Shanmuganathan Gunalan Thesis PDF

Structural Behaviour and Design of
Cold-formed Steel Wall Systems
under Fire Conditions
By
Shanmuganathan GUNALAN
School of Urban Development

Faculty of Built Environment and Engineering
Queensland University of Technology
A Thesis Submitted to the School of Urban Development,

Queensland University of Technology
in Partial Fulfillment of the Requirements for the Degree of
DOCTOR of PHILOSOPHY
December 2011
Keywords
KEYWORDS
Wall, Stud, Gypsum plasterboard, Insulation, Light gauge steel frame (LSF), Cold-
formed steel, Fire tests, Load ratios, Load-bearing, Structural behaviour, Finite
element analyses (FEA), Compression, Beam-column, Temperature, Furnace, Fire
design, Lipped-channel section, Local buckling and Flexural buckling.
iii
Abstract
ABSTRACT
In recent times, light gauge steel framed (LSF) structures, such as cold-formed steel
wall systems, are increasingly used, but without a full understanding of their fire
performance. Traditionally the fire resistance rating of these load-bearing LSF wall
systems is based on approximate prescriptive methods developed based on limited
fire tests. Very often they are limited to standard wall configurations used by the
industry. Increased fire rating is provided simply by adding more plasterboards to
these walls. This is not an acceptable situation as it not only inhibits innovation and
structural and cost efficiencies but also casts doubt over the fire safety of these wall
systems. Hence a detailed fire research study into the performance of LSF wall
systems was undertaken using full scale fire tests and extensive numerical studies. A
new composite wall panel developed at QUT was also considered in this study,
where the insulation was used externally between the plasterboards on both sides of
the steel wall frame instead of locating it in the cavity.
Three full scale fire tests of LSF wall systems built using the new composite panel
system were undertaken at a higher load ratio using a gas furnace designed to deliver
heat in accordance with the standard time temperature curve in AS 1530.4 (SA,
2005). Fire tests included the measurements of load-deformation characteristics of
LSF walls until failure as well as associated time-temperature measurements across
the thickness and along the length of all the specimens. Tests of LSF walls under
axial compression load have shown the improvement to their fire performance and
fire resistance rating when the new composite panel was used. Hence this research
recommends the use of the new composite panel system for cold-formed LSF walls.
The numerical study was undertaken using a finite element program ABAQUS. The
finite element analyses were conducted under both steady state and transient state
conditions using the measured hot and cold flange temperature distributions from the
fire tests. The elevated temperature reduction factors for mechanical properties were
based on the equations proposed by Dolamune Kankanamge and Mahendran (2011).
These finite element models were first validated by comparing their results with
experimental test results from this study and Kolarkar (2010). The developed finite
element models were able to predict the failure times within 5 minutes.
v
Abstract
The validated model was then used in a detailed numerical study into the strength of
cold-formed thin-walled steel channels used in both the conventional and the new
composite panel systems to increase the understanding of their behaviour under non-
uniform elevated temperature conditions and to develop fire design rules. The
measured time-temperature distributions obtained from the fire tests were used. Since
the fire tests showed that the plasterboards provided sufficient lateral restraint until
the failure of LSF wall panels, this assumption was also used in the analyses and was
further validated by comparison with experimental results. Hence in this study of
LSF wall studs, only the flexural buckling about the major axis and local buckling
were considered. A new fire design method was proposed using AS/NZS 4600 (SA,
2005), NAS (AISI, 2007) and Eurocode 3 Part 1.3 (ECS, 2006). The importance of
considering thermal bowing, magnified thermal bowing and neutral axis shift in the
fire design was also investigated. A spread sheet based design tool was developed
based on the above design codes to predict the failure load ratio versus time and
temperature for varying LSF wall configurations including insulations.
Idealised time-temperature profiles were developed based on the measured

temperature values of the studs. This was used in a detailed numerical study to fully
understand the structural behaviour of LSF wall panels. Appropriate equations were
proposed to find the critical temperatures for different composite panels, varying in
steel thickness, steel grade and screw spacing for any load ratio. Hence useful and
simple design rules were proposed based on the current cold-formed steel structures
and fire design standards, and their accuracy and advantages were discussed. The
results were also used to validate the fire design rules developed based on AS/NZS
4600 (SA, 2005) and Eurocode Part 1.3 (ECS, 2006). This demonstrated the
significant improvements to the design method when compared to the currently used
prescriptive design methods for LSF wall systems under fire conditions.
In summary, this research has developed comprehensive experimental and numerical

thermal and structural performance data for both the conventional and the proposed
new load bearing LSF wall systems under standard fire conditions. Finite element
models were developed to predict the failure times of LSF walls accurately. Idealized
hot flange temperature profiles were developed for non-insulated, cavity and
vi
Abstract
externally insulated load bearing wall systems. Suitable fire design rules and spread
sheet based design tools were developed based on the existing standards to predict
the ultimate failure load, failure times and failure temperatures of LSF wall studs.
Simplified equations were proposed to find the critical temperatures for varying wall
panel configurations and load ratios. The results from this research are useful to both
structural and fire engineers and researchers. Most importantly, this research has
significantly improved the knowledge and understanding of cold-formed LSF load-
bearing walls under standard fire conditions.
vii
Publications
PUBLICATIONS
Refereed International Conference Papers
1. Gunalan, S. and Mahendran, M. (2010), Structural and Fire Behaviour of a

New Light Gauge Steel Wall System, proceedings of the 6th International
Conference on Structures in Fire, Michigan State University, USA, pp. 43-50.
2. Gunalan, S. and Mahendran, M. (2010), Numerical Modeling of Load
Bearing Steel Stud Walls under Fire Conditions, proceedings of the 21st
Australasian Conference on the Mechanics of Structures and Materials,
Victoria University, Melbourne, Australia, pp. 501-506.
QUT Conference Papers
1. Gunalan, S. and Mahendran, M. (2009), Fire Tests of Steel Wall Systems, 3rd
BEE Postgraduate Research Conference on Smart Systems: Technology,
Systems and Innovation, Queensland University of Technology, Brisbane,
Australia, pp. 30-35.
Proposed International Journal Papers (to be submitted by November 2011)
1. Gunalan, S., Kolarkar, P. and Mahendran, M. (2011a), Experimental Study

on the Structural and Thermal Performance of Load Bearing Wall Systems,
ASCE Journal of Structural Engineering.
2. Gunalan, S. and Mahendran, M. (2011b), Development and Validation of
Finite Element Models of LSF Walls under Fire Conditions, Engineering
Structures.
3. Gunalan, S. and Mahendran, M. (2011c), Development of Fire Design rules
to Predict the Ultimate Failure Capacities of LSF Wall Panels Subjected to
Non-uniform Temperature Distributions, ASCE Journal of Structural
Engineering.
4. Gunalan, S. and Mahendran, M. (2011d), Idealised Time Temperature
Profiles and the Development of Simplified Method to Predict the Failure
Temperatures and Times of LSF Walls under Fire Conditions, Fire Safety
Journal.
ix
Table of Contents
TABLE OF CONTENTS
Keywords …………………………………………………………………………. iii

Abstract ………………………………………………………….....……………… v
Publications ………………………………………………………………..……… ix
Table of Contents ……………………………………………………..…...……… xi
List of Figures ………………………………………………………..……….… xvii
List of Tables ……….……….…………………………………………….…… xxxv
List of Symbols …………………………………………………………..……… xlv
Statement of Original Authorship …………………………………………….… xlix
Acknowledgement…….….…….…….…….……………………………...………. li
CHAPTER 1
1. INTRODUCTION1-1
1.1. General .......................................................................................................... 1-1
1.2. Cold-formed Steel Members ......................................................................... 1-2
1.3. Light Gauge Steel Frame (LSF) Walls.......................................................... 1-3
1.3.1. Elements Used ......................................................................................... 1-3
1.3.2. Plasterboards ........................................................................................... 1-4
1.3.3. Behaviour of LSF Walls in Fire .............................................................. 1-4
1.3.4. Applications of LSF ................................................................................ 1-5
1.4. Fire Safety ..................................................................................................... 1-5
1.4.1. Integrity ................................................................................................... 1-5
1.4.2. Insulation ................................................................................................. 1-6
1.4.3. Structural Adequacy ................................................................................ 1-6
1.4.4. Fire Resistance of LSF Walls.................................................................. 1-6
1.5. Composite Wall Panels ................................................................................. 1-7
1.6. Research Problem.......................................................................................... 1-8
1.7. Research Objectives and Scope .................................................................... 1-9
1.8. Thesis Contents ........................................................................................... 1-10
xi
Table of Contents
CHAPTER 2
2. LITERATURE REVIEW ........................................................................... 2-1

2.1. Introduction ................................................................................................... 2-1
2.2. Cold-formed Steel Members ......................................................................... 2-1
2.2.1. General .................................................................................................... 2-1
2.2.2. Cold-formed Steel Design Standards ...................................................... 2-2
2.3. Light Gauge Steel Frame (LSF) Wall Assemblies ........................................ 2-3
2.3.1. Fire Resistance of LSF Members ............................................................ 2-3
2.3.2. Current Fire Resistance Design Methods for LSF Walls ........................ 2-4
Concentrically Loaded Compression Members ................................................ 2-5
Combined Axial Compression and Bending at Ambient Temperature............. 2-6
2.4. Mechanical and Thermal Properties ............................................................ 2-12
2.4.1. Cold-formed Steels ................................................................................ 2-12
2.4.2. Plasterboards ......................................................................................... 2-18
2.5. Previous Experimental Studies.................................................................... 2-21
2.6. Thermal Modelling ...................................................................................... 2-29
2.7. Structural Modelling ................................................................................... 2-31
2.8. Design Rules Proposed by Other Researchers ............................................ 2-42
2.9. Literature Review Findings ......................................................................... 2-59
CHAPTER 3
3. EXPERIMENTAL STUDY ON THE STRUCTURAL AND THERMAL

PERFORMANCE OF LOAD-BEARING WALL SYSTEMS ............... 3-1
3.1. Introduction ................................................................................................... 3-1
3.2. Test Specimens .............................................................................................. 3-3
3.2.1. Test Specimens 1 and 2 ........................................................................... 3-7
3.2.2. Test Specimen 3 .................................................................................... 3-11
3.3. Test Set-up and Procedure........................................................................... 3-12
3.3.1. Gas Furnace ........................................................................................... 3-12
3.3.2. Loading Arrangement ........................................................................... 3-14
xii
Table of Contents
3.3.3. Displacement Measurement .................................................................. 3-16

3.3.4. Temperature Measurement.................................................................... 3-17
3.4. Test Procedure............................................................................................. 3-18
3.5. Observations and Results ............................................................................ 3-21
3.5.1. Test Specimen 1 (External Insulation - Glass Fibre; Load Ratio 0.2) .. 3-21
3.5.2. Test Specimen 2 (External Insulation - Glass Fibre; Load Ratio 0.4) .. 3-37
3.5.3. Test Specimen 3 (External Insulation - Rock Fibre; Load Ratio 0.4) ... 3-53
3.6. Discussions.................................................................................................. 3-68
3.6.1. Effect of Load Ratios on Wall Specimens Externally Insulated with Glass
Fibre ...................................................................................................... 3-68
3.6.2. Effect of Glass Fibre and Rock Fibre Insulations ................................. 3-72
3.6.3. Stud Temperatures and Failures ............................................................ 3-75
3.6.4. Failure Direction of LSF Walls ............................................................. 3-76
3.6.5. Plasterboard Performance ..................................................................... 3-78
3.7. Conclusions ................................................................................................. 3-80
CHAPTER 4
4. DEVELOPMENT AND VALIDATION OF FINITE ELEMENT

MODELS OF LSF WALLS ....................................................................... 4-1
4.1. Introduction ................................................................................................... 4-1
4.2. Finite Strip Analyses of LSF Wall Studs at Ambient Temperature .............. 4-2
4.3. Finite Element Analyses of LSF Wall Studs at Ambient Temperature ........ 4-4
4.3.1. Element Type .......................................................................................... 4-4
4.3.2. Element Size............................................................................................ 4-6
4.3.3. Mechanical Properties ............................................................................. 4-8
4.3.4. Loading and Boundary Conditions ....................................................... 4-10
4.3.5. Initial Geometric Imperfections ............................................................ 4-16
4.3.6. Residual Stresses ................................................................................... 4-19
4.3.7. Analysis Methods .................................................................................. 4-21
4.3.8. Validation of Finite Element Models .................................................... 4-23
4.4. Finite Element Analyses under Fire Conditions ......................................... 4-28
4.4.1. Element Type and Size.......................................................................... 4-29
xiii
Table of Contents
4.4.2. Loading and Boundary Conditions ....................................................... 4-29

4.4.3. Mechanical Properties ........................................................................... 4-29
4.4.4. Temperature Distribution ...................................................................... 4-34
4.4.5. Initial Geometric Imperfections ............................................................ 4-37
4.4.6. Residual Stresses ................................................................................... 4-37
4.5. Simulation of Test 1 .................................................................................... 4-38
4.5.1. Finite Element Analyses under Transient State Conditions .................. 4-38
4.5.2. Finite Element Analysis under Steady State Conditions ....................... 4-52
4.8. Simulation of Tests Performed by Kolarkar (2010) under Fire Conditions 4-68
4.8.1. Simulation of Test 1* (1x1 Plasterboard) ............................................. 4-69
4.8.2. Simulation of Test 2* (2x2 Plasterboards) ............................................ 4-72
4.8.3. Simulation of Test 3* (Cavity Insulation - Glass Fibre) ....................... 4-75
4.8.4. Simulation of Test 4* (Cavity Insulation - Rock Fibre)........................ 4-78
4.8.5. Simulation of Test 5* (Cavity Insulation - Cellulose Fibre) ................. 4-81
4.8.6. Simulation of Test 6* (Composite Panel - Rock Fibre) ........................ 4-84
4.8.7. Simulation of Test 7* (Composite Panel - Cellulose Fibre) ................. 4-87
4.9. Discussion ................................................................................................... 4-90
4.9.1. Finite Element Analyses under Transient State Conditions .................. 4-90
4.9.2. Finite Element Analyses under Steady State Conditions ...................... 4-91
4.10. Conclusions ................................................................................................ 4-98
CHAPTER 5
5. REVIEW AND DEVELOPMENT OF FIRE DESIGN RULES FOR

AXIALLY LOADED LSF WALL STUDS ............................................... 5-1
5.1. Introduction ................................................................................................... 5-1
5.2. Predictions of Stud Lateral Deflection .......................................................... 5-3
5.2.1. Klippstein (1980)..................................................................................... 5-3
5.2.2. Cooke (1987) ........................................................................................... 5-4
5.2.3. Gerlich et al. (1996) ................................................................................ 5-4
5.2.4. Ranby (1999) ........................................................................................... 5-5
xiv
Table of Contents
5.2.5. Feng and Wang (2005b) .......................................................................... 5-6

5.2.6. Zhao et al. (2005) .................................................................................... 5-6
5.2.7. Lateral Deflections Predicted by FEA and Previous Researches............ 5-7
5.3. Previous Fire Design Rules based on AISI Design Manual ....................... 5-13
5.3.1. Klippstein (1980b) ................................................................................ 5-13
5.3.2. Gerlich et al. (1996) .............................................................................. 5-14
5.3.3. Discussion of Previous Fire Design Rules based on AISI Design
Manual… .............................................................................................. 5-20
5.4. Design Rules in Eurocode 3 ........................................................................ 5-23
5.4.1. Eurocode 3 Part 1.3 (ECS, 2006) .......................................................... 5-23
5.4.2. Eurocode 3 Part 1.2 (ECS, 2005) .......................................................... 5-24
5.5. Previous Fire Design Rules based on Eurocode 3 ...................................... 5-26
5.5.1. Ranby (1999)......................................................................................... 5-26
5.5.2. Wang and Davies (2000)....................................................................... 5-27
5.5.3. Kaitila (2002) ........................................................................................ 5-27
5.5.4. Feng and Wang (2005b) ........................................................................ 5-35
5.5.5. Zhao et al. (2005) .................................................................................. 5-38
5.5.6. Discussion of Previous Fire Design Rules based on Eurocode 3.......... 5-46
5.6. Proposed Fire Design Rules based on Eurocode 3...................................... 5-55
5.6.1. Section Capacities of LSF Wall Studs under Fire Conditions .............. 5-55
5.6.2. Member Capacities of LSF Wall Studs under Fire Conditions ............ 5-65
5.6.3. Section Moment Capacities of LSF Wall Studs under Fire Conditions 5-66
5.6.4. Interaction of Compression and Bending .............................................. 5-70
5.6.5. Thermal Bowing, Neutral Axis Shift and Magnification Effects ......... 5-76
5.7. Design Rules given in AS/NZS 4600 (SA, 2005) ....................................... 5-77
5.7.1. Concentrically Loaded Compression Members at Ambient Temperature
....................... ........................................................................................ 5-77
5.7.2. Combined Axial Compression and Bending at Ambient Temperature 5-78
5.8. Proposed Fire Design Rules based on AS/NZS 4600 (SA, 2005) .............. 5-79
5.9. Comparison of Previous and Proposed Fire Design Rules ......................... 5-85
5.9.1. Failure Times Obtained from Fire Design Rules ................................ 5-85
5.9.2. Modifications Recommended for Previous Fire Design Rules ............. 5-93
5.10. Conclusion ................................................................................................. 5-94
xv
Table of Contents
CHAPTER 6
6. IDEALISED TIME-TEMPERATURE PROFILES AND

PARAMETRIC STUDIES OF LSF WALLS ........................................... 6-1
6.1. Introduction ................................................................................................... 6-1
6.2. Idealised Time - Temperature Profiles .......................................................... 6-2
6.3. Details of Finite Element Models Used in the Parametric Study ................ 6-10
6.4. Elastic Buckling Analyses at Ambient Temperature .................................. 6-13
6.5. Influence of Screw Spacing......................................................................... 6-15
6.6. Validity of Plasterboard Restraint to Hot Flanges under Fire Conditions .. 6-25
6.7. Influence of Steel Grade on the Fire Resistance of LSF Wall Studs .......... 6-36
6.8. Influence of Steel Thickness on the Fire Resistance of LSF Wall Studs .... 6-45
6.9. Critical Hot Flange Temperature of LSF Wall Studs (Simplified Method) 6-54
6.10.Comparison of FEA Results with Fire Design Rules .................................. 6-65
6.11.Failure Times Obtained from Test, FEA, Fire Design Rules and the
Simplified Method....................................................................................... 6-88
6.12.Failure Temperatures Obtained from Test, FEA and Fire Design Rules .... 6-95
6.13.Conclusions ................................................................................................. 6-99
CHAPTER 7
7. CONCLUSIONS AND RECOMMENDATIONS .................................... 7-1

7.1. Full Scale Fire Tests ...................................................................................... 7-5
7.2. Finite Element Modelling.............................................................................. 7-5
7.3. Fire Design Rules .......................................................................................... 7-6
7.4. Idealised Time - Temperature Profiles and Parametric Studies .................... 7-7
7.5. Future Research ............................................................................................. 7-9
xvi
List of Figures
LIST OF FIGURES
CHAPTER 1
Figure 1.1: Light Gauge Steel Framing (LSF) Systems............................................ 1-2

Figure 1.2: Cold-formed Steel Sections .................................................................... 1-3
Figure 1.3: LSF Wall System .................................................................................... 1-4
Figure 1.4: Composite Wall Panels Developed by Kolarkar and Mahendran, 2008.1-7
CHAPTER 2
Figure 2.1: LSF Wall and the Steel Stud Cross-section ............................................ 2-3
Figure 2.2: Thermal Conductivity of Plasterboard ................................................. 2-19
Figure 2.3: Thermal Properties of Plasterboards .................................................... 2-20
Figure 2.4: Test Set-up for Combined Axial Loading and Bending Used by
Gerlich et al. (1996).............................................................................. 2-22
Figure 2.5: Typical Steel Frame Layout of Wall Specimens (Alfawakhiri, 2001) . 2-24
Figure 2.6: Cross-sectional Details of the Wall Specimen (Alfawakhiri, 2001) .... 2-24
Figure 2.7: Composite Wall Panels ........................................................................ 2-27
Figure 2.8: Structural Failure Modes of Alfawakhiri (2001) .................................. 2-31
Figure 2.9: Stud End Conditions and Deflections (Alfawakhiri, 2001).................. 2-33
Figure 2.10: Boundary Conditions used by Feng et al. (2003b) ............................. 2-36
Figure 2.11: Temperature Distributions used by Feng et al. (2003b) ..................... 2-37
Figure 2.12: Boundary Conditions and Loading for Column Models Subject to
Flexural Buckling (Kaitila, 2002) ...................................................... 2-38
Figure 2.13: Boundary Conditions Adopted for Steel Studs (Zhao et al., 2005) .... 2-39
Figure 2.14: Temperature Distribution Used for Numerical Analysis .................... 2-40
Figure 2.15: Heating Condition of Steel Studs for Parametric Studies .................. 2-41
Figure 2.16: Total Horizontal Deflection for Load-bearing Systems According
to Gerlich et al. (1996) ....................................................................... 2-44
Figure 2.17: Neutral Axis Shifts in Lipped Channel Section under Non-uniform
High Temperatures (Feng et al., 2003b) ............................................ 2-47
Figure 2.18: Behaviour of Column due to Non-uniform Temperature Effects
(Feng and Wang, 2005b) .................................................................... 2-50
xvii
List of Figures
CHAPTER 3
Figure 3.1: LSF Wall Frame and Connection Details ............................................... 3-3
Figure 3.2: Thermocouple Fixings ............................................................................ 3-4
Figure 3.3: Plasterboard and Insulation Fixings........................................................ 3-6
Figure 3.4: Wall Making Process of Test Specimens 1 and 2................................... 3-8
Figure 3.5: Wall Making Process of Test Specimen 3 ............................................ 3-11
Figure 3.6: Furnace and Loading Frame ................................................................. 3-13
Figure 3.7: Loading System .................................................................................... 3-15
Figure 3.8: Deflection Measurements Using LVDTs ............................................. 3-16
Figure 3.9: Thermocouple Locations ...................................................................... 3-18
Figure 3.10: Test specimens before Testing............................................................ 3-19
Figure 3.11: Test Specimen 1 after the Fire Test .................................................... 3-22
Figure 3.12: Failure of Studs after the Fire Test ..................................................... 3-23
Figure 3.13: Time - Temperature Plots of Plasterboard Surfaces ........................... 3-27
Figure 3.14: Time - Temperature Profiles of Flange and Web Surfaces of
Central Studs in Test Specimen 1 ...................................................... 3-28
Figure 3.15: Time - Temperature Profiles across Central Studs at Mid-height in
Test Specimen 1 ................................................................................. 3-30
Figure 3.16: Average Temperature Profiles across Central Studs at Mid-height
in Test Specimen 1 ............................................................................. 3-31
Figure 3.17: Temperature Difference across Central Studs at Mid-height ............. 3-31
Figure 3.18: Axial Deformation Profiles for Studs of Test Specimen 1 ................. 3-33
Figure 3.19: Lateral Deflection - Time Plots of Test Specimen 1 .......................... 3-34
Figure 3.20: Axial Compression Load - Time Profile of Test Specimen 1 ............ 3-35
Figure 3.23: Time - Temperature Plots of Plasterboard Surfaces ........................... 3-43
Figure 3.25: Time - Temperature Profiles across the Central Studs at Mid-height
in Test Specimen 2 ............................................................................. 3-46
Figure 3.26: Average Temperature Profiles across Central Studs ......................... 3-47
xviii
List of Figures
Figure 3.27: Temperature Difference across Central Studs ................................... 3-47

Figure 3.29: Lateral Deflection - Time Profiles of Test Specimen 2 ...................... 3-50
Figure 3.33: Time - Temperature Profiles of Plasterboard Surfaces ..................... 3-58
Figure 3.35: Time - Temperature Profiles across the Central Studs at Mid-height
in Test Specimen 3 ............................................................................. 3-61
Figure 3.36: Average Temperature Profiles across Central Studs ......................... 3-62
Figure 3.37: Temperature Difference across Central Studs at Mid-height ............. 3-63
Figure 3.39: Lateral Deflection - Time Profiles of Test Specimen 3 ...................... 3-65
Figure 3.41: Time - Temperature Profiles for Test Specimens 1 and 2 .................. 3-69
Figure 3.42: Time - Temperature Profiles for the Central Studs ........................... 3-70
Figure 3.43: Lateral Deflection - Time for the Central Studs at Mid-height in
Test Specimens 1 and 2 ...................................................................... 3-71
Figure 3.44: Time - Temperature Profiles for Test Specimens 2 and 3 .................. 3-72
Figure 3.45: Time - Temperature Profiles for the Central Studs ........................... 3-73
Figure 3.46: Lateral Deflection - Time for the Central Studs ................................ 3-74
CHAPTER 4
Figure 4.1: Cross-sectional Dimensions and Boundary Conditions of the LSF

Wall Stud Used in CUFSM Analyses ..................................................... 4-2
Figure 4.2: Results from Finite Strip Analyses ......................................................... 4-3
Figure 4.3: Comparison of Load - Deflection Curves Using Different Shell
Element Types ........................................................................................ 4-6
Figure 4.4: Effect of Element Size on Memory and Analysis Time for Elastic
Buckling and Nonlinear Analyses .......................................................... 4-7
xix
List of Figures
Figure 4.5: Stress – Strain Models ............................................................................ 4-8

Figure 4.6: Stress – Strain Models Used in FEA .................................................... 4-10
Figure 4.7: Comparison of Nonlinear Responses for Elastic-Perfect-Plastic and
Strain Hardening Models ...................................................................... 4-10
Figure 4.8: Test Specimen and Set-up Used by Kolarkar (2010) ........................... 4-11
Figure 4.9: Modelling of LSF Walls ....................................................................... 4-11
Figure 4.10: Loading and Boundary Conditions Used in FEA ............................... 4-12
Figure 4.11: Eccentric Loading and Support Conditions ........................................ 4-13
Figure 4.12: Effect of Eccentric Loading and Support Conditions ......................... 4-14
Figure 4.13: Simulation of Plasterboard Restraints ................................................ 4-15
Figure 4.14: First Eigen Mode from Bifurcation Buckling Analysis ...................... 4-16
Figure 4.15: Initial Geometric Imperfection (Schafer and Pekoz, 1996) ................ 4-18
Figure 4.16: Effect of Initial Geometric Imperfection on Ultimate Load ............... 4-19
Figure 4.17: Flexural Residual Stress Distributions Assumed in FEA ................... 4-20
Figure 4.18: Comparison of the Effect of Riks On and Off Methods ..................... 4-22
Figure 4.19: Ultimate Load versus Deformation Curves from the Riks On
Method… ............................................................................................ 4-22
Figure 4.20: Comparison of Load - Deformation Curves from Test and FEA ....... 4-24
Figure 4.21: Elastic Buckling Modes Obtained from Finite Strip and Finite
Element Analyses ............................................................................... 4-25
Figure 4.22: Failure Modes from Test and Nonlinear Finite Element Analyses .... 4-26
Figure 4.23: Stress Distribution of the Stud at Failure ............................................ 4-27
Figure 4.24: Reduction in Yield Stress based on Dolamune Kankanamge and
Mahendran (2011) and Eurocode 3 Part 1.2 (ECS, 2005) .................. 4-31
Figure 4.25: Reduction in Elastic Modulus based on Dolamune Kankanamge
and Mahendran (2011) and Eurocode 3 Part 1.2 (ECS, 2005) ........... 4-32
Figure 4.26: Stress - Strain Curves of G500 Steels at Different Temperatures ...... 4-33
Figure 4.27: Temperature Distributions Assumed in the Previous Studies ............ 4-35
Figure 4.28: Temperature Distribution Used in FEA .............................................. 4-35
Figure 4.29: Average Temperature Profiles Used in FEA for the Central Studs
of Test 1 .............................................................................................. 4-39
Figure 4.30: Axial Shortening of Stud 3 after Step 1 (Axial Compressive Load)
under Transient State Conditions........................................................ 4-40
xx
List of Figures
Figure 4.31: Thermal Bowing and Expansion of Stud 3 under Transient State
Conditions........................................................................................... 4-41
Figure 4.32: Response for the Static Stabilize Method for Stud 2 of Test 1 .......... 4-43
Figure 4.33: FEA Results for the Central Studs of Test 1 under Transient State
Conditions........................................................................................... 4-45
Figure 4.34: Prediction of Lateral Deflection at Mid-height for Stud 2 of Test 1
under Transient State Conditions ....................................................... 4-46
Figure 4.35: Failure Modes from Test and FEA for Stud 3 of Test 1 under
Transient State Conditions.................................................................. 4-48
Figure 4.36: Stress Distribution from FEA for Stud 3 of Test 1 under Transient
State Conditions.................................................................................. 4-49
Figure 4.37: von Mises Stress Results from FEA for Stud 3 of Test 1 under
Transient State Conditions.................................................................. 4-50
Figure 4.38: Variation of Applied Load from Test and FEA for Stud 3 of Test 1
under Transient State Conditions ....................................................... 4-51
Figure 4.39: FEA Results for Stud 3 of Test 1 under Steady State Conditions ...... 4-53
Figure 4.40: Stress Distribution from FEA for Stud 3 of Test 1 under Steady
State Conditions.................................................................................. 4-54
Figure 4.41: Failure Modes from FEA for Stud 3 of Test 1 under Steady State
Conditions........................................................................................... 4-56
Figure 4.42: Failure Modes from FEA for Stud 3 of Test 1 ................................... 4-57
Figure 4.43: Applied Load versus Lateral Deflection Plots from FEA for Stud 3
of Test 1 under Steady State Conditions ............................................ 4-57
of Test 2 .............................................................................................. 4-58
Conditions........................................................................................... 4-59
Figure 4.46: Variation of Applied Load from Test and FEA under Transient
State Conditions for Stud 3 of Test 2 ................................................. 4-60
Figure 4.48: Failure Modes from Test and FEA under Transient and Steady
State Conditions for Stud 3 of Test 2 ................................................. 4-62
xxi
List of Figures
of Test 3 .............................................................................................. 4-63
Conditions ........................................................................................... 4-64
Figure 4.51: Variation of Applied Axial Compression Load from Test and FEA
under Transient State Conditions for Stud 2 of Test 3 ....................... 4-65
Figure 4.53: Failure Modes from Test and FEA for Stud 2 of Test 3 ..................... 4-67
Figure 4.54: Temperature Profiles Used in FEA for Stud 2 of Test 1* .................. 4-69
Figure 4.55: FEA Results for Stud 2 of Test 1* under Steady State Conditions .... 4-70
Figure 4.56: Failure Modes from Test and FEA under Steady State Conditions
for Stud 2 of Test 1* ........................................................................... 4-71
for Stud 2 of Test 2* ........................................................................... 4-74
for Stud 2 of Test 3* ........................................................................... 4-77
for Stud 2 of Test 4* ........................................................................... 4-80
for Stud 3 of Test 5* ........................................................................... 4-83
for Stud 2 of Test 6* ........................................................................... 4-86
xxii
List of Figures
for Stud 2 of Test 7* ........................................................................... 4-89
Figure 4.75: Variation of Load Ratio with Time under Steady State Conditions ... 4-93
Figure 4.76: Variation of Load Ratio with Hot Flange Temperature from FEA
under Steady State Conditions ............................................................ 4-95
Figure 4.77: Comparison of Failure Times from FEA and Tests ........................... 4-97
CHAPTER 5
Figure 5.1: Estimated Failure Deflection of Studs in Wall Panels ........................... 5-3
Figure 5.2: Lateral Deflections of LSF Wall Stud .................................................... 5-5
Figure 5.3: Lateral Deflection with Time from Experiment, FEA and Previous
Fire Design Rules for Test 1 ................................................................... 5-8
Figure 5.6: Lateral Deflection with Time from Experiment and Previous Fire
Design Rules for Test 1* ........................................................................ 5-9
Design Rules for Test 2* ...................................................................... 5-10
Design Rules for Test 5* .................................................................... 5-11
xxiii
List of Figures
Figure 5.13: Variation of Load Ratio with Time from FEA, Klippstein (1980)
and Gerlich et al. (1996) for Test 1 .................................................... 5-15
and Gerlich et al. (1996) for Test 1* .................................................. 5-16
Figure 5.23: Influence of Effective Area in Kaitila’s (2002) Uniform
Temperature Method for Test 1 ......................................................... 5-28
Figure 5.24: Variation of Load Ratio with Time from FEA, Ranby (1999) and
Kaitila (2002) for Test 1 ..................................................................... 5-30
Kaitila (2002) for Test 2 ..................................................................... 5-30
Kaitila (2002) for Test 3 ..................................................................... 5-31
Kaitila (2002) for Test 1* ................................................................... 5-31
Kaitila (2002) for Test 2* ................................................................... 5-32
xxiv
List of Figures
Kaitila (2002) for Test 3*................................................................... 5-32
Kaitila (2002) for Test 4*................................................................... 5-33
Kaitila (2002) for Test 5*................................................................... 5-33
Kaitila (2002) for Test 6*................................................................... 5-34
Kaitila (2002) for Test 7*................................................................... 5-34
Figure 5.34: Effective Cross-section of LSF Wall Stud ......................................... 5-36
Figure 5.35: Variation of Load Ratio with Time from FEA, Feng and Wang
(2005b) and Zhao et al. (2005) for Test 1 .......................................... 5-41
(2005b) and Zhao et al. (2005) for Test 1* ........................................ 5-42
Figure 5.45: Variation of Ultimate Load with Time from Previous Fire Design
Rules for Test 1 .................................................................................. 5-48
xxv
List of Figures
Figure 5.46: Variation of Effective Eccentricity with Time from Previous Fire
Design Rules for Test 1 ...................................................................... 5-49
Figure 5.47: Variation of Ultimate Failure Load with Time for Local Buckling
from Previous Fire Design Rules for Test 1 ....................................... 5-50
Figure 5.48: Variation of Effective Area with Time for Compression Capacity
from Previous Fire Design Rules for Test 1 ....................................... 5-50
Figure 5.49: Variation of Section Moment Capacity with Time from Previous
Fire Design Rules for Test 1 .............................................................. 5-53
Figure 5.50: Variation of Effective Area with Time for Section Moment
Capacity from Previous Fire Design Rules for Test 1 ....................... 5-53
Figure 5.51: Variation of Reduction Factor χ with Time from Previous Fire
Figure 5.52: Variation of Reduction Factor kxx with Time from Previous Fire
Figure 5.53: Ultimate Failure Mode of Short LSF Wall Stud under Fire
Conditions.. ........................................................................................ 5-56
Figure 5.54: Variation of Ultimate Local Buckling Capacity for Test 1 ................ 5-57
Figure 5.55: Variation of Ultimate Local Buckling Capacity for Test 1* .............. 5-57
Figure 5.58: Variation of Yield Stress with Time ................................................... 5-59
Figure 5.59: Variation of Load Ratio with Time for Test 1 .................................... 5-60
Figure 5.60: Variation of Load Ratio with Time for Test 1* .................................. 5-60
Figure 5.63: von Mises Stress Distribution at Failure for Test 4* .......................... 5-62
Figure 5.64: Proposed Method for the Selected Options ........................................ 5-65
Figure 5.65: Interaction of Compression and Bending for Test 1 ........................... 5-68
Figure 5.66: Interaction of Compression and Bending for Test 1* ......................... 5-68
Figure 5.69: Variation of Bending Moment Capacity for Test 1 ............................ 5-72
Figure 5.70: Variation of Ultimate Compression Capacity for Test 1 .................... 5-72
xxvi
List of Figures
Figure 5.71: Variation of Bending Moment Capacity for Test 1* .......................... 5-73
Figure 5.72: Variation of Ultimate Compression Capacity for Test 1* .................. 5-73
Figure 5.78: Variation of Load Ratio with Time from FEA and Proposed Fire
CHAPTER 6
Figure 6.1: Idealised Time - Temperature Profiles for LSF Walls ........................... 6-6
xxvii
List of Figures
Figure 6.9: Local Buckling Mode for Cases 1 to 4 ................................................. 6-14
Figure 6.10: Flexural Buckling Mode about the Major Axis for Case 5 ................ 6-14
Figure 6.11: Ultimate Failure Modes Obtained for LSF Wall Studs ...................... 6-16
Figure 6.12: Ultimate Load Variation with Time for 1.15 mm G500 Steel Studs .. 6-17
Figure 6.13: FEA Results for 1.15 mm G500 Steel Studs Lined on Both Sides
by a Single Layer of Plasterboard with Different Screw Spacing...... 6-21
by Two Layers of Plasterboard with Different Screw Spacing .......... 6-21
by Two Layers of Plasterboard with Glass Fibre Used as Cavity
Insulation ............................................................................................ 6-22
by Two Layers of Plasterboard with Rock Fibre Used as Cavity
Insulation ............................................................................................ 6-22
by Two Layers of Plasterboard with Cellulose Fibre Used as Cavity
Insulation ............................................................................................ 6-23
by Two Layers of Plasterboard with Glass Fibre Used as External
Insulation ............................................................................................ 6-23
by Two Layers of Plasterboard with Rock Fibre Used as External
Insulation ............................................................................................ 6-24
by Two Layers of Plasterboard with Cellulose Fibre Used as
External Insulation with Different Screw Spacing ............................. 6-24
xxviii
List of Figures
Figure 6.22: FEA Results for 1.15 mm G500 Steel Studs Lined by a Single
Layer of Plasterboard at 300 mm Screw Spacing .............................. 6-31
Figure 6.23: FEA Results for 1.15 mm G500 Steel Studs Lined by Two Layers
of Plasterboard at 300 mm Screw Spacing ........................................ 6-31
of Plasterboard at 300 mm Screw Spacing with Glass Fibre ............. 6-32
of Plasterboard at 300 mm Screw Spacing with Rock Fibre ............. 6-32
Figure 6.26: FEA Results for 1.15 mm G500 Steel Studs Lined by Two Layers of
Plasterboard at 300 mm Screw Spacing with Cellulose Fibre ..........6-33
of Plasterboard at 300mm Screw Spacing with Glass Fibre ............. 6-33
Figure 6.28: FEA Results for 1.15 mm G500 Steel Studs Lined by Two layers
of Plasterboard at 300 mm Screw Spacing with Rock Fibre ............ 6-34
Figure 6.29: FEA Results for 1.15 mm G500 Steel Studs Lined by Two layers of
Plasterboard at 300 mm Screw Spacing with Cellulose Fibre .......... 6-34
Figure 6.31: FEA Results for 1.15 mm LSF Wall Studs Lined on Both Sides by
a Single Layer of Plasterboard at 300 mm Screw Spacing ................ 6-41
Two Layers of Plasterboard at 300 mm Screw Spacing .................... 6-41
Figure 6.33: FEA Results for 1.15 mm LSF Wall Studs Lined on Both Sides
by Two Layers of Plasterboard at 300 mm Screw Spacing with
Glass Fibre ........................................................................................ 6-42
Two Layers of Plasterboard at 300 mm Screw Spacing with Rock
Fibre .................................................................................................. 6-42
Two Layers of Plasterboard at 300 mm Screw Spacing with
Cellulose Fibre Used as Cavity Insulation ......................................... 6-43
Two Layers of Plasterboard at 300 mm Screw Spacing with Glass
Fibre .................................................................................................. 6-43
xxix
List of Figures
Fibre .................................................................................................. 6-44
Cellulose Fibre Used as External Insulation ...................................... 6-44
Figure 6.40: FEA Results for G250 Steel Studs with Different Thicknesses Lined
on Both Sides by a Single Layer of Plasterboard at 300 mm Screw
Spacing ............................................................................................... 6-50
on Both Sides by Two Layers of Plasterboard at 300 mm Screw
Spacing ............................................................................................... 6-50
Figure 6.42: FEA Results for G250 Steel Studs with Different Thicknesses
Lined on Both Sides by Two Layers of Plasterboard at 300 mm
Screw Spacing with Glass Fibre Used as Cavity Insulation .............. 6-51
Screw Spacing .................................................................................... 6-51
Screw Spacing .................................................................................... 6-52
Screw Spacing .................................................................................... 6-52
Screw Spacing .................................................................................... 6-53
Screw Spacing .................................................................................... 6-53
xxx
List of Figures
Figure 6.48: Comparison of FEA Results with Previous Studies for 1.15 mm
G500 Steel Studs Lined on Both Sides by a Single Layer of
Plasterboard ........................................................................................ 6-57
G500 Steel Studs Lined on Both Sides by Two Layers of
Plasterboard ........................................................................................ 6-57
Plasterboard with Glass Fibre, Rock Fibre or Cellulose Fibre
Used as Cavity Insulation…………………………………………...6-58
G500 Steel Studs on Both Sides by Two Layers of Plasterboard
with Glass Fibre, Rock Fibre or Cellulose Fibre Used as External
Insulation ............................................................................................ 6-58
Plasterboard ........................................................................................ 6-59
Plasterboard ...................................................................................... 6-59
Plasterboard with Glass Fibre, Rock Fibre or Cellulose Fibre Used
as Cavity Insulation ............................................................................ 6-60
as External Insulation ......................................................................... 6-60
Plasterboard ........................................................................................ 6-61
Plasterboard ........................................................................................ 6-61
xxxi
List of Figures
as Cavity Insulation ............................................................................ 6-62
as External Insulation ......................................................................... 6-62
Figure 6.60: Comparison of Proposed Fire Design Rules and the Simplified
Method for 1.15 mm G500 Steel Studs Lined on Both Sides ............ 6-70
Method for 1.15 mm G500 Steel Studs Lined on Both Sides by Two
Layers of Plasterboard with Glass Fibre Used as Cavity Insulation .. 6-71
Layers of Plasterboard with Rock Fibre Used as Cavity Insulation .. 6-71
Figure 6.64: Comparison of Proposed Fire Design Rules and the Simplified Method
for 1.15 mm G500 Steel Studs Lined on Both Sides by Two Layers
of Plasterboard with Cellulose Fibre Used as Cavity Insulation ........ 6-72
of Plasterboard with Glass Fibre Used as External Insulation ........... 6-72
of Plasterboard with Rock Fibre Used as External Insulation ........... 6-73
of Plasterboard with Cellulose Fibre Used as External Insulation ..... 6-73
xxxii
List of Figures
of Plasterboard with Rock Fibre Used as Cavity Insulation .............. 6-78
of Plasterboard with Cellulose Fibre Used as Cavity Insulation ....... 6-79
Layers of Plasterboard with Rock Fibre Used as Cavity Insulation .. 6-85
of Plasterboard with Cellulose Fibre Used as Cavity Insulation ....... 6-86
xxxiii
List of Figures
Figure 6.84: Reduction Factors for Yield Strength at Elevated Temperatures ....... 6-98
xxxiv
List of Tables
LIST OF TABLES
CHAPTER 2
Table 2.1: Strength Reduction Factors for Cold-formed Steels ................................ 2-9
Table 2.2: Strength Reduction Factors for Cold-formed Steel at Elevated
Temperatures (SCI, 1993)...................................................................... 2-10
Table 2.3: Fire Resistance of Typical Floors, Walls and Partitions ....................... 2-11
Table 2.4: Reduction Factors Recommended in Eurocode 3 Part 1.2 (ECS, 2005) 2-15
Table 2.5: Reduction Factors for Steel Load-bearing Elements at High
Temperatures Proposed by Zhao et al. (2005) ....................................... 2-16
Table 2.6: Material Properties Used by Feng et al. (2003a) ................................... 2-20
Table 2.7: Details of Full Scale Fire Tests of Gerlich et al. (1996) ........................ 2-21
Table 2.8: Fire Test Specimens, Failure Modes and Failure Times of Panels
(Feng and Wang, 2005a) ........................................................................ 2-25
Table 2.9: Details of LSF Wall Specimens Tested by Kolarkar (2010) ................. 2-28
Table 2.10: Failure Criteria for Calculating the Maximum Bending Moment
Capacity (Feng et al., 2003b) .............................................................. 2-49
CHAPTER 3
Table 3.1: Details of LSF Wall Specimens Tested by Kolarkar (2010) ................... 3-1
Table 3.2: Details of Test Specimen Configurations in the Current Study............... 3-2
Table 3.3: Temperature and Lateral Deflection Measurements for Test Specimen 1
............................................................................................................ ..3-36
.............................................................................................................. 3-52
............................................................................................................ ..3-67
Table 3.6: Failure Times of Test Specimens ........................................................... 3-68
Table 3.7: Thermal and Structural Responses of Test Specimen 1......................... 3-75
Table 3.10: Temperatures of Pb2 Surfaces at Failure ............................................. 3-80
xxxv
List of Tables
CHAPTER 4
Table 4.1: Comparison of Different Types of Four Node Shell Elements ................ 4-5
Table 4.2: Effect of Mesh Sizes on Analysis Time, Memory and Ultimate Load .... 4-7
Table 4.3: Comparison of Ultimate Loads of Studs with Different Eccentricities . 4-13
Table 4.4: Effects of Different Cases of Plasterboard Restraints ............................ 4-15
Table 4.5: Effect of Initial Imperfection Amplitude on the Ultimate Load ............ 4-18
Table 4.6: Failure Time and Temperature Predicted by FEA for Stud 3 of Test 1
under Steady State Conditions ............................................................... 4-54
Table 4.9: Fire Tests Conducted by Kolarkar (2010).............................................. 4-68
Table 4.10: Failure Time and Temperature Predicted by FEA for Stud 2 of
Test 1* under Steady State Conditions................................................. 4-71
Table 4.17: Failure Time Predicted by FEA under Transient State Conditions ..... 4-90
Table 4.18: Failure Times Predicted by FEA under Steady State Conditions ........ 4-96
xxxvi
List of Tables
CHAPTER 5
Table 5.1: Failure Times Obtained from Experiments and FEA .............................. 5-2
Table 5.2: Critical Stress and Effective Area ......................................................... 5-22
Table 5.3: Summary of Effective Areas under Concentric Compression
Calculated based on Eurocode 3 Part 1.3 (ECS, 2006) ......................... 5-37
Table 5.4: Summary of Effective Areas under Bending about the Major Axis
Calculated based on Eurocode 3 Part 1.3 (ECS, 2006) ......................... 5-37
Table 5.5: Summary of Bending Moment Capacities ............................................. 5-38
Table 5.6: Mechanical Properties Used in Different Effective Areas ..................... 5-56
Table 5.7: Yield Stresses at Hot and Cold Flange Temperatures ........................... 5-63
Table 5.8: Comparison of FEA with Current and Proposed Design Rules ............. 5-64
Table 5.9: Stress Distribution for Bending Moment Capacities ............................. 5-67
Table 5.10: Effective Eccentricities Considered in the Current Study ................... 5-76
Table 5.11: Effective Areas for Previous and Proposed Fire Design Rules ........... 5-86
Table 5.12: Comparison of Previous and Proposed Fire Design Rules .................. 5-87
Table 5.13: Failure Times Predicted by FEA and the Previous Fire Design Rules
for Load Ratio of 0.2 ............................................................................ 5-88
for Load Ratio of 0.4 ............................................................................ 5-89
for Load Ratio of 0.7 ............................................................................ 5-90
Table 5.16: Failure Times Predicted by the Proposed Fire Design Rules Using
AS/NZS 4600 (SA, 2005) .................................................................... 5-91
Table 5.17: Failure Times Predicted by Proposed Fire Design Rules Using
Eurocode 3 Part 1.3 (ECS, 2006) ......................................................... 5-92
CHAPTER 6
Table 6.1: LSF Wall Test Configurations with Different Insulation Materials ........ 6-2
Table 6.2: Idealised Time - Temperature Values up to 100oC .................................. 6-3
Table 6.3: Parameters Considered in the Investigation of LSF Walls .................... 6-13
Table 6.4: Overview of Sections 6.5 to 6.8 ............................................................. 6-13
xxxvii
List of Tables
Table 6.5: Ultimate Loads Obtained from FEA for 1.15 mm G500 Steel Studs
Lined on Both Sides by Two Layers of Plasterboard ............................ 6-17
Table 6.6: FEA Results for 1.15 mm G500 Steel Studs .......................................... 6-18
Table 6.7: FEA Results for 1.15 mm G500 Steel Studs .......................................... 6-18
Table 6.8: FEA Results for 1.15 mm G500 Steel Studs Lined on Both Sides by
Two Layers of Plasterboard with Glass Fibre Used as Cavity
Insulation ................................................................................................ 6-18
Two Layers of Plasterboard with Rock Fibre Used as Cavity
Insulation ................................................................................................ 6-19
Two Layers of Plasterboard with Cellulose Fibre Used as Cavity
Insulation .............................................................................................. 6-19
Two Layers of Plasterboard with Glass Fibre Used as External
Insulation .............................................................................................. 6-19
Two Layers of Plasterboard with Rock Fibre Used as External
Insulation .............................................................................................. 6-20
Two Layers of Plasterboard with Cellulose Fibre Used as External
Insulation .............................................................................................. 6-20
Table 6.14: FEA Results for 1.15 mm G500 Steel Studs Lined by a Single
Layer of Plasterboard at 300 mm Screw Spacing ................................ 6-28
Table 6.15: FEA Results for 1.15 mm G500 Steel Studs Lined by Two Layers
of Plasterboard at 300 mm Screw Spacing ........................................... 6-28
of Plasterboard at 300 mm Screw Spacing with Glass Fibre Used as
Cavity Insulation .................................................................................. 6-28
of Plasterboard at 300 mm Screw Spacing with Rock Fibre Used as
Cavity Insulation .................................................................................. 6-29
xxxviii
List of Tables
of Plasterboard at 300 mm Screw Spacing with Cellulose Fibre
Used as Cavity Insulation ..................................................................... 6-29
of Plasterboard at 300 mm Screw Spacing with Glass Fibre Used as
External Insulation ............................................................................... 6-29
of Plasterboard at 300 mm Screw Spacing with Rock Fibre Used as
Table 6.21: FEA Results for 1.15 mm G500 Steel Studs Lined by Two Layers of
Plasterboard at 300 mm Screw Spacing with Cellulose Fibre Used as
Table 6.22: FEA Failure Times of LSF Wall Panels .............................................. 6-35
Table 6.23: FEA Results for 1.15 mm LSF Wall Studs Lined on Both Sides by a
Single Layer of Plasterboard at 300 mm Screw Spacing ..................... 6-38
Table 6.24: FEA Results for 1.15 mm LSF Wall Studs Lined on Both Sides by
Two Layers of Plasterboard at 300 mm Screw Spacing ...................... 6-38
Fibre Used as Cavity Insulation ........................................................... 6-38
Two Layers of Plasterboard at 300 mm Screw Spacing with Cellulose
Fibre Used as Cavity Insulation as Cavity Insulation .......................... 6-39
Fibre Used as External Insulation ........................................................ 6-39
xxxix
List of Tables
Cellulose Fibre Used as External Insulation ........................................ 6-40
Table 6.31: FEA Results for G250 Steel Studs Lined on Both Sides by a Single
Layer of Plasterboard at 300 mm Screw Spacing ................................ 6-47
Table 6.32: FEA Results for G250 Steel Studs Lined on Both Sides by Two
Layers of Plasterboard at 300 mm Screw Spacing ............................... 6-47
Layers of Plasterboard at 300 mm Screw Spacing with Glass Fibre
Layers of Plasterboard at 300 mm Screw Spacing with Rock Fibre
Table 6.35: FEA Results for G250 Steel Stud Lined on Both Sides by Two
Layers of Plasterboard at 300 mm Screw Spacing with Cellulose
Layers of Plasterboard at 300 mm Screw Spacing with Glass Fibre
Used as External Insulation .................................................................. 6-48
Layers of Plasterboard at 300 mm Screw Spacing with Rock Fibre
Used as External Insulation .................................................................. 6-49
Layers of Plasterboard at 300 mm Screw Spacing with Cellulose
Table 6.39: Comparison of Proposed Fire Design Rules and the Simplified
Method for 1.15 mm G500 Steel Studs Lined on Both Sides .............. 6-67
Layers of Plasterboard with Glass Fibre Used as Cavity Insulation .... 6-67
xl
List of Tables
Layers of Plasterboard with Rock Fibre Used as Cavity Insulation .... 6-68
Table 6.43: Comparison of Proposed Fire Design Rules and the Simplified Method
of Plasterboard with Cellulose Fibre Used as Cavity Insulation .......... 6-68
of Plasterboard with Glass Fibre Used as External Insulation ............. 6-68
of Plasterboard with Rock Fibre Used as External Insulation ............. 6-69
of Plasterboard with Cellulose Fibre Used as External Insulation ....... 6-69
Layers of Plasterboard with Rock Fibre Used as Cavity Insulation .... 6-75
xli
List of Tables
of Plasterboard with Rock Fibre Used as External Insulation.............. 6-76
Layers of Plasterboard with Rock Fibre Used as Cavity Insulation..... 6-82
of Plasterboard with Rock Fibre Used as External Insulation.............. 6-83
Table 6.63: Failure Times Predicted for 1.15 mm G500 Steel Studs Lined on
Both Sides by Plasterboard for a Load Ratio of 0.2 ............................. 6-89
xlii
List of Tables
Table 6.72: FEA Failure Temperatures of LSF Wall Panels for a Load Ratio
of 0.2..................................................................................................... 6-97
of 0.4..................................................................................................... 6-97
of 0.7..................................................................................................... 6-98
xliii
List of Symbols
LIST OF SYMBOLS
A gross cross-sectional area of the stud

Aeff effective area of the steel cross-section
Aeff,i effective element area
Ai area of element i
A*e elasticity-modulus-weighted effective stud section area in compression
Bx bending stiffness
C torsional stiffness
Cmx, Cmy coefficients for unequal end moments
Cs specific heat
Cw warping stiffness
c distance from the centroidal axis of the ‘modulus-weighted’ section to the
extreme fibre of the cold flange
d stud section depth
E elastic modulus
(EI)fi,eff effective flexural stiffness of the cross-section
ET, EO modulus of elasticity at elevated and ambient temperatures, respectively
Ei,θ modulus of elasticity of each plate element
e effective eccentricity
ex, ey shift of neutral axis in the major and minor axis, respectively
e P∆T P-∆ deflection
e0 deflection related to thermal bowing
e∆E shift of neutral axis about the major axis
e∆T maximum mid-height horizontal deflection due to temperature gradient
across the section
FalT allowable stress for axial load at failure temperature
FyH, FyC yield strengths of steel at temperatures TH and TC, respectively
Fyx calculated steel stress at height z
fH, fC compressive stresses at the extreme fibre of the hot flange and cold flange,
respectively
foc elastic buckling stress
fy yield strength
xlv
List of Symbols
fyT, fyo yield stress at temperature T and room temperature, respectively

G shear modulus
I second moment of area of the gross section
Ib second moment of area of the full, unreduced cross-section about the
bending axis
Ieff effective second moment of area of the steel cross-section
Ii moment of inertia of the element i about its own neutral axis parallel
to flanges
Ii,θ. second moment of area of each element
It torsion constant
Ix second moment of area about the major axis
Iw warping constant
I* elasticity-modulus-weighted moment of inertia of the unreduced stud
section about the neutral axis parallel to flanges
i coefficient of restraint along the column at the flange in tension
kE,θ slope of the linear elastic range at steel temperature θ
ks thermal conductivity
kxx, kxy, kyx, kyy interaction factors
ky,θ reduction factors for the steel yield strength
L member height
Lθ buckling length of the stud in fire situation
l buckling length
le effective length of member
leb effective length in the plane of bending
Mbx, Mby nominal member moment capacities about the major and minor axes,
respectively
Mfi,Rd,x, Mfi,Rd,y design bending moment resistance that is subjected to bending
moment only about the major and the minor axis, respectively
Mx,Ed, My,Ed applied bending moments about the major and minor axes,
respectively
Mx,eff, My,eff stud bending moment capacities at non-uniform elevated temperatures
about the major and the minor axis, respectively
NEd applied axial load
xlvi
List of Symbols
Nb,fi,t,Rd design resistance at time t of a compression member with a non-uniform

temperature distribution
Nb,fi,θ design resistance of a compression member with a uniform steel
temperature θ
Nc nominal member capacity of the member in compression
Ncr elastic critical force for the relevant buckling mode based on the gross
cross-sectional properties
Neff stud axial compression capacity at elevated temperatures
Nfi,c,Rd, design resistance load of the steel stud
Nfi,cr Euler buckling load
Nfi,Rd design buckling resistance of a compression member
Ns nominal section capacity of the member in compression
Nu ultimate axial compressive load
nH, nC elastic modulus reduction factor for temperature TH and TC, respectively
ni reduction factor for E at temperature Ti
PT predicted failure load of the stud at an elevated temperature T
Pa applied axial load
r radius of gyration of the full, unreduced cross-section
Sx plastic section modulus about major axis
S*eH, S*eC elasticity-modulus-weighted effective stud section modulus in bending
that causes compression of hot flange and cold flange, respectively
T temperature
TA average stud temperature
TCF cold flange temperature
THF hot flange temperature
Ti temperature of element
Weff,x, Weff,y effective section modulus for the maximum compressive stress in an
effective cross-section that is subject only to moment about the major and
minor axes, respectively
xi distance of element i from the extreme fibre of the cold flange
y distance from the effective plate element i to the neutral axis at ambient
temperature
y CF distance from the supported colder flange to the effective neutral axis
xlvii
List of Symbols
Zx elastic section modulus about major axis
α imperfection factor based on the relevant buckling curve

α thermal expansion coefficient for steel
αT thermal expansion coefficient for steel
αnx, αny moment amplification factors
γM1 partial factor for resistance of members to instability assessed by
member checks
γM,fi partial factor for the relevant material property, in the fire situation
δT total mid-height deflection of the stud at the time of failure
θ appropriate elevated temperature
θmax maximum steel temperature reached at time t
λc non-dimensional slenderness
λθ non-dimensional slenderness at temperature θ
φb, φc capacity reduction factors
∆Mx,Ed, ∆My,Ed additional moments due to shift of centroidal axis

∆T temperature difference across the stud section
∆z measured total horizontal deflection at height z
χ appropriate reduction factor for buckling resistance
χfi reduction factor for flexural buckling in the fire design situation
χLT reduction factor due to lateral torsional buckling
χx χy reduction factors due to flexural buckling
xlviii
STATEMENT OF ORIGINAL AUTHORSHIP
The work contained in this thesis has not been previously submitted for a degree or
diploma at 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.
Shanmuganathan Gunalan
Signed: ______________
Date : ______________
xlix
Acknowledgements
ACKNOWLEDGEMENTS
The author wishes to express sincere gratitude to his supervisor, Professor Mahen
Mahendran for his inspiration, guidance, invaluable expertise and continuous support
in many ways over the past three years. This study would not have been success to
this level without such assistance. The author would also like to thank Dr. Sabrina
Fawzia and Dr. Jung Kwan Seo for their assistance.
Author would like to thank Queensland University of Technology (QUT) and

