesh size
produced similar outputs to that of model with a 3 mm x 4 mm mesh size. Therefore, 4
mm x 8 mm mesh size was considered for the discretization of the mullion couple in the
full model. The aluminium sheets were discretized using element sizes of 25 mm x 25

5-38
 Development and validation of finite element models

mm. Furthermore, the gasket was discretized so that four elements are present in the
thickness direction as shown in Figure 5.27. Along the length of the gasket, i.e., in the
direction of the mullion span, 3 mm mesh length was used.

Since many contact interactions are present in the full model than in the simple model, a
comparison was made between the results of models with the selected mesh sizes and
finer mesh sizes. In the model with finer meshes, the mesh sizes used for the mullions
and aluminium sheets were 3 mm x 4 mm and 10 mm x 10 mm, respectively. The
gaskets were meshed so that six elements were present in the thickness direction, and
the length of the elements was 1.5 mm. Figure 5.28 shows that the results obtained for
the models with selected mesh sizes and finer mesh sizes agreed well. Therefore, the
selected mesh size can be used.

S4R elements
(Mullions)

S4R elements
(aluminium sheets)

C3D8RH elements
(gaskets)

OR

Under negative wind action Under positive wind action

Figure 5.27: Discretized full model in ABAQUS/Standard

5-39
 Development and validation of finite element models

-900
-800
-700
Pressure (Pa)

-600
-500
-400
-300
-200 Selected mesh
-100 Fine mesh
0
0 -5 -10 -15 -20 -25 -30
Displacement (mm)
(a) Pressure versus Y-axis displacement

(b) von-Mises stress plot from the model with selected mesh (Just before failure)

(c) von-Mises stress plot from the model with finer mesh (Just before failure)
Figure 5.28: Comparison of results from full models with different mesh sizes
under negative wind action

5-40
 Development and validation of finite element models

5.4.7 Contact and constraints

The full model includes many components of the facade assembly, such as mullions,
gaskets and aluminium sheets. There is a possibility that many of these components can
come into contact with each other. In order to ensure that all the possible contact
interactions are included, and to find the possible interacting surfaces so that they can be
modelled using contact pair approach (see Figure 5.29(a)), general contact approach
was adopted first. Then, a comparison was made between the analysis results of the
mullion couple based on general contact and contact pair approaches. Figure 5.29(b)
reveals that the pressure versus displacement curves of the mullion couple agreed well
for both cases. Therefore, both the contact pair and general contact approaches can be
used in analyzing the mullion couples. However, the general contact approach is
adopted since it reduces the difficulties and possible mistakes that could happen when
defining the contact interactions of the complex mullion couple’s behaviour.
Furthermore, to analyze the full model in ABAQUS/Standard, a friction coefficient of
0.4 and hard contact with default penalty enforcement were used to define the tangential
and normal behaviour, respectively.

(a) Full model with defined possible contact pairs

Figure 5.29: Comparison of results from full models with general and contact pair
approaches

5-41
 Development and validation of finite element models

4000
3500
3000
Pressure (Pa)

2500
2000
1500
1000
General contact
500
Contact pair
0
0 10 20 30 40 50 60 70
Displacement (mm)
(b) Pressure versus Y-axis displacement
Figure 5.29: Comparison of results from full models with general and contact pair
approaches
When the full model was analysed under positive wind action using the assumed
friction coefficient, the capacity of the mullion couple was over-predicted by the model.
Therefore, a sensitivity study was performed by assuming the friction coefficient of the
aluminium to aluminium interface to be 0.3 (Engineershandbook, 2017) and that of the
aluminium to gasket to be 0.05, 0.1, 0.2 and 0.3. When the friction coefficient of the
aluminum to gasket interface was increased, the capacity of the mullion couple
increased. Based on this study, the aluminium to gasket interface friction coefficient
was taken as 0.1. Refer Appendix B.6 for more details.

5.4.8 Material properties

The mechanical properties and the material model used for the mullion couple are the
same as used in the simple model (refer Section 5.3.8). The mechanical properties and
material model for the other components such as gaskets and aluminium sheets were
taken from the currently available standards and previous literature (SA, 1997,
Hidallana-Gamage, 2015, KP, 2017, Cambridge, 2003).

This investigation mainly focusses on the behaviour and ultimate flexural capacity of
mullion couples. The aluminum sheets were taken into consideration since they
influenced the behaviour and capacity of the mullion couple. They did not fail, or show
plastic deformations during the tests, thus actual mechanical properties of these

5-42
 Development and validation of finite element models

components are not important in modelling. Therefore, a linear elastic material model
was adopted for the aluminium sheets. The density and the elastic modulus of the
aluminium sheets were taken as 2700 kg/m3 and 69300 MPa, respectively (SA, 1997).

The gaskets are made of rubber like materials, such as innoprene and silicone, which
show hyperelastic behaviour. They have little compressibility compared to the shear
flexibility (Figure 5.30). These materials can undergo large deformations without any
change in their volume. Relative compressibility of a material can be assessed using the
K0/μ0 ratio, where K0 is the initial bulk modulus and μ0 is the initial shear modulus.
Furthermore, the relationship between Poisson’s ratio and the K0/μ0 ratio is given in
Equation (5.9) (DS, 2014a). It is evident from Equation (5.9) that as the K0/μ0 ratio
tends to infinity (or a higher value) the Poisson’s ratio approaches 0.5, which is the limit
for a fully incompressible material.

͵‫ܭ‬଴ Τߤ଴ െ ʹ
Xൌ ή ---------------- (5.9)
͸‫ܭ‬଴ Τߤ଴ ൅ ʹ
where:
E = modulus of elasticity in compression
X = Poisson’s ratio
K0 = initial bulk modulus
μ0 = initial shear modulus

0.80
0.70
0.60
0.50
Stress (MPa)

0.40
0.30
0.20
Compression
0.10
Shear
0.00
0 20 40 60 80 100 120
Strain (%)
Figure 5.30: Stress versus strain curves of DC983 silicone rubber (Dow Corning,
2017)

5-43
 Development and validation of finite element models

It is difficult to measure the modulus of elasticity of elastomeric materials using routine

methods. This is because special techniques are required for the experimental set-up and
to measure the elongation (Kunz and Studer, 2006). However, the deformation
behaviour of these materials can be expressed in terms of their Shore A hardness value,
which can be easily measured (Kunz and Studer, 2006). The Shore A hardness value of
these materials is often reported by manufacturers. The basic relation between the Shore
A hardness and the elastic modulus is given by Equation (5.10). Kunz and Studer
(2006) investigated, and further improved this with a correction factor as given in
Equation (5.11).

ͳ െ Xଶ ‫ܥ‬ଵ ൅ ‫ܥ‬ଶ ή ݄ܵ஺

‫ܧ‬ൌ ή ----------------(5.10)
ʹ ή ܴ ή ‫ܥ‬ଷ ͳͲͲ െ ݄ܵ஺

ͳ െ Xଶ ‫ܥ‬ଵ ൅ ‫ܥ‬ଶ ή ݄ܵ஺ ----------------(5.11)

‫ܧ‬ൌ ή ή ሺʹǤ͸ െ ͲǤͲʹ ή ݄ܵ஺ ሻ
ʹ ή ܴ ή ‫ܥ‬ଷ ͳͲͲ െ ݄ܵ஺

where:
E = modulus of elasticity in compression
X = Poisson’s ratio
ShA = Shore A hardness
R = radius of the indenter used to measure the Shore A hardness value (0.395 mm)
C1 = 0.549 N
C2 = 0.07516 N
C3 = 0.025 mm

One of the functions of the gasket is to transfer the wind load acting on the aluminium
sheet to the mullions through compression. In order to find a suitable value for the
modulus of elasticity of the gasket component, the Shore A hardness values available in
the Manufacturers’ data sheets and the available literature were used. The Shore A
hardness of 1730B type innoprene gasket (a grade of gasket material used to produce
the gaskets) is 81A based on material data sheet obtained from Kumho Polychem (KP,
2017). Thus the compressive modulus of elasticity based on Equation (5.11) is about
13.6 MPa. The modulus elasticity of silicone rubber specified in the Cambridge
University materials data book (Cambridge, 2003) varies in the range of 5 to 20 MPa.
Based on these, the modulus of elasticity of gasket components was taken as 10 MPa in
finite element simulations. The value of Poisson’s ratio was assumed to be 0.4995.

5-44
 Development and validation of finite element models

Furthermore, the density of the gaskets was considered to be 1.1E-9 ton/mm3 or 1100
kg/m3 (Hidallana-Gamage, 2015).

In order to investigate the effect of the modulus of elasticity of gaskets, finite element
simulations were performed under positive wind action by varying the value of the
modulus of elasticity from 5 to 50 MPa, whereas the other parameters of the model
were kept the same for all the models. The results showed that the predicted ultimate
capacity of the mullion couple increased when the modulus of elasticity of the gaskets
was increased. The maximum difference noticed between the highest and the lowest
predictions was about 15%. However the failure mode of all the models were similar.
Refer Appendix B.6 for more details of the results and comparison.

5.4.9 Type of finite element analysis

In Section 5.3.9, various analysis methods that are available in ABAQUS were
assessed, and found that the quasi-static implicit approach works better for analyzing
the mullion couple’s behaviour using the simple models. Therefore, the quasi-static
implicit approach was also adopted in analyzing the full models. Unlike the simple
model, in the full model, the load from the aluminium sheets is transferred to the
mullions by means of contact interaction. That is, there are no structural connections
between the gaskets and aluminium sheets. Therefore, the kinetic energy (KE) to
internal energy (IE) ratio (KE/IE) was found to be higher than 100% during the initial
stage of loading (Figure 5.31(b)). However, this effect is not throughout the loading.

When quasi-static analysis is performed, the kinetic energy (KE) of the system should
be less than about 5-10% of the internal energy (IE) in order to ensure that the inertia
effects are minimal (DS, 2014c). To assess this, a finite element simulation was
performed, where the mass of the aluminium sheet was scaled down by 10 times (270
kg/mm3 instead of 2700 kg/m3). Figure 5.31(a) reveals that although the initial energy
ratio was higher, it did not affect the capacity of the mullion couple. However, the
scaled mass of the aluminium sheet (270 kg/m3) was used in analyzing the other full
models. Furthermore, the energy ratio remained far below 1% except during the initial
loading and after failure (Figure 5.31(b)).

5-45
 Development and validation of finite element models

-1000
-900
-800
-700
Pressure (Pa)

-600
-500
-400
-300
270 kg/m3
-200
-100 2700 kg/m3
0
0 -5 -10 -15 -20 -25 -30
Displacement (mm)
(a) Pressure versus Y-axis displacement

100
90 270 kg/m3
80 2700 kg/m3
70
100 * KE/IE

60
50
40
30
20
10
0
0 -100 -200 -300 -400 -500 -600 -700 -800 -900
Pressure (Pa)
(b) Energy ratio versus pressure
Figure 5.31: Comparison of results from models with different aluminium sheet
density

5-46
 Development and validation of finite element models

5.5 Results and discussion

5.5.1 General

Full and simple models of the full-scale tests of facades made of 650-027 (male) and
650-028 (female) mullion couples without any clips were developed. The measured
thicknesses of the tested mullion sections were used in the finite element models. In
order to compare and validate the finite element models against experimental results,
pressure versus displacement characteristics of the mullion couple and the deformed
shapes were also obtained from the finite element analyses. The displacement locations
of the mullion couple considered in the finite element models are identical to those used
to measure those displacements in the full-scale tests (Refer Section 4.3.2 of Chapter 4).

5.5.2 Simple model

Test 1: Mullion couple under negative wind action

In the finite element analysis simulation, when the load was increased, the male and
female mullions first rotated to lean on each other. Thereafter, both of them started to
rotate together in the clockwise direction. After a certain loading the female mullion
separated from the male mullion, but both mullions continued to rotate in the same
direction. Finally both of them failed in lateral-torsional buckling. This behaviour of the
mullion couple is exactly the same as in Test 1 (Figure 5.32). Figure 5.33(a)
demonstrates that the pressure versus male and female mullions’ Y-axis displacement
curves are predicted by this model with a reasonable accuracy. Finite element analysis
(FEA) prediction is about 10% lower than the test ultimate capacity. This could be due
to variation in the test results. The pressure versus male and female mullions’ X-axis
displacement curves also reasonably agreed with the test results as shown in Figure
5.33(b). Furthermore, the ratio between the kinetic energy (KE) to internal energy (IE)
of the mullion couple for the quasi-static implicit analysis is shown in Figure 5.34.
Although it is higher at the beginning, it remained less than 1% throughout the analysis
until failure.

5-47
 Development and validation of finite element models

Leaning

Test Finite element analysis

(a) Mullions lean on each other

Male
Female mullion
mullion

Female Male mullion

Test Finite element analysis

(b) Mullions rotating together

Separation Separation

Test Finite element analysis

(c) Separation of female and male mullions

Figure 5.32: Mullion couple behaviour in FEA (simple model) and Test 1

5-48
 Development and validation of finite element models

Female mullion Male mullion Female mullion Male mullion

Test Finite element analysis

(d) Failure of mullions

Male mullion

Female mullion

(e) Three dimensional view of the failed mullion couple in FEA

Figure 5.32: Mullion couple behaviour in FEA (simple model) and Test 1

-1000
-900
-800
-700
Pressure (Pa)

-600
-500
-400
-300 Test (female)
Test (male)
-200
FEA (female)
-100 FEA (male)
0
0 -10 -20 -30 -40
Displacement (mm)
(a) Pressure versus Y-axis displacement
Figure 5.33: Comparison of results from FEA (simple model) and Test 1

5-49
 Development and validation of finite element models

-1000
-900
-800
-700
Pressure (Pa)

-600
-500
-400
Test (female)
-300 Test (male)
-200 FEA (female)
-100 FEA (male)
0
-5 15 35 55 75 95
Displacement (mm)
(b) Pressure versus X-axis displacement
Figure 5.33: Comparison of results from FEA (simple model) and Test 1

100
90
80
70
100 * KE/IE

60
50
40
30
20
10
0
0 -200 -400 -600 -800
Pressure (Pa)
Figure 5.34: Energy ratio for FEA (simple model) - Test 1

Test 2: Mullion couple under positive wind action

In the finite element analysis simulation, when the mullion couple was loaded under
positive wind action, the gap between the male and female mullions started to open up
(Figure 5.35(a)). However, this behaviour was not observed in the test until about 1500
Pa. This can be due to the presence of seal strip in the test (Figure 5.35(b)). After a
certain loading, the compression flange of the male and female mullions come into
contact and started to lean on each other as shown in Figure 5.35(b). Finally, both
mullions failed in buckling, where the female mullion also rotated in the direction of the
male mullion (Figure 5.35(c)). Although the failure mode of the mullion couple in FEA

5-50
 Development and validation of finite element models

matched well with the test, the ultimate failure capacity predicted by the FEA was
higher than the test (Figure 5.36). It is because the full lateral restraint provided at the
compression flanges gives more rigidity against the torsional rotation and lateral
movement of the mullions. In reality, the lateral restraint given by the aluminium sheets
to the mullions compression flange could be elastic. Therefore, instead of using full
lateral restraint, spring lateral restraint was adopted. Details of the determination of an
approximate spring stiffness are given in Appendix B.3.

Opening

Test Finite element analysis

(a) Early stage of loading

Leaning on
each other

Compression
flanges

Test Finite element analysis

(b) After 1500 Pa

Figure 5.35: Mullion couple behaviour in FEA (simple model) and Test 2

5-51
 Development and validation of finite element models

Female Male
Female mullion Male mullion mullion
mullion

Test Finite element analysis

(c) At failure

Male mullion

Female mullion

(d) Three dimensional view of the failed mullion couple in FEA

Figure 5.35: Mullion couple behaviour in FEA (simple model) and Test 2

3500
3000
2500
Pressure (Pa)

2000
1500
1000
Test
500
FEA
0
0 50 100 150 200
Displacement (mm)
(a) Pressure versus Y-axis displacement
Figure 5.36: Comparison of results from FEA (simple model) and Test 2

5-52
 Development and validation of finite element models

3500

3000

Pressure (Pa) 2500

2000

1500
Test (Female)
1000 Test (Male)
FEA (Female)
500
FEA (Male)
0
-150 -100 -50 0 50
Displacement (mm)
(b) Pressure versus X-axis displacement
Figure 5.36: Comparison of results from FEA (simple model) and Test 2

When the mullion couple was analysed with the spring lateral restraint, where a
stiffness value of 0.32 N/mm/mm was used (refer Appendix B.3), the ultimate failure
capacity reduced compared to the model with full lateral restraint, and it agreed well
with the test capacity (Figure 5.37(a)). The difference between the capacities obtained
from the test and FEA is about 3%. Also the kinetic energy (KE) to internal energy (IE)
ratio of the system remained below 1% throughout the analysis as shown in Figure 5.38.
Although similar to Test 2, the male mullion rotated and failed in the anti-clockwise
direction, while the female mullion failed by rotating in the clockwise direction
(opposite to the male mullion’s failure direction). It is evident from Figure 5.37(b) that
the X-axis displacement of the male and female mullions increased suddenly when the
mullions failed in buckling. In addition, experimental observation showed that the
compression flange of the male mullion displaced away from the aluminium sheets at
failure (see Figure 4.19 in Section 4). Therefore, in the test the male mullion’s
movement away from the aluminium sheet at failure could have forced the female
mullion to rotate and fail in its direction.

5-53
 Development and validation of finite element models

3500

3000

2500
Pressure (Pa)

2000

1500

1000 Test
FEA (Female)
500
FEA (Male)
0
0 50 100 150 200
Displacement (mm)

(a) Pressure versus Y-axis displacement

3000

2500

2000
Pressure (Pa)

1500

Test (Female)
1000
Test (Male)
500 FEA (Female)
FEA (Male)
0
-150 -100 -50 0 50 100 150
Displacement (mm)

(b) Pressure versus X-axis displacement

Figure 5.37: Comparison of results from FEA (simple model) with spring lateral
restraint and Test 2

5-54
 Development and validation of finite element models

100
90
80
100 * KE/IE 70
60
50
40
30
20
10
0
0 500 1000 1500 2000 2500 3000
Pressure (Pa)
Figure 5.38: Energy ratio for FEA (simple model) - Test 2

Table 5.3 presents a summary of the ultimate bending capacities under negative wind
action determined using ADM (AA, 2015), finite element analyses, and tests. The
capacities of the individual male and female mullions were calculated using ADM (AA,
2015) as per the current practice. The capacity of the mullion couple without
considering the interaction between the male and female mullions is taken as the
summation of their individual design capacities. The capacity of the mullion couple
considering the interaction between the male and female mullions is based on FEA and
Test 1. It is evident from the results that the capacity predicted using ADM (AA, 2015)
is over-conservative when compared with those obtained from FEA and test. The
capacity of the mullion couple predicted using separate FEA of the male and female
mullions is approximately equal to that predicted by the FEA of the mullion couple
where the male and female interactions were considered (1.94 and 2.03 kNm,
respectively). Also the capacities of the mullion couple predicted by FEA with and
without the interaction of the male and female mullions are closer to the capacity
obtained from the test as shown in Table 5.3.

5-55
 Development and validation of finite element models

Table 5.3: Ultimate bending moment capacity predictions under negative wind
action (kNm)

Member ADM FEA Test ADM/ FEA

Single male mullion 0.54 1.10 - 0.49

Single female mullion 0.32 0.84 - 0.38
Capacity of the mullion couple
based on the capacity of the 0.86 1.94 - 0.44
single mullions
Capacity of the mullion couple
considering the interaction of - 2.03 2.30 -
male and female mullions

Pressure versus displacement curves of the single and coupled mullions obtained from
FEA under negative wind loading are shown in Figure 5.39, where the single mullion
analyses were performed on individual mullions using the same modeling techniques
used for the coupled mullion. Although the capacity predicted by adding the analysis
results of single male and female mullions are closer to the capacity predicted through
FEA of mullion couple, the respective displacements of the individual mullions are
overestimated by the FEA of single mullions. Therefore, FEA of single mullions needs
further investigation to verify its applicability in practical design.

-1000
-900
-800
-700
Pressure (Pa)

-600
-500
-400 Female (couple)
-300 Male (couple)
-200 Female (single)
-100 Male (single)
0
0 -50 -100 -150 -200 -250
Displacement (mm)
(a) Pressure versus Y-axis displacement
Figure 5.39: Pressure versus displacement curves for single and coupled mullions
under negative wind action
5-56
 Development and validation of finite element models

-1000
-900
-800
Pressure (Pa) -700
-600
-500
-400 Female (couple)
-300 Male (couple)
-200 Female (single)
-100 Male (single)
0
-200 -150 -100 -50 0 50 100 150 200
Displacement (mm)
(b) Pressure versus X-axis displacement
Figure 5.39: Pressure versus displacement curves for single and coupled mullions
under negative wind action

Similarly, a comparison is made for the mullion couple under positive wind action. It is
evident from Table 5.4 that the capacity prediction based on ADM (AA, 2015) is much
less than the FEA predictions or test result. This is because in ADM (AA, 2015) the
mullions are considered to be unrestrained. Also the capacity prediction under positive
wind action is less than that under the negative wind action since the positive wind load
acts towards the shear centre of the mullions. Furthermore, unlike in negative wind
action, the interaction of the mullions result in about 74% higher capacity than that
without the interaction, where the individual male and female mullions were analysed
with the determined spring stiffness. However, the actual capacity of the mullion couple
obtained from test and FEA is less than that of yield capacity. Therefore, the capacity of
the mullion couple cannot be considered as the section moment capacity. Here the
section moment capacity of the mullion couple could be more than the yield capacity
since these mullion sections are generally non-slender. It should be noted that the
pressure versus displacement curves for the positive wind action case is not plotted
since it yields similar argument as that of negative wind action case.

5-57
 Development and validation of finite element models

Table 5.4: Ultimate bending moment capacity predictions under positive wind
action (kNm)

Yield Test/ Yield

Member ADM FEA Test capacity capacity

Single male mullion 0.30 2.08 - 4.83 -

Single female mullion 0.21 2.22 - 4.72 -
Capacity of the mullion couple
based on the capacity of the 0.51 4.30 - 9.55 -
single mullions
Capacity of the mullion couple
considering the interaction of - 7.50 7.31 - 0.77
male and female mullions

5.5.3 Full model

In the conducted tests, the mullions were laterally restrained by the aluminium sheets
through friction between the aluminium sheet and the gaskets. The magnitude of the
frictional force depends on the normal pressure and the friction coefficient of the
contacting interface. When the frictional force limit is exceeded, lateral sliding of the
mullions can occur, which was noticed during the conducted positive wind pressure
tests. A significant sliding of the mullions on aluminium sheets was observed prior to
the lateral torsional buckling failure of the mullion couple in Tests 2 and 4 (refer
Section 4). The developed full model was able to simulate this complex behaviour of
the mullion couple under both positive and negative wind actions, and the results are
given next.

Test 1: Mullion couple under negative wind action

The developed full model of the mullion couple was analysed under negative wind
action. Figure 5.40 shows a comparison of pressure versus displacement curves for the
mullion couple in Test 1 and FEA. The ultimate capacity predicted by the FEA of full
model is about 15% less than the test results. Although the full model shows a little
stiffer behaviour than the simple model, both models simulated the structural behaviour
of the mullion couple reasonably well. The behaviour of the mullion couple predicted
by the full model is similar to that predicted by the simple model. Refer Section 5.5.2
for the discussion on the behaviour of the mullion couple in Test 1 and FEA. Figures

5-58
 Development and validation of finite element models

5.41 and 5.42 show the kinetic energy (KE) to internal energy (IE) ratio of the model
and the failure mode of the mullion couple, respectively.

-1000
-900
-800
Pressure (Pa)

-700
-600
-500
Test (Female)
-400
Test (Male)
-300
-200 FEA (Female)

-100 FEA (Male)

0
0 -10 -20 -30 -40 -50
Displacement (mm)
(a) Pressure versus Y-axis displacement

-1000
-900
-800
-700
Pressure (Pa)

-600
-500
-400 Test (Female)
-300 Test (Male)
-200 FEA (Female)
-100 FEA (Male)

0
-5 15 35 55 75 95
Displacement (mm)
(b) Pressure versus X-axis displacement

Figure 5.40: Comparison of results from FEA (full model) and Test 1

5-59
 Development and validation of finite element models

100
90
80
70
100 * KE/IE

60
50
40
30
20
10
0
0 -100 -200 -300 -400 -500 -600 -700 -800
Pressure (Pa)

Figure 5.41: Energy ratio versus pressure for FEA (full model) - Test 1

Male mullion

Female mullion

Figure 5.42: Predicted failure mode of the mullion couple by full model – Test 1

The load that is transferred from the aluminium sheet to the mullion couple was
assumed to be trapezoidal, and was used in the simple model (refer Section 5.3.4 and
Appendix B.2). In full model, the load was simulated as uniform pressure, which
represents the real conditions. In order to ensure that the load from the aluminium sheets
to the mullions can be represented by the assumed trapezoidal loading, the ratio between
the load that is transferred to the mullion couple and the applied total load on the

5-60
 Development and validation of finite element models

aluminium sheets was assessed. Figure 5.43 shows that although the load that was
transferred to the mullion couple in the full model was slightly higher in the early stage
of loading, it agreed well with the theoretical assumption as described in Appendix B.2
(red line in Figure 5.43) in the rest of the analysis. Therefore, the assumption of
trapezoidal loading is valid.

1
0.9 Based on full model
0.8
0.7
Load ratio

0.6
Theoretical
0.5
0.4
0.3
0.2
0.1
0
0 -200 -400 -600 -800 -1000
Pressure (Pa)
Figure 5.43: Load carried by the mullion couple in the full model - Test 1

Test 2: Mullion couple under positive wind action

The developed full model of the mullion couple was analysed under positive wind
action. Figure 5.44 compares the pressure versus displacement curves of the mullion
couple from Test 2 and FEA. The ultimate capacity predicted by the full model is about
7% higher than the test results. The full model shows slightly stiffer behaviour than the
test result (Figure 5.44(b)). However, the failure mode of the mullion couple predicted
by the full model agreed well with the test result (Figure 5.46). That is, the rotation of
both the male and female mullions together in the anti-clockwise direction and the
sliding of the mullions on the aluminium sheets prior to failure observed during the test
were simulated by the full model. This is because the full model simulated the real
lateral restraint that was given to the mullions by the aluminium sheets through
frictional interaction. Figure 5.45 shows the kinetic energy (KE) to internal energy (IE)
ratio of the model.

5-61
 Development and validation of finite element models

3500

3000

2500
Pressure (Pa)

2000

1500

1000 Test (Female)

Test (Male)
500 FEA (Female)
FEA (Male)
0
0 50 100 150 200
Displacement (mm)
(a) Pressure versus Y-axis displacement

3500

3000

2500
Pressure (Pa)

2000

1500
Test (female)
1000
Test (male)
500 FEA (male)
FEA (female)
0
-100 -80 -60 -40 -20 0 20
Displacement (mm)
(b) Pressure versus X-axis displacement
Figure 5.44: Comparison of results from FEA (full model) and Test 2

5-62
 Development and validation of finite element models

100
90
80
70
100 * KE/IE 60
50
40
30
20
10
0
0 500 1000 1500 2000 2500 3000 3500
Pressure (Pa)
Figure 5.45: Energy ratio versus pressure for FEA (full model) - Test 2

At zero pressure Before failure

(a) Test 2

Male mullion

Female mullion

Sliding of mullions

(b) FEA
Figure 5.46: Predicted failure modes of the mullion couple by full model and Test 2

5-63
 Development and validation of finite element models

In order to verify the assumption regarding the trapezoidal loading for the simple
models, a comparison was made between the theoretical and FEA based load ratios for
Test 1 model (Figure 5.43). A similar comparison was made for Test 2 model as shown
in Figure 5.47. Although the load that was transferred to the mullion couple was slightly
higher in the early stage of loading, it agreed well with the theoretical assumption in the
rest of the analyses. Therefore, the trapezoidal loading assumption is further verified.

1.0
0.9
0.8
0.7
Load ratio

0.6
0.5
0.4
0.3
0.2
0.1
0.0
0 500 1000 1500 2000 2500 3000
Pressure (Pa)
Figure 5.47: Load sharing to the mullion couple in the full model - Test 2

5.6 Conclusion

Full and simple models of the tested mullion couple without clips or hollow stiffeners
were developed in this study. For the mullion couple under negative wind action, the
pressure versus displacement behavior throughout the loading, the ultimate capacity and
the failure mode of the mullion couple predicted by the finite element analysis (FEA)
based on both the full and simple models agreed reasonably well with the test results.
The differences in the ultimate capacities predicted by full and simple models were
about 15% and 10%, respectively, and the FEA predictions were conservative. These
comparisons demonstrated that the developed simple model can be successfully used to
predict the structural capacity of mullion couples under negative wind action.
Furthermore, FEA were performed separately for the male and female mullion models
without considering their interaction, and the addition of the individual male and female
capacities were compared with the capacity of the mullion couple (male and female
mullions together), where the interaction between the male and female mullions was
considered. The difference was small, thus showing that the male and female mullion
5-64
 Development and validation of finite element models

interaction did not significantly increase the capacity of the mullion couple under
negative wind action.

Simple model of the mullion couple with full lateral restraint under positive wind action
predicted the failure mode reasonably well, but the ultimate failure pressure was
overestimated. This is due the assumption of the availability of full lateral restraint to
the mullions by the aluminium sheets. In reality, due to the elastic behaviour of the
gasket that connects the aluminium sheets and the mullions, the aluminium sheets may
not provide full lateral restraint to the mullions. Therefore, an appropriate spring
stiffness value was calculated and used with the models. The ultimate capacity predicted
in this case agreed well with the test capacity, with a difference of less than 3%.
However, the failure mode of the female mullion was different to that obtained in the
test for this case. Based on these, simple model with full lateral restraint cannot be used
since it will result in unsafe predictions. Furthermore, FEA were performed separately
for the male and female mullion models without considering their interaction, and the
addition of the individual male and female capacities were compared with the capacity
of the mullion couple (male and female mullions together), where the interaction
between the male and female mullions was considered. The capacity increment due to
the interaction was about 50%, thus showing that the capacity of the mullion couple
under positive wind action is influenced by the interaction. However, a detailed
investigation is needed to verify this, and it will be undertaken using a finite element
analysis based parametric study.

The developed full model under positive wind action predicted the failure capacity
reasonably well, where it was slightly higher than the test capacity (about 7%). Also, the
structural behaviour of the mullion couple was predicted accurately by the full model.
However, using the full model to investigate a wide range of mullion couples could
result in unconservative predictions since it involves many uncertainties than that of a
simple model. Therefore, the simple model with the appropriate spring lateral restraint
could be considered as a safe approach to predict the ultimate capacity of the mullion
couples under positive wind action.

5-65
This Page is Intentionally Left Blank
 Parametric study of mullion couples under negative wind action

Chapter 6

6 PARAMETRIC STUDY OF MULLION COUPLES

WITHOUT CLIP OR HOLLOW SYSTEMS UNDER
NEGATIVE WIND ACTION
6.1 General

In Chapter 5, full and simple finite element models of a mullion couple without any clip
or hollow systems were developed and validated using the full-scale test results. The
simple finite element (FE) model is more economical than the full FE model in terms of
time and computer resources, thus simple FE models were used in this parametric
study. However, full finite element models were developed and analysed for some
selected mullion couples, which included the captive and structural glazing systems.
Then the results of the full FE models were compared with the results of their
respective simple FE models to further verify the developed finite element modelling
(FEM) and analysis approaches. This parametric study investigates the bending
moment capacity of a range of mullion couple sections, with varying shapes, glazing
types and depths, with different spans (from 1800 mm to 4200 mm) under the negative
wind action. In addition, the effects of gap between the mullions, lateral restraint, load
distribution profile, ratio of the loads acting on male and female mullions and
intermediate transoms were investigated in detail. Finally, the capacity predictions
based on the current design approaches were compared with the predictions from FE
models, and suitable recommendations are given for the design of aluminium mullion
couples used in facades.

6.2 Mullion sections and influential parameters

Mullion geometries used in facades can vary based on the type of glazing system used,
which are captive or structural (Figure 6.1). The depth of the sections commonly varies
between 100 mm and 200 mm depending on the architectural and structural
requirements. In order to study the mullion couple behaviour and capacity in detail, 20
mullion couples representing different types, shapes and sizes were considered in this
chapter (Figure 6.1). The mullion sections considered will also cover a large range of
slenderness for the local and global buckling of the mullion couples used in facade
systems. Furthermore, it should be noted that all the mullion couple sections used in this

6-1
 Parametric study of mullion couples under negative wind action

study are currently used in the industry. The details of the mullion couple sections used
are given in Table 6.1, whereas the section properties of the mullion couples are given
in Appendix C.1. Figure 6.2 shows the naming convention used to label the mullion
couples.

MC1 MC2 MC3 MC4 MC5

MC6 MC7 MC8 MC9 MC10

MC11 MC12 MC13 MC14 MC15

MC16 MC17 MC18 MC19 MC20

Figure 6.1: Shapes of mullion couples

6-2
 Parametric study of mullion couples under negative wind action

Table 6.1: Details of mullion couples

Name Type of Glazing Depth (mm) Gap (mm) Remark

MC1 102 3 These mullion couples
MC2 102 1 comprise the systems
MC3 102 3 shown in Figures
MC4 117 3 6.3(a) and (b).
MC5 152 3
MC6 152 1
Captive
MC7 152 3
MC8 152 3
MC9 152 3
MC10 152 3
MC11 172 6
MC12 200 3
MC13 92 3 These mullion couples
MC14 142 3 are similar to the
MC15 152 3 system shown in
MC16 162 8 Figure 6.3(c).
Structural
MC17 172 6
MC18 168 6
MC19 182 3
MC20 190 3

Note: The mullion couples are named as follows:

MC 1
Mullion
Number
Couple

Figure 6.2: Naming convention used to label mullion couples

The list of mullion couple sections given in Table 6.1 comprises captive glazing
mullions with gasket only, captive glazing mullions with gasket and weather seal and
structural glazing mullions with silicone sealant. Brief details of the different types of
mullion systems are given next.

Captive glazing mullions with gasket only

In the captive glazed system, rubber gaskets are used, where the gaskets act as a
cushion to the glass panes and there is no structural bond between the gaskets and glass
and gaskets and mullions (Figure 6.3(a)). The sequence of installation of the captive
glazing system is: 1). Inserting the inside or the outside gasket 2). Placing the glass
pane in place 3). Inserting the remaining outside or the inside gaskets. This is because
pushing the glass panes between the two gaskets is more difficult than inserting the

6-3
 Parametric study of mullion couples under negative wind action

gaskets after placing the glass pane. Furthermore, the wind load that is acting on the
glass panes is transferred to the mullions through the inside or outside gaskets. That is,
the inside or outside gaskets are compressed under the positive and negative wind
actions, respectively, while transferring the wind load from glass panes to the mullions.