Australian Research Council (ARC) for providing financial support to this research.
Thanks also to the School of Urban Development and the Faculty of Built
Environment and Engineering at QUT for providing necessary facilities and
technical support.
Many thanks to the structural laboratory staff members and technicians, particularly
Mr Jonathan James and Mr. David for their assistance with the full scale fire tests.
Also many thanks to staffs of high performance computing (HPC) and research
support services for providing necessary facilities and support. Special thanks to Mr.
Mark Barry for his great help with HPC facilities.
The author wishes to thank senior postgraduate students, Dr. Prakash Kolarkar, Dr.
Sivapathasunderam Jeyaragan, Dr. Tharmarajah Anapayan, Dr. Poologanathan
Keerthan, Dr. Yasintha Banduka Heva and Dr. Nirosha Dolamune Kankanamge for
their help during my experimental work and numerical studies. It is also important to
thank fellow postgraduate students, Mr. Balachandren Baleshan, Mr. Anthony
Ariyanayagam and Mr. Sivapalan Sabesan for their support and contribution to this
research, and other post graduate students for their friendship at QUT.
Finally, the author wishes to express his sincere appreciation to his parents,
especially his mother who has been a constant source of encouragement, wife who
exhibited immense patience and willingness to work around my schedule, and
brother and sister who were happy to guide and help in any means. Their support has
been of immense help in the completion of this thesis.
li
Introduction
CHAPTER 1
1. INTRODUCTION
1.1. General
The importance of fire safety design has been realised due to the ever-increasing loss
of properties and lives due to structural damage during fire events every year.
Research into the structural design of buildings exposed to fire has been performed
especially over the last couple of decades, in recognition of the requirements for fire
safety and fire resistance. In the same period, the development of fire engineering
principles has brought significant reduction to the cost of fire protection. However,
the traditional method of using fire protection material is still continuing although it
is approximate and conservative.
The use of cold-formed steel is increasing rapidly around the world due to its various
engineering applications. Cold-formed steel members can be produced in a wide
variety of section profiles and the most commonly used are the C - (channels) and the
Z - sections. They can be assembled in various combinations to provide load-bearing
wall and floor systems for buildings. The commonly used light gauge steel frame
(LSF) load-bearing wall systems are made of cold-formed thin-walled steel lipped
channels. These LSF wall systems are widely accepted in industrial and commercial
building construction (see Figure 1.1).
Under fire conditions, cold-formed thin-walled steel sections heat up quickly

resulting in fast reduction in their strength and stiffness. Therefore cold-formed thin-
walled steel stud sections are commonly used in planer structural wall systems with
plasterboard on both sides as fire protection. This provides protection for steel studs
during building fires, delaying the temperature rise in the wall cavity. The growth of
LSF systems is expected to increase the demand for economical solutions, where
specific performance is required, such as in the area of fire resistance. Innovative fire
protection systems are therefore essential without simply adding on more
plasterboards, which is inefficient.
1-1
Introduction
(a) (b)
Images from (a) http://www.artuji.com (b) http://www.grayhawk-ky.com [Accessed July 22, 2011]
Figure 1.1: Light Gauge Steel Framing (LSF) Systems
Achieving sufficient fire resistance to prevent or delay the spread of fire and to
ensure building integrity so that occupants can safely evacuate and fire fighters
perform their duties is a major issue when using LSF systems as load-bearing walls.
There is a need to develop new LSF wall systems with increased fire resistance
rating to replace the conventional LSF wall systems with plasterboard protection and
cavity insulation.
1.2. Cold-formed Steel Members
In steel construction, there are two main families of structural members. One is the
group of hot-rolled shapes and members built up of plates and girders, currently in
wide use in steel constructions. The other, rapidly growing in importance, is
composed of sections cold-formed from steel sheet, strip, plates, or flat bars in roll-
forming machines or by press brake or bending brake operations. These are known as
cold-formed steel structural members.
The method of manufacturing is important as it differentiates these products from

hot-rolled steel sections such as I-sections, channel and angle sections and hollow
sections. The versatility of the different shapes and sizes of cold-formed steel
sections allows the sections to be used effectively as primary wall studs, floor joists,
roof trusses and partitions (see Figure 1.2). The thickness of steel sheets or strip
generally used in cold-formed steel structural members ranges from 0.4 mm to about
6.4 mm.
1-2
Introduction
Image from http://www.globalspec.com [Accessed July 22, 2011]
Figure 1.2: Cold-formed Steel Sections
Over the last couple of decades, cold-formed and thin-walled steel sections have
been used extensively in residential, industrial and commercial buildings as primary
load-bearing structural components. The reasons for the popularity of cold-formed
steel members include their wide range of applications, high strength to weight ratio,
economy of transportation and handling, ease of fabrication, accurate detailing and
quick as well as simple erection or installation.
The use of cold-formed steel sections in building construction has grown, as a result
of increased design and development work on the structural applications of cold-
formed steel sections, allowing direct replacement of masonry and timber with cold-
formed steel sections. Thin steel products are used extensively in the building
industry, ranging from purlins and lintels, to roof sheeting and floor decking.
1.3. Light Gauge Steel Frame (LSF) Walls
1.3.1. Elements Used

In LSF wall construction, small size steel members are used. In a wall panel, studs
(vertical members) and tracks (horizontal members) are normally used. Studs carry
the vertical load with tracks connecting the studs to make the frame. These members
are manufactured by cold rolling steel strips and hence fire resistance must be based
on protective materials, by far the most common being gypsum plasterboard.
Gypsum plasterboard has fire resistance properties better than most of the other
similar materials because of the moisture in the gypsum crystals. Figures 1.3 (a) and
(b) show an LSF wall system clad with plasterboard on both sides.
1-3
Introduction
Track
Stud
(a) LSF wall (b) LSF clad with Plasterboards

Image from http://demo.thejupitech.com
[Accessed July 22, 2011]
Figure 1.3: LSF Wall System
1.3.2. Plasterboards
When gypsum plasterboard lining is heated during a fire, temperature on the exposed
face will increase steadily until about 100°C, at which time there will be a delay
while the water of crystallization is driven off. As the heating continues, the 100oC
temperature plateau will progress slowly through the plasterboard until the entire
board has been dehydrated. Hence this protects the steel frame from heat.
Temperatures within the board will rise steadily after dehydration is completed,
leading to increased temperatures in the cavity and in the framing members.
1.3.3. Behaviour of LSF Walls in Fire

Usually the LSF walls are exposed to fire attack from one side. As a result,
temperature distributions in a thin-walled steel section can be highly non-uniform
under fire conditions. This non-uniform temperature distribution will induce
complicated structural behaviour of studs. Thermal bowing and non-uniform
distribution of strength and stiffness of steel are some of them. These effects caused
by non-uniform temperature distributions will compound the already complex
structural behaviour of thin-walled steel columns (studs) due to local buckling,
distortional buckling, flexural buckling and their complex interactions with the
varying levels of support provided by plasterboard sheeting during fire.
1-4
Introduction
1.3.4. Applications of LSF

LSF systems are non-organic and they resist the deficiencies of frame made of wood
such as shrinking, warping and pest infestations. They are ideal for easy and quick
assembly into steel wall, floor, and roof panels. These panels are manufactured in
quality-controlled factories and transported to the construction site for fast erection.
The frame members can be supplied in exact lengths, eliminating labour work in on-
site cutting. The pre-punched holes aid to run pipes and electrical wires quickly.
Large quantities are easily handled on the construction site since it weighs less than
the traditional wood studs. The LSF system is non-combustible and it can be
designed to resist fire, earthquakes and storm events. Unlike wood, it does not
require treatment for termites with harmful chemicals and it does not shrink to cause
nail-pops or squeaky floors. Due to these reasons LSF systems are widely accepted
in residential, industrial and commercial building construction.
1.4. Fire Safety

The design process for fire resistance requires verification that the provided fire
resistance exceeds the design fire safety requirements. Verification may be in the
time, temperature or strength domain. Most often verification of the fire-resistance of
light weight frame structures is in the time domain, where proprietary ratings are
compared with the code specified fire resistance, or with the calculated equivalent
time of a complete burnout. The fire resistance of assemblies made with gypsum-
based panel products depends on several important interrelated properties such as the
insulating capacity of the board, the ability of the board to remain in place, resistance
to shrinkage and the ability of the core material to resist ablation from the fire side
during extreme fire exposure. The failure criteria can be in terms of structural
adequacy, integrity or insulation.
1.4.1. Integrity
Integrity is defined in AS 1530.4 (SA, 2005) as the ability to resist flames or smoke
passing through the section. Assessment of integrity must be done in full-scale
testing because small-scale tests cannot assess factors such as shrinkage in large
sheets of gypsum plasterboard or cracking due to structural deformations. Large-
scale testing is also necessary to assess the resistance of the gypsum plasterboard to
falling off from walls during fire.
1-5
Introduction
1.4.2. Insulation
Based on the ISO 834 (ISO, 1999) and AS 1530.4 (SA, 2005) criteria, the assembly
is considered to have failed the test by the insulation criterion when the average
temperature rise on the unexposed surface exceeds 140°C, or the maximum
temperature rise at any point exceeds 180°C.
1.4.3. Structural Adequacy

The structural adequacy criteria shall be deemed to have occurred when either a
collapse occurs or when the deflection exceeds the limit. The strength of LSF
assemblies is mainly in the steel members themselves and not the lining materials.
Lining materials are essential for providing lateral stability to the structural members,
but their contribution to overall strength and stiffness is small.
1.4.4. Fire Resistance of LSF Walls

With the increasing use of LSF walls in load-bearing applications, the demand for
systems with improved fire resistance ratings has increased. The fire resistance of a
load-bearing wall is its ability to withstand exposure to fire. It is usually expressed in
terms of its fire resistance rating, which is the length of time the wall can withstand
exposure to fire in a standard fire resistance test without losing its load-bearing or
fire separating functions.
The strength of steel walls depends on the temperature of the studs and the level of
lateral stability provided to the studs by the lining materials. As a fire progresses,
steel framing will lose strength due to increased temperatures, but long periods of
fire resistance can be achieved if the lining on the fire side remains in place.
Resistance to fire also depends on how much heat is transferred across the cavity and
through the lining on the unexposed side. In addition to this, critical factors affecting
the fire resistance are the thickness of the gypsum plasterboard, the quality of the
board material, the details of the construction and the fixings and the standard of
workmanship.
Calculation of the structural behaviour of steel structures is more difficult than

calculation of thermal behaviour. Computational models for walls must include
second-order effects in order to predict overall buckling, and models for steel walls
1-6
Introduction
must additionally be able to predict local buckling of thin steel sections restrained by
degrading lining materials. At present, the design of thin-walled steel structures in
fire is based on the standard fire resistance test results of manufacturers and there are
very few independent research studies.
1.5. Composite Wall Panels
Past research has produced contradicting results about the benefits of cavity
insulation to the fire rating of LSF wall systems. Because of the very low
conductivity of the cavity insulating material as compared to steel, most of the heat
gets directed along and across the steel studs which act as the heat sink thus raising
their body temperatures much faster than in the case of non-cavity insulated
specimens. This makes the presence of cavity insulation a threat to the survival of
steel frame under fire conditions. In many tests, it was concluded that the cavity
insulation reduces the fire resistance of LSF wall panels and there was a need to
develop new LSF wall systems with increased fire rating. This resulted in the
development of a new composite panel by Kolarkar and Mahendran (2008). The new
LSF wall system was built with the insulation sandwiched between the plasterboards
on either side of the steel wall frame as shown in Figure 1.4 instead of being placed
in the cavity.
Externally insulated wall panels such as the new composite panel can offer a much
higher level of protection to the studs as they are installed on the fire side of the studs
thus minimizing the transfer of heat by radiation and conduction. Hence the quality
of insulation used externally would directly influence the level of fire protection
offered to the LSF wall studs.
Insulation Studs Cavity Studs Plasterboards
Figure 1.4: Composite Wall Panels Developed by

Kolarkar and Mahendran (2008)
1-7
Introduction
1.6. Research Problem

In recent times, LSF wall systems are increasingly used in low-rise and multi-storey
buildings, but without a full understanding of their fire performance. Fire rating of
these LSF wall systems is provided simply by adding more plasterboards to the LSF
walls with cavity insulation. This approach is totally inefficient and we need new
LSF systems with improved fire protection. The relationship between the fire
resistance rating and the various parameters such as the load ratio, number of
plasterboards, types of insulation, steel grade and thickness and screw spacing, is not
well understood. Hence the LSF wall design continues to be based on time
consuming and expensive full scale fire tests. Their construction practice is not
optimized and the walls are being built up, too conservatively or with the risk of fire
attack. This resulted in extensive research by the International community in the area
of fire resistance of steel wall panel systems.
Most of this research was conducted in countries like Canada, United Kingdom,
Europe and New Zealand. However, their research was restricted to the traditional
LSF wall systems used in these countries. Further, this research also did not lead to a
full understanding of the structural and thermal behaviour of even the traditional LSF
wall systems. There were conflicting observations and outcomes, and simple design
models that can accurately predict the fire resistance were not developed. In
Australia, there has not been any research in this area except the recent work at QUT
by Kolarkar (2010). Until this work, there has never been any research addressing the
load-bearing LSF walls made of thin and high strength steels.
Kolarkar and Mahendran (2008) developed a new composite panel system in which
insulation was used externally between plasterboards instead of the conventional
cavity insulation located within the stud space. They conducted full scale tests to
investigate the effectiveness of the LSF walls protected by the new composite panel
system in comparison with cavity insulation. Their tests demonstrated that the use of
this new composite panel system improved the fire rating of LSF wall systems.
However, conducting full scale fire tests is time consuming and expensive and hence
only a limited number of tests has been undertaken while numerical and theoretical
analyses of the new LSF wall systems under fire conditions have not been
undertaken.
1-8
Introduction
1.7. Research Objectives and Scope

This research project proposes full scale fire tests and extensive numerical and
theoretical analyses to investigate the structural and fire performance of both the
traditional and the new composite load-bearing LSF wall systems under standard fire
conditions, improve the understanding of the effect of relevant thermal and structural
parameters, and develop simple design rules within the current cold-formed steel
standard provisions for use by structural and fire engineers.
The specific tasks of this study are to,

1) Investigate the structural and thermal behaviour of new composite LSF wall
panels with higher load ratios using full scale fire tests.
2) Develop accurate numerical (finite element and finite strip) models to predict
the structural failures of thin-walled cold-formed steel studs used in LSF
walls under fire conditions by calibrating and using the available full scale
test results.
3) Undertake detailed numerical analyses of LSF wall studs under non-uniform
temperature distributions to develop a full understanding and knowledge of
the behaviour of LSF wall systems under fire conditions.
4) Compare the structural and fire performance of the new LSF wall system
with that of conventional LSF wall system, to verify the predictions of
Kolarkar (2010) based on full scale tests and numerical simulations.
5) Verify the superior performance of load-bearing LSF walls using composite
panels with different steel grades and thicknesses.
6) Develop idealized time-temperature profiles for different wall configurations
(various plasterboard and insulation arrangements).
7) Investigate the structural and fire behaviour of LSF wall panels in terms of
various structural and thermal parameters including steel grade and thickness,
screw spacing, plasterboard restraint conditions and wall configurations.
8) Investigate the suitability of using AS/NZS 4600 (SA, 2005) NAS* (AISI,
2007) and Eurocode 3 Part 1.3 (ECS, 2006) for the design of cold-formed
LSF wall studs subjected to non-uniform elevated temperature conditions and
develop improved fire design rules for load-bearing LSF walls under fire
conditions.
1-9
Introduction
9) Develop spread sheet based design tools using the developed fire design rules
to predict the fire resistance rating of LSF wall systems.
10) Develop simplified design methods to predict the critical temperatures and
failure times of LSF wall studs as a function of the above mentioned
parameters.
Note: * - Since the design rules in AS/NZS 4600 (SA, 2005) and NAS (AISI,
2007) that are relevant to this study are identical, this thesis makes references
only to AS/NZS 4600 in the following chapters.
1.8. Thesis Contents

The outline of this thesis is as follows:
Chapter 1 presents the introduction to cold-formed LSF systems and fire safety. It
explains the behaviour of LSF walls and studs under non-uniform
temperature distributions and introduces the new composite LSF wall
panel developed at QUT to improve the fire performance.
Chapter 2 presents a detailed literature review of the current body of knowledge in
this field. It describes the past research conducted on mechanical
properties at elevated temperature, experimental study of LSF wall panels
and numerical modelling of thermal and structural behaviour. It also
summarises the fire design rules developed in previous researches on the
ultimate capacities of LSF wall studs under fire conditions.
Chapter 3 presents the details of the experimental study, which was conducted to
investigate the thermal and structural performance of load bearing steel
stud wall assemblies lined with external insulation under different load
ratios.
Chapter 4 presents the details of the numerical (finite element and finite strip)
models of LSF walls developed in this research and their validation under
steady state and transient state conditions.
Chapter 5 presents the details of the analyses to predict the failure times and lateral
deflections of LSF wall studs under non-uniform temperature conditions.
The fire design rules developed in previous researches are discussed and
1-10
Introduction
new fire design rules are proposed based on AS/NZS 4600 (2005) and
Eurocode 3 Part 1.3 (ECS, 2006).
Chapter 6 presents the idealised time-temperature profiles based on 10 full scale fire
tests. This chapter validates the fire design rules developed in Chapter 5
for different steel grades and thicknesses and develops simplified design
methods to predict critical temperature and failure times of LSF wall
studs.
Chapter 7 presents the significant findings of this research and the recommendations
made for further research.
1-11
Literature Review
CHAPTER 2
2. LITERATURE REVIEW
2.1. Introduction
This literature review provides the necessary background to the thermal and
structural behaviour of Light Gauge Steel Frame (LSF) wall assemblies at elevated
temperatures. It also provides a brief review of special characteristics and design
considerations of cold-formed steel structural members. The literature review then
explores the current knowledge and understanding of LSF wall assemblies including
the previous research on LSF wall assemblies at elevated temperature and the
appropriateness of currently available design procedures. The thermal performance
of LSF wall panels is discussed briefly. Furthermore descriptions of the structural
behaviour of LSF walls and design methods which have been used by engineers are
also discussed. Finally, the literature review summarises the current state of
knowledge and contributions in this area of research of LSF walls.
2.2. Cold-formed Steel Members
2.2.1. General
Cold-formed steel structural members can be used very efficiently in many
applications where conventional hot-rolled members prove to be uneconomical.
However, cold-formed sections are usually thinner than hot-rolled sections and have
different buckling modes of failure and deformation, which are not commonly
encountered in normal structural steel design. Local buckling may become a major
concern in the case of cold-formed steel members, with buckling at stresses well
below the yield point. It is observed that although the initial buckling stress is
reached, the member continues to take more load developing compression stresses
well in excess of those at which local buckling first appeared. The behaviour of the
cold-formed steel members will be more complicated with the effect of non-uniform
elevated temperatures. In addition, the cold-forming process often produces
geometric imperfections and residual stresses that are quite different from those of
traditional hot-rolled and welded members. Therefore specific design specifications
2-1
Literature Review
are required for cold-formed and thin-walled structural members, particularly for fire
conditions.
2.2.2. Cold-formed Steel Design Standards

During the 1930s the acceptance and development of cold-formed steel construction
in the industry faced difficulties due to the lack of an appropriate design
specification. Design clauses for cold-formed steel structural members were first
introduced in the American Iron and Steel Institute (AISI) Specification in 1946.
This was followed by extensive research on cold-formed steel structures at Cornell
University by Professor G. Winter. The British Steel Standard was modified in 1961
to include the design of cold-formed steel members based on the work by Professor
A.H. Chilver.
The Australian Standard for the design of cold-formed steel members AS 1538 (SA,
1974) was first published in 1974 based on the 1968 edition of the American
Specification. It was reproduced using the 1980 and 1986 editions of the American
Iron and Steel Institute Specification in 1988. In 1991, the AISI introduced the limit
states method in their 1986 specification called the Load and Resistance Factor
Design Specification (LRFD). The limit states design is based on load factors and
resistance factors applied to the loads and design strengths, respectively, such that
the factored design load does not exceed the factored strength.
In 1990, Standards Australia published the limit states design standard for steel
structures AS 4100 based on the load factor and capacity factor approach as used for
LRFD in the AISI Standards. Standards New Zealand produced the limit states
design standard for steel structures NZS 3404 based mainly on the Australian
Standards AS 4100 in 1992. In 1993, Standards Australia and Standards New
Zealand commenced work on a common limit design standard for cold-formed steel
structures. The new standard AS/NZS 4600 (SA, 2005) was based mainly on the
latest AISI Specification. The British Standard BS 5950 Part 5 (BSI, 1998) also
specifies guidance for the design of cold-formed structural steel work. Eurocode 3
Part 1.3 (ECS, 2006) and Canadian Standards are the other international standards
providing design guidance for cold-formed steel structures.
2-2
Literature Review
2.3. Light Gauge Steel Frame (LSF) Wall Assemblies

The traditional method of LSF wall construction is with Light Timber Framing
(LTF) and sheet material linings. However, there has been an increasing demand in
the commercial arena for prefabricated Light Gauge Steel Frame (LSF) systems. The
replacement of timber with steel becomes more prevalent in regions where timber
resources are scare and also in commercial or community applications where other
advantages such as speed of assembly, dimensional stability and fire resistance are
more important. The LSF wall panels are typically constructed by first connecting
studs and tracks with rivets or screws to form the frame, and then connecting
sheathing boards to the frame with self-drilling screws. Lee (1999), Telue (2001) and
Tian et al. (2007) present some of the studies on the structural (axial compressive)
capacity of these LSF wall panels at ambient temperature. Insulation materials are
added when additional fire rating is required. These panels can be easily assembled
to fabricate load-bearing or non-load-bearing LSF walls. The LSF wall and the
temperature distribution across the steel stud section when it is exposed to fire from
one side is shown in Figure 2.1.
Fire side
Larger
Original
Tracks Centroid
Studs Centroid with

Temperature non-uniform
distribution temperature
Ambient side Smaller

(a) LSF wall (b) Stud cross-section
Figure 2.1: LSF Wall and the Steel Stud Cross-section
2.3.1. Fire Resistance of LSF Members

The three fire resistant requirements, namely Stability, Insulation and Integrity
should be satisfied when walls are used in fire rated construction. LSF walls lose
their strength at elevated temperatures since they are made of thin steel sheets.
2-3
Literature Review
Therefore LSF walls are usually covered with plasterboards and insulation materials
to make a wall assembly that can withstand the required fire exposure.
According to Milke (1999), fire resistance of structural assemblies depends on fire

exposure conditions, material properties at elevated temperatures, thermal response
of the structure and structural response of the heated assembly. Heat transmission
limits were established to prevent the ignition of combustibles in contact with the
unexposed side of the assembly (Schwartz and Lie, 1985). The simple design method
to deal with fire resistance of LSF wall panels is by limiting the steel stud
temperatures to 400°C which ensures that the steel yield strength is not reduced to
less than about 60% of ambient values (Gerlich et al., 1996). At present, although
some analytical studies have been performed to predict the structural performance of
LSF wall panels subjected to the standard fire condition the fire-resistance rating of
load-bearing steel wall panels are still assigned based on the results of standard full
scale furnace tests.
The studs of the load-bearing walls are under axial compressive load, which will be
concentric at ambient conditions. However, when the wall is subjected to fire from
one side, the studs will bend towards the furnace due to non-uniform temperature
distribution (see Figure 2.1). Local buckling of studs at elevated temperatures, and
varying stiffness within the cross-section due to non-uniform temperature
distribution across the stud depth leads to a shift in the neutral axis of the cross-
section. All of these mean that the LSF wall studs will be subjected to bending in
addition to the applied axial compressive load under fire on one side.
2.3.2. Current Fire Resistance Design Methods for LSF Walls

The analysis and design of LSF walls are extremely complex and time consuming.
Hence past researches and standards recommend simplifying this by the design of
individual members. In a load-bearing wall, the critical member is the stud which
carries the axial compressive load. When the wall is subjected to fire on one side, the
studs will be subjected to a combined loading of axial compression and bending as
described in the last section. Hence the studs of LSF walls have to be designed as
beam-columns in fire situations. The following section of this chapter will present
the current design methods of beam columns under ambient and fire conditions.
2-4
Literature Review
2.3.2.1. AS/NZS 4600 (SA, 2005)
Concentrically Loaded Compression Members
The capacity of LSF wall studs subject to axial compression according to AS/NZS
4600 (SA, 2005) can be calculated using Clause 3.4 as
N * ≤ φc N s (2.1)
N * ≤ φc N c (2.2)
where
φc is the capacity reduction factor;
Ns is the nominal section capacity of the member in compression calculated using
N s = Ae f y where Ae is the effective area at yield stress fy;
Nc is the nominal member capacity of the member in compression calculated using

N c = Ae f n where Ae is the effective area at the critical stress fn;
f n shall be determined from,
(
f n = 0.658 λc f y
2
) for λc ≤ 1.5 (2.3a)
 0.877 
f n =  2  f y for λc > 1.5 (2.3b)
 λc 
where
fy
λc is the non-dimensional slenderness and can be calculated as λc = where foc is
f oc
the least of the elastic flexural, torsional and flexural-torsional buckling stresses
calculated as follows,
The elastic buckling stress foc shall be determined from Clause 3.4.2 as
π 2E
f oc = 2
(2.4)
 le 
 
r
where le is the effective length of member and r is the radius of gyration of the full,
unreduced cross-section.
2-5
Literature Review
Combined Axial Compression and Bending at Ambient Temperature
The capacity of the LSF wall stud subject to both axial compression and bending
moment can be calculated using Clause 3.5.1 as
N* C mx M x* C my M *y
+ + ≤1 (2.5)
φc N c φb M bxα nx φb M byα ny
where Nc is the nominal member capacity of the member in compression; Cmx and
Cmy are the coefficients for unequal end moments whose values shall be taken as 1
conservatively;
Mbx and Mby are the nominal member moment capacities about the major and minor
axes, respectively;
φb and φc are the capacity reduction factors whose values shall be taken as 1 in this
case;
αnx and αny are the moment amplification factors and can be calculated using
N* π 2 EI b
α = 1− where Ne is the elastic buckling load N e = where Ib is the second
Ne (leb )2
moment of area of the full, unreduced cross-section about the bending axis and leb is
the effective length in the plane of bending.
2.3.2.2. Eurocode 3 Part 1.3 (ECS, 2006)
Axial Compression
This European standard is for cold-formed steel structures at ambient temperature.

For axial compression members, the design buckling resistance Nb,Rd at ambient
temperature is given by Clause 6.3.1 of Eurocode 3 Part 1.1 (ECS, 2005). The design
buckling resistance Nb,Rd for compression member shall be obtained from Clause
6.3.1.1 of Eurocode 3 Part 1.1 (ECS, 2005) as.
N b, Rd = χAeff f y γ M 1 (2.6)
where
Aeff is the effective area of the cross-section and fy is the yield strength;
γM1 is the partial factor for resistance of members to instability assessed by member
checks, which was assumed to be equal to 1.
2-6
Literature Review
χ is the appropriate reduction factor for buckling resistance. The value of χ for the
appropriate non-dimensional slenderness λ should be determined from the relevant
buckling curve according to
1
χ= but χ ≤ 1
2
φ + φ2 −λ
where φ =
1
2
[ (
1 + α λ − 0 .2 + λ
2
) ]
Aeff f y
λ=
N cr
α is the imperfection factor based on the relevant buckling curve; for lipped C-
section α = 0.34;
Ncr is the elastic critical force for the relevant buckling mode based on the gross
cross-sectional properties and can be calculated as,
π 2 EI
N cr = (2.7)
l2
where E is the elastic modulus; I is the second moment of area of the gross section
and l is the buckling length.
Bending and Axial Compression
For combined axial load and bending moments, the general design equations are,
N Ed k xx (M x , Ed + ∆M x , Ed ) k xy (M y , Ed + ∆M y , Ed )
+ + ≤1 (2.8a)
χ x f y Aeff / γ M 1 χ LT f yWeff , x / γ M 1 f yWeff , y / γ M 1
N Ed k yx (M x , Ed + ∆M x , Ed ) k yy (M y , Ed + ∆M y , Ed )
+ + ≤1 (2.8b)
χ y f y Aeff / γ M 1 χ LT f yWeff , x / γ M 1 f yWeff , y / γ M 1
where NEd is the applied axial load according to Eurocode 3 Part 1.3 (ECS, 2006) and
fy is the basic yield strength;
Aeff is the effective cross-sectional area that is subject only to compression;
checks, which was assumed to be equal to 1;
Mx,Ed and My,Ed are the applied bending moments about the major and minor axes;
∆Mx,Ed and ∆My,Ed are the additional moments due to shift of centroidal axis;
2-7
Literature Review
Weff,x and Weff,y are the effective section modulus for the maximum compressive stress
in an effective cross-section that is subject only to moment about the major and
minor axes, respectively;
χx and χy are the reduction factors due to flexural buckling;
χLT is the reduction factor due to lateral torsional buckling from Clause 6.3.2 of
Eurocode 3 Part 1.1 (ECS, 2005); For members not susceptible to torsional
deformation χLT would be equal to 1;
kxx kxy kyx and kyy are the interaction factors calculated according to Annex A of
2.3.2.3. Eurocode 3 Part 1.2 (ECS, 2005)

This design code is specifically developed for hot-rolled steel structures in a fire
situation. However, it states that the methods given are also applicable to cold-
formed steel members and sheeting within the scope of Eurocode 3 Part 1.3 (ECS,
2006). Hence the relevant design rules are considered in this research. The simplified
method specified in Clause 4.2.1 of Eurocode 3 Part 1.2 (ECS, 2005) requires that
the design effect of actions for the fire design situation is less than or equal to the
corresponding design resistance of the steel member for the fire design situation at
time t. In this code the temperature is denoted by θ.
Compression Members with Class 3 Cross-sections

The design buckling resistance Nb.f1t.Rd of compression members is defined in Clause
4.2.3.2 as
N b, fi ,t , Rd = χ fi Ag k y ,θ f y / γ M , fi (2.9)
with χ fi =
1
2
, ϕθ =
1
2
[ ]2
1 + α λ θ + λ θ , α = 0.65 235 f y
ϕ θ + ϕθ − λ θ 2
λ θ = λ [k y ,θ k E ,θ ]0.5
Aeff f y
and λ =
N cr
where
χfi is the reduction factor for flexural buckling in the fire design situation and should
be taken as the lesser of the values calculated about the major and minor axes;
2-8
Literature Review
ky,θ and kE,θ are the reduction factors for the steel yield strength and the slope of the
linear elastic range, respectively at steel temperature θ;
γM,fi is equal to 1 which is the partial factor for the relevant material property, in the
fire situation; λ θ is the non-dimensional slenderness at temperature θ.
when designing using nominal fire exposure, the design resistance Nb,fi,t,Rd at time t of
a compression member with a non-uniform temperature distribution may be taken as
equal to the design resistance Nb,fi,θ of a compression member with a uniform steel
temperature θ equal to the maximum steel temperature θmax reached at time t.
Members with Class 4 Cross-sections

For the design under fire conditions the design yield strength of steel should be taken
as the 0.2% proof strength. For thin-walled elements, local buckling phenomena
become important. Precise design methods have not been included in this design
code for the fire design of cold-formed steel structures. Hence the temperature of
cold-formed (Class 4) cross-sections was limited to 350ºC at any time. This is a
conservative approach which leads to uneconomical designs. However, more
advanced calculation models and test results indicate that the critical temperature of a
particular cold-formed steel structure or structural member is higher than this
limiting value of 350ºC.
2.3.2.4. BS 5950 Part 8 (BSI, 1990)

BS 5950 Part 8 is the British code of practice for fire resistant design of structural
steel work in buildings. This code is specifically defined for hot-rolled structures at
elevated temperatures, and does not allow its use for cold-formed steel members. The
strength reduction factors for cold-formed steel members may be taken from Table
15 of BS 5950 Part 8 (BSI, 1990). This table is reproduced in this chapter as Table
2.1.
Table 2.1: Strength Reduction Factors for Cold-formed Steels
Temp (°C)
200 250 300 350 400 450 500 550 600
Strain (%)
0.5 0.945 0.890 0.834 0.758 0.680 0.575 0.471 0.370 0.269
1.5 1.000 0.985 0.949 0.883 0.815 0.685 0.556 0.453 0.349
2.0 1.000 1.000 1.000 0.935 0.867 0.730 0.590 0.490 0.390
Note: Intermediate values may be obtained by interpolation
2-9
Literature Review
2.3.2.5. SCI Publication (1993)

Chapter 4 of this publication on “Building Design using Cold-Formed Steel Sections:
Fire Protection” deals with planar protection to walls and floors. For thin-walled
sections the average temperature can be calculated using the standard fire
temperature and the unexposed surface temperature. This average temperature can be
used with Table 3 of this publication, where it presents the strength reduction factors
for cold-formed steel at elevated temperatures. This table is reproduced here as Table
2.2.
Also Table 4 of this publication defines the fire resistance time with respect to
different parameters such as the number of plasterboards, protection thickness, type
of plasterboard and insulation. This table is reproduced as Table 2.3 in this Chapter.
Using Table 2.3, designers can determine fire resistance of typical floors and walls
while Table 2.2 can be used to estimate the strength of cold-formed steel members at
the calculated maximum temperature in fire conditions. However, these design
methods are prescriptive and may be over-simplified resulting in either over
conservative or unsafe design.
It is proposed to maintain the ratio of wall height to wall thickness less than or equal
to 25 for load-bearing steel studs. It is also proposed that the loads that may be
applied to the studs at the fire limit state should be taken as not greater than 0.4 times
the stud axial capacity based on the slenderness of the wall under normal condition.
Table 2.2: Strength Reduction Factors for Cold-formed Steel at Elevated

Temperatures (SCI, 1993)
Temperature (°C) 200 250 300 350 400 450 500 550 600
0.5% strain 0.95 0.89 0.83 0.76 0.68 0.58 0.47 0.37 0.27
1.5% strain 1.00 0.99 0.95 0.88 0.82 0.69 0.56 0.45 0.35
2-10
Literature Review
Table 2.3: Fire Resistance of Typical Floors, Walls and Partitions Comprising
Cold-formed Steel Sections and Planar Board Protection, and Heated from One
Side Only (SCI, 1993)
Fire Resistance (hours)

Number Protection
Form of Fire
of layers thickness Notes
construction Plasterboard resistant
of board (mm)
board†
1 12.5 - 0.5 -
Floors with
2 12.5 0.5 1.0 + 60 mm glass
ceiling
wool mat**
protection 2 15 - 1.5 -
Non-load- 1 12.5 0.5 0.5 -

bearing 1 12.5 0.5 1.0 + 25 mm glass
wool mat*
walls
1 15 0.5 1.0 -
(partitions)
2 12.5 1.0 1.5 -
(number of 2 12.5 1.0 2.0 Boxed section
layers per depth > 60 mm
face) 2 15 1.5 2.0 -
Load- 1 12.5 - 0.5 -
bearing 2 12.5 0.5 1.0 -
walls 2 15 - 1.5 -
† ‘Fireline’ or ‘Firecheck’ board or similar
* Glass wool mat is required for insulation purposes for more than 30 minutes fire resistance
** For floors, the glass wool mat is only necessary for fire resistant suspended ceilings
2-11
Literature Review
2.4. Mechanical and Thermal Properties

Thermo-physical and mechanical properties were widely reported as a function of
temperature for several structural and reinforcing steels (Lie, 1972, 1992; Pettersson
et al., 1976; Pettersson, 1986; Kirby and Preston, 1988 and Schleich, 1993). In
addition to these, the more related expressions to this study are discussed next.
2.4.1. Cold-formed Steels

Anderberg (1986) concluded that when modelling material behaviour of steel, a
steady state behavioural model can predict the transient test under any given fire
process and load. According to Gerlich et al. (1996) the crystalline structure of
carbon steels typically used in construction changes at temperatures above
approximately 650°C. However, according to past test results, the failure of load-
bearing LSF systems is expected to happen before crystalline steel structure changes
become a factor.
2.4.1.1. Yield Strength

Klippstein (1980b) carried out experimental work on the yield strength of cold-
formed steel framing members as a function of temperature. This data was used to
develop the following equation, which was used by Gerlich et al. (1996).
fyT = 1 − 5.3T + 4.0T

2
− 1.9T
3
+ 1.7T
4
(2.10)
4 6 8
fyo 10 10 10 1011
where fyT and fyo are the yield stress at temperature T and room temperature,
respectively.
Dolamune Kankanamge and Mahendran (2011) undertook a study to investigate the

mechanical properties of cold-formed steels at elevated temperatures. Based on the
yield strength results obtained from tensile coupon tests at various temperatures, a set
of equations was developed for low and high strength steels.
For low strength steels (G250, G300),
f y ,T
20 ≤ T ≤ 200 o C = −0.0005T + 1.01 (2.11a)
f y , 20
200 < T ≤ 800 o C

f y ,T
f y , 20
(
= 25 1.16 − T 0.022 ) (2.11b)
2-12
Literature Review
Equations 2.11 (a) and (b) present the proposed equations for reduction factors
( f y ,T f y , 20 ) of low strength steels, where f y ,T and f y , 20 are the 0.2% proof stresses
at elevated and ambient temperatures, respectively, and T is the temperature.
As the first option Equations 2.12 (a) to (c) present the proposed equations for
reduction factors ( f y ,T f y , 20 ) of high strength steels (G500 and G550). The
equations were developed without considering the results of 0.42 mm G550 steel.
Since 0.42 mm G550 steel is unlikely to be used in load-bearing structural members,
this approach is justifiable.
For high strength steels (Option 1),
f y ,T  (T − 20 )4.56 
20 ≤ T < 300 C o
= 1 −  (2.12a)
f y , 20  1x1010 T 
300 ≤ T < 600 C o

f y ,T 
= 0.95 −
(T − 300 ) 
1.45
 (2.12b)
f y , 20  7.76T 
f y ,T
600 ≤ T ≤ 800 o C = −0.0004T + 0.35 (2.12c)
f y , 20
In the second option linear equations for 20oC to 300oC and 600oC to 800oC
temperature ranges and one non-linear curve for 300oC to 600oC were proposed
(Equations 2.13 (a) to (c)).
f y ,T
20 ≤ T < 300 o C = −0.000179T + 1.00358 (2.13a)
f y , 20
300 ≤ T < 600 o C

f y ,T 
= 0.95 −
(T − 300 ) 
1.45
 (2.13b)
f y , 20  7.76T 
f y ,T
600 ≤ T ≤ 800 o C = −0.0004T + 0.35 (2.13c)
f y , 20
2-13
Literature Review
As an alternative to Equations 2.12 and 2.13, three simple linear equations were
developed for the three main regions as given in Equations 2.14 (a) to (c).
f y ,T
20 ≤ T < 300 o C = −0.000179T + 1.00358 (2.14a)
f y , 20
f y ,T
300 ≤ T < 600 o C = −0.0028T + 1.79 (2.14b)
f y , 20
f y ,T
600 ≤ T ≤ 800 o C = −0.0004T + 0.35 (2.14c)
f y , 20
2.4.1.2. Elastic Modulus

Klippstein (1980b) obtained experimental data for the modulus of elasticity for cold-
formed steel studs. Gerlich et al.’s (1996) study fitted a polynomial to this data which
gave an expression as follows
ET = 1 − 3.0T + 3.7T
2
− 6.1T
3
+ 5.4T
4
(2.15)
4 7 9
EO 10 10 10 1012
where ET and Eo are the modulus of elasticity at temperature T and room

temperature, respectively
New empirical equations were developed for elastic modulus with respect to the
temperature by Dolamune Kankanamge and Mahendran (2011). Deterioration of
elastic modulus with increasing temperature directly influences the performance of
the structural member as it reduces the stiffness.
In this study it was found that the influence of steel grade and thickness on the
modulus of elasticity reduction factors is negligible and that there was not any
identifiable trend of reduction of elastic modulus with respect to the steel thickness
or grade. Hence neither steel thickness nor steel grade was included in developing
the predictive equations. Two linear equations were developed for the two identified
temperature regions to predict the elastic modulus reduction factors at elevated
temperatures (Equations 2.16 (a) and (b)).
2-14
Literature Review
For low and high strength steels,
ET
20 ≤ T ≤ 200 o C = −0.000833T + 1.0167 (2.16a)
E 20
ET
200 < T ≤ 800 o C = −0.00135T + 1.1201 (2.16b)
E 20
The equations developed by Dolamune Kankanamge and Mahendran (2011) to
predict the reduction factors at elevated temperatures for yield stress and elastic
modulus were used in the current study. This is because these are the latest equations
which represent the cold-formed steel which were used in the current study.
The reduction factors for the mechanical properties such as yield strength,
proportional limit and Young modulus proposed by Eurocode 3 Part 1.2 (ECS, 2005)
at different temperature levels are shown in Table 2.4 (Wang, 2002).
Table 2.4: Reduction Factors Recommended in Eurocode 3 Part 1.2 (ECS, 2005)
Steel Reduction factor for the

Temperature design yield strength of cold-formed slope of the linear elastic
T (°C) Class 4 sections kp0.2T = fp0.2T/fyb range kET = EaT/Ea
20 1.00 1.00
100 1.00 1.00
200 0.89 0.90
300 0.78 0.80
400 0.65 0.70
500 0.53 0.60
600 0.30 0.31
700 0.13 0.13
800 0.07 0.09
900 0.05 0.0675
1000 0.03 0.045
1100 0.02 0.0225
1200 0.00 0.00
In the study of Zhao et al. (2005), steels with different specifications were grouped
into two types (Types A and B). Only Type B is used for load-bearing purposes,
which includes grade S350 steel with labels Medium C2, Large C3 and S350GD+Z.
For this type of steels made by manufacturers of west Europe according to European
standard, the reduction factors given in Eurocode 3 Part 1.2 (ECS, 2005) are
expected to be replaced by those proposed by Zhao et al. (2005) for the fire design of
lightweight steel members (see Table 2.5).
2-15
Literature Review
Table 2.5: Reduction Factors for Steel Load-bearing Elements at High

Temperatures Proposed by Zhao et al. (2005)
Reduction factor for

Steel the slope of the
Temperature effective 0.2 % proof
proportional linear elastic
T (°C) yield strength strength
limit kpT=fpT/fy range
kyT=fyT/fy k0.2pT=f0.2pT/fy
kET=EaT/Ea
20 1.000 1.000 1.000 1.000
100 1.000 1.000 1.000 1.000
200 1.000 0.807 0.896 0.900
300 1.000 0.613 0.793 0.800
400 0.890 0.374 0.616 0.680
500 0.570 0.263 0.407 0.450
600 0.340 0.130 0.229 0.250
700 0.180 0.059 0.117 0.110
800 0.070 0.032 0.049 0.080
900 0.053 0.024 0.037 0.060
1000 0.035 0.016 0.025 0.040
1100 0.018 0.008 0.013 0.020
1200 0.000 0.000 0.000 0.000
2.4.1.3. Coefficient of Thermal Expansion

Steel will expand considerably when exposed to high temperatures. When the steel
wall is exposed to fire from one side, thermal bowing will be developed due to the
presence of non-uniform temperatures across the steel section. Hence the prediction
of coefficient of thermal expansion is necessary for the analysis of light steel frame
wall panels at elevated temperatures.
According to Eurocode 3 Part 1.2 (ECS, 2005) the coefficient of thermal expansion
α should be determined using the following expressions.
α = 1.2 × 10 −5 + 0.4 × 10 −8 T − 2.416 × 10 −4 / T for 20°C ≤ T ≤ 750°C (2.17a)

α = 1.1 × 10 −2 / T for 750°C ≤ T ≤ 860°C (2.17b)
α = 2 × 10 −5 − 6.2 × 10 −3 / T for 860°C ≤ T ≤ 1200°C (2.17c)
In the study of Zhao et al. (2005) the temperature dependent value of the coefficient
of thermal expansion was taken as that given in Equations 2.17 (a) to (c) and was
applied to the whole structure. Lawson and Newman (1990) provided an
approximate constant αT value of 14x10-6 °C-1 for use in simple models in the
2-16
Literature Review
temperature range of 200°C to 600°C (12x10-6 °C-1 at room temperature). Gerlich

(1995) in the study of fire resistance of light steel frame walls used the relationship
proposed by Lie (1992) on the coefficient of thermal expansion of steel as
αT = (0.004T + 12) x 10-6 for T < 1000°C (2.18)
These values are approximately the same as those given in Eurocode 3 Part 1.2
(ECS, 2005).
2.4.1.4. Thermal Conductivity: ks (W/m°C)

The temperature rise of a steel member as a result of heat flow is a function of the
thermal conductivity and specific heat of the material. The following equation was
used by Gerlich et al. (1996) to predict thermal conductivity as a function of
temperature.
ks = - 0.022T + 48 for 0 < T < 900°C (2.19)
According to Alfawakhiri (2001), steel framing plays a minor role in the heat transfer
mechanism, hence the accurate determination of thermo-physical properties of steel,
such as specific heat Cs and thermal conductivity ks is of little importance for the
thermal modelling of LSF walls exposed to fire. Hence an approximate constant
value of 37.5 W/(m°C) was suggested for ks.
Alternative expressions for thermal conductivity of steel ks are presented in Eurocode

3 Part 1.2 (ECS, 2005) where the variation of ks (W/m°C) with the grade of steel
being ignored.
ks = 54 – 0.0333 T for T < 800 °C (2.20a)

ks = 27.3 for T ≥ 800 °C (2.20b)
2.4.1.5. Specific Heat: Cs (J/kg °C)

Lawson and Newman (1990) proposed the variation of Cs (J/kg°C) in relation to
temperature as follows:
Cs = 38.0 x 10-8 T2 + 2.0 x 10-4 T + 0.47 for 20 °C ≤ T ≤ 725 °C (2.21)
2-17
Literature Review
A constant value of 600 J/kg°C was suggested by Gerlich et al. (1996) for
temperatures below 600°C whereas an approximate constant value of 520 J/kg°C
was suggested by Alfawakhiri (2001). However, Eurocode 3 Part 1.2 (ECS, 2005)
suggests the following expressions for the specific heat (J/kgK).
C s = 425 + 7.73 × 10 −1 T − 1.69 × 10 −3 T 2 + 2.22 × 10 −6 T 3 for20°C ≤ T ≤ 600°C (2.22a)
C s = 666 + 13002 (738 − T ) for 600°C ≤ T ≤ 735°C (2.22b)
Cs = 545 + 17820 (T − 731) for 735°C ≤ T ≤ 900°C (2.22c)
C s = 650 for 900°C ≤ T ≤ 1200°C(2.22d)
2.4.2. Plasterboards
Gypsum board is widely used for interior lining of domestic housing and commercial
office buildings, and is the most common lining material used to provide light frame
structures with fire resistance. Most gypsum boards are made with a thickness
between 10 and 20 mm. Typical Gypsum board has a density between 550 and 850
kg/m3. There are three broad types of gypsum board, usually known as Regular
board, Type X board, and Special purpose boards (Buchanan, 2001).
Sultan (1996) reported that fall-off of plasterboard occurs when the unexposed face
of the board reaches about 600°C (Buchanan and Gerlich, 1997). The temperature at
which the gypsum boards lose their restraining capacity depends on the type of board
used. However, according to Ranby (1999) a common temperature of 550°C was
proposed. In the numerical study of Kaitila (2002), the boundary conditions
providing lateral restraints at both flanges were assumed to be valid until 600°C.
Thermal properties of gypsum plasterboard are required if finite element analyses of
thermal performance are to be undertaken.
2.4.2.1. Thermal Conductivity

Thermal conductivity depends on the density of the gypsum board. Its value above
about 400°C will be affected by the presence of shrinkage cracks in the gypsum
board. Thomas et al. (1994) summarized the data for the thermal conductivity and
enthalpy of glass-fibre reinforced gypsum plasterboard as a function of temperature.
Thomas et al.’s (1994) values for the thermal conductivity and enthalpy of gypsum
plasterboard are presented in Figure 2.2 (a). The same material properties were used
2-18
Literature Review
by Gerlich et al. (1996) in their study. The expressions proposed by Sultan (1996) for
thermal conductivity based on tests conducted on Type X gypsum board specimens
are plotted by Alfawakhiri (2001) (see Figure 2.2 (b)).
(a) Gypsum Board (Thomas et al., 1994)
(b) Type X Gypsum Board Core (Sultan, 1996)
Figure 2.2: Thermal Conductivity of Plasterboard
2.4.2.2. Specific Heat

Thomas et al.’s (1994) Enthalpy values of gypsum plasterboard are presented in
Figure 2.3 (a). Enthalpy values represent the summation of the product of specific
heat and temperature, expressed as per unit of volume. This was used by Gerlich et
al. (1996) in their study. The expressions for specific heat proposed by Sultan (1996)
were plotted by Alfawakhiri (2001) as shown in Figure 2.3 (b)
The material properties used by Feng et al. (2003a) in their numerical analysis of
thermal performance of cold-formed thin-walled steel panel systems in fire are
presented in Table 2.6.
2-19
Literature Review
(a) Specific Volumetric Enthalpy (Thomas et al., 1994)
(b) Specific Heat (Sultan, 1996)

Figure 2.3: Thermal Properties of Plasterboards
Table 2.6: Material Properties Used by Feng et al. (2003a)

Density Thermal conductivity
Material Specific heat (J/kg°C)
(kg/m3) (W/m°C)
925.04 at 10°C
0.2 at 10°C
941.5 at 95°C
Gypsum 0.218 at 150°C
727.1 24572.32 at 125°C
board 0.103 at 155°C
953.14 at 155°C
0.3195 at 1200°C
1097.5 at 900°C
Ca=425+7.73x10-1T
-1.69x10-3T2+2.22x10-6T3
λa=54-3.33x10-2xT (20°C ≤ T ≤ 600°C)
(20°C ≤ T ≤ 800°C) Ca=666+13002/(738-T)
Steel 7850
λa=27.3 (600°C ≤ T ≤ 735°C)
(800°C ≤ T ≤ 1200°C) Ca=545+17820/(T-731)
(735°C ≤ T ≤ 900°C)
Ca=650 (900°C ≤ T ≤ 1200°C)
Mineral
25 0.036 840
wool
2-20
Literature Review
2.4.2.3. Moisture Content

Pure gypsum consists of calcium sulphate with free water at equilibrium moisture
content (approximately 3%) and chemically combined water of crystallisation
(approximately 20%). Its chemical formula is CaS04.2H20 (calcium sulphate
dihydrate). When exposed to fire the free water and chemically combined water is
gradually driven off at temperatures above approximately 100 oC. This causes a
temperature ‘plateau’ on the unexposed face of the lining. As exposure to moisture
may affect the performance of plasterboard linings, the plasterboard should be
installed in well ventilated areas protected from moisture penetration as
recommended by Boral Limited. Hence in the design of plasterboard lined walls it is
assumed that the fire performance is not affected by the moisture content as it is
unlikely to increase during service.
2.5. Previous Experimental Studies

In this section, previous experimental studies of cold-formed steel assemblies such as
LSF walls under fire conditions are presented and discussed. Gerlich et al. (1996)
tested three full length both lined and unlined specimens under combined axial
loading and bending. The details of the test specimens are given in Table 2.7. Cold-
formed steels with Grades of 300 and 450 and with thicknesses of 1.15 mm and 1
mm were used in this study. Spacing of the studs was 600 mm and frames had a
central row of nogs. Cuts were introduced in the end bays to minimize load transfer
to the cooler external studs and specimen holder.
Table 2.7: Details of Full Scale Fire Tests of Gerlich et al. (1996)
Test Number 1 2 3
Wall height (mm) 2850 3600 3600
Steel grade (MPa) 300 450 450
Framing type C-section Lipped C-section Lipped C-section
Framing (mm) 76 x 32 x 1.15 102 x 51 x 1.0 102 x 51 x 1.0
Load (kN/stud) 6 16 12
Lining thickness exposed 16.0 mm 12.5 mm 12.5 mm
Lining thickness
16.0 mm 12.5 mm 9.5 mm
unexposed
Fire curve ISO 834 ISO 834 severe
Structural failure (min.) 72 44 32
Integrity failure (min.) 78 45 32
Insulation failure (min.) Not reached Not reached Not reached
2-21
Literature Review
The steel frames were lined on both faces with a single layer of glass-fibre reinforced
gypsum plasterboard fixed vertically to all studs with self-drilling screws spaced at
300 mm centres. Insulation material was not used in this study and the vertical sheet
joints were formed over studs and the horizontal joints were avoided. The test set-up
is shown in Figure 2.4. Specimens were mounted horizontally and the axial load was
applied by a manually operated hydraulic jack at the top channel level. Testing
described by Klippstein (1980a, b) in accordance with the requirements of ASTM
E119 (1995) did not allow vertical expansion to occur and as a result the applied
loads during the fire tests increased to almost twice the initial load (Gerlich et al.,
1996). Therefore Klippstein’s (1980a, b) test results could not be used for
comparison purposes with others’ results. However, in Gerlich et al.’s (1996) study
the bottom platen of the specimen holder was free to move up and down, and the
vertical thermal expansion of the steel framing members was allowed to occur freely.
Figure 2.4: Test Set-up for Combined Axial Loading and Bending Used by
Gerlich et al. (1996)
All three test walls of Gerlich et al. (1996) failed by buckling of the ambient side
stud flange of the specimen near mid-height. These flanges were under compressive
stress since the wall was moving towards the furnace due to thermal bowing. Tests 1
and 2 failed by flexural buckling about the major axis. This was initiated by local
buckling of the compression flanges. However, Test 3 showed flexural torsional
buckling due to the loss of lateral restraint to the compression flange by ambient side
plasterboards. In the measured horizontal deflection, double curvature near failure
was not observed and it was concluded that the thermal deflections override
rotational restraint provided by stud-to-channel connections and load relocation.
2-22
Literature Review
In the study of Gerlich et al. (1996), too many variables were incorporated into a few
test specimens. More than one variable between the tests made the comparison
difficult and uncertain. The central row of nogs made the test specimens unique and
the results cannot be used for common LSF frames without nogs. Also, only one
layer of lining was used in this study.
Kodur and Sultan (2001) tested 14 steel wall panels with single or double layers of
gypsum boards on both sides with a thickness of 12.7 mm except in two tests where a
thickness of 15.9 mm was used. The nominal metal thicknesses of 0.84 and 0.912
mm were used. Membrane and Cross bracing were used to provide shear resistance
and stud spacing was maintained as 406 mm except in one test where it was 610 mm.
Glass fibre, rock fibre and cellulose fibre were used as insulation materials. The
applied load varied from 52.4 kN to 156.7 kN and a maximum fire resistance of 102
minutes was obtained. Local buckling was the dominant structural failure mode.
However, overall buckling was also observed in two tests. It was found that wall
assemblies without interior (cavity) insulation provided higher fire resistance
compared to cavity insulated assemblies. The cellulose fibre was found to be a better
insulation than glass fibre. The wall assemblies with wider stud spacing had higher
fire resistance than the narrow spaced walls due to high thermal conductivity of steel.
In the study of Alfawakhiri (2001), three standard fire resistance tests of LSF walls
were carried out. The metal thickness of 0.912 mm with minimum yield strength of
228 MPa was used. Cross-sectional details of the wall specimen and the typical steel
frame layout for wall specimens are shown in Figures 2.5 and 2.6. Structural failure
resulted in overall buckling towards the furnace for non-insulated walls by
compressive failure of the cold flange near mid-height. According to Alfawakhiri
(2001), this is because the heat penetrated well into the ambient side of the steel to
reduce its strength below the applied stress. For insulated walls, the failure was away
from the furnace by compressive failure of the hot flange. This is because the cavity
insulation delayed the transfer of heat to the ambient side and the acceleration of fire
side steel temperature reduced the strength below the applied stress. Similar to some
previous studies (Kodur and Sultan, 2001) it also concluded that the cavity insulation
reduces the fire resistance of load-bearing LSF walls.
2-23
Literature Review
Figure 2.5: Typical Steel Frame Layout of Wall Specimens (Alfawakhiri, 2001)
Figure 2.6: Cross-sectional Details of the Wall Specimen (Alfawakhiri, 2001)
In Alfawakhiri’s (2001) study each stud had four perforations and the studs were
found to be failing at these particular points. Thin sheets were used with low steel
yield strength of 228 MPa. Also cross-bracing was used on the ambient side to
strengthen the steel stud panel and it was assumed that the flexural-torsional buckling
and weak axis buckling failure modes were prevented by these lateral restraints such
as bridging and blocking. Therefore further study is required on the LSF wall panels
that normally do not use such bracing and perforation.
In the study of Feng and Wang (2005a) eight tests of loaded full-scale LSF wall
panels were carried out (two tests at ambient temperature and six tests exposed to the
standard fire condition on one side). Overall panel size was 2200x2000x125 mm.
One layer of 12.5 mm gypsum board was used with interior insulation. Details of the
2-24
Literature Review
fire test specimens, failure modes and failure times are listed in Table 2.8. Grade
S350 steels with thicknesses of 1.2 mm and 2 mm were used with each having two
holes, one near the top and one near the bottom. Load ratios of 0.2, 0.4, and 0.7 were
considered in this study. The temperature development in the steel channels was not
affected by the applied loads in the panel. Near failure, the panel lateral deflections
increased rapidly even though the steel temperature differences were reducing. This
was explained by the deterioration of mechanical properties of steel at increasing
steel temperatures. The failure mode of all the tests was found to be global buckling
about the major axis except for one test whose failure was initiated by local buckling
at the top service holes. At high temperatures, plasterboards on the fire side lost their
ability to restrain the studs, whereas the unexposed gypsum plasterboards were able
to prevent the flexural buckling of steel channels about the minor axis.
Table 2.8: Fire Test Specimens, Failure Modes and Failure Times of Panels
(Feng and Wang, 2005a)
Load Load per Channel failure Failure
Panel ID Channel size
ratio jack (kN) mode time (min)
Panel-1 100x56x15x2 0.2 20.6 G(F-LT) 44
Panel-2 100x56x15x2 0.4 41 G(F-T) 39
Panel-3 100x56x15x2 0.7 72 G(F-T) 31
Panel-4 100x54x15x1.2 0.2 12 G(F-T) 36
Panel-5 100x54x15x1.2 0.4 23 L 25
Panel-6 100x54x15x1.2 0.7 41 G(F-T) 22
Note: G - Global buckling; F-T – Flexural-torsional buckling about the major axis; F- LT -
Flexural buckling about the major axis and some torsional buckling; L - Local buckling
A testing program was carried out by Zhao et al. (2005) in which 30 different
specimens were tested. All these specimens had a size of 1200 mm in width and
2800 mm in height and consisted of two studs spaced at 600 mm with plasterboards
connected on one or both sides of steel studs. The yield strength of the steel used was
402 MPa. Also steel grades of S280 and S350 were used in the parametric study.
This study concentrated on the mechanical behaviour of steel sections, considering
different parameters such as cross-section size, load condition, eccentricity of
loading, internal insulation, heating condition, type of plasterboard and connection
condition between steel studs and plasterboards. Full scale tests were carried out and
the temperatures and deformations were recorded.
2-25
Literature Review
The walls were lined with plasterboards on both sides except two medium sections
which were lined on single side. Two fire tests were performed on non-perforated
small steel stud section (100x50x0.6). Fire board type plasterboards were used with
and without cavity insulation. A load ratio of 0.4 was maintained concentrically. Fire
test performed on perforated TC-section (150x1.2) was concentrically loaded with a
load ratio of 0.4. The large steel stud section (250x80x2.5) and AWS steel stud
section (150x1.2) were concentrically loaded with load ratios of 0.4 and 0.6. The
medium sections (150x57x1.2) were loaded concentrically and eccentrically with
load ratios from 0.2 to 0.6.
In this study two supporting systems were used to place the specimens to create an
idealized restrained condition. Each of them was composed of two UAP steel profiles
fixed together to a steel plate in which steel studs are inserted. In CTICM tests, the
upper steel end plate was simply supported by a roll and the lower steel end plate
was rigidly supported on two hydraulic jacks whereas in VTT tests, the upper steel
end plate was rigidly supported by a steel frame and the lower steel end plate was
simply supported through a roll on a mobile steel box beam pushed up with the help
of hydraulic jacks. Studs were considered as hinged at one end and restrained against
rotation at the other end since the supporting end plates allow specimen to rotate out
of plane but prevent the in- plane rotation of the specimen.
According to this study, 1 mm thick steel sheet added between exposed plasterboards
and the steel studs of the specimen seem to increase the maximum collapse
temperature of steel studs. Also the type of plasterboards (fire or standard) influenced
the collapse time but the same maximum temperature seems to be obtained at the
point of collapse. This is an important observation as it implies that steel stud failure
is essentially governed by the maximum temperature that studs reached during a fire.
Steel studs supported by plasterboards on both sides failed by global flexural
buckling with local buckling on the unexposed flange. For small and large sections
the exposed side flange temperature at failure is higher in the specimen with
insulation than the specimen without insulation. For medium sections the failure
temperature is higher in specimens without insulation. This was explained by
different failure modes of steel studs between non-insulated specimens. Medium
2-26
Literature Review
sections failed by deforming towards the cold side while small and large sections
failed towards the hot side.
In some of the concentric loaded tests, the failure mode was flexural buckling about
the major axis initiated by local buckling (compressive failure) of the cold flange
near mid-height due to the deflection of the specimen towards the furnace. This is
similar to the overall buckling mode obtained for the specimens with eccentric
loading on the exposed side. In other tests, a reversal was observed in the lateral
movement near the end of the test, indicating failure of the studs away from the
furnace. Hence the failure resulted in overall buckling towards the unexposed side
due to the compressive failure of the hot flange. These observations clearly indicate
the complex stud failure modes due to combined thermal and structural effects and
are similar to those observed by other researchers (Gerlich et al., 1996, Alfawakhiri,
2001 and Feng and Wang, 2005a). It is important to investigate why studs collapse in
such a manner, ie. stud failure towards and away from the furnace.
Recently Kolarkar (2010) undertook research into the fire resistance of LSF wall
panels based on nine full scale tests of load-bearing walls and nine small scale tests
of non-load-bearing walls. The test frames were made of 1.15 mm G500 cold-formed
steels whereas the plasterboard used had a nominal thickness of 16 mm. Glass fibre,
rock wool and cellulosic fibre were used as the insulation materials. The traditional
system of putting the insulation inside the cavity was found to be inefficient and an
innovative composite panel was introduced to increase the fire resistance of the wall
panels. The idea was to use the insulation layer outside the steel fame as shown in
Figure 2.7. Test results showed that the new composite panel improved the fire rating
of LSF wall panels.
Insulation Cavity Plasterboards
Figure 2.7: Composite Wall Panels Developed by