Line of action
of the loading Interior of a building (Inside)

Glazing
Load transfer
Glass pane tape Weather
to mullions
seal

Outside gaskets Exterior of a building (Outside) Outside gaskets

(a) Captive glazing mullions – Gasket (b) Captive glazing mullions - Gasket and
only weather seal

Interior of a building (Inside)

Glazing
tape Structural silicone
sealant bites

Weather seal
Glass panes

Exterior of a building (Outside)

(c) Structural glazing mullions

Figure 6.3: Captive and structural glazing mullions

Captive glazing mullions with gasket and weather seal

Sometimes mullion sections are used, where gaskets are used outside and glazing tape
together with weather seal inside (Figure 6.3(b)). In this system, weather seal is bonded
to the glass panes and mullions, which is not designed for any structural purpose. But

6-4
 Parametric study of mullion couples under negative wind action

there is no bond between the gaskets and the glass panes and gaskets and the mullions.
Furthermore, the wind load that is transferred from glass panes to the mullions is
similar to the captive glazing mullions with gasket only.

Structural glazing mullions

In the structurally glazed system, glass panes are bonded to the mullions using
structural silicone sealant bite (see Figure 6.3(c)), and therefore, the mullions are
structurally connected to the glass panes. In this system, the area of aluminium exposed
to the outside is less compared to the captive glazing systems. Therefore, it provides
more glazed surface, thus giving architecturally pleasing appearance. Furthermore,
unlike in the captive glazing system both the negative and positive wind loads from the
glass panes are transferred to the mullions through the sealant bites. The sealant bites
are stretched and compressed under the negative and positive wind actions,
respectively.

In this study, the mullion couples given in Table 6.1, which included the facade systems
shown in Figure 6.3, were analysed using the simple FE models with full lateral
restraint. Many parameters, such as the gap between the mullions, lateral restraint, load
distribution profile, ratio of the loads acting on male and female mullions and
intermediate transoms, were assessed in detail in order to gain a comprehensive
understanding of the structural behaviour and capacity of the mullion couples.
Furthermore, full FE models were developed for some selected mullion couples so that
the developed simple and full finite element analysis (FEA) procedures can be further
verified. Finally, the FEA predictions of the mullion couples were compared with the
current design approach in order to provide suitable design guidelines.

6.3 Finite element modelling of mullion couples

The procedure of FE model development and parameters used in both simple and full
FE models were given in Chapter 5. Idealization of the mullion couple, boundary
conditions, loading method, finite element type, mesh details, contacts and constraints,
material properties and the type of finite element analysis procedure were discussed in
detail and suitable approaches were specified. However, a brief summary of the finite
element modelling procedures is given next.

6-5
 Parametric study of mullion couples under negative wind action

Idealization of the mullion couples, glass and gaskets/sealants

Mullions used in facades are thin-walled members. Therefore, shell idealization was
used to model the mullion couples in both the full and simple finite element models.
Similarly, the glass panes in full FE models were modelled using shell elements as
described in Section 5.3.2. In Section 5.4.2, the gaskets were idealized into a simple
geometry and then modelled using solid elements. The same approach was used in this
section in order to idealize and model the irregular shape gasket geometries. The
structural silicone sealant bites, see Figure 6.3(c), were also modelled using solid
elements.

Boundary conditions and loading method

Facade mullions are simply supported at their top and bottom ends, thus their boundary
conditions are symmetric about their mid-span. Load is also symmetric when no
intermediate transom is used. Therefore, half span of the mullion couple was modeled,
which reduced the computational time and resource significantly. Full span of the
mullion couple was modeled in Section 6.5.5 when an intermediate transom was used.
Refer Section 5.2 for more details on the symmetry of the facade system.

Finite element type and mesh

Shell idealization was used in order to model the mullion couples and the glass, thus
S4R finite elements were used. Since the Poisson’s ratio of the gasket or silicone
sealant materials is closer to 0.5, C3D8R solid hybrid element was used to model the
gaskets and sealant bites. Furthermore, the mullion couples and the glass panes were
discretized using a mesh size of 4 mm x 8 mm and 25 mm x 25 mm, respectively. The
mesh size of gaskets or silicone sealant bite were selected so that at least four elements
are present in their cross section dimensions (along the thickness and the width),
whereas 3 mm mesh size was used along their length (i.e., in the direction of the
mullion span). Refer Sections 5.3.6 and 5.4.6 for more details.

Contacts and constraints

It was found that both the general contact and contact pair approaches in ABAQUS
predicted the same structural behaviour and capacity of the mullion couples well (refer
Sections 5.3.7 and 5.4.7 for more details). Therefore, either the general contact or the
contact pair approaches were used in the parametric study of the mullion couples

6-6
 Parametric study of mullion couples under negative wind action

depending on the convergence issues. Furthermore, the frictional coefficient between

aluminium to aluminium and gasket to glass were considered as 0.3 and 0.1,
respectively. Refer Sections 5.3.7 and 5.4.7 for more details on the modelling
techniques of the general and contact pair algorithms.

Material properties

In this parametric study, material properties obtained from standards and literature were
used to model the mullion couples. Mullions are made of aluminium, thus the yield
strength, ultimate strength, density, Young’s modulus and Poisson’s ratio used are 172
MPa, 207 MPa, 2700 kg/m3, 69,300 MPa and 0.33, respectively. The focus of this study
is to investigate the structural behaviour and capacity of the mullion couples. This
means that the actual strength properties of glass, gaskets or structural silicone bites are
not required. However, the Young’s modulus and the Poisson’s ratio of these materials
are necessary for the accurate modelling of mullion couples. Therefore, Young’s
modulus values of 70,000 MPa, 10 MPa and 0.5 MPa were used for glass, gasket and
silicone bites, respectively. Similarly, Poisson’s ratio values of 0.22, 0.4995 and 0.4995
were used for glass, gasket and silicone bites, respectively. Refer Sections 5.3.8 and
5.4.8 for more details on the material properties.

Type of analysis procedure

Different types of static analysis procedures are available in ABAQUS. However, the
quasi-static dynamic implicit analysis procedure was found to be the most suitable
option to analyse the mullion couple behaviour, and this worked well for all the
analyses considered. Therefore, the quasi-static dynamic implicit analysis procedure
was used in this parametric study. Refer Sections 5.3.9 and 5.4.9 for more details on the
analysis techniques considered.

6.4 Comparison of full and simple finite element models

Simple FE model was used in this parametric study since it is more economical than
full FE model in terms of time and resources. However, to further validate the simple
and full finite element model development and the analysis procedures, simple and full
FE models of different facade systems (Figures 6.3(a) to (c)) were modelled and
analysed for comparison. The span of the mullion couples considered in the analyses is

6-7
 Parametric study of mullion couples under negative wind action

3600 mm, which is commonly used in practical applications. The comparisons between
the full and simple FE models of the mullion couples are given next.

Figures 6.4(a) and (b) show that the failure modes predicted by both full and simple FE
models of mullion couple MC5 are the same. That is, the separation of female mullion
from the male mullion prior to lateral-torsional buckling is well predicted. Figure 6.5
shows a comparison of the obtained moment versus displacement curves of the two
types of FE models. It is evident from the curves that the stiffness of the system in both
models are about the same, and the ultimate moment capacity of the mullion couple is
well predicted. Therefore, either full or simple FE model could be used in order to
predict the structural behaviour and capacity of the mullion couple system shown in
Figure 6.3(a).

(a) Full FE model (b) Simple FE model

Figure 6.4: Failure modes predicted by full and simple FE models of mullion
couple MC5

2.5

2.0
Moment (kNm)

1.5

1.0

0.5 Full model

Simple model - UDL & Full restraint
0.0
0 -10 -20 -30 -40
Y-axis displacement (mm)

Figure 6.5: Moment versus displacement curves predicted by full and simple FE
models of mullion couple MC5

6-8
 Parametric study of mullion couples under negative wind action

Mullion couple MC12 comprises gaskets outside and weather seals inside as shown in
Figure 6.3(b). Since weather seals are not designed to carry any loads from the wind
actions, they were not modelled in the full FE model. Under the negative wind action,
the bottom gaskets are effectively acting to transfer the wind load to the mullions
through compression, thus only the outside gaskets were modeled. It is evident from
Figures 6.6(a) and (b) that both the full and simple FE models predicted the structural
behaviour and capacity of the mullion couple in a similar manner to mullion couple
MC5. The stiffness and the ultimate bending capacity values agree reasonably well as
shown in Figure 6.7. Therefore, either full or simple FE model could be used to predict
the structural behaviour and capacity of the mullion couple system shown in Figure
6.3(b).

(a) Full FE model (b) Simple FE model

Figure 6.6: Failure modes predicted by full and simple FE models of mullion
couple MC12

7
6
Moment (kNm)

5
4
3
2
1 Full model
Simple model - UDL & Full restraint
0
0 -5 -10 -15 -20 -25
Y-axis displacement (mm)

Figure 6.7: Moment versus displacement curves predicted by full and simple FE
models of mullion couple MC12

Unlike mullion couples MC5 and MC12, mullion couple MC15 is a double glazed
structurally glazed unitized facade system. In this system, the weather seal, see Figure

6-9
 Parametric study of mullion couples under negative wind action

6.3(c), is not designed to carry any loads similar to mullion couple MC12. However, the
weather seal was modelled in the full FE model since its presence could provide some
degree of rotational restraint, i.e. the mullions will be restrained at two locations (at the
structural silicone sealant bite and weather seal locations). In this case, three different
models were developed: 1). Mullion couple with single glass and only the structural
silicone sealants 2). Mullion couple with double glass, structural silicone sealant bite
and weather seal 3). Only the mullion couple. Restraints at two different locations in a
mullion would provide some rotational restraint, which can give a higher capacity than
the mullion with only one lateral restraint. However, Figures 6.8 (a) and (b) show that
the failure modes from the two different full FE models are the same. Also the stiffness
of the system is similar for both full FE models (Figure 6.10), and the ultimate bending
moment capacity of the full FE model with weather seal is not significantly higher than
that without the seal (approximately 10% variation). Therefore, it is adequate to model
only the structural silicone sealant bite.

Further comparisons were made between the results of full and simple FE models of
mullion couple MC15. Figures 6.9(a) and (b) show that failure modes predicted by both
the full and simple FE models are the same. Also, the stiffness of the system and the
ultimate bending moment capacity are similar as shown in Figure 6.10. In addition, the
simple FE model was analyzed with spring lateral restraint, which was determined as
per Appendix B.3. Figure 6.10 shows that the effectiveness of the lateral restraint to the
mullion couple did not significantly affect the ultimate bending moment capacity.
Therefore, either full or simple FE model could be used in order to predict the structural
behaviour and capacity of the mullion couple system shown in Figure 6.3(c).

(a) Single glazing (b) Double glazing

Figure 6.8: Failure modes predicted by full FE model of mullion couple MC15
with single and double glazing

6-10
 Parametric study of mullion couples under negative wind action

(a) Full FE model (b) Simple FE model

Figure 6.9: Failure modes predicted by full and simple FE models of mullion
couple MC15

5
Moment (kNm)

2
Full model - Double glass
1 Full model - Single glass
Simple model - UDL & Full restraint
0
0 -10 -20 -30 -40 -50
Y-axis displacement (mm)

Figure 6.10: Moment versus displacement curves predicted by full and simple FE
models of mullion couple MC15

It is evident from the moment versus displacement curves and the failure modes
obtained that both the full and simple FE models predict the structural behaviour and
capacity of the mullion couples reasonably well. However, the development and
analysis of full FE models needs more time for idealization and modelling of the facade
components and computer resources than simple FE models. Therefore, simple FE
models will be used in the parametric study to determine the ultimate failure capacities
of the mullion couples used in unitized facade systems.

6-11
 Parametric study of mullion couples under negative wind action

6.5 Effects of different parameters on the structural behaviour and

capacity of mullion couples

6.5.1 Gap between the mullions

In unitized facades, the panels are assembled with a gap between the male and female
mullions of the mullion couple, see Figure 6.11, in order to allow for thermal and other
building movements. This gap could vary depending on the design requirements. In
order to assess the effect of the gap in the structural behaviour and the ultimate bending
capacity of the mullion couples, two types of mullion couples (a captive and a structural
glazing mullion couple) were considered. The gaps between the male and female
mullions considered are 1 mm, 3 mm, 6 mm and 10 mm. Also, two spans (short and
long spans) were considered since it would give a comprehensive understanding of the
effect of gap.

Figure 6.11: Typical gap between male and female mullions

Figures 6.12 (a) and (b) show the moment versus displacement characteristics of
mullion couple MC5 with short and long spans, respectively. Similarly, Figures 6.13 (a)
and (b) show the moment versus displacement characteristics of mullion couple MC15.
Despite the type and the span of the mullion couples, the patterns in the variation of the
ultimate moment capacities are, in general, similar for all the cases. Furthermore, the
tangent (stiffness) of the moment versus displacement curves did not change with the
gaps although the ultimate moment capacities varied. Therefore, the deflection of the
mullion couple was not influenced by the gap between the mullions. The results in
Figures 6.12 and 6.13 show that the ultimate bending moment capacity increases with

6-12
 Parametric study of mullion couples under negative wind action

increasing gap. This is because the larger gap between the mullions will delay the
interaction between their compression flanges.

6 2.5

5 2.0
Moment (kNm)

Moment (kNm)
4
1.5
3
1.0
2 Gap = 1 mm Gap = 1 mm
Gap = 3 mm Gap = 3 mm
1 0.5
Gap = 6 mm Gap = 6 mm
Gap = 10 mm Gap = 10 mm
0 0.0
0 -5 -10 -15 -20 0 -5 -10 -15 -20 -25
Y-axis displacement (mm) Y-axis displacement (mm)
(a) 1800 mm span (b) 4200 mm span

Figure 6.12: Moment versus displacement curves of mullion couple MC5 with
different gaps

10 5.5
9 5.0
8 4.5
7 4.0
Moment (kNm)

Moment (kNm)

6 3.5
3.0
5
2.5
4 2.0
Gap = 1 mm Gap = 1 mm
3 1.5
Gap = 3 mm Gap = 3 mm
2 1.0 Gap = 6 mm
Gap = 6 mm
1 Gap = 10 mm 0.5 Gap = 10 mm
0 0.0
0 -5 -10 -15 -20 -25 -30 0 -10 -20 -30 -40 -50
Y-axis displacement (mm) Y-axis displacement (mm)
(a) 1800 mm span (b) 4200 mm span
Figure 6.13: Moment versus displacement curves of mullion couple MC15 with
different gaps

The maximum differences in the ultimate bending moment capacities of MC5 and
MC15 with short span are about 6% and 8%, while those for the long span are about
17% and 20%, respectively. This shows that the effect of gap on the ultimate bending
moment capacity of the mullion couples increases with the span length. This could be

6-13
 Parametric study of mullion couples under negative wind action

because of the higher slenderness of the longer mullions. Furthermore, when a building
is in service, the gap between the mullions can vary due to the thermal expansion or
contraction of the transoms. Therefore, it is advisable to use a minimum gap (in the case
of negative wind action), which depends on design requirements, when the capacity of
the mullion couple is determined using FEA.

6.5.2 Effectiveness of lateral restraint provided by glass pane

It is possible to consider that the glass pane provides full lateral restraint to the
mullions. However, the shear flexibility of the gaskets in captive glazing mullions or
the sealant bites in structural glazing mullions may not provide adequate lateral restraint
to fully prevent the lateral displacement of the mullions. Figure 6.14, for instance,
shows the application of a spring lateral restraint to the mullions. In Chapter 5, it was
found that the capacity prediction of the mullion couple with full lateral restraint under
negative wind action agreed well with the test results. However, it is still not well
understood whether the effectiveness of the lateral restraint given by the glass panes
influences the ultimate bending moment capacity of the mullion couple under negative
wind action. Therefore, two mullion couples (MC5 and MC15) with 1800 mm and 4200
mm spans were analyzed by varying the spring constant of the lateral restraint from
fully effective to 1/40 N/mm/mm.

Spring lateral
restraint

Figure 6.14: Typical spring lateral restraint under negative wind action

It is evident from Figures 6.15 and 6.16 that the ultimate bending moment capacity of
the mullion couple reduces when the effectiveness of the lateral restraint is reduced.
However, its influence on the stiffness of the mullion couple is insignificant. For the
mullion couple MC5, the reductions in the ultimate bending moment capacities for

6-14
 Parametric study of mullion couples under negative wind action

1800 mm and 4200 mm spans are 22% and 8%, respectively. But in the mullion couple
MC15, the reductions in the ultimate bending moment capacities for 1800 mm and 4200
mm spans are 11% and 17%, respectively. This shows that the effect of lateral restraint
on the ultimate bending moment capacity of the mullion couples could be significant
either for short or long spans. Therefore, it is advisable to use an appropriate reduction
factor that can be used conservatively for any span, when determining the ultimate
bending moment capacity of the mullion couples.

6 2.0

5
1.5

Moment (kNm)
Moment (kNm)

3 1.0
Full restraint Full restraint
2 k=3/2 N/mm/mm k=3/2 N/mm/mm
k=5/16 N/mm/mm 0.5 k=5/16 N/mm/mm
1 k=1/8 N/mm/mm k=1/8 N/mm/mm
k=1/40 N/mm/mm k=1/40 N/mm/mm
0 0.0
0 -5 -10 -15 -20 0 -10 -20 -30
Y-axis displacement (mm) Y-axis displacement (mm)
(a) 1800 mm span (b) 4200 mm span
Figure 6.15: Moment versus displacement curves of mullion couple MC5 with
different lateral restraint levels

9 4.5
8 4.0
7 3.5
Moment (kNm)

Moment (kNm)

6 3.0
5 2.5
4 2.0
Full restraint Full restraint
3 k=3/2 N/mm/mm 1.5 k=3/2 N/mm/mm
2 k=5/16 N/mm/mm 1.0 k=5/16 N/mm/mm
1 k=1/8 N/mm/mm 0.5 k=1/8 N/mm/mm
k=1/40 N/mm/mm k=1/40 N/mm/mm
0 0.0
0 -5 -10 -15 -20 -25 0 -10 -20 -30 -40 -50 -60
Y-axis displacement (mm) Y-axis displacement (mm)
(a) 1800 mm span (b) 4200 mm span
Figure 6.16: Moment versus displacement curves of mullion couple MC15 with
different lateral restraint levels

6-15
 Parametric study of mullion couples under negative wind action

6.5.3 Load distribution profile

Profile of the load that is transferred to a mullion couple depends on the dimensions of
the facade panel. That is, the aspect ratio (AR = panel height (h)/ panel width (b))
defines whether the load distribution profile is triangular or trapezoidal distribution.
When the panel height is relatively larger than the panel width (h >> b means h–b | h),
the load distribution approaches uniform distribution (Figure 6.17(b)). Furthermore, for
a particular span of mullion couples, the load distribution profile could vary depending
on the panel width. This implies that analysis needs to be performed when the panel
width is changed. It is convenient to assume uniform distribution in an analysis since it
is easier to define a uniform load distribution than varying load distribution. Therefore,
analyses were performed in order to investigate the effect of load distribution profiles
on the ultimate bending capacity of mullion couples.

Mullion couple Mullion couple

(a) Trapezoidal distribution (a) Uniform distribution

Figure 6.17: Load distributions

It is evident from Figures 6.18 and 6.19 that the bending moment capacity of the
mullion couple increases as the load distribution profile changes from uniform to
trapezoidal load distribution (i.e. the AR changes from λ to 1), irrespective of the span
or the type of mullion couples. However, the effect on the stiffness of the mullion
couple is insignificant. Among the mullion couples considered, the maximum
difference on the bending moment capacities is 9%, which occurred in the long span of

6-16
 Parametric study of mullion couples under negative wind action

MC5 mullion couple. But, in general the aspect ratio of facade panels could be in the
range of 2 to 3, for which the maximum difference in the bending moment capacities is
4%. Therefore, a uniform load distribution could be considered in the analyses of
mullion couples to predict their bending moment capacities.

6 3.0

5 2.5

Moment (kNm)
Moment (kNm)

4 2.0

3 1.5

2 AR = ∞ 1.0 AR = ∞
AR = 3 AR = 3
1 AR = 2 0.5 AR = 2
AR = 1.5 AR = 1.5
AR = 1 AR = 1
0 0.0
0 -5 -10 -15 -20 0 -5 -10 -15 -20 -25
Y-axis displacement (mm) Y-axis displacement (mm)
(a) 1800 mm span (b) 3600 mm span

Figure 6.18: Moment versus displacement curves of mullion couple MC5 with
different load distribution profiles

9 6
8
5
7
Moment (kNm)

Moment (kNm)

6 4
5
3
4
3 AR = ∞ 2 AR = ∞
AR = 3 AR = 3
2 AR = 2 AR = 2
AR = 1.5 1
1 AR = 1.5
AR = 1 AR = 1
0 0
0 -5 -10 -15 -20 0 -10 -20 -30 -40
Y-axis displacement (mm) Y-axis displacement (mm)
(a) 1800 mm span (b) 3600 mm span

Figure 6.19: Moment versus displacement curves of mullion couple MC15 with
different load distribution profiles

6-17
 Parametric study of mullion couples under negative wind action

6.5.4 Ratio of the loads acting on male and female mullions

It is common to have uniform/equal panel widths in a unitized facade. However, there

are instances where different panel widths are used as shown in Figure 6.20(a). When
different panel widths are used (assume a>b), the load that is transferred to the male and
female mullions at a particular wind pressure will be different as shown in Figure
6.20(b). The analyses performed on the mullion couples so far considered equal loads
on the male and female mullions, thus the effect of the ratio of the loads acting on male
and female mullions is not known. Therefore, analyses were performed on the mullion
couples MC5 and MC15 by considering different male and female mullions’ load
distribution ratios (a:b) of 1:0, 3:1, 2:1, 1:1, 1:2, 1:3 and 0:1. Among these load
distribution ratios, 1:0 or 0:1 refers to the case of load acting only on one mullion.

Load Load
(Fmale) (Ffemale)

(a) Different panel widths (b) Load transferred to the mullion couple

Figure 6.20: Typical loads acting on male and female mullions of a mullion couple

Note: For instance, the loads that are transferred to the male (Fmale) and female (Ffemale)
mullions for a unit length at a pressure P is equal to (P x a / 2) and (P x b / 2),
respectively. This means the load distribution ratio for the male and female mullions of
the mullion couple Fmale : Ffemale = a : b.

It is evident from Figures 6.21 and 6.22 that the failure mode of a mullion couple could
vary depending on the load distribution ratio. That is, mullion couples rotated in the
clockwise direction and failed by lateral-torsional buckling when the load on the female

6-18
 Parametric study of mullion couples under negative wind action

mullion was higher than that on the male mullion. On the other hand, mullion couples
rotated in the anti-clockwise direction and failed by lateral-torsional buckling when the
load on the male mullion was higher than that on the female mullion. Both mullion
couples showed approximately similar patterns of moment versus displacement
behaviour to the different load distribution ratios considered (Figures 6.23 and 6.24).
When the load on the male mullion was increased more than that on the female mullion,
the moment capacity increased (see load distribution ratio 2:1) and then decreased (see
load distribution ratios 1:0 and 3:1). But, when the load on the female mullion was
increased more than that on the male mullion, the moment capacity continuously
decreased. This is because, the female mullion tries to deflect more than the male
mullion due to the higher load transfer on it, thus transferring some load to the top
flange of the male mullion. This will induce higher torsional rotation of the male
mullion causing premature failure since the load that is transferred to the male mullion
is acting towards its shear centre. However, in all the cases the stiffness of the mullion
couples was not affected, which proves that the male and female mullions act together.
These observations show that the ratio of the loads acting on male and female mullions
(or load distribution ratio) significantly influence the structural behaviour and capacity
of mullion couples.

Female
Anti-clockwise mullion
Male rotation
mullion Clockwise
rotation
Male
Female mullion
mullion

(a) Load distribution ratio = 1:2 (b) Load distribution ratio = 2:1

Figure 6.21: Failure modes of mullion couple MC5 with load distribution ratios of
1:2 and 2:1

6-19
 Parametric study of mullion couples under negative wind action

Male
mullion Female
Anti-clockwise mullion
Clockwise rotation
rotation

Female Male
mullion mullion

(a) Load distribution ratio = 1:2 (b) Load distribution ratio = 2:1

Figure 6.22: Failure modes of mullion couple MC15 with load distribution ratios
of 1:2 and 2:1

Most of the mullion couples considered in this study show a failure mode as shown in
Figure 6.21(a) or 6.22(a) under equal loading on male and female mullions. That is,
male mullion pushes the female mullion, thus lateral-torsional buckling failure is
initiated on female mullion. But, mullion couples used in practical applications can be
of any shape, and their failure mode can be of any type. Therefore, the capacity of
mullion couples under different load distribution ratios can be determined using FEA.

7 2.5
6
2.0
5
Moment (kNm)

Moment (kNm)

4 1.5
a:b=1:0
3 a:b=1:0 a:b=3:1
a:b=3:1 1.0
a:b=2:1
2 a:b=2:1
a:b=1:1 a:b=1:1
a:b=1:2 0.5 a:b=1:2
1 a:b=1:3 a:b=1:3
a:b=0:1 a:b=0:1
0 0.0
0 -5 -10 -15 -20 -25 0 -10 -20 -30 -40
Y-axis displacement (mm) Y-axis displacement (mm)
(a) 1800 mm span (b) 4200 mm span

Figure 6.23: Moment versus displacement curves of mullion couple MC5 with
different load distribution ratios (a:b)

6-20
 Parametric study of mullion couples under negative wind action

10 6
9
5
8
Moment (kNm)
7

Moment (kNm)
4
6
5 a:b=1:0 3
a:b=1:0
4 a:b=3:1 a:b=3:1
a:b=2:1 2 a:b=2:1
3 a:b=1:1 a:b=1:1
2 a:b=1:2 1 a:b=1:2
1 a:b=1:3 a:b=1:3
a:b=0:1 a:b=0:1
0 0
0 -5 -10 -15 -20 -25 0 -20 -40 -60 -80
Y-axis displacement (mm) Y-axis displacement (mm)
(a) 1800 mm span (b) 4200 mm span
Figure 6.24: Moment versus displacement curves of mullion couple MC15 with
different load distribution ratios (a:b)

6.5.5 Intermediate transoms

Intermediate transoms are sometimes used in order to achieve a higher capacity by

reducing the effective length of longer mullions. It is common to have a single
intermediate transom as shown in Figure 6.25(a). However, there are instances where
more than one transom is used, for e.g. as shown in Figure 6.29(a). In this section, the
effects of intermediate transoms on the structural behaviour and capacity of mullion
couples are given.

a. Single intermediate transom

Figures 6.25(a) and (b) show the load tributary and the load that is transferred to the
mullion couple with a single transom within its span. The point load “P” refers to the
load from the transom, which is shown in light blue in Figure 6.25(a). In FE modelling,
the presence of transom was represented by providing appropriate lateral restraints to
the mullions at the transom locations. The actual load transfer, which is shown in Figure
6.25(b), was used in FE modelling. Furthermore, the load transfer to the mullion couple
is not symmetrical about the mid-span, thus the full span of the mullion couple was
modelled. In addition, equal panel width of 1200 mm was considered in this case, thus
the magnitudes of the load transferred to the male and female mullions are similar at a
particular wind pressure loading. Providing intermediate transoms reduces the unbraced
length (or effective length) of the mullion couples. Therefore, analyses were performed

6-21
 Parametric study of mullion couples under negative wind action

on the mullion couple MC5 of 3600 mm span, considering different transom locations
L1 of 900, 1200 and 1800 mm (Figure 6.25).
These loads act as point
loads on mullions

Mid-span
Transom

(a) Load transfer from glass panes (b) Load and bending moment diagrams

Figure 6.25: Load and moment profiles of mullion couples with one transom

Figure 6.26 shows that the mullion couple failed by lateral-torsional buckling when the
lateral restraint was provided at 900 mm. The same failure mode was obtained when the
transom locations were changed (at 1200 mm and 1800 mm). In addition, the failure
occurred at a location where the maximum moment (Mmax) is present. However, this
does not mean that failure will always occur where the maximum moment is present. It
is evident from Figure 6.27 that the ultimate failure capacity of the mullion couple
increases as the unbraced length (L2) reduces. Therefore, it is obvious that the reduction
in the unbraced length would increase the ultimate capacity of the mullion couples.
Furthermore, there are slight variations in the stiffness of the mullion couple when
transom location is changed (see Figure 6.27). This could be due to the fact that when
the transom is moved towards the mid-span, the point load “P” acting on the mullion
couple increases, and the location of the point load also moves towards the mid-span.
That is, variation in the profile of the distributed load and the magnitude of the point
load (see Figure 6.25(b)) could have resulted in such variations in stiffness (deflection
of beams may differ under the same magnitude of distributed and point loads).

6-22
 Parametric study of mullion couples under negative wind action

Support
L1 = 900 mm
Transom location

h - L1 = 2400 mm

| 1990 mm

Support

Figure 6.26: Failure mode of mullion couple MC5 with an intermediate transom at
900 mm

10
9
Reaction = R1+R2 (kN)

8
7
6
5
4
3 L1 = 0 mm
2 L1 = 900 mm
1 L1 = 1200 mm
L1 = 1800 mm
0
0 -10 -20 -30 -40
Y-axis displacement (mm)
Figure 6.27: Mullion couple MC5 with and without an intermediate transom

Note: In the above figure, total applied load (total reaction = R1+R2) versus
displacement curves were plotted instead of moment versus displacement. It is because,
when the intermediate transom location is changed, the locations where maximum
moment and deflection occur also change, which is difficult to decide during the
modelling phase. Therefore, total applied load and the corresponding mid-span
deflection were used in all the analyses for comparison.

6-23
 Parametric study of mullion couples under negative wind action

Moment gradient is another fact that can influence the ultimate bending moment
capacity of a beam. In ADM (AA, 2015), the factor Cb accounts for the effect of
moment gradient, where the elastic lateral-torsional buckling moment capacity Me is
proportional to the factor Cb (refer Appendix C.2). When transom is moved from
support to mid-span, the bending moment distribution of Segment L2 approaches
approximately a triangular shape. This means the factor Cb will increase from 1.14 to
1.35. In order to investigate this, the capacities obtained from the above analyses were
compared with the respective simply supported segment capacities. It is evident from
Table 6.2 that the use of transom increases the capacity of the mullion couple
significantly, but Table 6.3 shows that the effect of moment gradient is not significant
as expected. This may be due to the presence of torsional forces, and therefore, it is
advisable not to consider the factor Cb to be more than 1.0 in designs.

Load diagram

Bending moment diagram

Figure 6.28: SPACE GASS (SG, 2014) analysis result for the unit pressure loading

Table 6.2: Moment capacity predictions for mullion couple MC5 with and without
an intermediate transom

Total span h = 3600 mm

Length of the larger segment – L2 (mm) 2700 2400 1800
Moment capacity of the mullion couple with an intermediate
3.51 3.79 5.20
transom at L2 from bottom end (Mtr)
Moment capacity of the mullion couple without any
2.21
intermediate transom (M)
Ratio of the moment capacities (Mtr / M) 1.59 1.71 2.35
Note: The magnitudes of the ultimate moments from the mullion analyses were
obtained using the unit pressure output, which was obtained using SPACE GASS (SG,
2014) as shown in Figure 6.28.

6-24
 Parametric study of mullion couples under negative wind action

Table 6.3: Moment gradient factor Cb of mullion couple MC5 based on FEA and
ADM (AA, 2015)

Total span h = 3600 mm

Based on FEA:
Length of the larger segment – L2 (mm) 2700 2400 1800
Moment capacity of the mullion couple with an
3.51 3.79 5.20
intermediate transom at L2 from bottom end (Mtr)
Moment capacity of the mullion couple with a simply
3.32 3.89 5.09
supported span length of L2 (Mss,L2)
Ratio of the moment capacities (Mtr / Mss,L2) 1.06 0.97 1.02

Based on ADM (AA, 2015):

Moment values of a beam Moment at h/4 - MA 1.11 1.01 0.78
with an intermediate transom Moment at h/2 - MB 1.75 1.66 1.39
at L2 based on linear elastic
beam analysis using SPACE Moment at 3h/4 - MC 1.84 1.73 1.75
GASS software (SG, 2014) Maximum moment - Mmax 1.87 1.87 1.94
Moment gradient factor Cb in accordance with ADM
(AA, 2015)
1.14 1.20 1.35
ͳʹǤͷ‫ܯ‬௠௔௫
‫ܥ‬௕ ൌ
ʹǤͷ‫ܯ‬௠௔௫ ൅ ͵‫ܯ‬஺ ൅ Ͷ‫ܯ‬஻ ൅ ͵‫ܯ‬஼

b. Two or more intermediate transoms

The number of intermediate transoms and their locations within the span of a mullion
couple depend on the design requirements. In the analyses of mullion couples with two
transoms, transoms were located at an equal distance from the top and bottom ends of
mullion couples, (see Figure 6.29(a)), since it will be the worst scenario. This is
because, the moment gradient for the mid-segment, (L2 in Figure 6.29(a)), will approach
uniform bending, which is the critical case for a beam subject to bending. In addition,
there is symmetry in transom arrangement about the mid-span since the transoms are
located at the same distance from the top and bottom ends of mullion couples (Figure
6.29(b)). Therefore, it is sufficient to model half the span of the mullion couple in FEM.
Furthermore, the results of mullion couples with two transoms could then be compared
with the results of the respective mullion couples with single transom in order to get a
better understanding of the effects of different bending moment gradients.

6-25
 Parametric study of mullion couples under negative wind action

Mullion couples MC5 and MC15 with 4200 mm overall span were considered. Such a
large span was considered since providing two transoms will make the segment lengths
too short. Five transom locations (1200 mm, 900 mm, 600 mm, 300 mm and 0 mm)
were considered, where the corresponding mid-segment lengths (L2) are 1800 mm,
2400 mm, 3000 mm, 3600 mm and 4200 mm, respectively. Figures 6.30(a) and (b)
show that the mullion couples failed in lateral-torsional buckling, which is the same for
the mullion couple with single transom as shown in Figure 6.26. The moment versus
displacement curves of the mullion couples are given in Figures 6.31(a) and (b). The
capacity of the mullion couple increases as the mid-segment length L2 decreases.
Therefore, it can be stated that reduced segment length in a mullion couple will result in
a higher capacity. In addition, a comparison was made between the capacities of a
mullion couple with two transoms and the capacity of a simply supported mullion
couple with a span that is equal to the respective mid-segment length. It is expected that
the ratio of the moment capacities should be approximately 0.877. However, it is
evident from Table 6.4 that the actual ratios from FEA varies significantly, thus it
cannot be determined theoretically. This may be due to the couple interaction and the
presence of torsion on the male and female mullions.

Transom

(a) Load transfer from glass panes (b) Load transferred to the mullion couple

Figure 6.29: Load and moment distribution profiles of mullion couples with two
intermediate transoms
\

6-26
 Parametric study of mullion couples under negative wind action

Transom location Transom location

Male Male
mullion
mullion
Support Support

Female Female
mullion mullion
Mid-span Mid-span

(a) Mullion couple MC5 (b) Mullion couple MC15

Figure 6.30: Failure modes of mullion couples MC5 and MC15 with two
intermediate transoms

4.5
4.0
3.5
Moment (kNm)

3.0
2.5
2.0
L2 = 4200 mm
1.5 L2 = 3600 mm
1.0 L2 = 3000 mm
0.5 L2 = 2400 mm
L2 = 1800 mm
0.0
0 -10 -20 -30 -40
Y-axis displacement (mm)

(a) Mullion couple MC5

9
8
7
Moment (kNm)

6
5
4
L2 = 4200 mm
3 L2 = 3600 mm
2 L2 = 3000 mm
L2 = 2400 mm
1
L2 = 1800 mm
0
0 -20 -40 -60 -80
Y-axis displacement (mm)

(b) Mullion couple MC15

Figure 6.31: Moment versus displacement curves of mullion couples MC5 and
MC15 with two intermediate transoms

6-27
 Parametric study of mullion couples under negative wind action

Table 6.4: Moment gradient of mullion couples with two intermediate transoms

Total span h = 4200 mm

Mullion couple MC5 MC15
Length of the larger segment – L2 (mm) 2400 1800 2400 1800
Mullion with intermediate transom (Mtr) 2.84 3.97 7.18 8.16
Simply supported mullion with a span of L2 (Mss,L2) 3.89 5.09 7.09 8.21
Ratio of the moment capacities (Mtr/ Mss,L2) 0.73 0.78 1.01 0.99

The effect of using one and two intermediate transoms on the ultimate bending moment
capacity of the mullion couples was separately assessed. However, a comparison was
made between the results of mullion couples with one and two transoms since it would
give a better understanding of the effects of moment gradient on the bending capacity of
mullion couples. In order to assess this, a 3600 mm total span (h) mullion couple was
analysed with one transom at 1200 mm, and two transoms at 600 mm from both ends.
The respective larger segment length of the mullion couple under both cases will be
2400 mm. Figure 6.32 shows that the moment capacity of the mullion couple with
single transom is higher than the other one (about 18%). In fact, this is justifiable based
on their moment gradient, and therefore, a segment that is present in the end span would
achieve higher capacity than that present in the mid-span.

4.5
4.0
3.5
Moment (kNm)

3.0
2.5
2.0
1.5
1.0
One transom, L1 = 1200 mm
0.5
Two transom, L1 = 600 mm
0.0
0 -10 -20 -30 -40
Y-axis displacement (mm)

Figure 6.32: Moment versus displacement curves of mullion couple MC5 with one
and two intermediate transoms

6-28
 Parametric study of mullion couples under negative wind action

6.6 Comparison of FEA and design rule predictions

6.6.1 General

Effects of various criteria on the structural performance and capacity of some mullion
couples were assessed through numerous FE simulations. It showed that many factors,
such as load distribution profile, lateral restraint, gaps between the mullions, ratio of the
loads acting on male and female mullions and intermediate transoms, influence the
bending moment capacity of the mullion couples. Currently, the bending moment
capacities of the mullion couples are determined using available aluminium design
standards such as ADM (AA, 2015) and AS 1664 (SA, 1996), which do not give any
provisions to account for the aforementioned factors. AS 1664 (SA, 1996) is technically
equivalent to the outdated version of the Aluminium Association publication ADM
2010 (AA, 2010). ADM 2015 (AA, 2015) is the latest version, and therefore, only the
ADM (AA, 2015) was considered in this study. Furthermore, the capacity of the
mullion couple is determined by adding the individual bending moment capacities of
the male and female mullions, which are determined separately using the design
standards. The accuracy/efficiency of the currently used design approach of mullions is
not known. Therefore, the predictions from ADM (AA, 2015) are compared with those
from finite element analyses (FEA).

6.6.2 FEA and Design rule predictions

Section moment capacities of the mullion couples need to be determined in order to

calculate their member capacities. To calculate the section moment capacities, local
buckling checks need to be performed. It is evident from Figure 6.33 that mullion
sections are of complex shaped, and it is difficult to categorize the individual elements
of the sections for element slenderness calculations using ADM (AA, 2015). Therefore,
analysis using CUFSM (Li and Schafer, 2010) is considered as an accurate method to
determine the local buckling stress of the entire cross-section (Kesawan and
Mahendran, 2019). Then this can be used to determine the section moment capacity of
the section using the direct strength method given in ADM (AA, 2015).

6-29
 Parametric study of mullion couples under negative wind action

Figure 6.33: CUFSM model of mullion couple MC10

The mullion sections considered in this study are commercially used sections. They can,
in general, achieve yielding (or yield capacity) without any local buckling. However, to
verify this, the mullion couples MC10 and MC12 were analysed using CUFSM
software in order to determine the buckling stresses of their cross-sections. The male
mullion of mullion couple MC10 has a longer compression element than most of the
other mullion sections and the male and female mullions of mullion couple MC12 have
a deeper web element than the other mullions. It is evident from Figures 6.34 and 6.35
that the local buckling stress of the mullion sections are higher than the yield stress
(load factor is greater than 1.0). This means that the section moment capacity of the
mullion couples can be considered as their yield capacity. Therefore, yield capacities
were considered when determining the member moment capacity of the mullion
couples.

(a) Male mullion

Figure 6.34: CUFSM analysis results of mullion couple MC10

6-30
 Parametric study of mullion couples under negative wind action

(b) Female mullion

Figure 6.34: CUFSM analysis results of mullion couple MC10

(a) Male mullion

(b) Female mullion

Figure 6.35: CUFSM analysis results of mullion couple MC12

Mullion couples (MC1 to MC20) given in Table 6.1 were analyzed using the simple FE
models with simple support conditions at the ends. It was shown in Section 6.5 that
many factors influence the bending moment capacity of the mullion couple. Since there
6-31
 Parametric study of mullion couples under negative wind action

is no benchmark for these factors, the worst case was considered in FEA. That is, for
instance, the gaps between the mullions were selected between 1 mm and 6 mm
depending on the mullion couple, which is smaller than the actual gap giving
conservative predictions. Furthermore, uniform load distribution was considered since it
gives conservative predictions as shown in Figure 6.19. However, a fixed value for the
effectiveness of the lateral restraint cannot be assumed, thus the mullion couples were
analyzed with full lateral restraint. Then the effect of lateral restraint can be included by
using suitable reduction factors, which could vary among mullion couples. Finally, the
FEA predictions given in Table 6.5 were compared with the nominal ADM (AA, 2015)
predictions given in the same table. Refer Appendix C.2 for sample calculations of the
design rule predictions.

Table 6.5: Moment capacity predictions of FEA and ADM (AA, 2015)

FEA prediction (MFEA) ADM prediction (MADM)

Mullion Span (mm) Span (mm)
couple
1800 2400 3000 3600 4200 1800 2400 3000 3600 4200
MC1 3.36 2.39 1.73 1.38 1.17 2.16 1.36 0.97 0.74 0.6
MC2* 1.32 1.01 0.80 0.68 N/A 2.29 1.45 1.03 0.79 0.65
MC3* 3.22 2.69 2.30 2.00 1.78 5.27 4.41 3.38 2.63 2.14
MC4 3.98 3.14 2.60 2.27 2.02 5.37 4.03 2.9 2.23 1.81
MC5 5.09 3.89 2.86 2.21 1.82 3.41 2.07 1.44 1.08 0.86
MC6 4.65 3.39 2.57 2.07 1.74 3.77 2.29 1.6 1.22 0.97
MC7 6.15 4.59 3.30 2.55 2.12 5.39 3.67 2.5 1.87 1.47
MC8 6.16 4.16 2.72 2.05 1.74 5.86 4.06 2.78 2.07 1.63
MC9 7.06 5.59 4.17 3.25 2.73 6.77 5.13 3.77 2.82 2.23
MC10* 1.67 1.20 1.00 0.90 0.86 6.68 4.6 3.15 2.35 1.85
MC11 13.31 10.68 8.37 6.47 5.29 13.44 10.42 8.3 6.17 4.86
MC12 10.26 9.36 7.46 5.91 4.86 8.47 5.69 3.9 2.9 2.28
MC13 3.61 3.03 2.50 2.03 1.67 2.64 2.08 1.55 1.23 1.01
MC14 6.29 5.26 4.23 3.32 2.71 4.15 2.88 2.01 1.52 1.22
MC15 8.21 7.09 6.03 5.02 4.17 5.4 4.14 2.97 2.29 1.87
MC16 14.24 12.50 10.93 9.49 8.14 10.63 8.54 6.7 5.12 4.11
MC17 13.78 12.02 10.45 8.97 7.59 10.15 8 6.17 4.69 3.75
MC18 15.60 13.62 11.99 10.55 9.25 12.07 9.69 8.06 6.33 5.04
MC19 17.29 15.24 13.38 11.65 10.01 13.04 10.46 8.77 6.8 5.38
MC20 11.58 9.80 8.16 6.64 5.40 9.02 6.68 4.65 3.51 2.79
*ADM predictions for these mullion couples are significantly higher than the FEA predictions
(MADM / MFEA ratio is greater than 1.0).

6-32
 Parametric study of mullion couples under negative wind action

It is evident from Table 6.6 that the ADM predictions are, in general, conservative than
the FEA predictions for the mullion couples that show interaction (see Figure 6.36(a)).
For the most likely spans (3000 mm or 3600 mm) the FEA predictions show about 40 to
100% increment than the ADM predictions. But, this increment is for the full lateral
restraint condition, and thus it did not account for the reduction caused by the
effectiveness of lateral restraint. However, for example, it should be noted that the
mullion couples MC5 and MC15 showed approximately 56% increment even with the
highest reduction of 22% they showed (refer Section 6.5.2). Furthermore, they showed
the same reduction for a low level lateral restraint condition considered in this study,
but the actual restraint could be higher. Therefore, the ADM predictions are more
conservative than the FEA predictions although it gives reasonable predictions for short
spans (1800 mm).

Table 6.6: Comparison of FEA and ADM (AA, 2015) predictions of mullion
couples that show interaction

Mullion MADM / MFEA

Max Min Avg. COV
couple 1800 2400 3000 3600 4200
MC1 0.64 0.57 0.56 0.54 0.51 0.64 0.51 0.56 0.09
MC5 0.67 0.53 0.50 0.49 0.47 0.67 0.47 0.53 0.15
MC6 0.81 0.68 0.62 0.59 0.56 0.81 0.56 0.65 0.15
MC7 0.88 0.80 0.76 0.73 0.69 0.88 0.69 0.77 0.09
MC8 0.95 0.98 1.02 1.01 0.94 1.02 0.94 0.98 0.04
MC9 0.96 0.92 0.90 0.87 0.82 0.96 0.82 0.89 0.06
MC11 1.01 0.98 0.99 0.95 0.92 1.01 0.92 0.97 0.04
MC12 0.83 0.61 0.52 0.49 0.47 0.83 0.47 0.58 0.25
MC13 0.73 0.69 0.62 0.61 0.60 0.73 0.60 0.65 0.09
MC14 0.66 0.55 0.48 0.46 0.45 0.66 0.45 0.52 0.17
MC15 0.66 0.58 0.49 0.46 0.45 0.66 0.45 0.53 0.17
MC16 0.75 0.68 0.61 0.54 0.50 0.75 0.50 0.62 0.16
MC17 0.74 0.67 0.59 0.52 0.49 0.74 0.49 0.60 0.17
MC18 0.77 0.71 0.67 0.60 0.54 0.77 0.54 0.66 0.14
MC19 0.75 0.69 0.66 0.58 0.54 0.75 0.54 0.64 0.13
MC20 0.78 0.68 0.57 0.53 0.52 0.78 0.52 0.62 0.18
Max 1.01 0.98 1.02 1.01 0.94
Min 0.64 0.53 0.48 0.46 0.45
Avg. 0.79 0.71 0.66 0.62 0.59
COV 0.14 0.20 0.26 0.28 0.27

Table 6.7 shows that ADM predictions for the mullion couples in which the male and
female mullions behave individually, see Figure 6.36(b), are significantly higher than
those from FEA predictions. This means that the ADM predictions are unsafe for such

6-33
 Parametric study of mullion couples under negative wind action

mullion couples. It could be due to the effect of the individual rotation caused by
torsion. That is, in all the other mullion couples given in Table 6.6, the male and female
mullions rotated towards each other to lean at their unrestrained compression flange
before rotating together and then failure. Therefore, it is advisable to use validated FEA
to predict the capacity of such mullion couples. Furthermore, Tables 6.6 and 6.7 show
that COV of the ADM predictions varies significantly between mullion couples than
spans. This means that using the current design approach to determine the moment
capacity of mullion couples under negative wind action could be risky if the section
characteristics of the mullions are not considered (load and shear centre locations).

Table 6.7: Comparison of FEA and ADM (AA, 2015) predictions of mullion
couples in which male and female mullions behave individually

MADM / MFEA
Mullion Span (mm) Max Min Avg. COV
couple
1800 2400 3000 3600 4200
MC2 1.73 1.44 1.29 1.16 N/A 1.73 1.16 1.41 0.17
MC3 1.64 1.64 1.47 1.32 1.20 1.64 1.2 1.45 0.13
MC4 1.35 1.28 1.12 0.98 0.90 1.35 0.9 1.13 0.17
MC10 4.00 3.83 3.15 2.61 2.15 4 2.15 3.15 0.25
Max 4.00 3.83 3.15 2.61 2.15
Min 1.35 1.28 1.12 0.98 0.9
Avg. 2.18 2.05 1.76 1.52 1.42
CoV 0.56 0.58 0.53 0.49 0.46

Male and female mullions of mullion couples MC2, MC3, MC4 and MC10 did not
show any interaction. That is, in these mullion couples the shear centre lies inside the
line of the action of the load, thus resulting in outward rotation (see Figure 6.36(b)).
Furthermore, mullions behave individually due to the outward rotation, and hence the
capacity of the mullion couple cannot be the summation of the male and female
mullions’ capacities (Figure 6.37(a)). This is because, when one mullion reached the
failure load, the other mullion might not have reached the failure state as shown in
Figure 6.37(b). In this case, the capacity of the mullion couple will depend on the
weaker mullion. In fact, the capacity of the mullion couple is equal to two times the
capacity of the female mullion as shown in Figure 6.37(a). In addition, the capacity
based on ADM for these mullion couples resulted in unsafe predictions as given in
Table 6.7. Therefore, the moment capacity of the mullion couples that show similar

6-34
 Parametric study of mullion couples under negative wind action

characteristics as described above cannot be determined using ADM, and thus FEA
based solution shall be used.

Finally, one of the main objectives of this investigation was to assess the effectiveness
of the lateral restraint given by leaning action of the male and female mullions. We
found in this study that the leaning action is not an effective interaction since either the
female mullion separated from the male mullion after certain loading for all the mullion
couples given in Table 6.6 or the mullion couples given in Table 6.7 did not show any
interaction. Furthermore, Figure 6.38 reveals that the capacity of the mullion couples
reduces as the span is increased. It should be, otherwise, constant if the lateral restraint
given by the leaning action was effective. Therefore, unless a validated FEA is used the
interaction of the male and female mullions cannot be considered in mullion designs.

Line of action of the load

Line of action of the load
Leaning and
rotating together Individual rotation

(a) Mullions showing interaction (b) Mullions behaving individually

Figure 6.36: Combined and individual behaviour of male and female mullions

6-35
 Parametric study of mullion couples under negative wind action

3.0 1.8
1.6
2.5
1.4
Moment (kNm)

Moment (kNm)
2.0 1.2
1.0
1.5
0.8
1.0 0.6
Mullion couple 0.4 Male mullion
0.5
2 x Female mullion 0.2 Female mullion
0.0 0.0
0 -20 -40 -60 0 -20 -40 -60
Y-axis displacement (mm) Y-axis displacement (mm)
(a) Mullion couple (b) Individual mullions
Figure 6.37: Moment versus displacement curves of mullion couple MC3

12
Mullion couple MC5
Moment capacity (kNm)

10 Mullion couple MC12

Mullion couple MC15
8
6
4
2
0
1800 2400 3000 3600 4200
Span (mm)
Figure 6.38: Moment capacity predictions of FEA versus span of mullion couples

6.6.3 Capacity charts for mullion couples under negative wind action

It was found that many factors influence the structural behaviour and capacity of
mullion couples. It is difficult to incorporate the effects of these factors into the current
design rules since some of them did not show a specific trend. For instance, when the
spring constant of the lateral restraint was varied mullion couple MC5 showed a higher
reduction in the capacity for 1800 mm span, whereas it was for 4200 mm span of
mullion couple MC15. This means that the effect of this factor can be significant for
any span (no specific trend with span length), and thus cannot be incorporated into the
design rules. Therefore, at this stage developing capacity charts using the validated FE
models is considered as a key solution.

6-36
 Parametric study of mullion couples under negative wind action

Based on FEA results, suitable capacity charts can be developed for each mullion
couple so that they can be used directly in design. For example, Figure 6.39(a) shows a
capacity chart developed for the mullion couple MC5 with and without an intermediate
transom. To develop this chart analyses were performed using simple FE model of the
mullion couple with full lateral restraint and no transom (TSR = 1 as shown in Figure
6.39(a)), where the spans considered varied between 1800 mm and 4200 mm in an
interval of 600 mm. Then the results obtained from the FEA were converted to
uniformly distributed load considering a unit panel width. It should be noted that the
effects of gap and load distribution profile are already included in this chart by using
the minimum gap (given in Table 6.1) and uniform load distribution profile in FEM.
But, the effect of lateral restraint condition is incorporated using a separate reduction
factor kelr as shown in Figure 6.39(b). The effect of the ratio of the loads acting on male
and female mullions (load distribution ratio) should be considered as a specific case
since it cannot be incorporated into the chart by using a conservative approach. In
addition to these a suitable capacity reduction factor (I) should be used by designers to
obtain the design capacity.

(a) Capacity versus span curves

Figure 6.39: Capacity chart for mullion couple MC5 under negative wind action

6-37
 Parametric study of mullion couples under negative wind action

(b) Reduction factors and conditions

Figure 6.39: Capacity chart for mullion couple MC5 under negative wind action

It was found that the effect of moment gradient is not significant. This means, for
instance, the capacity of a mullion couple with a span of 3600 mm and a transom at its
mid-span (1800 mm) is equal to the capacity of the mullion couple with a span that is

6-38
 Parametric study of mullion couples under negative wind action

equal to the larger segment length (1800 mm). This approach was used to develop other
curves that give the capacity of the mullion couple with a transom at any location
(Figure 6.39(a)). In addition to this, serviceability limit state capacity curves can also be
plotted as shown in the same figure so that the critical failure criterion can be easily
identified in designs. A similar approach can be used to develop capacity charts for any
case and mullion couple systems to make the design process easier for the facade
industry. Refer Appendix C.4 for more capacity charts developed for some other
mullion couples investigated in this chapter.

6.7 Conclusion

Full and simple FE models of captive and structural glazing mullion couples were
developed and analyzed, and the results were compared to further validate the
developed full and simple finite element modelling approaches. This comparison
showed that both the full and simple FE models predicted the structural behaviour and
capacities of mullion couples reasonably well. However, considering the time and
computing resources, the simple FE model was chosen for the parametric study. Then
the effect of many parameters, such as gap, lateral restraint, load distribution profile,
load distribution ratio and intermediate transoms, were investigated in order to gain a
comprehensive understanding of the structural behaviour and capacity of mullion
couple systems under negative wind action. Thereafter, numerous finite element
analyses were conducted on commercially used mullion couples. In total 20 mullion
couples, consisting of captive and structural glazing mullions, were analyzed using
simple FE models. Then the results were compared with corresponding ADM (AA,
2015) predictions.

Firstly, the influence of gap (for the practical range) and load distribution profile were
found to be less significant than the other factors. The capacity of the mullion couple
increased when the possible critical values of the factors were used. This means that the
effects of gap and load distribution profile are on the conservative side, and hence they
can be incorporated into the FE models so that their effects will be accounted for in the
analysis results. Further, the effects of lateral restraint and load distribution ratio on the
bending moment capacity of the mullion couples were found to be significant, and
therefore they need to be treated separately. That is, the effect of lateral restraint is

6-39
 Parametric study of mullion couples under negative wind action

incorporated as a reduction factor, whereas the effect of load distribution ratio can be
treated as a special case.

The use of one intermediate transom increased the capacity of the mullion couples
significantly. However, no significant advantage was observed due to the moment
gradient. Similarly, the use of two intermediate transoms increased the capacity, but its
effect was less than that with one intermediate transom. This is because when two
intermediate transoms are used, the mid-segment approaches the case of uniform
moment distribution. Furthermore, the moment gradient for the mullion couple with
two transoms varied significantly, and therefore, its effect cannot be determined using a
theoretical approach.

Comparisons between ADM (AA, 2015) and FEA predictions showed that the ADM
predictions are conservative for the mullions when their shear centre was located
outside the line of action of loading. That is, for the mullion couples in which male and
female mullions interact together by leaning action, ADM (AA, 2015) gave
conservative predictions. But ADM (AA, 2015) gave unsafe predictions for the
mullions when their shear centre was located inside the line of action of loading.
Overall, the current design rules in ADM (AA, 2015) could provide either conservative
or unconservative predictions for the mullion couples under negative wind action.
Furthermore, the interaction of the male and female mullions was found to be
ineffective since either female mullion separated after certain loading or male and
female mullions rotated individually. Therefore, interaction of the mullions cannot be
considered in mullion design unless a validated FEA is used.

6.8 Recommendations

Based on the parametric study reported in this chapter, the following recommendations
are made in relation to the structural design of mullion couples under negative wind
action.

a. Mullion couples with shear centre outside the line of action of loading:
The capacity prediction based on design standards for such mullion couples was
conservative, thus these mullion couples can be designed safely in accordance with
the standards. However, if the mullion couples need to be designed economically,
finite element analysis based solutions shall be used. But suitable FEA approaches

6-40
 Parametric study of mullion couples under negative wind action

should be followed by taking conservative assumptions for the gap (minimum gap)
and load distribution profile (uniform profile) during the development of FE
models.

b. Mullion couples with shear centre inside the line of action of loading:
The capacity prediction based on design standards for such mullion couples was
unconservative, thus these mullion couples cannot be designed safely in accordance
with the standards. However, finite element analysis based solutions can be used by
following the recommendation given in (a) above.

c. Mullion couples with one intermediate transom:

The capacity of a mullion couple is increased significantly when an intermediate
transom is used. In this case, the current design standards can be used to design
them conservatively. However, the benefits of moment gradient cannot be
considered. If the mullion couples need to be designed economically, finite element
analysis based solutions shall be used. But suitable FEA approaches should be used
by taking conservative assumptions for the gap, load distribution profile and
boundary conditions during the development of FE models.

d. Mullion couples with two intermediate transoms:

Similar to (c) above, the capacity of a mullion couple is increased significantly
when two intermediate transoms are used. However, the effect of moment gradient
obtained from FEA did not show any correlation with theoretical prediction
(predictions from FEA was either higher or lower than the theoretical prediction).
This means that the real effect of moment gradient cannot be predicted using the
design standards. Therefore, suitable FEA approaches can be used by following the
recommendation given in (c) above.

6-41
This Page is Intentionally Left Blank
 Parametric study of mullion couples under positive wind action

Chapter 7

7 PARAMETRIC STUDY OF MULLION COUPLES

WITHOUT CLIP OR HOLLOW SYSTEMS UNDER
POSITIVE WIND ACTION
7.1 General

In Chapter 6, a parametric study of mullion couples under negative wind action was
performed using simple finite element (FE) models. The effects of many different
parameters on the structural behaviour and capacity of mullion couples were
investigated. This chapter presents a parametric study of the same mullion couples
under positive wind action conducted using simple FE models. The factors, such as gap
between the mullions, effectiveness of lateral restraint and different levels of loads on
male and female mullions, were investigated in order to assess their effects on the
structural behaviour and capacity of the mullion couples under positive wind action.
Thereafter, the yield moment capacities of the mullion couples were compared with the
predictions obtained from the FEA of the respective mullion couples with full lateral
restraint in order to assess the validity of currently used design assumption. Finally, the
bending moment capacities predicted using ADM (AA, 2015) were compared with the
FEA predictions. Based on the above comparisons suitable recommendations are given
for the design of aluminium mullion couples used in facades under positive wind action.

7.2 Mullion sections and influential parameters

In Table 7.1, the details of the mullion couples and the gap between the male and female
mullions used in FEA are given. It should be noted that the leg of the male mullion of
the mullion couples MC1, MC2, MC3, MC5 and MC10 was removed since most of the
mullion sections used in the industry do not have this leg and the purpose of this leg is
to provide weather seal (refer Section 4.3 of Chapter 4 for more details). In fact, this leg
would interlock with female mullion during the positive wind action, which is not
reliable when the gap between the mullions is increased as shown in Figure 7.1. The
section properties of the mullion couples are given in Appendix C.1.

The effects of the gap between the mullions, lateral restraint, ratio of the loads acting on
male and female mullions, load distribution profile and intermediate transoms, were

7-1
 Parametric study of mullion couples under positive wind action

assessed for the negative wind action in Chapter 6. Among these factors, the effects of
load distribution profile and intermediate transom would be the same for both wind
actions, thus they were not investigated again under positive wind action. This is
because these factors do not affect the male and female mullion assembly. Therefore,
the effects of the remaining factors were investigated under the positive wind action.

Table 7.1: Details of mullion couples

Name Type of Depth (mm) Gap
Glazing (mm)
MC1 102 3
MC2 102 4
MC3 102 6
MC4 117 6
MC5 152 3
MC6 Captive 152 3
MC7 152 6
MC8 152 6
MC9 152 6
MC10 152 6
MC11 172 10
MC12 200 6
MC13 92 6
MC14 142 6
MC15 152 6
MC16 Structural 162 16
MC17 172 12
MC18 168 10
MC19 182 12
MC20 190 6

Interlocking
reduces

Male mullion with the leg Male mullion without the leg
Figure 7.1: Removing the leg of male mullion

7-2
 Parametric study of mullion couples under positive wind action

7.3 Effects of different parameters on the structural behaviour and

capacity of mullion couples

7.3.1 Gap between the mullions

The effects of gap on the structural behaviour and capacity of mullion couples MC5 and
MC15 under negative wind action were investigated in Chapter 6. It was found that the
capacity of the mullion couple increased with increasing gap. However, the effect of
gap under positive wind action cannot be justified based on this since the interaction
between male and female mullions could reduce with increasing gap (Figure 7.4).
Therefore, analyses were performed on mullion couples MC5 and MC15, where the gap
between the male and female mullions was considered as 1 mm, 3 mm, 6 mm and 12
mm. In addition, two spans (short and long spans) were considered since it would give a
comprehensive understanding of the effect of gap. Mullion couple MC15 was analysed
with 2400 mm span instead of 1800 mm span due to convergence issues.

Figures 7.2 and 7.3 show the moment versus displacement characteristics of mullion
couples MC5 and MC15 with different gaps. Despite differences in the type and span of
the mullion couples, the patterns in the variation of the ultimate moment capacities are,
in general, similar for all the cases. Figures 7.2 and 7.3 show that the ultimate moment
capacity reduces with increasing gap. Furthermore, the tangent (stiffness) of the
moment versus displacement curves did not vary significantly with the gap although the
ultimate moment capacities varied. Therefore, the deflection of the mullion couple was
not influenced by the gap between the mullions.

The maximum differences in the ultimate bending moment capacities of mullion

couples MC5 and MC15 with short span are about 5.3% and 3.9%, while those for long
span are about 42% and 1.8%, respectively. The significant reduction (40%) occurred
for the mullion couple MC5 with long span although the interaction between mullions
was lost for both short and long spans when a 10 mm gap was used. This is because, the
long span mullions are slender so that the interaction between them helped to improve
the capacity of the mullion couple (see Figure 7.4(b)), whereas the short span mullions
failed by local buckling (see Figure 7.4(a)), thus the influence of interaction was not
significant. In addition, the capacity of the mullion couple MC15 did not vary
significantly since the interaction between the male and female mullions was present

7-3
 Parametric study of mullion couples under positive wind action

(see Figures 7.5(a) to (c)). The gap used between the mullions will vary depending on
the type of mullion couple, structural and architectural requirements, and among facade
contractors. It was found from the above comparisons that the effect of gap on the
ultimate bending moment capacity of the mullion couples could be significant either for
short or long span mullion couples. Therefore, it is advisable to use the maximum gap
(in the case of positive wind action) when the capacity of the mullion couple is
determined using FEA.

10 9
9 8
8 7

Moment (kNm)
Moment (kNm)

7 6
6 5
5
4
4
Gap = 1 mm 3 Gap = 1 mm
3
Gap = 3 mm 2 Gap = 3 mm
2 Gap = 6 mm
Gap = 6 mm
1 1
Gap = 10 mm Gap = 10 mm
0 0
0 20 40 60 80 0 20 40 60 80 100 120
Y-axis displacement (mm) Y-axis displacement (mm)
(a) 1800 mm span (b) 4200 mm span

Figure 7.2: Moment versus displacement curves of mullion couple MC5 with
different gaps

12 10
11 9
10 8
9
Moment (kNm)

Moment (kNm)

8 7
7 6
6 5
5 4
4 Gap = 1 mm Gap = 1 mm
3
3 Gap = 3 mm Gap = 3 mm
2 2
Gap = 6 mm Gap = 6 mm
1 Gap = 10 mm 1 Gap = 10 mm
0 0
0 20 40 60 80 100 0 20 40 60 80 100 120 140 160
Y-axis displacement (mm) Y-axis displacement (mm)
(a) 2400 mm span (b) 4200 mm span
Figure 7.3: Moment versus displacement curves of mullion couple MC15 with
different gaps

7-4
 Parametric study of mullion couples under positive wind action

Male
mullion

Individual
rotation

Interaction Female
Interaction No interaction
mullion
Gap = 3 mm Gap = 6 mm Gap = 10 mm
(a) 1800 mm span

Interaction
Interaction
No interaction
Gap = 3 mm Gap = 6 mm Gap = 10 mm
(b) 4200 mm span
Figure 7.4: Failure modes of mullion couple MC5 with different gaps

Male
mullion

Female
mullion

Interaction Interaction Interaction

Gap = 3 mm (b) Gap = 6 mm (c) Gap = 10 mm

Figure 7.5: Failure modes of mullion couple MC15 with different gaps
(4200 mm span)

7-5
 Parametric study of mullion couples under positive wind action

7.3.2 Effectiveness of lateral restraint provided by glass pane

The effects of lateral restraint on the bending moment capacity of mullion couples MC5
and MC15 under negative wind action were investigated in Chapter 6. It was found that
the effectiveness of lateral restraint influenced the bending moment capacity of the
mullion couples under negative wind action, where the restraint was available at the
tension flange of the male and female mullions. During the positive wind action the
lateral restraint is available at the compression flange of the mullions, thus the effects of
lateral restraint on the capacity could be different. Therefore, mullion couples MC5 and
MC15 were analyzed under positive wind action by varying the spring constant of the
lateral restraint from fully effective to 1/40 N/mm/mm in a similar manner to the
negative wind action case (Section 6.5.2). In addition, two spans (short and long spans)
were considered since it would give a comprehensive understanding of the effect of
lateral restraint. Figure 7.6, for instance, shows the application of a spring lateral
restraint to the mullions.

Spring lateral
restraint

Figure 7.6: Typical spring lateral restraint under positive wind action

It is evident from Figures 7.7 and 7.8 that the ultimate bending moment capacity of the
mullion couple reduces when the effectiveness of the lateral restraint is reduced.
However, its influence on the stiffness of the mullion couple is insignificant, which is
the same observation for the negative wind action (Section 6.5.2). In addition, the effect
of lateral restraint is high under positive wind action than negative wind action since the
presence of lateral restraint is at the compression flange of the mullions. For the mullion
couple MC5, the maximum reduction in the ultimate bending moment capacities for
1800 mm and 4200 mm spans is 50%, i.e. when a spring constant of 1/40 N/mm/mm
was used. In comparison, the corresponding reductions were 22% and 8% for negative
wind action (Chapter 6). But for the mullion couple MC15, the maximum reductions in

7-6
 Parametric study of mullion couples under positive wind action

the ultimate bending moment capacities for 2400 mm and 4200 mm spans are 32% and
42%, respectively. The corresponding reductions for negative wind action were 11%
and 17% (chapter 6). This shows that the effect of lateral restraint on the ultimate
bending moment capacity of the mullion couples is significant for both short and long
spans. Therefore, it is advisable to use an appropriate reduction factor that can be used
conservatively for any span, when determining the ultimate bending moment capacity
of the mullion couples under positive wind action.

10 9
9 8
8 7

Moment (kNm)
Moment (kNm)

7 6
6
5
5
4
4 Full restraint Full restraint
3 k=3/2 N/mm/mm
3 k=3/2 N/mm/mm
2 k=5/16 N/mm/mm 2 k=5/16 N/mm/mm
k=1/8 N/mm/mm 1 k=1/8 N/mm/mm
1 k=1/40 N/mm/mm
k=1/40 N/mm/mm
0 0
0 10 20 30 40 50 0 20 40 60 80 100 120 140
Y-axis displacement (mm) Y-axis displacement (mm)
(a) 1800 mm span (b) 4200 mm span
Figure 7.7: Moment versus displacement curves of mullion couple MC5 with
different lateral restraint levels

12 10
11 9
10 8
9
Moment (kNm)

Moment (kNm)

8 7
7 6
6 5
5 4
Full restraint Full restraint
4 3
k=3/2 N/mm/mm k=3/2 N/mm/mm
3 k=5/16 N/mm/mm k=5/16 N/mm/mm
2 2
k=1/8 N/mm/mm 1 k=1/8 N/mm/mm
1 k=1/40 N/mm/mm k=1/40 N/mm/mm
0 0
0 20 40 60 80 0 20 40 60 80 100 120 140
Y-axis displacement (mm) Y-axis displacement (mm)
(a) 2400 mm span (b) 4200 mm span
Figure 7.8: Moment versus displacement curves of mullion couple MC15 with
different lateral restraint levels

7-7
 Parametric study of mullion couples under positive wind action

7.3.3 Ratio of the loads acting on male and female mullions

The effect of load distribution ratio (ratio of the loads acting on male and female
mullions as shown in Figure 7.9) under negative wind action was investigated in
Chapter 6. It was found that the load distribution ratio significantly influenced the
structural behaviour and capacity of the mullion couples under negative wind action.
The interaction between the male and female mullions under positive wind action is
different to that under negative wind action as stated in Sections 7.3.1 and 7.3.2, thus
the effect of load distribution ratio cannot be justified based on the results obtained for
the negative wind action case. Therefore, FE analyses were performed on mullion
couples MC5 and MC15 by considering different male and female mullions’ load
distribution ratios (a:b) of 1:0, 3:1, 2:1, 1:1, 1:2, 1:3 and 0:1.