Kolarkar and Mahendran (2008)
2-27
Literature Review
Table 2.9 shows the details and the failure times of test specimens tested by Kolarkar
(2010). The first specimen was tested in ambient condition whereas all other
specimens were tested under standard fire conditions.
However, Kolarkar’s study was limited to experimental work and further numerical
modelling of the new walls is required to investigate the possibility of improving the
new composite system further. Hence there is a need to investigate the load-bearing
LSF walls made of the new composite panel to fully understand their structural and
thermal behaviour and to improve their fire resistance rating.
Table 2.9: Details of LSF Wall Specimens Tested by Kolarkar (2010)
Test Configuration Condition Insulation Failure Time

(min.)
1(P) Ambient None -
2(P) Fire None 53
3(P) Fire None 111
Glass Fibre
4(P) Fire 101
(Cavity Insulation)
Rock Fibre
5(P) Fire 107
(Cavity Insulation)
Cellulose Fibre
6(P) Fire 110
(Cavity Insulation)
181
Glass Fibre
7(P) Fire (Unexpected
(External Insulation)
furnace failure)
Rock Fibre
8(P) Fire 136
Cellulose Fibre
9(P) Fire 124
2-28
Literature Review
2.6. Thermal Modelling

This section presents the details of previous research on thermal modelling of LSF
wall systems. Gerlich et al. (1996) carried out heat transfer modelling using a finite
element analysis program TASEF to predict the steel frame temperatures. At very
high temperatures, some opening of the exposed sheet joints due to deflection of the
framing members and ablation (erosion due to heating) of the exposed linings
allowed hot gases into the cavity. However, these effects of accelerated rise in
measured temperatures towards the end of the tests were not modelled by TASEF.
As a result, differences between predictions and measurements were observed at high
temperatures. Also TASEF does not model mass transfer (moisture movement).
Hence the predicted horizontal deflections using TASEF temperatures exceeded
those calculated from measured temperatures.
A computer program TRACE was developed and used by Alfawakhiri (2001) to

conduct numerical simulations of temperature histories. The thermal properties
gained from literature review were calibrated to produce a good match of numerical
and test results. Hence it was believed that these apparent thermal properties, to some
degree, implicitly account for physical phenomena other than heat transfer, such as
mass transfer, phase change, etc. The presence of the steel frame was neglected in the
heat transfer simulations. The spalling of gypsum boards was modelled by removing
it from the simulation at a user-specified time. Mathematical and numerical analyses
of dehydration of gypsum plasterboards exposed to fire was carried out by
Belmiloudi and Meur (2005), and it was found that the radiative heat transfer
between the unexposed surface and the surrounding cannot be neglected.
Feng et al. (2003a) used the experimental study results of fire tests to validate the
thermal analysis capabilities of ABAQUS. In some of these systems, one or more
layers of gypsum boards on the fire exposed side were removed in numerical studies
to consider possible fall-off of gypsum boards. Also the results of a parametric study
using ABAQUS to examine the thermal performance of steel stud systems with
different numbers of gypsum boards on the exposed and unexposed sides were
presented. This study also assumed a uniform temperature distribution along the stud
height. Feng et al. (2003a) concluded that ABAQUS can be used to simulate the
2-29
Literature Review
temperature profile in light steel frame wall panels under standard fire conditions,
including cavity radiation, by adopting the appropriate thermal boundary conditions
and thermal properties, provided there is no integrity failure of the gypsum boards. It
was also found that the temperature profile of steel wall panel was not affected much
by the shape of the thin-walled steel cross-section. The effect of lips on temperature
distribution can be ignored, provided their width is small. It was found that the
thermal performance of wall panels was not significantly affected by the types of
interior insulation. However, it was noticed that, not having interior insulation gave
poor fire performance, which is a contradicting result compared to Kodur and Sultan
(2001) and Alfawakhiri (2001).
In the study of Zhao et al. (2005) different computer programs such as ABAQUS,
ANSYS, FLUENT and SAFIR were used to investigate the validity of heat transfer
analysis. The results obtained from these different computer models showed a good
agreement between them and it was considered that all these computer models are
available for heat transfer analysis if one of them is validated against tests. It was
assumed that conduction is the main heat transfer mechanism in the steel studs and
plasterboards. Convection and radiation act essentially for heat transfer from fire to
plasterboards. As simplification, radiation effects within the void between the
plasterboards were taken into account and convection effects within the plasterboards
were neglected. In numerical models, non-linearities due to temperature dependency
of material properties and boundary conditions were taken into account. The height
and the cross-section size of the stud were considered as parameters affecting the
thermal behaviour. However, the mass transfer in materials such as moisture
movement was not simulated. A new set of thermal properties (specific heat and
conductivity) of the plasterboards used by them were obtained, after some numerical
investigations, which led to a good estimation of stud heating compared to test
results. However, these proposed thermal properties were not presented in their
report.
This brief review on thermal modelling shows that it is possible to obtain reasonably
accurate results from thermal modelling of LSF wall systems using the many
different numerical tools available to fire researchers despite the complex LSF
system and behaviour and associated simplifications in modelling.
2-30
Literature Review
2.7. Structural Modelling

In this section the details of past research on structural modelling of LSF wall
systems and the analytical and numerical tools used by them. The analytical methods
to evaluate the structural response of LSF wall panels range from graphical solutions,
empirical correlations and iteration methods to finite element computer models (Lie,
1992; Fleischmann, 1995; White, 1995 and Zhao, 2000).
In the study of Alfawakhiri (2001), the stud was assumed to be hinged at the ends in
the structural model. It was assumed that the flexural-torsional and weak axis
buckling failure modes were prevented by bridging and blocking. Hence the
structural failure resulted in overall buckling towards the furnace for non-insulated
walls and the dominant failure mode was the compressive failure of the cold flange
near mid-height (see Figure 2.8 (a)). The structural failure of the insulated walls was
the overall buckling away from the furnace and the dominant failure mode was the
compressive failure of the hot flange near the support (see Figure 2.8 (b)).
(a) Non-insulated walls (b) Insulated walls
Figure 2.8: Structural Failure Modes of Alfawakhiri (2001)
This study considered the thermal bowing deflections and their magnifications. An
expression was developed for thermal bowing curvature φ by assuming a linear
gradient of thermal elongation strain across the stud section.
ϕ = αT ∆T d (2.23)
2-31
Literature Review
where αT is the thermal expansion coefficient for steel = (12+0.004TA)x10-6 (Lie

1992); d is the stud section depth; TA is the average stud temperature and ∆T is the
temperature difference across stud section.
The modulus of elasticity deteriorates at the hot flange due to the development of
temperature gradient across the stud. Hence the centroid of the steel section shifts
towards the cold flange. Also the rotation of stud ends associated with stress-free
thermal bowing causes the shift of the load towards the hot flange as shown in
Figures 2.9 (a) and (b). In the study of Alfawakhiri (2001) these effects were
incorporated by a model eccentricity ‘e’, which was expressed as
e = (1-KR)φβ-2 (2.24)
where KR is a reduction coefficient equals to 0.6;

β2 = P/EI*
where E is steel modulus of elasticity at room temperature (203000 MPa);
I* is the elasticity-modulus-weighted moment of inertia of the unreduced stud section
about the neutral axis parallel to flanges. This accounts for the variation of the
modulus of elasticity due to the temperature variation from the hot flange to the cold
flange. I* can be quantified numerically by dividing the stud section into a
sufficiently large number (q) of two-dimensional elements.
[ ]
q
I * = ∑ ni I i + Ai ( xi − c )
2
(2.25)
i =1
where ni is the reduction factor for E at temperature Ti; Ii is the moment of inertia of
the element i about its own neutral axis parallel to flanges and Ai is the area of
element i;
xi is the distance of element i from the extreme fibre of the cold flange;
Ti is the temperature of element i = TCF + ( ∆T xi / d) where TCF is the cold flange
temperature and ∆T is the temperature difference across the steel stud;
c is the distance from the centroidal axis of the ‘modulus-weighted’ section to the
q q
extreme fibre of the cold flange, calculated from c = ∑ ni Ai xi ∑n A i i
i =1 i =1
2-32
Literature Review
The shape of the stress-free initial imperfection y1(z), caused by thermal bowing was
proved to be equals to 0.5 φ z (H-z). The secondary lateral deflection, y2(z) was
developed by the vertical load P acting with an eccentricity e. Hence the total
deflection y(z) (see Figure 2.9 (c)) was found to be as follows
y(z) = y1(z) + y2(z) = (φβ-2–e) [tan(0.5βH) sin(βz) + cos(βz) -1] (2.26)
(a) Uniform (b) Non-uniform

heating heating (c) Thermal bowing and
secondary deflection
Figure 2.9: Stud End Conditions and Deflections (Alfawakhiri, 2001)
Alfawakhiri (2001) developed a computer programme STUD incorporating the

above theoretical equations. It was used to model the structural behaviour of load-
bearing LSF walls subject to fire exposure from one side. Hence the mid-height
lateral deflections were simulated for different tests. These simulated values were
used with S136 (CSA, 1994) to predict the structural failure. The STUD program
was enriched with the code method to predict the failure time, by conducting failure
checks at every time step. However, the temperature variation along the stud height
was neglected by Alfawakhiri (2001).
The following criterion was used for insulated walls to check the critical section at
stud perforations near stud ends for the compressive failure of the hot flange.
 P P[e − y (0.2 H )] 
f H = n H  * + *
 ≥ FyH (2.27a)
A
 e S eH 
2-33
Literature Review
For non-insulated walls, the section at mid-height was checked for the compressive
failure of the cold flange.
 P Py (0.4 H ) 
f C = nC  * + *
 ≥ FyC (2.27b)
A
 e S eC 
where fH and fC are the compressive stresses at the extreme fibre of the hot flange and
cold flange, respectively; nH and nC are the elastic modulus reduction factor for
temperature TH and TC, respectively; FyH and FyC are the yield strengths of steel at
temperatures TH and TC, respectively; A*e is the elasticity-modulus-weighted
effective stud section area in compression; S*eH and S*eC are the elasticity-modulus-
weighted effective stud section modulus in bending that causes compression of hot
flange and cold flange, respectively.
In the study of Alfawakhiri (2001) the width of compression elements in effective

stud cross-sections was reduced in accordance with Clause 5.6.2 of S136 (CSA
1994) to account for local buckling effects. The effective cross-section dimensions
were based on steel properties at room temperature and compressive stress f = Fy
since they were assumed to be insensitive to temperature. Using these effective
cross-sections, in a similar way to the evaluation of I*, the temperature dependent
‘modulus-weighted’ properties A*e , S*eH and S*eC were calculated.
q
Ae* = ∑ ni Ai (2.28)
i =1
*
=∑
q
[
ni I i + Ai ( xi − c )
2
]
S eH
i =1
(d − c ) (2.29)
*
S eC =∑
q
[
ni I i + Ai ( xi − c )
2
] (2.30)
i =1
c
Alfawakhiri’s (2001) STUD program conservatively assumes the calculated

deflections to remain constant whenever ∆T decreases with time near failure. In
STUD simulations, the values of ∆ and y(z) at any given time step are not allowed to
be less than in the previous step. This measure was first suggested by Gerlich (1995)
2-34
Literature Review
in order to account for the creep and stress relaxation phenomena in steel studs at
temperatures higher than 400°C. However, the limitation of this measure is that, it
cannot simulate the reversal in lateral deflection of studs.
Lee et al. (2001) presented results of steady-state fire tests and finite element
modelling of the buckling behaviour of thin-walled compression members with pin
end supports at elevated temperatures. The results of the tests and the finite element
analyses were used to determine the plate buckling coefficient at elevated
temperatures. It was also found that finite element analysis could be successfully
used to model the behaviour of thin-walled compression members in fire.
In the study of Feng et al. (2001), the length of all the tested lipped channel sections
was only 400 mm and hence their research was limited to local buckling. The results
of a numerical investigation of the behaviour and failure load of cold-formed thin-
walled steel channel columns in fire are reported in Feng et al. (2003b). They
investigated the axial strength of cold-formed thin-walled channel sections under
non-uniform high temperatures in fire. The non-uniform temperature distributions
are based on the results of a thermal analysis of thin-walled stud panels carried out
by Feng et al. (2003a).
Feng et al. (2003b) used the finite-element package ABAQUS to generate results of
failure loads of columns of lipped channel with different slenderness and non-
uniform temperature profiles, including two simplified temperature profiles. The
finite element model that was used to simulate the behaviour of short columns tested
under uniform elevated temperatures was extended to model the long columns with
non-uniform temperature conditions. However, the accuracy of ABAQUS
simulations was not fully validated due to the unavailability of test results. Hence the
accuracy of the numerical model is questionable. This study used only one lipped
channel cross-section (100x54x15x1.2). S4R shell elements were used in ABAQUS.
The rigid plate attached to each end of the column can deform in the axial direction,
but twisting is prevented. Two horizontal restraints were applied at the centroid of
the channel section at each end. The boundary conditions used in the analysis are
shown in Figure 2.10. The lateral restraint is provided on both sides (Hot flange and
Cold flange).
2-35
Literature Review
The stress–strain relationships of cold-formed steel at high temperatures used in

ABAQUS analyses were based on the model in Eurocode 3 Part 1.2 (ECS, 2005), but
using the measured yield stress and elastic modulus of the same lipped channel from
Feng et al. (2001). The finite element analyses had been performed under the steady-
state condition for convenience of comparison of ABAQUS results with design
calculations. The stud temperatures were based on the results of a temperature
distribution study by Feng et al. (2003a) on cold-formed thin-walled steel panel
systems. An eigenvalue buckling analysis was undertaken first, in which the buckling
modes were obtained and the deflection profile of the lowest buckling mode was
used to determine the stud initial imperfections. Later the temperatures and loads
were applied consecutively.
(a) Case 1 (b) Case 2 (c) Case 3
Figure 2.10: Boundary Conditions used by Feng et al. (2003b)
It was seen that under fire conditions, the maximum initial imperfection does not
have any noticeable effect on the behaviour of studs due to the dominance of thermal
bowing. Hence a value of L/1000 has often been used similar to other researchers.
This study revealed that, by increasing the temperatures of a column, as a result of
thermal bowing and a shift of the neutral axis, the failure mode changes from local
buckling at ambient temperature to either a combination of local buckling, flexural
buckling and bi-axial bending or a combination of torsional–flexural buckling and bi-
axial bending at high temperatures.
For long columns, thermal bowing effect becomes critical due to non-uniform high
temperatures compared to short columns where it is sufficient to use reduced strength
and elastic modulus of steel at elevated temperatures. The non-uniform temperature
2-36
Literature Review
field in the cross section of a column was simplified into two cases (see Figures 2.11
(a) and (b)) and Case 1 was found to be suitable for hand calculations, where uniform
temperatures in the flanges and lips and a linear temperature distribution in the web
were assumed.
(a) Case 1 (b) Case 2

Figure 2.11: Temperature Distributions used by Feng et al. (2003b)
The ABAQUS simulation results were also compared with predictions by using a
design method based on the limiting temperature method of Lawson (1993).
Lawson’s method of limiting the temperature was found to be not suitable to
calculate the critical temperatures of cold-formed thin-walled columns under non-
uniform high temperatures.
Local buckling of simply supported plate sections and lipped channel sections was
modelled in Kaitila (2002). The lipped channel section was simply supported in the
direction of the plane of the flanges along all the edges of the web plate. In order to
have perfectly symmetric conditions in the model, the load was applied to the
member at both ends of the plate. The load was divided only between the nodes of
the web, and not all the nodes at the end of the member. However, the reason for this
application is not explained in their report. At each end, the vertical displacements
were restrained at the loaded nodes. At mid-height of the member, the axial
displacements were restrained at the central web node.
The flexural buckling of lipped channel columns in fire conditions was modelled by
Kaitila (2002). The temperature was assumed to be constant in each flange and vary
linearly over the height of the web. The lip was assumed to be at the same
temperature as the corresponding flange. The average temperature in the cross-
2-37
Literature Review
section was used in the determination of the ultimate loads. The mechanical
properties used in the analytical estimations were based on the transient test data
from Outinen et al. (2001). The finite element models of C-sections included rigid
end plates connected to each node at each end. The members were considered as pin-
ended. The boundary conditions used in this study are shown in Figure 2.12.
Figure 2.12: Boundary Conditions and Loading for Column Models Subject to
Flexural Buckling (Kaitila, 2002)
In Kaitila’s (2002) study S8R5 shell elements were used. An imperfection sensitivity
analysis was performed in respect to flexural and local imperfections for the constant
temperature models. It was shown that the magnitude of the studied type of local
imperfections has an effect on the compression stiffness of the members whereas the
magnitude of global flexural imperfections has more influence on the ultimate
strength. A second set of finite element analyses was carried out on columns subject
to an increasing temperature gradient and a constant axial compressive load. The
mesh for these columns was made denser in the centre area expecting the most
important deformations to occur at mid-height. Three-dimensional rigid beam
elements with two nodes were used.
In the analysis the rigid elements opposed the expansion of the structural elements
during temperature increase close to the ends. Hence large stress concentrations were
developed. To eliminate this problem, the temperatures close to the column ends
were modelled at lower temperatures than the rest of the column. In addition to these
studies, torsional-flexural buckling of lipped channel columns in fire conditions was
2-38
Literature Review
also considered by Kaitila (2002). The boundary conditions were similar to Figure
2.12 except that the condition u2 = 0 along the length of the column was applied to
only one flange.
In the study of Zhao et al. (2005), the structural numerical model was developed
using the finite element softwares such as ABAQUS, SAFIR and ANSYS. The
comparison between these different computer codes showed a good agreement
between them. The boundary conditions used in this study are shown in Figure 2.13.
At room temperature the yield strength and modulus of elasticity were taken as 402
MPa and 210000 MPa, respectively. Poisson ratio was taken as a constant value of
0.3. The steel stud was modelled with shell element and the axial load was applied as
a surface load with a nominal eccentricity of 5 mm for specimens under concentric
load. The lateral displacements of screw positions were restrained to simulate the
connection of steel stud with plasterboard. This boundary condition was applied to
the centre of both exposed and unexposed flanges for walls lined with plasterboards
on both sides. The behaviour of steel studs is simulated with two different end
conditions, one hinged at both ends and the other fixed at one end and hinged at the
other end. Both ends of steel stud are modelled using a stiffer material with the value
of modulus of elasticity taken as a temperature-independent value of 210x105 MPa.
Figure 2.13: Boundary Conditions Adopted for Steel Studs (Zhao et al., 2005)
2-39
Literature Review
Temperatures along the edge stiffeners were taken as equal to those of the web at the
corresponding height. The modelled temperatures were piecewise linear over section
height as shown in Figure 2.14 (a). The temperatures applied in VTT tests were
based on the temperatures recorded at mid-height of the columns during the tests.
For CTICM test results, a temperature gradient has been assumed along the length of
the steel studs (see Figure 2.14 (b)). Initial imperfection obtained from eigenvalue
buckling analysis was used in numerical simulations which consist of sinusoidal
waves in the web with a maximum amplitude of 1 mm.
(a) Across the section (b) Along the stud length

Figure 2.14: Temperature Distribution Used for Numerical Analysis
(Zhao et al., 2005)
The difficulty of obtaining a good agreement with test results was revealed in this
numerical study of the concentrically loaded steel stud. For these tests, the failure
mode of specimens predicted numerically was often different from experimental
failure modes. In the numerical study, some specimens buckled on the exposed side
while in the fire test steel studs buckled towards the unexposed side. This difference
was attributed to imperfections and loading eccentricity on the exposed side because
of different contact effects with end plates due to differential thermal elongations of
steel stud. The boundary conditions and partial restraints provided by the adjoining
structures, namely the plasterboards, were not accurately modelled. Except for the in-
plane boundary conditions applied at the screw locations in the flanges, the
contribution of the plasterboards to the structure by providing axial and out-of-plane
restraint to the stud was ignored. The plasterboards gradually degrade and may reach
a condition at which they are no longer capable of preventing lateral buckling during
2-40
Literature Review
exposure to fire. However, the modelling of possible restraining characteristics and

their temperature dependencies were neglected in the study of Zhao et al. (2005).
In the parametric study, the temperature profiles were over-simplified as shown in
Figure 2.15 and the validity of this assumption has to be investigated. From the
parametric study, critical temperatures were found as a function of stud sections, stud
lengths, heating and loading conditions.
(a) Uniform heating condition
(b) With temperature gradient
Figure 2.15: Heating Condition of Steel Studs for Parametric Studies

(Zhao et al., 2005)
Past research indicates that the numerical models used to evaluate the structural
response of LSF wall panels were mainly based on finite element analyses. The
structural modelling of LSF wall systems was simplified by the use of individual
members (studs) in all studies. However, most of the models were not fully validated
due to lack of experimental results. Also due to the complexity of the problem the
failure locations in the tests were incorrectly predicted while the structural behaviour
of steel studs including their failure modes at elevated temperature was not fully
understood.
2-41
Literature Review
2.8. Design Rules Proposed by Other Researchers

Many fire resistance design guidelines include equations adapted from elementary
mechanics to address buckling and moment capacities of individual components
exposed to fire. The principal difference in applying the same equations from
ambient temperature structural calculations for fire resistance evaluations is adjusting
the mechanical property values to reflect their dependence on temperature (Milke,
1999). This section presents the various design rules proposed by other researches of
LSF wall systems in the past, and critically reviews them.
The design model proposed by Klippstein (1980b) is based on the allowable stress
method given by the AISI design manual (AISI, 1986) and ASTM-E119 (ASTM,
1973). ASTM-E119 (ASTM, 1973) presents the development of a design fire similar
to the ISO 834 (ISO, 1999) standard fire (Ranby, 1999). In the study of Klippstein
(1980b) it was assumed that the failure by weak axis flexural buckling or torsional
buckling will be prevented by the gypsum board cladding on the internal and external
faces of the wall. However, the ability of the gypsum boards to carry any vertical
loads was neglected. The stress-strain curve of the steel material was assumed to be
linear up to the yield strength. The failure condition given by Klippstein (1980b) is,
1
PT = (2.31)
1 δT
+
 23  S x FyT
A FalT 
 12 
where PT is the predicted failure load of the stud at an elevated temperature T; A and
Sx are the gross cross-sectional area and the major axis section modulus; FalT is the
allowable stress for axial load at failure temperature; δT is the total mid-height
deflection of the stud at the time of failure and FyT is the yield strength at failure
temperature.
The proposed design method by Klippstein (1980b) can be used with reasonable
accuracy to predict the failure temperatures of this particular type of steel wall up to
about 650oC. However, it is heavily dependent on empirical determination of the
variation of the temperature of the stud and the mid-height lateral deflection during
2-42
Literature Review
the fire. The method is also limited to the type of section used (regular C section), the
amount of insulation, cladding and other physical characteristics.
Lawson (1993) adopted the limiting temperature method in BS 5950 Part 8 (BSI,
1990) for hot-rolled steel structures to cold-formed thin-walled steel structures. The
limiting temperature is a function of the load ratio of the structural member. The load
ratio is the ratio between the load on member at the fire limit state and the load
carrying capacity of that member under normal loading. Usually hot-rolled columns
have a uniform temperature distribution at failure due to fire attack from all sides.
However, cold-formed thin-walled columns will be subject to severe non-uniform
temperature distributions. Hence the BS 5950 Part 8 (BSI, 1990) method is, strictly
speaking, not applicable to cold-formed thin-walled studs as Lawson (1993) did
(Feng et al., 2003b).
Gerlich et al. (1996) used a thinner unexposed lining (9.5 mm) for a test to achieve
an insulation failure. However, it was found that the unexposed lining failed to
provide lateral support after it has degraded. With the comparison of TASEF
modelling and test results, Gerlich et al. (1996) recommended a 3 mm thickness of
undamaged gypsum to be retained in order to restrain lateral buckling. According to
Gerlich et al. (1996), when steel temperatures reach critical levels (> 300-400oC) the
fire exposed side plasterboard will reduce its ability to prevent buckling of the studs
due to degradation. Hence in the design model of LSF wall at fire conditions, the
lateral restraint provided by the linings on the fire side was ignored. The torsional-
flexural buckling was neglected in this model and it was found that the failure mode
of steel studs in LSF systems exposed to fire was generally governed by the buckling
of compression flange on the ambient side of the wall assembly.
In order to predict stud bending moments, analytical methods were proposed for
calculating the thermal deflection arising from a temperature gradient across the steel
members and the superimposed deflection due to P-∆ effects (see Figure 2.16).
Hence the model presented by Gerlich et al. (1996) considered the additional
deflection caused by bending moment.
2-43
Literature Review
Axial load
P X ∆1
P
Figure 2.16: Total Horizontal Deflection for Load-bearing Systems According to

Gerlich et al. (1996)
Cooke (1987) considered the thermal bowing of simply supported steel members due
to a temperature gradient across the section and derives the following expression for
mid-height deflection,
e∆T = αL ∆T
2
(2.32)
8d
where e∆T is the mid-height deflection due to thermal bowing; α is the thermal
expansion coefficient for steel (°C-1); L and d are the member height and depth,
respectively, and ∆T is the temperature difference across the member.
This equation was used by Gerlich et al. (1996) to predict the mid-height deflection
of studs. It reasonably predicts the mid-height deflection at lower temperatures,
provided that the steel studs are free to rotate and expand at both ends. However,
after longer fire exposure the temperature difference across the steel member reduces
but actual deflections do not return to the calculated levels because of plastic
deformations of the steel.
The stress-free thermal deflection was treated as an initial eccentricity e∆T by

Gerlich et al. (1996). The initial bending moment P e∆T results in a further horizontal
deflection e P∆T . Therefore the total horizontal displacement ( e∆T + e P∆T ) of the
member will be the sum of the thermal deflection and the deflection due to P-∆
2-44
Literature Review
effects. The P-∆ component was predicted analytically by solving the moment
equilibrium equation as
 
 1
e P∆T = e∆T − 1 (2.33)
 µ L 
 cos 2 
Pa
with µ is
ET I X
where
ET is the elastic modulus of steel as a function of temperature; L and Ix are the wall
height and second moment of area of the cross-section; Pa is the applied axial load;
e∆T and e P∆T are the initial eccentricity (thermal deflection) and P-∆ deflection.
Calculated stresses in the studs are given by the following equations,

For the flange on the fire side:
Fyz =
Pa
−
(Pa ∆ z ) (2.34)
A Zx
For the flange on the ambient side:
Fyz =
Pa
+
(Pa ∆ z ) (2.35)
A Zx
where Fyx is the calculated steel stress at height z; Pa is the applied axial
compression load and ∆z is the measured total horizontal deflection at height z; A and
Zx are the cross sectional area of the stud and section modulus about major axis.
Three design codes, AS 1538 (SAA, 1988), BS 5950 Part 5 (BSI, 1987) and the AISI
design manual (AISI, 1991) were used by Gerlich et al. (1996) to predict the limit
state condition of cold-formed steel structures at room temperature. It was concluded
that the AISI design manual provides the most recent and reliable source for
predicting the ultimate limit state conditions of cold-formed steel studs at room
temperature. Therefore based on the AISI design manual equations, Gerlich et al.
(1996) developed a spreadsheet by adjusting the input values for yield strength and
modulus of elasticity to take into account the effect of temperature using the
reduction factors given by Klippstein (1980b). With the spreadsheet, a critical
temperature was found at which the predicted strength of the stud was equal to the
2-45
Literature Review
applied axial load. This temperature was compared with the compression flange
temperature on the ambient side of the wall to find the time to failure. The TASEF
predictions were used to find the time until which the unexposed lining will provide
lateral restraint to the compression flange. Hence the lower value obtained from these
analysis was considered as the predicted failure time.
Gerlich et al.’s (1996) study was limited to the steel grades of 300 and 450 with the
thicknesses of 1.15 mm and 1 mm. The stud failure was checked only at the mid-
height of the wall. However, many other researchers (Alfawakhiri, 2001, Zhao et al.,
2005, Feng and Wang, 2005a) found that the failure occurred at different points
along the stud height. Also Gerlich et al. (1996) compared the critical temperature
found by the model, with the compression flange temperature on the ambient side of
the wall, to find the time to failure. This is not sufficient and consideration should be
given to the fire exposed side steel temperature.
Ranby (1999) found that the basic design equations given in Eurocode 3 Part 1.3
(ECS, 2006) for flexural and flexural-torsional buckling at ambient temperature
could be used at elevated temperatures if appropriately reduced mechanical
properties at elevated temperatures are used. In the study of Ranby (1999) it was
found that for a C-section with a web height d = 100 mm, Klippstein’s (1980b)
method gives higher resistance values when compared to the methods proposed by
Gerlich et al. (1996) and Ranby (1999). On the other hand, when the web height was
doubled (d = 200 mm), Klippstein’s (1980b) method was conservative compared to
Gerlich et al.’s (1996) model. Ranby’s (1999) own model generally gave values
between the other two models.
Wang and Davies (2000) used the design equations in Eurocode 3 Part 1.3 (ECS,
2006) to carry out a theoretical study to calculate the fire resistance of thin-walled
cold-formed members. Uniform temperature values were assumed for flanges and
lips on both the fire and cold sides. The temperature distribution in the web was
assumed, such that the distribution of steel strength and stiffness is linear. Eurocode
3 Part 1.3 (ECS, 2006) was used to evaluate the effective widths of flanges and lips,
replacing the ambient temperature properties by elevated temperature properties. The
average steel strength and stiffness values are used for webs due to changing steel
2-46
Literature Review
properties. The design method was validated against the results of three tests on
channel columns by Gerlich et al. (1996). In this study it was found that the
temperature gradient can be beneficial or detrimental hence it was concluded that it
may not always be safe to follow the SCI (1993) recommendations that only give
improved failure temperatures. The ambient temperature approach was found to be
suitable for adoption under fire conditions by taking into account reductions in the
strength and stiffness of steel at elevated temperatures, and the additional bending
moments due to thermal bowing and shift in the neutral axis.
In the study of Feng et al. (2003b) the ambient temperature design method for cold-
formed thin-walled columns in Eurocode 3 Part 1.3 (ECS, 2006) was found to be
suitable to predict the capacity of cold-formed thin-walled columns at elevated
temperature. However, consideration was given to reductions in the strength and
stiffness of steel at elevated temperatures, additional bending moments due to
thermal bowing, partial plasticity and a shift in the neutral axis (see Figure 2.17).
Figure 2.17: Neutral Axis Shifts in Lipped Channel Section under Non-uniform
High Temperatures (Feng et al., 2003b)
A uniform temperature distribution was assumed for flanges and lips and the method
in Eurocode 3 Part 1.3 (ECS, 2006) was directly used to evaluate their effective
widths, but using the elevated temperature properties. The weighted average steel
stiffness value was used for web to calculate the effective width due to different steel
properties at non-uniform temperature distributions. Hence in this study both local
and global buckling effects were considered. The proposed Eurocode 3 Part 1.3
(ECS, 2006) method was used with both first occurrence of material yield and partial
plastic behaviour and validated against ABAQUS results. Overall, the agreement
2-47
Literature Review
between partial plasticity results and ABAQUS results was better than with the first
yield results.
For the studs exposed to fire conditions, compression is on the cold side which has a
higher strength, and tension is on the hot side with a lower strength at stud mid-
height. Hence partial plasticity was considered whereby tension stresses at the
extreme fibres have reached yield and the maximum compression stress at the
extreme fibre is equal to the yield stress, as shown in Table 2.10 (2). Alternatively,
the lower of either first compression yield or tension yield was considered as shown
in Table 2.10 (1). Both were considered in the proposed design calculations of the
maximum bending moment capacity Mx,eff. At the supports first yield of the
compression flange was adopted when calculating the maximum bending moment
capacity Mx,eff.
In this study the bending moment about the major axis at the column centre was
considered as a result of a shift of the centroid and thermal bowing. However, near
the column ends, this was only due to neutral axis shift. Both cross-sections were
checked using the following design equations.
At the support about the major axis

M x , sd + ∆M x , sd = − Pe x (2.36)
At the mid-height about the major axis

α∆TL2
M x , sd + ∆M x , sd = − P(e∆T − e∆E ) with e∆T = (2.37)
8d
About the minor axis
M y , sd + ∆M y ,sd = Pe y (2.38)
where
ex and ey are the shift of neutral axis in the major and minor axis, respectively;
e∆T is the maximum lateral deflection due to thermal bowing
∆T is the temperature difference across stud section;
d and L are the column section depth and height, respectively.
2-48
Literature Review
Table 2.10: Failure Criteria for Calculating the Maximum Bending Moment
Capacity (Feng et al., 2003b)
This study compared the limiting temperature method of Lawson (1993) and
concluded that this method is not suitable to calculate the critical temperatures of
LSF under non-uniform temperature conditions. However, it was found that the
failure temperatures in a column with either single or double layers of gypsum
boards are very similar at the same load ratio. Hence the effect of using double layers
of gypsum board was simply to delay the time to reach the same temperatures in the
steel section with a single layer of gypsum board.
Feng and Wang (2005b) presented a theoretical analysis to predict lateral deflections
and failure times. The local and global buckling effects were considered after taking
into account their thermal bowing deflections and associated magnifications. This
study considered the combination of bi-axial bending and axial compression. When
checking the stud load carrying capacity the partial plasticity was included.
2-49
Literature Review
The thermal bowing deflections in the plane of the flanges were neglected. Also it
was assumed that the load is applied through the original centroid of the stud cross-
section and the stud has simple rotational supports at ends. The behaviour of column
due to non-uniform temperature effect according to Feng and Wang (2005b) is
shown in Figure 2.18.
Figure 2.18: Behaviour of Column due to Non-uniform Temperature Effects

(Feng and Wang, 2005b)
For a simply supported column, pure thermal bowing deflection curve is given by
α∆T α∆TL
ν0 = − z2 + z (2.39)
2d 2d
where d is the overall depth of the cross-section
Using Energy theory and the boundary conditions of the steel studs at mid-height and
supports, the mechanical deflection profile ν1 was proved to be as,
2α∆T
PL
4δ m 2 4δ m 3d
ν1 = − 2 z + z with δ m = (2.40)
L L 64 16
3
EI − P
L 3L
Hence the total lateral deflection related to the original centroid of the gross cross-
section (load point) of the steel stud was given by y = ν 0 + ν 1 − e x in which ex is the
shift of neutral axis about the major axis.
2-50
Literature Review
The maximum lateral deflection at mid-height (z=L/2), which is the magnified

thermal bowing deflection proved by Feng and Wang (2005b) is given by,
2α∆T
PL
3d α∆T 2
e=− + L − ex (2.41)
64 16 8d
EI − P
L3 3L
For LSF steel studs at elevated temperatures, design calculations are given in
Eurocode 3 Part 1.2 (ECS, 2005), which simply refers to Eurocode 3 Part 1.3 (ECS,
2006). In addition to the global buckling check according to the equations given in
Eurocode 3 Part 1.3 (ECS, 2006), local capacity should also be checked, and
Eurocode 3 Part 1.2 (ECS, 2005) recommends using effective cross-section at
ambient temperature without modification for temperature effects. However, Feng
and Wang (2005b) introduced a new method to allow for temperature effects on the
effective cross-section (referred to as modified Eurocode 3 Part 1.3).
The proposed equations are

P k (M + ∆M x ) k y (M y + ∆M y )
+ x x + ≤1 (2.42)
χ min N eff M x ,eff M y ,eff
P k (M x + ∆M x ) k y (M y + ∆M y )
+ LT + ≤1 (2.43)
χ lat N eff χ LT M x ,eff M y ,eff
where
Neff is the stud axial compression capacity at elevated temperatures;
Mx,eff and My,eff are the stud bending moment capacities at non-uniform elevated
temperatures about the major and the minor axis, respectively.
The bending moment about the major axis due to thermal bowing and the shift of
centroid was expressed as P(ν0+ ν1-ey).
Hence the maximum bending moment at mid-height, with magnification of thermal
 2α∆T 
 PL 
3d α ∆TL2
bowing is ∆M χ ,max = P + − ex  (2.44)
 16 EI − 4 P 8d 
 L3
3L 
α∆TL2 
Without magnification of thermal bowing ∆M χ ,max = P − ex  (2.45)
 8d 
The bending moment at the support ∆M y ,max = − Pe y (2.46)
2-51
Literature Review
Equations 2.44 to 2.46 were used in Equations 2.42 and 2.43 to find the influence of
thermal bowing and magnification effect in the design of LSF members at elevated
temperatures.
In this study, uniform temperature distribution is assumed for flanges and lips. The
web temperature distribution was assumed as linear variation. This study also
assumed a uniform temperature distribution along the stud height. Based on this
research, it was recommended that for design calculations, it is not necessary to
consider the effect of increasing thermal bowing deflections due to axial compression
and reducing elastic modulus of steel at elevated temperatures. Modified Eurocode 3
Part 1.3 (ECS, 2006) was found to be giving slightly better agreement with test
results, however, Eurocode 3 Part 1.2 (ECS, 2005) was found to be easier to
implement because at non-uniform elevated temperatures it does not require
additional calculations of effective areas of thin-walled cross-sections.
According to Feng and Wang (2005b) the neutral axis shift of the major axis had
some effect and may change the panel failure position from the mid-height to the
support. However, test results showed that the time and the point at which the panels
failed matched well, when the neutral axis shift was ignored. The effect of shift of
the minor axis was found to be very small on the prediction of panel failure times
and could be ignored in order to simplify routine design. Adopting the partial
plasticity design method gave slightly higher stud load carrying capacities.
Feng and Wang (2005b) extended Feng et al.’s (2003b) study to 1.2 mm and 2 mm
thickness steels with the grade of S350. However, higher steel grades were not
investigated. Also in this study, each stud had holes near the supports, and studs
without holes were not investigated. The main weakness of this study is the
complexity of design rules developed. In the conclusion of this study itself, it was
mentioned that the elevated temperature calculation methods are extremely complex
and an alternative method should be sought.
According to Kaitila (2002), the model is developed as two parts. First the design
model was based on equations that consider flexural buckling along the stronger axis
only and the torsional-flexural buckling was not considered as a critical failure mode.
2-52
Literature Review
This was when the temperature stays under the level of calcination of the
plasterboards. Hence the lateral restraint to the studs by the plasterboards was taken
into account throughout the fire. Secondly, the torsional-flexural buckling mode was
checked. This is when the temperature during the fire goes above the level of
calcination of the plasterboards (equal to approximately 550oC). Hence the lateral
restraint was no longer valid on the fireside.
First, the model for flexural buckling was developed by modifying the equation for
bending and axial compression given in Eurocode 3 Part 1.3 (ECS, 2006). It is
simplified by Kaitila (2002) as
χf y Aeff
Nu =
1+
(
k x e d − y CF χAeff ) (2.47)
I eff
In which
χ=
1
2
[ ( )
≤ 1 with ϕ = 0.5 1 + α λθ − 0.2 + λθ
2
] where α = 0.34
ϕ + ϕ − λθ2
1l f y Aeff
λ=
π i E A
µx Nu
kx = 1− ≤ 1.5 with µ x = λ (2 β M , x − 4 ) ≤ 0.90 and β M , x = 1.3
χf y Aeff
where
Nu is the ultimate axial compressive load (NSd=Nu);
e is the effective eccentricity due to the changes in effective elastic material
properties and thermal bowing and d is the height of the web;
Aeff and Ieff are the effective area and second moment of area of the steel cross-
section;
y CF is the distance from the supported colder flange to the effective neutral axis;
fy is 0.2% proof stress corresponding to the average temperature over the steel
section.
The deflection caused by the two different effects (shift of neutral axis due to
changes in effective modulus of elasticity and deflection of column due to
temperature gradient) of the thermal gradient was expressed as eo = e∆T – e∆E. The
2-53
Literature Review
total effective eccentricity is not known at the start. The temperature gradient causes
a deflection. The magnitude of the deflection e∆T due to the thermal gradient is
calculated using Equation 2.32. However, the axial load is applied at the original
cross-section centroid. Due to this effective eccentricity, bending moment will be
developed. This leads to more deflection and therefore an increase in the bending
moment, and so on. Therefore the problem has to be solved iteratively, until the
value of eo in Equation 2.48 remains constant throughout the iteration.
1
e = e0 (2.48)
N
1− u
N cr , x
where eo is the original effective eccentricity due to the thermal gradient and should
be taken as a positive value,
Ncr,x is the critical Euler buckling load about the major axis given by
π 2 ∑ EI gr
N cr , x = where ΣEIgr is the sum of the bending stiffness weighted with the
L2
variation of the modulus of elasticity across the section.
Secondly the model incorporated the studs with one (fire side) flange unrestrained
(Restrained torsional-flexural buckling). The ultimate load was also calculated again
using the equation given in Eurocode 3 Part 1.3 (ECS, 2006). This is simplified as
χf y Aeff
Nu =
1−
(
k x e d − y CF χAeff ) (2.49)
I eff
The parameters were defined as Equation 2.47 except the effective slenderness,
which was calculated as
fy
λ=
N cr N cr e d − y CF
+
( ) (2.50)
Aeff I eff
where the torsional-flexural buckling load Ncr was calculated based on StBK-K2
(1983) and Nylander (1956). The equation was derived using the energy method as,
2-54
Literature Review
1  d 2 π 2 
N cr = 2

 x
B + C w  2 + C
 (2.51)
d 8d  4 l 
+ i p2 − e
4 3π
where Bx is the bending stiffness equal to EIx for bending about the major axis, where
E is the modulus of elasticity and Ix is the second moment of area about the major
axis; Cw is the warping stiffness equal to EIw, where Iw is the warping constant; C is
the torsional stiffness equal to GIt, where G is the shear modulus and It is the torsion
constant and i is the coefficient of restraint along the column at the flange in tension.
Simple calculation rules for assessing the fire resistance of lightweight steel studs
maintained by boards were proposed by Zhao et al. (2005). The approach is similar
to the proposed method by Eurocode 3 Part 1.3 (ECS, 2006) for lightweight
structures at room temperature. Mechanical properties were modified according to
temperature and the thermal bowing effects were considered in this study. The stress-
strain relationships of cold-formed steel at elevated temperature were assumed to be
perfectly linear elastic-plastic in the study of Zhao et al. (2005). Also the yield stress
was taken as the 0.2% proof stress at any temperature level.
Simplified design method for studs under flexural buckling was developed by Zhao
et al. (2005). The buckling resistance of steel studs maintained by plasterboards
under compression, Nfi,Rd, was expressed as follows:
( )
N fi , Rd = χ λθ N fi ,c , Rd (2.52)
where χ =
1
2
[ (
≤ 1 with ϕ = 0.5 1 + α λθ − 0.2 + λθ ) 2
]
ϕ + ϕ − λθ 2
λθ = ∑A f i y ,θ ,i N fi ,cr with N fi ,cr = π 2 (EI ) fi ,eff Lθ2 and (EI ) fi ,eff = ∑ Ei ,θ I i ,θ

i i
N fi ,c , Rd = ∑ Aeff ,i f y ,θ ,i
i
where χ is the reduction factor for buckling resistance; λ θ is the relative slenderness
at elevated temperature and Nfi,c,Rd, is the design resistance load of the steel stud; Ai
and Aeff,i are the initial element area and effective element area, respectively; fy,θ,i is
2-55
Literature Review
the 0.2% proof yield strength of steel at temperature θi; Nfi,cr is the Euler buckling
load and Lθ is the buckling length of the stud in fire situation; (EI)fi,eff is the effective
flexural stiffness of the cross-section; Ei,θ is the modulus of elasticity of each plate
element; θ is the appropriate elevated temperature and Ii,θ. is the second moment of
area of each element; α is an imperfection factor, depending on the appropriate
buckling curve (as function of the buckling axis and geometrical characteristics).
The reduction factor should be determined in the same way as at room temperature,
but using an appropriate buckling curve. It should be reminded that Eurocode 3 Part
1.1 (ECS, 2005) considers three different column buckling curves for different types
of column and Eurocode 3 Part 1.3 (ECS, 2006) recommends the buckling curve ‘b’
for C-sections.
The effects of local buckling are taken into account on the basis of the effective
widths of individual plate elements. The reduction factor ρ is determined from:
ρ = 1 when λ p ≤ 0.673 (2.53a)
1 − 0.055(1 + ψ ) λ p
ρ= when λ p ≥ 0.673 but ρ ≤ 1 (2.53b)
λp
in which the plate slenderness λ p is defined by
λp =
bp ( )
12 1 − ν 2 f y
t π Ek ρ
2
where t is the plate thickness and kσ is the plate buckling factor which depends on the
edge condition and stress distribution of the plate (kσ = 4 for elements under uniform
compression).
The stability of a lightweight steel member subjected to a combination of axial

compression and bending is checked by Zhao et al. (2005) as,
N sd k x (M x ,sd + ∆M x , sd ) k y (M y , sd + ∆M y , sd )
+ + ≤1 (2.54)
N fi , Rd M fi , Rd , x M fi , Rd , y
2-56
Literature Review
where Nfi,Rd, is the design buckling resistance of a compression member (Equation

2.52); Mfi,Rd,x and Mfi,Rd,y are the design bending moment resistance (based on the
effective cross-section) that is subjected to bending moment only about the major
and the minor axis, respectively and all other parameters are defined in Equation 2.8.
To simplify calculations, when compression is on the hot side of stud, the moment
resistance Mfi,Rd of the cross-section is calculated with yielding occurring on the hot
flange and a linear stress distribution in the web. When compression is on the cold
side, the moment resistance is calculated with yielding occurring on both flanges
with bi-linear stress-distribution in the web on the hot side.
The bending moment at the end of stud about the major axis is:
M x , sd + ∆M x , sd = N sd (e − e∆E ) (2.55)
At the mid-height, the bending moment about the major axis is:
M x , sd + ∆M x , sd = N sd (e − e∆E + e∆T ) (2.56)
Both cross-sections were checked in the design calculations
∑E θ y A i, i i
1 α T ∆TL2
with e∆E = i
, e P∆T = e∆T and e∆T =
∑E θ Ai
i, i
1−
N sd 8d
N fi ,cr
where e is the effective eccentricity of the applied load; e∆E is the shift of neutral axis
about the major axis; e∆T is the maximum mid-height horizontal deflection due to
temperature gradient across the section; e0 is the deflection related to thermal
bowing; NSd is the applied load; Nfi,cr is the Euler buckling load; ∆T is the
temperature difference across the stud section in °C; d is the web depth; L is the
column height; αT is the thermal expansion coefficient of steel; y is the distance from
the effective plate element i to the neutral axis at ambient temperature.
In this study the effects of local buckling and distortional buckling are taken into
account according to Eurocode 3 Part 1.3 (ECS, 2006) using the effective width for
all plate elements. The flange stiffeners and the flanges on each side of the cross-
section were assumed to be at the same temperature. Therefore the method in
2-57
Literature Review
Eurocode 3 Part 1.3 (ECS, 2006) was directly used to evaluate their effective widths,
but using the elevated temperature properties to replace those at ambient
temperature. The weighted average steel resistance (strength and stiffness) value is
used to calculate the web effective width.
In most of the tests carried out by Zhao et al. (2005), the failure locations were
wrongly predicted and the behaviour of steel studs at elevated temperature was not
fully understood. For some tests, the difference between calculations and
experimental results was explained by the fact that the end restraint conditions at the
bottom of studs turn out to be more hinged support condition rather than fixed as
planned. The critical temperatures obtained from the numerical analysis were
compared with the values obtained from the simplified methods and it was found that
in some cases the variation was up to 30 percent. Also, the failure time obtained for
all of the tests did not exceed the benchmark of 90 minutes except one test in which
it went up to 105 minutes. Therefore further research is needed to find the critical
temperatures of the newly developed composite panels where the failure time
exceeds 130 minutes.
These observations indicate that in the past research studies the many important
effects such as the thermal bowing effects, the second order deflection due to
bending, the neutral axis shift due to temperature gradient, and the eccentricity of
loading due to curvature have not been considered fully. The numerical and
calculation models were proposed based on AISI design manual and Eurocode 3 Part
1.3 (ECS, 2006) by giving considerations to reduction in strength and stiffness of
steel at elevated temperatures. The models were developed mainly for flexural
buckling about the major axis. Local buckling and flexural-torsional buckling were
also considered in some studies. However, these models are very complicated and
their usability in routine design work is not possible.
2-58
Literature Review
2.9. Literature Review Findings

In this section the important literature review findings are presented.
1) According to Feng et al. (2003a), the interior (cavity) insulation was found to be
giving positive impact on the fire resistance of steel wall panels. However, in the
studies of Kodur and Sultan (2001) and Alfawakhiri (2001), wall assemblies without
cavity insulation provided higher fire resistance compared to cavity insulated
assemblies. Hence the past researches were not able to conclude the efficiency of
traditional approach of placing the insulation inside the cavity. Recently Kolarkar
and Mahendran (2008) developed a new composite panel system, where the
insulation was placed outside the steel frame and it was found that the fire resistance
has improved considerably. However, Kolarkar and Mahendran’s (2008) study was
limited to an experimental study with a load ratio of 0.2. Hence a further
experimental study is needed with higher load ratios. Also numerical and theoretical
analyses are needed to fully understand the improvements offered by the new system.
2) Gerlich et al.’s (1996) study was limited to steel grades of 300 and 450 with
thicknesses of 1.15 mm and 1 mm. The nominal metal thicknesses of 0.84 and 0.912
mm were used by Kodur and Sultan (2001). In the study of Alfawakhiri (2001) thin
sheets with a metal thickness of 0.912 mm were used with low steel yield strength of
228 MPa. Feng et al.’s (2003b) study was limited to a thickness of 1.2 mm. Feng and
Wang (2005) extended their study to 1.2 mm and 2 mm thickness steels with the
grade of S350. In Zhao et al.’s (2005) study, steel thicknesses of 0.6, 1.2 and 2.5 mm
were considered. The yield strength of the steel used was 402 MPa and steel grade of
S280 and S350 were used in the parametric study. Hence it can be concluded that the
behaviour of LSF wall panels at elevated temperature was investigated for lower
steel grade and for limited thicknesses in the past research. Therefore further research
is needed on LSF wall systems made of steel with varying thicknesses and higher
steel grades.
3) All three tests of Gerlich et al.’s (1996) study were found to have failed towards
the furnace. In the studies of Alfawakhiri (2001) and Zhao et al. (2005), the studs
failed towards the furnace as well as away from the furnace. The studs failed at mid-
2-59
Literature Review
height for Gerlich et al.’s (1996) tests. However, according to Alfawakhiri (2001)
and Zhao et al. (2005), the studs failed at mid-height as well as near supports. In
most of the previous studies the failure locations of the tests were wrongly predicted
and the behaviour of steel studs at elevated temperature was not fully understood.
Hence further experimental, numerical and theoretical studies are required in the
field of fire exposed LSF members in order to fully understand their behaviour and
failure modes during fire tests.
4) Most of past researches were carried out outside Australia, representing specific
materials and method of construction used in those countries. Except Kolarkar and
Mahendran (2008), there has not been any research on LSF walls in Australia. Hence
the behaviour of Australian LSF wall systems has to be investigated in detail.
5) Uniform temperature values were assumed for flanges and lips of lipped channel
studs on both the hot and cold sides in the studies of Wang and Davies (2000),
Alfawakhiri (2001), Kaitila’s (2002), Feng et al. (2003b), Feng and Wang (2005b)
and Zhao et al. (2005). However, in Zhao et al.’s (2005) parametric study using
ABAQUS, the temperatures along the edge stiffeners were taken as equal to those of
the web at the corresponding height. Wang and Davies (2000) assumed a temperature
distribution in the web, such that the distribution of steel strength and stiffness is
linear. In other studies such as Alfawakhiri (2001), Kaitila’s (2002), Feng et al.
(2003b), Feng and Wang (2005b) and Zhao et al. (2005) study, linear temperature
variation was assumed over the height of the web. The temperature variation along
the height of the stud was neglected in the previous researches except by Zhao et al.
(2005). In their study also, the importance of introducing the temperature variation
along the height was not discussed. In the parametric study by Zhao et al. (2005), the
temperature profiles with time were over-simplified and the validity of this
assumption has to be investigated. These observations demonstrate the need to study
the true temperature profiles across the studs under fire conditions and to recommend
a simplified temperature profile for use in numerical and theoretical studies of LSF
wall systems.
2-60
Literature Review
6) With the comparison of TASEF modelling and test results Gerlich et al. (1996)
recommended a 3 mm thickness of undamaged gypsum to be retained to provide
lateral buckling restraint. According to Gerlich et al. (1996), when steel temperatures
reach critical levels (> 300-400oC) the fire exposed side plasterboard will reduce its
ability to prevent buckling of the studs due to degradation. Hence in the design
model of LSF wall under fire conditions, the lateral restraint provided by the linings
on the fire side was ignored. Sultan (1996) reported that fall-off of plasterboard
occurs when unexposed face of the board reaches about 600°C (Buchanan and
Gerlich (1997)). According to Ranby (1999) a common temperature of 550°C was
proposed. In the numerical study of Kaitila (2002), the boundary conditions
providing lateral restraints at both flanges were assumed to be valid until 550oC.
According to Kaitila (2002), the model is developed as two parts. The lateral restraint
to the studs by the plasterboards was taken into account throughout the fire for the
first model. In the second model it was ignored. The lateral restraint is provided on
both sides (Hot flange and Cold flange) in the study of Feng et al. (2003b). However,
according to Feng and Wang (2005b), at high temperatures plasterboards on the fire
side lose their ability to restrain the studs. In the model of Zhao et al. (2005), the
lateral displacements of screw positions were restrained to simulate the connection of
steel stud with plasterboard. This boundary condition was applied to the centre of
both exposed and unexposed flanges for walls lined with plasterboards on both sides.
Hence the true lateral restraint provided by plasterboards at elevated temperatures
has to be investigated and determined for accurate use in analytical and numerical
models of LSF wall studs, since it influences their fire design of a steel stud
considerably.
7) The thermal bowing effects were considered in almost all researches including
Klippstein (1980b), Cooke (1987), Gerlich et al. (1996), Ranby (1999), Wang and
Davies (2000), Alfawakhiri’s (2001), Kaitila (2002), Feng et al. (2003b), Feng and
Wang (2005b) and Zhao et al. (2005). However, the second order deflection due to
bending was not considered by Klippstein (1980b) and Feng et al. (2003b). Based on
Feng and Wang’s (2005b) research, it was recommended that for design calculations,
it is not necessary to consider the effect of increasing thermal bowing deflections due
to axial compression and reducing elastic modulus of steel at elevated temperatures.
The neutral axis shift due to temperature gradient across the stud section was
2-61
Literature Review
considered by Ranby (1999), Alfawakhiri’s (2001), Wang and Davies (2000), Kaitila
(2002), Feng et al. (2003b), Feng and Wang (2005b) and Zhao et al. (2005).
According to Feng and Wang’s (2005b) the test results showed that the time and the
point at which the panels failed matched well, when the neutral axis shift was
ignored. The eccentricity of loading due to curvature was considered in the studies of
Alfawakhiri’s (2001) and Zhao et al. (2005) was neglected by all others. These
observations indicate that the effect of thermal bowing, their magnification, neutral
axis shift about the major and minor axes, and the shift of loading point due to
curvature were not investigated fully. Hence this has to be investigated further in
order to identify the inter-relationship and importance of these effects.
8) In Alfawakhiri’s (2001) study, the studs had four perforations and they were found
to be failing at these particular points. In the study of and Wang (2005) also, each
stud had holes near the supports, and studs without holes were not investigated.
Gerlich et al.’s (1996) test frames had a central row of nogs. Cross-bracing was used
in Alfawakhiri’s (2001) study and it was assumed that the flexural-torsional buckling
and weak axis buckling failure modes are prevented by this lateral restraint.
Therefore further study is necessary to investigate the behaviour of stud panels
without perforation and/or bracing.
9) Gerlich et al. (1996) proposed a critical temperature in relation to the ambient side
steel section. However, the critical temperature in relation to the fire exposed side
steel section has to be investigated. Also this model was validated against only three
full scale tests. Feng et al.’s (2001) model which was used to simulate the behaviour
of short columns tested under uniform elevated temperatures was extended to model
the long columns with non-uniform temperature conditions (Feng et al., 2003b).
However, the weakness of Feng et al.’s (2001) study is that the model was not
validated due to unavailability of test results. The critical temperatures obtained from
the numerical analysis of Zhao et al.’s (2005) were compared with the values
obtained from the simplified methods and it was found that in some cases the
variation was up to 30 percent. These observations indicate that in past researches,
the numerical models were not fully validated due to lack of experimental results and
the complexity of the problem. Hence further study is required to develop numerical
models which are fully validated against various experimental results.
2-62
Literature Review
10) The design models proposed by Klippstein (1980b) and Gerlich et al. (1996)
were based on AISI design manual (AISI, 2007). Lawson (1993) adopted the limiting
temperature method in BS 5950 Part 8 (BSI, 1990) for hot-rolled steel structures to
cold-formed thin-walled steel structures. All other studies including Ranby (1999),
Kaitila (2002), Wang and Davies (2000), Feng et al. (2003b), Feng and Wang
(2005b) and Zhao et al., (2005) proposed calculation methods based on the equations
of Eurocode 3 Part 1.3 (ECS, 2006) by giving considerations to reduction in strength
and stiffness of steel at elevated temperatures in these studies. However, the
calculation methods proposed are different to each other and the validity of these
methods to Australian LSF members (including those in the new composite panel
system) have to be checked. In the previous researches local buckling effects were
considered by using the effective cross-section at ambient temperature without
modification for temperature effects. However, Klippstein (1978), Feng and Wang
(2005b) and Zhao et al. (2005) introduced a new method to allow for temperature
effects on the effective cross-section. However, the past researches did not consider
the suitability of AS/NZS 4600 (SA, 2005) design rules for LSF wall studs at
elevated temperatures. Hence further study has to be performed in order to check the
validity of AS/NZS 4600 (SA, 2005), AISI design manual and Eurocode 3 Part 1.3
(ECS, 2006) in the design of Australian LSF wall studs at elevated temperatures.
11) In the study of Gerlich et al. (1996), too many variables were incorporated into
limited test specimens. More than one variable between the tests made the
comparison difficult and uncertain. Also, only one layer of lining was used in this
study. The main weakness of Feng and Wang’s (2005b) study is the complexity of
design rules developed. In the conclusion of this study itself, it was mentioned that
the elevated temperature calculation methods are extremely complex and an
alternative method should be sought.
These observations demonstrate the need to expand the study of LSF walls at
elevated temperatures by incorporating varying thicknesses and higher steel grades.
The behaviour of stud panels without perforation and/or bracing also has to be
investigated. Numerical and theoretical analyses are needed to fully understand the
improvements offered by the new composite system incorporating further
experimental study with higher load ratios. The possible buckling modes of the LSF
2-63
Literature Review
members at elevated temperatures have to be investigated in detail in order to

investigate the temperature effects on LSF wall studs. The lateral restraint provided
by plasterboards at elevated temperatures has to be investigated. Most of the failure
locations of the past tests were incorrectly predicted and the numerical models were
not fully validated due to lack of experimental results and the complexity of the
problem. Also the previously developed elevated temperature calculation methods
are extremely complex. Hence further study is required to recommend a simplified
temperature profile for use in numerical and theoretical studies and to develop simple
models which are fully validated against various experimental results. In addition,
the accuracy of relevant design rules in AS/NZS 4600 (SA, 2005), NAS (AISI, 2007)
and Eurocode 3 Part 1.3 (ECS, 2006) must be investigated in the design of Australian
LSF wall systems at elevated temperatures including the new composite panel
system developed recently at QUT.
2-64
Experimental Study
CHAPTER 3
3. EXPERIMENTAL STUDY ON THE STRUCTURAL AND THERMAL

PERFORMANCE OF LOAD-BEARING WALL SYSTEMS
3.1. Introduction
A full scale experimental study was conducted in the Fire Research Laboratory of
Queensland University of Technology to evaluate the fire resistance of LSF wall
assemblies. A total of nine full scale fire tests were carried out by Kolarkar (2010).
Table 3.1 gives the details of the load-bearing LSF wall specimens tested by them
under a constant load of 15 kN/stud during the fire test and their fire ratings. His
results demonstrated the improved fire performance of LSF wall assemblies when
insulation was used externally between plasterboards instead of using in the cavity.
Table 3.1: Details of LSF Wall Specimens Tested by Kolarkar (2010)
Failure Time
Test Configuration Condition Insulation
(min.)
1(P) Ambient None -
2(P) Fire None 53
3(P) Fire None 111
4(P) Fire Glass Fibre 101

(Cavity Insulation)
5(P) Fire Rock Fibre 107

(Cavity Insulation)
6(P) Cellulose Fibre 110

Fire
(Cavity Insulation)
7(P) Fire Glass Fibre 181 (Unexpected

(External Insulation) furnace failure)
8(P) Fire Rock Fibre 136

9(P) Fire Cellulose Fibre 124

3-1
Experimental Study
In Kolarkar’s (2010) experimental study, Test 7(P) was affected by an unexpected

furnace failure and the results are not useable. Hence in this experimental study it
was decided to repeat Test 7(P) as Test 1. Even though many tests were conducted
by Kolarkar (2010), he did not consider the effect of load ratio. Hence two additional
tests were also conducted at a higher load ratio of 0.4 to study the effect of different
load ratios on the fire resistance of load-bearing LSF wall systems. A total of three
test specimens were built and tested. All three test walls were built with the
insulation sandwiched between the plasterboards on both sides of the steel wall
frame, as proposed by Kolarkar and Mahendran (2008). The insulations used were
glass and rock fibres. Test specimens were carefully built in a similar way to that of
Kolarkar (2010) for comparison purposes. Table 3.2 gives the details of the three full
scale load-bearing test wall specimens used in this study.
Table 3.2: Details of Test Specimen Configurations in the Current Study
Test Configuration Load Ratio External Insulation
0.2
1 Glass Fibre
(15 kN/stud)
0.4
2 Glass Fibre
(30 kN/stud)
0.4
3 Rock Fibre
(30 kN/stud)
This chapter presents the details of the experimental study, which was carried out to
investigate the thermal and structural performance of load-bearing steel wall
assemblies lined with external insulation under different load ratios. Details of the
results, including the temperatures and deflections measured during the tests, are
presented along with the stud failure modes. The results were also analyzed and
presented to identify the influence of different insulation materials. The failure time
variation in terms of load ratio is also discussed in this chapter.
3-2
Experimental Study
3.2. Test Specimens
All the steel wall frames used in the full scale load-bearing wall assemblies were
built to a height of 2400 mm and width of 2100 mm to represent a typical LSF wall
in a building. Four studs of lipped channel sections (90 x 40 x 15 x 1.15 mm) were
used at a spacing of 600 mm. These studs were attached to the top and bottom tracks
made of un-lipped channel sections (92 x 50 x 1.15 mm) using 12 mm long self
drilling wafer head screws (see Figures 3.1 (a) to (d)). All the studs and tracks used
were fabricated from G500 galvanized steel sheets having a nominal base metal
thickness (BMT) of 1.15 mm and a minimum specified yield strength of 500 MPa.
Their measured BMT and yield strengths were 1.15 mm and 569 MPa, respectively.
Track Interior Studs Track
Exterior Studs
(a) Before attaching any plasterboards
40 50
15
90 92
15
40 50 Stud Track Track Stud
t = 1.15 mm
(b) Stud (c) Track (d) Screws connecting stud and track
Figure 3.1: LSF Wall Frame and Connection Details
3-3
Experimental Study
K type thermocouple wires were installed to measure the temperature variations

across the wall (over the plasterboard and steel surfaces) and along the stud lengths.
Figures 3.2 (a) to (d) demonstrate the way the thermocouples were attached.
On the studs, the thermocouple wires were attached to the hot flange, web and cold
flange. These wire ends (hot junctions) were then pressed flat against the steel to
measure its surface temperature. The wires were taken to the ambient side through
tiny holes drilled in the unexposed plasterboards.
On the plasterboard surfaces the thermocouple wires were knotted in a way that it
would be in place while making the wall and during the fire test. Screws were used to
hold the thermocouple wires on to the cavity surfaces since there was a chance of
cavity wires getting pulled out during wall making.
(a) On the flange of stud (b) On the web and flanges of the stud
(c) Over the plasterboard (d) Inside the cavity
Figure 3.2: Thermocouple Fixings
3-4
Experimental Study
The plasterboards used were 1200 mm in width by 2400 mm in length with a

thickness of 16 mm and a mass of 13 kg/m2 (see Figure 3.3 (a)). These plasterboards
are manufactured by Boral Plasterboard under the product name of Fire-stop to suit
the requirements of Australian Standard AS/NZS 2588 (SA, 1998), and installed in
accordance with the requirements of AS/NZS 2589.1 (SA, 1997).
The base layer plasterboards were attached to the studs by 25 mm long drill point
screws (see Figure 3.3 (b)), which required much less effort than needle point screws
reducing the chance of stud distortion. The base layers of plasterboards were
installed vertically on either side of the steel frame with staggered vertical joints over
the centre line of stud flanges for proper fixing as shown in Figure 3.3 (c). These
screws were spaced at 200 mm centres along the plasterboard edges and 300 mm
centres along the intermediate studs in the field of the plasterboards. A minimum
edge distance of 10 - 15 mm was maintained for all the screws from plasterboard
edges.
Edge strips were used to create an external cavity to be filled with insulation as
shown in Figure 3.3 (d). The thickness of the insulation layer in all the specimens is
maintained at 25 mm. Hence two 13 mm thick plasterboard strips were used to form
the external cavity (see Figure 3.3 (e)).
The outer layer consisted of plasterboard sheets installed horizontally with the joint
at mid-height of the wall. The outer layer plasterboards were attached to the studs by
70 mm long self-drilling bugle head screws (see Figure 3.3 (b)) spaced at 300 mm
centres in the field of the plasterboard and penetrating the studs. The joints were
sealed with 50 mm wide reinforced paper tape and covered with two coats of joint
compound. To prevent any heat loss from the edge of the wall, the exterior studs
were covered with plasterboards and insulation material as shown in Figure 3.3 (f).
3-5
Experimental Study
Used to fix plasterboards
Used to fix frames
(a) Plasterboards (b) Screws
200 mm
200 mm
Edge strips
(c) Staggered screws along vertical (d) Edge strips to hold insulation
joints
Plasterboards
Insulation
(e) Insulation layer supported with (f) Layer of insulation next to

edge strips external studs
Figure 3.3: Plasterboard and Insulation Fixings
3-6
Experimental Study
3.2.1. Test Specimens 1 and 2

For these test specimens, the steel stud frame was lined by two layers of
plasterboards on both sides with an external insulation. This was achieved by placing
the insulation between the base and face layer plasterboards. Details of assembling
these specimens are given next.
The base layer of plasterboard on the fire side was attached first along with its
associated thermocouples to the steel frame as shown in Figure 3.4 (a). The vertical
joints were sealed next. Plasterboard strips of 40 mm and 80 mm in width and 13 mm
in thickness were then fixed to the base layer plasterboards on each side along the
perimeter of the wall, so as to generate a cavity to be later filled with external
insulation. Two layers of strips were used to obtain a total cavity depth of 26 mm
(see Figure 3.4 (b)). This was followed by the placing of a single layer of 25 mm
thick glass fibre mat in the cavity formed on the fire side of the wall along with
additional thermocouples to measure the temperatures on either side of the insulation
during the fire test as shown in Figure 3.4 (c). The face layer plasterboards were then
fixed and sealed on the fire side (sandwiching the insulation between the face and
base layer plasterboards) taking care to pass all the thermocouple wires onto the
ambient side of the wall through holes in the base layer of fire side plasterboards.
This layer was fixed using 70 mm long plasterboard screws with bugle heads (see
Figure 3.3 (b)) spaced at 300 mm centres in the field of plasterboard. As the screws
used were not of the self drilling type, it was necessary to pre-drill before using the
screws.
The wall was then turned upside down in order to fix the ambient side plasterboards
as shown in Figure 3.4 (d). The base layer plasterboard of the ambient side was fixed
along with its thermocouples to the steel frame thus closing the wall cavity. Before
fixing this plasterboard, all the thermocouples attached to the base layer plasterboard
on the fire side were carefully passed on to the ambient side through small holes
drilled at appropriate locations in the base layer plasterboard on the ambient side (see
Figures 3.4 (e) and (f)). This was followed by fixing of edge strips and insulation on
the ambient side as for the fire side (see Figure 3.4 (g)). Finally the outer layer
ambient side plasterboards were fixed as for the fire side (see Figure 3.4 (h)).
3-7
Experimental Study
Plasterboard piece (750 x 2400) Full plasterboard (1200 x 2400)
Vertical joints Plasterboard piece (150 x 2400)
(a) After fixing the fire side base layer
Edge strips Sealed vertical joints
Thermocouples (Ins-Pb2)
(b) Edge strips were placed to hold insulation
Glass fibre insulation Thermocouples (Pb1-Ins)
(c) After placing the fire side glass fibre insulation
Figure 3.4: Wall Making Process of Test Specimens 1 and 2
3-8
Experimental Study
Track projecting
150 mm each side
Thermocouples on studs Fire side thermocouples
(d) Ready to fix ambient side plasterboards
Thermocouples on interior studs
Thermocouples on exterior studs
(e) Thermocouples were pulled out to the ambient side
(f) After fixing the ambient side base layer
3-9
Experimental Study
Washers to hold insulation
(g) After placing ambient side glass fibre insulation
Horizontal joint
Full plasterboards (1200 x 2400)
(h) After fixing the ambient side external layer
3-10
Experimental Study
3.2.2. Test Specimen 3
The construction of Test Specimen 3 was identical to that of Test Specimen 2 in all
respects, except for rock fibre insulation used as external insulation. The insulation
layers placed on the fire and ambient sides are shown in Figures 3.5 (a) and (b),
respectively.
(a) After placing the fire side Rock fibre insulation
(b) After placing the ambient side Rock fibre insulation
Figure 3.5: Wall Making Process of Test Specimen 3
3-11
Experimental Study
3.3. Test Set-up and Procedure
3.3.1. Gas Furnace

A propane fired gas furnace used by Kolarkar and Mahendran (2008) was used to
conduct the fire tests of wall specimens. The furnace has internal dimensions of 2.1
m width, 2.4 m height and 0.3 m depth. The front face of the furnace showing all the
six burners can be seen in Figures 3.6 (a) and (b).
The furnace was mounted on a carriage so that it could roll on wheels. To start the
test the carriage was moved forward to make contact with the frame holding the test
wall specimen, thereby completing the combustion chamber. On starting the furnace
the wall was exposed to heat from one side as desired. The firing system protects the
wall from any localized flame impingement and ensures a more uniform distribution
of the temperature over the wall surface mostly by radiation.
The furnace was designed to deliver heat in accordance with AS 1530.4 (SA, 2005)
as given by the following equation
Tt – To = 345 log10 (8t+1)
where t Elapsed time in minutes

Tt Furnace temperature (oC) at time t
To Ambient temperature (oC) at the start of the test
Four observation ports were provided on the rear side of the furnace for observing
the structural response of test walls. These were helpful to observe the plasterboard
deterioration and their fall off time.
The specimen holder containing the wall assembly was sealed against the furnace in
order to maintain the furnace pressure by at least 2 Pa greater than the atmospheric
pressure over the top two thirds of the wall specimen. The positive pressure helped in
preventing the drawing of outside cold air into the combustion chamber.
3-12
Experimental Study
Deep universal beam Burners Loading frame
Furnace
RHS
Universal beam Universal column
(a) Front view
Furnace controls Furnace
Loading frame
(b) Side view
Figure 3.6: Furnace and Loading Frame
3-13
Experimental Study
3.3.2. Loading Arrangement

The loading frame essentially consisted of two columns firmly bolted to the ground
and a universal beam (UB) connecting the two columns to form an ‘H’ shaped portal
frame. A second universal beam (UB) was bolted to the floor (see Figure 3.6). Four
jacks were mounted on this UB at a spacing of 600 mm.
The loading frame was specially designed to load the individual studs of a wall
specimen directly from the bottom side. The shafts of the jack were co-axially guided
through a hollow sleeve running across a rectangular hollow section (RHS) attached
between the two columns and fixed parallel to the second UB. The RHS supporting
the hollow sleeve ensured a vertical displacement of the jack without any angular
movement (see Figure 3.7 (a)).
The test wall specimen was mounted in a manner to ensure that the centroids of the
studs lined up with the centroids of the loading plates. This was possible as the
spacing of the jacks was identical to the spacing of the studs (ie. 600 mm). A special
arrangement was also made using a sliding plate to facilitate movement of any
individual jack along with its shaft and sleeve to the extent of 20 mm on either side
so as to achieve a better accuracy in lining up of the centroids of the loading plate
and the studs (see Figure 3.7 (b)).
The loading frame was built such that when the test specimen was mounted inside
the frame, the bottom track was resting on the four loading plates and the upper track
would be pressing against the top UB. The upper track is then firmly clamped to the
UB at the top to give additional support (see Figure 3.7 (c)).
The verticality of the wall was checked at different locations over the wall as shown
in Figure 3.7 (d). The jacks were then connected to a single hydraulic pump (see
Figure 3.7 (e)). Use of a single pump ensured equal loading on all four studs as the
same hydraulic pressure operated these jacks. A pressure transducer was attached to
the pump to obtain the load applied to the studs by the jacks.
3-14
Experimental Study
Loading Sleeve
plate
RHS
Jack Shaft
(a) Loading plate (b) Adjustable shaft and sleeve
Top track Wall
(c) Frame clamped to the top UB (d) Checking the verticality
Loading plates
Jack Pump
(e) Hydraulic jacks connected to pump
Figure 3.7: Loading System
3-15
Experimental Study
3.3.3. Displacement Measurement

In order to measure the out-of-plane movements of the wall specimen during the test,
many Linear Variable Displacement Transducers (LVDT) were used (see Figure 3.8
(a)). They were placed at 0.25H, 0.50H and 0.75H (where ‘H’ is the height of the test
specimen) of each stud. The transducers were attached to wooden beams that were
firmly connected to the columns. A total of 12 LVDTs were placed to measure the
lateral deflection of four studs. However, in the analysis of results only eight LVDTs
were used (For interior studs: Top, middle and lower level, for exterior studs: only at
middle level).
In order to measure the axial shortening of studs, four LVDTs were used with one
LVDT placed under the loading plate as shown in Figure 3.8 (b). The LVDTs were
fixed as close as possible to the studs to minimize experimental errors.
LVDT
Wooden
beam
(a) Lateral deflection measurement
Thermocouple wires
LVDT
(b) Axial deformation measurement
Figure 3.8: Deflection Measurements Using LVDTs
3-16
Experimental Study
3.3.4. Temperature Measurement

K type thermocouples were used to measure the temperature development across the
wall specimens. The stud temperatures were measured at three levels for interior
studs, namely at 0.25 H, 0.50 H and 0.75 H. At each level three thermocouples were
attached per stud to measure the temperatures of the hot flange, web and cold flange,
thus giving a total of nine thermocouples per stud. For exterior studs they were
attached only at 0.5 H. Therefore a total of 24 thermocouples per frame were used to
measure the temperatures of studs. These thermocouples allowed the determination
of average stud temperature and the temperature gradient of the stud across its cross-
section and length.
To measure the temperatures across the wall, three additional sets of thermocouples
were attached at the mid-height of the assembly between the studs. Each set was used
to measure the temperatures of seven locations across the wall as shown in Figure
3.9. These are the surface of fire side face layer (FS), surface between face
plasterboard layer and insulation layer (Pb1-Ins), surface between insulation and fire
side base plasterboard layer (Ins-Pb2), cavity surface of the fire side base
plasterboard layer (Pb2-Cav), cavity surface of the ambient side base plasterboard
layer (Pb3-Cav), surface between ambient side base plasterboard layer and
insulation layer (Pb3-Ins) and the surface between the insulation layer and ambient
side face plasterboard layer (Ins-Pb4).
To measure the average temperature on the unexposed side of the wall, five
thermocouples were positioned on this surface, one at the centre of the area and one
at the centre of each quarter section as mentioned in AS 1530.4 (SA, 2005). The
temperatures measured by these thermocouples indicated the level of heat
penetration across the specimens. To measure the temperature at other points on the
ambient side an infrared gun was used.
In order to measure the furnace temperature, four K – type furnace thermocouples

were symmetrically placed inside the furnace chamber within a vertical plane 100
mm from the exposed surface of the specimen.
3-17
Experimental Study
Fire exposed side

Pb1
Ins.
Pb2
Stud 1 Stud 2 Stud 3 Stud 4
Pb3
Ins.
Pb4
Ambient side
Figure 3.9: Thermocouple Locations
3.4. Test Procedure

First the specimen was installed inside the loading frame. The studs were centered
over the individual jacks and the wall was checked for its verticality using spirit
levels. After properly positioning the wall, the top track was fastened to the top beam
using G - clamps on either side. This was done to retain the correct positioning of the
wall during testing. Figures 3.10 (a) to (c) show Test Specimens 1, 2 and 3,
respectively, before the load was applied.
A load ratio is the ratio of the applied load in the fire test to the collapse load
determined by the ambient temperature test. A load ratio of 0.2 was chosen for Test
Specimen 1 while it was 0.4 for Test Specimens 2 and 3. Hence 15 kN per stud for
Test Specimen 1 and 30 kN per stud for Test Specimens 2 and 3 were applied
gradually at a constant rate by the hydraulic jacks. For all the specimens the applied
load was reached within 15 minutes. Initially the studs were loaded simultaneously to
about 50% of the target applied load and then unloaded, to remove any residual
strains and initial slackness, which could be present in the system during the
assembly of the wall specimen. The studs were then reloaded in increments of 2 kN.
Load and displacement readings were recorded by the Edcar software at the end of
each load increment. Figure 3.10 (d) shows the side view of the test setup before the
furnace was moved close to the specimen.
The furnace was then rolled forward to close the combustion chamber of the furnace
with the wall forming the fourth side of the chamber facing the burners. This
arrangement ensured that only one face of the test specimen was exposed to elevated
temperatures. The width of the wall was designed to be less than the width of the
3-18
Experimental Study
furnace opening by 20 mm such that there would be a gap of 10 mm on either side of

the wall, to ensure free vertical edges. The gaps were packed with Isowool, a non-
restraining and non-combustible mineral fibre such that the lateral displacement of
the wall would not be restricted due to frictional forces.
The target applied load was held constant at room temperature for a few minutes
before the furnace was started and was maintained throughout the fire test. This
ensured the free vertical expansion of the wall when exposed to elevated
temperatures. During the fire test, the furnace temperature was regulated such that
the average temperature recorded by the control thermocouples inside the furnace
followed the standard cellulosic temperature-time fire curve in accordance with AS
1530.4 (SA, 2005). During the fire test the vertical and lateral displacements of the
wall, and the temperature readings from all the thermocouples were taken at intervals
of 1 minute. The test specimen was considered to have failed when the oil pressure in
the jacks could not be maintained. This was also confirmed via the Edcar load-
displacement graph, which showed a rapid unloading phase. The test was stopped
immediately following the failure of the wall and the time to failure was then
recorded.
Top level LVDTs
Middle level
LVDTs
Lower level
LVDTs
LVDTs to measure
axial deformation
(a) Test Specimen 1
Figure 3.10: Test specimens before Testing
3-19
Experimental Study
Deep UB beam
(b) Test Specimen 2 (c) Test Specimen 3
Furnace Loading
frame
Data logger Pump Wall specimen
(d) Side view

Figure 3.10: Test Specimens before Testing
3-20
Experimental Study
3.5. Observations and Results
3.5.1. Test Specimen 1 (External Insulation - Glass Fibre; Load Ratio 0.2)
Each stud of the wall specimen was loaded in compression to 15 kN, (total load 60
kN) and the wall specimen was then subjected to the standard fire test. The test was
stopped after 118 minutes since the wall specimen could not support the applied load
of 15 kN after this time. The test wall specimen failure was due to the structural
collapse of studs before any insulation or integration failure occurred.
3.5.1.1. Visual Observations

The test specimen showed no signs of lateral displacement during the application of
the axial compression load to each stud. After 3 minutes from starting the furnace,
smoke was escaping from the top of the wall specimen. This was probably due to the
burning of the plasterboard paper on the exposed surface. After about 10 minutes,
more smoke and steam were seen to escape from the top corners of the wall
specimen. The presence of steam in the mixture of gases was confirmed by the
observation of water particles on the loading frame after condensation.
From the beginning of the fire test, the wall specimen was observed to be bending
towards the furnace. However, near the failure, the lateral deflection started to
reverse its direction and finally the studs bent outwards and away from the furnace,
resulting in the breaking of plasterboards (see Figure 3.11 (a)).
Parts of the exposed plasterboards (Pb1 and Pb2) were seen to have fallen off near
the centre of the wall specimen exposing the Pb3-Cav surface (see Figure 3.11 (b)).
However, the temperature measurements indicate that Pb2 must have fallen off due
to increasing lateral deflection at the failure point. The glass fibre insulation used
between Plasterboards 1 and 2 had completely disappeared leaving only some molten
glass traces. The base layer plasterboard on the fire side was seen to have shrunk.
This lead to the opening of plasterboard joints by 10 -15 mm exposing the studs. The
front view shows that the flexural buckling about the minor axis and the torsional
buckling of studs were fully prevented by the lateral support offered by the
plasterboards throughout the test. The exposed plasterboards were stripped off and
the debris removed to expose the frame (see Figure 3.11 (c)). This view shows the
3-21
Experimental Study
presence of local buckling waves along the studs, confirming the occurrence of local
buckling of studs before the ultimate failure.
(a) Cracks formed in ambient side plasterboards
Stud 1
Stud 2
Stud 3
Stud 4
(b) Partial collapse of exposed (c) After removing the exposed

plasterboards plasterboards and remains of
external insulation
Figure 3.11: Test Specimen 1 after the Fire Test
3-22
Experimental Study
Local buckling waves were seen in the web, hot flange, hot lip and cold lip of Stud 1
(see Figure 3.12 (a)). Stud 2 showed inelastic local buckling of the entire cross-
section (web, hot flange and lip, cold flange and lip) close to the mid-height
suggesting the complete mobilization of the capacity of the cross-section (see Figure
3.12 (b)). Stud 3 behaved in the same manner as Stud 2 (see Figure 3.12 (c)).
However, Stud 4 was seen to display local buckling of only the web element near
mid-height (see Figure 3.12 (d)). This observation also confirms that the local
buckling occurred first, which led to the failure of studs.
(a) Stud 1 (b) Stud 2
(c) Stud 3 (d) Stud 4

Figure 3.12: Failure of Studs after the Fire Test
3-23
Experimental Study
3.5.1.2. Time - Temperature Profiles
Plasterboard surfaces (see Figure 3.13)
Average temperature of the interface surface between the exposed Plasterboard

1 and insulation (Pb1-Ins)
The temperature profile of this surface showed a rapid rise in temperature within 4
minutes of starting the test (first phase) reaching a temperature of about 82oC by the
end of 5 minutes. Beyond 5 minutes the second phase started with the temperature
increasing gradually to about 100oC by the end of 13 minutes. This was followed by
a very rapid rise in temperature (third phase) crossing 400oC in 25 minutes and
640oC in 35 minutes. The sudden increase in the temperature of the interface
between the exposed plasterboards and insulation was probably caused by the heat
blocked and redirected by the adjoining layer of insulation.
Beyond 35 minutes the temperature gradient started to reduce (fourth phase) with the
temperature crossing 700oC by 45 minutes and 752oC by 53 minutes. This is because
the insulation has started to melt by this time and the heat was used for this chemical
process rather than for the temperature rise. A rise in temperature at about 54 minutes
indicates that the melting process was over by this time. The Pb1-Ins curve between
35 and 54 minutes was different from the curve, where the insulation was used inside
the cavity (Kolarkar, 2010).
From here onwards this curve approached the fire side temperature curve gradually.
Visual observations and careful analyses of individual thermocouple measurements
revealed that the left side plasterboard pieces (Pb1) have fallen off at 86 minutes.
Similarly the middle plasterboard pieces (Pb1) have fallen off at 106 minutes.
However, the right side plasterboard pieces (Pb1) appears to have fallen off in steps
at 86 and 106 minutes. The average temperature of the interface was about 1070oC
by the end of the test.
3-24
Experimental Study
Average temperature of the interface surface between the insulation and

exposed base layer Plasterboard 2 (Ins-Pb2)
This interface responded to the initial rise in temperature (first phase) for 6 minutes
from the start of the furnace and reached a temperature of approximately 74oC
rapidly by around 8 minutes and then remained constant (second phase) up to about
25 minutes. Beyond this time the third phase started with the temperature rising
almost linearly with respect to time reaching 775oC by about 54 minutes. Following
this, this curve approached the fire side temperature curve. A closer look at the
individual thermocouples revealed that Ins-Pb2-R thermocouple recorded very rapid
temperature gains and intersected the Pb1-Ins-R curve at 44 minutes indicating the
complete burning of the insulation at mid-height between Studs 3 and 4. The
intersection of curves Pb1-Ins-M and Ins-Pb2-M at about 47 minutes indicates the
burning out of the insulation at mid-height of the wall specimen between Studs 2 and
3.
Average temperature on the cavity facing surface of the exposed Plasterboard 2

(Pb2-Cav)
The initial rise in temperature of this surface was followed by a plateau extending up
to 52 minutes. A rapid rise in temperature was noticed at 53 minutes, reaching 255oC
by 61 minutes. This confirms the insulation burn out time, which was about 54
minutes after the furnace was started. Individual analysis of temperature
measurements revealed that the rapid temperature rise of the middle thermocouple
started at 52 minutes and this confirms the middle insulation burning out time of 47
minutes. The rapid temperature rise of the right side thermocouple started at 50
minutes and this confirms the right side insulation burning out time of 44 minutes.
Similarly the rapid temperature rise of the left side thermocouple started at 50
minutes and this indicates that the left side insulation burning out time is nearly 50
minutes. One may argue that the rapid rise of Pb2-Cav temperature curve was just a
natural process of plasterboard after 120oC and this has no connection with insulation
burning time. However, a closer look of Pb1-Ins shows that the rapid rise continued
up to about 50 minutes until it was very close to the fireside (FS) temperature. On the
other hand Pb2-Cav’s rapid increase was stopped after 10 minutes and then it
3-25
Experimental Study
continued with a gradual increase. This behavior indicates that the rapid temperature
rise of the Pb2-Cav is due to the insulation burn out and not because of its natural
process. The third phase started beyond 61 minutes with the temperature increasing
gradually and crossing 595oC by the end of the test. A temperature difference of
approximately 450oC was present across the thickness of Plasterboard 2 by the end of
the test, indicating the sustained integrity of the exposed base plasterboard layer until
the failure of the test specimen. It appears to have provided sufficient lateral restraint
to the studs as the studs did not suffer from torsional buckling or flexural buckling
about the minor axis.
Average temperature on the cavity facing surface of the ambient Plasterboard 3

(Pb3-Cav)
The transmission of heat across the cavity was almost instantaneous due to radiation
making the temperature profile of Pb3-Cav surface follow very closely but just on
the underside of the Pb2-Cav surface profile. At failure, the temperature difference
between Pb2-Cav and Pb3-Cav was 40oC.
Average temperature of the interface surface between base layer Plasterboard 3

and insulation (Pb3-Ins)
The temperature of this surface increased gradually reaching around 133oC and
remained almost constant up to 107 minutes. The third phase started beyond this time
with the temperature increasing linearly with respect to time reaching 314oC by the
failure time. This may be because all the first layer plasterboards have fallen off by
106 minutes.

Plasterboard 4 (Ins-Pb4)
This surface was well protected from the fire and remained under 100oC up to 117
minutes from the start of the test. The temperature was just above 110oC at the end of
the fire test.
3-26
Experimental Study
Average temperature on the ambient side of unexposed Plasterboard 4
The temperature on the unexposed face of the wall specimen remained under 70oC
(well below the insulation failure temperature) throughout the fire test.
1200
1000
Temperature ( o C)
800
600
400
200
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min)
AS 1530.4 Furnace FS Pb1-Ins Ins-Pb2

Pb2-Cav Pb3-Cav Pb3-Ins Ins-Pb4 AS
(a) Average Time - Temperature profiles of plasterboard surfaces
1200
1000
Temperature ( o C)
800
600
400
200
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min)
AS 1530.4 Furnace FS-L FS-M FS-R Pb1-Ins-L
Pb1-Ins-M Pb1-Ins-R Ins-Pb2-L Ins-Pb2-M Ins-Pb2-R Pb2-Cav-L
Pb2-Cav-M Pb2-Cav-R Pb3-Cav-L Pb3-Cav-M Pb3-Cav-R Pb3-Ins-L
Pb3-Ins-M Pb3-Ins-R Ins-Pb4-L Ins-Pb4-M Ins-Pb4-R
(b) Time -Temperature profiles across the left, middle and right sections of
plasterboard surfaces
Figure 3.13: Time - Temperature Plots of Plasterboard Surfaces in

Test Specimen 1
3-27
Experimental Study
Steel surfaces: Temperature variation of hot flange, web and cold flange
Figures 3.14 (a) to (c) show the time-temperature profiles of hot flanges, webs and
cold flanges of the central studs (Studs 2 and 3). The temperatures along the length
were seen to be almost uniform in both studs until the insulation was burnt (about 50
minutes). However, from there onwards the mid-height stud readings increased
rapidly than the top stud. Due to thermal bowing, the mid-height stud must have
moved closer to the furnace, resulting in higher temperature readings at mid-height.
The top hot flange reading gradually approached the middle hot flange reading and
finally merged with it due to the heat conducted along the length of the stud. Figure
3.15 gives the temperature variation across the central stud cross-sections at mid-
height. The hot flange temperature of the third stud at mid-height increased rapidly
from 45 minutes which was soon after the right side glass fibre was burnt. In a
similar way the hot flange temperature of the second stud at mid-height increased
rapidly from 48 minutes. The hot flange temperature of the third stud showed further
steep rise after 107 minutes which was soon after the first layer plasterboard had
fallen off. At failure the hot flanges of Studs 2 and 3 at mid-height were recorded as
582oC and 664oC. Stud 3 experienced higher temperatures at failure due to the
vertical joint placed against it.
700
600
500
Temperature ( oC)
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min)
S2-Top S3-Top S2-Middle S3-Middle
(a) Time - Temperature profiles of hot flange surfaces of central studs

Central Studs in Test Specimen 1
3-28
Experimental Study
600
500
Temperature ( o C)
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min)
(b) Time - Temperature profiles of web surfaces of central studs
600
500
Temperature ( o C)
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min)
(c) Time-Temperature profiles of cold flange surfaces of central studs

3-29
Experimental Study
700
600
500
Temperature ( o C)
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min)
S2-Hot Flange S3-Hot Flange S2-Web

S3-Web S2-Cold Flange S3-Cold Flange
Figure 3.15: Time - Temperature Profiles across Central Studs at Mid-height in

Test Specimen 1
Average temperature of the steel studs and the temperature variation across
stud depth
The variation of the average temperature and the temperature difference between hot
and cold flanges are shown in Figures 3.16 and 3.17, respectively. The average
temperature was near ambient temperature for about 8 minutes followed by a gradual
rise up to 47 minutes. By this time all the mid-level insulation was melted, the curve
showed a steep rise up to 75 minutes. This was followed by a steeper curve, which
was after the entire insulation was burnt out.
The initial temperature difference across Stud 3 was nearly 20oC. A rapid increase
was noticed after 44 minutes which was soon after the right side insulation was
burnt. Similarly the insulation at middle and left sides burnt at 47 and 50 minutes,
respectively, and this was followed by a rapid increase in temperature difference
across Stud 2. The Stud 3 curve was above the Stud 2 curve due to the presence of a
vertical joint along Stud 3. These studs reached a temperature difference of 150oC by
71 minutes. This temperature difference was nearly maintained up to about 30
minutes and at failure the temperature difference across Stud 3 showed a sudden
increase while that of Stud 2 was dropping. The reduction of temperature difference
3-30
Experimental Study
between hot and cold flanges is due to the heat conducted across the stud. However,
the sudden increase of temperature difference across Stud 3 is explained by the rapid
increase in hot flange temperature of Stud 3 near failure due to the vertical joint of
Pb2. The temperature differences across Stud 2 and 3 near failure were 71 and
174oC, respectively.
600
500
Temperature ( o C)
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min)
Stud 2 Stud 3
Figure 3.16: Average Temperature Profiles across Central Studs at Mid-height

in Test Specimen 1
180
160
140
Temperature ( o C)
120
100
80
60
40
20
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min)
Stud 2 Stud 3
Figure 3.17: Temperature Difference across Central Studs at Mid-height

in Test Specimen 1
3-31
Experimental Study
3.5.1.3. Behaviour of Wall Specimen
The studs when loaded at ambient temperature up to the target load (15 kN/stud)
showed an average axial shortening of 2 mm while it was slightly more than 3 mm
for Stud 3 (see Figure 3.18 (a)). The lateral deformation was negligibly small, which
indicates that the eccentricity of loading about the major axis was negligibly small
before the furnace was started. On starting the furnace the wall was observed to
expand gently up to 75 minutes. Thereafter the thermal expansion was accelerated
due to the rapid increase in average stud temperature. Maximum deformation of 10
mm expansion was reached at failure as shown in Figure 3.18 (b).
After the furnace was started the wall bent towards the furnace with Studs 2 and 3
reaching a maximum central deflection of 21 and 19 mm at 96 and 99 minutes,
respectively (see Figure 3.19). Beyond which both of them reversed and deformed
very rapidly in the outward direction leading to failure of the wall specimen at 118
minutes.
The lateral deflection of the wall was influenced by the placement of vertical joint in
addition to the temperature difference across the studs. The stud which was attached
to the full plasterboard had screws at 300 mm spacing while the stud having the
vertical joint had two sets of staggered screws at 200 mm spacing, making the
effective spacing between screws as 100 mm. In this test, the vertical joint was
placed against Stud 3. Hence the expansion of the hotter side was restricted
compared to Stud 2. This resulted in less lateral deflection of Stud 3 even though it
had a higher temperature difference across the stud.
Figure 3.20 shows the average load applied to the studs. After 118 minutes the load
could not be maintained further. Hence the furnace was stopped and the wall
specimen was considered to have failed.
3-32
Experimental Study
0.00
-0.50
Axial Deformation (mm)
-1.00
-1.50
-2.00
-2.50
-3.00
-3.50
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00
Load (kN)
(a) Axial deformation - Load profiles at ambient temperature
12.00
10.00
8.00
6.00
4.00
2.00
0.00
-2.00
-4.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min)
(b) Axial deformation - Time profiles at elevated temperatures
Figure 3.18: Axial Deformation Profiles for Studs of Test Specimen 1
3-33
Experimental Study
0.00
-5.00
Lateral Deflection (mm)
-10.00
-15.00
-20.00
-25.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min)
Stud 2 Stud 3 Stud 4
(a) Lateral deflection - Time profiles at upper level
0.00
-5.00
-10.00
-15.00
-20.00
-25.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min)
(b) Lateral deflection - Time profiles at middle level
Figure 3.19: Lateral Deflection - Time Plots of Test Specimen 1
3-34
Experimental Study
Lateral Deflection (mm) 0.00
-4.00
-8.00
-12.00
-16.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min)
(c) Lateral deflection - Time profiles at lower level

Figure 3.19: Lateral Deflection - Time Plots of Test Specimen 1
16.00
15.50
15.00
Load (kN)
14.50
14.00
13.50
13.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min)
Figure 3.20: Axial Compression Load - Time Profile of Test Specimen 1
3-35
Experimental Study
The temperatures of hot and cold flanges, temperature difference across the stud
cross-section and the lateral deflection measurements at regular intervals are shown
in Table 3.3.
Table 3.3: Temperature and Lateral Deflection Measurements for

Test Specimen 1
Hot Flange (HF) Cold Flange (CF) HF - CF Lateral Deflection

o o o
Time ( C) ( C) ( C) (mm)
(min)
Stud 2 Stud 3 Stud 2 Stud 3 Stud 2 Stud 3 Stud 2 Stud 3
30 70 70 55 54 15 16 8 7
60 218 261 109 128 109 133 11 11
90 432 424 281 269 151 155 20 19
118 582 664 511 490 71 174 21* 19*
* - indicates the maximum values obtained from the test and not the value at failure time,
because a sudden reversal occurred near the failure point.
3-36
Experimental Study
3.5.2. Test Specimen 2 (External Insulation - Glass Fibre; Load Ratio 0.4)
Each stud of the wall specimen was loaded in compression up to 30 kN (total load
120 kN), and the wall specimen was then subjected to the standard fire test. The test
was stopped after 108 minutes since the wall specimen could not support the applied
load of 30 kN after this time. The test wall specimen failure was due to the structural

The behaviour of Test Specimen 2 was very similar to that of Test Specimen 1. It
showed no signs of lateral displacement during the initial application of the axial
compression load to each stud. After 4 minutes from starting the furnace, smoke was
escaping from the top of the wall specimen. This was probably due to the burning of
the plasterboard paper on the exposed surface. After about 12 minutes, more smoke
and steam were seen to escape from the top corners of the wall specimen. The
presence of steam in the mixture of gases was confirmed by the observation of water
particles on the loading frame after condensation as for Test Specimen 1.
From the beginning of the fire test, the wall was observed to be bending towards the
furnace. However, near the failure, the lateral deflection started to reverse its
direction and finally the studs bent outwards and away from the furnace, resulting in
the breaking of plasterboards (see Figure 3.21 (a)).
The exposed plasterboards were seen to have fallen off near the centre of the wall
specimen exposing the Pb2 (see Figure 3.21 (b)). The glass fibre insulation used
between Plasterboards 1 and 2 had completely disappeared leaving only some molten
glass traces. The base layer plasterboard on the fire side was seen to have shrunk.
This led to the opening of plasterboard joints by 10 -15 mm and thus exposing the
studs. The flexural buckling about the minor axis and the torsional buckling of studs
were fully prevented by the lateral support offered by the plasterboards. The exposed
plasterboards were stripped off and the debris removed to expose the frame (see
Figure 3.21 (c)). This view shows the influence of Stud 3 in the wall failure.
3-37
Experimental Study
(a) Cracks in ambient side plasterboards
Stud 1
Stud 2
Stud 3
Stud 4

external insulation

3-38
Experimental Study
Studs 1 and 4 did not fail in Test Specimen 2 and nor experienced any local buckling
(see Figures 3.22 (a) and (d)). Stud 2 showed local buckling of the entire cross-
section other than its cold flange close to the mid-height, suggesting the complete
mobilization of the capacity of the cross-section (see Figure 3.22 (b)). Local waves
were observed in the web even away from mid-height. Stud 3 bent severely about the
major axis and showed local buckling of the entire cross-section near mid-height (see
Figure 3.22 (c)). Local buckling waves were observed in the web even away from
mid-height. This observation also confirms that the local buckling of studs occurred
first, which led to their ultimate failure.
(a) Stud 1 (b) Stud 2
(c) Stud 3 (d) Stud 4

3-39
Experimental Study

The only difference between Test Specimens 1 and 2 was in the applied load.
Therefore the measured Time - Temperature profiles were very similar as expected.
Average temperature of the interface surface between the exposed

Plasterboards 1 and insulation (Pb1-Ins)
The temperature profile of this surface showed a rapid rise in temperature within 4
minutes of starting the test (first phase) reaching a temperature of about 90oC by the
end of 6 minutes. Beyond 6 minutes the second phase started with the temperature
increasing gradually to about 100oC by the end of 11 minutes. This was followed by
a very rapid rise in temperature (third phase) crossing 400oC in 25 minutes and
690oC in 34 minutes. The sudden increase in the temperature of the interface
between the exposed plasterboards and insulation was probably caused by the heat
blocked and redirected by the adjoining layer of insulation similar to that of Test
Specimen 1.
temperature crossing 770oC by 51 minutes. This is because the insulation has started
to melt by this time and the heat was used for this chemical process rather than for
the temperature rise. A rise in temperature at about 51 minutes indicates that the
melting process was over by this time. The Pb1-Ins curve between 35 and 51 minutes
was different from the curve, where the insulation was used inside the cavity
(Kolarkar, 2010).
From here onwards this curve approached the fire side temperature curve. Visual
observations and careful analyses of individual thermocouple measurements revealed
that the left side plasterboard pieces have fallen off at 88 minutes. Similarly the
middle plasterboard pieces have fallen off at 85 minutes. However, the right side
plasterboard appears to have fallen off in steps at 89 and 105 minutes. The average
temperature of the interface was about 1052oC by the end of the test.
3-40
Experimental Study

This interface responded to the initial rise in temperature (first phase) for 6 minutes
from the start of the furnace and reached a temperature of approximately 77oC and
then remained constant (second phase) up to about 24 minutes. Beyond this time the
third phase started with the temperature rising almost linearly with respect to time
reaching 714oC by about 53 minutes. Following this, this curve approached the fire
side temperature curve.
A closer look at the individual thermocouples revealed that Ins-Pb2-L thermocouple

recorded very rapid temperature gains and intersected the Pb1-Ins-L curve at 56
minutes indicating the complete burning of the insulation at mid-height between
Studs 1 and 2. The intersection of curves Pb1-Ins-M and Ins-Pb2-M at about 52
minutes indicates the burning out of the insulation at mid-height of the wall between
Studs 2 and 3. According to Pb2-Cav temperature profile, the right side insulation
was burnt at about 48 minutes.

(Pb2-Cav)
to 55 minutes. A rapid rise in temperature was noticed at 56 minutes, reaching 217oC
by 63 minutes.
Individual analysis of temperature measurements revealed that the rapid temperature

rise of the left side thermocouple started at 60 minutes and this confirms the left side
insulation burning out time of 56 minutes. The rapid temperature rise of the middle
thermocouple started at 57 minutes and this confirms the middle insulation burning
out time of 52 minutes. Similarly the rapid temperature rise of the right side
thermocouple started at 52 minutes and this indicate that the right side insulation
burning out time is nearly 48 minutes. This is similar to that of Test Specimen 1.
3-41
Experimental Study
The third phase started beyond 63 minutes with the temperature increasing gradually
and crossing 497oC by the end of the test. A temperature difference of approximately
550oC was present across the thickness of Plasterboard 2 by the end of the test
indicating the sustained integrity of the exposed base plasterboard layer until the
failure of the specimen, i.e. Plasterboard 2 did not fall off although it would have
softened. However, it appears to have provided sufficient lateral restraint to the studs
as the studs did not suffer from torsional buckling or flexural buckling about the
minor axis as in Test 1.

(Pb3-Cav)
The transmission of heat across the cavity was almost instantaneous due to radiation
making the temperature profile of Pb3-Cav surface follow very closely but just on
the underside of the Pb2-Cav surface profile. At failure, the temperature difference

The temperature of this surface was at ambient temperature for the first 12 minutes.
It then increased gradually reaching about 124oC by the failure time.

This surface was well protected from the fire and remained under 100oC until failure.
3-42
Experimental Study
1200
1000
Temperature ( o C)
800
600
400
200
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min)

1200
1000
Temperature ( o C)
800
600
400
200
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min)
(b) Time - Temperature profiles across the left, middle and right sections of
Figure 3.23: Time - Temperature Plots of Plasterboard Surfaces in

Test Specimen 2
3-43
Experimental Study
cold flanges of the central studs (Studs 2 and 3). The temperatures along the length
were seen to be almost uniform in both studs until the insulation was burnt.
Thereafter the mid-height stud temperature values were higher than the top and lower
stud readings due to thermal bowing effect. However, the temperature values along
the stud merged gradually due to heat conducted across the studs. Figure 3.25 gives
the temperature variation across the central stud cross-sections at mid-height. The hot
flange temperature of the third stud at mid-height increased rapidly from 43 minutes
which was soon after the right side glass fibre was burnt. In a similar way the hot
flange temperature of the second stud at mid-height increased rapidly from 55
minutes. Temperature of the hot flange of third stud showed further steep rise after
105 minutes which was soon after the first layer plasterboard has fallen off. The top
hot flange reading gradually approached the middle hot flange reading and finally
merged with it due to the heat conducted along the length of the stud. At failure the
temperatures of middle hot flanges of Studs 2 and 3 were recorded as 505oC and
554oC. Stud 3 experienced higher temperatures at failure due to the vertical joint
placed against it.
600
500
Temperature ( o C)
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min)
S2-Top S3-Top S2-Middle S3-Middle S2-Lower S3-Lower

3-44
Experimental Study
500
400
Temperature ( o C)
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min)
400
300
Temperature ( o C)
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min)
(c) Time - Temperature profiles of cold flange surfaces of central studs

3-45
Experimental Study
600
500
Temperature (oC)
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min)
Figure 3.25: Time - Temperature Profiles across the Central Studs at Mid-
height in Test Specimen 2
stud depth
temperature was near ambient temperature for about 6 minutes followed by a steeper
curve, which was after the entire insulation was burnt out.
The initial temperature difference across Stud 3 was nearly 20oC up to 41 minutes
and then a rapid increase was noticed after this reaching 164oC at 72 minutes.
Similarly Stud 2 showed a rapid increase in temperature difference at about 50
minutes. Stud 3 was above 150oC up to 88 minutes at which time the left
plasterboard had fallen off and hence Stud 2 overtook Stud 3 to reach a maximum
temperature difference of 155oC after 95 minutes. The Stud 3 curve was above the
Stud 2 curve due to the presence of a vertical joint along Stud 3.
Near failure the temperature difference across Stud 3 showed a sudden increase
while that of Stud 2 was dropping. The sudden increase in temperature difference
3-46
Experimental Study
across Stud 3 was due to the rapid increase in its hot flange temperature because of
the vertical joint of Pb2. However, the temperature difference across Stud 2 was
reducing due to the heat conducted across the stud. At failure, the temperature
differences across the Studs 2 and 3 were 134 and 168oC, respectively.
700
600
500
Temperature ( o C)
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min)
Stud 2 Stud 3
Figure 3.26: Average Temperature Profiles across Central Studs at

Mid-height in Test Specimen 2
180
160
140
Temperature ( o C)
120
100
80
60
40
20
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min)
Stud 2 Stud 3
Figure 3.27: Temperature Difference across Central Studs at

3-47
Experimental Study
showed an average axial shortening of slightly more than 3 mm (see Figure 3.28 (a)).
The lateral deformation was negligibly small, which indicates that the eccentricity of
loading about the major axis was negligibly small before the furnace was started.
On starting the furnace the wall was observed to expand gently up to 75 minutes.
Thereafter the thermal expansion was accelerated due to the rapid increase in average
stud temperature. A maximum deformation of 9 mm was reached at failure as shown
in Figure 3.28 (b). In comparison with Test 1, the curves of Studs 1 and 4 were
separated gradually from the other two and showed a time lag of 10 minutes.
After the furnace was started the wall bent towards the furnace rapidly up to 15
minutes. This may be due to the whole wall expansion on the fire side since the
average temperature was observed to be only at ambient temperature for the first 6
minutes. For Studs 2 and 3, this was followed by a gradual increase up to 50 and 80
minutes, respectively, since the temperature difference across the studs was only
20oC. Again the lateral deflection of Studs 2 and 3 started to increase rapidly up to 95
minutes and by this time the temperature difference across the studs was nearly
150oC. Studs 2 and 3 reached a maximum central deflection of 22 and 20 mm at 99
minutes, respectively (see Figure 3.29). Beyond which both of them reversed and
deformed very rapidly in the outward direction leading to failure of the wall
specimen at 108 minutes. The lateral deflection curve reversed only once, compared
to Test 1 where the curve reversed its direction twice.
The lateral deflection of Stud 2 was above Stud 3 after 90 minutes irrespective of the
temperature difference across the stud depth. This is considered to be for the same
reason as explained for Test 1. The variation of the average load applied to the studs
is shown in Figure 3.30. After 108 minutes the load could not be maintained further.
Hence the furnace was stopped and the wall specimen was considered to have failed.
3-48
Experimental Study
0.00
-0.50
-1.00
-1.50
-2.00
-2.50
-3.00
-3.50
-4.00
0.00 5.00 10.00 15.00 20.00 25.00 30.00
Load (kN)
10.00
8.00
6.00
4.00
2.00
0.00
-2.00
-4.00
0 10 20 30 40 50 60 70 80 90 100 110
Time (min)
3-49
Experimental Study
0.00
-5.00
-10.00
-15.00
-20.00
-25.00
0 10 20 30 40 50 60 70 80 90 100 110
Time (min)
0.00
-5.00
-10.00
-15.00
-20.00
-25.00
0 10 20 30 40 50 60 70 80 90 100 110
Time (min)
Figure 3.29: Lateral Deflection - Time Profiles of Test Specimen 2
3-50
Experimental Study
0.00
-5.00
-10.00
-15.00
-20.00
0 10 20 30 40 50 60 70 80 90 100 110
Time (min)
(a) Lateral deflection - Time profiles at lower level

32.00
31.50
31.00
30.50
Load (kN)
30.00
29.50
29.00
28.50
28.00
0 10 20 30 40 50 60 70 80 90 100 110
Time (min)
3-51
Experimental Study
in Table 3.4.

Test Specimen 2
Hot Flange (HF) Cold Flange (CF) HF - CF Lateral Deflection

Time (oC) (oC) (oC) (mm)
(min)
30 71 71 56 56 15 15 8 7
60 144 271 84 221 60 50 11 11
90 374 402 222 254 152 148 20 19
108 505 554 371 386 134 168 22* 20*
* - indicates the maximum values obtained from the test and not the value at failure time,
3-52
Experimental Study
3.5.3. Test Specimen 3 (External Insulation - Rock Fibre; Load Ratio 0.4)
Each stud of the wall specimen was loaded in compression up to 30 kN (total load
120 kN), and the wall specimen was then subjected to the standard fire test. The test
was stopped after 134 minutes since the wall specimen could not support the applied
load of 30 kN after this time. The wall test specimen failure was due to the structural

The test specimen showed no signs of lateral displacement during the application of
the axial compression load to each stud. After 3 minutes from starting the furnace,
smoke was escaping from the top of the wall specimen similar to the previous tests
due to the burning of the plasterboard paper on the exposed surface. After about 10
minutes, more smoke and steam were seen to escape from the top corners of the wall.
As wall specimens observed in previous tests, the presence of steam in the mixture of
gases was confirmed by the observation of water particles on the loading frame after
condensation.
From the beginning of the fire test, the wall specimen was observed to be bending
towards the furnace. This continued until the failure and resulted in failing towards
the furnace, which was different to that of Tests 1 and 2 (see Figure 3.31 (a)).
The face layer exposed plasterboards were seen to have fallen off near the centre of
the wall exposing the rock fibre insulation (see Figure 3.31 (b)). The rock fibre
insulation used between Plasterboards 1 and 2 was still there without any integral
failure. The base layer plasterboard on the fire side was seen to have shrunk. This
lead to the opening of plasterboard joints by 10 -15 mm, thus exposing the studs. The
front view shows that the flexural buckling about the minor axis of studs and the
torsional buckling were fully prevented by the lateral support offered by the
plasterboards throughout the test. The exposed plasterboards were stripped off and
the debris removed to expose the frame (see Figure 3.31 (c)). This view shows the
presence of local buckling waves along the studs, confirming the occurrence of local
buckling of studs before the ultimate failure.
3-53
Experimental Study
(a) Partial collapse of ambient side plasterboards
Stud 2
Stud 3

external insulation
3-54
Experimental Study
Stud 1 did not fail in this test while Stud 4 experienced local web buckling along the
length. Stud 2 bent severely about the major axis (moved towards the furnace) and
showed local buckling of the entire cross-section (web, hot flange and lip, cold
flange and lip) near the mid-height suggesting the complete mobilization of the
capacity of the cross-section (see Figure 3.32 (a)). Local failure was also observed at
the support as shown in Figure 3.32 (b). Stud 3 showed local buckling waves even
away from mid-height. At the support the entire section has failed due to local failure
(see Figures 3.32 (c) and (d)).
(a) Stud 2 middle (b) Stud 2 top
(c) Stud 3 middle (d) Stud 3 top
3-55
Experimental Study
Average temperature of the interface surface between the exposed Plasterboard

1 and insulation (Pb1-Ins)
The temperature profile of this surface showed a rapid rise in temperature (first
phase) reaching a temperature of about 90oC by the end of 6 minutes. Beyond 6
minutes the second phase started with the temperature increasing gradually to about
95oC by the end of 12 minutes. This was followed by a very rapid rise in temperature
(third phase) crossing 400oC in 25 minutes and 600oC in 35 minutes. The sudden
increase in the temperature of the interface between the exposed plasterboards and
insulation was probably caused by the heat blocked and redirected by the adjoining
layer of insulation as for Test Specimens 1 and 2.
temperature crossing 900oC by 84 minutes. This is because the heat was used for the
chemical process of rock fibre rather than for the temperature rise. The insulation did
not melt this time and hence the slope is steeper than in the previous two test curves.
A rise in temperature at about 87 minutes indicates that the chemical process was
completed by this time. The Pb1-Ins curve between 35 and 87 minutes is different
from the curve, where the insulation was used inside the cavity (Kolarkar, 2010).
From here onwards this curve approached the fire side temperature curve gradually.
Visual observations and careful analyses of individual thermocouple measurements
have revealed that the left side plasterboard (Pb1) pieces have fallen off at 86 and
109 minutes. Similarly the right side plasterboard pieces have fallen off at 96
minutes. However, only small pieces of middle plasterboard have fallen off at 86 and
110 minutes, leaving majority of the plasterboard in place until failure. The average
temperature of the interface was about 1053oC by the end of the test.
3-56
Experimental Study

This interface responded to the initial rise in temperature (first phase) and reached a
temperature of approximately 70oC rapidly by about 5 minutes and then remained
constant (second phase) up to about 24 minutes. Beyond this time the third phase
started with the temperature rising almost linearly with respect to time reaching
400oC by 65 minutes and 600oC by about 86 minutes. It reached a temperature of
800oC by 108 minutes. The left side and middle Ins-Pb2 curves showed a sudden
temperature rise at 87 and 110 minutes respectively while the right side Ins-Pb2
curve showed a temperature rise at 97 minutes. These also confirm the plasterboards
fall off time. At failure, a temperature difference of 140oC was realized between Pb1-
Ins and Ins-Pb2. This indicates the integrity of the rock fibre insulation throughout
the test.

(Pb2-Cav)
to 76 minutes. A rapid rise in temperature was noticed after this, reaching 500oC at
failure. A temperature difference of approximately 400oC was present across the
thickness of Plasterboard 2 towards the end of the test indicating the sustained
integrity of the exposed base plasterboard layer (Pb2) up to the failure of the wall
specimen.

(Pb3-Cav)
The transmission of heat across the cavity was almost instantaneous due to radiation,
and hence the temperature profile of Pb3-Cav surface followed very closely on the
underside of the Pb2-Cav surface profile. At failure, the temperature difference
3-57
Experimental Study

The temperature of this surface increased gradually reaching around 127oC and
remained almost constant up to 130 minutes. The third phase started beyond this time
with the temperature increasing linearly with respect to time reaching 164oC by the
failure time.

This surface was well protected from the fire and remained under 100oC until the end
of the fire test.
1200
1000
Temperature ( o C)
800
600
400
200
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min)


Figure 3.33: Time - Temperature Profiles of Plasterboard Surfaces in
Test Specimen 3
3-58
Experimental Study
1200
Temperature ( o C) 1000
800
600
400
200
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min)
(b) Time - Temperature profiles across the left, middle and right sections of
Figure 3.33: Time - Temperature Profiles of Plasterboard Surfaces in

Test Specimen 3
cold flanges of the central studs. Initially the temperatures along the length were seen
to be almost uniform in both studs. After about 70 minutes the curves separated and
the mid-height stud temperature readings were higher than the top and lower stud
readings due to thermal bowing effect. However, this difference was reduced with
time due to the heat conducted across the stud. Figure 3.35 gives the temperature
variation across the central stud cross-sections at mid-height.
The hot flange temperature of the second stud at mid-height started to increase
rapidly from 68 minutes. In a similar way the hot flange temperature of the third stud
at mid-height started to increase rapidly from 74 minutes. At failure the temperatures
of hot flanges of Studs 2 and 3 at mid-height were recorded as 556oC and 523oC,
respectively. Stud 2 experienced higher temperatures at failure due to the vertical
joint placed against it.
3-59
Experimental Study
600
500
Temperature ( o C)
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min)
500
450
400
Temperature ( o C)
350
300
250
200
150
100
50
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min)

3-60
Experimental Study
500
450
400
Temperature ( o C)
350
300
250
200
150
100
50
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min)
(c) Time - Temperature profiles of cold flange surfaces of central studs

600
500
Temperature ( o C)
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min)
Figure 3.35: Time - Temperature Profiles across the Central Studs at Mid-
height in Test Specimen 3
3-61
Experimental Study
stud depth
temperature was near ambient temperature for about 7 minutes followed by a gradual
rise up to 85 minutes. By this time the rock fibre became ineffective and the curve
was steeper until failure.
The initial temperature difference across Studs 2 and 3 were nearly 20oC for 55 and
68 minutes, respectively. This duration is much longer than for the previous two
tests. A rapid increase in temperature difference was noticed after this time. The Stud
2 curve was above Stud 3 curve due to the vertical joint along Stud 2.
Studs 2 and 3 reached a maximum temperature difference of 147oC and 124oC by

105 and 115 minutes, respectively. At failure the temperature difference across Stud
2 showed a sudden increase due to the vertical joint of Pb2. However, the
temperature difference across Stud 3 was dropping. This is due to the heat conducted
across the stud depth.
800
700
600
Temperature ( o C)
500
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min)
Stud 2 Stud 3
Figure 3.36: Average Temperature Profiles across Central Studs at

3-62
Experimental Study
160
140
120
Temperature ( o C)
100
80
60
40
20
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min)
Stud 2 Stud 3
Figure 3.37: Temperature Difference across Central Studs at Mid-height

in Test Specimen 3

showed an average axial shortening of 3.5 mm (see Figure 3.38 (a)). The lateral
deformation was negligibly small which can lead to the conclusion that the
eccentricity of loading about the major axis was negligibly small before the furnace
was started. On starting the furnace the wall was observed to expand gently up to 87
minutes. Thereafter the thermal expansion was accelerated due to the rapid increase
in average stud temperature. A maximum deformation of 8 mm expansion was
reached at about 126 minutes, which was then reversed near the failure as shown in
Figure 3.38 (b). After the furnace was started the wall bent towards the furnace with
Studs 2 and 3 reaching a maximum central deflection of 29 and 30 mm (see Figure
3.39). This led to the failure of the studs in the same direction where as a reversal of
lateral movement of studs occurred in Tests 1and 2 near the failure point.
In this test, the vertical joint was placed against Stud 2. Hence the expansion of the
hotter side was restricted compared to Stud 3. This resulted in smaller lateral
deflection of Stud 2 even though it had a higher temperature difference across the
3-63
Experimental Study
stud. Figure 3.40 shows the variation of applied load with time. After 118 minutes
the load could not be maintained further. Hence the furnace was stopped and the wall
specimen was considered to have failed.
0.00
-0.50
-1.00
-1.50
-2.00
-2.50
-3.00
-3.50
-4.00
-4.50
0.00 5.00 10.00 15.00 20.00 25.00 30.00
Load (kN)
10.00
8.00
6.00
4.00
2.00
0.00
-2.00
-4.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min)
3-64
Experimental Study
-5
-10
-15
-20
-25
-30
-35
-40
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Tim e (m in)
0
-5
-10
-15
-20
-25
-30
-35
-40
-45
-50
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min)
3-65
Experimental Study
0.00
-5.00
-10.00
-15.00
-20.00
-25.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min)
(a) Lateral deflection - Time profiles at lower level
32.00
31.00
30.00
Load (kN)
29.00
28.00
27.00
26.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min)
3-66
Experimental Study
in Table 3.5.