Load Load
(Fmale) (Ffemale)

Load distribution ratio = Fmale / Ffemale

(a) Different panel widths (b) Load transferred to the mullion couple

Figure 7.9: Typical loads acting on male and female mullions of a mullion couple

It is evident from Figures 7.10 and 7.11 that the failure mode of a mullion couple could
vary depending on the load distribution ratio. That is, the mullion couples rotated in the
anti-clockwise direction and failed by lateral-torsional buckling when the load on
female mullion was higher than that on male mullion. On the other hand, the mullion
couples rotated in the clockwise direction and failed by lateral-torsional buckling when
the load on male mullion was higher than that on female mullion. Both mullion couples
showed approximately similar patterns of moment versus displacement behaviour to the
different load distribution ratios considered (Figures 7.12 and 7.13), where the capacity
7-8
 Parametric study of mullion couples under positive wind action

reduces as the load distribution ratio changes from 1:1 to 1:0 or 0:1. However, in all the
cases the stiffness of the mullion couples was not significantly affected, which proves
that the male and female mullions acted together. These observations show that the load
distribution ratio significantly influences the structural behaviour and capacity of
mullion couples as observed for negative wind action (Chapter 6).

Most of the mullion couples considered in this study showed a failure mode as shown in
Figure 7.10(a) or 7.11(a) under equal loading on male and female mullions. That is, the
male mullion pushes the female mullion, thus rotating together in the anti-clockwise
direction prior to failure by lateral-torsional buckling. But, the mullion couples used in
practical applications can be of any shape, and their failure mode can be of any type.
Therefore, the moment capacity of mullion couples under different load distribution
ratios can be determined using FEA by taking conservative assumptions.

Anti-clockwise
rotation

Clockwise
rotation
Female Male
Male mullion mullion Female
mullion mullion

(a) Load distribution ratio = 1:2 (b) Load distribution ratio = 2:1

Figure 7.10: Failure modes of mullion couple MC5 with load distribution ratios of
1:2 and 2:1

Anti-clockwise
rotation

Female
mullion Male Individual
mullion rotation
Male Female
mullion mullion

(a) Load distribution ratio = 1:2 (b) Load distribution ratio = 2:1

Figure 7.11: Failure modes of mullion couple MC15 with load distribution ratios of
1:2 and 2:1

7-9
 Parametric study of mullion couples under positive wind action

10 8
9 7
8
6
Moment (kNm)

Moment (kNm)
7
6 5
5 a:b=1:0 4 a:b=1:0
4 a:b=3:1 3 a:b=3:1
a:b=2:1 a:b=2:1
3 a:b=1:1 2 a:b=1:1
2 a:b=1:2 a:b=1:2
1 a:b=1:3 1 a:b=1:3
a:b=0:1 a:b=0:1
0 0
0 20 40 60 80 0 20 40 60 80 100 120
Y-axis displacement (mm) Y-axis displacement (mm)
(a) 1800 mm span (b) 4200 mm span

Figure 7.12: Moment versus displacement curves of mullion couple MC5 with
different load distribution ratios (a:b)

12 10
11 9
10 8
9
Moment (kNm)

Moment (kNm)

7
8
7 6
6 5
a:b=1:0 a:b=1:0
5 a:b=3:1 4 a:b=3:1
4 a:b=2:1 3 a:b=2:1
3 a:b=1:1 a:b=1:1
a:b=1:2 2
2 a:b=1:2
a:b=1:3 1
1 a:b=0:1 a:b=1:3
0 0
0 20 40 60 80 0 20 40 60 80 100 120 140 160
Y-axis displacement (mm) Y-axis displacement (mm)
(a) 1800 mm span (b) 4200 mm span
Figure 7.13: Moment versus displacement curves of mullion couple MC15 with
different load distribution ratios (a:b)

7.4 Comparison of FEA and current design predictions

7.4.1 General

Effects of gap between mullions, lateral restraint and ratio of the load on male and
female mullions on the moment capacity were investigated. It was found in Section 7.3
that these factors significantly influenced the moment capacity of the mullion couples.
The effects of these factors are not considered in the current mullion designs since past

7-10
 Parametric study of mullion couples under positive wind action

studies or current design standards have not considered them. Therefore, the moment
capacity of a mullion couple is determined by adding the individual bending moment
capacities of the male and female mullions in a similar manner to the negative wind
action case. It should be noted that the capacity of the mullion couple under positive
wind action is currently taken as the yield/section moment capacity since the
compression flange of the mullions is assumed to be fully laterally restrained. This is
questionable. In fact, this is one of the aims of this study. Furthermore, the moment
capacity prediction method in ADM (AA, 2015) does not consider the presence of any
lateral restraint (unrestrained case). The accuracy of the currently used assumption or
the design standard predictions is not known, thus the predicted moment capacities from
ADM (AA, 2015) and the yield moment capacities of the mullion couples are compared
with those obtained from FEA.

7.4.2 FEA and current design predictions

Local buckling check needs to be performed for the elements of a cross-section in order
to find the section moment capacity, which is then used to calculate the member
moment capacity. In Chapter 6, CUFSM (Li and Schafer, 2010) analyses were
performed for some mullion sections in order to perform the local buckling check for
the entire cross-section under negative wind action. The same approach was used in this
section to perform the local buckling check for the mullion sections subjected to
positive wind action.

The mullion sections considered in this study are commercially used sections. They can,
in general, reach yielding (or yield capacity) without any local buckling. However, to
verify this under positive wind action, the same mullion couples (MC10 and MC12)
were analysed using CUFSM software. In this case, mullions with full lateral restraint
and no lateral restraint (unrestrained) were considered, which relate to the currently
used design assumption and design standard predictions, respectively. It is evident from
Figures 7.14 and 7.15 that the local buckling stress of the mullion sections is
substantially more than yield stress (the load factor is greater than 1.0). This means that
the section moment capacity of the mullion couples can be considered as yield moment
capacity. Therefore, the yield moment capacity was considered when determining the
member moment capacity of the mullion couples.

7-11
 Parametric study of mullion couples under positive wind action

(a) Male mullion

(b) Female mullion

Figure 7.14: CUFSM analysis results of mullion couple MC10

(a) Male mullion

Figure 7.15: CUFSM analysis results of mullion couple MC12

7-12
 Parametric study of mullion couples under positive wind action

(b) Female mullion

Figure 7.15: CUFSM analysis results of mullion couple MC12

Table 7.2: Moment capacity predictions of FEA and Yield moment capacity

Yield moment
FEA prediction (MFEA)
Mullion capacity (My)
couple Span (mm)
All spans
1800 2400 3000 3600 4200
MC1 5.06 4.86 4.42 3.85 3.19 4.34
MC2* 4.31 3.87 3.27 2.62 N/A 4.06
MC3* 8.21 8.06 7.29 6.33 5.28 6.74
MC4 9.48 8.88 8.17 7.58 6.76 8.74
MC5 9.48 9.46 9.07 8.40 7.79 8.26
MC6 N/A 9.41 9.05 8.37 7.56 7.30
MC7 N/A N/A N/A 9.05 8.25 8.74
MC8 11.45 11.23 10.53 9.65 8.69 9.05
MC9 11.68 11.68 11.26 10.58 9.83 9.99
MC10* 8.19 7.12 6.18 5.60 5.46 9.11
MC11 22.09 21.72 20.95 19.85 18.57 17.53
MC12 19.01 19.26 18.86 18.09 17.14 15.68
MC13 4.21 3.94 3.44 2.88 2.44 3.66
MC14 8.91 8.48 7.90 7.15 6.28 7.38
MC15 10.59 10.65 10.16 9.50 8.77 7.80
MC16 13.66 12.53 10.64 8.85 7.48 11.75
MC17 17.16 16.83 16.16 15.51 14.39 11.50
MC18 15.06 14.64 12.51 11.28 9.08 12.95
MC19 20.52 20.80 20.62 20.07 19.17 13.09
MC20 19.17 18.21 17.50 16.66 15.58 14.78

7-13
 Parametric study of mullion couples under positive wind action

The mullion couples (MC1 to MC20) were analyzed under positive wind action with
full lateral restraint. In the FEA, uniform load distribution and equal loads on male and
female mullions were considered in a similar manner to the negative wind loading case
(Chapter 6). The gaps between the mullions were selected between 3 and 16 mm
depending on the mullion couple as used by G.James Glass and Aluminium. This will
give conservative predictions since the moment capacity of the mullion couple under
positive wind action reduces with increasing gap (refer Section 7.3.1 for more details).
The moment capacity predictions obtained from FEA (MFEA) and the yield moment
capacities (My) are given in Table 7.2, whereas Table 7.3 shows a comparison of the
predicted moment capacities. Refer Appendix C.3 for a sample calculation of the yield
moment capacity.

Table 7.3: Comparison of FEA predictions and Yield moment capacity

MFEA / My Min
Mullion couple
1800 2400 3000 3600 4200 (MFEA/My)
MC1 1.17 1.12 1.02 0.89 0.74 0.74
MC2 1.06 0.95 0.81 0.65 N/A 0.65
MC3 1.22 1.20 1.08 0.94 0.78 0.78
MC4 1.09 1.02 0.94 0.87 0.77 0.77
MC5 1.15 1.14 1.10 1.02 0.92 0.92
MC6 N/A 1.29 1.24 1.15 1.04 1.04
MC7 N/A N/A N/A 1.04 0.94 0.94
MC8 1.27 1.24 1.16 1.07 0.96 0.96
MC9 1.17 1.17 1.13 1.06 0.98 0.98
MC10 0.90 0.78 0.68 0.61 0.60 0.60
MC11 1.26 1.24 1.20 1.13 1.06 1.06
MC12 1.21 1.23 1.20 1.15 1.09 1.09
MC13 1.15 1.08 0.94 0.79 0.67 0.67
MC14 1.21 1.15 1.07 0.97 0.85 0.85
MC15 N/A 1.37 1.30 1.22 1.12 1.12
MC16 1.16 1.07 0.91 0.75 0.64 0.64
MC17 1.49 1.46 1.41 1.35 1.25 1.25
MC18 1.16 1.13 0.97 0.87 0.70 0.70
MC19 1.57 1.59 1.58 1.53 1.46 1.46
MC20 1.30 1.23 1.18 1.13 1.05 1.05
Min (MFEA/My) 0.90 0.78 0.68 0.61 0.60
COV 0.13 0.15 0.19 0.22 0.24

It is evident from Table 7.3 that the moment capacity reduces when the span is
increased for all the mullion couples. Most of the mullion couples (13 out of 20) could
not achieve the yield moment capacity even with the full lateral restraint condition,

7-14
 Parametric study of mullion couples under positive wind action

where the lowest MFEA/My ratio calculated was 0.60 (Table 7.3). This proves that using
the yield/section moment capacity for the mullion couples subject to positive wind
action is not correct. It is questionable why mullion couples could not achieve the yield
moment capacity even when glass provides full lateral restraint to their compression
flange. This is because, the male and female mullions of a mullion couple are subject to
both bending and torsion, and the restraint given by glass pane is not available at their
extreme compression fiber. Furthermore, Table 7.3 shows that the capacity ratio
(MFEA/My) varies significantly among the mullion couples, and increases with the span
(COV increases from 0.13 to 0.24 when the span increases from 1800 mm to 4200 mm).
Therefore, a reliable correlation between the capacity predictions could not be made.

Table 7.4: Moment capacity predictions of FEA and ADM (AA, 2015)

FEA prediction (MFEA) ADM prediction (MADM)

Mullion Span (mm) Span (mm)
couple
1800 2400 3000 3600 4200 1800 2400 3000 3600 4200
MC1 5.06 4.86 4.42 3.85 3.19 1.13 0.77 0.59 0.48 0.41
MC2* 4.31 3.87 3.27 2.62 N/A 1.44 0.96 0.72 0.58 0.49
MC3* 8.21 8.06 7.29 6.33 5.28 4.52 3.2 2.43 1.96 1.65
MC4* 9.48 8.88 8.17 7.58 6.76 5.01 3.22 2.35 1.84 1.52
MC5 9.48 9.46 9.07 8.40 7.79 1.51 1.01 0.76 0.61 0.51
MC6 N/A 9.41 9.05 8.37 7.56 1.75 1.17 0.88 0.72 0.6
MC7 N/A N/A N/A 9.05 8.25 2.52 1.64 1.21 0.97 0.81
MC8 11.45 11.23 10.53 9.65 8.69 2.72 1.78 1.32 1.06 0.9
MC9 11.68 11.68 11.26 10.58 9.83 4.45 2.87 2.1 1.66 1.38
MC10* 8.19 7.12 6.18 5.60 5.46 4.89 3.05 2.18 1.68 1.36
MC11 22.09 21.72 20.95 19.85 18.57 8.8 5.67 4.17 3.31 2.76
MC12 19.01 19.26 18.86 18.09 17.14 3.93 2.56 1.89 1.51 1.27
MC13 4.21 3.94 3.44 2.88 2.44 1.64 1.22 0.99 0.83 0.72
MC14 8.91 8.48 7.90 7.15 6.28 2.01 1.37 1.05 0.85 0.73
MC15 10.59 10.65 10.16 9.50 8.77 2.96 2.09 1.65 1.37 1.19
MC16 13.66 12.53 10.64 8.85 7.48 6.07 4.29 3.34 2.76 2.38
MC17 17.16 16.83 16.16 15.51 14.39 5.52 3.79 2.92 2.4 2.06
MC18 15.06 14.64 12.51 11.28 9.08 6.37 4.42 3.44 2.86 2.46
MC19 20.52 20.80 20.62 20.07 19.17 6.14 4.24 3.3 2.74 2.37
MC20 19.17 18.21 17.50 16.66 15.58 4.4 2.99 2.29 1.87 1.59
*Shear centre of the male and female mullions lies inside the line of action of loading.

7-15
 Parametric study of mullion couples under positive wind action

In Chapter 6 it was found that ADM (AA, 2015) predictions were conservative and
unsafe for the mullion couples given in Tables 7.5 and 7.6, respectively. In order to
assess its accuracy in the case of positive wind action, the moment capacities of the
mullion couples (MC1 to MC20) were calculated using ADM (AA, 2015) guidelines
and compared with FEA predictions. The predictions obtained from ADM (AA, 2015)
and FEA are given in Table 7.4, whereas Tables 7.5 and 7.6 show a comparison
between ADM (AA, 2015) and FEA predictions (MADM/MFEA) for the mullion couples
with shear centre outside and inside the line of action of the loading, respectively. Refer
Appendix C.3 for sample calculations of ADM (AA, 2015) moment capacity
predictions.

Table 7.5: Comparison of FEA and ADM (AA, 2015) predictions of mullion
couples with shear centre outside the line of action of loading

Mullion MADM / MFEA

Max Min Avg. COV
couple 1800 2400 3000 3600 4200
MC1 0.22 0.16 0.13 0.12 0.13 0.22 0.12 0.15 0.27
MC5 0.16 0.11 0.08 0.07 0.07 0.16 0.07 0.10 0.38
MC6 N/A 0.12 0.10 0.09 0.08 0.12 0.08 0.10 0.20
MC7 N/A N/A N/A 0.11 0.10 0.11 0.10 0.10 0.06
MC8 0.24 0.16 0.13 0.11 0.10 0.24 0.10 0.15 0.37
MC9 0.38 0.25 0.19 0.16 0.14 0.38 0.14 0.22 0.44
MC11 0.40 0.26 0.20 0.17 0.15 0.40 0.15 0.23 0.43
MC12 0.21 0.13 0.10 0.08 0.07 0.21 0.07 0.12 0.45
MC13 0.39 0.31 0.29 0.29 0.30 0.39 0.29 0.31 0.14
MC14 0.23 0.16 0.13 0.12 0.12 0.23 0.12 0.15 0.30
MC15 0.28 0.20 0.16 0.14 0.14 0.28 0.14 0.18 0.32
MC16 0.44 0.34 0.31 0.31 0.32 0.44 0.31 0.35 0.16
MC17 0.32 0.23 0.18 0.15 0.14 0.32 0.14 0.21 0.35
MC18 0.42 0.30 0.27 0.25 0.27 0.42 0.25 0.30 0.22
MC19 0.30 0.20 0.16 0.14 0.12 0.30 0.12 0.18 0.38
MC20 0.23 0.16 0.13 0.11 0.10 0.23 0.10 0.15 0.35
Max 0.44 0.34 0.31 0.31 0.32
Min 0.16 0.11 0.08 0.07 0.07
Avg. 0.30 0.21 0.17 0.15 0.15
COV 0.31 0.35 0.42 0.47 0.53

It is evident from Table 7.5 that the ADM predictions are conservative compared to the
FEA predictions for the mullion couples for which the shear centre lies outside the line
of action of loading (Figure 7.16(a)). For the most likely spans (3000 mm or 3600 mm)
the FEA predictions are about 200 to 1100% higher than the ADM (AA, 2015)
predictions. But, this increment is for the full lateral restraint condition, and thus it did

7-16
 Parametric study of mullion couples under positive wind action

not account for the reduction caused by the effectiveness of lateral restraint. However,
even with the highest reduction of 50% obtained in Section 7.3.2, the FEA prediction of
mullion couple MC5 showed approximately 525% increment. However, the actual
lateral restraint could be more effective than what was considered in Section 7.3.2.
Therefore, the ADM (AA, 2015) predictions are more conservative than the FEA
predictions.

Table 7.6: Comparison of FEA and ADM (AA, 2015) predictions of mullion
couples with shear centre inside the line of action of loading

MADM / MFEA
Mullion Span (mm) Max Min Avg. COV
couple
1800 2400 3000 3600 4200
MC2 0.33 0.25 0.22 0.22 N/A 0.33 0.22 0.26 0.21
MC3 0.55 0.40 0.33 0.31 0.31 0.55 0.31 0.38 0.27
MC4 0.53 0.36 0.29 0.24 0.22 0.53 0.22 0.33 0.37
MC10 0.60 0.43 0.35 0.30 0.25 0.60 0.25
Max 0.60 0.43 0.35 0.31 0.31
Min 0.33 0.25 0.22 0.22 0.22
Avg. 0.50 0.36 0.30 0.27 0.26
CoV 0.23 0.22 0.20 0.16 0.17

Table 7.6 shows that ADM (AA, 2015) predictions for the mullion couples with shear
centre inside the line of action of the loading (Figure 7.16(b)) are better than those
obtained for the mullion couples with shear centre outside the line of action of loading.
That is, for the 3000 mm span, FEA predictions of these mullion couples showed about
185 to 350% increment, which is less than that obtained for the mullion couples with
shear centre outside the line of action of the loading. In fact, this was the same in the
case of negative wind loading case, but it was an unsafe prediction (Chapter 6). These
mullion couples showed individual rotation during the early stage of loading, i.e. male
and female mullions rotated in the clockwise and anti-clockwise directions. However,
the female mullion rotated in the opposite direction (in the direction of the male
mullion) since the male mullion could have transferred some load at the tension flange
of the female mullion (see Figure 7.16(b)), which induces a torsional rotation in the
clockwise direction. This could be the reason why for the mullion couple under positive
wind action the individual mullion capacities did not show any significant difference as
that obtained under negative wind action. The capacity of the mullion couple cannot be

7-17
 Parametric study of mullion couples under positive wind action

the summation of the male and female mullions’ capacities (Figure 7.17(a)), and will
depend on the weaker mullion. Therefore, for instance, the capacity of the mullion
couple is equal to twice the capacity of the female mullion as shown in Figure 7.17(a),
which is the same for the negative wind loading case (Chapter 6).

Male Female Male Female

Line of action of the load

Load sharing region

Line of action of the load

Interaction
No interaction
Initial stage At failure

(a) Mullions showing interaction (b) Mullions behaving individually

Figure 7.16: Combined and individual behaviour of male and female mullions

7 4.0
6 3.5
3.0
Moment (kNm)

Moment (kNm)

5
2.5
4
2.0
3
1.5
2 1.0
Mullion couple Male mullion
1 0.5
2 x Female mullion Female mullion
0 0.0
0 20 40 60 80 100 0 20 40 60 80 100
Y-axis displacement (mm) Y-axis displacement (mm)
(a) Mullion couple (b) Individual mullions
Figure 7.17: Moment versus displacement curves of the mullion couple MC3

7-18
 Parametric study of mullion couples under positive wind action

The moment capacity predictions of FEA under positive wind action are higher than
those under negative wind action. However, the ADM (AA, 2015) predictions show that
the capacities under positive wind action are less than those under negative wind action
(refer Table 7.4 of this section and Table 6.5 of Chapter 6). This is because, load is
acting towards the shear centre during the positive wind action, thus lower than that
under negative wind loading case.

7.4.3 Capacity charts for mullion couples under positive wind action

It was found that various factors influence the structural behaviour and capacity of
mullion couples. It is difficult to incorporate the effects of these factors into the current
design rules since some factors did not show any pattern in relation to their effects.
Therefore, developing capacity charts as shown in Figure 7.18(a) is considered as a key
solution, which is the same for the mullion couples under negative wind action (Chapter
6).

(a) Capacity versus span curves

Figure 7.18: Capacity chart for mullion couple MC5 under positive wind action

7-19
 Parametric study of mullion couples under positive wind action

(b) Reduction factors and conditions

Figure 7.18: Capacity chart for mullion couple MC5 under positive wind action

To develop such capacity charts, FEA were performed using simple FE models of the
mullion couples with full lateral restraint and no transom. Then the results obtained
from the FEA were converted to a uniformly distributed load by considering a unit
panel width. It should be noted that, the effect of load distribution and gap were
included into the FE model by taking conservative assumptions, i.e. uniform loading
and larger gap (refer Sections 7.3.1 and 7.3.2). But, the effect of lateral restraint

7-20
 Parametric study of mullion couples under positive wind action

condition is incorporated using a separate reduction factor kelr as shown in Figure

7.18(b). The effect of the ratio of the loads acting on male and female mullions (load
distribution ratio) should be considered as a specific case since it cannot be incorporated
into the chart within a conservative approach. In addition to these a suitable capacity
reduction factor (I) should be used by designers to obtain the design capacity.

In Chapter 6 it was found that the effect of moment gradient is not significant (refer
Section 6.3 of Chapter 6 for more details). Therefore, the capacity curves for mullion
couples with an intermediate transom was developed using the capacities of the mullion
couples without any transom. This approach was used to develop capacity curves of the
mullion couples with an intermediate transom under positive wind action (Figure
7.18(a)). A similar approach can be used to develop capacity charts for any other cases
and mullion couple systems to make the design process easier for the facade industry.
Refer Appendix C.5 for more capacity charts developed for some other mullion couples
investigated in this chapter.

7.5 Conclusion

Effects of several key parameters such as the gap, lateral restraint, and load distribution
ratio were investigated in order to gain a comprehensive understanding of the structural
behaviour and capacity of mullion couple systems under positive wind action.
Thereafter, detailed FEA were conducted on the same mullion couples used in Chapter
6, which consisted of 20 mullion couples including both captive and structural glazing
mullions. The moment capacities obtained from FEA were then compared with the yield
moment capacities and the capacities obtained from ADM (AA, 2015).

Firstly, it was found that the gap between the mullions could significantly affect the
capacity under positive wind action, where the moment capacity decreased with
increasing gap. However, the effect could be small for the gaps in the practical range.
Since the capacity reduces with increasing gap, the maximum gap provided for a
mullion couple in practical applications can be used in FEA. In addition, a uniform load
distribution was used in FEA in a similar manner to that used in Chapter 6. Hence the
effects of gap and load distribution profile are on the conservative side, and they will be
accounted for in the analysis results. Further, the effects of lateral restraint and load
distribution ratio on the bending moment capacity of the mullion couples were found to

7-21
 Parametric study of mullion couples under positive wind action

be significant, and therefore they need to be treated separately. That is, the effect of
lateral restraint is incorporated as a reduction factor, whereas the effect of load
distribution ratio can be treated as a special case.

One of the main objectives of this research study is to assess whether the glass pane
provides full lateral restraint so that the mullion couples could achieve their
yield/section moment capacity under positive wind action. In order to investigate this,
the moment capacities of the mullion couples obtained from FEA were compared with
the corresponding yield moment capacities. This comparison showed that majority of
the mullion couples (13 out of 20) could not achieve their yield moment capacity,
particularly for the long spans, even with full lateral restraint. In fact, this shows that the
moment capacity of mullion couples cannot be taken as their yield moment capacity
assuming that glass pane provides full lateral restraint. Finally, comparisons were made
between ADM (AA, 2015) and FEA predictions. It showed that the ADM predictions
are more conservative for all the mullion couples irrespective of their shear centre
location (inside or outside the line of action of loading).

7.6 Recommendations

Based on the parametric study reported in this chapter, the following recommendations
are made in relation to the structural design of the mullion couples under positive wind
action.

x The moment capacities of mullion couples with full lateral restraint predicted by
FEA were less than the yield moment capacities, thus assuming yield/section
moment capacity is not recommended for the design of mullion couples under
positive wind action.
x The capacity prediction based on ADM (AA, 2015) was conservative, thus these
mullion couples can be designed safely in accordance with the standards. However, it
should be noted that the moment capacity from the design standard under positive
wind action is less than that obtained for negative wind action. If the mullion couples
need to be designed economically, finite element analysis based solutions shall be
used. But suitable FEA approaches should be followed by taking conservative
assumptions for the gap and load distribution profile in the development of FE
models.

7-22
 Numerical investigation of mullion couples with clip and/or hollow systems

Chapter 8

8 NUMERICAL INVESTIGATION OF MULLION

COUPLES WITH CLIP AND/OR HOLLOW SYSTEMS
8.1 General

Full and simple finite element (FE) models of a mullion couple without any clip or
hollow systems were developed and validated using full-scale test results in Chapter 5.
In this chapter, the same mullion couple with a clip at mid-span was developed and
analysed under both negative and positive wind actions using the validated finite
element analysis (FEA) approaches. The results from FEA were then compared with the
respective full-scale test results, which further validated the developed FE modeling and
analysis techniques. Following the validation, analyses were performed on different
mullion couples with clip system, hollow system and clip and hollow systems using the
simple FE models. Thereafter, the moment capacity predictions from the analyses were
compared with the capacities of respective mullion couples without any clip or hollow
systems investigated in Chapters 6 and 7. In addition, the moment capacity predictions
based on currently used design approaches and assumptions were compared with the
predictions from FEA. Furthermore, a brief discussion on the importance of
serviceability performance criteria is also given. Finally, suitable recommendations are
given for the economical and safe design of mullion couples with clip and/or hollow
systems subject to wind actions.

8.2 Validation of the FE models of mullion couples with a clip

The FEA techniques used in Chapter 5 to model the mullion couple without any clip or
hollow systems were used to model the mullion couple with a clip at its mid-span.
Although both full and simple FE models predicted the structural behavior and capacity
well, the simple FE model was preferred due to its simplicity and economy in terms of
time and computer resources. Therefore, simple FE models were used in the parametric
study of mullion couples without any clip or hollow systems, which is the most
commonly used mullion assembly (Chapters 6 and 7). In this chapter also, simple FE
models were used in the parametric study of the mullion couples with clip and/or
hollow systems. However, both full and simple FE models of the mullion couple tested
in Tests 3 and 4a were developed and analysed for comparison with the test results since

8-1
 Numerical investigation of mullion couples with clip and/or hollow systems

it would further increase the reliability of our validated FE models and analysis
techniques. Details of the FE modelling and the results and discussion of the tests and
FEA are given in this section.

8.2.1 Development of FE models

Key concepts used to idealize the full-scale test panels of the mullion couple without
any clip or hollow stiffener systems have been discussed in Chapter 5. Also, the
development of FE models, full and simple FE models, were discussed in detail. In
order to analyse the mullion couple with a clip at mid-span, the same techniques used in
Chapter 5 were used. Therefore, the details of the idealization of the facade assembly,
boundary conditions, selection of finite elements and analysis procedures, mesh, and
contact constraints will not be discussed in this chapter. However, the details of the FE
modelling of clip are discussed. The measured thicknesses and stress-strain curves of
the tested mullion sections given in Appendix B.1 were used in the FE models.

Modelling the clip

Clips are also thin-walled aluminium elements, thus idealized as similar to the mullions.
Figures 8.1(a) and (b) show the assembly of the mullion couple with a clip and the
idealized sections, respectively. As seen in Figure 8.1(a), there are two separate clip
parts. In FE simulations, contact interactions were defined between those two clip parts
in a similar manner to those defined for mullions in Section 5.3.7, so that the opening of
mullion couples will be prevented by the clip in simulations. Furthermore, the clip parts
are inserted into the screw ports of the male and female mullions in order to keep them
in place (Figure 8.1(a)). This was simulated in FEA by tying the edges of the clip parts
to the surface of the mullions in the screw ports (Figure 8.1(b)). Refer Appendix B.1 for
the dimensions of the clip parts used in the FE models.

The main function of the clip system is to prevent the opening of the mullion couple
(refer Chapter 4). Visual observations showed that the clip did not fail or undergo any
visible yielding during the tests, thus the measured mechanical properties of the clip
system are not needed in FE modelling. Therefore, the yield and ultimate tensile
strengths were assumed as 172 and 207 MPa, respectively, based on AS 1664 (SA,
1997). The density, Poisson’s ratio and the material model used for clips in ABAQUS
are the same as those used for the mullions in Section 5.3.7 of Chapter 5.

8-2
 Numerical investigation of mullion couples with clip and/or hollow systems

Clip parts Tied edge (Clip)

Tied surface
Screw ports (Mullion)

(a) Actual geometry (b) Idealized geometry

Figure 8.1: Mullion couple with a clip at mid-span

8.2.2 Results and discussion

8.2.2.1 Simple model

Test 3: Mullion couple with a clip at mid-span under negative wind action

In the FEA simulation, when the load was increased, the male and female mullions
rotated to lean on each other. Thereafter, both mullions started to rotate together in the
clockwise direction. After a certain loading the female mullion separated from the male
mullion, but both mullions continued to rotate in the same direction. Finally both of
them failed by lateral-torsional buckling, where the female mullion separated from the
male mullion significantly at 490 mm (450 mm in Test 3 – see Figure 8.2(d)) away from
the mid-span prior to failure (Figure 8.2(e)). This is because the presence of clip at mid-
span prevents the separation of female mullion from the male mullion, and thus the
failure occurred at a location away from the mid-span. This behaviour of the mullion
couple is the same as that observed in Test 3 except during the initial loading, where the
mullion couple rotated together in the test due to possible frictional effect on the clip
(Figures 8.2(a) to (e)). Figure 8.3(a) demonstrates that the pressure versus male and
female mullions’ Y-axis displacement curves are predicted by this model with a
reasonable accuracy. FEA prediction is about 18% lower than the test ultimate capacity.
This could be due to variations in the test results. The pressure versus male and female
mullions’ X-axis displacement curves also agreed reasonably well with the test results
as shown in Figure 8.3(b). Furthermore, the ratio between the kinetic energy (KE) to
internal energy (IE) of the mullion couple for the quasi-static implicit analysis is shown

8-3
 Numerical investigation of mullion couples with clip and/or hollow systems

in Figure 8.4. Although it is higher at the beginning, it remained less than 1%

throughout the analysis until failure.

Leaning

Negative wind action

View from outside the pressure
chamber (Mid-span region)
Test FEA
(a) Mullions lean on each other

Male
Female mullion
Female mullion Male mullion mullion

Test FEA
(b) Mullions rotating together

Separation

Separation

Test FEA

(c) Separation of female and male mullions

Figure 8.2: Mullion couple behaviour in FEA (simple model) and Test 3

8-4
 Numerical investigation of mullion couples with clip and/or hollow systems

Significant separation of female

mullion at 450 mm from the mid-span

~450 mm

Clip prevented further

No further separation at mid-span separation at mid-span
Test FEA
(d) Failure of mullions

~ 490 mm
Top support
Male
mullion
Separation of female mullion
away from mid-span
Female Mid-span
mullion

(e) Three dimensional view of the failed mullion couple in FEA

Figure 8.2: Mullion couple behaviour in FEA (simple model) and Test 3

-1400
-1200
Pressure (Pa)

-1000
-800
-600
Test (female)
-400 Test (male)
-200 FEA (female)
Test (male)
0
0 -10 -20 -30 -40 -50 -60 -70
Displacement (mm)
(a) Pressure versus Y-axis displacement
Figure 8.3: Comparison of results from FEA (simple model) and Test 3

8-5
 Numerical investigation of mullion couples with clip and/or hollow systems

-1400
-1200

Pressure (Pa)
-1000
-800
-600
Test (female)
-400 Test (male)
FEA (female)
-200
FEA (male)
0
-5 15 35 55 75 95
Displacement (mm)
(b) Pressure versus X-axis displacement
Figure 8.3: Comparison of results from FEA (simple model) and Test 3

Note: The directions of X and Y axes are shown in Figure 8.2(a).

100
90
80
100 * KE/IE

70
60
50
40
30
20
10
0
0 -200 -400 -600 -800 -1000
Pressure (Pa)
Figure 8.4: Energy ratio for FEA (simple model) - Test 3

Test 4a: Mullion couple with a clip at mid-span under positive wind action

Mullion couple with a clip at mid-span was analysed under positive wind action, where
the spring constant used in FEA of Test 2 was used for the lateral restraint (Section
5.5.2 of Chapter 5). During the early stage of loading, the male and female mullions
rotated to open until the clip is activated (Figure 8.5(a)). Thereafter, both of them started
to rotate together in the anti-clockwise direction (Figure 8.5(b)). Finally both of them
failed by lateral-torsional buckling (Figures 8.5(c) and (d)). This behaviour of the
mullion couple is the same as observed in Test 4a except during the initial loading

8-6
 Numerical investigation of mullion couples with clip and/or hollow systems

where the mullion couple rotated together due to possible seal friction (Figures 8.5(a) to
(e)). Figure 8.6(a) shows that the pressure versus Y-axis displacement curves are
predicted by this model with good accuracy. FEA prediction is about 12% higher than
the test ultimate capacity. This could be due to variations either in the test results or the
spring lateral restraint used, which cannot be determined accurately (refer Appendix B.3
for more details). The pressure versus X-axis displacement curves also agreed
reasonably well with the test results as shown in Figure 8.6(b). Furthermore, the ratio
between the kinetic energy (KE) to internal energy (IE) of the mullion couple for the
quasi-static implicit analysis is shown in Figure 8.7. Although it is higher at the
beginning, it remained less than 1% throughout the analysis until failure.

Rotation

Clip activated

Positive wind action

View from outside the pressure

chamber (Mid-span region)

Test FEA
(a) Early stage of loading

Gap
Aluminium beam used Male
to fix the LVDTs Female mullion
mullion

Unrestrained tension flanges

Test FEA
(b) After 1800 Pa

Figure 8.5: Mullion couple behaviour in FEA (simple model) and Test 4a

8-7
 Numerical investigation of mullion couples with clip and/or hollow systems

Gap did not further increase

Lateral movement

Sliding of Gap
mullion f h

Gap did not further increase

Test FEA
(c) Sliding of mullions

Mullion slided significantly on

the aluminium sheet

Aluminium sheet popped out

Female mullion Male mullion

Test FEA
(d) At failure

Female
mullion

Top
support
Male
mullion

Mid-span

(e) Three dimensional view of the failed mullion couple in FEA

Figure 8.5: Mullion couple behaviour in FEA (simple model) and Test 4a

8-8
 Numerical investigation of mullion couples with clip and/or hollow systems

3500
3000

Pressure (Pa)
2500
2000
1500
Test (female)
1000 Test (male)
500 FEA (female)
FEA (male)
0
0 20 40 60 80 100
Y-axis displacement (mm)

(a) Pressure versus Y-axis displacement

3500
3000
Pressure (Pa)

2500
2000
1500
Test (female)
1000 Test (male)
500 FEA (female)
FEA (male)
0
5 -5 -15 -25 -35 -45
X-axis displacement (mm)

(b) Pressure versus X-axis displacement

Figure 8.6: Comparison of results from FEA of simple model and Test 4a

100
90
80
100 * KE/IE

70
60
50
40
30
20
10
0
0 500 1000 1500 2000 2500 3000 3500
Pressure (Pa)

Figure 8.7: Energy ratio for FEA (simple model) - Test 4a

8-9
 Numerical investigation of mullion couples with clip and/or hollow systems

8.2.2.2 Full model

Test 3: Mullion couple with a clip at mid-span under negative wind action

The developed full model of the mullion couple was analyzed under negative wind
action. The behaviour of the mullion couple predicted by the full model is similar to that
predicted by the simple model (Figure 8.8). Figures 8.9(a) and (b) show a comparison of
pressure versus displacement curves for the mullion couple in Test 3 and FEA, where
the results agreed reasonably well. The ultimate capacity predicted by the FEA of full
model is about 18% less than the test results, which could be due to variations in the test
results as similar to that in the simple model comparison (Section 8.2.3.1). Figure 8.10
shows the kinetic energy (KE) to internal energy (IE) ratio of the model, where it
remained less than 1% throughout the analyses although it is higher at the beginning.

Top support Separation of female mullion

away from mid-span

~ 402 mm

Clip prevented Male mullion

further separation
Female mullion
at mid-span Mid-span

Figure 8.8: Mullion couple behaviour in FEA (full model) and Test 3

-1400
-1200
Pressure (Pa)

-1000
-800
-600
Test 1 (female)
-400 Test 1 (male)
-200 FEA-Full (female)
FEA-Full (Male)
0
0 -20 -40 -60 -80
Y-axis displacement (mm)

(a) Pressure versus Y-axis displacement

Figure 8.9: Comparison of results from FEA (full model) and Test 3

8-10
 Numerical investigation of mullion couples with clip and/or hollow systems

-1400
-1200

Pressure (Pa)
-1000
-800
-600 Test 3-Female
-400 Test 3-Male
FEA-Full (female)
-200
FEA-Full (male)
0
-5 15 35 55 75 95
X-axis displacement (mm)

(b) Pressure versus X-axis displacement

Figure 8.9: Comparison of results from FEA (full model) and Test 3

100
90
80
100 * KE/IE

70
60
50
40
30
20
10
0
0 -200 -400 -600 -800 -1000
Pressure (Pa)

Figure 8.10: Energy ratio for FEA (full model) - Test 3

Test 4a: Mullion couple with a clip at mid-span under positive wind action

The developed full model of the mullion couple was analyzed under positive wind
action. Failure mode of the mullion couple predicted by the full model agreed well with
the test result (Figures 8.11(a) and (b)). The rotation of both the male and female
mullions together in the anti-clockwise direction and the sliding of the mullions on the
aluminium sheets prior to failure observed during the test were simulated by the full
model as evident from these figures. This is because the full model simulated the real
lateral restraint that was given to the mullions by the aluminium sheets through
frictional interaction. Figures 8.12(a) and (b) show the pressure versus displacement

8-11
 Numerical investigation of mullion couples with clip and/or hollow systems

curves of the mullion couple from Test 4a and FEA, where the full model shows
slightly stiffer behaviour than the test result. Furthermore, the ultimate capacity
predicted by the full model is about 2% less than the test result. Figure 8.13 shows the
kinetic energy (KE) to internal energy (IE) ratio of the model, where it remained less
than 1% throughout the analyses although it is higher at the beginning.

At zero pressure Before failure

(a) Test

Male
Female mullion
mullion

Sliding of
mullions

(b) FEA
Figure 8.11: Mullion couple behaviour in FEA (full model) and Test 4a

8-12
 Numerical investigation of mullion couples with clip and/or hollow systems

3000

2500

2000

Pressure (Pa)
1500

1000 Test (female)

Test (male)
500 FEA (female
FEA (male)
0
0 10 20 30 40 50 60
Y-axis displacement (mm)
(a) Pressure versus Y-axis displacement
3000

2500

2000
Pressure (Pa)

1500

1000 Test-Female
Test-Male
500 FEA-Female
FEA-Male
0
5 0 -5 -10 -15 -20
X-axis displacement (mm)
(b) Pressure versus X-axis displacement
Figure 8.12: Comparison of results from FEA (full model) and Test 4a

100
90
80
70
100 * KE/IE

60
50
40
30
20
10
0
0 500 1000 1500 2000 2500 3000
Pressure (Pa)
Figure 8.13: Energy ratio for FEA (full model) - Test 4a

8-13
 Numerical investigation of mullion couples with clip and/or hollow systems

8.3 Mullion couples with clip system

8.3.1 General

In Section 8.2, the mullion couple with a clip at mid-span was modelled and validated
against the full scale test results. It was also shown that both the full and simple FE
models of the mullion couple predicted the structural behaviour and capacity well.
However, simple FE model was preferred in the parametric study considering the time
and computer resources as observed with the parametric study of mullion couples
without any clip or hollow systems conducted in Chapters 6 and 7. In this parametric
study, six different types of mullion couples comprising structural and captive glazing
systems with spans of 1800 mm, 3000 mm and 4200 mm were considered. It should be
noted that, these mullion couples were selected from the mullion couples investigated in
Chapter 6, which represent different glazing systems and mullion sizes. Refer Chapter 6
for the geometry and sizes of the mullion couples. Figures 8.14(a) to (e) show the cross-
sectional views of the mullion couple models from ABAQUS software.

(a) MC1 (b) MC2 (c) MC5

(d) MC12 (e) MC13 (f) MC15

Figure 8.14: Mullion couples with a clip at mid-span

The clip system shown in Figure 8.14 is a separate piece of extrusion (Figure 8.1(a)),
which can be used with any mullion couple sections. In industry practice, clips with a

8-14
 Numerical investigation of mullion couples with clip and/or hollow systems

length of 100 mm are used at the mid-span of mullion couples. It is because the
dominant separation between male and female mullions occur at the mid-span of a
mullion couple (refer Chapter 4 for more details). Therefore, in this parametric study
mullion couples with a 100 mm long clip at their mid-span were analyzed. Furthermore,
in some instances, designers use a continuous clip system, in which the mullion couple
itself has the clip feature as shown in Figure 8.15(b). This clip system eliminates the
labour work associated with the installation of the separate clip parts. However, it
should be noted that both clip systems are not frequently used due to the difficulties in
removal of the unitized facade mullions in case of renovation or replacement.

Separation Separation

Resistance to separation of
No resistance to separation mullions (Figure 8.22(b))
of mullions (Figure 8.22(a))

(a) Mullion couple without clip (b) Mullion couple comprising

system continuous clip
Figure 8.15: Mullion couple with and without continuous clip system

8.3.2 Mullion couples with clip under negative wind action

In this section, results obtained from FEA of the mullion couples with a clip at their
mid-span under negative wind action are compared with the respective capacities of the
mullion couples without clip. Mullion couples MC1, MC2, MC5 and MC12 are captive
glazing mullions, whereas the other two are structural glazing mullions. It was found in
Chapter 6 that all the mullion couples except MC2 showed interaction (shear centre lies
inside the line of action of the loading). Mullion couple MC2 showed individual
rotation since its shear centre lies outside the line of action of the loading. Comparison
of the results of mullion couples with and without clip under negative wind action is
given next.

8-15
 Numerical investigation of mullion couples with clip and/or hollow systems

Comparison of mullion couples with and without clip

Figures 8.16(a) and (b) show the predicted failure modes of the mullion couple MC5,
which is a captive glazing mullion couple, with and without a clip. The failure mode of
the mullion couple without clip is lateral-torsional buckling (Figure 8.16(a)). The failure
mode of the mullion couple with a clip is still the same, but the location of the failure is
not at its mid-span (Figure 8.16(b)). In this case, lateral-torsional buckling failure
occurred at a location away from the mid-span. This is because, the presence of clip
prevents the separation of female mullion from the male mullion, and thus the failure
occurred at a location away from the clip location. The failure modes of the structural
glazing mullion couple MC15 with and without a clip are also the same as those of the
mullion couple MC5 with and without a clip (Figures 8.16 and 8.17). Since the mullion
couples MC1, MC12 and MC13 show similar behaviour (interaction) as mullion
couples MC5 and MC15, they are not discussed.

Separation of female Separation of female mullion

mullion at mid-span away from mid-span

Male
Male
mullion
mullion Female
mullion Female
Mid-span Mid-span mullion

(a) Without a clip (b) With a clip

Figure 8.16: Predicted failure modes of mullion couple MC5 with and without a
clip under negative wind action

Separation of female Separation of female mullion

mullion at mid-span away from mid-span

Male
Male
mullion
mullion Female
Female
mullion
Mid-span Mid-span mullion

(a) Without a clip (b) With a clip

Figure 8.17: Predicted failure modes of mullion couple MC15 with and without a
clip under negative wind action

8-16
 Numerical investigation of mullion couples with clip and/or hollow systems

Mullion couple MC2 shows individual rotation (or no leaning action) as shown in
Figure 8.18(a). This means that the moment capacity of the mullion couple will depend
on the mullion with lower capacity, which can be either male or female mullion (refer
Chapter 6 for more details). When a clip is used at the mid-span of such mullion couples
the separation that occurs between the male and female mullions due to individual
rotation will be prevented. This means that the mullion couple can rotate only in one
direction (clockwise or anti-clockwise), which depends on the level of torsional force
acting on the mullions and their rigidity against torsional rotation. For instance, Figure
8.18(b) shows that the male and female mullions rotated together in the clockwise
direction prior to failure. In this case, the female mullion pulled the male mullion. This
can be justified based on Figure 8.19, where the female mullion has higher eccentricity
than the male mullion, but their torsional constants are approximately similar (848 mm4
versus 939 mm4 for the female and male mullions, respectively).

Mullions rotating
Male
together
mullion

Female Male
mullion mullion

Individual rotation Female

Mid-span of mullions mullion
Mid-span

(a) Without a clip (b) With a clip

Figure 8.18: Predicted failure modes of mullion couple MC2 with and without a
clip under negative wind action

Shear centre

Line of action of
the loading

Load
Note: Dimensions are in mm

Figure 8.19: Load and shear centre details of mullion couple MC2

8-17
 Numerical investigation of mullion couples with clip and/or hollow systems

Moment capacity predictions based on FEAs of the mullion couples with and without a
clip under negative wind action are given in Table 8.1. It is evident from Table 8.1 that
the moment capacity increment varied between 3% and 37% for all the mullion couples
except for mullion couple MC2 when a clip was used at their mid-span. The increments
for the most commonly used span range (3000 mm) varied between 4% and 26%, but
mostly increments were limited to 15%. Higher increments were obtained mainly for
4200 mm span, but overall the increments are not significant (Figure 8.20). Considering
the fact that Hoglund (2015) and Skejic et al. (2017) assume the unbraced length of a
mullion couple can be considered as half the span if a clip is used at its mid-span
(Figure 8.21), thus a higher capacity increment is expected. For instance, expected
capacity of mullion couple MC5 with a span of 3000 mm and a clip at its mid-span
should be more than the capacity of the mullion couple (without clip) with half the span
length (1500 mm). However, the capacity of the mullion couple with a span of 3000
mm and a clip at its mid-span (3.36 kNm) is significantly lower than even the capacity
of the mullion couple (without clip) with 1800 mm span (5.09 kNm). Therefore, the use
of clip cannot reduce the unbraced length to the extent expected, and thus cannot
significantly increase the capacity of such mullion couples under negative wind action.

Table 8.1: Moment capacities of the mullion couples with and without a clip at
mid-span based on FEA under negative wind action

Without a clip M (kNm) With a clip Mclip (kNm) Mclip / M

Mullion
couple Span (mm) Span (mm) Span (mm)
1800 3000 4200 1800 3000 4200 1800 3000 4200
MC1 3.36 1.73 1.17 3.58 2.18 1.59 1.07 1.26 1.35
MC2* 1.32 0.80 N/A 2.91 2.20 N/A 2.20 2.74 N/A
MC5 5.09 2.86 1.82 5.78 3.36 2.28 1.14 1.18 1.25
MC12 11.70 7.46 4.86 12.19 7.75 5.44 1.04 1.04 1.12
MC13 3.61 2.50 1.67 3.85 2.81 2.29 1.07 1.13 1.37
MC15 8.21 6.03 4.17 8.48 6.39 4.57 1.03 1.06 1.10
* This mullion couple shows no interaction since the shear centres of the male and female
mullions lie inside the line of action of the loading (Figure 8.19).

Figure 8.20 shows that the capacity of mullion couple MC2 increased significantly
compared to the other mullion couples when a clip was used at its mid-span (between
120% and 174%). This is due to the counter balance in the torsional forces that are
acting on the male and female mullions (Figures 8.18 and 8.19). That is, during the
negative wind action the male and female mullions rotate in the anti-clockwise and
clockwise directions, respectively (Figure 8.18(a)), and the presence of clip prevents

8-18
 Numerical investigation of mullion couples with clip and/or hollow systems

this individual rotation by pulling each other, thus reducing the torsional rotation of the
mullions. It should be noted that the individual rotation of the mullions (no interaction)
makes this mullion couple different from the other mullion couples (load and shear
centre locations). Therefore, the capacity of the mullion couple increased significantly
compared to the other mullion couples. In this case, mullion couple rotated in the
direction in which higher torsional force was acting, and then failed by lateral-torsional
buckling (Figure 8.18(b)). Moreover, moment capacity increments show an increasing
trend with increasing span. This observation is the same for the other mullion couples
also.

3.0
2.8 MC1
Moment capacity ratio (Mclip / M)

MC2
2.6 MC5
2.4 MC12
2.2 MC13
2.0 MC15
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
1800 3000 4200
Span (mm)

Figure 8.20: Moment capacity ratio of mullion couples with and without a clip
versus span based on FEA under negative wind action

Clip location (mid-span)

Figure 8.21: Reduction in unbraced length due to the use of clip under negative
wind action

8-19
 Numerical investigation of mullion couples with clip and/or hollow systems

Table 8.2: Moment capacities of the mullion couples with and without a clip based
on ADM (AA, 2015) under negative wind action
Mclip / M
Mullion Without a clip M (kNm) With a clip Mclip (kNm)
couple Span (mm) Span (mm) Span (mm)
1800 3000 4200 1800 3000 4200 1800 3000 4200
MC1 2.16 0.97 0.6 3.88 2.84 1.92 1.80 2.93 3.20
MC2* 2.29 1.03 0.65 3.38 2.60 1.56 1.48 2.52 2.39
MC5 3.41 1.44 0.86 6.81 4.74 2.97 2.00 3.29 3.45
MC12 8.47 3.9 2.28 14.56 10.96 7.80 1.72 2.81 3.42
MC13 2.64 1.55 1.01 3.90 3.24 2.52 1.48 2.09 2.50
MC15 5.40 2.97 1.87 8.39 6.79 5.10 1.55 2.29 2.73

Table 8.3: Moment capacities of the mullion couples with a clip based on FEA and
ADM (AA, 2015) under negative wind action

Mclip,ADM (kNm) Mclip,FEA (kNm) Mclip,ADM / Mclip,FEA

Mullion
couple Span (mm) Span (mm) Span (mm)
1800 3000 4200 1800 3000 4200 1800 3000 4200
MC1 3.88 2.84 1.92 3.58 2.18 1.59 1.08 1.30 1.21
MC2 3.38 2.60 1.56 2.91 2.20 N/A 1.16 1.18 N/A
MC5 6.81 4.74 2.97 5.78 3.36 2.28 1.18 1.41 1.30
MC12 14.56 10.96 7.80 12.19 7.75 5.44 1.19 1.41 1.43
MC13 3.90 3.24 2.52 3.85 2.81 2.29 1.01 1.15 1.10
MC15 8.39 6.79 5.10 8.48 6.39 4.57 0.99 1.06 1.12

Earlier in this section, it was shown that assuming a reduced unbraced length by
considering the presence of a clip is not applicable. However, in order to assess this
assumption in relation to the currently used design approach, the moment capacities of
the mullion couples with and without a clip were obtained based on ADM (AA, 2015)
and compared in Table 8.2. In addition, the moment capacities of the mullion couples
with a clip obtained based on ADM (AA, 2015) and FEA were also compared since it
would give an improved understanding of the assumption (Table 8.3). It is evident from
Table 8.2 that the capacity increments of the mullion couples varied between 48% and
245%, which is significantly higher than that obtained based on FEA predictions (varied
between 3% and 37%). In addition, Table 8.3 shows that the capacity of the mullion
couples with a clip at their mid-span predicted by FEA are in general lower than the
ADM (AA, 2015) predictions, which are unsafe. Therefore, based on both comparisons
it can be stated that assuming a reduced unbraced length by considering the presence of
a clip is not acceptable, and the capacity of the mullion couples cannot be increased

8-20
 Numerical investigation of mullion couples with clip and/or hollow systems

significantly by using clips. However, serviceability performance of the mullion

couples, such as opening, can be improved by using clips (e.g. Figures 8.16(a) and (b)).

Effect of using continuous clip

Mullion couple MC5 was modified in order to introduce the continuous clip feature in
the mullion couple as shown in Figure 8.15. In this clip system, the clip feature is
available throughout the length of the mullion couple since it is a part of the mullion
sections. In order to investigate the influence of the continuous clip system on the
structural behaviour and capacity of the mullion couples, mullion couples MC5 with
and without the clip system were analysed. The span of the mullion couple considered
was 3000 mm, which is a commonly used mullion span. Figures 8.22(a) and (b) show
that the female mullion separated from the male mullion in the mullion couple without
clip, whereas both mullions rotated together when the continuous clip was present. In
addition, it is evident from Figure 8.22(b) that the use of continuous clip prevents the
separation of the male and female mullions throughout the length. However, the use of
continuous clip did not increase the capacity significantly. The capacity of the mullion
couple with continuous clip was increased only by 11% (Figure 8.23), which is
negligible when compared to the increments expected based on currently used
assumptions as given in Table 8.2. Therefore, the clip systems (continuous clip or
separate clip parts) cannot be used to increase the ultimate moment capacity of mullion
couples.

Separation of female mullion

(a) Mullion couple without clip system

Figure 8.22: Predicted failure modes of mullion couple MC5 with and without
continuous clip system under negative wind action

8-21
 Numerical investigation of mullion couples with clip and/or hollow systems

No separation between
mullions

(b) Mullion couple comprising continuous clip

Figure 8.22: Predicted failure modes of mullion couple MC5 with and without
continuous clip system under negative wind action

5.0
4.5
4.0
Moment (kNm)

3.5
3.0
2.5
2.0
1.5
1.0 Without clip
0.5 With continuous clip
0.0
0 -10 -20 -30 -40
Y-axis displacement (mm)

Figure 8.23: Moment versus displacement curves of mullion couple MC5 with and
without continuous clip system under negative wind action

8.3.3 Mullion couples with clip under positive wind action

Comparison of mullion couples with and without clip

During the positive wind action, the clips are available in the tension region of the
mullions, which is expected to provide no improvement to the ultimate capacity of the
mullion couples. However, the presence of clip would prevent the opening of the
mullion couples during the positive wind action, thus improving the serviceability
performance. In order to investigate the effect of clips on the ultimate capacity of the

8-22
 Numerical investigation of mullion couples with clip and/or hollow systems

mullion couples, the mullion couples analysed in Section 8.3.2 were analysed under
positive wind action.

Figures 8.24(a) and (b) show the predicted failure modes of the mullion couple MC5
with and without a clip under positive wind action. The failure mode of the mullion
couple without clip is lateral-torsional buckling (Figure 8.24(a)). The failure mode of
the mullion couples with a clip is still the same (Figure 8.24(b)). Failure modes of
mullion couples MC15 and MC2 were also the same as those obtained for the mullion
couple MC5 (Figures 8.24 to 26). Furthermore, in this case, the opening of mullion
couples MC5 and MC15 was prevented. In mullion couple MC2, separation occurs at
the bottom flange region due to the nature of the position of shear centre and line of
action of the loading, thus the presence of clip could not prevent it (Figure 8.26(b)).
Mullion couples MC1, MC12 and MC13 have similar structural behaviour as mullion
couples MC5 and MC15, and thus they are not discussed here.

Mullion couple No separation

opened up

Clip

Male Female Male Female

mullion mullion mullion mullion

(a) Without a clip (b) With a clip

Figure 8.24: Predicted failure modes of mullion couple MC5 with and without a
clip under positive wind action

Mullion couple No separation

opened up

Clip

Male Female Male Female

mullion mullion mullion mullion

(a) Without a clip (b) With a clip

Figure 8.25: Predicted failure modes of mullion couple MC15 with and without a
clip under positive wind action

8-23
 Numerical investigation of mullion couples with clip and/or hollow systems

No separation
No separation
Clip

Male Female
Male Female
mullion mullion
mullion mullion
Separation at bottom
Separation at bottom
flange region
flange region

(a) Without a clip (b) With a clip

Figure 8.26: Predicted failure modes of mullion couple MC2 with and without a
clip under positive wind action

Moment capacity predictions based on FEAs of the mullion couples with and without a
clip under positive wind action are given in Table 8.4. It is evident from Table 8.4 that
the moment capacity increments varied between 0% and 39% for all the mullion
couples. The increments for the most commonly used span (3000 mm) varied between
0% and 26%, but mostly the increments were limited to 10%. In addition, the
increments show an increasing trend with the span length, but the increments are not
significant. Furthermore, the capacity increments show significant variations between
the mullion couples (Figure 8.27). Therefore, it is concluded that the ultimate moment
capacity of mullion couples under positive wind action cannot be increased significantly
by using clips. However, clips can improve the serviceability performance of mullion
couples (Figures 8.24 to 26).

Table 8.4: Moment capacities of the mullion couples with and without a clip based
on FEA under positive wind action

Mullion Without clip M (kNm) With a clip Mclip (kNm) Mclip / M

couple Span (mm) Span (mm) Span (mm)
1800 3000 4200 1800 3000 4200 1800 3000 4200
MC1 5.06 4.42 3.19 5.40 4.74 3.48 1.07 1.07 1.09
MC2 4.46 3.46 N/A 4.49 3.47 N/A 1.01 1.00 N/A
MC5 9.48 9.07 7.79 9.76 9.51 8.14 1.03 1.05 1.04
MC12 19.01 18.86 17.14 18.94 19.7 18.27 1.00 1.04 1.07
MC13 4.21 3.44 2.44 5.02 4.33 3.38 1.19 1.26 1.39
MC15 10.59 10.16 8.77 11.01 10.55 9.18 1.04 1.04 1.05

8-24
 Numerical investigation of mullion couples with clip and/or hollow systems

3
MC1

Moment capacity ratio (Mclip / M)

2.8
2.6 MC2
2.4 MC5
2.2 MC12
2 MC13
1.8
1.6 MC15
1.4
1.2
1
0.8
0.6
0.4
0.2
0
1800 2400 3000 3600 4200
Span (mm)

Figure 8.27: Moment capacity ratio of mullion couples with and without a clip
versus span based on FEA under positive wind action

Effect of using continuous clip

Mullion couple MC5 with and without continuous clip (Figure 8.15) was analysed
under negative wind action in Section 8.3.2. Analyses were performed on the same
mullion couple with and without continuous clip in order to investigate the influence of
continuous clip system on the structural behaviour and capacity of the mullion couple
under positive wind action. In this case, the span used was 4200 mm since the mullion
couple with the 3000 mm span is subject to local buckling, and thus the real effect of
clip cannot be evaluated. Figures 8.28(a) and (b) show the predicted failure modes of
the mullion couple with and without the continuous clip. It is evident from Figure
8.28(b) that the use of continuous clip prevents separation of the male and female
mullions, which is the same as that obtained when a 100 mm clip was used. However,
the use of continuous clip did not increase the capacity significantly as that observed
under negative wind action. That is, the mullion couple with continuous clip increased
the capacity by 10% (Figure 8.29), whereas it was 4% when a 100 mm clip was used
(Table 8.4). Therefore, the clip systems (continuous clip or separate clip parts) cannot
be used to increase the ultimate bending capacity of mullion couples, but they can
improve the serviceability performance of the mullion couples under positive wind
action.

8-25
 Numerical investigation of mullion couples with clip and/or hollow systems

Opening up of
mullion couple

(a) Mullion couple without clip system

No separation between
mullions

(b) Mullion couple comprising continuous clip

Figure 8.28: Predicted failure modes of mullion couple MC5 with and without
continuous clip system under positive wind action

10
9
8
7
Moment (kNm)

6
5
4
3
2 Without clip
1 With continuous clip
0
0 20 40 60 80 100 120 140
Y-axis displacement (mm)

Figure 8.29: Moment versus displacement curves of mullion couple MC5 with and
without continuous clip system under positive wind action

8-26
 Numerical investigation of mullion couples with clip and/or hollow systems

8.4 Mullion couples with hollow system

8.4.1 General

Mullion couples without any clip or hollow systems are more commonly used in facade
applications. However, hollow systems, which can be hollow stiffeners inserted inside
the open mullion couples or hollow mullions as shown in Figures 8.30(a) and (b),
respectively, are sometimes used in order to increase the ultimate moment capacity of
the mullion couples. This is because the open mullion couples fail by lateral-torsional
buckling at lower stress levels, particularly under negative wind action, when their
unbraced length is increased. For instance, the moment capacity of mullion couple
MC5 with a span of 4200 mm predicted using FEA under negative wind action is 1.82
kNm (Chapter 6), and the respective yield moment capacity is 7.48 kNm (Table 8.6).
This implies that only 24% of the yield capacity is utilized by this mullion couple when
the span length of 4200 mm is used. If mullion couple MC5 is to be used and the
required design moment capacity is greater than 1.82 kNm, the mullion couple needs to
be strengthened. This could be done by using a hollow stiffener as shown in Figure
8.30(a). On the other hand, if a new mullion couple section is to be extruded and used, it
could be designed in a similar manner to the mullion couple sections as shown in Figure
8.30(b).

Hollow male Hollow female

Hollow stiffener

(a) Hollow stiffener (b) Hollow mullions

Figure 8.30: Hollow stiffener and hollow mullions

To date, the effect of using hollow system on the structural behaviour and capacity of
mullion couples has not been studied. Therefore, analyses were performed on mullion
couples MC2 and MC5 comprising: 1) hollow male and female mullions, 2) hollow
male and open female mullions, in order to investigate the effect of hollow systems on

8-27
 Numerical investigation of mullion couples with clip and/or hollow systems

the structural behaviour and capacity of mullion couples. It should be noted that hollow
female and open male mullions were not considered since they are not commonly used
in practice. One of the main concerns of using hollow mullions is regarding the
difficulties in fabrication of the frame. That is, the use of hollow system will increase
the workmanship of making the screw connections shown in Figure 8.31, since hollow
mullions need additional cutting/machining work. This prevents the designer from
choosing the hollow mullions in unitized façade constructions.

Screws

Figure 8.31: Screw connections in mullion couples

MC5 MC2 MC5 MC2

(a) Both hollow (b) Hollow male

Figure 8.32: Mullion couples with hollow systems used in analyses

Kesawan et al. (2018) conducted a FEA based study on a mullion couple with different
types of hollow system, and found that it is efficient to have a hollow system similar to
that shown in Figure 8.32(a). In fact, this type of hollow system eliminates the
difficulties associated with facade construction, such as screw connections, mentioned
earlier. Therefore, the hollow system shown in Figure 8.32(a) is mainly considered in
this study.

8-28
 Numerical investigation of mullion couples with clip and/or hollow systems

8.4.2 Mullion couples with hollow under negative wind action

Mullion couples with hollow male and female mullions

The mullion couples comprising hollow male and female mullions were analysed under
negative wind action, where the thicknesses of the hollow part considered were 1.8, 2, 3
and 4 mm. Figures 8.33(a) and (b) show the moment versus displacement curves of the
mullion couples MC5 and MC2, respectively. It is evident from Figures 8.33(a) and (b)
that the addition of a hollow system to the open mullion sections significantly increased
their moment capacities. Mullion couples MC5 and MC2 showed an increment of 452%
and 491%, respectively, even with an addition of 1.8 mm thick hollow part. However,
the stiffness of the mullion couples did not show any noticeable variations since the
hollow part is located closer to the neutral axis, and thus the second moment of area of
the mullion sections did not increase significantly (refer Table 8.5).

12 6
11
10 5
9
Moment (kNm)

Moment (kNm)

8 4
7
6 3
5
4 Open mullions 2 Open mullions
Hollow - 1.8 mm Hollow - 1.8 mm
3
Hollow - 2 mm Hollow - 2 mm
2 Hollow - 3 mm 1
Hollow - 3 mm
1 Hollow - 4 mm Hollow - 4 mm
0 0
0 -50 -100 -150 -200 -250 0 -30 -60 -90 -120 -150
Y-axis displacement (mm) Y-axis displacement (mm)
(a) Mullion couple MC5 (b) Mullion couple MC2
Figure 8.33: Moment versus displacement curves of mullion couples MC5 and
MC2 with different hollow sizes under negative wind action

It is evident from Table 8.5 that torsional constant (J) of the mullions increased by more
than 44 and 16 times for the mullion couples MC5 and MC2, respectively. This helped
the hollow mullion couples to achieve a higher capacity than that with the respective
open mullion couples. That is, the capacity of mullion couples MC5 and MC2 with
hollow mullions are more than 5.5 and 5.9 times the capacity of the respective open
mullion couples, respectively. Furthermore, the moment capacity per unit area (Mt/ATotal
in Table 8.5) of the mullion couples with and without hollow were calculated in order to

8-29
 Numerical investigation of mullion couples with clip and/or hollow systems

evaluate the efficiency of the mullion couples. This shows that for the mullion couple
MC5 the material is efficiently used when 1.8 mm hollow is added to the mullion
couples MC5 (Mt/ATotal ratio is 6.78, which is the highest). For the mullion couple MC2
addition of 3 mm hollow gave the highest efficiency (3.62), which is only about 7%
higher than that obtained with the 1.8 mm hollow (3.39). Therefore, mullion couples
with 1.8 mm thick hollow part were used in the investigations conducted further. It
should be noted that the addition of a 1.8 mm hollow part increased the material usage
by 25% and 22% for mullion couples MC5 and MC2, respectively, giving significant
capacity increment (Table 8.5).

Table 8.5: Moment capacities of the mullion couples with different hollow sizes
based on FEA under negative wind action
t A Ixx x E5 J Mt Jt / Mt / Mt / ATotal
Mullion 2 4
4
(mm) (mm ) (mm ) (mm ) (kNm) Jt=0 Mt=0 (Nmm/mm2)
Mullion couple MC5
M 614 17.94 998 1.0
0 1.82 1.0 1.54
F 570 16.85 1142 1.0
M 762 18.52 50407 50.5
1.8 10.05 5.5 6.78
F 721 17.50 50637 44.3
M 778 18.58 54357 54.5
2 10.22 5.6 6.74
F 738 17.55 54403 47.6
M 861 18.87 71146 71.3
3 10.83 6.0 6.44
F 820 17.83 71435 62.6
M 943 19.16 84879 85.0
4 11.05 6.1 5.99
F 902 18.10 85234 74.6
Mullion couple MC2
M 511 6.32 939 1.0
0 0.68 1.0 0.70
F 460 5.68 848 1.0
M 619 6.72 15009 16.0
1.8 4.02 5.9 3.39
F 568 6.07 15040 17.7
M 631 6.76 16301 17.4
2 4.19 6.2 3.46
F 580 6.11 16250 19.2
M 691 6.93 22094 23.5
3 4.82 7.1 3.62
F 640 6.27 22065 26.0
M 752 7.08 27365 29.1
4 5.24 7.7 3.61
F 700 6.42 27406 32.3
Note:
t - Thickness of the hollow part, A – Area, Ixx - Second moment of area, J - Torsional
constant, Mt - Moment capacity prediction of mullion couple with hollow thickness t (t= 0,
1.8, 2, 3 and 4 mm), ATotal - Summation of the area of male and female mullions, M - Male
mullion and F - Female mullion

8-30
 Numerical investigation of mullion couples with clip and/or hollow systems

Table 8.6 shows a comparison made between the moment capacities of the mullion
couples predicted by ADM (AA, 2015) and FEA in order to provide an improved
understanding of mullion couples with hollow system. In ADM (AA, 2015) capacity
calculations the yield capacity of these mullions was taken as their section capacity
(refer Chapter 6 for more details). In addition, the moment capacities predicted by FEA
and the respective yield moment capacities were also compared (Table 8.6). It is evident
from Table 8.6 that moment capacity increments obtained for mullion couple MC5 with
hollow based on ADM (AA, 2015) are in the similar order as that obtained based on
FEA (Table 8.5). That is, for instance, increments obtained for mullion couple MC5
with 1.8 mm hollow based on ADM (AA, 2015) and FEA are 5.41 and 5.5, respectively.
Moment capacity increments obtained for mullion couple MC2 based on ADM (AA,
2015) are significant (3.33 to 4.44), although they are less than that obtained based on
FEA. This could be due to the fact that ADM (AA, 2015) prediction for mullion couple
MC2 with open mullions (0.79) is higher than the FEA prediction (0.68), but ADM
(AA, 2015) predictions are conservative when hollow mullions are used (Table 8.5).
Overall, it can be stated that ADM (AA, 2015) capacity predictions for hollow mullions
are conservative irrespective of whether mullion couples show interaction or not.