Test Specimen 3
Hot Flange (HF) Cold Flange (CF) HF-CF Lateral Deflection

o o o
Time ( C) ( C) ( C) (mm)
(min)
30 74 74 18 18 56 56 5 6
60 113 108 81 75 32 33 7 8
90 266 239 137 141 129 98 12 13
120 445 435 307 317 138 118 22 24
134 556 523 432 420 124 103 29 30
3-67
Experimental Study
3.6. Discussions
Table 3.6 gives the fire resistance ratings (in minutes) of the load-bearing LSF wall
specimens tested under a constant axial compression load during the fire test. This
experimental study has thus extended the work of Kolarkar (2010) on the fire
performance of LSF walls using external insulation. It has also confirmed the
superior performance of LSF walls using external insulation over cavity insulation
(compare results in Tables 3.1 and 3.6), and provided the additional results required
for glass fibre external insulation and a higher load ratio of 0.4. Detailed results of
time-temperature profiles and structural behavioural characteristics of studs in LSF
walls obtained from this study and Kolarkar (2010) can now be used in the numerical
analyses of LSF wall system. Following sections present some of the main findings
by comparing the results of Tests 1 to 3.
Table 3.6: Failure Times of Test Specimens
Load External Failure Time

Test Configuration Failure
Ratio Insulation (min)
1 0.2 Glass Fibre Structural 118
2 0.4 Glass Fibre Structural 108
3 0.4 Rock Fibre Structural 134
3.6.1. Effect of Load Ratios on Wall Specimens Externally Insulated with Glass
Fibre
Test Specimens 1 and 2 were identical except for the applied load level. Therefore
the temperature profile across the wall panels (see Figure 3.41) was almost identical.
The time-temperature curves are the same, but with a time lag of 5 minutes due to
experimental variation, which is acceptable. Therefore it is considered that these
measured temperature profiles can be used for such wall specimens with varying
load ratios in numerical analyses.
3-68
Experimental Study
1200
1000
Temperature ( oC)
800
600
400
200
0
0 20 40 60 80 100 120
Tim e (m in.)
LR 0.2 FS LR 0.2 Pb1-Ins LR 0.2 Ins-Pb2 LR 0.2 Pb2-Cav

LR 0.2 Pb3-Cav LR 0.2 Pb3-Ins LR 0.2 Ins-Pb4 LR 0.2 AS
LR 0.4 FS LR 0.4 Pb1-Ins LR 0.4 Ins-Pb2 LR 0.4 Pb2-Cav
LR 0.4 Pb3-Cav LR 0.4 Pb3-Ins LR 0.4 Ins-Pb3 LR 0.4 AS
Figure 3.41: Time - Temperature Profiles for Test Specimens 1 and 2
The temperature profiles across Studs 2 and 3 (central studs) are shown in Figures
3.42 (a) and (b), respectively. Test 2 curves of Stud 2 are similar to Test 1 curves
with a time lag of 10 minutes. The time lag was developed due to experimental
variation, which is within the acceptable range. The temperature profiles of Stud 3
for Tests 1 and 2 are identical with a negligibly small time lag. At failure the
temperatures of web and flanges were higher for Wall Specimen 1 compared to Test
Specimen 2 due to lower applied loads.
For numerical analyses, an idealised time-temperature profiles will be developed for

studs. The idealised time-temperature profile will be based on these two tests and
will be used for the wall specimens with glass fibre as external insulation.
The curves of lateral deflections versus time for Studs 2 and 3 from Tests 1 and 2 are
shown in Figures 3.43 (a) and (b). Each figure shows that the curves are similar for
Tests 1 and 2. After about 100 minutes, both test specimens reversed their lateral
deflection and moved away from the furnace. Hence Test Specimen 2 failed after
108 minutes, however, near the failure point Test Specimen 1 reversed its direction
for the second time and survived another 10 minutes.
3-69
Experimental Study
600
500
Temperature ( oC)
400
300
200
100
0
0 20 40 60 80 100 120
Tim e (m in.)
LR 0.2 HF LR 0.2 W LR 0.2 CF

LR 0.4 HF LR 0.4 W LR 0.4 CF
(a) Stud 2
700
600
500
Temperature ( oC)
400
300
200
100
0
0 20 40 60 80 100 120
Tim e (m in.)
LR 0.2 HF LR 0.2 W LR 0.2 CF
LR 0.4 HF LR 0.4 W LR 0.4 CF
(b) Stud 3
Figure 3.42: Time - Temperature Profiles for the Central Studs at

Mid-height in Test Specimens 1 and 2
3-70
Experimental Study
Lateral Deflection (mm) 0.00
-5.00
-10.00
-15.00
-20.00
-25.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)
LR 0.2 Lower LR 0.2 Middle LR 0.2 Top

(a) Stud 2
0.00
-5.00
-10.00
-15.00
-20.00
-25.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)

(b) Stud 3
Figure 3.43: Lateral Deflection - Time for the Central Studs at Mid-height in
Test Specimens 1 and 2
3-71
Experimental Study
3.6.2. Effect of Glass Fibre and Rock Fibre Insulations

The only difference between Test Specimens 2 and 3 was the type of insulation used.
In Test Specimen 3 rock fibre was used instead of glass fibre in Test Specimen 2.
The temperature profiles across the wall are shown in Figure 3.44. The curve Ins-Pb2
of Test Specimen 3 was completely different from that of Test Specimen 2. In Test
Specimen 2 the glass fibre melted completely after 56 minutes leaving only few
molten pieces. Hence the curves Pb1-Ins and Ins-Pb2 came closer and merged with
each other. However, in Test Specimen 3, the full layer of rock fibre was in place
until failure. Hence the curve Pb2-Ins was well below the curve Pb1-Ins. This
resulted in low temperature in the other thermocouple positions of Test Specimen 3
and a delay in failure. The temperature profiles across Studs 2 and 3 are shown in
Figures 3.45 (a) and (b), respectively. They are similar in shape, however, a time
delay of 25 minutes was observed between them resulting in longer fire resistance
rating for Test Specimen 3. At failure the temperatures across the studs were nearly
the same for Test Specimens 2 and 3, since the studs in both tests were stressed to the
same level by the applied load of 30 kN. For future numerical analysis, an idealised
time-temperature profiles across the stud will be developed for an LSF wall that is
externally insulated by rock fibre. The idealised time-temperature profiles will be
based on Test 3 curves and those obtained by Kolarkar (2010) for the wall specimen
with external rock fibre insulation and an applied load ratio of 0.2.
1200
1000
Temperature ( o C)
800
600
400
200
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min.)
GF 0.4 FS GF 0.4 Pb1-Ins GF 0.4 Ins-Pb2 GF 0.4 Pb2-Cav

GF 0.4 Pb3-Cav GF 0.4 Pb3-Ins GF 0.4 Ins-Pb4 GF 0.4 AS
RF 0.4 FS RF 0.4 Pb1-Ins RF 0.4 Ins-Pb2 RF 0.4 Pb2-Cav
RF 0.4 Pb3-Cav RF 0.4 Pb3-Ins RF 0.4 Ins-Pb3 RF 0.4 AS
Figure 3.44: Time - Temperature Profiles for Test Specimens 2 and 3

3-72
Experimental Study
600
500
Temperature ( o C)
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min.)
GF HF GF W GF CF RF HF RF W RF CF
(a) Stud 2
600
500
Temperature ( o C)
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min)
GF HF GF W GF CF RF HF RF W RF CF
(b) Stud 3
Figure 3.45: Time - Temperature Profiles for the Central Studs at

3-73
Experimental Study
The lateral deflections of Studs 2 and 3 are shown in Figures 3.46 (a) and (b),
respectively. Test Specimen 3 behaved in a different way compared to the other two
specimens. It failed with rapid lateral deflection towards the furnace after reaching a
maximum lateral deflection of 30 mm whereas the other specimens’ lateral
deflections increased rapidly in the opposite direction away from the furnace near the
failure point.
0.00
-5.00
-10.00
-15.00
-20.00
-25.00
-30.00
-35.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min)
GF Lower GF Middle GF Top
RF Lower RF Middle RF Top
(a) Stud 2
0.00
-5.00
-10.00
-15.00
-20.00
-25.00
-30.00
-35.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min)
GF Lower GF Middle GF Top

RF Lower RF Middle RF Top
(b) Stud 3
Figure 3.46: Lateral Deflection - Time for the Central Studs at
3-74
Experimental Study
3.6.3. Stud Temperatures and Failures

Tables 3.7 to 3.9 give a comparison of the thermal and structural responses of the
central studs at the end of 60, 90 and 120 minutes from the start of the test. Also
values are given at respective failure time of each specimen. Studs of Test Specimen
1 reached higher temperatures compared to those in Test Specimens 2 and 3. This is
because the first specimen was only subjected to half the load of others. The
temperature values near the stud failure of the studs of Test Specimens 2 and 3 are in
a narrow range of 500 – 560oC. The hot flange failure temperatures of these studs
(Stud 3 in Test 2 and Stud 2 in Test 3) are very close to each other (i.e. 554 and
556oC). For these studs the temperature differences between hot and cold flanges
were 168 and 124oC, respectively. This may mean that stud failure is mostly
governed by the (maximum) hot flange temperature than the temperature difference
between hot and cold flanges. Hence we can conclude that structurally similar wall
panels will fail once their studs reach a particular temperature and the fire resistance
can be increased only by delaying the maximum temperature in the studs. This is
confirmed by the increase in fire resistance time of Test Specimen 3, which was
achieved by the delay in temperature rise on studs due to better insulation.
In fact, in all three tests, the stud that had vertical plasterboard joint was subjected to
more heat flow due to the opening of joints in Plasterboard 2. Hence the temperatures
of these studs (Stud 3 in Tests 1 and 2, and Stud 2 in Test 3) were higher than those
of other studs and thus the wall failure was also influenced and initiated by these
studs.
Table 3.7: Thermal and Structural Responses of Test Specimen 1
HF (oC) CF (oC) HF-CF (oC) L.D (mm)

Time (min.)
S2 S3 S2 S3 S2 S3 S2 S3
60 218 261 109 128 109 133 11 11
90 432 424 281 269 151 155 20 19
118 582 664 511 490 71 174 21* 19*
3-75
Experimental Study

Time (min.)
S2 S3 S2 S3 S2 S3 S2 S3
60 144 271 84 221 60 50 11 11
90 374 402 222 254 152 148 20 19
108 505 554 371 386 134 168 22* 20*

Time (min.)
S2 S3 S2 S3 S2 S3 S2 S3
60 113 108 81 85 32 23 7 8
90 266 239 137 141 129 98 12 13
120 445 435 307 317 138 118 22 24
134 556 523 432 420 124 103 29 30
HF - Hot Flange; CF - Cold Flange; L.D. - Lateral Deflection; S - Stud

*- indicates the maximum values obtained from the test and not the value at failure time,
3.6.4. Failure Direction of LSF Walls

As discussed earlier, all the wall specimens deflected towards the furnace as a result
of thermal bowing, but Test Specimens 1 and 2 reversed the direction near the failure
point and failed by deflecting away from the furnace. A possible explanation for this
behaviour is given next.
Figure 3.47 shows the position of the centroid of stud at mid-height at elevated non-
uniform temperatures. It will be at the geometric centre of the stud at the beginning
of the tests. Once the fire test is commenced, thermal bowing will be induced
because of non-uniform temperature distribution across the stud. The position of
3-76
Experimental Study
centroid at mid-height will therefore move towards the fire side due to this thermal
bowing. In addition to this the effective centroid of the stud (neutral axis) will shift
towards the ambient side due to the loss of stiffness (reduced elasticity modulus) of
the hotter side steel.
Position of stud
Position of stud at non-uniform
at ambient temperature
temperature e∆T
e e∆E
Ambient side Fire side
X Z Y
X: Position of centroid at the beginning of tests, and the point of Load
application throughout the test.
Y: Position of centroid after taking into account thermal bowing.
Z: Position of centroid after taking into account thermal bowing and neutral axis
shift.
e∆T Eccentricity due to thermal bowing
e∆E Eccentricity due to neutral axis shift
e Actual eccentricity
Figure 3.47: Location of Effective Centroid of Studs in Relation to the Point of

Load Application during Fire Tests
The net eccentricity can be expressed by e = e∆T - e∆E. The position of Z and the
value for e will vary according to the values of e∆T and e∆E which depend on the
temperature distribution of the stud cross-section. During the initial stages of the test,
thermal bowing will be induced by the temperature variation across the wall, even
before the temperatures of the stud has increased significantly. Hence e∆T will be
more than e∆E, which will result in a positive value for e. Therefore the wall will
move towards the furnace initially. However, near failure, non-uniform temperature
distribution will be developed in the stud cross-section and hence e∆E will be
developed. Therefore the positive value of e will reduce and the LSF wall will start
to move away from the furnace.
3-77
Experimental Study
The failure direction also depends on the comparative strengths of cold and hot
flanges near the failure time. This can be explained by investigating the behaviours
of Tests 2 and 3 which were subjected to the same load ratio of 0.4. Test 3 survived
more time due to the superior performance of rock fibre compared to glass fibre (134
minutes compared to 108 minutes). However, at failure the cold flange temperature
is higher in Test 3 (432oC) than in Test 2 (386oC) (see Tables 3.8 and 3.9). This is
due to the heat transfer from hot side to the cold side during this extra time.
Therefore the structural failure of Test 3 was towards the furnace by compressive
failure of the cold flange near mid-height. This is because the heat penetrated well
into the ambient side of the steel stud to reduce the strength of cold flange below the
applied stress. However, in Test 2 there was not enough time for the heat to be
transferred from hot flange to cold flange and hence the acceleration of fire side
temperature reduced the strength of hot flange below the applied stress. This resulted
in overall buckling away from the furnace by the local buckling failure of the hot
flange.
3.6.5. Plasterboard Performance

In the design of LSF wall studs, the assumption of lateral restraint provided by the
fire side base layer plasterboard (Pb2) plays a significant role. According to BS 5950
(BSI, 1998) a total lateral restraint of 3% of the compressive load in the critical
flange is required. However, the load resistance at each point of restraint should not
be less than 1%. In the study of Klippstein (1980b) it was assumed that the failure by
weak axis flexural buckling or torsional buckling will be prevented by the gypsum
board cladding on the internal and external faces of the wall.
According Gerlich et al. (1996), when steel temperatures reach critical levels (> 300-
400oC) the fire exposed side plasterboard will reduce its ability to prevent buckling
of the studs due to degradation. Gerlich et al. (1996) used a thinner unexposed lining
(9.5 mm) for a test to achieve an insulation failure. However, it was found that the
unexposed lining failed to provide lateral support after it had degraded. Hence with
the comparison of TASEF modelling and test results, Gerlich et al. (1996)
recommended a 3 mm thickness of undamaged gypsum to be retained to provide
lateral buckling.
3-78
Experimental Study
According to Kaitila (2002), the model is developed as two parts. First the design
model was based on equations that consider flexural buckling along the stronger axis
only and the torsional-flexural buckling was not considered as a critical failure mode.
This was when the temperature stays under the level of calcination of the
plasterboards. Hence the lateral restraint to the studs by the plasterboards was taken
into account throughout the fire. Secondly, the torsional-flexural buckling mode was
checked. This is when the temperature during the fire test goes above the level of
calcination of the plasterboards (equal to approximately 550oC). Hence the lateral
restraint was no longer considered to be valid on the fire side.
In the studies of Feng et al. (2003b) and Zhao et al. (2005) the exposed plasterboards
were assumed to provide lateral restraint to the studs. Alfawakhiri (2001) assumed
that the flexural-torsional and weak axis buckling failure modes are prevented by
bridging and blocking. Hence the question of lateral restraint by plasterboards was
not raised.
In all three wall specimens, Pb1 started to fall off at about 85 minutes and the entire
Pb1 collapsed by 110 minutes. After the reinforcing paper was burnt small cracks
were observed in the plasterboard. These cracks propagated with time due to high
temperature development. Finally in all three wall specimens Pb1 fell off when the
unexposed face of the board reached about 900°C (slightly lesser value for Test
Specimen 3). This accelerated the temperature rise of the remaining parts and
interfaces of the wall. This observation is helpful to update the findings of Sultan
(1996) who reported that plasterboard fall-off occurs when the unexposed face of the
plasterboard reaches about 600°C (Buchanan and Gerlich, 1997). According to the
current experimental study, the plasterboard fall-off occurred when the unexposed
face of the board reaches about 900°C.
In all three wall specimens, the integrity of Pb2 was maintained until the failure
point. It should be noted that even in Test Specimen 1 Pb2 was present until the
failure point. However, it fell-off at the last minute due to severe lateral deflection at
failure. For the current study, the temperatures of the cavity side (unexposed) surface
of Pb2 at failure point are shown in Table 3.10. For all the wall specimens, these
temperatures are less than 600°C. In Test Specimens 2 and 3, these temperatures
3-79
Experimental Study
were about 500°C. These temperatures are less than the value recommended by
Sultan (1996). Hence it can be assumed that the plasterboard did not loose its
integrity until failure. In addition to this the minimum temperature difference
between the faces of Pb2 was about 400°C until failure. This may also suggest that
the Pb2 did not fully calcinate to lose its ability to provide lateral restraint until
failure. This is verified in the experimental study where lateral or torsional buckling
failure modes were not observed in the studs of any of the three wall specimens.
Hence in the numerical modelling of the LSF walls tested in this study, lateral
restraint by plasterboards will be considered for both (hot side and cold side) flanges
of the studs until failure.
Table 3.10: Temperatures of Pb2 Surfaces at Failure

Insulation Cavity Difference Maximum Temperature
Test
side (°C) side (°C) (°C) of the Stud (°C)
1 1055 595 460 664
2 1055 497 558 554
3 911 500 411 556
3.7. Conclusions
This chapter has presented the details of three full scale load-bearing wall specimens
built and fire tested to study the thermal and structural performance of the load-
bearing wall assemblies lined with two layers of plasterboards and external
insulation. The analysis of the test results showed that the proposed cold-formed steel
wall systems with external insulation provided considerably increased fire resistance
rating with smaller lateral deformations than the conventional cavity insulated wall
systems. The effects of load ratios were also investigated in this study by considering
load ratios of 0.2 and 0.4. The temperature distributions of the studs were alike for
thermally similar wall panels although they were tested with different load ratios.
Based on the experimental results, it was concluded that the plasterboard fall-off
occurred when the unexposed face of the board reached about 900°C. Details of
experimental results including the temperature and deflection profiles measured
during the tests are presented along with the stud failure modes. These results will be
useful for the finite element modeling and validations.
3-80
Finite Element Models
CHAPTER 4
4. DEVELOPMENT AND VALIDATION OF FINITE ELEMENT MODELS

OF LSF WALLS
4.1. Introduction
Performing full scale fire tests of LSF walls is very difficult, expensive and time
consuming. Therefore finite element analyses (FEA) have been used by many
researchers in recent times to study the behaviour of LSF walls under fire conditions
(Kaitila, 2002, Feng et al. 2003b, Zhao et al. 2005). For this purpose suitable finite
element models of LSF walls were developed in this research and validated using the
available results of full scale fire tests. Such validated finite element models can then
be used to simulate the true behaviour of LSF walls under fire conditions. This
approach will considerably improve the efficiency and effectiveness of this research
into the complex structural and fire behaviour of LSF walls. In this research,
ABAQUS Standard Version 6.9 (HKS, 2009) was used in the development of finite
element models of LSF walls. MSC/PATRAN was used as a pre-processor to create
the input file. It was also used as a post-processor to obtain and analyse the results.
Finite element models of LSF walls under fire conditions were developed in two
stages. A model was developed first to simulate the behaviour of LSF wall studs at
ambient temperature. It was then extended to simulate the behaviour of LSF wall
studs under fire conditions. The finite strip analysis program CUFSM was used for
elastic buckling analyses of LSF wall studs at ambient temperature while the finite
element program ABAQUS was used for both elastic buckling and nonlinear
analyses under ambient and fire conditions. Elastic buckling analysis results from
ABAQUS were compared with the corresponding results from CUFSM. From the
analyses the critical buckling load and the associated buckling mode were identified.
Following this, nonlinear analyses of LSF wall studs were undertaken and their
results were compared with test results. Comparisons of the results from both elastic
buckling and nonlinear analyses thus enabled a full validation of the developed
numerical models. This chapter presents the details of the numerical models of LSF
walls developed in this research and their validation.
4-1
4.2. Finite Strip Analyses of LSF Wall Studs at Ambient Temperature
Finite strip models using CUFSM were used to validate the bifurcation buckling
analyses of finite element models using ABAQUS. CUFSM is commonly used for
the purpose of conducting elastic buckling analyses of thin-walled members. It
employs the well known finite strip method to provide solutions for the cross-
sectional stability of these members. The LSF wall stud used in the experimental
study reported in Chapter 3 was used in the finite strip modelling. Three cases of
LSF wall studs were considered as shown in Figure 4.1. In Case 1 the LSF wall stud
was analysed without considering the lateral stability (in the plane of wall) provided
by the plasterboards. On the other hand, Cases 2 and 3 considered the relevant lateral
restraints provided by the plasterboards. Hence the lateral movements (UZ) of
selected nodes were restrained along the member length. In Case 2, all the flange
nodes were restrained whereas only the middle node in each flange was restrained in
Case 3.
Lateral movement (UZ) is restrained

40
15
Y
X
90 t = 1.15 mm
Z
15
40
(a) Dimensions (b) Case 1 (c) Case 2 (d) Case 3
Figure 4.1: Cross-sectional Dimensions and Boundary Conditions of the LSF

Wall Stud Used in CUFSM Analyses
Figure 4.2 presents the results from CUFSM in the form of buckling plots (Load
factor versus half-wave lengths) for all three cases. The load factor shown in Figure
4.2 is the elastic buckling load expressed as a ratio of the squash load, which is
127.87 kN for the stud section considered in this study. The buckling plots for Cases
2 and 3 are about the same irrespective of the number of nodes restrained. In other
words, there is negligible difference between restraining one node of the flange
versus restraining the entire flange to simulate the restraints provided by the
plasterboard.
4-2
As expected, when the half-wave length is more than 200 mm, the load factors for
the studs with plasterboard restraint (Cases 2 and 3) are much higher compared to
that for the stud without plasterboard restraint (Case 1). When the member length is
less than 1000 mm, the governing load factor is 0.31 with a local buckling mode in
all three cases. In our case where the member length is 2400 mm, plasterboard plays
a major role. In this case the governing load factor for Case 1 is 0.12 and the failure
mode is flexural-torsional buckling. On the other hand the load factors for Cases 2
and 3 are still the same with a value of 0.31. Figure 4.2 (b) presents the buckling
modes and load factors in all three cases. When the member length is 2400 mm, the
governing buckling load factor in each case is circled.
Case 2
Case 1
Case 3
(a) Load factor versus half-wave length
HWL 70 HWL 2400 HWL 70 HWL 2400 HWL 70 HWL 2400

LF 0.31 LF 0.12 LF 0.31 LF 0.85 LF 0.31 LF 0.80
(a) Case 1 (b) Case 2 (c) Case 3
(b) Buckling modes of studs at different half-wave lengths

Figure 4.2: Results from Finite Strip Analyses
4-3
Figures 4.2 (a) and (b) clearly demonstrate that the use of plasterboards on both sides
of the studs in the LSF walls prevents the studs buckling prematurely by a flexural
torsional mode (load factor of 0.12). Since the major axis flexural buckling mode
(about Z axis) occurs at a higher load factor of 0.8 for a stud length of 2400 mm,
local web buckling with a load factor of 0.31 becomes the critical buckling mode of
plasterboard restrained studs in LSF walls.
4.3. Finite Element Analyses of LSF Wall Studs at Ambient Temperature

Many numerical softwares are available to predict the structural behaviour of thin-
walled structural members subjected to different actions. ABAQUS is one of them
that uses a finite element analysis (FEA) technique. It will provide accurate results
for LSF wall studs provided their loading and boundary conditions and mechanical
properties are modelled correctly. The axial compression load capacities of LSF wall
studs are high at ambient temperature compared to those at elevated temperatures.
Fire resistance ratings of these studs are determined at varying ratios of the ambient
temperature load capacity (load ratio) under fire conditions. Therefore a suitable
finite element model was developed first to determine the axial compression load
capacity of LSF wall studs at ambient temperature. The present study used a 1.15
mm thick lipped channel with external dimensions of 90x40x15 mm, as shown in
Figure 4.1.
Convergence studies were conducted first to determine the suitable element type and
sizes that simulates a closer behaviour of the actual LSF wall test. Appropriate
mechanical properties, initial geometric imperfections and residual stresses were
used in these analyses. Centre-line dimensions were used in both finite strip and
finite element models of LSF wall studs. Elastic bifurcation buckling and nonlinear
analyses were conducted using ABAQUS. The Riks method was used in the
nonlinear analyses. The developed model was validated by comparing its results with
the full scale test results obtained by Kolarkar (2010).
4.3.1. Element Type

The LSF wall studs are made of thin cold-formed steel sheets. In finite element
modelling this can be represented by shell elements such as S4, S4R, S4R5, S8R5
4-4
and S9R5. In their numerical studies, Kaitila (2002) and Feng et al. (2003b) used
S8R5 and S4R elements, respectively.
Element type S4 is a fully integrated, general-purpose shell element that allows for
transverse shear. The element S4R is a four node, quadrilateral, stress/displacement
shell element with reduced integration and a large-strain formulation. The doubly
curved general-purpose shell element S4R gives robust and accurate solutions in
most applications and allows transverse shear deformations. On the other hand, the
shell element S4R5 cannot be used in cases where transverse shear deformation is
not negligible. The S4R5 element has one integration location per element while the
S4 element type has four integration locations per element. S4R5, S8R5 and S9R5
elements use five degrees of freedom per node with four, eight and nine nodes,
respectively.
Higher order elements need much more processing time and memory requirements.
Therefore it was decided to select a four-node element type for this numerical study.
Table 4.1 shows the elastic buckling and nonlinear analysis results based on these
element types with a mesh size of 4 mm x 4 mm. For nonlinear analyses, the analysis
time is higher for S4R element compared to S4R5 element. Similarly the analysis
time is higher for S4 element compared to S4R element. The disk usage and memory
required are much higher for the S4 element than the other two four-node element
types (S4R and S4R5). There is a slight variation in the elastic buckling and ultimate
loads when S4R5 and S4R elements were used. However, the difference between
these loads is negligible when S4R and S4 elements are used. Therefore S4R element
type was selected for further studies since it provides closer results as S4 with less
memory space and time. This element type ensured sufficient degrees of freedom for
buckling deformations of light gauge cold-formed steel compression members.
Table 4.1: Comparison of Different Types of Four Node Shell Elements

Analysis Time (min.) Memory (Gb) Load (kN)
Element Elastic Elastic Elastic
Nonlinear Nonlinear Nonlinear
Type Buckling Buckling Buckling
Analysis Analysis Analysis
Analysis Analysis Analysis
S4R5 4.3 16.4 0.4 0.4 39.8 76.7
S4R 7.9 24.2 0.4 0.4 39.8 77.1
S4 7.9 28.0 1.5 1.5 39.8 77.1
4-5
Figure 4.3 shows the ultimate load versus out-of-plane deflection of the stud for each
element type used in the analysis. The behaviour of element types S4R and S4 was
the same and resulted in the same ultimate load. The load-deflection curve for S4R5
is slightly different to others with a slightly lower ultimate load. These observations
confirm the choice of S4R elements in the analyses of LSF wall studs.
80
70
Axial Compression Load (kN)
60
50
40
30
20
10
0
0 2 4 6 8 10 12 14 16
Out-of-plane Deflection (mm)
S4R5 S4R S4
Figure 4.3: Comparison of Load - Deflection Curves Using Different Shell

Element Types
4.3.2. Element Size

Element size used is an important factor in FEA. A finer mesh with smaller elements
will produce more accurate results. However, it may not give the most economical
simulation since it needs higher processing time and memory. Therefore the
optimum size of the element was found by a convergence study as shown in Table
4.2. Since this is a comparison study, residual stresses were not included in the
analyses. The ratio of analysis time and element size, and the ratio of memory and
element size increased rapidly when the mesh size was decreased below 4 mm.
However, at this time the percentage difference in the ultimate load was very low.
Figures 4.4 (a) and (b) show the influence of element size in the memory and time,
respectively. They show that the memory and analysis time were very high when a 2
mm x 2 mm mesh size was used. Therefore 4 mm x 4 mm mesh size was selected in
the analyses of LSF wall studs, which gave the optimum solution with acceptable
processing time and memory.
4-6
Table 4.2: Effect of Mesh Sizes on Analysis Time, Memory and Ultimate Load
Analysis Time Memory Ultimate Load
Element
∆ Time ∆ Memory
Size min. Gb kN % Difference
∆ Size ∆ Size
10 4.44 0 0.99 0 78.61 0
6 13.45 2.3 2.69 0.4 77.18 1.8
5 19.69 6.2 4.17 1.5 76.95 0.3
4 31.97 12.3 6.52 2.4 76.71 0.3
2 149.85 58.9 25.54 9.5 76.39 0.4
30
25
20
Memory (Gb)
15
10
0
0 2 4 6 8 10 12
Element Size (mmxmm)
Bifurcation Buckling Analysis Nonlinear Analysis
(a) Memory
160
140
120
Time (min.)
100
80
60
40
20
0
0 2 4 6 8 10 12
Element Size (mmxmm)
Bifurcation Buckling Analysis Nonlinear Analysis
(b) Analysis time

Figure 4.4: Effect of Element Size on Memory and Analysis Time for Elastic
Buckling and Nonlinear Analyses
4-7
4.3.3. Mechanical Properties
The use of accurate mechanical properties is important in finite element analyses.

Mechanical properties are characterised by elastic modulus, stress-strain relationship
and Poisson’s ratio. To enable the comparison of results from the FEA and tests, the
mechanical properties of test specimens should be used in FEA. Therefore the
measured values of elastic modulus and yield strength were used in FEA. Kolarkar
(2010) reported that the yield strength and the elasticity modulus of 1.15 mm G500
steel used in the LSF walls were 569 MPa and 213520 MPa at ambient temperature
based on a series of tensile coupon tests. These measured values were used in FEA.
However, the Poisson’s ratio of steel was assumed as 0.3.
One of the important data required in the modelling of cold-formed steel members is
the stress-strain (σ - ε) curve. There are two stress-strain models used in the finite
element analyses, namely elastic-perfect-plastic and strain hardening models as
shown in Figure 4.5. The elastic-perfect-plastic model assumes a constant yield stress
in the plastic range whereas the strain hardening model includes strain hardening in
the inelastic range. The stress-strain curves from the tensile coupon tests show the
presence of strain hardening behaviour for 1.15 mm G500 steel used in this research.
Therefore there is a need to determine whether a simpler elastic-perfect-plastic model
can be used.
600
500
Stress (MPa)
400
Strain Hardening Model

300
Elastic-Perfect-Plastic Model
200
100
0
0.000 0.002 0.004 0.006 0.008 0.010 0.012 0.014
Strain
Figure 4.5: Stress – Strain Models
4-8
Tensile coupon tests give the nominal stress-strain curves. On the other hand,
ABAQUS requires a stress-strain relationship in terms of true stress and logarithmic
plastic strain. Therefore the engineering stress-strain data obtained from tensile
coupon tests ( σ eng and ε eng ) are converted to the true stress and logarithmic plastic
strain values ( σ true and ε true

pl
) using the following equations.
σ true = σ eng (1 + ε eng ) (4.1a)
σ true
ε true
pl
= ln(1 + ε eng ) − (4.1b)
E
To assess the suitability and accuracy of the material models, a numerical study was
conducted for LSF wall studs. Figure 4.6 shows the elastic-perfect-plastic material
model used in the FEA. It also shows the actual tensile coupon test results and the
converted true stress – true strain values used in FEA as strain hardening model.
Elastic buckling analyses were conducted first, which were followed by non-linear
analyses based on the two mechanical property models. S4R element type was used
in the simulations. Initial geometric imperfections and residual stresses were also
included in the analyses. Figure 4.7 shows the nonlinear responses based on these
two mechanical property models. Both models gave the same ultimate failure load.
From these results it is confirmed that there is no significant effect on the nonlinear
analysis when the simplified elastic-perfect-plastic model was used. Local buckling
failures of LSF wall studs are likely to occur before the strain hardening effect
becomes important, ie. larger strains. Thus, the use of simplified elastic-perfect-
plastic material model did not affect the nonlinear behaviour and ultimate load of
LSF wall studs. Therefore the elastic-perfect-plastic material model was used to
simulate the LSF wall studs via ABAQUS *PLASTIC option.
4-9
700
600
500
Stress (MPa)
400 Young's Modulus

Stress - Strain Curve, FEA - EPP
300 Stress - Strain Curve, Exp - Eng
Stress - Strain Curve, FEA - SH
200
100
0
0 1 2 3 4 5 6
Strain (%)
Figure 4.6: Stress – Strain Models Used in FEA
80
70
60
50
40
30
20
10
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Out-of-plane Deflection (mm)
Elastic-perfect-Plastic Model Strain Hardening Model
Figure 4.7: Comparison of Nonlinear Responses for Elastic-Perfect-Plastic and

Strain Hardening Models
4.3.4. Loading and Boundary Conditions

Appropriate boundary conditions must be included in order to simulate the behaviour
of test specimens using FEA. Figure 4.8 shows the details of tested LSF walls at
ambient temperature (Kolarkar, 2010). Each stud was loaded by a hydraulic jack
placed below it. Applied load and displacement readings were recorded at the end of
4-10
each load increment. The specimen was assumed to have failed when the oil pressure
in the loading jacks could not be maintained. Pinned end conditions were used in the
test. End plates were used to transfer the load from the jack to the studs via their
geometric centroids so that no additional moments were created.
Loading Frame
Wall Specimen
LVDTs
Loading Jacks
Figure 4.8: Test Specimen and Set-up Used by Kolarkar (2010)
Considering the previous numerical studies (Kaitila, 2002, Feng et al. 2003b and
Zhao et al. 2005) and the experimental behaviour of studs, one of the two middle
studs was considered in the analyses by taking into account the appropriate loading
and boundary conditions as shown in Figure 4.9.
Top Track
Studs
Stud
4 Bottom Track
3
2
Stud 1
Load
(a) LSF Wall (b) Stud
Figure 4.9: Modelling of LSF Walls
4-11
In the numerical study of Kaitila (2002), rigid end plates were used at each end and
the member was considered to be pin-ended. The reference point for the rigid end
section was the original centroid of the gross cross-section. Zhao et al. (2005)
considered two support conditions separately. In the first case pinned support
condition was assumed for both ends whereas in the second case fixed support
condition was used in one end and pinned support condition was used at the other
end. In their numerical analyses, both ends were modelled using a stiffer material
with E as 21000 GPa. In Feng et al.’s (2003b) numerical study, a rigid plate was
attached to each end of the stud and pinned support condition was used as in Kaitila
(2002). In the experimental study of Kolarkar (2010), the end support conditions of
LSF wall studs were maintained as pinned. Hence in the current numerical study also
pinned support conditions were simulated as shown in Figure 4.10. Rigid plates
made of R3D4 elements were attached to each end of the stud and twisting about
these plates (ROTX) was restrained. The ends of the stud were restrained in the two
major axial directions (UY and UZ). The axial displacement (UX) was restrained at
one end of the member. Axial compressive load was applied at the centroid of the
section at the other end.
Pinned
Restrained DOF
Support
“1234”
Load
Rigid Plate
Restrained
DOF 3
Stud
Restrained
DOF “234” Y
X
300 mm
Z
Load
Rigid Plate
Figure 4.10: Loading and Boundary Conditions Used in FEA
4-12
4.3.4.1. The Influence of Eccentric Loading and Boundary Conditions
The influence of (major axis) eccentric loading due to experimental errors was
investigated using two cases as shown in Figure 4.11. In Case 1, the loading was
applied with an eccentricity. However, in Case 2, both the loading and the support
condition were applied with an eccentricity.
Load Eccentric Eccentric

Loading Loading
Eccentric
Support Support
Case 1 Case 2
Figure 4.11: Eccentric Loading and Support Conditions
Table 4.3 and Figure 4.12 show the effect of eccentric loading and boundary
conditions on the ultimate load. The ultimate load is reduced to a large extent when
both ends are eccentric. The ultimate load is reduced by 13.3% when this eccentricity
is 5 mm. This eccentricity is likely to occur in the testing of LSF walls under fire
conditions where the loading point moves towards the furnace due to thermal
bowing. Similarly, a 5 mm loading eccentricity is possible in the testing of LSF
walls, which leads to a 5.1% reduction in the ultimate load. These observations
demonstrate the difficulties in simulating the behaviour of LSF walls under fire
conditions.
Table 4.3: Comparison of Ultimate Loads of Studs with Different Eccentricities
Case - 1 2
Eccentricity (mm) 0 1 2 5 1 2 5
Ultimate Load (kN) 77.1 76.6 76.6 73.1 76.6 75.1 66.8
% Difference - 0.6 0.6 5.1 0.6 2.6 13.3
4-13
78
76
Ultimate Load (kN)
74
72
70
68
66
0 1 2 3 4 5 6
Eccentricity (mm)
Case 1 Case 2
Figure 4.12: Effect of Eccentric Loading and Support Conditions
4.3.4.2. The Influence of Plasterboards

The simulation of support provided by plasterboards plays a major role in the finite
element modelling of LSF wall studs. In Kaitila’s (2002) model the lateral
displacement (UZ) of the stud was restrained at each node of the web-flange edges
along the length. Zhao et al. (2005) restrained the lateral displacement (UZ) only at
the flange centre where the screws were located. In the study of Feng et al. (2003b), a
range of effective lengths about the minor axis from 300 to 2000 mm was
considered, ie, stud length between screws and the full length. Thus they indirectly
simulated the possible restraining effects instead of modelling the plasterboard
restraints at the screw locations.
In the tests of Kolarkar (2010), the gypsum plasterboards were found to be effective
in restraining the studs from torsional buckling and flexural buckling about the minor
axis. The connection of steel stud to plasterboard was represented by restraining the
lateral displacement (in the plane of wall) of both flanges at 300 mm intervals along
the length. This boundary condition was applied to several regions of the flanges
including single node (Case 1) and single row of nodes across the section (Case 2) as
shown in Figure 4.13. Feng et al. (2003b) and Zhao et al. (2005) used the method
given in Case 1 for their numerical study of LSF wall studs. Case 3 was simulated
with plasterboard restraint applied to a single row of nodes along the length which is
similar to Kaitila (2002). The restraints to the in-plane horizontal and twisting
4-14
deformations of the stud were considered in Case 4. The plasterboards are unlikely to
share any axial compression load applied to the LSF walls and hence they were not
included in the model. However, their restraint to the in-plane horizontal deformation
of the studs was included as described above.
Restrained DOF 3 Restrained DOF 3 and 4
Case 1 Case 2 Case 3 Case 4
Figure 4.13: Simulation of Plasterboard Restraints
Table 4.4 shows both the elastic buckling and the ultimate loads obtained for these
four different cases. All these cases resulted in the same elastic buckling load of 39.8
kN. The ultimate load was the same for Cases 1 and 2 with a value of 77.1 kN. This
indicates that there is negligible difference between restraining one node of the
flange and restraining the entire flange to simulate the restraints provided by the
plasterboard. The ultimate loads obtained with Cases 1 and 4 are also the same. This
shows that the additional twisting restraint is not helping the stud and hence it can be
ignored. However, Case 3 resulted in a higher ultimate load of 82.9 kN. Therefore it
is concluded that restraining the node along the full length as in Kaitila’s (2002)
study will provide a slightly higher ultimate load. Hence either Case 1 or Case 2 can
be used to simulate the plasterboard restraint on LSF wall studs. However, it was
decided to use Case 1 since it represents the actual condition in LSF wall studs.
Table 4.4: Effects of Different Cases of Plasterboard Restraints
Elastic Buckling and Ultimate Loads (kN)

Failure Load
from Test Case 1 Case 2 Case 3 Case 4
(kN)
E U E U E U E U
79 39.8 77.1 39.8 77.1 39.8 82.9 39.8 77.1
E – Elastic Buckling Load; U – Ultimate Load
4-15
4.3.5. Initial Geometric Imperfections

Light gauge cold-formed steel sections are likely to have larger initial geometric
imperfections since they deform easily during handling, fabrication and construction.
Local imperfections affect the local or distortional buckling capacities, and are
expressed in terms of plate width and/or thickness. Global imperfections can also
affect the member capacity and are expressed in terms of the member length. These
imperfections can have a large influence on the ultimate failure load of studs.
According to Schafer and Pekoz (1998), the strength of a cold-formed steel member
is particularly sensitive to its imperfections that are in the shape of eigen modes.
Therefore the relevant initial geometric imperfection was included in the nonlinear
analyses by introducing them to the appropriate buckling mode obtained from the
bifurcation buckling analyses. The initial geometric imperfections in the studs were
applied via ABAQUS *IMPERFECTION option by modifying the nodal coordinates
using a field created by scaling appropriate buckling eigenvectors obtained from an
elastic bifurcation buckling analysis. Figure 4.14 shows the first eigen mode
representing local web buckling as obtained from the bifurcation buckling analysis.
The local web buckling is predominant in this mode and also in all the test results of
Kolarkar (2010). Therefore this eigen mode was selected to introduce the initial
geometric imperfection with an amplitude value, which is suitable for local
imperfection.
Mid-span
Support
Figure 4.14: First Eigen Mode from Bifurcation Buckling Analysis
Many researchers have used the measured geometric imperfections to study the
effect of geometric imperfections on the ultimate strength of cold-formed steel
4-16
members. However, it is not possible in the current study due to the unavailability of
such imperfection data.
Kaitila (2002) modelled LSF wall studs and compared his results with Eurocode 3
Part 1.3 (ECS, 2006) calculations. In the modelling of LSF wall studs he used both
local and global initial imperfections. According to him, without a small global
imperfection, the structure is too ideal and in most analyses, the deformations are
limited to axial shortening of the compressed column and the ultimate load is
overestimated. However, this assumption is not supported by test results. In his
study, for local buckling, one of the most symmetrical eigenvalue buckling modes
was chosen as the basis for the local web buckling imperfection for subsequent
nonlinear static post-buckling analyses. The buckled shape is highly symmetric and
he used b/100 and b/200 as the amplitudes of local imperfection. The eigenvalue
buckling mode corresponding to strong axis Euler buckling was also selected and he
used L/1000, L/750, L/500 and L/400 as the amplitudes of global imperfection.
However, these values will not be considered in the current study since the the local
web buckling near the support was predominant in the lowest eigen mode and also in
the test results of Kolarkar (2010).
In the study of Feng et al. (2003b) the initial deflections were according to the
deflection profile of the lowest eigen value. This mode showed only local buckling
along the stud. Flexural or flexural-torsional buckling was not seen in the lowest
eigen mode. However, they used L/1000, L/500 and L/200 as the amplitude of initial
geometric imperfection, which is not acceptable. They also used t as the amplitude of
initial geometric imperfection, while Zhao et al. (2005) used 1 mm in their numerical
study. AS 4100 (SA, 1998) suggests b/150 as the maximum tolerance for the web
elements, where ‘b’ is the web depth.
The magnitudes of imperfections of the web (stiffened) and flange (unstiffened)

elements for local buckling were also suggested by Schafer and Pekoz (1996). They
suggested a local plate imperfection amplitude of 0.006b for stiffened elements. An
alternative value of 6te-2t is also suggested by them. For unstiffened elements, a
value of (0.014 b/t + 0.5)t was suggested. In the above expressions, b = plate width, t
= thickness and d1 and d2 are the maximum geometric imperfections in the web and
4-17
flange elements, respectively (see Figure 4.15). In Schafer and Pekoz’s (1996) study
d1 is referred to as Type 1 imperfection in a stiffened element such as the web in this
research while d2 is referred to as Type 2 imperfection in an unstiffened element such
as the flange in this research. Figure 4.15 shows the two types of imperfections. The
above expressions gave geometric imperfections of the same order as the values
reported by Young and Rasmussen (1995) for press braked plain C-sections.
b
d2
b
d1
(a) Local Web Imperfection (b) Local Flange Imperfection
Figure 4.15: Initial Geometric Imperfection (Schafer and Pekoz, 1996)
Table 4.5 shows the amplitudes considered in the current study for local initial
imperfections. The effect of initial imperfection amplitude on the ultimate capacity
was investigated using nonlinear finite element analyses. Table 4.5 also shows the
influence of the initial imperfection amplitude on the ultimate compression capacity
of the stud. As anticipated, larger imperfections reduce the load carrying capacity of
the stud. Figure 4.16 compares the test and FEA results with varying initial
geometric imperfections. According to this figure the use of 0.006b as the initial
local imperfection agrees well with the test results. Hence this value was adopted for
the LSF wall stud section used in the current study.
Table 4.5: Effect of Initial Imperfection Amplitude on the Ultimate Load
Imperfection 0.006b b/150 6te-2t t

1 mm
Amplitude (0.533 mm) (0.592 mm) (0.692 mm) (1.15 mm)
Ultimate Load
77.3 77.1 76.7 75.4 74.7
(kN)
% Difference* 2.2 2.4 2.9 4.6 5.4
* Comparison with Test Results
4-18
80
70
60
Ultimate Load (kN)
50
Test
40
FEA
30
20
10
0
0.0006b b/150 6te-2t 1 t
Initial Geometric Imperfection (mm)
Figure 4.16: Effect of Initial Geometric Imperfection on Ultimate Load
4.3.6. Residual Stresses

The residual stress is an important parameter influencing the axial compressive
strength of steel studs as this can cause premature yielding, and reduce their strength.
Two types of residual stresses are identified, namely, membrane and flexural residual
stresses. The membrane residual stress is negligibly small in cold-formed steel
members as it does not involve any heat treating or welding in the manufacturing
process. Therefore the membrane residual stresses can be ignored in cold-formed
lipped channel sections according to Schafer and Pecoz (1998). Based on the
measurements of residual stresses in lipped channel sections, Schafer and Pekoz
(1998) made suitable recommendations for the flexural residual stress distributions in
cold-formed steel sections. They proposed suitable residual stress distributions for
both press-braked and roll-formed specimens as shown in Figure 4.17 (a).
The residual stress models of Schafer and Pekoz (1998) have higher residual stresses
at the rounded corners of lipped channel sections. However, the press-braked
sections used in this research had sharp corners, and the corner radii were negligibly
small. Therefore the corner regions with higher residual stresses due to press braking
can be neglected. Hence a new set of residual stresses was proposed by Ranawaka
(2006) for lipped channel sections, where the rounded corners were ignored (see
4-19
Figure 4.17 (b)). These proposed values were used in the current numerical study at
ambient temperature.
The initial residual stresses were created using the SIGINI Fortran user subroutine
and executed using ABAQUS *INITIAL CONDITIONS option, with TYPE =
STRESS. This SIGINI user subroutine is generally used when the residual stresses
are complicated. In this subroutine the local components of the initial stress were
defined as a function of the global coordinates. Since the global coordinates were
used to define the local stress components, member imperfections had to be included
in determining the residual stresses. To vary the flexural residual stresses through the
thickness, they were applied as a function of the integration point numbers through
the thickness. Five integration points were defined through the thickness of each
element to simulate the accurate distribution of residual stresses.
(a) Models Proposed by Schafer and Pekoz (1998)

17 %
8%
17%
8%
17%
(b) Model Proposed by Ranawaka (2006)
Figure 4.17: Flexural Residual Stress Distributions Assumed in FEA
4-20
However, the effect of residual stresses was found to be small based on finite
element analyses of studs at ambient temperature. The ultimate loads obtained with
and without residual stresses were 77.1 kN and 76.7 kN, respectively, ie, a
percentage difference between these ultimate loads is only 0.5. The effect of residual
stresses will be even more insignificant at elevated temperatures (Lee, 2004). Hence
residual stresses will not be considered in the numerical modelling of studs under fire
conditions.
4.3.7. Analysis Methods
Two types of analyses, namely bifurcation buckling and nonlinear analyses, were
conducted using ABAQUS. The bifurcation buckling analyses were used to
determine the elastic buckling loads and modes. The relevant buckling modes were
then used to include the initial geometric imperfections in the nonlinear analyses.
Finally the nonlinear analyses were used to determine the ultimate loads and
deformations. The initial, minimum and maximum load increments were controlled
appropriately to obtain smooth loading and accurate load factors.
In the nonlinear analyses, both the Riks on method and Riks off method were used to
determine the ultimate compression load. In the Riks on method, force induced
loading was used whereas in the Riks off method, displacement induced loading was
used. Load versus axial deformation curves from both methods are plotted in Figure
4.18. The plot for the Riks on method reversed after achieving the ultimate
compression load. On the other hand, the Riks off method did not reverse after
achieving the highest load. However, both methods produced the same load-
deformation curve until failure and achieved the same ultimate compression load.
The failure modes were also the same in both methods. Therefore the Riks on
method was used in the study of LSF wall studs at ambient temperature.
Figure 4.19 shows the load versus axial shortening and out-of-plane deflection
curves of the LSF wall stud under axial compression load. The out-of-plane
deflection of the web was measured near the support where large local buckling
waves were observed. Although the axial shortening reversed after achieving the
4-21
ultimate load, the out-of-plane deflection curve confirms that the analysis has
reached the ultimate load.
80
70
60
50
40
30
20
10
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Axial Shortening (mm)
Riks On Riks Off
Figure 4.18: Comparison of the Effect of Riks On and Off Methods
80
70
60
50
40
30
20
10
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Deformation (mm)
Axial Shortening Out-of-Plane Deflection
Figure 4.19: Ultimate Load versus Deformation Curves from the Riks On
Method
4-22
4.3.8. Validation of Finite Element Models

Precious time and resources can be saved by using finite element analyses instead of
full scale tests. However, the validity of finite element models must be established
before undertaking detailed parametric studies. Therefore it is essential to ensure that
the finite element model can be used to simulate the desired buckling and ultimate
strength behaviour of cold-formed steel studs at ambient temperature. It is also
important to be able to extend this model to simulate the behaviour of studs under
fire conditions.
The results from finite strip analyses (CUFSM) and Kolarkar’s (2010) full scale tests
were used here to validate the results of finite element analyses (FEA) at ambient
temperature. Elastic buckling loads obtained from finite strip analyses and finite
element analyses (ABAQUS) were 39.5 kN and 39.8 kN, respectively. On the other
hand, the ultimate failure load recorded in the full scale test and FEA were 79.0 kN
and 77.3 kN, respectively. This shows that the developed finite element model
accurately predicted both the elastic buckling and non-linear ultimate capacities of
studs subjected to pure axial compression. In addition to these validations, the load-
deflection curves and failure modes obtained from FEA were compared with
Kolarkar’s (2010) test results.
4.3.8.1. Load - Deflection Curves

Axial shortenings of all the four studs in the LSF walls were measured in the tests.
They were measured at the loading plates placed at the bottom of the studs. This
deformation can also be obtained from the nonlinear finite element analyses. For
comparison purposes, the axial deformation of the reference node of the rigid plate
attached to the stud was taken from finite element analyses. Figure 4.20 (a) compares
the load versus axial deformation curves from tests and FEA. The axial deformations
obtained from test include the compaction of packing materials kept between the
LSF wall and the loading frame. Therefore the axial shortening measured in tests is
higher than the FEA results. However, when the initial section of the test load versus
axial deformation curves was removed, their agreement with FEA results improved
considerably as shown in Figure 4.20 (b). The observed higher axial shortening in
test compared to FEA could have been due to an overall initial imperfection in the
stud. Such an imperfection and other deviations in testing could have lead to larger
4-23
lateral deflections during the test, which in turn lead to also larger axial shortening.
Having considered these facts, it can be concluded that the load-deflection curves
from FEA and tests showed a reasonably good agreement.
80
60
40
20
0
0 5 10 15 20
Axial Deform ation (m m )
Stud 1 Stud 2 Stud 3 Stud 4 FEA
(a) Actual
80
60
40
20
0
0 5 10 15 20
Axial Deform ation (m m )
Stud 1 Stud 2 Stud 3 Stud 4 FEA
(b) When initial part is removed
Figure 4.20: Comparison of Load - Deformation Curves from Test and FEA
4-24
4.3.8.2. Failure Modes
Accuracy of the developed models was also assessed by comparing the failure modes
from FEA, finite strip analyses and tests. Initially the elastic buckling load and mode
obtained from CUFSM were compared with those from the bifurcation buckling
analysis of FEA. Local web buckling was observed as shown in Figure 4.21. This
figure shows a close agreement with these two types of numerical analyses in terms
of elastic buckling load and mode.
(a) CUFSM (b) ABAQUS

Elastic buckling load = 39.5 kN Elastic buckling load = 39.8 kN
Figure 4.21: Elastic Buckling Modes Obtained from Finite Strip and Finite
Element Analyses
Figure 4.22 (a) compares the failure modes obtained from full scale test and FEA.
According to Kolarkar (2010), the studs failed by the local buckling of web and
flanges at the base close to the loading point. Nonlinear finite element analyses also
showed the same local web buckling behaviour at failure as shown in Figure 4.22
(b). Larger web deformations could be seen near the support as in the tests. Figure
4.22 (c) shows the deformed stud at failure.
4-25
Local buckling
failure
(b) FEA
(a) Test
(c) Deformed Stud at Failure
Figure 4.22: Failure Modes from Test and Nonlinear Finite Element Analyses
4-26
Figure 4.23 shows the stress components (von Mises) of the LSF wall stud near
failure. The highest von Mises stress was observed near the support where the load
was applied. A closer look at this figure indicates that yielding has occurred in the
web element near the support. These comparisons demonstrate that the ultimate load
of stud has been achieved during the nonlinear FEA.
Figure 4.23: Stress Distribution of the Stud at Failure
This section has described the details of finite element models developed for the
investigation into the structural behaviour and capacity of LSF wall panels at ambient
temperature. The finite element model provided reasonable comparisons with test
results. It accurately predicted both the elastic buckling and non-linear ultimate
capacities of studs subjected to pure compression. The models accounted for all the
significant behavioural effects including material inelasticity, local buckling
deformations, residual stresses, and initial geometric imperfections. The models were
validated using a series of comparisons of elastic buckling and ultimate load capacity
results with corresponding results from an established finite strip analysis program
CUFSM and test results, respectively. It is therefore reasonable to assume that the
validated model can be used for the development of finite element model under fire
conditions.
4-27
4.4. Finite Element Analyses under Fire Conditions
In the previous sections, the finite element program ABAQUS has been successfully
used to simulate the behaviour of LSF wall studs at ambient temperature. This
section extends the application of finite element analyses to LSF wall studs under fire
conditions. Kaitila (2002), Zhao et al. (2005) and Feng et al. (2003b) have also used
ABAQUS in their numerical studies of LSF wall studs under fire conditions.
Finite element analyses of structural steel members under fire conditions can be
conducted under two conditions, namely steady state and transient state conditions.
In the steady state modelling, the temperature distributions in the steel cross-section
are raised to the target levels and then kept unchanged. A load is then applied in
increments until failure. In the transient state modelling, the load is applied in
increments until it reaches the target load (a particular load ratio). Following this, the
temperature distribution is input in a time frame. Feng et al. (2003b) used steady
state modelling whereas Kaitila (2002) and Zhao et al. (2005) used transient state
modelling in their numerical analyses.
Full scale fire tests were undertaken based on transient state conditions. In order to
validate the FEA using the test results, it is important to simulate the complete
loading history as used in the tests. Therefore finite element analyses were performed
under transient state conditions where the stud was subjected to a pre-determined
axial compression load first. The stud was then exposed to the measured temperature
distributions (profiles) obtained from full scale fire tests. On the other hand, in order
to develop suitable fire design rules, we should know the load carrying capacity of
the stud for a given temperature profile. In this case the analyses were performed
under steady state conditions where the stud was subjected to a pre-defined
temperature profile first and then the load was applied incrementally until it failed.
Many analyses were conducted in closer time intervals to simulate the test under
steady state conditions. In both types of analysis, the measured temperature profiles
of LSF wall studs from Kolarkar (2010) and this research (Chapter 3) were used.
4-28
4.4.1. Element Type and Size

For FEA of LSF wall studs under fire conditions, the element type S4R was used
with a mesh size of 4mm x 4mm as for ambient temperature conditions discussed
earlier.
4.4.2. Loading and Boundary Conditions

Loading conditions of the model under fire conditions were similar to the model at
ambient temperature as discussed in Section 4.3.4. Rigid plates made of R3D4
elements were attached to each end of the stud and pinned support conditions were
assumed for both ends. That is, the ends of the stud were restrained in the two major
axial directions (UY and UZ). The axial displacement (UX) was restrained at one
end of the member. Axial compressive load was applied at the centroid of the section
at the other end. Twisting about the axis of the member (ROTX) was restrained at
both ends.
The plasterboard restraint was simulated by restraining the appropriate translational

displacement (UZ) in every 300 mm as discussed in Section 4.3.4.2. The in-plane
lateral restraint provided by the plasterboards was considered on both sides of the
studs in the model, that is, it was assumed that the plasterboard on the hot flange side
also provided sufficient lateral restraint until the stud failure. This assumption is the
same as used by Kaitila (2002), Feng et al. (2003b) and Zhao et al. (2005) in their
finite element models.
4.4.3. Mechanical Properties

Mechanical properties are very important in the finite element analyses of
compression members at elevated temperatures. The mechanical properties
significantly influence their elastic buckling and ultimate strength behaviour because
they deteriorate rapidly with increasing temperature.
4.4.3.1. Yield Strength

Outinen (1999), Lee et al. (2003), Mecozzi and Zhao (2005), Eurocode 3 Part 1.2
(ECS, 2005), Ranawaka and Mahendran (2009a) and Chen and Young (2007) have
proposed suitable equations for yield strength reduction factors of cold-formed steels.
Their proposals determine the yield strength reduction factors as a function of
temperature. Based on the yield strength results obtained from tensile coupon tests at
4-29
various temperatures in the research conducted at QUT (Dolamune Kankanamge,

2010 and Kolarkar, 2010), it was found that none of the proposals can be used to
predict the yield strength reduction factors of high grade Australian cold-formed
steels at elevated temperatures except the proposals of Ranawaka and Mahendran
(2009a).
Ranawaka and Mahendran (2009a) showed that there were differences between the
low and high strength steels in the reduction of yield strength with temperature and
hence developed separate equations for low and high strength steels. Although
Ranawaka and Mahendran’s (2009a) equations predicted the yield strength reduction
factors reasonably well, it was identified that they have to be improved in some
regions to improve the accuracy. Therefore a new set of equations to determine the
yield strength reduction factors of high strength steels was developed by Dolamune
Kankanamge and Mahendran (2011) after considering all the results obtained by
QUT researchers including the results of Kolarkar (2010). These modifications to
Ranawaka and Mahendran’s (2009a) equations were based on the reduction factors
of yield strengths obtained from the 0.2% proof stress method.
The yield strength reduction factors of high strength steels show three main regions:
two nonlinear regions (20oC – 300oC and 300oC – 600oC) and one linear region
(600oC – 800oC). Three different equations were therefore proposed by Dolamune
Kankanamge and Mahendran (2011) for these three main regions. Equations 4.2 (a)
to (c) present the proposed equations for reduction factors (fy,T / fy,20) of high strength
steels, where fy,T and fy,20 are the 0.2% proof stresses at elevated and ambient
temperatures, respectively, and T is the temperature.
f y ,T  (T − 20 )4.56 
20 ≤ T < 300 Co
= 1 −  (4.2a)
f y , 20  1x1010 T 
300 ≤ T < 600 o C

f y ,T 
= 0.95 −
(T − 300)1.45 
 (4.2b)
f y , 20  7.76T 
f y ,T
600 ≤ T ≤ 800 o C = −0.0004T + 0.35 (4.2c)
f y , 20
4-30
Figure 4.24 shows the reduction factors for the mechanical properties based on
Dolamune Kankanamge and Mahendran (2011) and Eurocode 3 Part 1.2 (ECS,
2005). The predictive equations developed by Dolamune Kankanamge and
Mahendran (2011) were used in the current numerical study since they are more
accurate for the cold-formed steels used in Australia. It is also shown by Dolamune
Kankanamge and Mahendran (2011) that a good agreement exists between the results
of other QUT research and the proposed equations. Therefore it is recommended to
use these modified equations to determine the yield strength reduction factors of high
strength steels (G500) at any given temperature.
600
500
Yield Stress (MPa)
400
300
200
100
0
0 100 200 300 400 500 600 700 800
Temperature (oC)
Dolamune Kankanamge (2010) Eurocode 3 Part 1-2 (ECS, 2005)
Figure 4.24: Reduction in Yield Stress based on Dolamune Kankanamge and

Mahendran (2011) and Eurocode 3 Part 1.2 (ECS, 2005)
4.4.3.2. Elastic Modulus

Deterioration of elastic modulus with increasing temperature directly influences the
performance of the structural member as it reduces the stiffness. Several researchers
(Outinen, 1999, Lee et al., 2003, Ranawaka and Mahendran, 2009a and Chen and
Young, 2007) have developed predictive equations for elastic modulus reduction
factors as a function of temperature. It was found that none of these equations
accurately predicted the elastic modulus reduction factors. Therefore Dolamune
Kankanamge and Mahendran (2011) developed new empirical equations for elastic
modulus with respect to temperature. There are two main regions in which reduction
factors vary linearly: 20oC - 200oC and 200oC - 800oC. Two linear equations were
4-31
proposed for the two identified temperature regions to predict the elastic modulus
reduction factors at elevated temperatures (Equations 4.3 (a) and (b)).
ET
20 ≤ T ≤ 200 o C = −0.000835T + 1.0167 (4.3a)
E20
ET
200 < T ≤ 800 o C = −0.00135T + 1.1201 (4.3b)
E 20
Figure 4.25 shows the elastic modulus reduction factors based on Dolamune
Kankanamge and Mahendran (2011) and Eurocode 3 Part 1.2 (2005). The predictive
equations developed by Dolamune Kankanamge and Mahendran (2011) were used in
this study for the above mentioned reasons.
250000
200000
E lastic M odulus (M P a)
150000
100000
50000
0
0 100 200 300 400 500 600 700 800
o
Temperature ( C)
Dolamune Kankanamge (2010) Eurocode 3 Part 1-2 (ECS, 2005)
Figure 4.25: Reduction in Elastic Modulus based on Dolamune Kankanamge

and Mahendran (2011) and Eurocode 3 Part 1.2 (ECS, 2005)
4.4.3.3. Stress - Strain Curves

The stress-strain model at elevated temperature is usually based on the Ramberg-
Osgood stress-strain model and is given as shown in Equation 4.4 (a) where εT is the
strain corresponding to a given stress fT at temperature ( T ), ET and fy,T are elastic
modulus and yield strength, respectively, and ηT and β are two parameters. Ranawaka
and Mahendran (2009a) proposed β to be taken as 0.86 and two equations for ηT
depending on the grade of steel as given in Equation 4.4 (b).
4-32
ηT
f  f y ,T  f T 
ε T = T + β    (4.4a)
ET  
 T  f y ,T
E 
where
η T = −3.05 x10 −7 T 3 + 0.0005T 2 − 0.2615T + 62.653 20 ≤ T ≤ 800 o C (4.4b)
β = 0.86
Dolamune Kankanamge and Mahendran (2011) showed that the stress-strain curve
model proposed by Ranawaka and Mahendran (2009a) accurately predicted the
stress-strain curves of cold-formed high strength steels based on the results of the
previous studies at QUT. Hence it is recommended that Ranawaka and Mahendran’s
(2009a) proposed stress-strain model for G550 steels can be used for 1.15 mm thick
G500 cold-formed steels. Figure 4.26 shows the stress-strain curves for different
temperatures obtained from the equations proposed by Ranawaka and Mahendran
(2009a) (Equations 4.4 (a) and (b)). For this purpose, fy,T and ET in Equations 4.4 (a)
and (b) were obtained from Equations 4.2 (a) to (c) and Equations 4.3 (a) and (b),
respectively, and the measured ambient temperature mechanical properties of 1.15
mm G500 steel (fy,20 = 569 MPa, E20 = 213520 MPa). These engineering stress-strain
data obtained from the proposed equations (σeng and εeng) were then converted to the
true stress and logarithmic plastic strain values (σtrue and ε true
pl
) as discussed in
Section 4.3.3 for use in FEA.
700
020 oC
600 100 oC
200 oC
500
o
300 C
Stress (MPa)
400 o
330 C
300 400 oC
o
500 C
200
600 oC
100 o
700 C
o
0 800 C
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Strain
Figure 4.26: Stress - Strain Curves of G500 Steels at Different Temperatures
4-33
4.4.3.4. Thermal Expansion

Steel will expand considerably when exposed to high temperatures. Therefore
thermal bowing will be developed due to the presence of non-uniform temperatures
across the stud section. Hence the prediction of coefficient of thermal expansion is
necessary for the analysis of light steel frame wall panels at elevated temperatures.
Feng et al. (2003b) assumed a constant value of 1.4x10-5 °C-1 for the coefficient of
thermal expansion at higher temperatures. In the study of Kaitila (2002) the values
proposed by Outinen et al. (2001) and Eurocode 3 Part 1.2 (ECS, 2005) were used.
Zhao et al. (2005) also used the temperature dependant coefficient of thermal
expansion given in Eurocode 3 Part 1.2 (ECS, 2005). According to Eurocode 3 Part
1.2 (ECS, 2005) the coefficient of thermal expansion α is given by Equation 4.5.
These values were used in the current finite element modelling under fire conditions.
α = 1.2x10-5 + 0.4x10-8T – 2.416x10-4/T for 20ºC ≤ T ≤ 750ºC (4.5)
4.4.4. Temperature Distribution

The accuracy of using two different methods of simplifying the non-uniform
temperature distribution was investigated by Feng et al. (2003b). In the first
simplification the lip and flange elements on each side of the cross-section were
assumed to have the same temperature. The web had a linear distribution of
temperatures as shown in Figure 4.27 (a). In the second simplification each plate
element of the cross-section was assumed to have a linear distribution (see Figure
4.27 (b)). Feng et al. (2003b) concluded that the non-uniform temperature
distribution in the cross-section of a column may be simplified by assuming uniform
temperatures in the flanges and lips on both the fire and cold sides, and a linear
temperature distribution in the web.
On the other hand in the numerical study of Zhao et al. (2005) the lip temperature
was taken as equal to that of the web at the corresponding height. The temperature
was assumed to be constant in each flange while the web temperature was considered
to vary linearly from the hot flange to the centre of the web, and then again vary
linearly at a slower rate from the centre of the web to the cold flange (see Figure 4.27
(c)).
4-34
(a) Simplification 1 (b) Simplification 2 (c) Zhao et al. (2005)

Feng et al. (2003b) Feng et al. (2003b)
Figure 4.27: Temperature Distributions Assumed in the Previous Studies
Simplification 1 for the non-uniform temperature distributions in LSF wall studs as

proposed by Feng et al. (2003b) makes it possible to develop hand calculation
methods to evaluate the column strength at non-uniform temperatures. Therefore in
the current study also this method was adopted for numerical modelling and later in
the development of fire design rules. i.e, at any time during the analyses, the non-
uniform temperature distribution in the cross-section of a stud was simplified by
assuming uniform temperatures in the flanges and lips on both fire and cold sides. A
linear temperature distribution was assumed in the web as shown in Figure 4.28.
HF
Linear
Variation
CF
Figure 4.28: Temperature Distribution Used in FEA
In order to simulate the fire test conditions, the measured temperature profiles
obtained from the full scale fire tests were used as was done in the previous studies
(Kaitila, 2002 and Zhao et al., 2005). The temperatures of the steel stud profile at
4-35
mid-height and quarter points were measured throughout the test. Therefore average
measured temperature values were input and these temperatures were assumed to be
constant over the entire stud length (see Figure 4.28).
The temperature loading was created as an amplitude curve with respect to step time.
An amplitude curve allows arbitrary time variations of load, displacement, and other
prescribed variables to be given throughout a step (using step time) or throughout the
analysis (using total time). By default, the values of loads, boundary conditions, and
predefined fields either change linearly with time throughout the step (ramp function)
or they are applied immediately and remain constant throughout the step. ABAQUS
offers different ways to define an amplitude curve: Tabular definition method was
selected to define the temperature loading amplitude curve as a table of values at
convenient points on the time scale. ABAQUAS interpolates linearly between these
values as needed. By default in ABAQUAS / Standard, if the time derivatives of the
function must be computed, some smoothing is applied at the time points where the
time derivatives are discontinuous.
The temperature loads with time were created using *AMPLITUDE, NAME=name,
DEFINITION=TABULAR option. PATRAN pre-processor allows defining
amplitude curves as field function. A field is a set of data defined by a relationship
between one or more independent variables. The fields available in PATRAN
support up to three dimensions and are divided into three types: spatial, material
property, non-spatial fields. Field can be created either from tabular input,
mathematical relationships expressed in function or as a scalar or vector results on
collection of finite elements.
Spatial fields are commonly used to control application of temperatures in the

Load/Boundary conditions application, although they can also be applied to
displacements and other generalized loads. Spatial fields can be scalar or vector in
nature and can be applied in either real or parametric space. They will vary over the
coordinates of the selected coordinate system. In this model, spatial field was used to
create linearly varying temperature on the web. Non-spatial Fields are principally
used to specify time varying data. Time dependent loads and boundary conditions are
defined via non-spatial Fields. Three different non-spatial fields (cold flange, hot
4-36
flange and web) were created to apply the measured temperature loads with time.
The default size of all tabular fields is 30 entries in each dimension although it can be
increased up to 1000 in the Options forms. Therefore the measured temperature
values obtained from tests import as comma separated value (CSV) files. This
provides compatibility with spreadsheet programs such as Microsoft Excel.
4.4.5. Initial Geometric Imperfections

Kaitila’s (2002) model included both local and global initial geometric
imperfections. Zhao et al. (2005) and Feng et al. (2003b) used 1 mm and L/1000,
respectively, as their amplitudes of initial imperfections. However, due to the
dominance of thermal bowing the initial geometric imperfection is not expected to
have any significant effect on the behaviour and strength of LSF wall studs at
elevated temperatures. In the finite element analyses of studs under fire conditions,
the initial step was an eigen buckling analysis at ambient temperature, in which the
buckling modes were obtained and the deformed shape corresponding to the lowest
buckling mode was used to input the stud initial imperfections. In addition to the
eigen buckling mode at ambient temperature, two other modes were also considered
in the study. The first mode was obtained by analysing the stud with the deteriorated
material property at hot flange temperature. Second mode was obtained by analysing
the stud with varying temperature dependant material properties across the section.
However, the nonlinear analyses with these different eigen buckling modes resulted
in the same ultimate load. Therefore in further studies, the local buckling mode
obtained from the bifurcation analysis at ambient temperature will be used with an
imperfection value of 0.006b.
4.4.6. Residual Stresses

The effect of residual stress was considered in the model at ambient temperature.
With increasing temperatures, the residual stresses in the studs decrease rapidly (Lee,
2004). Therefore the residual stresses were not considered in this model as confirmed
by other researchers (Kaitila, 2002, Feng et al., 2003b and Zhao et al., 2005).
4-37
4.5. Simulation of Test 1
4.5.1. Finite Element Analyses under Transient State Conditions

Initially, an eigen buckling analysis at ambient temperature was performed, from
which the buckling modes were obtained. The deformation profile of the lowest
buckling mode was used to input the stud initial imperfections as discussed in
Section 4.4.5. The nonlinear analysis was different to the analysis at ambient
temperature. Here the analyses were performed in two steps with Riks off method. In
the first step the load was increased to the target level and in the second step the
measured temperatures were applied. The Riks off method is helpful in this situation,
because after achieving the target load the next step can be started, which is not
possible with the Riks on method.
The model creation was similar to that at ambient temperature. However, the
definition of material properties, application of temperature loading and the method
of analyses were different. The load, support conditions, plasterboard restraints and
temperature distributions were applied as boundary conditions. After the target load
was reached, while changing the temperatures in a time frame, the stud was analysed
by varying the mechanical and thermal properties of steel at the relevant
temperatures. In other words the material properties were defined as a function of
temperature and the temperature values across the section were defined as a function
of time. This was modelled using the facility called “Field” in ABAQUS. Several
fields (tabular input) were created to input the mechanical properties (elastic
modulus, yield stress and thermal expansion coefficient) at different temperatures.
The temperature loading was also created using two fields with respect to time (for
hot and cold flanges and web temperatures).
The analysis was conducted in a time frame to arrive at a deformation versus time
curve for each test. The accuracy of the developed finite element models was verified
by using the experimental deformation curves of LSF wall studs under fire
conditions. Hence these numerical analyses were used to fully understand the
improvements offered by the new composite panel system. The use of accurate
numerical models as described above allowed the inclusion of various complex
4-38
thermal and structural effects such as thermal bowing, local buckling and neutral axis
shift at elevated temperatures.
The measured average temperature profiles shown in Figure 4.29 were used in the
finite element analyses of Test 1 studs under transient state conditions. In this figure,
Studs 2 and 3 are the central studs in the tested LSF wall. Linear variation was
assumed for the web from hot flange temperature to cold flange temperature. The
lips were assumed to be at the same temperature as their corresponding flanges.
700
600
500
Temperature (oC)
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)
Stud 2 Hot Flange Stud 2 Cold Flange Stud 3 Hot Flange Stud 3 Cold Flange
of Test 1
4.5.1.1. Deflected Studs of Test 1 at Different Time Intervals
Figure 4.30 shows the shortened stud after the axial loading step. Up to this point, the
stud was not subjected to any temperature loading. Figure 4.31 (a) shows the stud
after 30 minutes of the temperature loading step. At this point, the stud has elongated
due to the temperature loading. We can also observe the thermal bowing towards the
fire side due to the temperature difference across the stud. Figures 4.31 (b) and (c)
show the deflected stud at 90 and 115 minutes, respectively, and as expected the
thermal bowing has considerably increased.
4-39
Axial shortening
Figure 4.30: Axial Shortening of Stud 3 after Step 1 (Axial Compressive Load)
under Transient State Conditions
Ambient side
Thermal bowing
Fire side
(a) After 30 minutes of Step 2 (Temperature)
Conditions
4-40
Ambient side
Thermal bowing
Fire side
(b) After 90 minutes of Step 2 (Temperature)
Thermal expansion
Ambient side
Thermal bowing
Fire side
(c) After 115 minutes of Step 2 (Temperature)
Conditions
4-41
4.5.1.2. Automatic Stabilization and Damping Factor based on the Dissipated

Energy Fraction
In FEA under transient state conditions, numerical problems were experienced near
the test failure time. It was assumed that in such cases global load control methods
such as the modified Riks method may not work properly. Thus ABAQUS offers the
option to stabilize this class of problems by applying damping throughout the model
in such a way that the viscous forces introduced are sufficiently large to prevent
instantaneous buckling or collapse but small enough not to affect the behaviour
significantly while the problem is stable. ABAQUS provides an automatic
mechanism for stabilizing unstable quasi-static problems through the automatic
addition of volume-proportional damping to the model.
The problem was stable at the beginning of the step and the instabilities may have
developed near the failure time. However, the damping factor was applied
throughout the entire step. While the model is stable, viscous forces and therefore the
viscous energy dissipated are very small. Thus, the additional artificial damping has
no effect. If a local region goes unstable, the local velocities increase and,
consequently, part of the strain energy then released is dissipated by the applied
damping. ABAQUS can, if necessary, reduce the time increment to permit the
process to occur without the unstable response causing very large displacements. In
most applications the first increment of the step is stable without the need to apply
damping. The damping factor is then determined in such a way that the dissipated
energy for a given increment with characteristics similar to the first increment is a
small fraction of the extrapolated strain energy. The fraction is called the dissipated
energy fraction and has a default value of 2.0 × 10–4. Alternatively, the dissipated
energy fraction can be specified for automatic stabilization directly using
*STABILIZE = dissipated energy fraction option. This method was used to stabilize
the current problem.
Figure 4.32 shows the stable nonlinear response near the failure time when the
stabilize method was used. A range of values from 1.0 × 10–11 to 1.0 × 10–9 was used
as the dissipated energy fraction. The value 1.0 × 10–11 was found to be good enough
to stabilise the model while having minimal viscous energy dissipated. There is not
4-42
much difference between the original FEA results and the FEA results with 1.0×10–11
as the dissipated energy fractions. Hence a value of 1.0×10–11 was used to determine
the failure time.
20.00
10.00
0.00
-10.00
-20.00
-30.00
0 20 40 60 80 100 120
Time (min.)
Experiment FEA
FEA Stabilize 0.00000000001 FEA Stabilize 0.0000000001
FEA Stabilize 0.000000001
(a) Variation of axial deformation with time
10.00
0.00
-10.00
-20.00
-30.00
0 20 40 60 80 100 120
Time (min.)
Experiment FEA
FEA Stabilize 0.00000000001 FEA Stabilize 0.0000000001
FEA Stabilize 0.000000001
(b) Variation of lateral deflection with time
Figure 4.32: Response for the Static Stabilize Method for Stud 2 of Test 1
4-43
Figure 4.33 shows the axial deformation and lateral deflection versus time curves
obtained from FEA under transient state conditions. During the fire test, the axial
deformation was positive and increased with time due to thermal expansion.
However, at failure the stud no longer can carry the applied load and hence the axial
deformation decreased (shortened). This behaviour was also found in FEA and the
sharp fall in axial deformation indicates that Studs 2 and 3 failed after 118 and 115
minutes, respectively in FEA. The sharp fall in lateral deflection experienced in the
test was not observed in FEA. This is due to the difference in failure locations
between the test and FEA, which will be discussed later in Section 4.5.1.4.
The lateral deflections obtained from FEA and test followed the same trend even
though there are some differences between them during the early stages of the test. In
the initial period of the test, the temperature gradient across the stud was very low
compared to the temperature gradient across the LSF wall. The predicted lateral
deflection using FEA was based on the temperature gradient across the stud. Hence
the lateral deflection including the thermal bowing was less in this case. However,
during the early stages of the test the LSF wall will act as a rigid structure and the
thermal bowing will be induced due to the temperature gradient across the wall.
Hence the lateral deflection including the thermal bowing was more in this case. The
FEA results show that the lateral deflection of Stud 2 was decreasing near the failure
while it was increasing for Stud 3. This difference in thermal bowing near failure is
due to the change in temperature difference in these studs. Figure 4.29 shows that the
temperature difference between the hot and cold flanges decreased in Stud 2 near
failure, however, it increased in Stud 3. Due to this difference in temperature profiles
the studs behaved in a slightly different manner in FEA near failure. However, this
isolated behaviour of stud is practically not possible in full scale test.
Regardless of these differences, the agreement of these load-deformation curves with

corresponding experimental curves is very good compared to the previous numerical
studies of LSF walls under fire conditions (Zhao et al. 2005). This is due to several
reasons including the use of accurate reduction factors for mechanical properties at
elevated temperatures and appropriate temperature profiles. This will be discussed in
detail in Section 4.9.1.
4-44
20.00
10.00
0.00
-10.00
-20.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)
Stud 2 Exp. Stud 3 Exp. Stud 2 FEA Stud 3 FEA
10.00
0.00
-10.00
-20.00
-30.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)
Conditions
4-45
4.5.1.3. Thermal Bowing and Magnified Thermal Bowing
Cooke (1987) considered thermal bowing of simply supported steel members due to
temperature gradient across the section and derives the following equation for mid-
height deflection (Gerlich, 1995).
α∆TL2
e ∆T = (4.5)
8d
where
e∆T is the mid-height deformation due to thermal bowing;
α is the expansion coefficient for steel;
L is the member length;
∆T is the temperature difference across the member;
d is the member depth.
In the presence of axial compressive load in the LSF wall stud, a bending moment
will be generated due to this thermal bowing. This will cause the stud to bend further
towards the furnace. This additional horizontal deflection due to P-∆ effects is called
as the magnified thermal bowing.
10.00
Magnified
thermal bowing
0.00
-10.00
-20.00
-30.00
0 20 40 60 80 100 120
Time (min.)
Experiment Gerlich (1995) FEA
Figure 4.34: Prediction of Lateral Deflection at Mid-height for Stud 2 of Test 1

4-46
Figure 4.34 shows the lateral deflection obtained from the test, FEA and Equation
4.5. It is interesting to see that the lateral deflection predicted by FEA follows the
same trend as Equation 4.5. The difference between these curves is due to the
magnified thermal bowing (P-∆ effect) as discussed earlier. This magnified thermal
bowing is directly proportional to thermal bowing (Gerlich, 1995 and Zhao et al.,
2005). This behaviour can be seen in the lateral deflection predicted by FEA. In other
words Figure 4.34 confirms that the lateral deflection predicted by FEA includes
both thermal bowing and P-∆ effects.
4.5.1.4. Failure Modes

Figure 4.35 shows the failure modes obtained from test and FEA. In the test the stud
failed by moving away from the furnace. However, FEA predicted the stud failure
towards furnace. Possible reason for this is the omission of loading eccentricity in
FEA. The contact surface of the stud and the loading plate will move towards the hot
side in the test due to thermal bowing. This will induce an eccentric loading towards
the hot side. This eccentric loading and the effect of neutral axis shift will cause the
stud to reverse its movement towards the furnace. Therefore in the final stage of the
test, there is a possibility of the stud to bend away from the furnace for most of the
tests. However, this eccentric loading was not considered in FEA and this resulted in
failures towards the furnace due to temperature gradient in all the simulations. This
behaviour is also experienced in the study of Zhao et al. (2005).
Another possible reason for this is explained by the location of local buckling failure.
The test and FEA showed local buckling along the stud near the failure time. In the
test, the stud was found to be failed due to the local buckling of hot flange at mid-
height in 118 minutes. However, without the use of a stabilize factor the ultimate
failure mode could not be achieved in FEA, since the analysis stopped with
numerical problems after 118 and 115 minutes for Studs 2 and 3, respectively. These
times were considered as the failure times of these studs and the failure times were
confirmed by the use of a stabilize factor. When stabilize factors were used, the
numerical problem was eliminated and FEA predicted the failure location as hot
flange at supports for Test 1. Severe local buckling was observed at both ends of the
stud. Similar behaviour was also observed in the numerical study of Zhao et al.
(2005).
4-47
The reason for this behaviour is due to the assumption in temperature distribution. In
Test the temperatures were high at mid-height compared to the supports since the
stud mid-height is close to the furnace due the thermal bowing. Hence the hot flange
is critical at mid-height in the test. However in FEA, the variation of temperature
distribution along the stud was not considered. Hence the hot flange is critical at
support. This will be further explained in Section 4.5.2. The stud in the test moved
away from the furnace when the stud is failed due to the local buckling of hot flange
at mid-height. However, in FEA the stud was bent towards hot side due to thermal
bowing, while it failed due to local buckling of hot flange at support.
Cold Side
Hot Side
(a) Test
Cold Side
Hot Side
(b) FEA
Figure 4.35: Failure Modes from Test and FEA for Stud 3 of Test 1 under
Transient State Conditions
4.5.1.5. Stress Distribution
The von Mises stress distribution of the stud at failure is shown in Figure 4.36. The
maximum stresses were observed in the cold flange near the support. However, it is
not possible to say that yielding occurred in cold flange only because at this time the
temperature dependant yield stress will also be higher at cold flange. It should be
noted that the strength (yield stress) of the steel is dependant on the applied
4-48
temperature values. Therefore the contours of the yield stress also would be similar
to Figure 4.28. A closer look at the support indicates that the von Mises stress
contours are in the horizontal direction in the web near the support, which are very
similar to the contours of the temperature distribution shown in Figure 4.28. This
indicates that yielding of the entire cross-section has occurred near the support (i.e,
von Mises stress reached the temperature dependant yield stress).
Figure 4.36: Stress Distribution from FEA for Stud 3 of Test 1 under Transient
State Conditions
Figure 4.37 (a) shows the variation of yield stress and von Mises stress with respect
to time for a node close to the support. The reduction in yield stress is very low up to
90 minutes when the temperature of the node is less than 300oC. However, beyond
90 minutes the yield stress deteriorates rapidly due to temperature rise. At the
beginning the von Mises stress depends on the stress developed due to applied
loading (load ratio of 0.2). When the increasing temperature distribution was applied
the von Mises stress increased gradually until it reached the temperature dependant
yield stress. After this, the von Mises stress could not increase beyond the yield
stress, and thus followed the path of temperature dependent yield stress.
4-49
Figure 4.37 (b) shows the variation of von Mises stress of all the nodes in the web
near the support. According to this figure, the von Mises stress reached the yield
stress after 95 minutes for majority of the nodes in the web element. After this time
the stress distribution had spread to the stronger side of the cross-section (towards
cold flange) and the stud survived another 20 minutes before the hot flange
underwent significant yielding at failure.
600
500
400
Stress (MPa)
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110 120
von MisesTime (min.)

Stress Yield Stress
(a) Variation of yield stress and von Mises stress
(b) Variation of von Mises stresses for all the nodes in the web near the support
Figure 4.37: von Mises Stress Results from FEA for Stud 3 of Test 1 under
Transient State Conditions
4-50
4.5.1.6. Variation of Applied Load
Figure 4.38 shows the variation of load with respect to total time according to test
and FEA. The term total time is used here which includes the time taken for Steps 1
and 2. Step 1 of the analysis (application of load) is shown in the first minute. In the
actual test, comparatively more time was allowed for this step. This load was
maintained in both test and FEA and the temperature was applied in Step 2. The
applied load of 15 kN/stud was fluctuating in the actual test since the load was
controlled manually using hydraulic jacks. However, a smooth line was obtained for
FEA. In Step 2 the temperature was increased using the measured temperature
profiles obtained from the full scale fire test. The actual test was stopped after 118
minutes when the stud could no longer sustain the applied load of 15 kN. In FEA the
command given was to maintain the applied load while increasing the temperature.
However, after 115 minutes ABAQUS could not sustain the load of 15 kN at that
high temperature and the load started to decrease rapidly due to the severe local
buckling observed at the support. Therefore the failure time was recorded as 115
minutes in FEA.
16
14
12
Step 2 Failure
10
Load (kN)
8 Step 1
6
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130
Total Time (min.)
Experiment FEA
Figure 4.38: Variation of Applied Load from Test and FEA for Stud 3 of Test 1
4-51
4.5.2. Finite Element Analysis under Steady State Conditions
Finite element analyses were also conducted under steady state conditions to confirm
the failure time under transient state conditions. In addition to this, it is convenient to
compare these results (loads obtained from FEA under steady state conditions) with
design calculations. The stud with the vertical plasterboard joint against it
experienced higher temperatures than other studs in the wall panel. Therefore this
stud was expected to fail before other studs. In the FEA under transient conditions
both central studs were modelled and it was proved that the stud with the vertical
plasterboard joint failed earlier. Therefore this critical stud was considered in the
finite element analyses under steady state conditions.
Initially an eigen buckling analysis of the stud at ambient temperature was

performed, in which the buckling modes were obtained and the deformation profile
of the lowest buckling mode was used to input the stud initial imperfections.
Alternative buckling modes were also considered as discussed in Section 4.4.5.
However, they resulted in the same ultimate load.
The nonlinear finite element analyses of the stud under steady state conditions were
performed in two steps. In the first step, temperatures profiles were applied and in
the second step the load was applied. The Riks off method and the Riks on method
were used for the first and second steps, respectively. The Riks on method was used
in the loading (second) step to determine the ultimate load which is different to the
analysis under transient state conditions. In the transient state analyses Riks off
method was used for loading step in order to achieve the target load (not the ultimate
load).
The fire model under steady state conditions was created similar to the fire model
under transient state conditions. However, the method of analyses was different as
explained before. The steel stud temperatures were based on the average temperature
distribution for the critical stud obtained from the experimental study. This analysis
was also conducted with the temperature distribution at stud mid-height.
4-52
The analysis was conducted in a time frame to arrive at a load ratio versus time curve
for each test as shown in Figure 4.39 (a). The variation of load ratio with respect to
the hot flange temperature was shown in Figure 4.39 (b). In FEA the failure time was
obtained by finding the time at which the load ratio was equal to 0.2. Table 4.6
shows the failure times and hot flange temperatures (at failure) obtained from FEA
using both the average temperature and the mid-height temperature distributions.
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 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)
Stud 3 Average Stud 3 Mid-height
(a) Variation of load ratio with time
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 100 200 300 400 500 600 700
o
Temperature ( C)
(b) Variation of load ratio with hot flange temperature
Figure 4.39: FEA Results for Stud 3 of Test 1 under Steady State Conditions
4-53
under Steady State Conditions
Failure Time (min.) Max. Temperature at Failure (oC)

Temperature
used in FEA
Test FEA Test FEA
Average 115 656 602

118
Mid-height 115 664 597
Figure 4.40 shows the stress distribution obtained from FEA under steady state
conditions, which indicates yielding near the support. This is very similar to the
results obtained from FEA under transient state conditions.
Figure 4.40: Stress Distribution from FEA for Stud 3 of Test 1 under Steady
State Conditions
In FEA non-uniform temperature distributions were applied to the stud cross-section

as explained before. In this case bending moment will be developed at mid-height
due to the net effect of thermal bowing and neutral axis shift. Hence the cold flange
at mid-height will be subjected to compressive stresses due to axial compression and
bending. On the other hand the hot flange at mid-height will be subjected to axial
compressive stress and tensile stress due to bending. Therefore the cold flange at
mid-height is stressed more compared to the hot flange at mid-height in LSF studs
subjected to non-uniform temperature distributions across the stud cross-section.
4-54
On the other hand the bending moment will be developed at the support only due to
the neutral axis shift. Hence the hot flange at the support will be subjected to
compressive stresses due to axial compression and bending. However, the cold
flange at support will be subjected to axial compressive stress and tensile stress due
to bending. Therefore the hot flange at the support is stressed more compared to the
cold flange at the support in LSF studs subjected to non-uniform temperature
distributions across the stud cross-section. However, it should be noted that the yield
stress of steel reduces with time and hence the strength of cold flange is also higher
than the hot flange. Therefore the failure locations of LSF studs in FEA can be either
in the cold flange at mid-height or the hot flange at support. This depends on the
induced stresses and material yield strength at elevated temperatures.
Figure 4.41 shows the failure modes obtained from FEA under steady state
conditions at different times during the test. The failure location for Stud 3 of Test 1
changes from cold flange at mid-height to hot flange at support when elevated
temperatures are increased. Initially when non-severe elevated temperatures are
applied, the stud will act as a column with initial geometric imperfection about the
major axis. In this case the stud failure will be initiated by the global imperfection
and hence the failure occurred at mid-height cold flange. Therefore the fire resistance
of the LSF studs with higher load ratios are much less and they will fail even in non-
severe elevated temperatures. In this case the temperature gradient and corresponding
thermal bowing plays a major role in initiating the failure. When the stud is subjected
to severe elevated temperatures then the yield stress of hot flange will reduce
considerably and hence it outplays the effect of thermal bowing. In this case the
failure was at support hot flange. However, it should be noted that when the cold
flange temperature is very high as hot flange temperature, then the failure can be at
mid-height cold flange.
The failure modes obtained from FEA under steady and transient state conditions are
shown in Figure 4.42. This indicates that the predicted failure modes are very similar
in both types of analyses. The stud failed due to severe local buckling at the support
with large deformation of the hot flange
4-55
(a) t = 30 min.
(b) t = 60 min.
(c) t = 90 min.
(d) t = 100 min.
(e) t = 110 min.
(f) t = 115 min.
(g) t = 118 min.

Figure 4.41: Failure Modes from FEA for Stud 3 of Test 1 under Steady State
Conditions
4-56
Cold Flange Cold Flange
Hot Flange
Hot Flange
Local buckling failure near
the support
(b) Transient state
(a) Steady state
condition
condition
Figure 4.42: Failure Modes from FEA for Stud 3 of Test 1
Figure 4.43 shows the applied axial compression load versus lateral deflection at
mid-height under steady state conditions at different times during the test. Each curve
starts at a non-zero value of lateral deflection. This is the thermal bowing caused by
the temperature difference across the stud. The ultimate load decreased with time due
to increasing temperature distribution. Figure 4.43 can also be used to predict the
failure time of Test 2. When t=110 minutes, the predicted ultimate load is 30 kN.
This means that the same panel will fail at 110 minutes when the load ratio is 0.4
(Test 2). This failure time (110 min.) agrees well with the FEA results of Test 2.
80
70
60
50
40
30
20
10
0
0 5 10 15 20 25 30 35 40 45
t=10 min t=20 min. t=30 min. t=40 min. t=50 min.
t=60 min. t=70 min. t=80 min. t=90 min. t=100 min.
t=110 min. t=115 min. t=118 min.
Figure 4.43: Applied Load versus Lateral Deflection Plots from FEA for Stud 3
of Test 1 under Steady State Conditions
4-57

The model was created and analysed as for Test 1, but with a load ratio of 0.4. The
measured average temperature profiles shown in Figure 4.44 were used in the
analysis of Test 2 studs. Both central studs were used in FEA under transient state
conditions. The hot flange temperature of the third stud increased rapidly soon after
the right side glass fibre was burnt. In a similar way the hot flange temperature of the
second stud increased rapidly after a few minutes. Therefore Stud 3 was expected to
fail before Stud 2 in the finite element analyses.
500
400
Temperature (oC)
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)
of Test 2
Figure 4.45 shows the axial deformation and lateral deflection curves from FEA
under transient state conditions for Test 2. Rapid fall in axial deformation confirms
the failure at the support due to severe local buckling. This is similar to the behaviour
observed in Test 1. Without the use of a stabilize factor, the numerical problem was
experienced after 113 and 111 minutes for Studs 2 and 3, respectively. Therefore the
stabilize factor was used thereafter to confirm the failure time. From the beginning of
the fire test, the wall was observed to be bending towards the furnace. Near the
failure, the lateral deflection started to reverse its direction and finally the studs bent
away from the furnace. However, this was not predicted by FEA due to the omission
in loading eccentricity and change in failure location as discussed in Section 4.5.1.4.
4-58
20.00
10.00
0.00
-10.00
-20.00
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)
20.00
0.00
-20.00
-40.00
-60.00
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)
Conditions
4-59
Figure 4.46 shows the variation of applied load with respect to time for Stud 3 of
Test 2. The behaviour is very similar to Test 1 as discussed in Section 4.5.1.6. This
figure confirms the failure of Stud 3 at 111 minutes for Test 2 in FEA under transient
conditions.
30
25
20
Load (kN)
15
10
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Total Time (min.)
Experiment FEA
Figure 4.46: Variation of Applied Load from Test and FEA under Transient
State Conditions for Stud 3 of Test 2
Figures 4.47 (a) and (b) show the load ratio versus time and hot flange temperature
curves based on average and mid-height temperature distributions. These are based
on the results from FEA under steady state conditions. Table 4.7 shows the failure
time and hot flange temperatures predicted by FEA at a load ratio of 0.4.
The behaviour of Test Specimen 2 was very similar to that of Test Specimen 1. In the
test, Studs 1 and 4 did not fail in Test Specimen 2 nor experienced any local
buckling. Studs 2 and 3 bent severely about the major axis and showed local
buckling of the cross-section close to the mid-height. Local buckling waves were
observed in the web even away from mid-height. This observation also confirms that
local buckling of studs occurred first, which led to their ultimate failure. The failure
location was predicted as support and mid-height for transient and steady state
analyses, respectively (see Figure 4.48). It should be noted that, as discussed in
Figure 4.41, the failure location in FEA changes from mid-height to the support with
decreasing load ratios. Hence for a load ratio of 0.4 the failure location can be either
4-60
mid-height or support depending on their induced stress and temperature dependant

yield stress.
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 100 200 300 400 500 600
o
Tem perature ( C)
4-61

Temperature
used in FEA
Test FEA Test FEA
Average 110 481 505

108
(a) FEA under transient state conditions
(b) Test
(c) FEA under steady state conditions
Figure 4.48: Failure Modes from Test and FEA under Transient and Steady
State Conditions for Stud 3 of Test 2
4-62
The construction of Test Specimen 3 was identical to that of Test Specimen 2 in all
respects, except for rock fibre insulation used as external insulation. The FE model
was created and analysed as for Tests 1 and 2. The load ratio was 0.4. Figure 4.49
shows the time-temperature profiles of hot and cold flanges of the central studs.
These are the average measured temperature profiles used in the numerical
simulations of Test 3. Both central studs were used in FEA under transient state
conditions.
600
500
Temperature (oC)
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min.)
of Test 3
Figure 4.50 shows the axial deformation and lateral deflection versus time curves
from FEA under transient state conditions for Test 3. Without the use of a stabilize
factor the numerical problem was experienced after 132 and 136 minutes for Studs 2
and 3, respectively. Therefore a stabilize factor was used to confirm the failure time.
A rapid fall in axial deformation was experienced at failure due to severe local
buckling at the support. From the beginning of the fire test, the wall specimen was
observed to be bending towards the furnace. This continued until failure and resulted
in failing towards the furnace. In FEA also the stud failed towards the furnace.
4-63
20.00
10.00
0.00
-10.00
-20.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min.)
20.00
0.00
-20.00
-40.00
-60.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min.)
Conditions
4-64
Figure 4.51 shows the variation of applied load with respect to time in test and FEA.
The behaviour is very similar to Test 1 as discussed in Section 4.5.1.6. The test was
stopped after 134 minutes since the specimen could not support the applied load of
30 kN after this time. Figure 4.51 confirms the failure at 132 minutes in FEA under
transient conditions.
30
25
20
Load (kN)
15
10
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Total Time (min.)

Experiment FEA
Figure 4.51: Variation of Applied Axial Compression Load from Test and FEA
under Transient State Conditions for Stud 2 of Test 3
Figure 4.52 shows the load ratio versus time and hot flange temperature curves based
on average and mid-height temperature distributions of studs according to FEA under
steady state conditions. Table 4.8 shows the failure time and hot flange temperatures
predicted by FEA.
The failure modes obtained from FEA under transient and steady state conditions
were similar to the test failure mode. The failure location was predicted to be at the
support for transient state analysis. However, under steady state conditions, the
failure locations were predicted as both mid-height and support at 131 minutes (see
Figure 4.53). Severe local buckling was observed near mid-height and support. Fire
tests also showed that the stud failed due to local buckling at both mid-height and
support. In both test and FEA, the stud failed towards the furnace.
4-65
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Tim e (m in.)
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 100 200 300 400 500 600
Tem perature ( oC)
4-66

Temperature
used in FEA
Test FEA Test FEA
Average 131 548 519

134
(a) FEA under transient state conditions
(b) Test
(c) FEA under steady state conditions
Figure 4.53: Failure Modes from Test and FEA for Stud 2 of Test 3
4-67
4.8. Simulation of Tests Performed by Kolarkar (2010) under Fire Conditions

Kolarkar (2010) performed eight tests in the fire research laboratory of Queensland
University of Technology to evaluate the fire resistance of full scale load-bearing
wall panels. Among these tests, one test could not be used in further studies since an
unexpected furnace failure was experienced during the test. Therefore the remaining
seven fire tests as listed in Table 4.9 were used to validate the finite element models
developed in the current study. The composite panel with rock fibre as insulation
(Test 6*) was considered to have failed earlier in the test than expected. This is due
to the insufficient space for expansion between the loading frame and the panel.
Table 4.9: Fire Tests Conducted by Kolarkar (2010)

Load Failure
Test Configuration Insulation
Ratio Time (min.)
1* None 0.2 53
2* None 0.2 111
3* Glass Fibre 0.2 101
4* Rock Fibre 0.2 107

Cellulose
5* 0.2 110
Fibre
6* Rock Fibre 0.2 136#
Cellulose
7* 0.2 124
Fibre
( # ) – Experimental error
Earlier sections described the finite element analyses (FEA) conducted under both
steady and transient state conditions. However, it was observed that both these
methods provided similar failure times and hot flange temperatures at failure.
Therefore it was decided to use either steady or transient state condition. Finite
element analyses under steady state conditions provided an additional output, the
ultimate load capacity at different non-uniform temperatures. Therefore in order to
simulate the fire tests of Kolarkar (2010) the finite element analyses were conducted
under steady state conditions using both the average and the mid-height temperature
distributions of the stud based on the measurements in the fire tests.
4-68
4.8.1. Simulation of Test 1* (1x1 Plasterboard)

The LSF wall panel was lined on both sides by a single layer of plasterboard. The top
ends of the studs were not connected to the upper track using screws. Instead friction
fit connections were adopted to allow for the vertical expansion of the studs. The test
was stopped after 53 minutes as the wall specimen could no longer sustain the
applied load. The vertical plasterboard joint was placed against Stud 2 and hence this
was identified as the critical stud as discussed earlier. Therefore finite element
analyses of Stud 2 were conducted using both its mid-height and average temperature
distributions shown in Figure 4.54. During the test the Edcar software crashed for a
period of 8 minutes from 9 to 17 minutes resulting in the loss of readings. However,
the temperatures at the 15th minute were assumed to be the same as at the 9th minute.
This is a reasonable assumption considering the previous experimental
measurements.
700
600
500
Temperature ( o C)
400
300
200
100
0
0 10 20 30 40 50 60
Time (min.)
Hot Flange Mid-height Hot Flange Average

Cold Flange Mid-height Cold Flange Average
Figure 4.54: Temperature Profiles Used in FEA for Stud 2 of Test 1*
Figures 4.55 (a) and (b) show the variation of load ratio with respect to time and hot
flange temperature, respectively. The failure time was obtained by finding the time at
a load ratio of 0.2. This was found to be 53 minutes when the average temperature
distribution was used. Table 4.10 compares the failure times and the hot flange
temperatures obtained from FEA and test.
4-69
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 10 20 30 40 50 60
Tim e (m in.)
1.0
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0 100 200 300 400 500 600 700
o
Tem perature ( C)
Figure 4.55: FEA Results for Stud 2 of Test 1* under Steady State Conditions
4-70
Test 1* under Steady State Conditions

Temperature
used in FEA
Test FEA Test FEA
Average 53 595 593

53
Figure 4.56 shows the failure modes obtained from test and FEA under steady state
conditions for Stud 2. The reduced lateral support of the studs at elevated
temperatures caused the central studs to undergo global flexural torsional buckling
about the minor axis in the experimental study. However, in FEA the lateral support
was assumed until failure. Hence the stud did not buckle about its minor axis. Local
buckling was observed near the support in the test as shown in Figure 4.56 (a). In
FEA also the stud failed due to local buckling near the support (see Figure 4.56 (b)).
(a) Test
Local
buckling
(b) FEA
for Stud 2 of Test 1*
4-71
4.8.2. Simulation of Test 2* (2x2 Plasterboards)
Test Specimen 2* was constructed with two layers of plasterboard on either side of
the steel frame. The top ends of the studs were not connected to the upper track using
screws. Instead friction fit connections were adopted (Kolarkar, 2010). The test wall
specimen continued to maintain the applied load up to 111 minutes when it suddenly
failed. The vertical plasterboard joint was placed against Stud 2 and hence this stud
was identified as the critical stud. Therefore finite element analyses of Stud 2 were
conducted using both its mid-height and average temperature distributions shown in
Figure 4.57. When the stud was analyzed using average temperature distributions
under steady state conditions, the load ratio was greater than 0.2 at 111 minutes.
Therefore the average temperature values obtained from test were extrapolated after
111 minutes to determine the failure time.
700
600
500
Temperature ( o C)
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)
flange temperature, respectively. The stud failed in 115 minutes at a load ratio of 0.2
when the average temperature distribution was used. Table 4.11 compares the failure
times and the hot flange temperatures obtained from FEA and test.
4-72
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 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 100 200 300 400 500 600 700
o
Tem perature ( C)
4-73

Temperature
used in FEA
Test FEA Test FEA
Average 115 555 606

111
Figure 4.59 shows the failure modes obtained from FEA under steady state
conditions and test for Stud 2. In the test the studs were laterally displaced at the top
end. The friction fit joints provided at the top end of each stud failed to prevent the
slipping of the studs in the lateral deflection near failure. However, in FEA pinned
support conditions were assumed for both ends. Hence the stud did not move
laterally in FEA. The central studs also displayed distortional buckling in the top
portion of their heights and this was predicted well in FEA as shown in Figure 4.59.
Distortional (a) Test

buckling
(b) FEA
4-74
4.8.3. Simulation of Test 3* (Cavity Insulation - Glass Fibre)
The construction of Test Specimen 3* was very similar to that of Test Specimen 2*.
The differences were the use of cavity insulation and the connection of stud to top
track. After fixing the two plasterboards on the fire side, the cavity in the wall
between the studs was filled with two layers of 50 mm thick glass fibre mats
compressed to 90 mm thickness (depth of the cavity). Test Specimen 3* was
subjected to heat in the furnace for 101 minutes. The vertical plasterboard joint was
placed against Stud 2 and hence this stud was identified as the critical stud. Therefore
finite element analyses of Stud 2 were conducted using both its mid-height and
average temperature distributions shown in Figure 4.60.
700
600
500
Temperature ( o C)
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)
4-75
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 10 20 30 40 50 60 70 80 90 100 110
Tim e (m in.)
Stud 2 Average. Stud 2 Mid-height
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 100 200 300 400 500 600 700 800
o
Tem perature ( C)
4-76

Temperature
used in FEA
Test FEA Test FEA
Average 100 645 631

101
The plasterboards on the ambient side were intact giving good lateral support to the
studs throughout the test. In the test Studs 2 and 3 failed by local buckling
(compressive failure) close to the mid-height of the wall. The hot flange and the web
of Stud 2 buckled locally as shown in Figure 4.62 (a). However, in FEA the stud
failed by local buckling of the hot flange and web at support as shown in Figure 4.62
(b). The difference in the predicted location is due to the assumption of uniform
temperature distribution along the stud. This is further explained in Section 4.5.2.
(a) Test
Local
buckling
(b) FEA
4-77
4.8.4. Simulation of Test 4* (Cavity Insulation - Rock Fibre)
Construction of Test Specimen 4* was identical to that of Test Specimen 3* in all

respects, except for rock fibre being used as cavity insulation instead of glass fibre.
Test Specimen 4* was subjected to heat in the furnace for 107 minutes. Both fire side
plasterboards had fallen off at the end of the test. The rock fibre insulation was
almost fully intact through out the test. The ambient side plasterboards although in
good condition up to the end of the test cracked up when the wall failed by bowing in
the outward direction. The vertical plasterboard joint was placed against Stud 2 and
hence this stud was identified as the critical stud. Therefore finite element analyses
of Stud 2 were conducted using both its mid-height and average temperature
distributions shown in Figure 4.63.
700
600
500
Temperature ( o C)
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)
4-78
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 10 20 30 40 50 60 70 80 90 100 110
Tim e (m in.)
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 100 200 300 400 500 600 700
Tem perature (oC)
4-79
Table 4.13: Failure Time and Temperature Predicted by FEA for Stud 2
of Test 4* under Steady State Conditions

Temperature
used in FEA
Test FEA Test FEA
Average 105 648 633

107
Torsional failure was fully prevented in Studs 2 and 3 on account of the lateral
support provided by the plasterboards on both sides. The tracks were in good
condition and maintained good contact with the studs throughout the fire test. Studs 2
and 3 failed by local buckling of the hot flange close to the mid-height as shown in
Figure 4.65 (a). In FEA the stud failed by distortional buckling at mid-height and
local buckling at support as shown in Figure 4.65 (b).
(a) Test
Local
buckling
Distortional
buckling
(b) FEA
4-80
4.8.5. Simulation of Test 5* (Cavity Insulation - Cellulose Fibre)

Test Specimen 5* was built similar to Test Specimens 3* and 4*, but with cellulose
fibre as cavity insulation. After fixing the two layers of plasterboard on the fire side,
the cavity was fine sprayed with plain water to just moisten the cavity facing surface
of the plasterboard. This was quickly followed by a wet spray of cellulose fibre.
Spraying was stopped after the complete filling of the wall cavity with the insulation
material. The test procedures were similar to the previous tests and the fire test was
stopped at 110 minutes when the specimen failed to maintain the load. The vertical
plasterboard joint was placed against Stud 3 (not Stud 2 as other tests performed by
Kolarkar (2010)) and hence this stud was identified as the critical stud. Therefore
finite element analyses of Stud 3 were conducted using both its mid-height and
average temperature distributions shown in Figure 4.66.
800
700
600
Temperature ( o C)
500
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)

4-81
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 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 100 200 300 400 500 600 700 800
Tem perature ( oC)
4-82
Failure Time (min.) Max. Temperature (oC)

Temperature
used in FEA
Test FEA Test FEA
Average 109 652 632

110
Middle 104 780 639
In Studs 2 and 3 the local buckling of hot flange and web was observed in the test
(see Figure 4.68 (a)). Stud 3 also displayed local crushing of the cross-section near
the top support in the frame. In FEA when uniform temperature distributions were
assumed along the stud length, the critical hot flange is located at support and not at
the mid-height. This is because the hot flange at support will be subjected to the axial
compressive stress and the compressive stress due to bending (due to neutral axis
shift. On the other hand the hot flange at mid-height will be subjected to axial
compressive stress and tensile stress due to bending (due to the net effect of thermal
bowing and neutral axis shift). Hence in FEA the local buckling at support was
predicted as shown in Figure 4.68 (b).
Local (a) Test Local

buckling buckling
(b) FEA
4-83
4.8.6. Simulation of Test 6* (Composite Panel - Rock Fibre)

The construction of Test Specimen 6* required the rock fibre insulation to be laid
outside the cavity (external insulation) between the base and face layer plasterboards
on either side of the wall. After 132 minutes, the studs reached the free expansion
limit provided for it in the loading frame. Hence the loading plate of the studs came
into contact with the lower beam and the jack became non-functional. This meant
that beyond 132 minutes the load on the studs could have increased to more than 15
kN due to the prevention of studs thermal expansion. The specimen failed at 136
minutes after which the test was stopped.
The vertical plasterboard joint was placed against Stud 2 and hence this stud was
identified as the critical stud. Therefore finite element analyses of Stud 2 were
Therefore the average temperature values obtained from test were extrapolated
beyond 136 minutes to determine the failure time.
700
600
500
Temperature ( o C)
400
300
200
100
0
0 20 40 60 80 100 120 140 160
Time (min.)
4-84
when the average temperature distribution was used in FEA. This predicted failure
time is much higher than the test failure time of 136 minutes due to the shortcoming
in the test mentioned earlier. Table 4.15 compares the failure times and the hot flange
temperatures obtained from FEA and test.
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 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160
Tim e (m in.)
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 100 200 300 400 500 600 700
o
Tem perature ( C)
4-85

Temperature
used in FEA
Test FEA Test FEA
Average 154 509 606

136
In the test the tracks were seen to maintain good contact and connection with the
studs. The studs did not show any torsional or flexural buckling about the minor axis
due to adequate lateral support. Studs 2 and 3 exhibited local buckling of flange and
web near the mid-height (see Figure 4.71 (a)). In FEA also severe local buckling was
experienced near the mid-height as shown in Figure 4.71 (b). At this failure time the
cold flange temperature was also too severe and close to the hot flange temperature.
Hence the failure of stud in FEA this time was due to the local buckling of cold
flange at mid-height and not due to the local buckling of hot flange at support. In
other words when both hot and cold flanges have similar strengths, (the bending
stress due to) the net effect of thermal bowing and neutral axis shift out plays (the
bending stress due to) the effect of neutral axis shift.
(a) Test
(b) FEA
4-86
4.8.7. Simulation of Test 7* (Composite Panel - Cellulose Fibre)

The fixing of the base layer plasterboards onto the steel frame and the strips along
the periphery to develop the cavity was identical to that conducted in Test Specimen
6*. In this specimen 25 mm thick cubical spacers cut from plasterboard strips were
also positioned in the field of cavity. This was done to provide a firm support to the
face layer plasterboard which was subsequently attached after wet spraying the
cavity with cellulose fibre insulation. The test set-up and procedure were similar to
the previous tests. The wall specimen failed after 124 minutes for fire exposure.
The vertical plasterboard joint was placed against Stud 2 and hence this stud was
identified as the critical stud. Therefore finite element analyses of Stud 2 were
Therefore the average temperature values obtained from test were extrapolated
beyond 124 minutes to determine the failure time.
700
600
500
Temperature ( o C)
400
300
200
100
0
0 20 40 60 80 100 120 140
Time (min.)

4-87
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 10 20 30 40 50 60 70 80 90 100 110 120 130
Tim e (m in.)
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 100 200 300 400 500 600 700
o
Tem perature ( C)
4-88

Temperature
used in FEA
Test FEA Test FEA
Average 129 558 604

124
In the test the tracks were in good condition providing good support to the studs.
Studs 2 and 3 were severely damaged by local buckling of the hot flange and web as
shown in Figure 4.74 (a). In FEA also local buckling waves were observed over the
entire length of the stud (see Figure 4.74 (b)) even though the local buckling was
severe near the support.
(a) Test
(b) FEA
4-89
4.9. Discussion
4.9.1. Finite Element Analyses under Transient State Conditions

The analyses under transient state conditions were performed in a time frame to
obtain a deformation curve for each test. The failure modes and stress distributions
were also obtained from these analyses. The accuracy of the developed finite element
models was verified by using the deformation curves and failure times of LSF wall
panels from the fire tests of this research.
In the first two tests the studs failed by moving away from the furnace. On the other
hand, they failed towards the furnace in the third test. However, in FEA the studs
always failed by moving towards the furnace due to the omission of loading
eccentricity in the finite element modelling. This is also due to the variation in hot
flange failure location in test and FEA (mid-height versus support) due to the
assumption of uniform temperature distribution along the stud length. Both test and
FEA showed local buckling waves along the stud near failure. The ultimate failure
was caused by local buckling in both tests and FEA. Without the use of a stabilize
factor FEA under transient state conditions appeared to terminate due to numerical
problems near failure. Therefore a suitable stabilize factor was used to confirm the
failure time. Similar failure times were obtained with and without the usage of
stabilize factors. However, the use of a stabilize factor helped to obtain the failure
modes and to confirm the failure time in FEA as discussed in Section 4.5.1.2. Table
4.17 compares the failure times obtained from tests and FEA under transient state
conditions. The lower value of Studs 2 and 3 was taken as the governing failure time.
These failure times agreed well with the test results.
Table 4.17: Failure Time Predicted by FEA under Transient State Conditions
Failure Time (min.)

Load
Test Configuration Insulation FEA
Ratio Test
Stud 2 Stud 3
1 Glass Fibre 0.2 118 118 115
2 Glass Fibre 0.4 108 113 111
3 Rock Fibre 0.4 134 133 136
4-90
Kaitila (2002) and Zhao et al. (2005) conducted finite element analyses under
transient state conditions from which the failure time and deformation curves were
obtained. However, Kaitila’s (2002) finite element model of LSF wall studs with
non-uniform temperatures was not validated against relevant experimental results.
Therefore the accuracy of Kaitila’s (2002) numerical study could not be directly
compared with the accuracy of the results obtained in the current numerical study. In
the current study the failure time and deformation curves obtained from FEA are
very good compared to Zhao et al.’s (2005) numerical studies of LSF walls under
transient conditions.
One of the possible reasons for this is the use of accurate reduction factors for
mechanical properties at elevated temperatures. Zhao et al. (2005) used the reduction
factors for yield stress and elastic modulus based on Eurocode 3 Part 1.2 (ECS,
2005). However, in the current study the reduction factors were based on Dolamune
Kankanamge and Mahendran (2011), which are more accurate for the Australian
cold-formed steels used in this study. Another reason could be the correct assumption
of temperature profiles. The temperature profiles used in the numerical simulations
of Zhao et al.’s (2005) study correspond to the average measured temperatures at the
mid-height of the two studs. However, in this study the average measured
temperatures (top, mid-height and bottom) of the stud along its length were
considered. This eliminated the sharp rise and fall in the deformation curves with
time. In order to be on the conservative side the stud with the vertical plasterboard
joint against it was considered when deciding the failure time and maximum
temperature at failure. This is because the stud with the vertical joint against it had
the highest temperature compared to other studs due to the slight opening of the joint
at elevated temperatures.
4.9.2. Finite Element Analyses under Steady State Conditions

The FEA results under steady state conditions can be used to determine the ultimate
load of the stud for a particular temperature distribution over the stud cross section
(at a particular time). Therefore this type of analysis is convenient to compare the
results obtained from design calculations. The critical stud which had the vertical
4-91
plasterboard against it was analysed under steady state conditions as discussed

before. Figure 4.75 shows the variation of load ratios with respect to time using the
average and mid-height temperature distributions. Each test produced different load
ratio versus time graphs due to the variation in the assumed temperature profiles. The
load ratio did not reduce much for the first few minutes as expected. This is the time
when the plasterboard was dehydrated and hence the stud was protected from fire
and the maximum temperature in the stud was less than 100oC.
Following this the load ratio reduced rapidly with time due to mainly two reasons.
First reason was the increasing temperature with time, which leads to lower ultimate
capacity of LSF wall stud due to deteriorated mechanical properties. Second reason
was the thermal bowing and associated bending moment due to the development of
temperature difference between hot and cold flanges.
Near the failure time the lateral restraint provided by the plasterboards was taken into
account in FEA because the base layer of the fire side plasterboard maintained its
integrity until failure. However, some parts of the fire side face layer plasterboards
fell off and this resulted in a rapid increase in temperatures in the stud. Therefore a
rapid fall in load ratio was experienced in the load ratio versus time curve as shown
in Figure 4.75. This phase covers load ratios of 0.2 and 0.4. Therefore the predicted
failure times for load ratios of 0.2 and 0.4 were closer in most of the tests. In other
words the failure time difference between these load ratios was only few minutes for
all the tests.
There is a slight variation in load ratio versus time graphs when using the average
and mid-height stud temperature distributions. This is due to the slight variation in
the assumed temperature profiles. The mid-height stud temperature was more than or
equal to the average stud temperature for most of the tests. Therefore at any
particular time the load ratio obtained based on mid-height stud temperatures was
less than or equal to the load ratio obtained based on average stud temperatures.
4-92
1.0
0.9
0.8
0.7
0.6
Load Ratio
0.5
0.4
0.3
0.2
0.1
0.0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160
Tim e (m in.)
1x1 0.2 2x2 0.2 CI GF 0.2 CI RF 0.2 CI CF 0.2

CP RF 0.2 CP CF 0.2 CP GF 0.2 CP GF 0.4 CP RF 0.4
(a) Using average temperature along the stud
0.9
0.8
0.7
0.6
Load Ratio
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160
Tim e (m in.)
1x1 0.2 2x2 0.2 CI GF 0.2 CI RF 0.2 CI CF 0.2

(b) Using mid-height temperature of the stud
Figure 4.75: Variation of Load Ratio with Time under Steady State Conditions
4-93
Figures 4.76 (a) and (b) show the variation of load ratios with respect to the hot
flange temperatures. The load ratio versus hot flange temperature curves for different
tests merged well. These figures are very similar to the graphs obtained by Feng et al.
(2003b). For a given load ratio, the failure temperatures in a stud for all the
specimens are reasonably close. This means that structurally similar studs will fail at
the same critical failure temperature regardless of the number plasterboards and
insulation characteristics. The significant effect of plasterboards and different types
of insulation is simply to delay the time to reach that critical temperature in steel
studs. The curves will agree even better if cavity and externally insulated wall panels
are plotted separately.
A small reduction in the load ratio was observed even at lower temperatures of 100-
300oC, where there is not a reduction in yield stress. The reasons for this behaviour
can be mainly explained by two factors. First the elastic modulus has decreased even
at these low temperatures. This reduces the stiffness of the steel stud. Secondly, even
in this initial phase, the stud experienced some temperature deviation across the stud.
This results in larger lateral deflections and hence higher bending moment in the
stud. These effects caused the reduction in load ratios even at temperatures below
300oC.
The accuracy of the developed finite element models was investigated by comparing
the failure times of LSF members from FEA and test. In FEA the failure time was
obtained by finding the time at which the load ratio was equal to 0.2 and 0.4. Most of
the tests with a load ratio of 0.2 failed when the hot flange temperature was about
600oC in FEA. The specimens under a load ratio of 0.4 failed at about 520oC. This
shows that the current limiting temperature method based on 350oC is too
conservative and will lead to an over-design of LSF walls.
4-94
1.0
0.9
0.8
0.7
0.6
Load Ratio
0.5
0.4
0.3
0.2
0.1
0.0
0 100 200 300 400 500 600 700 800
o
Tem perature ( C)
1X1 0.2 2X2 0.2 CI GF 0.2 CI RF 0.2 CI CF 0.2

(a) Using average temperature along the stud
0.9
0.8
0.7
0.6
Load Ratio
0.5
0.4
0.3
0.2
0.1
0
0 100 200 300 400 500 600 700 800
Tem perature ( oC)
1X1 0.2 2X2 0.2 CI GF 0.2 CI RF 0.2 CI CF 0.2

(b) Using mid-height temperature of the stud
Figure 4.76: Variation of Load Ratio with Hot Flange Temperature from FEA
4-95
Table 4.18 compares the predicted FEA failure times with those obtained from the
full scale fire tests performed at QUT. Some variations were observed in the failure
times when using average and mid-height temperatures as shown in Figure 4.77.
When mid-height temperatures were used, conservative results were obtained with a
maximum deviation of 10%. However, the FEA results agreed well with the test
results when average temperatures of the stud were used. Except the composite panel
with rock fibre, the difference between the failure times from FEA and tests does not
exceed 5% (within 5 minutes error), which is a considerable improvement compared
to previous researches (Zhao et al., 2005). This is due to the accurate assumptions of
temperature profiles and mechanical properties used in FEA. Kaitila’s (2002) and
Feng et al.’s (2003b) numerical studies were performed under transient and steady
state conditions, respectively. However, their models were not validated using the
full scale fire tests of LSF wall panels. Therefore the accuracy of their numerical
studies cannot be directly compared with the accuracy of the current studies.
Table 4.18: Failure Times Predicted by FEA under Steady State Conditions
Failure Time (min.)