Table 8.6: Moment capacities of the mullion couples with different hollow sizes
based on FEA and ADM (AA, 2015) under negative wind action

t MFEA MADM My MADM,t / MADM / MFEA /

(mm) (kNm) (kNm) (kNm) MADM, t=0 MFEA My
Mullion couple MC5
0 1.82 0.86 7.48 1.00 0.47 0.24
1.8 10.05 4.65 7.99 5.41 0.46 1.26
2 10.22 4.84 8.03 5.63 0.47 1.27
3 10.83 5.61 8.25 6.52 0.52 1.31
4 11.05 6.07 8.46 7.06 0.55 1.31
Mullion couple MC2
0 0.68 0.79 4.07 1.00 1.16 0.17
1.8 4.02 2.63 4.66 3.33 0.65 0.86
2 4.19 2.74 4.72 3.47 0.65 0.89
3 4.82 3.19 4.99 4.04 0.66 0.97
4 5.24 3.51 5.24 4.44 0.67 1.00
Note:
MFEA – FEA prediction, MADM – ADM (AA, 2015) prediction and My – Yield
moment capacity

Table 8.6 shows that mullion couple MC5 achieved its yield moment capacity with an
addition of even 1.8 mm hollow (MFEA / My > 1). In this case, when the open mullion
8-31
 Numerical investigation of mullion couples with clip and/or hollow systems

couple is used only 24% of its yield moment capacity is utilized. But the addition of 1.8
mm hollow, which increased the material usage only by 25%, helps to utilize the whole
section. This is the same for mullion couple MC2, where 86% of the yield capacity is
utilized when 1.8 mm hollow is used compared to 17% for the open mullion couple.
Therefore, hollow system can be used to make the mullions more structurally efficient.

The hollow mullion couples were analysed with the lowest spring constant (k=1/40
N/mm/mm) for the lateral restraint considered in Chapter 6 or 7. Figures 8.34(a) and (b)
show a comparison of the moment versus displacement curves of the mullion couples
MC5 and MC2, respectively. It is evident from the figures that the stiffness of the
mullion couples was not affected by the effectiveness of lateral restraint. For the
mullion couple MC5, the reduction in the moment capacity is about 16%, whereas it is
about 15% for the mullion couple MC2. These reductions are in the similar range as that
obtained for the mullion couples with open section subject to negative wind action
(Chapter 6). Therefore, the effect of lateral restraint on the bending moment capacity of
mullion couples with hollow systems can be incorporated using the guidelines given in
Chapter 6.

11 5
10
9 4
8
Moment (kNm)

Moment (kNm)

7 3
6
5
4 2
3
2 Full restraint 1 Full restraint
1 k=1/40 N/mm/mm k=1/40 N/mm/mm
0 0
0 -40 -80 -120 -160 -200 0 -20 -40 -60 -80 -100
Y-axis displacement (mm) Y-axis displacement (mm)
(a) Mullion couple MC5 (b) Mullion couple MC2
Figure 8.34: Moment versus displacement curves of hollow mullion couples with
full and spring lateral restraints under negative wind action

Mullion couples with hollow male and open female mullions

Figures 8.35(a) and (b) show the predicted failure modes, whereas Figures 8.36(a) and
(b) show the moment versus displacement curves of the mullion couples. It is evident
from Figure 8.36(a) that the capacity of the mullion couple MC5 increased significantly
8-32
 Numerical investigation of mullion couples with clip and/or hollow systems

(210%) when hollow male mullion was used, but it is less than the capacity of the
mullion couple with both hollow. However, for the mullion couple MC2, the capacity
did not increase when hollow male mullion was used. This is because, this mullion
couple shows individual rotations (no interaction) under negative wind action since its
shear center lies inside the line of action of the loading (Figure 8.19), thus making one
mullion stronger would not be beneficial (Figure 35(b)). Furthermore, in all the cases
the stiffness of the mullion couples did not show any noticeable difference since the
addition of such hollow did not change the second moment of area of the mullion
sections significantly (Table 8.5). Based on these it can be stated that the capacity of
only the mullion couples that show interaction could be increased by using one hollow
mullion.

Hollow male
Interaction mullion

No interaction
Hollow male Open female
mullion Open female
mullion
mullion

(a) Mullion couple MC5 (b) Mullion couple MC2

Figure 8.35: Predicted failure modes of mullion couples MC5 and MC2 with a
hollow mullion under negative wind action

12 5

10 4
Moment (kNm)

Moment (kNm)

8
3
6
2
4
Open mullions Open mullions
2 Hollow male 1
Hollow male
Both hollow Both hollow
0 0
0 -40 -80 -120 -160 -200 0 -20 -40 -60 -80 -100
Y-axis displacement (mm) Y-axis displacement (mm)
(a) Mullion couple MC5 (b) Mullion couple MC2
Figure 8.36: Moment versus displacement curves of mullion couples MC5 and
MC2 with a hollow mullion under negative wind action

8-33
 Numerical investigation of mullion couples with clip and/or hollow systems

8.4.3 Mullion couples with hollow under positive wind action

Mullion couples with hollow male and female mullions

In the last section it was found that the mullion couples showed significant capacity
increment under negative wind action even with an addition of 1.8 mm thick hollow
part. In this section, the same mullion couples with the 1.8 mm thick hollow part is
investigated under positive wind action. The influence of the effectiveness of lateral
restraint was first investigated, where the mullion couples with hollow male and female
mullions were analysed with full and spring (1/40 N/mm/mm) lateral restraints (Chapter
6). Figures 8.37(a) and (b) show the moment versus displacement curves of mullion
couples MC5 and MC2, respectively. The capacity reduction for the mullion couple
MC5 is about 12%, whereas it is about 6% for the mullion couple MC2, which is
significantly less than that of the open mullion couple sections. This is because, for
instance, capacity reduction obtained for mullion couple MC5 in Chapter 7 was 50%,
whereas it was 12% for the same mullion couple with hollow. Therefore, the effect of
lateral restraint on the capacity of hollow mullion couples is less than that for open
mullion couples under positive wind action.

12 6

10 5
Moment (kNm)
Moment (kNm)

8 4

6 3

4 2

2 Full restraint 1 Full restraint

k=1/40 N/mm/mm k=1/40 N/mm/mm
0 0
0 30 60 90 120 150 180 210 240 0 20 40 60 80 100 120 140 160
Y-axis displacement (mm) Y-axis displacement (mm)
(a) Mullion couple MC5 (b) Mullion couple MC2
Figure 8.37: Moment versus displacement curves of hollow mullion couples with
full and spring lateral restraints under positive wind action

Mullion couples with hollow male and open female mullions

Mullion couples MC5 and MC2 with a hollow were also analysed under positive wind
action. It is evident from Figure 8.39(a) that the capacities of mullion couple MC5 did
not vary significantly between the different couple systems (open mullions, hollow male
8-34
 Numerical investigation of mullion couples with clip and/or hollow systems

and both hollow) when compared to mullion couple MC2. The capacity of mullion
couple MC2 increased significantly when a hollow mullion was used (82%), but the
capacity of the mullion couple with a hollow is still lower than the capacity of the
mullion couple with both hollow (12% less). This is because, in mullion couple MC5
male and female mullions interact (Figures 8.24(a) and 8.38(a)), whereas in mullion
couple MC2 both mullions rotate together without any interaction (Figure 8.26(a)), thus
stiffening the male mullion will help to share the load from the female mullion to male
mullion, thus achieving a higher capacity (Figures 8.38(b) and 8.39(b)). Overall, the
capacity of mullion couples could be increased significantly by using at least one
hollow mullion under positive wind action. It should be noted that for mullion couples
similar to MC2 it will not be beneficial since negative wind action will be the governing
(refer Section 8.4.2).

Open female
Open female
mullion
mullion
Hollow male Load sharing
mullion region
No interaction
Interaction
Hollow male
mullion

(a) Mullion couple MC5 (b) Mullion couple MC2

Figure 8.38: Predicted failure modes of mullion couples MC5 and MC2 with a
hollow mullion under positive wind action

12 6
10 5
Moment (kNm)

Moment (kNm)

8 4
6 3
4 2 Open mullions
Open mullions
2 Hollow male 1 Hollow male
Both hollow Both hollow
0 0
0 30 60 90 120 150 180 210 240 0 20 40 60 80 100 120 140 160
Y-axis displacement (mm) Y-axis displacement (mm)
(a) Mullion couple MC5 (b) Mullion couple MC2
Figure 8.39: Moment versus displacement curves of mullion couples MC5 and
MC2 with a hollow mullion under positive wind action

8-35
 Numerical investigation of mullion couples with clip and/or hollow systems

8.5 Mullion couples with clip and hollow systems

8.5.1 General

It is possible to increase the ultimate moment capacity of the mullion couples when one
hollow mullion and a clip are used. It is because, for instance, when mullion couple
MC2 with hollow male and open female mullions were analysed the female mullion
showed individual rotation (Section 8.4.2), thus using a clip would prevent it from
behaving individually, giving a higher capacity. Therefore, mullion couples MC5 and
MC2 with a hollow and a clip at their mid-span were analysed in this section in order
investigate the effect of using clip and hollow systems. Figures 8.40(a) and (b) show
mullion couples MC5 and MC2 with one hollow mullion and a clip, respectively.

Clip Clip

(a) MC5 (b) MC2

Figure 8.40: Mullion couples with a hollow and a clip

8.5.2 Mullion couples with hollow and clip under negative wind action

Figures 8.41(a) and (b) show the predicted failure modes of mullion couples MC5 and
MC2 with a hollow and a clip, respectively, under negative wind action. It is evident
from the figures that the separation of the female mullion at mid-span is prevented by
the clip. In fact, this has increased the capacity of the mullion couple significantly,
particularly for mullion couple MC2 (307%). The increment obtained for mullion
couple MC5 is not as high as that obtained for mullion couple MC2 since the male and
female mullions of this mullion couple show interaction (Figures 8.42(a) and (b)).
Therefore, capacity increments that can be obtained by using a clip for mullion couples
with a hollow in which mullions show interaction could be in the same order as that

8-36
 Numerical investigation of mullion couples with clip and/or hollow systems

obtained for such mullion couples without any hollow system. However, the capacity of
mullion couples in which male and female mullions rotate individually could be
increased significantly by using hollow and clip systems.

Clip
Clip Open
Open female
female

Hollow No separation at No separation at

Hollow
male mid-span mid-span
male

(a) Mullion couple MC5 (b) Mullion couple MC2

Figure 8.41: Predicted failure modes of mullion couples MC5 and MC2 with
hollow and clip systems under negative wind action

7 3.0
6 2.5
5
Moment (kNm)

Moment (kNm)

2.0
4
1.5
3
1.0
2
1 Without clip 0.5 Without clip
With clip With clip
0 0.0
0 -10 -20 -30 -40 -50 -60 0 -20 -40 -60 -80
Y-axis displacement (mm) Y-axis displacement (mm)
(a) Mullion couple MC5 (b) Mullion couple MC2
Figure 8.42: Moment versus displacement curves of mullion couples MC5 and
MC2 with hollow and clip systems under negative wind action

8.5.3 Mullion couples with hollow and clip under positive wind action

It is evident from Figure 8.38(a) that mullion couple MC5 with a hollow opened up
under positive wind action. In the last section it was found that a significant capacity
increment, particularly for mullion couple MC2 with a hollow, could be achieved when
a clip is used. Therefore, it is interesting to investigate the effect of using hollow and
clip system under positive wind action, and thus in this section analyses were performed
on mullion couples MC5 and MC2 comprising a hollow and a clip. Figures 8.43(a) and

8-37
 Numerical investigation of mullion couples with clip and/or hollow systems

(b) show the predicted failure modes of mullion couples MC5 and MC2 with a hollow
and a clip, respectively. The addition of a clip prevented the opening up of mullion
couple MC5 (Figures 8.38(a) and 8.43(a)), which has resulted in a slight (13%)
increment in capacity as shown in Figure 8.44(a). Furthermore, it is evident from
Figure 8.44 (b) that the capacity of mullion couple MC2 comprising a hollow did not
increase when a clip was used since the mullion couple even without a clip did not open
up (Figures 8.38(b)). Therefore, the capacity of the mullion couples with or without
hollow system cannot be increased by using clips under positive wind action.

No opening up of No opening up of
mullion couple mullion couple

(a) Mullion couple MC5 (b) Mullion couple MC2

Figure 8.43: Predicted failure modes of mullion couples MC5 and MC2 with
hollow and clip systems under positive wind action

12 6

10 5
Moment (kNm)
Moment (kNm)

8 4

6 3

4 2

2 Without clip 1 Without clip

With clip With clip
0 0
0 30 60 90 120 150 180 210 240 0 20 40 60 80 100 120 140 160
Y-axis displacement (mm) Y-axis displacement (mm)
(a) Mullion couple MC5 (b) Mullion couple MC2
Figure 8.44: Moment versus displacement curves of mullion couples MC5 and
MC2 with hollow and clip systems under positive wind action

8-38
 Numerical investigation of mullion couples with clip and/or hollow systems

8.6 Serviceability performance of mullion couples

In the structural design of mullions both ultimate limit state (ULS) and serviceability
limit state (SLS) failure criteria need to be considered (SA 2002 and SA, 1996). So far,
structural behaviour and ultimate capacity of mullion couples with and without clip
and/or hollow system have been investigated in detail, which was the main focus of this
study. However, in this section a brief discussion on the serviceability performance of
mullion couples is given since it would enhance the knowledge of mullion design. It
should be noted that the displacement characteristics of mullion couples discussed in
this section are based on the axis descriptions shown in Figure 8.45(a).

Y-axis displacement (deflection

in the loading direction) X-axis displacement (lateral
deflections)

(a) Axes of displacement of mullions

12
Theoretical deflection curve
10 Maximum
ULS moment
Moment (kNm)

6 5.20

4 2.28
SLS moment Open mullions
2 Hollow male - 1.8 mm thick
-38.3 Both hollow - 1.8 mm thick
0
0 -16.8 -40 -80 -120 -160 -200
Y-axis displacement (mm)
(b) Moment versus Y-axis displacement curves
Figure 8.45: Serviceability performance of mullion couple MC5 with and without
hollow systems under negative wind action (4200 mm span)

8-39
 Numerical investigation of mullion couples with clip and/or hollow systems

SLS criteria for the mullions specified in AS 1170.0 (SA, 2002) is span/250. This limit
is applied to the deflection of mullions in the direction of the wind loading
(perpendicular to the surface of the facade), which is the Y-axis displacement shown in
Figure 8.45(a). Furthermore, ULS and SLS wind speeds required for the wind pressure
calculations of commonly used building types (Importance level 2 based on BCA) are
V500 and V25, which are the wind speeds for 500 and 25 year return periods. The
maximum ratio between the ultimate and serviceability limit state pressures PULS and
PSLS, determined based on AS 1170.2 (SA, 2011) is 2.28 (PULS / PSLS = [V500]2 / [V25]2
since P DV2, where P and V are pressure and wind velocity, respectively). This means,
for instance, if the design SLS pressure is 2.28 kPa, then the design ULS pressure is
5.20 kPa. In other words, if the SLS design moment is 2.28 kNm, then the ULS design
moment is 5.20 kNm (Moment is proportional to load/pressure).

In order to understand how the ULS and SLS criteria are related to the mullion design,
the moment versus displacement curve obtained from FEA for mullion couple MC5 in
Section 8.4.3 was considered. The theoretical deflection curve determined for the
mullion couple with both hollow as per Appendix C.6 is also plotted in the same graph
as shown in Figure 8.45(b). In addition, for this 4200 mm span mullion couple, the
design SLS and ULS moments are also plotted in the same figure (red and black broken
lines).

Firstly, it is evident from Figure 8.45(b) that theoretical and FEA based stiffness of the
mullion couple agrees well, thus the SLS deflection of the mullion couples in the
direction of loading (Y-axis direction) can be theoretically determined using their
section properties. Unlike the displacement behaviour of mullion couples in the
direction of wind load, lateral behaviour (in the X-axis direction) is more complex due
to the influence of second order effects such as seal friction. For instance, Figure 8.3(b)
shows that the displacement behaviour of the mullion couple in the X-axis direction was
not captured as accurately as that in the direction of wind load (Y-axis), which is shown
in Figure 8.3(a). This was the same for the mullion couple without any clip or hollow
system investigated in Chapter 5. In this case, if the SLS capacity based on lateral
displacement behaviour of the mullion couple is determined using FEA predictions, it
will be conservative. In fact, no SLS limits are specified in the design standard for the
lateral displacement behaviour of mullion couples.

8-40
 Numerical investigation of mullion couples with clip and/or hollow systems

In Section 8.4.2 it was found that adding a hollow part to the male and female mullions
increased the capacity of the mullion couples significantly. For instance, the capacity of
the mullion couple MC5 with even a hollow part of 1.8 mm thick was 5.5 times the
capacity of the mullion couple without any hollow (Table 8.5). Similar order of
increment was obtained when a 4 mm thick hollow part was added. But, the addition of
the hollow parts increased the use of material by 25% and 56% for the mullion couples
with 1.8 mm and 4 mm thick hollow part, respectively. At this point it can be
questioned which hollow thickness should be chosen or whether the material will be
utilized efficiently if the mullion couple with hollow part is used. In order to give an
answer to these questions, the moment versus displacement curves of mullion couple
MC5 with and without hollow system predicted by FEA was compared with the SLS
criteria specified in AS 1170.2 (SA, 2002).

It is evident from Figure 8.45(b) that the design of mullion couple MC5 without any
hollow is governed by both SLS and ULS failure criteria. When hollow male is used,
ULS criteria deemed to satisfy or it would marginally fail. It is because a reduction
factor for the effectiveness of lateral restraint and a capacity reduction factor should be
applied with the FEA predictions in mullion designs, which could fail the mullion
couple with hollow male. When the mullion couple with both hollow was used, the
capacity of the mullion couple satisfies the SLS and ULS design requirements. In this
case, the capacity of the mullion couple considering the reduction of 16% obtained for
the lowest spring constant (Figure 8.34(a)) and a capacity reduction factor of 0.9
specified in AS 1664 (SA, 1996a), which is 10.05 x 0.86 x 0.9 = 7.6 kNm, is higher than
the ULS requirement (5.20 kNm). For the span considered, capacity of the mullion
couple more than 5.20 kNm will not be utilized in any design since the ULS design
bending moment is the maximum considering the maximum SLS failure criteria.
Therefore, based on the above discussions it can be stated that both the ULS and SLS
performance of mullion couples should be considered if efficient and economical
mullion sections need to be produced.

8-41
 Numerical investigation of mullion couples with clip and/or hollow systems

8.7 Conclusion

Full and simple FE models of the mullion couple with clip at its mid-span (Tests 3 and
4a) were developed under both negative and positive wind actions using the validated
FEA approaches. Structural behaviour and capacity of the mullion couple predicted by
FEA were compared with the test results and observations. The comparisons showed
that the predictions from both the full and simple FE models agreed reasonably well
with the test results. This further verified that the FEA approaches developed in this
study could be used to analyse any kind of mullion couple systems used in unitized
aluminium facades. Furthermore, considering the time and computing resources, the
simple FE model was chosen for the parametric study as was done in Chapters 6 and 7.

Firstly, it was found that the ultimate capacity of mullion couples without any hollow
systems cannot be increased significantly using clips under negative wind action. The
ADM (AA, 2015) predictions based on currently used assumptions were higher than the
FEA predictions, thus unsafe. This means that the currently used assumption (assuming
reduced unbraced length considering the presence of clip) is not applicable. Similarly,
the use of continuous clips also cannot increase the capacity of the mullion couples
significantly. This was the same for the mullion couples without any hollow systems
under positive wind action. Therefore, clip systems cannot be used to improve the
moment capacity of mullion couples without any hollow systems. However,
serviceability performance of the mullion couples, such as opening, could be improved
by using clips.

Mullion couples with hollows of varying thickness were investigated under negative
wind action, and the results showed that even the addition of hollow with the smallest
thickness (1.8 mm) increased the capacity of the mullion couples significantly. The
capacity of the mullion couples predicted by both FEA and ADM (AA, 2015) increased
significantly under negative wind action. In addition, the effect of lateral restraint was
significantly less compared to that obtained for the respective open mullion couples,
particularly under positive wind action. When mullion couples comprising only one
hollow mullion were analysed under negative wind action, the capacity of the mullion
couple that show interaction increased. In this case, the capacity of the mullion couple
in which the shear centre lies inside the line of action of the loading did not increase.
But, under positive wind action the use of a hollow increased the capacity of the

8-42
 Numerical investigation of mullion couples with clip and/or hollow systems

mullion couples, where the mullion couple in which the shear centre lies inside the line
of action of the loading increased significantly.

When the mullion couples comprising one hollow mullion were analysed with a clip at
their mid span under negative wind action, the capacity of the mullion couples
increased. In this case a significant increment was obtained for the mullion couple in
which the shear centre lies inside the line of action of the loading. However, the
increments obtained under positive wind action were not significant for both mullion
couples. Therefore, this implies that the capacity of the mullion couples could be
improved by using a hollow and clip, particular for the mullion couples in which the
shear centre of the mullions lies inside the line of action of the loading.

Finally, the serviceability performance of mullion couples in relation to the structural

design of mullions was discussed. It was found that the deflection of mullion couples in
the direction of wind loading can be determined theoretically using their section
properties in order to check the adequacy of SLS criteria specified in design standards.
In addition, the SLS capacities for the lateral displacement behaviour of mullion couples
cannot be determined accurately using FEA as for the direction of wind load since it is
influenced by second order effects such as seal friction. This is a limitation of this study.
In fact, no SLS limits are specified in the aluminium design standards for the lateral
displacement behaviour of mullion couples. Furthermore, it was found that both the
ULS and SLS performance should be considered if efficient and economical mullion
sections need to be produced.

8.8 Recommendations

Based on the study reported in this chapter, the following recommendations are made in
relation to the structural design of mullion couples with clip and/or hollow systems
subjected to both negative and positive wind actions.

a. Mullion couples comprising open mullions with and without clip system under
wind actions:

The use of clip, in general, did not increase the ultimate moment capacity of the mullion
couples significantly under both negative and positive wind actions. In addition, the
capacity of the mullion couples with a clip at their mid-span predicted using the current
design approach was unsafe under negative wind action. Therefore, it is recommended

8-43
 Numerical investigation of mullion couples with clip and/or hollow systems

that clip system and the currently used assumption regarding clip system cannot be
considered in the design of such mullions. If the mullion couples need to be designed
economically, finite element analysis based solutions can be used following the
recommendations given in Chapters 6 and 7.

b. Mullion couples with hollow system under wind actions:

The moment capacity of the mullion couples can be significantly increased by using
hollow male and female mullions under both negative and positive wind actions. For the
mullion couples with hollow male and female mullions, ADM (AA, 2015) predictions
also increased significantly, where they were conservative than FEA predictions
irrespective of the mullions’ shear centre locations. This means that ADM (AA, 2015)
can be used safely. When only one hollow mullion is used, the moment capacity of the
mullion couples in which the shear centre of the mullions lies outside the line of action
of loading can be increased, but needs to be determined using validated FEA
approaches. In this case, the capacity of the mullion couples in which the shear centre of
the mullions lies inside the line of action of loading cannot be increased under negative
wind action, but can be increased under positive wind action. Overall, it is
recommended to use validated FEA to predict the capacity of mullion couples with one
hollow mullion.

c. Mullion couples with clip and hollow systems under wind actions:

The moment capacity of the mullion couples with a hollow can be increased using clips
under negative wind action, particularly for the mullion couples in which the shear
centre lies inside the line of action of the loading. In this case, suitable FEA approaches
should be used to determine the capacity of the mullion couples. The moment capacity
of the mullion couples with a hollow did not increase significantly under positive wind
action when a clip was used. However, it can be used to improve the serviceability
performance of the mullion couples, such as openings.

d. Serviceability performance of mullion couples:

The deflection of mullion couples in the direction of wind loading can be determined
theoretically using their section properties. When developing new mullion sections SLS
criteria also should be considered to achieve efficient and economical sections.

8-44
 Conclusions and recommendations

Chapter 9

9 CONCLUSIONS AND RECOMMENDATIONS

This thesis has presented a detailed investigation on the structural behaviour and capacity
of aluminium mullion couples used in unitized building facades subject to negative
(suction) and positive (pressure) wind actions. It included both experimental and
numerical investigations of a range of commonly used aluminium mullion couples.
Firstly, the mechanical property tests of 6063-T6 aluminium alloy mullions were
conducted to determine the stress-strain characteristics of aluminium alloy including its
yield strength, ultimate strength and Young’s modulus. Full-scale wind loading tests were
then performed on unitized facades made of mullion couples with and without clip and
hollow systems. These tests provided a very good understanding of the structural
behaviour of mullion couples, and a solid base for the finite element analysis (FEA) based
investigations. Full and simple finite element (FE) models of the tested mullion couples
were developed and were validated against the full-scale test results, which is a significant
achievement in FE modelling of unitized facade mullion couples. Thereafter, a
comprehensive numerical parametric study was conducted on different types of mullion
couples, which included both captive and structural glazing mullions. The effects of using
clip and hollow systems were also investigated in detail. Finally, the capacity predictions
from the currently used design approaches based on ADM (AA, 2015) were compared
with the FEA results, and suitable recommendations were made for economical and safe
designs of unitized facade mullions under wind actions.

Overall, this research has developed comprehensive experimental and numerical capacity
data for a range of complex-shaped aluminium mullion couples used in unitized building
facades under wind actions including those with clip and hollow systems. More than 400
mullion couple analyses were performed, where 20 types of mullion couples with varying
spans were considered in this research. It has significantly improved the knowledge and
understanding of the structural behavior of aluminium mullion couples under wind
actions, based on which it has recommended suitable design methods. Design charts of
40 commercially used mullion couples have also been developed for use by facade
engineers.

Following conclusions and recommendations have been drawn based on the specific
topics investigated in this research.

9-1
 Conclusions and recommendations

9.1 Conclusions

9.1.1 Mechanical property tests

x The yield and ultimate strengths obtained from tensile coupon tests were higher
than the minimum values specified in AS/NZS 1664 (SA, 1997).
x The average modulus of elasticity obtained from bending tests was 69703 MPa,
which is slightly higher (0.6%) than that given in AS/NZS 1664 (SA, 1997a,b).
x Ramberg-Osgood model (Ramberg and Osgood, 1943) with strain-hardening
exponent (n) determined based on the model given in CEN (2007) for plastic range
analysis case was found to be the suitable model to define the stress-strain
characteristics of 6063-T6 aluminium alloy mullions.
x Equation given in CEN (2007) to predict the ultimate fracture strain (εu) and the
equation proposed by Su et al. (2014) to predict the strain at ultimate tensile strength
(εmax) were calibrated using the tensile coupon test results.
x A new equation was proposed to predict the 0.1% proof strength (f0.1) based on
0.2% proof strength (f0.2).

9.1.2 Full-scale experimental study

x When the mullion couple without any clip or hollow systems was subjected to
negative wind action, the female mullion separated from the male mullion after
certain loading, leading to failure at a lower capacity than expected (850 Pa versus
yield capacity of 3170 Pa).
x During positive wind action, mullions slid on the aluminium sheets leading to the
failure of mullion couple, and the capacity obtained from the test was 15% less than
the yield capacity of the mullion couple (2700 Pa versus yield capacity of 3170 Pa).
x When the mullion couple with a clip at its mid-span was subjected to negative wind
action, the capacity was only about 30% higher than that obtained without clip
(1125 Pa versus 850 Pa). Significant separation between the mullions was observed
at about 450 mm away from the mid-span as the presence of clip limited the
separation at mid-span.
x In the full-scale test of mullion couple with a clip at its mid-span under positive
wind action, mullions rotated together and slid on the aluminium sheets until
failure. In this case, the capacity of the mullion couple did not increase (2650 Pa

9-2
 Conclusions and recommendations

versus that obtained without clip of 2700 Pa). Furthermore, the presence of clip at
mid-span prevented the opening of mullion couple observed in the test of the
mullion couple without clip.
x When a hollow stiffener was used under negative wind action the performance of
the mullion couple improved significantly, but the test was not continued until
failure due to premature failure of the test assembly.
x The stiffness of the mullion couple under wind actions was found to be similar
whether a clip was used or not.

9.1.3 Development and validation of FE models of mullion couples without any

clip or hollow systems

x Under negative wind action, structural behaviour and capacity of the tested mullion
couple predicted by both full and simple FE models agreed reasonably well with
the test results. The difference in the ultimate capacities predicted by the simple and
full FE models were about 10% and 15%, respectively, where the FEA predictions
were conservative.
x The capacity of the mullion couple considering the interaction was only 5% higher
than that based on individual mullion analyses, thus the interaction of mullions did
not significantly increase the capacity under negative wind action.
x The simple FE model with full lateral restraint overestimated the capacity of the
mullion couple under positive wind action. An appropriate spring stiffness value
was calculated and used with the simple FE model. The ultimate capacity predicted
in this case agreed well with the test capacity, with a difference of less than 3%.
x The developed full FE model under positive wind action predicted the failure
capacity reasonably well, where it was slightly higher than the test capacity (about
7%). The structural behaviour of the mullion couple was also predicted reasonably
well by the full FE model.
x The capacity of the mullion couple considering the interaction was about 50%
higher than that based on individual mullion analyses, thus the interaction between
mullions significantly influenced the capacity of the mullion couple.

9-3
 Conclusions and recommendations

9.1.4 Parametric study of mullion couples without any clip/hollow systems under
negative wind action

x Full and simple FE models of captive and structural glazing mullion couples were
modelled and analyzed, and the results from both models agreed reasonably well.
x The moment capacity of mullion couples increased when the gap between mullions
was increased (Section 6.5.1), whereas it reduced when the load distribution profile
approached uniform loading. Therefore, using a smaller gap and uniform load
distribution profile will give conservative predictions.
x The effects of lateral restraint provided by the glass panes and load distribution ratio
(the ratio between the loads that are acting on male and female mullions) on the
bending moment capacity of the mullion couples were found to be significant.
x Provision of intermediate transom(s) significantly increased the capacity of mullion
couples. However, the changes in the moment diagram (or moment gradient) did
not seem to have contributed to the capacity increment (Section 6.5.5).
x ADM (AA, 2015) predictions were, in general, less than FEA predictions for the
mullion couples shown in Figure 9.1(a) (Type A), whereas they were unsafe for the
mullion couples shown in Figure 9.1(b) (Type B). Overall, ADM (AA, 2015)
predictions would be either conservative or unconservative.

9.1.5 Parametric study of mullion couples without any clip/hollow systems under
positive wind action

x The moment capacity of mullion couples decreased when the gap between mullions
was increased (Section 7.3.1), thus using a larger gap will give conservative
predictions.
x The effects of lateral restraint and load distribution ratio on the bending moment
capacity of mullion couples were found to be significant as observed under negative
wind action.
x Moment capacities obtained from FEA were less than their respective yield moment
capacity for the majority of the mullion couples even with full lateral restraint.
x ADM (AA, 2015) predictions were significantly less than FEA predictions for all
the mullion couples irrespective of their type (Type A or B in Figure 9.1).

9-4
 Conclusions and recommendations

9.1.6 Parametric study of mullion couples with clip and/or hollow systems

Mullion couples with clip system:

x Under negative wind action, structural behaviour and ultimate capacity of the test
mullion couple with a clip at its mid-span was predicted reasonably well by the
developed simple and full FE models. The difference between the capacities
predicted by test and FEA was about 18% for both models, where FEA predictions
were conservative. Under positive wind action, both simple FE model with the
lateral spring stiffness used in Chapter 5 and full FE model predicted the structural
behavior and capacity reasonably well. The difference between the capacities
predicted by test and simple FE model was about 12% (Test prediction was
conservative), whereas it was 2% for full FE model prediction (FEA prediction was
conservative).
x FE analyses of different mullion couples with clip systems showed that the capacity
of mullion couples cannot be significantly increased using clip system under
negative wind action.
x ADM (AA, 2015) predictions based on currently used assumption (unbraced length
of a mullion couple with a clip at its mid-span is half the span) were higher than the
FEA predictions, thus unsafe.
x The capacity of mullion couples cannot be significantly increased using clip system
under positive wind action, but the opening of mullion couples can be prevented.

Mullion couples with hollow system:

x FEA predictions showed that even the addition of 1.8 mm hollow to the open
mullions significantly increased the capacity of mullion couples. This was the same
observation for ADM (AA, 2015) based predictions as well.
x When hollow male and female mullions were used, the effect of lateral restraint on
the moment capacity was significantly less compared to that obtained for the
respective open mullion couples, particularly under positive wind action.
x When mullion couples comprising only one hollow mullion were analysed under
negative wind action, the capacity of “Type A” mullion couples (Figure 9.1(a))
increased, whereas the capacity of “Type B” mullion couples (Figure 9.1(b)) did
not increase. Under positive wind action, the capacity of both types of mullion
couples increased.

9-5
 Conclusions and recommendations

Mullion couples with clip and hollow systems:

x When mullion couples comprising a hollow mullion and a clip were analysed a
significant capacity increment was obtained for “Type B” mullion couple (307%).
x Despite the types of mullion couples (Type A or B), the maximum capacity
increment obtained under positive wind action was not significant (13%).
x It was found that the theoretical deflection of mullion couples in the direction of
wind loading determined based on their section properties agreed well with FEA
predictions. Furthermore, it was found that both the ULS and SLS (ultimate and
serviceability limit states) performance of mullion couples should be considered if
efficient and economical mullion sections need to be produced.

9.2 Recommendations for the design of mullion couples

In this research study, two major characteristics of unitized façade mullion couples were
found, which are mullion couples that show interaction (Type A) and that show individual
rotation/no interaction (Type B) as shown in Figure 9.1. When the shear centre of the
mullions lies outside the line of action of loading (Type A), mullion couples show
interaction (Figure 9.1(a)). Mullion couples show no interaction (Figure 9.1(b)) when the
shear centre of mullions lies inside the line of action of loading (Type B). Considering
these characteristics of mullion couples, recommendations/design guidelines are given in
this section for economical and safe bending capacity predictions of unitized facade
mullion couples under wind actions. This section provides design guidelines based on the
outcomes of this study to determine the bending moment capacity of mullion couples
used in unitized aluminium facades subject to wind actions. Three methods can be used
to predict the bending moment capacity of a mullion couple, which are full-scale testing,
finite element analysis and design rules. Details of the three different methods of capacity
predictions are given next.