Load
Test Configuration Insulation FEA
Ratio Test
Avg. Mid.Ht.
1 Glass Fibre 0.2 118 115 115
2 Glass Fibre 0.4 108 110 107
3 Rock Fibre 0.4 134 131 131
1* None 0.2 53 53 52
2* None 0.2 111 115 110
3* Glass Fibre 0.2 101 100 90
4* Rock Fibre 0.2 107 105 99

Cellulose
5* 0.2 110 109 104
Fibre
6* Rock Fibre 0.2 136# 154 153
Cellulose
7* 0.2 124 129 123
Fibre
( * ) - Tests conducted by Kolarkar (2010)
( # ) – Experimental error
4-96
160
Experimental error
140
120
FEA Failure Time (min.)
100 + 5%
80 - 5%
60
40
20
0
0 20 40 60 80 100 120 140 160
Test Failure Tim e (m in.)
(a) Using Average Temperature along the Stud
160
Experimental error
140
120
FEA Failure Time (min.)
+ 10%
100
80
- 10%
60
40
20
0
0 20 40 60 80 100 120 140 160
Test Failure Tim e (m in.)
(b) Using Mid-height Temperature of the Stud
Figure 4.77: Comparison of Failure Times from FEA and Tests
4-97
4.10. Conclusions
This chapter has described the finite element models developed to simulate the fire
tests of LSF wall panels conducted at QUT. The finite element analyses were used to
fully understand the improvements offered by the new composite system. The use of
accurate numerical models as described in this chapter allowed the inclusion of
various complex thermal and structural effects such as thermal bowing, local
buckling and shift of neutral axis at elevated temperatures. The finite element
analyses were conducted under both steady and transient state conditions using the
measured temperature profiles obtained from full scale fire tests. Following
conclusions are made from the analyses.
• The lateral restraint of the plasterboard increases the ultimate capacity of the
stud significantly. However, there is not much difference between restraining
one node of the flange versus restraining the whole flange in FEA.
• Both transient and steady state methods could be used to analyse the LSF
wall studs restrained by plasterboards under fire conditions.
• At higher load ratios, the net effect of thermal bowing and neutral axis shift
will act as an initial (global) geometric imperfection about the major axis.
The LSF wall stud is restrained about the minor axis and hence the failure
mode will be the flexural buckling about the major axis.
• At lower load ratios, the LSF wall stud with stand higher temperatures and
the yield stress of the steel reduces considerably. In this time the failure will
be due to the local buckling of either hot flange or cold flange depending on
their temperature dependent yield stress and the applied stress due to
compression and bending effects.
• In FEA, the failure location of LSF wall studs under fire condition change
from the cold flange at mid-height to the hot flange at support with
decreasing load ratios. The failure location under lower load ratios can also
be at the cold flange at mid-height when the cold flange temperature is as
high as the hot flange temperature. However, in full scale fire test, the mid-
height will be close to the furnace due to thermal bowing. Hence the
temperature at mid-height will be mostly greater than the temperature at
support. Therefore at lower load ratios the failure at hot flange will be mostly
at mid-height compared to support.
4-98
• FEA results confirmed the development of local buckling along the stud even
before the failure.
• When the stud was analysed under transient state conditions, the agreement
of the deflection versus time graphs with test results was very good compared
to the previous numerical studies of LSF walls under fire conditions (Kaitila,
2002, Feng et al. 2003b and Zhao et al. 2005) due to the use of accurate
temperature profiles and mechanical properties in FEA.
• When the stud was analysed under both steady and transient state conditions,
the results agreed well with the test results when average temperatures along
the stud were used. It was found that the developed finite element model
could be used to predict the failure time within 5 minutes, which is a
considerable improvement compared to previous researches (Kaitila, 2002,
Feng et al. 2003b and Zhao et al. 2005).
• In all the finite element analyses, the LSF test specimens under the load ratios
of 0.2 and 0.4 failed at about 600oC and 500oC, respectively. This shows that
the current limiting temperature method of using 350oC for cold-formed steel
structures is too conservative and will lead to conservative designs.
• A small reduction in load ratio was experienced even at lower temperatures
such as 100-300oC, where there is no reduction in yield stress. This is due to
the reduction in elastic modulus and the induced bending moment caused by
thermal bowing.
• Near failure, the hot flange temperatures increased rapidly for most of the
tests due to the fall off of some plasterboard pieces. Therefore a rapid fall in
load ratio was observed with time. This phase covers load ratios of 0.4 and
0.2. Hence the predicted failure times for load ratios of 0.2 and 0.4 were
closer to each other.
• The load ratio versus hot flange temperature curves merged reasonably well.
At the same load ratio, the hot flange temperatures in a stud at failure are very
close. This means that structurally similar studs will fail at the critical failure
temperature regardless of the number of plasterboards and types of insulation.
The significant effect of plasterboards and types of insulation is simply to
delay the time to reach that critical temperature in the steel stud.
4-99
Fire Design Rules
CHAPTER 5
5. REVIEW AND DEVELOPMENT OF FIRE DESIGN RULES FOR

AXIALLY LOADED LSF WALL STUDS
5.1. Introduction
In this chapter a detailed investigation into the prediction of axial compression

strength of LSF wall studs was performed to increase the understanding of their
behaviour under non-uniform temperature conditions. The measured time-
temperature profiles reported in Chapter 3 were used in this investigation. The
reduction factors for mechanical properties at elevated temperatures were based on
the equations proposed by Dolamune Kankanamge and Mahendran (2011). In the
experimental study, it was observed that the plasterboards provided sufficient lateral
restraint to LSF wall studs until their failure. This assumption was also used in the
finite element analyses and was verified using experimental results. Hence the
flexural buckling about the minor axis and flexural-torsional buckling were not
considered in the current study. This chapter reports the details of the analyses to
predict the failure times and lateral deflections of LSF wall studs under non-uniform
temperature conditions. The failure times of 10 full scale wall panels obtained from
both experiments and finite element analyses were used to investigate the accuracy
of fire design rules developed in the previous studies including Klippstein (1980),
Gerlich et al. (1996), Ranby (1999), Kaitila (2002), Feng and Wang (2005b) and
Zhao et al. (2005). Alfawakhiri’s (2001) study was based on Canadian cold-formed
steel design rules and hence it was not considered in the current study. Klippstein
(1980) and Gerlich et al. (1996) developed their fire design rules based on AISI
design manual. Ranby (1999), Kaitila (2002), Feng and Wang (2005b) and Zhao et
al. (2005) developed their fire design rules based on Eurocode 3. Hence after
analysing the merits and demerits of their proposals, a new method was proposed for
fire design based on AS/NZS 4600 (SA, 2005) and Eurocode 3 Part 1.3 (ECS, 2006).
In the current investigation, the importance of considering thermal bowing and its
magnification effects and the neutral axis shift in the routine fire design was also
investigated.
5-1
Fire Design Rules
Table 5.1 shows the details and failure times obtained from the experimental study
and the finite element analyses (FEA) using the measured temperature profiles.
These ten tests were investigated in the current study using fire design rules proposed
by previous research studies. For comparison purposes some of the parameters in the
fire design rules were assumed to be consistent with others. For example, different
thermal expansion coefficients were used by previous researches as discussed in
Chapter 4. However, for comparison purposes, the values given in Eurocode 3 Part
1.2 (ECS, 2005) were used in the current study. These values were also used in
Chapter 4. In the previous studies, effective areas of locally buckled studs were
found according to the design equations available during those times. However, in
the current study current equations given in AS/NZS 4600 (SA, 2005) and Eurocode
3 Part 1.3 (ECS, 2006) were used to find the effective element widths of LSF wall
studs. Appendices A and B show the calculation of effective areas at ambient and
elevated temperatures according to AS/NZS 4600 (SA, 2005), respectively.
Appendices C and D show the calculation of effective areas at ambient and elevated
temperatures according to Eurocode 3 Part 1.3 (ECS, 2006), respectively.
Table 5.1: Failure Times Obtained from Experiments and FEA
Load Failure Time (min.)

Test Configuration Insulation
Ratio Test FEA
1 Glass Fibre 0.2 118 115
2 Glass Fibre 0.4 108 110
3 Rock Fibre 0.4 134 131
1* None 0.2 53 53
2* None 0.2 111 115
3* Glass Fibre 0.2 101 100
4* Rock Fibre 0.2 107 105
5* Cellulose Fibre 0.2 110 109
6* Rock Fibre 0.2 136# 154
7* Cellulose Fibre 0.2 124 129
( * ) - Tests conducted by Kolarkar (2010) ( # ) - Experimental error

5-2
Fire Design Rules
5.2. Predictions of Stud Lateral Deflection
The LSF wall panel subjected to fire from one side will have a temperature gradient
across the wall. Due to this non-uniform temperature distribution across the stud, it
will be subjected to thermal bowing. This will induce an additional bending moment
in the stud and hence the deflection will increase further. This deflection will further
increase due to reducing elastic modulus of steel at elevated temperatures. This
additional deflection due to P-∆ effects and reduced elastic modulus is called the
magnification effects of thermal bowing. In order to predict the lateral deflection of
studs accurately, both the thermal bowing and its magnification effects should be
considered.
5.2.1. Klippstein (1980)

In the study of Klippstein (1980) the test results of previously conducted 10 fire tests
were inspected and correlated based on the measured temperature values. Figure 5.1
shows the estimated deflection versus time graphs for different configurations. This
empirical determination of mid-height lateral deflection cannot be applied to wall
panel construction that is significantly different from the study of Klippstein (1980)
in terms of insulation, plasterboards and other physical characteristics.
Figure 5.1: Estimated Failure Deflection of Studs in Wall Panels
5-3
Fire Design Rules
5.2.2. Cooke (1987)

Cooke (1987) considered thermal bowing of simply supported steel members due to
temperature gradient across the section and derived the following expression for mid-
height deflection,
αL2δT
e ∆T = (5.1)
8bw
where e∆T is the mid-height deflection due to thermal bowing; α is the thermal
expansion coefficient for steel (°C-1); L and bw are the stud height and web depth,
respectively in mm, and δT is the temperature difference across the member. This
equation was later used by many researchers including Gerlich et al. (1996), Ranby
(1999), Wang and Davies (2000), Kaitila (2002), Feng and Wang (2005b) and Zhao
et al. (2005) to determine the lateral deflection due to thermal bowing.
5.2.3. Gerlich et al. (1996)

Both the thermal bowing and its magnification effects were considered in the study
of Gerlich et al. (1996). Therefore the total horizontal displacement (e) of the
member will be the sum of the thermal deflection and the deflection due to P-∆
effects (e∆T + eP∆T). The stress-free thermal bowing was treated as an initial
eccentricity e∆T. Equation 5.1 was used by Gerlich et al. (1996) to predict the mid-
height thermal deflection (e∆T) of studs. However, calculated thermal deflections
were conservatively assumed to remain constant when temperature gradients
decrease. The initial bending moment P*e∆T results in a further horizontal deflection
eP∆T. The P-∆ component was predicted analytically by solving the moment
equilibrium equation as
 
 1 
e P∆T = e ∆T  − 1 (5.2)
 cos µL 
 2 
where eP∆T and e∆T are the deflections due to the magnification effects and the
thermal bowing, respectively, in mm;
N*
µ equal to (mm-1) where ET is the elastic modulus of steel at elevated
ET I X
temperature (MPa); L and Ix are the stud height (mm) and the second moment of area
of the cross-section, respectively and N* is the applied axial load (N).
5-4
Fire Design Rules
5.2.4. Ranby (1999)

According to Ranby (1999), the deflection caused by thermal bowing and neutral
axis shift was expressed as e∆T – e∆E (see Figure 5.2), where e∆T is the thermal bowing
deflection and e∆E is that due to the neutral axis shift due to changes in elastic
modulus caused by non-uniform temperature distributions. Due to this effective
eccentricity, bending moment will be developed. This leads to more deflection and
therefore an increase in the bending moment, and so on. Therefore the problem was
solved iteratively, until the value of e in Equation 5.3 (a) remains constant throughout
the iteration.
e = (e∆T − e∆E )
1
(5.3a)
N
1 − Ed
Ne
where NEd is the applied load according to Eurocode 3 Part 1.3 (ECS, 2006);
Ne is the critical Euler buckling load about the major axis given by
π 2 ∑ EI gr
Ne = (5.3b)
L2
where ΣEIgr is the sum of the bending stiffness weighted with the variation of the
modulus of elasticity across the section.
e∆E is the neutral axis shift given by,
t b t
(b f + bl ) E cf + bw w E w + (b f + bl )(bw − ) E hf
b 2 2 2
e ∆E = w − (5.3c)
2 (b f + bl )( E cf + E hf ) + bw E w
where bw, bf and bl are element widths of web, flange and lip, respectively; t is the
thickness; Ecf, Ew and Ehf are the elastic modulus of cold flange, web and hot flange,
respectively.
Hot y y y y
Side
x’
e x’
x x e x
e x
x’
Cold
Side
Original Neutral Thermal Resultant
position axis shift bowing e = e∆T – e∆E
e = e∆E e = e∆T
Figure 5.2: Lateral Deflections of LSF Wall Stud

5-5
Fire Design Rules
5.2.5. Feng and Wang (2005b)

Feng and Wang (2005b) found the lateral deflection of LSF wall studs subjected to
non-uniform temperature distributions using the Energy method. It was assumed that
the load is applied through the original centroid of the stud cross-section and the stud
has simply supported ends. The maximum lateral deflection at mid-height, which
includes the thermal bowing, magnification effects and neutral axis shift, is given by,
e = e∆T + e P∆T − e∆E (5.4a)
where e∆T is the deflection due to thermal bowing calculated using Equation 5.1; e∆E
is the neutral axis shift about the major axis; eP∆T is the magnification effects of
thermal bowing calculated as
2α∆T
N Ed L
3bw
e P∆T = (5.4b)
64 16
3
EI − N Ed
L 3L
where NEd is the applied load according to Eurocode 3 Part 1.3 (ECS, 2006); L and
bw are the stud length and web depth, respectively; E and I are the weighted average
elastic modulus and second moment of area of the cross-section, respectively. I can
[ ]
q
be calculated using I = ∑ ni I i + Ai xi2 where the cross-section is divided into q
i =1
parts, and for each part ni is the ratio between the reduced elastic modulus at
temperature Ti and the weighted average elastic modulus of the section.
5.2.6. Zhao et al. (2005)

According to Zhao et al. (2005) the mid-height lateral deflection e is defined as
e = (e L − e∆E + e∆T + e P∆T ) (5.5)
∑E θ y A i, i i
1
with e∆E = i
and e∆T + e P∆T = e∆T
∑E θ Ai
i, i
N
1 − Ed
Ne
where eL is the initial eccentricity of the applied load; e∆E is the neutral axis shift
about the major axis; e∆T and eP∆T are the maximum mid-height deflections due to
thermal bowing and its magnification effects, respectively; e∆T can be calculated
using Equation 5.1; NEd is the applied load according to Eurocode 3 Part 1.3 (ECS,
2006) and Ne is the Euler buckling load calculated according to Equation 5.3 (b).
5-6
Fire Design Rules
5.2.7. Lateral Deflections Predicted by FEA and Previous Researches

Figures 5.3 to 5.5 compare the lateral deflections obtained from experiments, FEA
and equations proposed by other researchers (Cooke (1987), Gerlich et al. (1996),
Ranby (1999) and Feng and Wang (2005b). The FEA under transient conditions were
not performed for the tests conducted by Kolarkar (2010). Therefore Figures 5.6 to
5.12 compare only the experimental results with previous researches. Measured
temperatures of hot and cold flanges obtained from experiments were used in the
predictions of lateral deflections using FEA and equations. The measured deflections
in experiments and FEA only include the thermal bowing and its magnification
effects (not the neutral axis shift). Therefore the neutral axis shift was not considered
in these figures for previous studies for comparison purposes with FEA and
experiments. Hence even though Ranby (1999) and Zhao et al. (2005) proposed
slightly different methods in predicting effective eccentricity in capacity calculations,
they were identical after ignoring neutral axis shift. Therefore only Ranby’s (1999)
method is plotted in Figures 5.3 to 5.12.
Cooke’s (1987) equation only predicts the lateral deflection due to thermal bowing.
However, the lateral deflections obtained from FEA and previous researches include
thermal bowing and its magnification effects. Therefore as expected their values
were always larger than the values predicted by Cooke (1987). The lateral deflections
obtained from FEA and previous researches followed the same trend even though
there are some differences between them and the experimental values during the
early stages of the test. In the initial period of the test, the temperature gradient
across the stud was very low compared to the temperature gradient across the LSF
wall. The predicted lateral deflection using FEA and previous researches were based
on the temperature gradient across the stud. Hence the lateral deflection including the
thermal bowing was less in this case. However, during the early stages of the test the
LSF wall will act as a rigid structure and the thermal bowing will be induced due to
the temperature gradient across the wall. Hence the lateral deflection including the
thermal bowing was more in this case. This effect will be more in LSF walls with
two layers of plasterboards compared to a single layer of plasterboard. Hence slightly
better agreement was obtained in Figure 5.6.
5-7
Fire Design Rules
-10
-20
-30
-40
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)
Experiment FEA
Cooke (1987) Gerlich (1995)
Ranby (1999) Feng and Wang (2005)
Fire Design Rules for Test 1
-10
-20
-30
-40
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)
Experiment FEA
5-8
Fire Design Rules
-10
-20
-30
-40
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min.)
Experiment FEA
-10
-20
-30
-40
-50
0 10 20 30 40 50 60
Time (min.)
Experiment Cooke (1987)
Gerlich (1995) Ranby (1999)
Feng and Wang (2005)
Design Rules for Test 1*
5-9
Fire Design Rules
0
-10
-20
-30
-40
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)
-10
-20
-30
-40
-50
-60
-70
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)
5-10
Fire Design Rules
Lateral Deflection (mm) -10
-20
-30
-40
-50
-60
-70
-80
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)

-10
-20
-30
-40
-50
-60
-70
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)

5-11
Fire Design Rules
-10
-20
-30
-40
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min.)
0
-10
-20
-30
-40
0 10 20 30 40 50 60 70 80 90 100 110 120 130
Time (min.)

5-12
Fire Design Rules
5.3. Previous Fire Design Rules based on AISI Design Manual
5.3.1. Klippstein (1980b)

The fire design model proposed by Klippstein (1980b) is based on the allowable
stress method given in the AISI design manual and ASTM E119. In this study, the
ability of the gypsum boards to carry any vertical loads was neglected. However, it
was assumed that the failure by weak axis flexural buckling or torsional buckling is
prevented by the plasterboards on both sides of the stud. The stress-strain curve of
steel was assumed to be linear up to the yield strength. The failure load N* at elevated
temperature was calculated using,
1
N* = (5.6a)
1 e
+
 23  S x FyT
A FalT 
 12 
where N* is the predicted failure load of the stud at an elevated temperature T; A and
Sx are the gross cross-sectional area and the major axis section modulus, respectively;
e is the total mid-height lateral deflection of the stud; FyT is the yield strength at
elevated temperature and FalT is the allowable stress for axial load at elevated
temperature which can be calculated using Equation 5.6 (b).
3 (QT FyT )  kL 
2 2
12
FalT = QT FyT −   (5.6b)
23 23 π 2 ET  r 
where QT is the ratio between the effective area at elevated temperature and the gross
area; FyT and ET are the yield strength and the modulus of elasticity at elevated
temperature, respectively; Average temperature of the stud was used to find QT, FyT
and ET. kL is the effective length of the stud and r is the radius of gyration about the
major axis.
The proposed fire design method by Klippstein (1980b) is heavily dependent on the
empirical determination of the mid-height lateral deflection during the fire by
Klippstein (1980b) as discussed in Section 5.2.1.
5-13
Fire Design Rules
5.3.2. Gerlich et al. (1996)

In this study, the torsional-flexural buckling was neglected and it was found that the
failure mode of steel studs in LSF wall systems exposed to fire was generally
governed by the buckling of compression flange on the ambient side of the wall
assembly. The stud failure was checked only at the mid-height cold flange of the
studs. The failure criterion was checked based on the AISI design manual using,
N* M*
+ =1 (5.7)
Pn M nx
where N* is the applied axial compression load; M*=N*e where e is the total
horizontal deflection at mid-height defined in Section 5.2.3; Pn=Aeff*fn where Aeff is
the effective area at ambient temperature and fn is the critical stress using the yield
stress at cold flange temperature; Mnx=fyT*Sx where fyT is the yield stress at cold
flange temperature and Sx is the gross section modulus about the major axis at
ambient temperature.
Gerlich et al. (1996) developed a spreadsheet by adjusting the input values for yield
strength and modulus of elasticity to take into account the effect of temperature using
the reduction factors given by Klippstein (1980b). Using the spreadsheet, a critical
temperature was found at which the predicted strength of the stud was equal to the
applied axial load. This temperature was compared with the compression (cold)
flange temperature of the wall to find the failure time. The TASEF predictions were
used to find the time until which the unexposed lining will provide lateral restraint to
the compression flange. Hence the lower value obtained from these analyses was
considered as the predicted failure time.
The load ratio is defined as the ratio between the ultimate capacities of LSF wall stud
at elevated and ambient temperatures. Figures 5.13 to 5.22 compare the variation of
load ratio with time according to FEA and fire design rules proposed by Klippstein
(1980b) and Gerlich et al. (1996). The load ratio versus time curve of FEA was
obtained by plotting the ratios of ultimate capacities of FEA at elevated and ambient
temperatures. Similarly the load ratio versus time curves of fire design rules were
obtained by plotting the ratios of ultimate capacities of fire design rules at elevated
and ambient temperatures.
5-14
Fire Design Rules
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)
FEA Klippstein (1980) Gerlich (1995)
and Gerlich et al. (1996) for Test 1
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)
5-15
Fire Design Rules
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min.)
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60
Time (min.)
and Gerlich et al. (1996) for Test 1*
5-16
Fire Design Rules
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)
5-17
Fire Design Rules
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)
5-18
Fire Design Rules
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min.)
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130
Time (min.)
5-19
Fire Design Rules
5.3.3. Discussion of Previous Fire Design Rules based on AISI Design Manual
Both Klippstein (1980) and Gerlich et al. (1996) developed fire design rules for LSF
wall studs using the AISI design manual. Appendices E and F show the fire design
calculations according to Klippstein (1980) and Gerlich et al. (1996), respectively.
Figure 5.1 shows the estimated deflection versus time graphs for different
configurations used by Klippstein (1980). This empirical determination of lateral
deflection cannot be applied to the current LSF wall panels, which are significantly
different from those considered in the study of Klippstein (1980). Hence the
measured experimental lateral deflection values were used in the investigation of
Klippstein (1980). During the early stages of the test the measured lateral deflection
was high compared to the predicted deflection as discussed in Section 5.2.7. Hence
the ultimate compression capacities of LSF wall studs according to Klippstein (1980)
were lower than FEA results during the initial stages of the tests (see Figures 5.13 to
5.22). During the final stages of many tests, the lateral deflection reversed its
direction and hence Klippstein’s (1980) design rule resulted in arbitrary values.
Gerlich et al. (1996) compared the critical temperature found by the model with the
compression flange temperature on the ambient side of the wall, to find the time to
failure. Therefore in their fire design rules, the calculations of critical stress and the
bending moment capacity were based on the yield stress at cold flange temperature.
However, the average temperature was used in the calculation of elastic modulus.
This resulted in the over-estimation of failure time of LSF wall panels using Gerlich
et al.’s (1996) fire design rules. For cavity insulated wall panels the cold flange
temperature is much lower than the hot flange temperature. Therefore in these cases
the predicted failure times using Gerlich et al.’s (1996) rules were much higher than
those from FEA (see Figures 5.18 to 5.20).
5.3.3.1. Effective Area of LSF Wall Studs at Elevated Temperature

The effect of local buckling is considered in the studies of Klippstein (1980) and
Gerlich et al. (1996) by calculating the effective area of the stud cross section.
Gerlich et al. (1996) considered the form factor Q (ratio between the effective and
gross areas) as a constant value even at elevated temperature. This means that he
used the effective area at ambient temperature in the design of LSF wall studs at
5-20
Fire Design Rules
elevated temperature as well. However, in the study of Klippstein (1980) the form
factor QT is defined in terms of elevated temperature. This is the ratio between the
effective area at elevated temperature and gross area at ambient temperature. In the
current study of these two fire design rules the effective areas were found using
AS/NZS 4600 (SA, 2005).
In order to find the form factor Q (at ambient temperature), the critical stress fn and
effective element widths were found using ambient temperature properties. In order
to find the form factor QT (at elevated temperature), the critical stress fn and effective
element widths were found using the elevated temperature properties of the stud.
Detailed calculations can be found in Appendices A and B for effective areas at
ambient and elevated temperatures, respectively. The stud section used in the current
study was considered with the elevated temperatures at failure for Stud 3 of Test 1
(Hot flange temperature of 656oC and cold flange temperature of 492oC). Table 5.2
summarizes the effective areas at ambient and elevated temperatures.
5.3.3.2. Calculation of Section Modulus

Both Klippstein (1980) and Gerlich et al. (1996) used gross section to calculate the
bending moment capacity. However, it is not recommended and the effective area
should be used to find the section modulus. Other researchers including Ranby
(1999), Kaitila (2002), Feng and Wang (2005b) and Zhao et al. (2005) also used
effective section modulus in their fire design rules.
5.3.3.3. Effective Eccentricity “e”

Klippstein (1980) used the estimated lateral deflection of the studs using his
experimental study. These lateral deflections cannot be used in the current study due
to the different wall configurations used. Therefore the measured experimental lateral
deflection values were used in the current study. Hence both these models were
assumed to consider thermal bowing and its magnification effects. However, their
models did not take into account the neutral axis shift of LSF wall stud cross-section
due to non-uniform temperature distribution, which is important in the fire design of
LSF wall panels.
5-21
Fire Design Rules
Table 5.2: Critical Stress and Effective Area based on

AS/NZS 4600 (SA, 2005)
Using Average Stud

Section 90x40x15x1.15 Using Ambient Temperature
Ag = 224.71 mm2 Temperature Hot Flange = 656oC
Cold Flange = 492oC
Elastic buckling stress about the
472.313 MPa 163.043 MPa
major axis fox
Elastic buckling stress about the
5661.827 MPa 1954.463 MPa
minor axis foy
Elastic flexural buckling stress foc
472.313 MPa 163.043 MPa
(Clause 3.4.2)
Critical foc 472.313 MPa 163.043 MPa
Non-dimensional slenderness λc 1.098 0.874
Critical stress fn 343.530 MPa 90.552 MPa
Effective web width bwe 47.180 mm 52.776 mm
Effective flange width bfe 35.936 mm 34.654 mm
Effective lip width ds 12.997 mm 12.521 mm
2
Effective area Ae 166.803 mm 176.210 mm2
5-22
Fire Design Rules
5.4. Design Rules in Eurocode 3
5.4.1. Eurocode 3 Part 1.3 (ECS, 2006)
5.4.1.1. Axial Compression
This European standard is for cold-formed steel structures at ambient temperature.

For axial compression members, the design buckling resistance Nb,Rd at ambient
temperature is given by Clause 6.3.1 of Eurocode 3 Part 1.1 (ECS, 2005). The design
buckling resistance Nb,Rd for compression member shall be obtained from Clause
6.3.1.1 of Eurocode 3 Part 1.1 (ECS, 2005) as.
N b, Rd = χAeff f y γ M 1 (5.8)
where
Aeff is the effective area of the cross-section and fy is the yield strength;
χ is the appropriate reduction factor for buckling resistance. The value of χ for the
appropriate non-dimensional slenderness λ should be determined from the relevant
buckling curve according to
1
χ= but χ ≤ 1
2
φ + φ −λ 2
where φ =
1
2
[ (
1 + α λ − 0 .2 + λ
2
) ]
Aeff f y
λ=
N cr
α is the imperfection factor based on the relevant buckling curve; for lipped C-
section α = 0.34.
Ncr is the elastic critical force for the relevant buckling mode based on the gross
cross-sectional properties and can be calculated as,
π 2 EI
N cr =
l2
where E is the elastic modulus; I is the second moment of area of the gross section
and l is the buckling length.
5-23
Fire Design Rules
5.4.1.2. Bending and Axial Compression

For combined axial load and bending moments, the general design equations are,
N Ed k xx (M x , Ed + ∆M x , Ed ) k xy (M y , Ed + ∆M y , Ed )
+ + ≤1 (5.9a)
χ x f y Aeff / γ M 1 χ LT f yWeff , x / γ M 1 f yWeff , y / γ M 1
N Ed k yx (M x , Ed + ∆M x , Ed ) k yy (M y , Ed + ∆M y , Ed )
+ + ≤1 (5.9b)
χ y f y Aeff / γ M 1 χ LT f yWeff , x / γ M 1 f yWeff , y / γ M 1
where NEd is the applied axial load according to Eurocode 3 Part 1.3 (ECS, 2006) and
fy is the basic yield strength;
Aeff is the effective cross-sectional area that is subject only to compression;
Mx,Ed and My,Ed are the applied bending moments about the major and minor axes (see
Figure 5.2);
∆Mx,Ed and ∆My,Ed are the additional moments due to shift of centroidal axis;
Weff,x and Weff,y are the effective section modulus for the maximum compressive stress
in an effective cross-section that is subject only to moment about the major and
minor axes, respectively;
χx and χy are the reduction factors due to flexural buckling from Section 5.4.1.1;
χLT is the reduction factor due to lateral torsional buckling from Clause 6.3.2 of
Eurocode 3 Part 1.1 (ECS, 2005); For memebers not susceptible to torsional
deformation χLT would be equal to 1.
kxx kxy kyx and kyy are the interaction factors calculated according to Annex A of
5.4.2. Eurocode 3 Part 1.2 (ECS, 2005)

This design code is specifically developed for hot-rolled steel structures in a fire
situation. However, it states that the methods given are also applicable to cold-
formed steel members and sheeting within the scope of Eurocode 3 Part 1.3 (ECS,
2006). Hence the relevant design rules are considered in this research. The simplified
method specified in Clause 4.2.1 of Eurocode 3 Part 1.2 (ECS, 2005) requires that
the design effect of actions for the fire design situation is less than or equal to the
corresponding design resistance of the steel member for the fire design situation at
time t. In this code the temperature is denoted by θ.
5-24
Fire Design Rules
5.4.2.1. Compression Members with Class 3 Cross-sections

The design buckling resistance Nb.f1t.Rd of compression members is defined in Clause
4.2.3.2 as
N b, fi ,t , Rd = χ fi Ag k y ,θ f y / γ M , fi (5.10)
with χ fi =
1
2
, ϕθ =
1
2
[ ]2
1 + α λ θ + λ θ , α = 0.65 235 f y
ϕ θ + ϕθ − λ θ 2
λ θ = λ [k y ,θ k E ,θ ]0.5 and λ =
Aeff f y
.
N cr
where
χfi is the reduction factor for flexural buckling in the fire design situation and should
be taken as the lesser of the values calculated about the y and z-axes;
ky,θ and kE,θ are the reduction factors for the steel yield strength and the slope of the
linear elastic range, respectively at steel temperature θ;
γM,fi is equal to 1 which is the partial factor for the relevant material property, in the
fire situation;
λ θ is the non-dimensional slenderness at temperature θ.
When designing using nominal fire exposure, the design resistance Nb,fi,t,Rd at time t
of a compression member with a non-uniform temperature distribution may be taken
as equal to the design resistance Nb,fi,θ of a compression member with a uniform steel
temperature θ equal to the maximum steel temperature θmax reached at time t.
5.4.2.2. Members with Class 4 Cross-sections

For the design under fire conditions the design yield strength of steel should be taken
as the 0.2% proof strength. For thin-walled elements, local buckling phenomena
become important. Precise design methods have not been included in this design
code for the fire design of cold-formed steel structures. Hence the temperature of
cold-formed (Class 4) cross-sections was limited to 350ºC at any time. This is a
conservative approach which leads to uneconomical designs. However, more
advanced calculation models and test results indicate that the critical temperature of a
particular cold-formed steel structure or structural member is higher than this
limiting value of 350ºC.
5-25
Fire Design Rules
5.5. Previous Fire Design Rules based on Eurocode 3
5.5.1. Ranby (1999)

Ranby (1999) found that the basic design equations given in Eurocode 3 Part 1.3
(ECS, 1996) for flexural and flexural-torsional buckling at ambient temperature
could be used at elevated temperatures if appropriately reduced mechanical
properties at elevated temperatures are used. The model is developed as two parts.
First the design model was based on equations that considered flexural buckling
about the major axis only and the torsional-flexural buckling was not considered as a
critical failure mode. This was when the temperature stays under the level of
calcination of the plasterboards. Hence the lateral restraint provided by the
plasterboards was taken into account throughout the fire event. Secondly, the
torsional-flexural buckling mode was checked. This is when the temperature during
the fire goes above the level of calcination of the plasterboards. Hence the lateral
restraint was no longer valid on the fire side. However, in the current study, only the
flexural buckling will be considered as discussed before.
The model for flexural buckling was developed by Ranby (1999) by modifying
Equations 5.9 (a) and (b) for bending and axial compression. In the present case, no
external bending moments are applied and the column can deflect only about the
major axis. Therefore M x ,Ed = M y , Ed = ∆M y , Ed = 0 and ∆M x , Ed = N Ed e where e is the
effective eccentricity due to thermal bowing, magnification effects and neutral axis
shift. This leads to the equation,
N Ed k N e
+ xx Ed ≤ 1 with β M , x = 1.3 (5.11)
χ x f y Aeff  I eff 
f y  
y 
 CF 
Then it was re-arranged as
χ x f y Aeff
N Ed = (5.12)
k xx e y CF χ x Aeff
1+
I eff
where Ieff is the effective second moment of area of the steel cross-section; y CF is the
distance from cold flange to the effective neutral axis; all other parameters are
defined in Section 5.4.1.2.
5-26
Fire Design Rules
5.5.2. Wang and Davies (2000)

Wang and Davies (2000) used the design equations in Eurocode 3 Part 1.3 (ECS,
1996) to calculate the fire resistance of thin-walled cold-formed members. Uniform
temperature values were assumed for flanges and lips on both the fire and cold sides.
The temperature distribution in the web was assumed, such that the distribution of
steel strength and stiffness was linear. Eurocode 3 Part 1.3 (ECS, 1996) was used to
evaluate the effective widths of flanges and lips, replacing the ambient temperature
properties by elevated temperature properties. The average steel strength and
stiffness values were used for webs due to changing steel properties. The design
method was validated against the results of three tests on channel columns by Gerlich
et al. (1996). The ambient temperature approach was found to be suitable for
adoption under fire conditions by taking into account the reductions in the strength
and stiffness of steel at elevated temperatures, and the additional bending moments
due to thermal bowing and neutral axis shift. Feng and Wang (2005b) extended this
study further and it will be discussed later in this chapter. Hence the method
proposed by Wang and Davies (2000) is not investigated further in this chapter.
5.5.3. Kaitila (2002)

Kaitila (2002) used Equation 5.8 given in Eurocode 3 Part 1.3 (ECS, 1996) to find
the buckling resistance of the stud subjected to uniform temperature.
N b, Rd = χAeff f yT (5.13)
where Aeff is calculated using ambient temperature properties and fyT is the yield
stress at elevated temperature; χ is found based on Equation 5.8
Kaitila’s (2002) uniform temperature method was used in the current study to check
its validity for use in the simplified calculations of the capacity of LSF wall studs
subjected to non-uniform temperatures. Maximum (hot flange) temperature of the
stud cross section was used in the calculations. Figure 5.23 shows the variation of
load ratios with time when uniform temperature method was used compared to FEA
results. At higher load ratios, this method does not predict the LSF wall stud
behaviour at non-uniform temperatures. On the other hand at lower load ratios (less
than 0.4) when the stud withstands higher temperatures the effect of non-uniform
5-27
Fire Design Rules
temperature distribution is negligible. Hence the uniform temperature method agrees

reasonably well with the non-uniform temperature method for all the tests for lower
load ratios. In the study of Kaitila (2002) the effective area at ambient temperature
was used in the fire design rules. In the current study the uniform temperature
method proposed by Kaitila (2002) was further investigated by incorporating
effective area at elevated temperatures (see Figure 5.23). This time the agreement
was better compared to the case where the effective area at ambient temperature was
used. However, the variation of predicted failure time from the FEA results at higher
load ratios is still not acceptable.
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)
FEA Using Aeff20 Using AeffT
Figure 5.23: Influence of Effective Area in Kaitila’s (2002) Uniform

Temperature Method for Test 1
The method proposed by Ranby (1999) was investigated by Kaitila (2002), and
minor changes were proposed for LSF wall studs subjected to non-uniform
temperature distributions. The section modulus should be calculated corresponding
to the flange that is furthest away from the effective centroid (Kaitila, 2002).
( )
Therefore Weff,x in Equation 5.9 (a) should be equal to I eff / d − y CF . i.e Ranby’s
( )
(1999) equation overestimated the capacity when I eff / y CF was used as Weff,x,com.
Hence the new equation proposed by Kaitila (2002) for LSF wall studs subjected to
non-uniform temperature is
5-28
Fire Design Rules
χ x f y Aeff
N Ed =
1+
( )
k xx e d − y CF χ x Aeff
(5.14)
I eff
where d is the web height and all other parameters are defined in Equation 5.12. The
major and minor axes are shown in Figure 5.2.
Ranby’s (1999) and Kaitila’s (2002) equations in combination with Eurocode design
equations were used to determine the axial compression capacities of LSF wall stud
at smaller time intervals based on the measured temperature profiles with time.
Appendices G and H show the sample calculations of Kaitila’s (2002) and Ranby’s
(1999) methods, respectively.
Figures 5.24 to 5.33 show the variation of load ratios from FEA, Ranby (1999) and
Kaitila (2002). Ranby’s (1999) equation and the modified equation proposed by
Kaitila (2002) for non-uniform temperature agreed well with the FEA results. These
two fire design rules agreed well with each other for non-insulated and externally
insulated wall panels (see Figures 5.24 to 5.28, 5.32 and 5.33). However, Ranby’s
(1999) method overestimated the stud capacity (see Figures 5.29 to 5.31) for cavity
insulated wall panels. This is because when the temperature difference is high
( ) ( )
between the hot and cold flanges, the difference between y CF and d − y CF is also
high in Equations 5.12 and 5.14.
In the initial period, the load ratio of the LSF wall stud is very sensitive to the
temperature difference between the hot and cold flanges. It can be clearly observed
in Figure 5.27. The load ratio dropped rapidly up to 6 minutes due to the increased
thermal bowing and corresponding bending moment. However, after 6 minutes the
temperature difference between the hot and cold flanges was reduced even though
the average stud temperature was increased slightly. Hence the load ratio increased
with time for a few minutes before it started to reduce due to increased stud
temperature.
5-29
Fire Design Rules
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)
FEA Ranby (1999)
Kaitila (2002) Uniform Kaitila (2002) Non-uniform
Kaitila (2002) for Test 1
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)
FEA Ranby (1999)
Kaitila (2002) Uniform Kaitila (2002) Non-unform
5-30
Fire Design Rules
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min.)
FEA Ranby (1999)
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60
Time (min.)
FEA Ranby (1999)
Kaitila (2002) for Test 1*
5-31
Fire Design Rules
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)
FEA Ranby (1999)

1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)
FEA Ranby (1999)

5-32
Fire Design Rules
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)
FEA Ranby (1999)

1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)
FEA Ranby (1999)
5-33
Fire Design Rules
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min.)
FEA Ranby (1999)
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130
Time (min.)
FEA Ranby (1999)
5-34
Fire Design Rules
5.5.4. Feng and Wang (2005b)

In the study of Feng and Wang (2005b) the local and global buckling effects were
considered after taking into account the neutral axis shift, thermal bowing and its
magnification effects. This study considered the combination of bi-axial bending and
axial compression. However, it was found that the effect of neutral axis shift about
the minor axis was very small on the prediction of panel failure times and could be
ignored in order to simplify routine fire design. Hence in the current study also
bending about the minor axis was not considered. Both flexural and torsional
buckling were considered in this study. However, in the current study plasterboards
were assumed to provide lateral restraint to the studs until failure as discussed in
Chapters 4 and 5. Therefore only the flexural buckling was investigated.
Equations 5.9 (a) and (b) used in Feng and Wang (2005b) was further simplified in
the current study after considering their conclusions as follows,
N Ed k N e
+ xx Ed = 1 (5.15)
χ x N eff M x ,eff
where kxx and χ x are defined in Equations 5.9 (a) and (b); Mx,eff is the stud section
moment capacity at non-uniform elevated temperatures about the major axis; e is
defined in Section 5.2.5; The stud cross-section at both mid-height and support were
checked using Equation 5.15. At mid-height e includes thermal bowing,
magnification effects and neutral axis shift while at the support it only includes the
neutral axis shift;
Neff is the axial compression capacity of the stud at elevated temperatures calculated
as N eff = f yT Aeff where fyT is the weighted average yield stress of the cross-section
using the non-uniform temperature distribution;

Parameters Aeff and Mx,eff are defined in Sections 5.5.4.1 and 5.5.4.2, respectively.
5.5.4.1. Calculation of Effective Area Aeff

Two methods were investigated by Feng and Wang (2005b) to find the ultimate
capacity of LSF wall studs at non-uniform elevated temperatures. For LSF wall studs
at elevated temperatures, fire design calculations are given in Eurocode 3 Part 1.2
(ECS, 2001), which simply refers to Eurocode 3 Part 1.3 (ECS, 2001). In addition to
the global buckling check using the equations given in Eurocode 3 Part 1.3 (ECS,
5-35
Fire Design Rules
2001), local buckling capacity should also be checked, and Eurocode 3 Part 1.2
(ECS, 2001) recommends using effective cross-section at ambient temperature
without modification for temperature effects. Therefore Feng and Wang (2005b)
used Equation 5.15 with effective area calculated at ambient temperature (referred to
as Eurocode 3 Part 1.2 method). In addition to this Feng and Wang (2005b)
introduced a new method to allow for temperature effects on the effective cross-
section (referred to as the modified Eurocode 3 Part 1.3 method). Therefore, a
uniform temperature distribution was assumed for flanges and lips and the method in
Eurocode 3 Part 1.3 (ECS, 2001) was directly used to evaluate their effective widths,
but using the elevated temperature properties. The weighted average elastic modulus
was used to calculate the effective width of web.
Figure 5.34 shows the effective cross-section of the LSF wall stud. Table 5.3 shows
the effective areas under concentric compression at ambient and elevated
temperatures according to Eurocode 3 Part 1.3 (ECS, 2006). Detailed calculations are
shown in Appendices C and D. Table 5.4 shows the effective areas under bending
about the major axis at ambient and elevated temperatures. Detailed calculations are
shown in Appendices I and J. These values are calculated for the section
90x40x15x1.15 used in the current study. The hot and cold flange temperature values
used are 656oC and 492oC, respectively (Failure temperatures of Stud 3 of Test 1).
Cold Side
beff1,cf beff2,cf
t tred,cf ceff,cf
y
deff1
deff2
t tred,hf ceff,hf
beff1,hf beff2,hf
Hot Side
Figure 5.34: Effective Cross-section of LSF Wall Stud
5-36
Fire Design Rules
Table 5.3: Summary of Effective Areas under Concentric Compression

Calculated based on Eurocode 3 Part 1.3 (ECS, 2006)
Section 90x40x15x1.15 Using Ambient Using Elevated

2
Ag = 224.71 mm Temperature Stud Temperature
Effective web width deff1 18.969 mm 25.011 mm
Effective web width deff2 18.969 mm 25.0011 mm
Effective flange width beff1,hf 16.103 mm 19.425 mm
Hot Effective flange width beff2,hf 18.279 mm 19.425 mm
Flange Effective lip width ceff,hf 14.425 mm 14.425 mm
Reduced thickness tred,hf 0.774 mm 1.127 mm
Effective flange width beff1,cf 16.628 mm
Cold Effective flange width beff2,cf Same as above 18.512 mm
Flange Effective lip width ceff,cf flange values 14.425 mm
Reduced thickness tred,cf 0.815 mm
Effective area Aeff 131.291 mm2 163.979 mm2
Table 5.4: Summary of Effective Areas under Bending about the Major Axis
Calculated based on Eurocode 3 Part 1.3 (ECS, 2006)
Section 90x40x15x1.15 Using Ambient Using Elevated

2
Ag = 224.7 mm Temperature Stud Temperature
Effective web width deff1
88.85 mm 88.85 mm
Effective web width deff2
Bending about x-x axis HF in CF in HF in CF in

Tension Tension Tension Tension
Effective flange width beff1,hf Gross Same as Gross Same as
Hot Effective flange width beff2,hf values Table values Table
Flange Effective lip width ceff,hf for 5.3 for 5.3
Reduced thickness tred,hf elements values elements values
Effective flange width beff1,cf Same as Gross Same as Gross
Cold Effective flange width beff2,cf Table values Table values
Flange Effective lip width ceff,cf 5.3 for 5.3 for
Reduced thickness tred,cf values elements values elements
Effective area Aeff (mm2) 207.275 207.275 209.410 223.931
5-37
Fire Design Rules
5.5.4.2. Calculation of Section Moment Capacity Mx,eff

The section moment capacities about the major axis were calculated separately for
stud mid-height and support in the study of Feng and Wang (2005b). For mid-height
calculations the two section moment capacities were found by considering the first
occurrence of material yield on the tension (hot) side (Mx,eff,f) and considering partial
plasticity so that the compression (cold) side reaches yield (Mx,eff,p). For support
calculations, only one value for section moment was calculated. This is because the
exposed flange is under compression and has a lower yield strength. Therefore only
the first occurrence of material yield on the compression (hot) side was considered
(Mx,eff,f1). Table 5.5 shows the section moment capacities using effective areas at
ambient and elevated temperatures (see Appendices I and J).
Table 5.5: Summary of Section Moment Capacities
Using Effective
Using Effective
Area at Elevated
Area at
Section 90x40x15x1.15 Temperature
Ambient
HF = 656oC
Temperature
CF = 492oC
Bending about x-x axis Hot Flange in tension (at mid-height)
Partial Plasticity Mx,eff,1 0.753 kNm 0.767 kNm
Bending about x-x axis Cold Flange in tension (at support)
First occurrence of Material Yield Mx,eff,2 0.305 kNm 0.367 kNm
5.5.5. Zhao et al. (2005)

Simple calculation rules for assessing the fire resistance of lightweight steel studs
maintained by boards were proposed by Zhao et al. (2005). The approach is based on
Eurocode 3 Part 1.3 (ECS, 2002) for lightweight structures at room temperature.
Mechanical properties were modified according to temperature and the thermal
bowing and its magnification effects were considered in this study. The stress-strain
relationships of cold-formed steel at elevated temperatures were assumed to be
perfectly elastic-plastic in the study of Zhao et al. (2005). Also the yield stress was
taken as the 0.2% proof stress at any temperature level.
5-38
Fire Design Rules
First the simplified design method for studs under flexural buckling was developed
by Zhao et al. (2005). The buckling resistance of steel studs maintained by
plasterboards under compression, Nfi,Rd, was expressed as follows:
( )
N fi , Rd = χ λθ N fi ,c , Rd (5.16)
where
χ is the reduction factor for buckling resistance defined in Equation 5.8; λ θ is the
relative slenderness at elevated temperature; Nfi,c,Rd is the design resistance load of
the steel stud.
λθ = ∑A f i y ,θ ,i N fi ,cr with N fi ,cr = π 2 (EI ) fi ,eff Lθ2 and (EI ) fi ,eff = ∑ Ei ,θ I i ,θ

i i
N fi ,c , Rd = ∑ Aeff ,i f y ,θ ,i (5.17)
i
where Ai and Aeff,i are the initial and effective element areas, respectively; fy,θ,i is the
0.2% proof yield strength of steel at temperature θi; Nfi,cr is the Euler buckling load
and Lθ is the buckling length of the stud in fire situation; (EI)fi,eff is the effective
flexural stiffness of the cross-section; Ei,θ is the modulus of elasticity of each plate
element; Ii,θ. is the second moment of area of each element and θ is the appropriate
elevated temperature.
The stability of a lightweight steel member subjected to a combination of axial

compression and bending is checked by Zhao et al. (2005) as,
N Ed k N e
+ xx Ed = 1 (5.18)
N fi , Rd M x ,eff
where Nfi,Rd, is the design buckling resistance of a compression member calculated

using Equation 5.16;
kxx is the modification factor equal to 1.0 according to Zhao et al. (2005);
Mx,eff is the design section moment resistance (based on the effective cross-section
that is subjected to bending moment only) about the major axis;
e is the effective eccentricity and calculated as e = (e L − e∆E ) at the support and
e = (e L − e∆E + e∆T ) at the mid-height, where eL, e∆E and e∆T are defined in Section
5.2.6.
5-39
Fire Design Rules
To simplify calculations, when compression was on the hot side of the stud, the
moment resistance Mx,eff of the cross-section was calculated with yielding occurring
on the hot flange and a linear stress distribution in the web. When compression was
on the cold side, the moment resistance was calculated with yielding occurring on
both flanges. In this case bi-linear stress-distribution will be experienced in the web
on the hot side (partial plasticity).
In this study the effects of local buckling and distortional buckling were taken into
account according to Eurocode 3 Part 1.3 (ECS, 2002) using the effective widths of
all the plate elements. The flanges and the lips (referred as flange stiffeners in
Eurocode) on each side of the cross-section were assumed to be at the same
temperature. Therefore the method in Eurocode 3 Part 1.3 (ECS, 2002) was directly
used to evaluate their effective widths, but using the elevated temperature properties
to replace those at ambient temperature. The weighted average steel resistance
(strength and stiffness) value was used to calculate the web effective width according
to Zhao et al. (2005). However, in the current study the mechanical properties at web
mid-height was used to determine the effective width of web for comparison
purposes.
Appendices I, J and K show the sample calculations to find the ultimate capacities of
LSF wall stud under fire conditions according to Feng and Wang’s (2005b) method
based on Eurocode 3 Part 1.2 (ECS, 2001), Feng and Wang’s (2005b) method based
on Eurocode 3 Part 1.3 (ECS, 2001) and Zhao et al.’s method (2005). Figures 5.35
to 5.44 show the variation of load ratio with time based on the fire design methods of
Feng and Wang (2005b) and Zhao et al. (2005). Feng and Wang’s (2005b) method
based on Eurocode 3 Part 1.3 (ECS, 2001) agreed well with the FEA results
compared to the methods based on Eurocode 3 Part 1.2 (ECS, 2001) and Zhao et al.
(2005).
5-40
Fire Design Rules
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)
FEA Feng and Wang (2005) EC3 P1.2

Feng and Wang (2005) EC3 P1.3 Zhao et al. (2005)
(2005b) and Zhao et al. (2005) for Test 1
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)

5-41
Fire Design Rules
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min.)

1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60
Time (min.)

(2005b) and Zhao et al. (2005) for Test 1*
5-42
Fire Design Rules
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)

1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)

5-43
Fire Design Rules
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)
5-44
Fire Design Rules
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min.)
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130
Time (min.)
5-45
Fire Design Rules
5.5.6. Discussion of Previous Fire Design Rules based on Eurocode 3

Design rules to find the ultimate capacity of LSF wall studs at elevated temperatures
were developed by Ranby (1999), Kaitila (2002), Feng and Wang (2005b) and Zhao
et al. (2005) using Eurocode 3 Parts 1.2 and 1.3. They used Equations 5.9 (a) and (b)
to find the combined effects of axial compression and bending moment. In the study
of Kaitila (2002), the method proposed by Ranby (1999) was improved with minor
modifications as discussed before. Therefore Kaitila’s (2002) method was
investigated in detail instead of Ranby (1999).
In all the previous studies except Feng and Wang (2005b), the minor axis bending
was not considered. This is due to the reduced effective length in the minor axis
direction due to plasterboard restraint. Feng and Wang (2005b) included the neutral
axis shift about the minor axis and corresponding bending moment. However, in her
study it was concluded that this effect is negligible and can be ignored in the fire
design of LSF wall studs. Therefore Equation 5.9 (a) can be reduced to
N Ed k xx (M x , Ed + ∆M x , Ed )
+ =1 (5.19)
χ x f y Aeff / γ M 1 χ LT f yWeff , x / γ M 1
The bending moment about the major axis is developed due to three separate effects
caused by non-uniform temperature distribution in LSF wall studs. They are the pure
thermal bowing due to temperature gradient, magnification effects due to P-∆ effects
and the neutral axis shift due to the deterioration of stiffness at non-uniform elevated
temperatures. If the effective eccentricity of this combination of effects is denoted as
“e” then the bending moment about the major axis is N Ed e . For members not
susceptible to torsional deformations, χ LT will be equal to 1.0. In addition to this,

with the assumption of γM1 equal to 1, Equation 5.19 can be reduced to
N Ed k N e
+ xx Ed = 1 (5.20)
χ x f y Aeff f yWeff , x
In Equation 5.20, the component f y Aeff is the ultimate failure load for local buckling
Neff and the component f yWeff , x is the section moment capacity Mx,eff of LSF wall
5-46
Fire Design Rules
stud. Hence the common equation used by all the previous researches including
Ranby (1999), Kaitila (2002), Feng and Wang (2005b) and Zhao et al. (2005) is
N Ed k N e
+ xx Ed = 1 (5.21)
χ x N eff M x ,eff
In order to find the ultimate load NEd of LSF wall stud at elevated temperature, the
parameters kxx, e, χ x , Neff and Mx,eff should be determined accurately by taking into
account the effects of non-uniform temperature distribution in the LSF wall studs.
The calculation of these parameters varied in the previous studies and hence different
ultimate loads were obtained for LSF wall studs at elevated temperatures.
5.5.6.1. Variation of Ultimate Load NEd

Figure 5.45 shows the variation of the ultimate load capacity of LSF wall stud with
respect to time from previous studies. Average measured temperature values of Test
1 (Stud 3) were used for hot and cold flange temperatures. At ambient temperature
(t=0) Kaitila’s (2002) method and Feng and Wang’s (2005b) method using Eurocode
3 Part 1.3 (ECS, 2001) resulted in the same ultimate capacity of 52.2 kN. However,
Feng and Wang’s (2005b) method using Eurocode 3 Part 1.2 (ECS, 2001) and Zhao
et al.’s (2005) method gave lesser ultimate load capacity of 44.6 kN and 44.0 kN,
respectively. The difference between the two methods proposed by Feng and Wang
(2005b) is due to the reduction factor χ x for flexural buckling resistance. In the
study of Zhao et al. (2005) an additional reduction factor was used in addition to the
reduction factor χ x given in Eurocode 3 Part 1.3 (ECS, 2002). This will be
explained in detail in Section 5.5.6.5.
The ultimate loads at ambient temperature were different using the design rules
proposed by Kaitila (2002), Feng and Wang (2005b) and Zhao et al.’s (2005).
However, they all merged together near the failure time when the stud temperature
was increased considerably. At elevated temperatures the influence of several
parameters (kxx, e, χ x , Neff and Mx,eff ) in the design of LSF wall studs is discussed in
detail in the following sections.
5-47
Fire Design Rules
60.0
50.0
Ultimate Load NEd (kN)
40.0
30.0
20.0
10.0
0.0
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
Kaitila (2002) Feng and Wang (2005) EC3 P1.2

Figure 5.45: Variation of Ultimate Load with Time from Previous Fire Design
Rules for Test 1
5.5.6.2. Variation of Effective Eccentricity e

Figure 5.46 shows the variation of effective eccentricity used to find the applied
bending moment in LSF wall studs subjected to non-uniform temperature conditions.
This eccentricity obtained from the previous research studies included thermal
bowing, magnification effects and neutral axis shift (see Section 5.2). However,
according to Figure 5.46, there is not much difference between the lateral deflections
obtained from these previous fire design rules. The difference between Figures 5.3
and 5.46 is the omission of neutral axis shift. The measured lateral deflection from
experimental study and the lateral deflection from FEA will not include the neutral
axis shift. Therefore in Figure 5.3 the neutral axis shift was not considered for
comparison purposes. However, in order to find the axial compression capacity of
LSF wall studs, neutral axis shift should be considered as shown in Figure 5.46. The
neutral axis shift depends on the effective area calculation. In the studies of Kaitila
(2002) and Feng and Wang (2005b) using Eurocode 3 Part 1.2 (ECS, 2001), effective
area was calculated at ambient temperature. Therefore the effective eccentricity is
also almost the same when these methods are used (see Figure 5.46). Similarly in the
studies of Zhao et al. (2005) and Feng and Wang (2005b) using Eurocode 3 Part 1.3
(ECS, 2001), effective area was calculated at elevated temperature. Therefore the
effective eccentricity is also almost the same when these methods are used (see
5-48
Fire Design Rules
Figure 5.46). The small variation between these methods is due to the different
calculation methods used in these studies as discussed in Section 5.2.
25.000
20.000
Effective Eccentricity e (mm)
15.000
10.000
5.000
0.000
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)

Figure 5.46: Variation of Effective Eccentricity with Time from Previous Fire
Design Rules for Test 1
5.5.6.3. Variation of Ultimate Load for Local Buckling Neff

Figure 5.47 shows the variation of ultimate failure load for local buckling using fire
design rules proposed by other researchers. Figure 5.48 shows the effective area used
by these researchers to find these capacities. Kaitila (2002) and Feng and Wang
(2005b) using Eurocode 3 Part 1.2 (ECS, 2001) used the effective area at ambient
temperature (131.3 mm2) to find the ultimate failure load for local buckling.
However, Feng and Wang (2005b) using Eurocode 3 Part 1.3 (ECS, 2001) and Zhao
et al. (2005) used the effective area at elevated temperature based on Eurocode 3 Part
1.3 (ECS, 2002). Therefore the effective area was changing with time as shown in
Figure 5.48. The effective area at elevated temperature was less than the effective
area at ambient temperature up to about 110 minutes (see Figure 5.48). Therefore the
ultimate failure load for local buckling is also less according Feng and Wang (2005b)
using Eurocode 3 Part 1.3 (ECS, 2001) and Zhao et al. (2005) as shown in Figure
5.47. Even though the same effective area was used in Kaitila (2002) and Feng and
Wang (2005b) using Eurocode 3 Part 1.2 (ECS, 2001), the ultimate failure load for
local buckling is slightly different to each other. This is due to the simplicity adopted
in these fire design rules to find the yield stress at elevated temperature. Kaitila
5-49
Fire Design Rules
(2002) used the yield stress at the average web temperature. However, a weighted
average yield stress was used by Feng and Wang (2005b). Regardless of this
difference, these two methods agreed well in the prediction of ultimate failure load
for local buckling.
80.0
70.0
Ultimate Failure Load for Local buckling
60.0
50.0
40.0
Neff (kN)
30.0
20.0
10.0
0.0
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)

Figure 5.47: Variation of Ultimate Failure Load with Time for Local Buckling
from Previous Fire Design Rules for Test 1
160.000
140.000
Effective Area A eff for Compression
120.000
100.000
Capacity (mm 2)
80.000
60.000
40.000
20.000
0.000
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)

Figure 5.48: Variation of Effective Area with Time for Compression Capacity
from Previous Fire Design Rules for Test 1
5-50
Fire Design Rules
5.5.6.4. Variation of Section Moment Capacity Mx,eff

Figure 5.49 shows the variation of section moment capacity Mx,eff with time using
previous fire design rules. Kaitila (2002) found the effective section modulus
corresponding to the flange that is furthest away from the effective centroid.
Therefore the ymax adopted in Kaitila’s (2002) study is the distance from effective
neutral axis to the hot flange. It should be noted that Ranby (1999) calculated the
effective section modulus in relation to the cold flange. In the study of Zhao et al.
(2005), when the compression is on the cold side (at stud mid-height), the moment
resistance was calculated with yielding occurring on both flanges. It is also assumed
that the bi-linear stress distribution in the web is on the hot side. In the study of Feng
and Wang (2005b) two different section moment capacities at mid-height were
calculated considering the first occurrence of material yield on the tension (hot) side
and partial plasticity so that the compression (cold) side reaches yield. However, for
comparison purposes only the latter was plotted in Figure 5.49.
Another difference between these fire design rules is the calculation method used to
find the section moment capacity. Kaitila (2002) found the section moment capacity
by simply multiplying the yield stress at average temperature by effective section
modulus. However, in the studies of Feng and Wang (2005b) and Zhao et al. (2005)
the section moment capacity was found by ∑f Ay
i i i where fi and Ai are the
developed stress and area, respectively, of the individual elements of the reduced
(effective) cross-section. yi is the distance from the effective neutral axis to the force
generated by f i Ai component of the individual elements (of the reduced cross-
section).
Figure 5.50 shows the variation of effective area used to calculate the section
moment capacity. Kaitila (2002) used the effective area for uniform compression (at
ambient temperature) to find the section moment capacity of LSF wall studs at
elevated temperature. Therefore large parts of web and tension flanges were lost in
the calculation of section moment capacity (see Figure 5.50). Hence in this study, the
section moment capacity was less compared to all other previous studies. In the study
of Feng and Wang (2005b) using Eurocode 3 Part 1.2 (ECS, 2001), the effective area
calculated for pure bending at ambient temperature was used to calculate the section
5-51
Fire Design Rules
moment capacity at elevated temperatures. In the studies of Zhao et al. (2005) and
Feng and Wang (2005b) using Eurocode 3 Part 1.3 (ECS, 2001), the effective area
for pure bending was calculated at elevated temperatures. As shown in Figure 5.50,
the effective area at ambient temperature was more than the effective area at elevated
temperatures for most of the time. Hence the section moment capacity is higher for
Feng and Wang (2005b) using Eurocode 3 Part 1.2 (ECS, 2001).
5.5.6.5. Variation of Reduction Factors Used in the Fire Design

Figure 5.51 shows the variation of reduction factor used in the previous fire design
rules for flexural buckling resistance. The usage of imperfection factor α to find the
reduction factor χ x for flexural buckling resistance is an important decision to make
in Eurocode 3. In Eurocode 3 Part 1.3 (ECS, 2006), buckling curve b is suggested for
channel sections to arrive at a value of 0.34 for imperfection factor α. However, in
Eurocode 3 Part 1.2 (ECS, 2005), the value of α is defined as 0.65 235 / f y . The
yield stress of the steel considered in the current study is 569 MPa. Hence an α value
of 0.418 is used in Eurocode 3 Part 1.2 (ECS, 2005). This resulted in lower reduction
factor χ x for flexural buckling resistance in the study of Feng and Wang (2005b)
using Eurocode 3 Part 1.2 (ECS, 2001) (see Figure 5.51). In the study of Zhao et al.
(2005) an additional reduction factor was used with Eurocode 3 Part 1.3 (ECS,
2002). This is the relative slenderness at elevated temperature λθ defined in Equation
5.16. Therefore the effective reduction factor is small compared to other researchers’
values as shown in Figure 5.51.
Figure 5.52 shows the variation of the modification factor kxx used in the previous
fire design rules to take account of non-uniform bending moment distributions in the
stud. In the studies of Kaitila (2002) and Feng and Wang (2005b) this factor was
calculated using the equation given in Eurocode 3 Part 1.3 (see Equations 5.9 (a) and
(b)). It should be noted that even in the study of Feng and Wang (2005b) using
Eurocode 3 Part 1.2 (ECS, 2001), they used the kxx factor given in Eurocode 3 Part
1.3. Eurocode 3 Part 1.3 (ECS, 2001) sets an upper limit of 1.5 for the modification
factor. When the modification factor is calculated according to Eurocode 3 Part 1.3
(ECS, 2001) for the measured temperature values of Test 1, the factor is always more
than 1.5. Hence the value is limited to 1.5 according to Eurocode 3 Part 1.3 (ECS,
5-52
Fire Design Rules
2001) (see Figure 5.52). Zhao et al. (2005) used a value of 1 for this modification
factor kxx.
250.000
Effective Area A eff for Section Moment Capacity
200.000
150.000
(mm 2)
100.000
50.000
0.000
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
Figure 5.49: Variation of Section Moment Capacity with Time from Previous
250.000
Effective Area A eff for Bending Moment Capacity
200.000
150.000
(mm 2)
100.000
50.000
0.000
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
Figure 5.50: Variation of Effective Area with Time for Section Moment
Capacity from Previous Fire Design Rules for Test 1
5-53
Fire Design Rules
0.9
0.8
0.7
Reduction Factor X
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)

Figure 5.51: Variation of Reduction Factor χ with Time from Previous Fire
1.6
1.4
1.2
Modification Factor k xx
1.0
0.8
0.6
0.4
0.2
0.0
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)

Figure 5.52: Variation of Reduction Factor kxx with Time from Previous Fire
5-54
Fire Design Rules
5.6. Proposed Fire Design Rules based on Eurocode 3

The fire design of LSF wall studs using Eurocode 3 was investigated in detail in the
studies of Ranby (1999), Kaitila (2002), Feng and Wang (2005b) and Zhao et al.
(2005). Among these fire design rules, Feng and Wang’s (2005b) proposals using
Eurocode 3 Part 1.3 agreed well with the current FEA and experimental results. The
major differences between these previous researches are the effective areas used and
the calculation of section moment capacities, which were discussed in the previous
sections of this chapter. Therefore in the current study it is important to decide
between the effective areas at ambient and elevated temperatures. This can be
concluded by investigating the section capacities of LSF wall studs under fire
conditions.
5.6.1. Section Capacities of LSF Wall Studs under Fire Conditions

The section capacity of LSF wall stud under fire condition is very hard to determine
using experimental set-up. The reason for this is the temperature gradient across the
stud when it is subjected to non-uniform temperature distributions. Thermal bowing,
neutral axis shift and magnification effects will be induced due to this temperature
gradient. This will result in bending moments in addition to the applied compression
force. Hence it is impossible to test a stud which is subjected to pure compression
with non-uniform temperature distributions. Therefore it was decided to model the
short LSF wall stud using ABAQUS. The half-wave length of the local buckling
waves was 70 mm according to the elastic buckling analyses of CUFSM. Hence a
stud height of 210 mm was modelled to simulate 3 half waves. The thermal bowing
and its magnification effects are eliminated in the FE model by using an extremely
low thermal expansion co-efficient (1x10-20 oC-1). The neutral axis shift and
corresponding bending moments are eliminated by applying the load at a pre-
determined centroid on the new neutral axis calculated based on gross sectional
dimension and reduced elastic modulus. The finite element analyses were carried out
under steady state conditions. In the first step the non-uniform temperature
distributions was applied. In the second step the load was applied to the new centroid
calculated according to the applied temperature. Figure 5.53 shows the ultimate
failure mode of short LSF wall stud under fire conditions. Pure local buckling mode
was obtained as expected.
5-55
Fire Design Rules
Figure 5.53: Ultimate Failure Mode of Short LSF Wall Stud under Fire
Conditions
The local buckling capacity according to Eurocode Part 1.3 (ECS, 2006) was
calculated by multiplying effective area with yield stress at elevated temperatures.
Three possible yield stress values for LSF wall stud subjected to non-uniform
temperature distributions were investigated. They are the yield stress at (web)
average stud temperature (fyweb), the weighted average yield stress of the gross
section (fybar) and the yield stress at hot flange temperature (fyhf). Similarly four
different effective areas were investigated. They are the effective area at ambient
(A20), elevated (At), web (Aw) and hot flange (Ahf) temperatures. Table 5.6 shows the
mechanical properties used in the calculation of these effective areas. Hence twelve
different cases were investigated with FEA results to predict the section capacity of
LSF wall studs under fire conditions (see Figures 5.54 to 5.57).
Table 5.6: Mechanical Properties Used in Different Effective Areas
Elements A20 At Aw Ahf

Web fy20, E20 fyweb, Eweb fyweb, Eweb fyhf, Ehf
Hot Flange and Lip fy20, E20 fyhf, Ehf fyweb, Eweb fyhf, Ehf
Cold Flange and Lip fy20, E20 fycf, Ecf fyweb, Eweb fyhf, Ehf
5-56
Fire Design Rules
80.0
Ultimate local buckling capacity (kN) 70.0
60.0
50.0
40.0
30.0
20.0
10.0
0.0
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
fyw eb*A20 fybar*A20 fyhf*A20 fyw eb*At fybar*At

fyhf*At fyw eb*Aw fybar*Aw fyhf*Aw fyw eb*Ahf
fybar*Ahf fyhf*Ahf FEA
Figure 5.54: Variation of Ultimate Local Buckling Capacity for Test 1
80.0
Ultimate local buckling capacity (kN)
70.0
60.0
50.0
40.0
30.0
20.0
10.0
0.0
0 10 20 30 40 50 60
Tim e (m in.)