9-6
 Conclusions and recommendations

Shear centre
Shear centre

Line of action of loading Line of action of loading

Individual rotation

Leaning and
rotating together

(a) Mullions showing interaction (Type A) (b) Mullions behaving individually (Type B)

Figure 9.1: Combined and individual behaviour of male and female mullions

9.2.1 Full scale testing

If the ultimate capacity of a mullion couple is to be determined through full-scale testing,

the following guidelines/recommendations shall be followed:

x Compliance with relevant standards (AS 2047 (SA, 2014), AS/NZS 4284 (SA, 2008)
or any other standards)
x Include the components such as infills, seal strips, gaskets/sealants and transoms (if
any)
x Represent the continuity of real unitized facade assembly
x Real support/boundary conditions of unitized facade system shall be used, or else the
support/boundary conditions used should be able to simulate the real scenario
x Complete facade panel should be subjected to wind pressure (positive)/suction
(negative) loading
x Simulate the width of the real unitized facade panels used in construction
x If alternative materials are used for infills (e.g. using aluminium sheets instead of glass
panes), the stiffness of infills used in testing should represent the real scenario
x Appropriate factor of safety shall be used with the capacity obtained from test

9-7
 Conclusions and recommendations

9.2.2. Finite element analysis

If the ultimate capacity of a mullion couple is to be determined through finite element

analysis, the following guidelines/recommendations shall be followed:

x Validated finite element modelling and analysis approaches shall be used

x Idealization of male and female mullions should not affect the interaction behaviour
of the mullion couple
x Support boundary conditions should simulate the real scenario
x Half span of mullion couples can be modelled if geometry, boundary and loading are
symmetric about the mid-span
x Mechanical properties used in modelling shall be either obtained from tests or the
minimum values specified in relevant design standards/manuals. If the mechanical
properties obtained from testing are used, consideration should be given for variations
in the properties between different batches of productions
x If only the mullion couple is modelled (simple FE model), suitable boundary
conditions to represent the glass restraint and loading position shall be used
x If the mullion couple, glass panes and gaskets/sealants are modelled (full FE model),
suitable boundary conditions and contact interactions shall be considered
x In both cases, the presence of transoms can be simulated using suitable boundary
conditions
x Appropriate capacity reduction factors shall be used

9.2.3 Design rules

If design rules are to be used, e.g. ADM (AA, 2015), guidelines given in Chapter F of
ADM (AA, 2015) can be used, but they are limited by the guidelines given next.

a. Mullion couples without any clip or hollow system (Figure 9.2)

Negative wind action: If mullions show interaction/leaning action (Type A), ADM (AA,
2015) guidelines can be used. If economical solutions are needed, FEA based solutions
should be used. In FEA consider a smaller gap used (between male and female mullions),
uniform load distribution and a suitable spring constant for the lateral restraint. In addition
a suitable capacity reduction factor shall be used to determine the design capacity.

9-8
 Conclusions and recommendations

For mullion couples that show individual rotation (Type B), ADM (AA, 2015) predictions
will be unsafe. Therefore, it is recommended to use FEA, following the guidelines given
above, to predict the moment capacity of such mullion couples. In this case, the capacity
of the mullion couple is twice the capacity of the mullion with the lower capacity.

Positive wind action: The capacity of the mullion couples cannot be considered as their
section/yield capacity by assuming that glass provides full lateral restraint. ADM (AA,
2015) prediction can be used, but the capacity under positive wind action will be even
less than that under negative wind action. Therefore, FEA based solution is recommended
for an economical and safe design. In FEA consider a larger gap used, uniform load
distribution and a suitable spring constant for the lateral restraint. In addition, a suitable
capacity reduction factor shall be used to obtain the design capacity.

Open male Open female

mullion mullion

Figure 9.2: Typical mullion couple without any clip or hollow systems

On the other hand, full-scale tests can be conducted following the guidelines given in
Section 9.2.1.

b. Mullion couples with clip system (Figure 9.3)

Clip parts

Figure 9.3: Typical mullion couple with clip system

Negative wind action: For the mullion couples that show interaction (Type A), the
capacity increment obtained with the use of clip was not significant compared to the

9-9
 Conclusions and recommendations

increments expected based on the currently used assumption (unbraced length of a

mullion couple with a clip at its mid-span is half the span). Therefore, the current
assumption is not applicable. If economical and safe solutions are needed, FEA based
solutions should be used.

ADM (AA, 2015) predictions were also unsafe for the mullion couples that show
individual rotation (Type B). Therefore, the current assumption is not applicable. If
economical and safe solutions are needed, FEA based solutions should be used.

Positive wind action: Capacity increment obtained with the use of clip was not
significant irrespective of whether mullion couples showed interaction or not. Therefore,
clip system cannot be used to increase the capacity. If economical and safe solutions are
needed, FEA based solutions can be used. Opening of mullion couples can be prevented
using a clip system, thus improving their serviceability performance.

On the other hand, full-scale tests can be conducted following the guidelines given in
Section 9.2.1.

c. Mullion couples with hollow system (Figure 9.4)

Hollow

Hollow

(a) Both hollow (b) Hollow male

Figure 9.4: Typical mullion couple with different hollow systems

Negative wind action: If hollow male and female mullions (both hollow) are used, the
capacity of the mullion couples can be significantly increased. In this case ADM (AA,
2015) can be used. If economical solutions are needed, FEA based solutions should be
used.

If only one hollow mullion is used (generally hollow male), the capacity of mullion
couples that show interaction (Type A) can be increased. However, the capacity of the
mullion couples that show individual rotation cannot be increased (Type B). ADM (AA,

9-10
 Conclusions and recommendations

2015) can be used to predict the capacity of mullion couples with a hollow. If economical
solutions are needed, FEA based solutions should be used.

Positive wind action: The capacity of the mullion couples can be increased by using
hollow male and female mullions. In this case ADM (AA, 2015) can be used. If
economical solutions are needed, FEA based solutions should be used.

One hollow mullion (hollow male) can also be used to achieve a significant capacity
increment, particularly for the mullion couples that show individual rotation. In this
case, ADM (AA, 2015) can be used. If economical solutions are needed, it is
recommended to use FEA.

On the other hand, full-scale tests can be conducted following the guidelines given in
Section 9.2.1.

d. Mullion couples with clip and hollow system (Figure 9.5)

Clip

Figure 9.5: Typical mullion couple with clip and hollow systems

Negative wind action: The capacity of a mullion couples can be significantly increased
by using a hollow and a clip, particularly for the mullion couples that show individual
rotation (Type B). In this case, ADM (AA, 2015) can be used without considering the
currently used assumption regarding the use of clip system. If economical solutions are
needed, FEA based solutions should be used.

Positive wind action: The capacity of the mullion couples, which comprise only one
hollow mullion, cannot be increased by using a clip. However, serviceability
performance, such as opening, can be improved. In this case, ADM (AA, 2015) can be
used. If economical solutions are needed, FEA based solutions should be used.

On the other hand, full-scale tests can be conducted following the guidelines given in
Section 9.2.1.

9-11
 Conclusions and recommendations

9.3 Future research

Following research projects can be undertaken in the future to extend the knowledge of
the behaviour and design of aluminium mullions:

x Experimental and numerical investigations of unitized facades made of

structural/captive glazing thermal break facade mullions
In thermal break facade systems, aluminium extrusions are present on both sides of
infills (interior and exterior) and are joined by thermal breaks (a low heat conductive
material such as glass-fibre reinforced polyamide). Under bending, a composite action
will be present between interior and exterior extrusions through the polyamide thermal
break, and thus it differs from the mullion couple systems investigated in this study.
Huang (2014)’s study on thermal break facade mullions is limited to only hollow
mullion sections. Therefore, detailed investigations are recommended.
x Experimental and numerical investigations of unitized facade mullions with
overhang (in curtain wall applications) under wind actions
This research investigated unitized facade mullions under bending with simply
supported end conditions, where bending moment is zero at the supports. However,
mullions with overhang are also used in curtain wall applications, which will induce
moments at the supports under wind actions. Therefore, detailed investigations are
recommended.
x Experimental and numerical investigations of shear and bearing capacities of
unitized facade mullions under wind actions
Lateral wind loads acting on facades are transferred to the main frames via end
connections, where shear and bearing actions are present. To ensure structural integrity
of facade systems, both shear and bearing capacities of the mullion couples should be
adequate. Hence a study on the shear and bearing capacities of mullion couples under
wind actions is recommended.
x Full-scale experimental study of unitized facades made of structural glazing
mullions under wind actions
In this research, full-scale tests were conducted on a captive glazing mullion couple
system, and the results were used to validate the developed FE models. Then the
validated FE models were used to analyse the structural glazing mullion couples. The
main focus was on the structural behaviour and capacity of mullion couples, and thus

9-12
 Conclusions and recommendations

the real structural behaviour of the whole structural glazing system (mullion couple,
infills and adhesives) under wind actions is not known. Therefore, it will be beneficial
to conduct an experimental study on unitized facades made of structural glazing
mullions.
x Effects of cyclic and earthquake loadings on the structural behaviour of unitized
facade mullions
In this research, wind actions were considered as static loads. The effects of cyclic
wind actions on the bending moment capacity of mullion couples were not considered.
In addition, the structural behaviour of mullion couples under earthquake loading was
also not addressed in this study. Therefore, it is recommended to conduct
investigations on the structural behaviour and capacity of mullion couples under cyclic
and earthquake loadings.
x Capacity reduction factor for the design of mullion couples based on non-linear
finite element simulations
In designs, a capacity reduction factor is used to allow for the variations in capacity
predictions. For example, in ADM (AA, 2015) a capacity reduction factor of 0.9 is
used when limit state design approach is used. It is still unclear whether the same
capacity reduction factor can be used with the capacities predicted by FE analyses.
Therefore, it is recommended to conduct a study to investigate this.
x Structural behavior and capacity of unitized facade mullions subject to wind
debris loading
In cyclonic regions, wind induced debris are common. The effects of wind debris
loading on the structural behaviour and capacity of mullion couples were not
investigated in this research. Hence a study on the effects of wind debris loading is
recommended.
x Fire performance of unitized aluminium curtain wall facades
Recently, several facade related fire incidents have occurred, and thus fire resistance
design of building facades has become important. In curtain wall facades, vertical fire
spreading is likely since unitized facade panels are connected to the slab edges
(through perimeter joints). Although considerations are given to prevent such fire
spread to the next levels by providing fire barriers, the structural integrity of mullions
in fire is not known. Therefore, a detailed investigation on the structural and fire
behaviour of curtain wall facades is recommended.

9-13
This Page is Intentionally Left Blank
 References

REFERENCES

The Aluminium Association (AA) (2010). Aluminium Design Manual (ADM), Part 1:
Specification for Aluminium Structures, The Aluminium Association, Washington,
D.C., USA.

The Aluminium Association (AA) (2015). Aluminium Design Manual (ADM), Part 1:
Specification for Aluminium Structures, The Aluminium Association, Arlington, VA,
USA.

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.

Belis, J. and Bedon, C. (2014). Strengthening Effect of Structural Sealants on the LTB
Behaviour of Glass Beams. Proc. of the Challenging Glass 4 & COST Action TU0905
Final Conference, Lausanne: Taylor Francis Ltd, pp. 97-111.

Cambridge (2003). Materials Data Book, 2003 Edition, Engineering department,

University of Cambridge, UK.

Clift, C. and Austin, W. (1989). Lateral Buckling in Curtain Wall Systems. Journal of
Structural Engineering, 115, pp. 2481-2495.

Dassault Systems Simulia Corp. (DS) (2014a). Abaqus Analysis User’s Manual.
Version 6.14, Providence, RI, USA.

Dassault Systems Simulia Corp. (DS) (2014b). Abaqus/CAE User’s Guide. Version
6.14, Providence, RI, USA.

Dassault Systems Simulia Corp. (DS) (2014c). Abaqus Keyword Manual. Version 6.14,
Providence, RI, USA.

De Martino, A., Landolfo, R. and Mazzolani, F. M. (1990). The Use of the Ramberg-
Osgood Law for Materials of Round-House Type. Materials and Structures, 23, pp. 59-
67.

Dow Corning, (2017). Stress-strain curves of DC983 type silicone sealant. Material data
sheet, Dow Corning, Korea.

R-1
 References

Engineershandbook, (2017). Reference Tables-Coefficient of friction. Retrieved from

http://www.engineershandbook.com/Tables/frictioncoefficients.htm.

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

Aluminium Structures - Part 1-1: General Structural Rules, EN 1999-1-1, European
Committee for Standardization, Brussels, Belgium.

Gardner, L. and Ashraf, M. (2006). Structural Design for Non-Linear Metallic

Materials. Engineering Structures, 28, pp. 926-934.

Gere, J. M. and Timoshenko, S. P. (1997). Mechanics of Materials. 4th ed. Boston,

USA: PWS Pub. Co.

Haines, D. W., Leban, J. M. and Herbe, C. (1996). Determination of Young's Modulus

for Spruce, Fir and Isotropic Materials by the Resonance Flexure Method with
Comparisons to Static Flexure and Other Dynamic Methods. Journal of Wood Science
and Technology, 30, pp. 253-263.

Hidallana-Gamage, H. D. (2015). Response of Laminated Glass Panels to Near Field

Blast Events. Doctoral dissertation, Queensland University of Technology, Brisbane,
Australia.

Hill, H. N. (1944). Determination of Stress-Strain Relations from "Offset" Yield

Strength Values. Washington, D.C., National Advisory Committee for Aeronautics,
USA.

Hoglund, T. (2015). CEN/TC 250/SC 09 – N 426 Technical Issues Covered by WGI for
Discussion at the SC9-Meeting in Stockholm 2015-04-25/29: Annex: Lateral Torsional
Buckling of Curtain Wall Profile, European Committee for Standardization (CEN),
Brussels, Belgium.

Huang, S. S. (2014). Investigation of Composite Facade Mullions. Doctoral dissertation,

University of Technology Sydney, Australia.

Huang, Y. and Young B. (2014). The Art of Coupon Tests. Journal of Constructional
Steel Research, 96, pp. 159-175.

R-2
 References

Hui, C., Zhu, Y., Wang, B., Wang, Y. and Tao, W. (2015). Experimental and
Theoretical Investigation on Mechanical Performance of Aluminum Alloy Beams in
Unit Curtain Walls. Advances in Structural Engineering, 18, pp. 2103-2115.

Jeyaragan, S. (2009). Flexural Behaviour and Design of the New Built-Up LiteSteel
Beams. Doctoral dissertation, Queensland University of Technology, Brisbane,
Australia.

Kaitila, O. (2007). Web Crippling of Thin-Walled Cold Formed Steel Cassettes. Journal
of Constructional Steel Research, 63, pp. 766-778.

Kesawan, S. and Mahendran, M. (2019). Buckling behaviour and design of complex-

shaped aluminium mullion sections. Advances in Structural Engineering, 22, pp. 547-
564.

Kesawan, S., Mahendran, M. and Sivaprakasam, T. (2017a). Numerical study of the

section moment capacity of complex–shaped aluminium mullions. Research report,
Queensland University of Technology, Brisbane, Australia.

Kesawan, S., Mahendran, M. and Sivaprakasam, T. (2017b). Structural performance of

mullion couples made of hollow mullions under negative wind action. Research report,
Queensland University of Technology, Brisbane, Australia.

Kesawan, S. and Mahendran, M. (2017). Member moment capacity of long span

aluminium mullions under wind suction loading. Research report, Queensland
University of Technology, Brisbane, Australia.

Kim, Y. (2003). Behaviour and Design of Aluminium Members in Bending. Research

report, Cornell University, Ithaca, NY, USA.

Kissell, J. R. and Ferry, R. L. (1995). Aluminum Structures: A Guide to Their

Specifications and Design. John Wiley & Sons, Inc., NY, USA.

Kumho Polychem (KP) (2017). General Properties of Innoprene. Material data sheet,
Kumho Polychem, Korea.

Kunz, J. and Studer, M. (2006). Determining the Modulus of Elasticity in Compression

via the Shore A Hardness. Kunststoffe, pp. 92-94.

R-3
 References

Lee, A. D., Shepherd, P., Evernden, M. C. and Metcalfe, D. (2017). Optimizing the
Cross-sectional Shapes of Extruded Aluminium Structural Members for Unitized
Curtain Wall Facades. Structures, 10, pp. 147-156.

Lee, A. D., Shepherd, P., Evernden, M. C. and Metcalfe, D. (2017). Optimizing the
Architectural Layouts and Technical Specifications of Curtain Walls to Minimize Use
of Aluminium. Structures, 13, pp. 8-25.

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. Proceedings of the 20th Int;l. Spec. Conf. on Cold-Formed Steel
Structures, St. Louis, MO, November, 2010.

Mahaarachchi, D. and Mahendran, M. (2005). Finite Element Analysis of LiteSteel

Beam Sections. Research Report, Queensland University of Technology, Brisbane,
Australia.

McCowan, D. B. and Kivela, J. B. (2010). Lessons Learned from Curtain Wall Failure
Investigations. Proc. of the Building Envelope Technology Symposium, RCI Inc., pp.
16-26.

Mondolfo, L. F. (1976). Aluminium Alloys: Structure and Properties. London; Boston,

UK; USA: Butterworths, pp. 3-8.

Mudie, S. and Rawlinson, S. (2010). Specialist costs: Building Envelopes. Magazine,

Available at: http://building.co.uk [Accessed 20 February, 2017].

Natário, P., Silvestre, N. and Camotim, D. (2014). Computational Modelling of Flange

Crushing in Cold-Formed Steel Sections. Thin-Walled Structures, 84, pp. 393-405.

Ramberg, W. and Osgood, W. (1943). Description of Stress–Strain Curves by Three

Parameters. Technical note No. 902. National Advisory Committee for Aeronautics,
Washington, D.C., USA.

Sadowski, A. J. and Rotter, J. M. (2013). Solid or Shell Finite Elements to Model Thick
Cylindrical Tubes and Shells under Global Bending. International Journal of
Mechanical Sciences, 74, pp. 143-153.

R-4
 References

Sanders, R. M. and Hargrove, C. A. (2012). Preventing and Treating Failure in Glazed

Curtain wall systems. Journal of Architectural Technology, 29, pp. 1-8.

ShapeDesigner SaaS (SD) (2013). Section Calculator Software. MechaTools

Technologies, Canada.

Skejic, D., Lukic, M., Buljan, N. and Vido, H. (2016). Lateral Torsional Buckling of
Split Aluminium Mullion. Key Engineering Materials, 710, pp. 445-450.

SPACE GASS 11.10 (SG) (2014). Structural Engineering Software. Integrated

Technical Software, Australia.

Standards Australia (SA) (1996a), Windows-Methods of test, Method 1: Test sample,

preparation for tests and test sequence, AS 4420.1, Sydney, Australia.

Standards Australia (SA) (1996b), Windows-Methods of test, Method 2: Deflection test,

AS 4420.2, Sydney, Australia.

Standards Australia (SA) (1997a), Aluminium Structures, Part 1: Limit State Design,
AS/NZS 1664.1, Sydney, Australia.

Standards Australia (SA) (1997b), Aluminium Structures, Part 2: Allowable Stress

Design, AS/NZS 1664.2, Sydney, Australia.

Standards Australia (SA) (2002), Structural Design Actions, Part 0: General Principles,
AS/NZS 1170.0, Sydney, Australia. Incorporating Amendment Nos. 1, 2, 3, 4 and 5

Standards Australia (SA) (2006), Glass in Buildings – Selection and Installation, AS

1288, Sydney, Australia. Incorporating Amendment Nos. 1, 2 and 3 (Reconfirmed
2016).

Standards Australia (SA) (2007), Metallic Materials-Tensile Testing at Ambient

Temperature, AS 1391, Sydney, Australia. Reissued incorporating Amendment No. 1
(August 2012).

Standards Australia (SA) (2008), Testing of Building Facades, AS/NZS 4284, Sydney,
Australia.

Standards Australia (SA) (2011), Structural Design Actions, Part 2: Wind Actions,
AS/NZS 1170.2, Sydney, Australia. Incorporating Amendment Nos. 1, 2, 3 and 4

R-5
 References

Standards Australia (SA) (2014), Windows and External Glazed Doors in Buildings, AS
2047, Sydney, Australia. Reissued incorporating Amendment No. 1 (February 2016).

Suansuwan, N. and Swain, M. V. (2001). Determination of Elastic Properties of Metal

Alloys and Dental Porcelains. Journal of Oral Rehabilitation, 28, pp. 133-139.

Su, M., Young, B. and L. Gardner. (2014). Testing and Design of Aluminium Alloy
Cross Sections in Compression. Journal of Structural Engineering, 140 (9): 04014047.

Sun, J. S., Lee, K. H. and Lee, H. P. (2000). Comparison of Implicit and Explicit Finite
Element Methods for Dynamic Problems. Journal of Materials Processing Technology,
105, pp. 110-118.

Ullrich Aluminium (UA), (2005). Commercial framing systems product manual.

Manual, Ullrich Aluminium Pty. Ltd., Australia.

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.

Yip, J. and Carrick, J. (1995). Behaviour of Coupled Mullions. Test Report, Building
Research Center, University of New South Wales, Sydney, Australia.

Zyla, A. (2010). Mullions: 451-016 (Male) and 451-017 (Female). Test Report,
Contract Glass & Aluminium Pty. Ltd.

R-6
 Appendix

APPENDIX A

A.1 Tensile coupon details

The measured thickness, width and other details of the tensile coupons are summarized
in Table A.1. The alphabets “A” to “C” or the numbers “1” to “5” (for specimens from
different batches) used to label the tensile coupons are for the convenience while testing
and managing the data.

Table A.1: Sample details

Type of sample No. of Nominal thickness

Label Location
specimens (mm)
450-044 2 A,B 1.80 Web
450-045 2 A,B 1.80 Web
650-027 2 A,B 2.10 Web
3 1 2.10 Web
650-028 2 A 2.10 Web
3 4,5 2.10 Web
651-012 2 A,B 2.50 Web
651-013 2 A,B 2.50 Web
651-014 2 A,B 2.30 Web
1 C 2 Flange
651-015 2 A,B 2.30 Web
651-537 2 A,B 2.00 Web
1 C 2.80 Flange
651-599 1 A 2.50 Web
1 C 2.80 Flange
851-512 2 A,B 2.50 Web
1 C 3.00 Flange
546-035 2 A,B 3.00 Web
1 C 3.50 Flange
546-036 2 A,B 3.00 Web
570-001 2 A,B 3.00 Web
1 C 4.00 Flange
570-053 2 A,B 3.00 Web
579-001 2 A,B 3.00 Web
579-002 2 A,B 3.00 Web
579-003 2 A,B 3.00 Web
1 C 3.50 Flange
579-004 2 A,B 3.00 Web
1 C 3.50 Flange

A-1
 Appendix

A.2 Stress-strain curves from tensile coupon tests

The measured stress-strain curves of 6063-T6 and 5005 aluminium alloys are presented
in this section.

300

250
Stress (MPa)

200

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
Strain (%)
450-044-A 450-044-B

(a.1) 450-044

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (%)
450-045-A 450-045-B

(a.2) 450-045
Figure A. 1: Measured stress versus strain curves of 6063-T6 aluminium alloy

A-2
 Appendix

300

Stress (MPa) 250

200

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (%)
546-035-A 546-035-B 546-035-C

(a.3) 546-035

300

250

200
Stress (MPa)

150

100

50

0
0 0.05 0.1 0.15
Strain (%)
546-036-A 546-036-B

(a.4) 546-036

Figure A. 1: Measured stress versus strain curves of 6063-T6 aluminium alloy

A-3
 Appendix

300

250
Stress (MPa)

200

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
Strain (%)
570-001-A 570-001-B 570-001-C

(a.5) 570-001

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (%)
570-053-A 570-053-B

(a.6) 570-053

Figure A. 1: Measured stress versus strain curves of 6063-T6 aluminium alloy

A-4
 Appendix

250

Stress (MPa) 200

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
Strain (%)
579-001-A 579-001-B

(a.7) 579-001

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (%)
579-002-A 579-002-B

(a.8) 579-002

Figure A. 1: Measured stress versus strain curves of 6063-T6 aluminium alloy

A-5
 Appendix

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (%)
579-003-A 579-003-B 579-003-C

(a.9) 579-003

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (%)
579-004-A 579-004-B 579-004-C

(a.10) 579-004

Figure A. 1: Measured stress versus strain curves of 6063-T6 aluminium alloy

A-6
 Appendix

250

Stress (MPa) 200

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
Strain (%)
651-012-A 651-012-B

(a.11) 651-012

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (%)
651-013-A 651-013-B

(a.12) 651-013

Figure A. 1: Measured stress versus strain curves of 6063-T6 aluminium alloy

A-7
 Appendix

300

250
Stress (MPa)

200

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
Strain (%)
651-014-A 651-014-B 651-014-C

(a.13) 651-014

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (%)
651-015-A 651-015-B

(a.14) 651-015

Figure A. 1: Measured stress versus strain curves of 6063-T6 aluminium alloy

A-8
 Appendix

300

Stress (MPa) 250

200

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
Strain (%)
651-537-A 651-537-B 651-537-C

(a.15) 651-537

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (%)
651-599-A 651-599-C

(a.16) 651-599

Figure A. 1: Measured stress versus strain curves of 6063-T6 aluminium alloy

A-9
 Appendix

250

200

150
Stress (MPa)

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
Strain (%)
851-512-A 851-512-B

(a.17) 851-512
Figure A. 1: Measured stress versus strain curves of 6063-T6 aluminium alloy

A-10
 Appendix

A.3 Comparison of Ramberg-Osgood model with test results

Comparison of the measured stress-strain curves and the Ramberg-Osgood model for
plastic analysis range given in EN 1999-1-1 (CEN, 2007) of 6063-T6 aluminium alloy
are shown in this section.

300

250
Stress (MPa)

200

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (%)
Test results Ramberg-Osgood model

(a.1) 450-044-A

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (%)
Test results Ramberg-Osgood model
(a.2) 450-045-A

Figure A. 2: EN 1999-1-1 (CEN, 2007) model for plastic region and test results

A-11
 Appendix

300

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (%)
Test results Ramberg-Osgood model

(a.3) 546-035-A

300

250
Stress (MPa)

200

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
Strain (%)
Test results Ramberg-Osgood model

(a.4) 546-036-A

Figure A. 2: EN 1999-1-1 (CEN, 2007) model for plastic region and test results

A-12
 Appendix

300

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
Strain (%)
Test results Ramberg-Osgood model

(a.5) 570-001-A

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (%)
Test results Ramberg-Osgood model

(a.6) 570-053-A

Figure A. 2: EN 1999-1-1 (CEN, 2007) model for plastic region and test results

A-13
 Appendix

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
Strain (%)
Test results Ramberg-Osgood model

(a.7) 579-001-A

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (%)
Test results Ramberg-Osgood model

(a.8) 579-002-A

Figure A. 2: EN 1999-1-1 (CEN, 2007) model for plastic region and test results

A-14
 Appendix

250

Stress (MPa) 200

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (%)
Test results Ramberg-Osgood model

(a.9) 579-003-A

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (%)
Test results Ramberg-Osgood model

(a.10) 579-004-A

Figure A. 2: EN 1999-1-1 (CEN, 2007) model for plastic region and test results

A-15
 Appendix

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
Strain (%)
Test results Ramberg-Osgood model

(a.11) 651-012-A

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (%)
Test results Ramberg-Osgood model

(a.12) 651-013-A

Figure A. 2: EN 1999-1-1 (CEN, 2007) model for plastic region and test results

A-16
 Appendix

300

250
Stress (MPa)
200

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
Strain (%)
Test results Ramberg-Osgood model

(a.13) 651-014-A

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (%)
Test results Ramberg-Osgood model

(a.14) 651-015-A

Figure A. 2: EN 1999-1-1 (CEN, 2007) model for plastic region and test results

A-17
 Appendix

300

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (%)
Test results Ramberg-Osgood model

(a.15) 651-537-A

250

200
Stress (MPa)

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (%)
Test results Ramberg-Osgood model

(a.16) 651-599-A

Figure A. 2: EN 1999-1-1 (CEN, 2007) model for plastic region and test results

A-18
 Appendix

250

Stress (MPa) 200

150

100

50

0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
Strain (%)
Test results Ramberg-Osgood model

(a.17) 851-512-A

Figure A. 2: EN 1999-1-1 (CEN, 2007) model for plastic region and test results

A-19
 Appendix

A.4 Summary of tensile test results for 6063-T6 aluminium alloy samples

The measured mechanical properties and the values of strain hardening exponents
calculated based on the provisions given in EN 1999-1-1 (CEN, 2007) for elastic and
plastic analysis cases (nElastic and nPlastic, respectively) are given in Table A. 2.

Table A. 2: Summary of test results

E f0.2 f0.1 fmax

Sample Label
(MPa) (Mpa) (Mpa) (Mpa)
Hmax Hu nElastic nPlastic

A 67065 210.3 207.3 228.4 0.06967 0.09780 48.4 42.5

650-027 B 63092 210.4 208.0 228.5 0.07508 0.10241 58.7 43.4
1 65824 192.1 189.3 210.6 0.06255 0.09350 47.2 36.8
A 64013 201.4 199.3 218.3 0.06252 0.08536 67.4 41.9
650-028 4 64266 214.8 211.3 233.1 0.07046 0.09694 42.3 42.9
5 64467 217.6 214.1 235.8 0.10709 0.07534 42.6 44.4
A 60549 237.8 232.8 258.4 0.09177 0.11447 33.1 45.4
450-044
B 61041 239.1 233.9 260.2 0.09434 0.13413 31.9 44.9
A 60887 217.7 213.1 234.3 0.07677 0.11467 32.6 48.8
450-045
B 60954 215.6 212.3 232.3 0.07475 0.09302 45.5 47.7
A 63971 210.7 208.0 230.6 0.07239 0.10345 53.2 39.3
651-013
B 63754 210.9 208.1 230.6 0.07221 0.1106 52.4 39.5
A 66679 234.8 231.3 258.1 0.09197 0.1213 46.0 40.0
651-014 B 64410 231.3 228.2 253.6 0.0919 0.12075 50.2 41.2
C 64834 232.7 229.1 256.9 0.08452 0.10299 44.4 37.3
A 67553 205.4 202.3 234.8 0.09404 0.12722 45.7 28.5
651-012
B 65275 202.8 199.8 232.3 0.09177 0.1207 45.7 27.9
A 65572 217.5 213.7 235.7 0.07588 0.10523 40.3 44.6
579-003 B 65567 219.8 216.2 238.0 0.0765 0.10248 41.8 45.2
C 67611 215.8 211.0 235.7 0.07629 0.10812 30.8 40.6
A 66765 208.1 204.1 227.0 0.07313 0.10127 35.6 40.8
579-004 B 66198 208.3 205.2 227.5 0.07208 0.11265 47.0 40.0
C 67734 209.4 205.7 228.3 0.07373 0.11142 38.9 41.2
A 68365 238.3 232.1 259.6 0.08377 0.12935 26.3 43.0
570-001 B 62851 234.5 229.0 256.0 0.08284 0.10791 29.2 42.0
C 66407 234.2 227.8 255.7 0.08496 0.12842 24.8 42.2
A 64050 227.9 220.0 252.7 0.08421 0.11686 19.6 35.7
546-035 B 62923 229.5 222.5 253.3 0.08437 0.11005 22.3 37.5
C 63722 235.3 227.8 259.5 0.07852 0.09391 21.5 36.9
A 65146 240.0 234.1 263.6 0.09121 0.11952 27.9 40.3
546-036
B 67147 239.1 233.2 261.7 0.09073 0.13497 27.8 41.8
A 70841 218.9 214.0 237.9 0.08171 0.11046 30.6 44.1
651-599
C 66110 219.3 215.4 237.7 0.07638 0.10645 38.1 44.7

A-20
 Appendix

Table A. 2: Summary of test results

E f0.2 f0.1 fmax

Sample Label
(MPa) (Mpa) (Mpa) (Mpa)
Hmax Hu nElastic nPlastic

A 63389 178.5 175.7 217.5 0.10113 0.12723 44.5 19.7

579-001
B 66153 183.3 180.4 219.9 0.09369 0.11695 43.0 20.9
A 63436 211.5 208.2 230.0 0.07325 0.10401 44.1 42.2
579-002
B 63356 213.9 210.9 232.1 0.07475 0.1118 49.9 43.6
A 68626 213.8 209.6 233.2 0.07004 0.09964 34.9 40.5
570-053
B 67502 216.9 213.4 235.0 0.07294 0.10909 41.8 44.4
A 65646 203.3 200.6 222.8 0.06906 0.09957 52.0 38.2
651-015
B 65150 202.5 199.6 222.1 0.06288 0.08084 48.1 36.7
A 65979 222.4 218.3 245.9 0.09021 0.11438 36.9 37.5
651-537 B 65245 220.0 215.4 244.2 0.08977 0.12007 32.5 36.1
C 66973 219.5 215.3 242.0 0.08207 0.09761 36.5 37.6
A 67921 212.7 208.2 237.0 0.09152 0.12911 32.3 35.0
851-512
B 65365 211.7 207.4 235.3 0.08758 0.13177 33.5 35.3
- 69452 138.0 133.2 157.7 0.02523 0.05634 19.7 18.3
5005
- 69504 137.4 132.6 156.6 0.03258 0.06436 19.5 20.8
Minimum 60549 179 176 211 0.0625 0.0808 - -
Maximum 70841 240 234 264 0.1011 0.1350 - -
Mean 65158 N/A N/A N/A 0.0799 0.1105 40 40
COV 0.033 N/A N/A N/A 0.122 0.115 0.267 0.15

A-21
 Appendix

A.5 Specimen under four-point bending

This section describes the derivation of the relationship between the M/I and 1/R at the
mid-span of a beam under four-point bending. M, I and R are the moment at mid-span,
second moment of area of the specimen and the radius of curvature of the deflected
specimen, respectively. Schematic and bending moment diagrams of a four-point
bending test are shown in Figure A.4.

Figure A. 3: Specimen under four-point bending

Region BC in Figure A.4 is subject to pure bending, i.e. shear deformations are not
present. Therefore, based on the classical (or Euler-Bernoulli) beam theory (Gere and
Timoshenko, 1997),

݀ଶ ‫ݕ‬
‫ܯ‬௫ ൌ ‫ ܫܧ‬ή
݀‫ ݔ‬ଶ
Also, the relationship between the bending moment and the beam curvature for an
Euler-Bernoulli beam can be written as,

‫ܧ ܯ‬ ͳ
ൌ  ՜ ‫ ܯ‬ൌ ‫ ܫܧ‬ή
‫ܫ‬ ܴ ܴ
Where,

A-22
 Appendix

ͳ ݀ଶ‫ݕ‬
ൌ
ܴ ݀‫ ݔ‬ଶ

At any point x in the region BC, Mx=W·a. Therefore,

݀ଶ ‫ݕ‬
ܹ ή ܽ ൌ ‫ ܫܧ‬ή
݀‫ ݔ‬ଶ

Integrate both sides by x,

݀‫ݕ‬
ܹ ή ܽ ή ‫ ݔ‬൅ ‫ܥ‬ଵ ൌ ‫ ܫܧ‬ή Ǣ ‫ܥ‬ଵ ݅‫ݐ݊ܽݐݏ݊݋ܿܽݏ‬
݀‫ݔ‬

At‫ ݔ‬ൌ ‫ܮ‬ൗʹ, ‫ܥ‬ଵ ൌ െܹ ή ܽ ή ‫ܮ‬ൗʹ

ௗ௬
ൌͲ
ௗ௫

Further integrate both sides by x,

‫ݔ‬ଶ
ܹ ή ܽ ή ൅ ‫ܥ‬ଵ ή ‫ ݔ‬൅ ‫ܥ‬ଶ ൌ ‫ ܫܧ‬ή ‫ݕ‬Ǣ ‫ܥ‬ଶ ݅‫ݐ݊ܽݐݏ݊݋ܿܽݏ‬
ʹ
At‫ ݔ‬ൌ Ͳ, ‫ ݕ‬ൌ Ͳ ‫ܥ‬ଶ ൌ Ͳ

Therefore,
‫ݔ‬ଶ ‫ܮ‬
ܹήܽή െ ܹ ή ܽ ή ή ‫ ݔ‬ൌ ‫ ܫܧ‬ή ‫ݕ‬
ʹ ʹ

At‫ ݔ‬ൌ ‫ܮ‬ൗʹ, consider the deflection as “'”. Therefore,

‫ܮ‬ଶ ‫ܮ ܮ‬
ܹήܽή െ ܹ ή ܽ ή ή ൌ െ‫ ܫܧ‬ή ο
ͺ ʹ ʹ

‫ܮ‬ଶ
െܹ ή ܽ ή ൌ െ‫ ܫܧ‬ή ο
ͺ

‫ܯ‬ ͺο ‫ܯ‬ ͳ ͳ ͺο
ൌ‫ܧ‬ή ଶ ܱܴ ൌ‫ܧ‬ή Ǣ ‫݁ݎ݄݁ݓ‬ ൌ ଶ
‫ܫ‬ ‫ܮ‬ ‫ܫ‬ ܴ ܴ ‫ܮ‬

Based on this derivation, the tangent of M/I versus 1/R graph will give the flexural
modulus of elasticity of the material.

A-23
This Page is Intentionally Left Blank
 Appendix

APPENDIX B

B.1 Section properties of the test mullions

B1.1 Measured section thicknesses of the mullions used in Test 1 and 2

Test 1

Figure B. 1: Measured thicknesses of sample taken from Test 1

Table B. 1: Measured section properties of sample taken from Test 1

Property 650-027 (Male) 650-028 (Female)

A (mm2) 5.848E+02 5.582E+02
Ixx (mm4) 1.676E+06 1.660E+06
Iyy 7.193E+04 3.724E+04
rx (mm) 53.53 54.539
ry (mm) 11.091 8.168
Zpx (mm3) 27983 27272
J (mm4) 8.667E+02 1.069E+03
Cw (mm6) 2.987E+08 1.212E+08
Xs (mm) -16.880 10.789
Ys (mm) 9.984 -0.737
Ex (mm) -8.912 10.952

B-1
 Appendix

Test 2

Figure B. 2: Measured thicknesses of sample taken from Test 2

Table B. 2: Measured section properties of sample taken from Test 2

Property 650-027 (Male) 650-028 (Female)

A (mm2) 5.706E+02 5.584E+02
Ixx (mm4) 1.645E+06 1.660E+06
Iyy 7.028E+04 3.717E+04
rx (mm) 53.69 54.53
ry (mm) 11.10 8.16
Zpx (mm3) 27343 27274
J (mm4) 8.104E+02 1.065E+03
Cw (mm6) 2.949E+08 1.212E+08
Xs (mm) -16.933 10.80
Ys (mm) 10.10 -0.52
Ex (mm) -8.372 10.562

B-2
 Appendix

B1.2 Comparison of the section properties of the original and idealized mullions

In the simplified models, the mullion sections were idealized into shell finite elements.
In order to assess the effect of idealization in the section properties such as area (A),
second moment of areas (Ixx and Iyy), torsional constant (J) and etc., a comparison is
made between the section properties of the original and the idealized mullion sections. It
is evident from Tables B.3 and B.4 that the section properties are not significantly
affected due to idealization.

Table B. 3: Section properties of male mullion - Test 1

Property Original Idealized

A (mm2) 5.848E+02 5.774E+02
Ixx (mm4) 1.676E+06 1.657E+06
Iyy 7.193E+04 7.20E+04
rx (mm) 53.53 53.57
ry (mm) 11.09 11.17
Zpx (mm3) 27983 27693
J (mm4) 8.67E+02 7.10E+02
Cw (mm6) 2.987E+08 2.999E+08
Xs (mm) -16.88 -17.02
Ys (mm) 9.98 9.62
Ex (mm) -8.912 -8.562

Table B. 4: Section properties of female mullion - Test 2

Property Original Idealized

A (mm2) 5.582E+02 5.472E+02
Ixx (mm4) 1.660E+06 1.632E+06
Iyy 3.724E+04 3.674E+04
rx (mm) 54.54 54.61
ry (mm) 8.17 8.19
Zpx (mm3) 27272 26814
J (mm4) 10.69E+02 10.12E+02
Cw (mm6) 1.212E+08 1.205E+08
Xs (mm) 10.79 10.89
Ys (mm) -7.37 -5.79
Ex (mm) 10.952 10.204

B-3
 Appendix

B1.3 Measured properties of the mullions used in Test 3 and 4a

Measured thicknesses

(a) Test 3

(b) Test 4a

Figure B. 3: Measured thicknesses of sample taken from Tests 3 and 4a

B-4
 Appendix

Measured stress-strain curves

300

250

200
Stress (MPa)

150

100
Test results
50 EN 1999-1-1 (CEN, 2007)

True stress-strain
0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (mm/mm)

(a) Male mullion

300

250

200
Stress (MPa)

150

100
Test results
50 EN 1999-1-1 (CEN, 2007)
True stress-strain
0
0 0.02 0.04 0.06 0.08 0.1
Strain (mm/mm)

(b) Female mullion

Figure B. 4: Stress versus strain curves of Test 3 mullions

B-5
 Appendix

300

250

200
Stress (MPa)

150

100
Test results
50 Ramberg-Osgood model
True stress-strain
0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (mm/mm)

(a) Male mullion

300

250

200
Stress (MPa)

150

100

Test results
50
Ramberg-Osgood model
True stress-strain
0
0 0.02 0.04 0.06 0.08 0.1 0.12
Strain (mm/mm)

(b) Female mullion

Figure B. 5: Stress versus strain curves of Test 4a mullions

B-6
 Appendix

B1.4 Cross sectional details of the clip parts

Figure B. 6: Thicknesses of the clip parts used in FE modelling

B.2 Determination of pressure on mullion couple

Test setup comprised two unitized facade wall panels of 1.45 m c/c width (b) and 3.96
m span length (h). A trapezoidal tributary area is considered for the load transferred
from aluminium sheets to the mullion couple as shown in Figure B.7. The reaction at
one end of the mullion couple (R) was obtained from FEA, whereas pressure was
measured in the tests. In order to compare the FEA and test results, the reaction
obtained from FEA is converted to pressure as given next. Also a relationship between
the moment at span (Mspan) and the end reaction (R) of the mullion couple is derived in
this section.

Figure B. 7: Load transfer from aluminum sheets to center mullion couple

B-7
 Appendix

ܾ
Tributary load area (At) = ሺʹ݄ െ ܾሻ
ʹ
ܾ
If applied pressure is p, reaction (R) = ሺʹ݄ െ ܾሻ ‫ ݌‬OR ‫ ݌‬ൌ Ͷܴൗሼܾ ή ሺʹ݄ െ ܾሻሽ
Ͷ

‫ݓ‬
Total reaction (2R) = ሺʹ݄ െ ܾሻ
ʹ

Magnitude of uniform load (w) = Ͷܴൗሺʹ݄ െ ܾሻ

ோή௛ ௪ή௛మ ௕ή௪ ௛ ௕

Maximum span moment (Mspan) = ቄെ ൅ െ ቀ െ ቁቅ
ଶ ଼ ଶήଶ ଶ ଶήଷ
ோή௛ ோή௛మ ோή௕
= ቄെ ൅ െ ሺ͵݄ െ ܾሻቅ
ଶ ଶሺଶ௛ି௕ሻ ଺ሺଶ௛ି௕ሻ

In the test setup,

b = 1.45 m
h = 3.96 m

Hence,
At = ሺʹ ൈ ͵Ǥͻ͸ െ ͳǤͶͷሻ ൈ ͳǤͶͷȀʹ ൌ ͶǤ͸ͻͳ݉ଶ

p = Ͷ ൈ ܴൗሼͳǤͶͷ ൈ ሺʹ ൈ ͵Ǥͻ͸ െ ͳǤͶͷሻሽ ൌ ܴൗʹǤ͵Ͷͷ ܲܽ

ଷǤଽ଺ൈோ ଷǤଽ଺మ ൈோ ଵǤସହൈோ

Mspan = ቄെ ൅ െ ሺ͵ ൈ ͵Ǥͻ͸ െ ͳǤͶͷሻቅ
ଶ ଶൈሺଶൈଷǤଽ଺ିଵǤସହሻ ଺ൈሺଶൈଷǤଽ଺ିଵǤସହሻ

= ͳǤͳͷͺ ή ܴܰ݉
B.3 Determination of the elastic spring constant for lateral restraint

An approximate value of the elastic spring constant was determined in this section
assuming that the gasket does not slip. Belis and Bedon (2014) derived a relationship
between the elastic modulus and stiffness of a silicone joint used in glass beams, where
they considered a silicone bide that deforms under shear as shown in Figure B.8(b). A
similar approach was adopted to determine an approximate stiffness for the lateral
restraint provided by the aluminium sheets to the mullions. For this purpose, a
rectangular portion of the gasket is considered as shown in Figure B.8(a). Shear stress
versus strain curve for the DC983 type silicone obtained from the manufacturer is
shown in Figure B.9, where 50% strain range is considered in this study.

B-8
 Appendix

(a) Gasket (b) Representative shear block

Figure B. 8: Shear deformation

0.25
y = 0.427x + 0.0137
0.2
Shear stresss

0.15

0.1

0.05

0
0 0.1 0.2 0.3 0.4 0.5 0.6
Shear strain
Figure B. 9: Shear stress versus strain curve of DC983 silicone rubber (Dow
Corning, 2017)

ο௬
Shear strain ɀ ൌ
௛೒

ி
Shear stress W ൌ ‫ܩ‬௦௜௟ ɀ ൌ
஺೒

Area of silicone ‫ܣ‬௚ ൌ ݈௚ ܾ௚

‫ܨ‬
‫ܭ‬௬ ൌ
ο‫ݕ‬

where:

‫ܭ‬௬ ൌ ݇௬ ݈௚

ky = stiffness of the lateral restraint per unit length (N/mm/mm)

B-9
 Appendix

Therefore,

‫ܭ‬௬ ο‫ݕ‬
‫ܩ‬௦௜௟ ɀ ൌ
‫ܣ‬௚

݇௬ ݈௚ ο‫݄ݕ‬௚
‫ܩ‬௦௜௟ ൌ
ο‫݈ݕ‬௚ ܾ௚

finally,

‫ܩ‬௦௜௟ ܾ௚
݇௬ ൌ
݄௚
using the derived equation,

hg = 3.88 mm
bg = 3.04 mm
lg = length of the gasket
ͲǤͶʹ͹ ൈ ͵ǤͲͶ
݇௬ ൌ
͵Ǥͺͺ

! ky = 0.334 N/mm/mm

In FEAs, 4 x 8 mm (width x length) mesh size was used. Elastic spring constant was
defined at the nodes of discretized finite element model of the mullions along their
length. Therefore, the stiffness per node was approximately taken as 2.5 N/mm, i.e., 8 x
0.334 (length of the mesh x stiffness per unit length).

B.4 Lateral restraint and load position

The load from aluminium sheets to mullions can be expected to transfer through the
point as shown in Figure B.10(a). Also the lateral restraint given by the aluminium
sheets to the mullions can be expected to act on the same point. However, since there is
a possibility for the load and the lateral restraint acting at different locations as the
mullions deform under different wind actions, different load and the lateral restraint
locations were investigated (Figure B.10). Figure B.11 shows that the predicted
ultimate capacities and the pressure versus displacement characteristics of the mullion
couple under negative wind action are within a closer agreement. Therefore, the load
and the lateral restraint location (Case 1) as shown in Figure B.10(a) can be considered
for the negative wind action.

B-10
 Appendix

Lateral
restraint

Load and lateral

restraint Load

(a) Case 1 (b) Case 2

Load and lateral Load and lateral

restraint restraint
(c) Case 3 (d) Case 4
Figure B. 10: Different load and lateral restraint locations under negative
wind action

B-11
 Appendix

-1000
-900
-800
-700
Pressure (Pa)-600
-500
-400
-300 Case 1
-200 Case 2
Case 3
-100
Case 4
0
0 -5 -10 -15 -20 -25 -30
Y-axis displacement (mm)

Figure B. 11: Pressure versus displacement curves of mullion couple with different
stiffness values under negative wind action

Load

Load and lateral restraint Lateral restraint

(a) Case 1 (b) Case 2
Figure B. 12: Different load and lateral restraint locations under positive
wind action

B-12
 Appendix

Load and lateral Load lateral

restraint restraint
(c) Case 3 (d) Case 4
Figure B. 12: Different load and lateral restraint locations under positive
wind action

During positive wind action, the load and the lateral restraint can be expected to act on
mullions as shown in Figure B.12(a). However, different possible locations for the load
and lateral restraint were assessed as that for the negative wind action (Figure B.12).
Figure B.13 shows that the predicted ultimate capacities of the mullion couple for the all
cases considered are within a closer agreement. The maximum difference in the
predicted capacities are less than 10% with respect to the Case 1. Therefore, the load
and the lateral restraint location (Case 1) as shown in Figure B.12(a) can be considered
for the positive wind action.

B-13
 Appendix

4000
3500
3000

Pressure (Pa) 2500

2000
1500
Case 1
1000 Case 2
500 Case 3
Case 4
0
0 30 60 90 120 150
Y-axis displacement (mm)

Figure B. 13: Pressure versus displacement curves of mullion couple with different
load and lateral restraint cases under positive wind action

B.5 Support boundary conditions

In the aluminium frame of a unitized facade panel, the mullions are connected with the
transoms using screws at their both ends. A typical mullion to transom connection of a
unitized facade panel is shown in Figure B.14. It is evident from this figure that the
connection is capable of providing effective lateral restraint to the mullions. Also this
connection can provide effective restraint against torsional rotation of the mullions.
Based on this, the boundary condition considered in finite element modelling of the
mullion couple U1=U2=UR3=0, where the U1, U2 and UR3 stands for translation along
the X axis, translation along the Y axis and rotation about the Z axis, respectively. The
boundary conditions were provided around the perimeter of the mullion couple.

However, since the connection between the mullions and transoms is by means of two
screws over a small region of the mullions, a sensitivity study was performed by
changing the U1 and the UR3 boundary conditions. That is, two addition analyses were
performed on the mullion couple which are: 1. U1=U2=0, where the U1 was restrained
around the perimeter, and UR3 was not restrained; 2. U1=U2=0, where the U1 was
restrained at the screw locations of the mullions. Figure B.15 reveals that there is no
significance regarding the end boundary conditions used. Therefore, the
U1=U2=UR3=0 will be used in the other finite element analyses.

B-14
 Appendix

Transom

Mullion Screw
connections

Figure B. 14: Typical mullion to transom connection

-1000
-900
-800
Pressure (Pa)

-700
-600
-500
-400
-300 U1=U2=UR3=0
-200 U1=U2=0
-100 U1=U2=0, U1 at screw points
0
0 -5 -10 -15 -20 -25
Y-axis displacement (mm)

(a) Under negative wind action

Figure B. 15: Pressure versus displacement curves of mullion couple with different
support conditions

B-15
 Appendix

3500

3000

2500
Pressure (Pa) 2000

1500
U1=U2=UR3=0
1000
U1=U2=0
500
U1=U2=0, U1 at screw points
0
0 20 40 60 80 100 120
Y-axis displacement (mm)

(b) Under positive wind action

Figure B. 15: Pressure versus displacement curves of mullion couple with different
support conditions

B.6 Young’s modulus of gaskets and frictional coefficient between gasket

and aluminium sheet

Full finite element models of the mullion couple was analysed by varying the frictional
coefficient, between aluminium sheet and gasket, and the modulus of elasticity of the
gasket. The predicted failure capacities of the mullion couple are given in Table B.5. It
reveals that the rate of change in the capacity of the mullion couple is significant with
respect to friction than that with the modulus of elasticity. Since the modulus of
elasticity of the gasket is in the range of 10 MPa, the frictional coefficient between the
aluminium sheet and gasket was considered as 0.1.

Table B. 5: Mullion couple under positive wind action with different Young’s
modulus of gasket and gasket-aluminium sheet frictional coefficient

Gasket-Aluminium friction
Test 2 model
F=0.05 F=0.1 F=0.2 F=0.3

5 2040 2680 3242 3370

Young’s modulus 10 2230 2910 3539 >3370

of gasket (N/mm2) 20 2320 3000 - -

50 2610 3220 - -

B-16
 Appendix

APPENDIX C

C.1 Section properties of the mullions obtained using shape designer software (SD, 2013)

Table C. 1: Section properties of the male and female mullions of mullion couples MC1 to MC20
A J
Name Type Ixx Iyy Cw Zpx Xs Ys Ytop Ybot βx Yn,b Yp,b
(mm2) (mm4)
MC1 Male 493 636230 66897 692 1.53E+08 15750 -17.8 10.2 52 49 -13.9 6.0 22.5
Female 477 639400 32702 874 6.61E+07 15896 10.1 1.4 50 52 -2.5 6.0 22.5
MC2 Male 498 600240 60127 879 1.82E+08 14438 -7.6 10.5 50 52 -21.6 6.0 41.5
Female 460 568150 36126 848 9.98E+07 13692 2.5 1.5 51 51 -1.7 6.0 41.5
MC3 Male 779 981350 185920 2875 5.51E+08 23275 -14.6 11.9 49 52 -22.5 6.0 41.5
Female 787 976970 114860 4046 2.85E+08 23797 3.9 -1.9 50 51 5.0 6.0 41.5
MC4 Male 829 1406100 176980 2798 7.83E+08 28819 -13.8 9.7 63 54 -6.2 22.7 40.7
Female 755 1322600 86048 2895 3.62E+08 27253 1.3 -14.9 64 53 40.5 22.7 40.7
MC5 Male 614 1794600 79527 998 3.13E+08 29596 -17.8 9.8 81 72 -7.3 6.0 22.5
Female 570 1685000 37960 1142 1.22E+08 27757 10.73 -0.77 79 73 10.7 6.0 22.5
MC6 Male 655 1727200 68154 1290 2.48E+08 29576 -14.9 21.0 73 79 -39.7 6.0 41.5
Female 647 1680600 67415 1268 2.47E+08 29009 -14.7 20.1 72 80 -38.7 6.0 41.5
MC7 Male 736 2007500 141860 1551 6.15E+08 33790 -20.9 25.9 76 76 -42.2 6.0 41.5
Female 676 1862000 65047 1507 2.66E+08 31430 11.0 6.5 77 76 -6.9 6.0 41.5
MC8 Male 799 2203300 162680 2046 6.77E+08 37322 -21.7 24.6 73 80 -43.9 6.0 41.5
Female 682 1893800 65713 1426 2.68E+08 31869 11.3 6.7 76 76 -7.4 6.0 41.5
MC9 Male 821 2248500 217710 2603 1.02E+09 36537 -28.3 16.4 77 74 -16.5 7.7 43.7
Female 731 2138800 82158 2622 3.53E+08 35728 13.1 -1.6 75 77 9.2 7.7 43.7

C-1
 Appendix

Table C.1: Section properties of the male and female mullions of mullion couples MC1 to MC20
A J
Name Type Ixx Iyy Cw Zpx Xs Ys Ytop Ybot βx Yn,b Yp,b
(mm2) (mm4)
MC10 Male
Female 736 1969900 91762 1966 4.35E+08 32498 -4.1 -2.8 75 78 1.7 6.0 84.2
MC11 Male 1293 4601300 444230 5985 2.59E+09 68583 -32.3 23.1 87 86 -34.7 7.7 43.7
Female 1128 4055000 162190 5584 8.28E+08 60625 16.1 4.5 88 84 -2.7 7.7 43.7
MC12 Male 929 4619500 184980 2803 1.29E+09 59247 -24.2 13.4 103 97 -15.3 7.7 25.7
Female 878 4421000 68534 3469 3.88E+08 56177 13.5 -4.3 98 102 9.7 7.7 25.7
MC13 Male 552 565420 127480 1715 1.48E+08 16356 -35.2 9.2 44 48 -10.4 15.7 15.7
Female 459 476080 42711 1569 2.75E+07 13868 18.9 -0.9 42 50 0.7 15.7 15.7
MC14 Male 623 1650300 139590 1506 4.11E+08 29429 -30.4 9.9 73 69 -6.7 15.7 15.7
Female 555 1471300 46362 1576 9.57E+07 26432 16.3 -8.8 66 77 15.4 15.7 15.7
MC15 Male 790 1909500 211370 2895 5.80E+08 33433 -35.0 22.1 73 79 -27.3 43.7 43.7
Female 689 1763100 79204 2770 1.12E+08 31917 20.0 14.5 69 83 -25.2 43.7 43.7
MC16 Male 1191 3499100 454190 6567 1.53E+09 60003 -41.9 12.2 69 93 -26.5 33.2 33.2
Female 975 2820100 138550 5001 2.68E+08 47956 23.8 5.2 70 92 -16.6 33.2 33.2
MC17 Male 1147 3490300 432420 4913 1.51E+09 57432 -41.1 20.9 76 96 -36.8 43.7 43.7
Female 982 2958700 150580 4361 2.73E+08 48781 23.5 10.9 75 97 -25.1 43.7 43.7
MC18 Male 1369 3728100 609410 5968 1.62E+09 61551 -43.7 25.3 75 94 -40.9 33.5 33.5
Female 1145 3168200 185400 5148 3.78E+08 52461 25.8 6.0 79 89 -2.7 33.5 33.5
MC19 Male 1365 4225300 651420 5892 1.65E+09 64093 -42.7 22.0 76 106 -47.7 33.5 33.5
Female 1112 3673700 204400 4615 4.63E+08 55591 26.7 0.3 80 102 -3.9 33.5 33.5
MC20 Male 1206 4280400 247340 4292 1.15E+09 60044 -27.6 7.2 98 92 5.0 15.7 15.7
Female 1057 3818300 81022 3968 2.94E+08 54857 15.4 -4.3 93 97 19.2 15.7 15.7

C-2
 Appendix

C.2 Sample bending moment capacity calculations for mullions under

negative wind action

A spreadsheet was developed in Excel to predict the moment capacities of the mullion
couples in accordance with ADM (AA, 2015) in Chapter 6. Sample calculations used to
develop the spread sheet based on ADM (AA, 2015) guidelines are given in this section.
The section properties of the mullion sections were obtained using SAAS shape
designer software (SD, 2013). The yield strength and Young’s modulus values of
aluminium were taken from AS1664 (SA, 1996a or b). Table C.2 shows the properties
of the mullion sections required for the moment capacity calculations (Table C.3).

Table C. 2: Section properties required for the design calculations

Section properties Male mullion Female mullion

yb 72 73.0
yo 9.8 -0.8
yn,b 6.0 6.0
go (yb + yo - yn,b) +75.8 +66.2
βx -7.3 10.7
Iy 79.53E+03 37.96E+03
J 998 1142
Cw 3.13E+08 1.22 E+08
E 69300 69300
Lb 3600 3600
Sxc 22.29E+03 21.23E+03
Fcy 172 172

Note: Figure C.1 shows a representation of the section properties.

Figure C. 1: Representation of the section properties

C-3
 Appendix

Table C. 3: Moment capacity calculations based on ADM (AA, 2015)

Since transverse load is within the unbraced segment, C1 = C2 = 0.5

Mullion Male Female

ܷ ൌ ‫ܥ‬ଵ ݃଴ ൅ ‫ܥ‬ଶ ߚ௫ Ȁʹ = 36.08 35.78

ͳʹǤͷ‫ܯ‬௠௔௫
‫ܥ‬௕ ൌ = 1.136 1.136
ʹǤͷ‫ܯ‬௠௔௫ ൅ ͵‫ܯ‬஺ ൅ Ͷ‫ܯ‬஻ ൅ ͵‫ܯ‬஼

‫ܥ‬௕ ߨ ଶ ‫ܫܧ‬௬ ଶ൅
ͲǤͲ͵ͺ‫ܮܬ‬ଶ௕ ‫ܥ‬௪
‫ܯ‬௘ ൌ ቎ܷ ൅ ඨܷ ൅ ቏ = 681502 397673 Nmm
‫ܮ‬ଶ௕ ‫ܫ‬௬ ‫ܫ‬௬

‫ܵܧ‬௫௖
ߣ ൌ ߨඨ = 149.6 191.1
‫ܯ‬௘

ଵൗ
‫ܨ‬௖௬ ଶ
‫ܤ‬௖ ൌ ‫ܨ‬௖௬ ൥ͳ ൅ ൬ ൰ ൩ = 190.1 190.1
ʹʹͷͲߢ

ଵൗ
‫ܤ‬௖ ‫ܤ‬௖ ଶ
= 0.996 0.996
‫ܦ‬௖ ൌ ൬ ൰
ͳͲ ‫ܧ‬
‫ܤ‬௖
‫ܥ‬௖ ൌ ͲǤͶͳ = 78.3 78.3
‫ܦ‬௖

ఒ గమ ாఒௌೣ೎
‫ܯ‬௡௠௕ ൌ ‫ܯ‬௡௣ ቀͳ െ ቁ ൅ if ɉd 
஼೎ ஼೎య

‫ܯ‬௡௠௕ ൌ ߨ ଶ ‫ܵܧ‬௫௖ Τߣଶ if ɉ൐

Since ɉ൐ ǡ

‫ܯ‬௡௠௕ ൌ ߨ ଶ ‫ܵܧ‬௫௖ Τߣଶ = 681208 397613 Nmm

? Bending moment capacity of the mullion couple = 0.68 + 0.40 = 1.08 kNm

C-4
 Appendix

C.3 Sample bending moment capacity calculations for mullions under

positive wind action

The calculation procedure used in this section is the same as that used in Section C.2.
Table C.4 shows the section properties of the mullions used in the calculations, whereas
the moment capacity calculations are given in Table C.5.

Table C. 4: Section properties required for the design calculations

Section properties Male mullion Female mullion

yb 72 73.0
yo 9.8 -0.8
yp,b 22.5 22.5
go (yb + yo - yp,b) -59.3 -49.7
βx -7.3 10.7
Iy 79.53E+03 37.96E+03
J 998 1142
Cw 3.13E+08 1.22 E+08
E 69300 69300
Lb 3600 3600
Sxc 24.96E+03 23.07E+03
Fcy 172 172

Note: Figure C.2 shows a representation of the section properties.

Figure C. 2: Representation of the section properties

C-5
 Appendix

Table C. 5: Moment capacity calculations based on ADM (AA, 2015)

Since transverse load is within the unbraced segment, C1 = C2 = 0.5

Mullion Male Female

ܷ ൌ ‫ܥ‬ଵ ݃଴ ൅ ‫ܥ‬ଶ ߚ௫ Ȁʹ = -31.48 -22.18

ͳʹǤͷ‫ܯ‬௠௔௫
‫ܥ‬௕ ൌ = 1.136 1.136
ʹǤͷ‫ܯ‬௠௔௫ ൅ ͵‫ܯ‬஺ ൅ Ͷ‫ܯ‬஻ ൅ ͵‫ܯ‬஼

‫ܥ‬௕ ߨ ଶ ‫ܫܧ‬௬ ଶ൅
ͲǤͲ͵ͺ‫ܮܬ‬ଶ௕ ‫ܥ‬௪
‫ܯ‬௘ ൌ ቎ܷ ൅ ඨܷ ൅ ቏ = 352664 259336 Nmm
‫ܮ‬ଶ௕ ‫ܫ‬௬ ‫ܫ‬௬

‫ܵܧ‬௫௖
ߣ ൌ ߨඨ = 220.0 246.7
‫ܯ‬௘

ଵൗ
‫ܨ‬௖௬ ଶ
‫ܤ‬௖ ൌ ‫ܨ‬௖௬ ൥ͳ ൅ ൬ ൰ ൩ = 190.1 190.1
ʹʹͷͲߢ

ଵൗ
‫ܤ‬௖ ‫ܤ‬௖ ଶ
= 0.996 0.996
‫ܦ‬௖ ൌ ൬ ൰
ͳͲ ‫ܧ‬
‫ܤ‬௖
‫ܥ‬௖ ൌ ͲǤͶͳ = 78.3 78.3
‫ܦ‬௖

ఒ గమ ாఒௌೣ೎
‫ܯ‬௡௠௕ ൌ ‫ܯ‬௡௣ ቀͳ െ ቁ ൅ if ɉd 
஼೎ ஼೎య

‫ܯ‬௡௠௕ ൌ ߨ ଶ ‫ܵܧ‬௫௖ Τߣଶ if ɉ൐

Since ɉ൐ ǡ

‫ܯ‬௡௠௕ ൌ ߨ ଶ ‫ܵܧ‬௫௖ Τߣଶ = 352778 259365 Nmm

? Bending moment capacity of the mullion couple = 0.35 + 0.26 = 0.61 kNm

C-6
 Appendix

C.4 Capacity charts for mullion couples under negative wind action

Capacity charts developed for some mullion couples, comprising structural and captive
glazing systems, based on the outcomes of Chapter 6 are shown in Figures C.3(a) to (e).
Figure C.4 shows reduction factors and conditions that need to be used with the capacity
chart of mullion couple MC1 in order to get its final design capacity. Similarly,
reduction factors and conditions can be developed for other mullion couples. Capacity
charts and associated Reduction Factors and Conditions have also been developed for
all other mullion couples (total of 40) mentioned in Chapters 6 and 7. They are included
in Research Reports 4 to 6 and are given to our Industry Partner.

(a) MC1
Figure C.3: Capacity charts for mullion couples under negative wind action
C-7
 Appendix

(b) MC2
Figure C.3: Capacity charts for mullion couples under negative wind action

C-8
 Appendix

(c) MC12
Figure C.3: Capacity charts for mullion couples under negative wind action

C-9
 Appendix

(d) MC13
Figure C.3: Capacity charts for mullion couples under negative wind action

C-10
 Appendix

(e) MC15
Figure C. 3: Capacity charts for mullion couples under negative wind action

C-11
 Appendix

Figure C. 4: Reduction factors and conditions of mullion couple MC1 under

negative wind action

C-12
 Appendix

C.5 Capacity charts for mullion couples under positive wind action

In a similar manner to Section C.4, capacity charts were developed for the same mullion
couples under positive wind action based on the outcomes of Chapter 7, which are
shown in Figures C.5(a) to (e). Figure C.6 shows reduction factors and conditions that
need to be used with the capacity chart of mullion couple MC1. Similarly, reduction
factors and conditions can be developed for other mullion couples.

(a) MC1
Figure C.5: Capacity charts for mullion couples under positive wind action

C-13
 Appendix

(b) MC2
Figure C.5: Capacity charts for mullion couples under positive wind action

C-14
 Appendix

(c) MC12
Figure C.5: Capacity charts for mullion couples under positive wind action

C-15
 Appendix

(d) MC13
Figure C.5: Capacity charts for mullion couples under positive wind action

C-16
 Appendix

(e) MC15
Figure C. 5: Capacity charts for mullion couples under positive wind action

C-17
 Appendix

Figure C. 6: Reduction factors and conditions of mullion couple MC1 under

positive wind action

C-18
 Appendix

C.6 Theoretical deflection of mullions

This section describes the derivation of the relationship between bending moment and
deflection under uniformly distributed loading (UDL). Schematic diagram of a simply
supported beam with UDL is shown in Figure C.7.

Figure C. 7: Simply supported beam with UDL

Based on Gere and Timoshenko (1997):

௪௅௫ ௪௫ మ
Moment at a distance x from support A, ‫ܯ‬௫ ൌ െ
ଶ ଶ

௪௅ ௅ ௪ ௅మ ௪௅మ
? Maximum moment, ‫ ܯ‬ൌ ‫ܯ‬௫ୀಽ ൌ ή െ ή ൌ
మ ଶ ଶ ଶ ଶమ ଼

௪௫
Deflection at a distance x from support A, ‫ݕ‬௫ ൌ ሺ‫ܮ‬ଷ െ ʹ‫ ݔܮ‬ଶ ൅ ‫ ݔ‬ଷ ሻ
ଶସாூ

௪ ௅ ௅మ ௅య ହ௪௅ర
? Maximum deflection, οൌ  ‫ݕ‬௫ୀಽ ൌ ή ή ቀ‫ܮ‬ଷ െ ʹ‫ ܮ‬ή ൅ ቁൌ
మ ଶସாூ ଶ ଶమ ଶయ ଷ଼ସாூ

͵ͺͶ‫ܫܧ‬
o ‫ ݓ‬ൌ  ο
ͷ‫ܮ‬Ͷ

͵ͺͶ‫ܫܧ‬ ‫ʹܮ‬
o ‫ ܯ‬ൌ  ቀ Ͷ οቁ ή
ͷ‫ܮ‬ ͺ

Ͷͺ‫ܫܧ‬
? ‫ ܯ‬ൌ  ήο
ͷ‫ʹܮ‬

where,
 = Young’s modulus of aluminium

= ƒŽ‡൅ ‡ƒŽ‡( ƒŽ‡and ‡ƒŽ‡ are second moment of area of male and female
mullions, respectively)

 = Span

C-19
This Page is Intentionally Left Blank