Figure 5.55: Variation of Ultimate Local Buckling Capacity for Test 1*
5-57
Fire Design Rules
80.0
70.0
60.0
50.0
40.0
30.0
20.0
10.0
0.0
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)

80.0
70.0
60.0
50.0
40.0
30.0
20.0
10.0
0.0
0 10 20 30 40 50 60 70 80 90 100 110
Time (m in.)

5-58
Fire Design Rules
The local buckling capacities were too low when fyhf was used as the yield stress.
This approach reduces the ultimate capacity considerably irrespective of the various
effective areas used. On the other hand the ultimate capacity results were arbitrary
when Ahf was used. Hence it is not recommended to use the hot flange temperature in
determining the effective widths of individual elements subjected to non-uniform
temperature distribution. Therefore these options were not taken into account for
further investigation. This reduced the number of cases to six from twelve.
Figure 5.58 shows the variation of yield stress with time. Four different tests are
considered in this investigation (T1 – Externally insulated LSF wall with glass fibre;
P1 – Non-insulated LSF wall with single layer of plasterboard; P2 – Non-insulated
LSF wall with double layers of plasterboard P3 – Cavity insulated LSF wall with
glass fibre). This figure compares the difference between the weighted average yield
stress (fybar) and the yield stress at average stud (web) temperature (fyweb). This figure
indicates that there is a noticeable difference between using fybar and fyweb. This
difference is very high for cavity insulated wall panels. The appropriate yield stress
value to represent the cross section at non-uniform temperature is fybar and Figure
5.58 suggests that it cannot be further simplified by using fyweb. This reduced the
number of cases to three from six.
600
500
Yield Stress (MPa)
400
300
200
100
0
0 20 40 60 80 100 120
Tim e (m in.)
T1 fyw eb T1 fybar P1 fyw eb P1 fybar

P2 fyw eb P2 fybar P3 fyw eb P3 fybar
Figure 5.58: Variation of Yield Stress with Time
5-59
Fire Design Rules
1.00
0.90
0.80
0.70
0.60
Load ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
fybar*A20 fybar*At fyw eb*Aw fybar*Aw FEA Proposed
Figure 5.59: Variation of Load Ratio with Time for Test 1
1.00
0.90
0.80
0.70
Load Ratio
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60
Tim e (m in.)
Figure 5.60: Variation of Load Ratio with Time for Test 1*
5-60
Fire Design Rules
1.00
0.90
0.80
0.70
Load Ratio
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
1.00
0.90
0.80
0.70
Load Ratio
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110
Tim e (m in.)
fybar*A20 fybar*At fybar*Aw FEA Series1

Proposed
5-61
Fire Design Rules
Figures 5.59 to 5.62 show the short listed options to find the local buckling capacity
of LSF wall studs with non-uniform temperature distribution. Here, the load ratios
are considered to investigate the non-dimensional values. At higher load ratios the
load ratio curve using the effective areas at elevated temperatures agreed well with
FEA results compared to the load ratio curve using effective area at ambient
temperature. However, at lower load ratios none of them agreed with FEA results.
This variation is much higher for cavity insulated wall panel (see Figure 5.62).
Figure 5.63 shows the von Mises stress distribution at failure for the cavity insulated
wall panel. Table 5.7 provides the hot and cold flange temperatures and the
corresponding yield stress for different tests. The yield stress of the hot and cold
flanges at failure (at 107 minutes) of Test 4* are 52 MPa and 269 MPa, respectively.
Figure 5.63 indicates that the entire hot flange has reached its yield strength capacity
at this time and triggered the failure. At this time the cold flange has not reached its
yield strength capacity. This indicates that the cold flange has some reserve capacity
at the time the stud failed. However, in the prediction of local buckling capacity at
elevated temperatures according to ∑ Ai f yi , traditionally it was assumed that the
entire stud cross-section has reached its yield stress. This resulted in the over-
estimation of local buckling capacity of studs with non-uniform temperature
distributions.
Cold Flange
Hot Flange
Figure 5.63: von Mises Stress Distribution at Failure for Test 4*
5-62
Fire Design Rules
Table 5.7: Yield Stresses at Hot and Cold Flange Temperatures
Time HF CF HF-CF fyHF fyCF

Test fyCF/fyHF
(min.) (oC) (oC) (oC) (MPa) (MPa)
0 20 20 0 569 569 1.00
30 71 56 15 569 569 1.00
60 182 111 71 565 569 1.01
70 218 132 86 561 568 1.01
1
90 368 240 128 450 558 1.24
97 418 296 122 363 543 1.50
110 537 394 143 162 405 2.50
118 656 492 164 50 236 4.72
0 20 20 0 569 569 1.00
9 97 74 23 569 569 1.00
20 184 88 96 565 569 1.01
1* 30 346 187 159 486 565 1.16
46 535 346 189 165 486 2.95
50 570 395 175 109 404 3.71
53 595 467 128 71 278 3.92
0 20 20 0 569 569 1.00
30 93 83 10 569 569 1.00
60 215 128 87 562 568 1.01
2* 80 354 259 95 473 554 1.17
90 416 313 103 367 531 1.45
105 497 408 89 227 381 1.68
111 555 446 109 133 314 2.36
0 20 20 0 569 569 1.00
60 165 91 74 567 569 1.00
80 388 162 226 416 567 1.36
4* 90 476 200 276 263 564 2.14
94 515 213 302 197 562 2.85
100 590 238 352 78 558 7.15
107 648 269 379 52 551 10.6
Table 5.8 shows the ultimate loads and load ratios obtained from FEA and prediction
(fybar*At). The load ratio agreement was good during the initial period of the tests.
However, the agreement was very poor during the last phase for all the tests. In this
period it was identified that the ratio of fyCF/fyHF was greater than 1.5 (see Table 5.7).
Therefore it was decided to limit the yield stress of cold flange to 1.5 fyHF during this
time in the determination of local buckling capacity. Appendix L shows the
calculations of the proposed method and Table 5.8 shows the results obtained. This
indicates a better agreement in load ratios with FEA results for local buckling
capacity of LSF wall studs subjected to non-uniform temperature distributions.
5-63
Fire Design Rules
Table 5.8: Comparison of FEA with Current and Proposed Design Rules
FEA fybar * At Proposed

Test Time
Load LR Load LR Load LR
0 77.9 1.00 74.7 1.00 74.7 1.00
30 77.1 0.99 73.3 0.98 73.3 0.98
60 75.6 0.97 70.3 0.94 70.3 0.94
70 74.7 0.96 68.9 0.92 68.9 0.92
1
90 64.7 0.83 59.7 0.80 59.3 0.79
97 57.1 0.73 53.4 0.71 52.9 0.71
110 32.1 0.41 37.8 0.51 30.0 0.40
118 13.9 0.18 20.4 0.27 12.4 0.17
0 77.9 1.00 74.7 1.00 74.7 1.00
9 76.6 0.98 72.5 0.97 72.5 0.97
20 75.7 0.97 70.6 0.95 70.6 0.95
1* 30 68.1 0.87 62.7 0.84 62.4 0.84
46 36.9 0.47 42.3 0.57 31.2 0.42
50 25.8 0.33 35.7 0.48 23.8 0.32
53 20.2 0.26 26.5 0.35 17.4 0.23
0 77.9 1.00 74.7 1.00 74.7 1.00
30 76.6 0.98 72.4 0.97 72.4 0.97
60 74.8 0.96 69.1 0.93 69.1 0.93
2* 80 65.4 0.84 59.6 0.80 59.3 0.79
90 56.5 0.73 52.5 0.70 52.1 0.70
105 39.0 0.50 39.2 0.52 37.1 0.50
111 26.8 0.34 31.7 0.42 25.8 0.35
0 77.9 1.00 74.7 1.00 74.7 1.00
60 75.6 0.97 70.9 0.95 70.9 0.95
80 64.8 0.83 61.4 0.82 60.6 0.81
4* 90 51.2 0.66 55.6 0.74 46.7 0.63
94 44.4 0.57 52.7 0.71 39.1 0.52
100 27.1 0.35 47.5 0.64 24.3 0.33
107 19.2 0.25 42.1 0.56 18.7 0.25
Figure 5.64 shows the proposed method incorporated with the short listed options to
find the local buckling capacity. It clearly indicates that the effective area at ambient
temperature (A20) is not suitable for determining the local buckling capacity of LSF
wall studs at elevated temperatures. This figure also shows that the simplification of
using effective area at average stud (web) temperature (Aweb) is too conservative
compared to the effective area at elevated temperature (At). Therefore it is concluded
that the effective element widths should be determined accurately using their
respective elevated temperature properties. If effective areas are calculated at
ambient or web temperatures, the calculations are slightly less involved. However,
5-64
Fire Design Rules
this is at the expense of relatively high inaccuracy in the predicted ultimate local
buckling capacity. Therefore this is not recommended. Figures 5.59 to 5.62 show the
close agreement between the FEA results and proposed method with effective area at
elevated temperature (At).
1.00
0.90
0.80
0.70
Load Ratio
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110
Tim e (m in.)
Using At Using Aw Using A20 FEA
Figure 5.64: Proposed Method for the Selected Options
5.6.2. Member Capacities of LSF Wall Studs under Fire Conditions
Ranby (1999) and Kaitila (2002) used the mechanical properties at average stud
(web) temperatures to find the ultimate compression capacities of LSF wall studs
subjected to non-uniform temperature distributions. However, Feng and Wang
(2005b) and Zhao et al. (2005) used the weighted average mechanical properties
based on gross section dimensions. Based on Figure 5.58, it is clear that the weighted
average mechanical properties should be used instead of the mechanical properties at
web temperature. Hence in the current study the weighted average mechanical
properties were used with Eurocode Part 1.3 (ECS, 2006) design rules to determine
the ultimate compression capacities of LSF wall studs under fire conditions.
Appendix L shows the ultimate compression capacity calculations of LSF wall studs
at elevated temperatures.
5-65
Fire Design Rules
5.6.3. Section Moment Capacities of LSF Wall Studs under Fire Conditions
Ranby (1999) and Kaitila (2002) used the basic formula f y , web Z eff , 20 to find the
section moment capacities of LSF wall studs subject to non-uniform temperature

distributions. They used the yield stress at web temperature and the effective section
modulus at ambient temperature. The section modulus was calculated based on the
effective element widths for pure compression. However, it is important that the
effective element widths based on pure bending are used. Feng and Wang (2005b)
suggested a more accurate method to find the section moment capacity. When
calculating Mx,eff at stud mid-height, compression is on the cold flange side with a
high yield strength and tension is on the hot side with a low yield strength. In this
case partial plasticity was considered whereby tensile stress in the extreme fibres has
reached yield and the maximum compression stress in the extreme fibre is equal to
the yield stress as shown in Table 5.9. Alternatively the lower of first yield in either
tension or compression is considered. In the current study the first option (partial
plasticity) suggested by Feng and Wang (2005b) was used in calculating the section
moment capacity at mid-height. At the supports since the hot flange is in
compression, the first yield of compression flange was adopted. Detail calculations
can be found in Appendices I and J. In the study of Zhao et al. (2005), at mid-height
when compression is on the cold side the moment resistance of the cross-section is
calculated with yielding occurring on both flanges with bi-linear stress distribution in
the web on the hot side. At the support when compression is on the hot side of the
stud, the moment resistance is calculated with yielding occurring on the hot flange
and linear stress distribution in the web. This method is similar to the method
proposed by Feng and Wang (2005b).
Figures 5.65 to 5.68 show the interaction of compression and bending in four
different tests using the ratio of applied axial compression force and axial
compression capacity (N*/Nu) and the ratio of applied bending moment and section
moment capacity (M*/Mu). The influence of bending is more at mid-height compared
to the support. Similarly the influence is more for LSF walls with cavity insulation
compared to the externally insulated wall panels (see Figures 5.65 and 5.68).
However, overall, these figures clearly indicate that the LSF wall studs under non-
uniform temperature distributions are dominated by compression than bending.
Therefore it was decided to propose a simplified method to calculate the section
5-66
Fire Design Rules
moment capacity without affecting the accuracy of ultimate compression capacities

of LSF wall studs. In the current proposal, the section moment capacity at mid-height
was calculated using the formula f y Z eff ,t where f y is the weighted average yield
stress and Z eff ,t is equal to I eff ,t / Ymax . In the studies of Feng and Wang (2005b) and
Zhao et al. (2005), partial plasticity was considered at mid-height whereby tensile
stress in the extreme fibres has reached yield and the maximum compression stress at
the extreme fibre is equal to the yield stress. Therefore in this scenario f y is suitable
to calculate the section moment capacity. I eff ,t is the weighted average second
moment of area (taking into account the variation of elastic modulus across the
section) calculated based on the effective element widths at elevated temperatures. In
the studies of Feng and Wang (2005b) and Zhao et al. (2005), the moment resistance
at the support was calculated with yielding occurring in the hot flange. In this case
f y , HF is suitable to calculate the section moment capacity instead of f y . Therefore it
is proposed to calculate the section moment capacity at support using the

formula f y , HF Z eff ,t . Detail calculations can be found in Appendix L.
Table 5.9: Stress Distribution for Section Moment Capacities
At Support
At Mid-
height
5-67
Fire Design Rules
1.0
0.9
0.8
0.7
N* / Nu , M* / M u
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
Mid-height N*/ Nu Mid-height M*/ Mu Support N*/ Nu Support M*/ Mu
Figure 5.65: Interaction of Compression and Bending for Test 1
1.0
0.9
0.8
0.7
N* / Nu , M* / M u
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0 10 20 30 40 50 60
Tim e (m in.)
Figure 5.66: Interaction of Compression and Bending for Test 1*
5-68
Fire Design Rules
1.0
0.9
0.8
0.7
N* / Nu , M* / M u
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
1.0
0.9
0.8
0.7
N* / Nu , M* / M u
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0 10 20 30 40 50 60 70 80 90 100 110
Tim e (m in.)
5-69
Fire Design Rules
Figures 5.69, 5.71, 5.73 and 5.75 show the variation of section moment capacities
(Mxeff) with time for four different tests according to Feng and Wang (2005b) and the
current proposal. Both mid-height and supports are considered in these figures. A
reasonable agreement was achieved between the accurate and simplified methods.
Figures 5.70, 5.72, 5.74 and 5.76 show the variation of ultimate compression
capacities (Neff) of LSF wall studs with time at mid-height and supports for four
different tests. It is very clear that the ultimate compression capacities were not
affected by using the simplified method for section moment capacities. This is due to
the fact that the LSF wall studs subjected to non-uniform temperature distributions
are dominated by compression rather than bending as shown in Figures 5.65 to 5.68.
Therefore it is concluded that the section moment capacities can be calculated using
the simplified method without affecting the accuracy of ultimate compression
capacities of LSF wall studs.
5.6.4. Interaction of Compression and Bending

The effect of thermal bowing can be considered as that of an initial geometric
imperfection of a slender member with pinned ends. The sinusoidal initial deformed
shape is represented by,
 πx 
y o = e∆T sin   (5.22)
L
where e∆T is the thermal bowing at mid-height and L is the stud height.
In this case the maximum total deformation of the member (e1) is obtained at mid-
height as (Luis et al., 2010),
e ∆T
e1 = (5.23)
N*
1−
N cr
where N* is the applied load and Ncr is the Euler buckling load.
Therefore the bending moment generated by thermal bowing and its magnification
effect is given by
N * e ∆T
N e1 =
*
(5.24)
N*
1−
N cr
5-70
Fire Design Rules
According to Eurocode 3 Part 1.3 (ECS, 2006) the bending moment due to the shift
of neutral axis about the major axis and its magnification effects is given by
N * e2 = k xx N * e∆E (5.25)
assuming the LSF wall stud is not susceptible to torsional deformation,
C mx
where k xx =
N*
1− χx
N cr
Cmx allows for the effects of non-uniform distribution of bending moment and
N*
applied axial compression load C mx = 0.79 + 0.21ψ + 0.36(ψ − 0.33) where
N cr
ψ = 1 in this study since the moment due to neutral axis shift is equal at both ends.
Therefore the total moment due to thermal bowing, neutral axis shift and their
magnification effects can be calculated as,
N * e ∆T N * e∆E C mx
M = N (e1 − e2 ) =
* *
− (5.26)
N* N*
1− 1− χx
N cr N cr
Equations 5.9 (a) and (b) were used to obtain the ultimate compression capacities of
LSF wall studs subjected to non-uniform temperature distributions. In the current
study, these equations are reduced to
N* M*
+ =1
χN eff M x,eff
where
Neff is obtained as discussed in Section 5.6.1;
χ is obtained as discussed in Section 5.6.2;
Mx,eff is obtained as discussed in Section 5.6.3
M* is obtained as discussed in Section 5.6.4.
It should be noted that in the studies of Ranby (1999) and Kaitila (2002), the
1
magnification effects were counted twice by the use of the factors and kxx.
N*
1−
N cr
5-71
Fire Design Rules
This is not recommended in the fire design of LSF wall studs and hence the proposed
method should be used.
3.5
Ultimate Section Moment Capacity
3.0
2.5
2.0
(kNm)
1.5
1.0
0.5
0.0
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
Mid-height - Feng and Wang (2005) Mid-height - Proposed

Support - Feng and Wang (2005) Support - Proposed
Figure 5.69: Variation of Section Moment Capacity for Test 1
60.0
Ultimate Compression Capacity (kN)
50.0
40.0
30.0
20.0
10.0
0.0
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)

Figure 5.70: Variation of Ultimate Compression Capacity for Test 1
5-72
Fire Design Rules
Ultimate Section Moment Capacity 3.5
3.0
2.5
2.0
(kNm)
1.5
1.0
0.5
0.0
0 10 20 30 40 50 60
Tim e (m in.)

Figure 5.71: Variation of Section Moment Capacity for Test 1*
60.0
50.0
40.0
30.0
20.0
10.0
0.0
0 10 20 30 40 50 60
Tim e (m in.)

Figure 5.72: Variation of Ultimate Compression Capacity for Test 1*
5-73
Fire Design Rules
3.5
Ultimate Section Moment Capacity
3.0
2.5
2.0
(kNm)
1.5
1.0
0.5
0.0
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)

60.0
50.0
40.0
30.0
20.0
10.0
0.0
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)

5-74
Fire Design Rules
Ultimate Section Moment Capacity 3.5
3.0
2.5
2.0
(kNm)
1.5
1.0
0.5
0.0
0 10 20 30 40 50 60 70 80 90 100 110
Tim e (m in.)

60.0
50.0
40.0
30.0
20.0
10.0
0.0
0 10 20 30 40 50 60 70 80 90 100 110
Tim e (m in.)
Mid-height - Feng et al. (2005) Mid-height - Proposed

Support - Feng et al. (2005) Support - Proposed
5-75
Fire Design Rules
5.6.5. Thermal Bowing, Neutral Axis Shift and Magnification Effects
When the axial compression load is applied at the original centroid of the cross-
section, the effective eccentricity in LSF wall studs under non-uniform temperature
conditions can be developed mainly due to thermal bowing, neutral axis shift and
magnification effects. In the studies of Klippstein (1980) and Gerlich et al. (1996),
only the thermal bowing and its magnification effects were considered in the fire
design of LSF wall studs. However, all three effects were considered in the studies of
Ranby (1999), Kaitila (2002), Feng and Wang (2005b) and Zhao et al. (2005). In the
current study also the effect of considering these parameters was investigated as four
different cases (see Table 5.10).
Figure 5.77 shows the variation of load ratios with respect to time using these four
different effective eccentricities. When the magnification effect of thermal bowing
was ignored the failure time was significantly overestimated. On the other hand when
the neutral axis shift was ignored the variation was small, although a better
agreement was obtained with FEA results (compare Cases 1 and 3). In the current
study it is recommended to include thermal bowing, magnification effects and
neutral axis shift in the fire design of LSF wall panels since they all play a major role
when the stud is subjected to non-uniform temperature distributions.
Table 5.10: Effective Eccentricities Considered in the Current Study
Case Effective Eccentricity

Thermal bowing, Magnification effects and Neutral axis shift
1
are included (e∆T +eP∆T – e∆E)
Thermal bowing and Neutral Axis Shift are included
2
(e∆T - e∆E )
Thermal bowing and Magnification effects are included
3
(e∆T +eP∆T )
4 Only Thermal bowing effect is included e∆T
5-76
Fire Design Rules
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)
FEA EC3 P1.3 Case 1 EC3 P1.3 Case 2

EC3 P1.3 Case 3 EC3 P1.3 Case 4
5.7. Design Rules given in AS/NZS 4600 (SA, 2005)
5.7.1. Concentrically Loaded Compression Members at Ambient Temperature

The capacity of LSF wall studs subject to axial compression can be calculated using
Clause 3.4 of AS/NZS 4600 (SA, 2005) as
N * ≤ φc N s (5.27)
N * ≤ φc N c (5.28)
where
φc is the capacity reduction factor;
Ns is the nominal section capacity of the member in compression calculated using
N s = Ae f y where Ae is the effective area at yield stress fy;
Nc is the nominal member capacity of the member in compression calculated using

N c = Ae f n where Ae is the effective area at the critical stress fn; fn shall be
determined from,
5-77
Fire Design Rules
(
f n = 0.658λc f y
2
) for λc ≤ 1.5 (5.29a)
 0.877 
f n =  2  f y for λc > 1.5 (5.29b)
 λc 
where
fy
λc is the non-dimensional slenderness and can be calculated as λc = where foc is
f oc
the least of the elastic flexural, torsional and flexural-torsional buckling stresses
calculated as follows,
The elastic buckling stress foc shall be determined from Clause 3.4.2 as
π 2E
f oc = 2
(5.30)
 le 
 
r
where le is the effective length of member and r is the radius of gyration of the full,
unreduced cross-section.
5.7.2. Combined Axial Compression and Bending at Ambient Temperature

The capacity of the LSF wall stud subject to both axial compression and bending
moment can be calculated using Clause 3.5.1 as
N* C mx M x* C my M *y
+ + ≤1 (5.31)
φc N c φb M bxα nx φb M byα ny
where Nc is the nominal member capacity of the member in compression calculated

calculated using the last section; Cmx and Cmy are the coefficients for unequal end
moments whose values shall be taken as 1 conservatively;
Mbx and Mby are the nominal member moment capacities about the x- and y-axes,
respectively;
φb and φc are the capacity reduction factors whose values shall be taken as 1 in this
case;
αnx and αny are the moment amplification factors and can be calculated using
N* π 2 EI b
α = 1− where Ne is the elastic buckling load N e = where Ib is the second
Ne (leb )2
moment of area of the full, unreduced cross-section about the bending axis and leb is
the effective length in the plane of bending.
5-78
Fire Design Rules
5.8. Proposed Fire Design Rules based on AS/NZS 4600 (SA, 2005)
As discussed in Section 5.6, Equation 5.32 can be used to obtain the ultimate
compression capacities of LSF wall studs subjected to non-uniform temperature
distributions.
N* M*
+ =1 (5.32)
Aeff ,t f n M x ,eff
where
Aeff,t is the effective area at elevated temperature;
fn is calculated based on the weighted average mechanical properties at elevated
temperatures;
Mx,eff is the section moment capacity calculated as discussed in Section 5.6.3;
The total moment M* due to thermal bowing, neutral axis shift and their
magnification effects is given by
C mx N * e N * e ∆T N * e∆E
M =*
= − (5.33)
α nx N* N*
1− 1−
N cr N cr
where Cmx is equal to 1 in the case of neutral axis shift as the moments developed in
this case is uniform while the total moment due to thermal bowing is shown to be
N * e ∆T
(see Equation 5.24).
N*
1−
N cr
Appendix M shows the calculations in detail to find the ultimate capacity of LSF
wall studs at elevated temperatures according to AS/NZS 4600 (SA, 2005).
Figures 5.78 to 5.87 compare the variation FEA load ratios with predicted load ratios
based on AS/NZS 4600 (SA, 2005) and Eurocode 3 Part 1.3 (ECS, 2006). A
reasonable agreement was obtained between the FEA results and the predictions. A
very good agreement was obtained between AS/NZS 4600 (SA, 2005) and Eurocode
3 Part 1.3 (ECS, 2006) although the latter is slightly conservative in predicting
failure times in the case of lower load ratios.
5-79
Fire Design Rules
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)
FEA EC3 Part 1.3 AS/NZS 4600
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)
5-80
Fire Design Rules
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min.)
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60
Time (min.)
5-81
Fire Design Rules
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)
5-82
Fire Design Rules
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110
Time (min.)
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)
5-83
Fire Design Rules
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160
Time (min.)
1
0.9
0.8
0.7
Load Ratio
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130
Time (min.)
5-84
Fire Design Rules
5.9. Comparison of Previous and Proposed Fire Design Rules
Table 5.11 shows the effective areas used in the previous and the proposed fire
design rules. It shows the different effective areas used in finding the compression
and section moment capacities. Table 5.12 shows the parameters and assumptions
used with these fire design rules including thermal bowing, neutral axis shift and
magnification effects.
5.9.1. Failure Times Obtained from Previous and Proposed Fire Design Rules
Tables 5.13 to 5.15 show the failure times obtained from the previous and the
proposed fire design rules for load ratios of 0.2, 0.4 and 0.7, respectively.
Klippstein’s (1980) method resulted in arbitrary failure times compared to test and
FEA results as discussed in Section 5.3.3. Hence failure times were not achieved in a
few cases. Gerlich’s (1995) fire design rules overestimate the failure times of LSF
wall studs considerably. This is due to the usage of cold flange temperature in the
fire design as discussed before.
Feng and Wang’s (2005b) fire design rules agreed reasonably well with the test and
FEA results. Among the two methods proposed by them, the fire design rules based
on Eurocode 3 Part 1.3 (ECS, 2001) agreed well with the test and FEA results.
However, Feng and Wang’s (2005b) method involves complex calculations and
hence is not suitable for routine designs. Zhao et al.’s (2005) fire design rules also
involve complex calculations as Feng and Wang (2005b). Ranby’s (1999) method
also slightly overestimated the failure times due to the assumption in finding the
distance from the neutral axis to the extreme fibre. Kaitila (2002) improved the fire
design rules proposed by Ranby (1999) by modifying the distance from the neutral
axis to the extreme fibre. Hence Kaitila’s (2002) predicted failure times agreed well
with test and FEA results. The uniform temperature method discussed in Kaitila
(2002) can be used when the load ratio is as low as 0.2. However, it overestimates
the failure times of LSF wall studs when the load ratio is more than 0.4.
Tables 5.16 and 5.17 compare the test and FEA results with the predicted failure
times using the proposed fire design rules based on AS/NZS 4600 (SA, 2005) and
Eurocode 3 Part 1.3 (ECS, 2006), respectively. The agreement was very good and it
5-85
Fire Design Rules
is concluded that the proposed fire design rules using AS/NZS 4600 (SA, 2005) and
Eurocode 3 Part 1.3 (ECS, 2006) accurately predict the failure times of LSF wall
panels subject to fire from one side.
Table 5.11: Effective Areas for Previous and Proposed Fire Design Rules
Previous Fire Effective Area for Section Modulus Zeff for

design Rules Compression Capacity Bending Capacity
Effective area was calculated
Klippstein
assuming uniform compression Gross area was used
(1980)
at elevated temperatures
Effective area was calculated
Gerlich et al.
assuming uniform compression Gross area was used
(1996)
at ambient temperatures
Effective area was calculated Effective area was calculated
Ranby (1999) assuming uniform compression assuming uniform compression
at ambient temperatures at ambient temperatures
Kaitila (2002) assuming uniform compression assuming uniform compression
at ambient temperatures at ambient temperatures
Feng and
Wang (2005)
assuming uniform compression assuming pure bending at
Eurocode 3
at ambient temperatures ambient temperatures
Part 1.2
Feng and
Wang (2005)
Eurocode 3
at elevated temperatures elevated temperatures
Part 1.3
Zhao et al.
(2005)
at elevated temperatures elevated temperatures
Proposed in Effective area was calculated Effective area was calculated
the Current assuming uniform compression assuming pure bending at
Study at elevated temperatures elevated temperatures
5-86
Fire Design Rules
Table 5.12: Comparison of Previous and Proposed Fire Design Rules
Fire Design Klippstein Gerlich et Ranby Kaitila Feng and Feng and Zhao et al. Proposed Fire
Rules (1980) al. (1996) (1999) (2002) Wang (2005) Wang (2005) (2005) Design Rule
Standard EC3 EC3 EC3 EC3 EC3 AS/NZS 4600
AISI AISI
Used Part 1.3 Part 1.3 Part 1.2 Part 1.3 Part 1.3 EC3 Part 1.3
Weighted Weighted Weighted Weighted
Yield stress Yield stress Yield stress Yield stress
average yield average yield average yield average yield
at average at cold at average at average
Yield Stress stress at stress at stress at stress at
stud flange stud stud
elevated elevated elevated elevated
temperature temperature temperature temperature
temperature temperature temperature temperature
Elastic Elastic Weighted Weighted Weighted Weighted
Elastic Elastic
modulus at modulus at average elastic average elastic average elastic average elastic
Elastic modulus at modulus at
average average modulus at modulus at modulus at modulus at
Modulus average stud average stud
stud stud elevated elevated elevated elevated
temperature temperature
temperature temperature temperature temperature temperature temperature
Thermal
Considered Considered Considered Considered Considered Considered Considered Considered
bowing
Magnification
Considered Considered Considered Considered Considered Considered Considered Considered
effects
Neutral axis Not Not
Considered Considered Considered Considered Considered Considered
shift Considered Considered
Capacity Mid-height Mid-height Mid-height Mid-height
Mid-height Mid-height Mid-height Mid-height
Checks at and Stud End and Stud End and Stud End and Stud End
5-87
Fire Design Rules
Table 5.13: Failure Times Predicted by FEA and the Previous Fire Design Rules for Load Ratio of 0.2
Failure Time (min.)

Gerlich Kaitila Kaitila Feng and Feng and
Test Configuration Insulation Klippstein Ranby Zhao et al.
Test FEA et al. (2002) (2002) Wang (2005) Wang (2005)
(1980) (1999) (2005)
(1996) U NU EC3 P1.2 EC3 P1.3
Glass
1 118 115 - 120 117 114 117 116 118 119
Fibre
Glass
2 - 116 - 125 119 115 117 118 119 121
Fibre
Rock
3 - 143 - 144 139 136 138 139 139 141
Fibre
1* None 53 53 - 56 55 51 55 55 56 56
2* None 111 115 - 122 116 113 116 117 117 119
Glass
3* 101 100 - 105 105 96 102 102 103 107
Fibre
Rock
4* 107 105 - - 116 99 109 109 110 120
Fibre
Cellulose
5* 110 109 - - 112 107 110 113 111 114
Fibre
Rock
6* 136# 154 - 155 152 149 152 153 155 154
Fibre
Cellulose
7* 124 129 - 133 130 126 129 130 131 132
Fibre
( * ) - Tests conducted by Kolarkar (2010); ( # ) - Experimental error; (U) - Uniform Temperature Method; (NU) - Non-uniform Temperature Method
Tests 1 and 1* - 7* were conducted under a load ratio of 0.2; Tests 2 and 3 were conducted under a load ratio of 0.4
5-88
Fire Design Rules
Failure Time (min.)

(1980) (1999) (2005)
(1996) U NU EC3 P1.2 EC3 P1.3
Glass
1 - 109 108 111 108 108 107 108 107 109
Fibre
Glass
2 108 110 - 113 110 110 109 111 109 111
Fibre
Rock
3 134 131 127 134 130 130 128 132 127 132
Fibre
1* None - 42 44 48 43 43 41 45 40 45
2* None - 107 105 109 105 106 105 107 104 108
Glass
3* - 88 100 90 92 90 88 91 87 92
Fibre
Rock
4* - 91 - 99 95 93 91 94 89 96
Fibre
Cellulose
5* - 101 103 107 104 103 102 104 100 104
Fibre
Rock
6* - 137 - 138 134 135 133 137 132 138
Fibre
Cellulose
7* - 119 117 121 117 117 116 119 115 119
Fibre
( * ) - Tests conducted by Kolarkar (2010); (U) - Uniform Temperature Method; (NU) - Non-uniform Temperature Method
5-89
Fire Design Rules
Failure Time (min.)

(1980) (1999) (2005)
(1996) U NU EC3 P1.2 EC3 P1.3
Glass
1 - 72 80 79 76 91 74 82 75 83
Fibre
Glass
2 - 85 84 89 87 100 86 93 85 93
Fibre
Rock
3 - 95 99 98 97 113 96 101 96 103
Fibre
1* None - 20 26 23 21 32 21 24 22 26
2* None - 63 67 69 66 83 64 75 64 74
Glass
3* - 62 63 65 63 77 62 65 63 68
Fibre
Rock
4* - 64 63 65 64 79 63 66 64 69
Fibre
Cellulose
5* - 64 64 67 64 82 63 68 64 72
Fibre
Rock
6* - 99 96 102 101 114 100 105 100 105
Fibre
Cellulose
7* - 87 92 91 89 104 88 93 89 95
Fibre
( * ) - Tests conducted by Kolarkar (2010); (U) - Uniform Temperature Method; (NU) - Non-uniform Temperature Method
5-90
Fire Design Rules
Table 5.16: Failure Times Predicted by the Proposed Fire Design Rules Using AS/NZS 4600 (SA, 2005)
Load Ratio = 0.2 Load Ratio = 0.4 Load Ratio = 0.7

Failure Time Failure Time Failure Time
Ratio Ratio Ratio
Test Configuration Insulation (min.) (min.) (min.)
Test FEA Prop. T/P F/P Test FEA Prop. T/P F/P Test FEA Prop. T/P F/P
Glass
1 118 115 117 1.009 0.983 - 109 109 - 1.000 - 72 82 - 0.878
Fibre
Glass
2 - 116 116 - 1.000 108 110 111 0.973 0.991 - 85 92 - 0.924
Fibre
Rock
3 - 143 137 - 1.044 134 131 132 1.015 0.992 - 95 101 - 0.941
Fibre
1* None 53 53 56 0.946 0.946 - 42 44 - 0.955 - 20 25 - 0.800
2* None 111 115 116 0.957 0.991 - 107 108 - 0.991 - 63 73 - 0.863
Glass
3* 101 100 97 1.041 1.031 - 88 91 - 0.967 - 62 67 - 0.925
Fibre
Rock
4* 107 105 101 1.059 1.040 - 91 94 - 0.968 - 64 67 - 0.955
Fibre
Cellulose
5* 110 109 108 1.019 1.009 - 101 103 - 0.981 - 64 70 - 0.914
Fibre
Rock
6* 136# 154 155 - 0.994 - 137 138 - 0.993 - 99 105 - 0.943
Fibre
Cellulose
7* 124 129 130 0.954 0.992 - 119 119 - 1.000 - 87 93 - 0.935
Fibre
Mean 0.998 1.003 0.994 0.984 0.908
-
COV 0.046 0.029 0.030 0.016 0.052
( * ) - Tests conducted by Kolarkar (2010); ( # ) - Experimental error; (T) - Test; (P) - Proposed; (F) - FEA
5-91
Fire Design Rules
Table 5.17: Failure Times Predicted by Proposed Fire Design Rules Using Eurocode 3 Part 1.3 (ECS, 2006)
Load Ratio = 0.2 Load Ratio = 0.4 Load Ratio = 0.7

Failure Time Failure Time Failure Time
Ratio Ratio Ratio
Test Configuration Insulation (min.) (min.) (min.)
Test FEA Prop. T/P F/P Test FEA Prop. T/P F/P Test FEA Prop. T/P F/P
Glass
1
Fibre 118 115 115 1.026 1.000 - 109 107 - 1.019 - 72 82 - 0.878
Glass
2
Fibre - 116 115 - 1.009 108 110 109 0.991 1.009 - 85 92 - 0.924
Rock
3
Fibre - 143 137 - 1.044 134 131 128 1.047 1.023 - 95 101 - 0.941
1* None
53 53 53 1.000 1.000 - 42 41 - 1.024 - 20 25 - 0.800
2* None
111 115 114 0.974 1.009 - 107 105 - 1.019 - 63 73 - 0.863
Glass
3*
Fibre 101 100 98 1.031 1.020 - 88 89 - 0.989 - 62 67 - 0.925
Rock
4*
Fibre 107 105 101 1.059 1.040 - 91 91 - 1.000 - 64 68 - 0.941
Cellulose
5*
Fibre 110 109 108 1.019 1.009 - 101 102 - 0.990 - 64 70 - 0.914
Rock
6*
Fibre 136# 154 153 - 1.007 - 137 133 - 1.030 - 99 105 - 0.943
Cellulose
7*
Fibre 124 129 128 0.969 1.008 - 119 116 - 1.026 - 87 93 - 0.935
Mean 1.011 1.014 1.019 1.013 - 0.906
COV 0.032 0.015 0.039 0.015 0.051
( * ) - Tests conducted by Kolarkar (2010); ( # ) - Experimental error; (T) - Test; (P) - Proposed; (F) - FEA
5-92
Fire Design Rules
5.9.2. Modifications Recommended for Previous Fire Design Rules

Klippstein’s (1980b) study was based on the old AISI design manual (AISI, 1968).
He also suggested using the measured deflection of LSF wall panels under fire
conditions. Hence this method is not recommended for the fire design of LSF wall
studs. In the study of Gerlich et al. (1996), the yield stress of the cold flange was
used to find the ultimate capacity of LSF wall studs at elevated temperatures.
However, this method overestimated the compression capacity of LSF wall studs. In
the cavity insulated wall panels, the cold flange temperature is much lower than the
hot flange temperature of the stud. Therefore the predicted failure time was very high
for this type of wall panels. Hence a modification is recommended for Gerlich et al.
(1996) fire design rules whereby the weighted average yield stress at elevated stud
temperature is used in the fire design of LSF wall studs.
Ranby (1999) and Kaitila (2002) used the effective area at ambient temperature to
find the ultimate capacity of LSF studs subject to non-uniform temperature
conditions. However, it is shown in the current study that effective area at elevated
temperature should be used. In the determination of section moment capacity, they
used the effective area for pure compression. This is not recommended and the
effective area for bending case should be calculated. In combining the effects of axial
compression and bending, they have taken the effects of magnification effects twice
in the form of kxx and Equation 5.3 (a). Hence it is recommended to use the
magnification factors given in Eurocode 3 Part 1.3 (Equation 5.26) and AS/NZS
4600 (Equation 5.33).
Feng and Wang (2005b) method to find the ultimate capacity of LSF studs under fire
conditions was based on Eurocode 3 Part 1.3 (ECS, 2001). However, Eurocode 3
Part 1.3 was revised in 2006 and hence new design rules are needed. Their
calculation methods to find the section moment capacities at non-uniform
temperatures were extremely complex. Hence a suitable method is proposed in the
current study using Eurocode 3 Part 1.3 (ECS, 2006) to reduce the complexity
involved in the fire design of LSF wall studs. Zhao et al.’s (2005) method is similar
to Feng and Wang’s method and hence has the same short comings. In addition they
used a reduction factor (relative slenderness at elevated temperature λ θ ) in the
5-93
Fire Design Rules
determination of ultimate capacity under compression. The reason for this is not
known and is not recommended based on this study.
5.10. Conclusion
This chapter has presented the details of an investigation of the fire design rules for
LSF wall studs. The behaviour of LSF wall studs subjected to non-uniform elevated
temperature conditions was analysed in detail. The lateral deflections of the studs due
to non-uniform temperature conditions were predicted using the available equations
proposed by previous researches. It was found that these deflections agreed
reasonably well with the lateral deflections obtained from finite element analyses.
Applications of the previously developed fire design rules based on Eurocode 3 Parts
1.2 and 1.3 to LSF wall studs were investigated in detail and suitable modification
were proposed where necessary. In the prediction of the local buckling (section)
capacity at elevated temperatures, traditionally it is assumed that the entire stud has
reached its yield stress. However, this resulted in considerable over-estimation of the
local buckling capacity of studs with non-uniform temperature distributions.
Therefore it is recommended that the yield stress of cold flange is limited to 1.5
times of the yield stress of hot flange in the local buckling capacity calculations. It is
also recommended that the weighted average mechanical properties are used instead
of the mechanical properties at the web temperature in the ultimate compression
capacity calculations of LSF wall studs under fire conditions. A Simplified method
was proposed to determine the section moment capacity that had minimal effect on
the accuracy of the ultimate compression capacity of LSF wall studs.
New fire design rules based on Eurocode 3 Part 1.3 (ECS, 2006) were proposed in
the current study with suitable allowances for the interaction effects of compression
and bending actions. Previously developed fire design rules based on AISI design
manual were also investigated in detail and suitable modifications were proposed
where necessary. New fire design rules were also developed for LSF wall studs
based on AS/NZS 4600 (SA, 2005). The accuracy of the proposed fire design rules
was verified with the available test and FEA results. The agreement of failure times
was very good compared to the complexity and assumptions involved in the fire
deign of LSF wall studs.
5-94
Parametric Studies
CHAPTER 6
6. IDEALISED TIME-TEMPERATURE PROFILES AND PARAMETRIC

STUDIES OF LSF WALLS
6.1. Introduction
Chapter 3 presented the experimental results of LSF wall panels under standard fire
conditions. The measured time-temperature profiles of the LSF wall studs obtained
from these full scale fire tests were then used in the development of finite element
models and fire design rules described in Chapters 4 and 5, respectively. In this
chapter idealised time-temperature profiles are proposed based on ten full scale fire
tests conducted in this research and Kolarkar (2010). These idealised time-
temperature profiles presented in this chapter were then used in further finite element
parametric studies of LSF wall studs under fire conditions.
Chapter 4 described the development of finite element models of LSF wall studs and
their validation. In this chapter the validated finite element models were used to
conduct an extensive parametric study to investigate the behaviour of LSF wall studs.
This parametric study included the effects of various parameters such as steel grade,
steel thickness, screw spacing, plasterboard restraint, various insulation materials and
load ratios.
Chapter 5 described the development of fire design rules based on AS/NZS 4600
(SA, 2005) and Eurocode 3 Part 1.3 (ECS, 2006) for the LSF wall studs made of 1.15
mm G500 steel. In this chapter the nonlinear analysis results of LSF wall studs from
the parametric study are used to compare and validate the above mentioned fire
design rules for varying steel grades and thicknesses. Finally, simple design rules are
proposed based on the parametric study to predict the failure hot flange temperatures
and failure times of LSF wall panels with varying wall configurations (single layer of
plasterboard, double layers of plasterboards, cavity insulted and externally insulated)
and structural parameters (steel grade and thickness) under different load ratios.
6-1
Parametric Studies
6.2. Idealised Time - Temperature Profiles
Table 6.1 shows the LSF wall test configurations and insulations used in this
research and Kolarkar (2010). The new externally insulated wall panels with glass
and rock fibres were tested under two load ratios of 0.2 and 0.4. Hence these ten full
scale fire tests were essentially conducted using eight different wall configurations
(see Table 6.1). Therefore the idealised time-temperature profiles were developed for
these eight different configurations using the measured hot flange and cold flange
temperature distributions along the stud. In the development of idealised time-
temperature profiles of studs with externally insulated glass and rock fibres, the
average temperature values of the two full scale fire tests were used (load ratios of
0.2 and 0.4 – Tests 1 & 2; Tests 3 & 6*). The critical stud in a LSF wall panel was
the stud, which had the vertical plasterboard joint against it. The temperature values
of this stud were high compared to other studs due to the opening of this vertical
joint at higher temperatures. Therefore the average temperatures along this stud were
considered in the development of all the idealised time-temperature profiles.
Table 6.1: LSF Wall Test Configurations with Different Insulation Materials
Failure Time
Test Configuration Insulation Location Load Ratio
(min.)
1 Glass Fibre External 0.2 118
2 Glass Fibre External 0.4 108
3 Rock Fibre External 0.4 134
1* None - 0.2 53
2* None - 0.2 111
3* Glass Fibre Cavity 0.2 101
4* Rock Fibre Cavity 0.2 107

Cellulose
5* Cavity 0.2 110
Fibre
6* Rock Fibre External 0.2 136#
Cellulose
7* External 0.2 124
Fibre
( * ) - Tests conducted by Kolarkar (2010) ( # ) - Experimental error
6-2
Parametric Studies
The experimental error in Test 6* is due to the lack of expansion space for the stud at
elevated temperatures. However, the temperature distribution of the stud is still
useful although the failure time was reduced. When the LSF wall was subject to
standard fire conditions, the hot and cold flange temperatures of the steel stud were
20oC for the initial few minutes. They then increased gradually to reach 100oC and
remained at the same temperatures during the plasterboard dehydration process.
After this the steel temperatures increased rapidly with time. Table 6.2 shows the
time-temperature values of hot and cold flanges up to 100oC. A linear variation of
temperature distribution was assumed between these times. Beyond 100oC,
Equations 6.1 to 6.8 represent the idealised time-temperature profiles for the LSF
wall panels with eight different configurations shown in Table 6.2, where T is the
average temperature in oC and t is the time measured in minutes.
Table 6.2: Idealised Time - Temperature Values up to 100oC
Time HFT Time CFT

W/C Index Configuration Insulation
(min.) (oC) (min.) (oC)
2 20 2 20
1 1x1 None 6 100 11 100
14 100 21 100
2 20 2 20
2 2x2 None 25 100 35 100
41 100 54 100
6 20 6 20
Glass
3 CI-GF 20 100 50 100
Fibre
52 100 65 100
6 20 6 20
Rock
4 CI-RF 25 100 50 100
Fibre
52 100 65 100
6 20 6 20
Cellulose
5 CI-CF 25 100 50 100
Fibre
52 100 65 100
6 20 6 20
Glass
6 CP-GF 40 100 55 100
Fibre
42 100 60 100
6 20 6 20
Rock
7 CP-RF 55 100 60 100
Fibre
70 100 80 100
6 20 6 20
Cellulose
8 CP-CF 45 100 60 100
Fibre
70 100 80 100
W/C - Wall Configuration; HFT - Hot Flange Temperature; CFT - Cold Flange Temperature
6-3
Parametric Studies
Figures 6.1 to 6.8 show the measured and idealised time-temperature profiles of hot
and cold flanges. They were based on the equations proposed in this study. Equation
6.1 (a) was used to represent the hot flange temperature values in Wall Configuration
1. The limit given as 15 ≤ t indicates that after 15 minutes the hot flange temperature
increased rapidly from 100oC. Similarly, the cold flange temperature increased
rapidly after 22 minutes according to Equation 6.1 (b). After 50 minutes the fire side
plasterboard has fallen off and hence the cold flange was also exposed to severe fire
due to the heat transmitted into the cavity. Therefore the cold flange temperature
increased rapidly and the temperature difference across the stud was reduced (see
Figure 6.1). It is unsafe to neglect this phase in the fire design of LSF wall panels and
hence this phase is included using Equation 6.1 (c). A similar argument was used to
develop the idealised time-temperature profiles of all other wall configurations.
Therefore additional equations were proposed for the final phases of Wall
Configurations 2, 3 and 5 (see Figures 6.2, 6.3 and 6.5). In Wall Configuration 1, the
cold flange temperature is expected to reach the hot flange temperature after a few
minutes from the plasterboard fallen off time. It is possible in LSF wall panels where
both sides are lined by a single layer of plasterboard. Hence an upper limit is
proposed for Equation 6.1 (c) as 60 minutes. After 60 minutes it is expected that the
cold flange temperatures will be the same as hot flange temperatures.
1) LSF wall lined on both sides by a single layer of plasterboard (Test 1*).
THF = -0.1066 t2 + 20.17 t - 165 (15 ≤ t) (6.1a)
TCF = 10.29 t - 125 (22 ≤ t ≤ 50) (6.1b)
TCF = 29.35 t - 1090 (50 < t ≤ 60) (6.1c)
TCF = THF (60 < t) (6.1d)
2) LSF wall lined on both sides by two layers of plasterboard (Test 2*).
THF = 6.35 t - 160 (42 ≤ t ≤ 110) (6.2a)
THF = 12.11 t - 790 (110 < t) (6.2b)
TCF = 6.07 t - 230 (55 ≤ t) (6.2c)
3) LSF wall lined on both sides by two layers of plasterboard with glass fibre used as
cavity insulation (Test 3*).
THF = 11.17 t - 490 (53 ≤ t) (6.3a)
TCF = 4.92 t - 225 (66 ≤ t ≤ 96) (6.3b)
TCF = 12.04 t - 915 (96 < t) (6.3c)
6-4
Parametric Studies
4) LSF wall lined on both sides by two layers of plasterboard with rock fibre used as
cavity insulation (Test 4*).
THF = 10.2 t - 435 (53 ≤ t) (6.4a)
TCF = 4.06 t - 165 (66 ≤ t) (6.4b)
5) LSF wall lined on both sides by two layers of plasterboard with cellulose fibre
used as cavity insulation (Test 5*).
THF = 8.94 t - 360 (53 ≤ t ≤ 106) (6.5a)
THF = 19.83 t - 1530 (106 < t) (6.5b)
TCF = 3.83 t - 150 (66 ≤ t ≤ 106) (6.5c)
TCF = 17 t - 1550 (106 < t) (6.5d)
6) LSF wall lined on both sides by two layers of plasterboard with glass fibre used as
external insulation (Tests 1 & 2).
THF = 0.001007 t3 - 0.1605 t2 + 12.15 t - 205 (43 ≤ t) (6.6a)
TCF = 0.0904 t2 – 9.56 t + 350 (61 ≤ t) (6.6b)
7) LSF wall lined on both sides by two layers of plasterboard with rock fibre used as
external insulation (Test 3 & 6*).
THF = - 0.000212 t3 + 0.0931 t2 – 5.47 t + 100 (71 ≤ t) (6.7a)
TCF = 0.0586 t2 – 6.69 t + 260 (81 ≤ t) (6.7b)
8) LSF wall lined on both sides by two layers of plasterboard with cellulose fibre
used as external insulation (Test 7*).
THF = -0.000286 t3 + 0.1024 t2 – 2.92 t - 100 (71 ≤ t) (6.8a)
TCF = 0.0846 t2 – 9.5 t + 320 (81 ≤ t) (6.8b)
Figures 6.1 to 6.8 show a close agreement between the measured and idealised time-
temperature profiles. Hence the proposed equations were used in the current
parametric study. It is also observed that the hot flange temperatures of cavity
insulted wall panels (see Figures 6.3 to 6.5) increased rapidly compared to externally
insulated wall panels (see Figures 6.6 to 6.8). In addition to this the difference
between the hot and cold flange temperatures also increased rapidly in cavity
insulated wall panels compared to externally insulated wall panels. Therefore it is
expected that the cavity insulated wall panels will be subjected to larger bending
moment effects and earlier failure times compared to externally insulated wall
panels. This will be further investigated in this chapter.
6-5
Parametric Studies
800
700
600
Temperature ( o C)
500
400
300
200
100
0
0 10 20 30 40 50 60 70
Time (min.)
Hot Flange (Proposed) Cold Flange (Proposed)

Hot Flange (Exp.) Cold Flange (Exp.)
Figure 6.1: Idealised Time - Temperature Profiles for LSF Walls

Lined on Both Sides by a Single Layer of Plasterboard
800
700
600
Temperature ( o C)
500
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130
Time (min.)


Lined on Both Sides by Two Layers of Plasterboard
6-6
Parametric Studies
800
700
600
Temperature ( o C)
500
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)


with Glass Fibre Used as Cavity Insulation
800
700
600
Temperature ( o C)
500
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)


with Rock Fibre Used as Cavity Insulation
6-7
Parametric Studies
800
700
Temperature ( C)
600
o
500
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110 120
Time (min.)


with Cellulose Fibre Used as Cavity Insulation
800
700
Temperature ( C)
600
o
500
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130
Time (min.)


with Glass Fibre Used as External Insulation
6-8
Parametric Studies
800
700
Temperature ( C)
600
o
500
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160
Time (min.)


with Rock Fibre Used as External Insulation
800
700
Temperature ( C)
600
o
500
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Time (min.)


with Cellulose Fibre Used as External Insulation
6-9
Parametric Studies
6.3. Details of Finite Element Models Used in the Parametric Study
A simply supported steel stud subject to axial compression was used in the
parametric study of cold-formed steel lipped channel sections subject to non-uniform
temperature distributions. The finite element models used in the parametric study are
similar to those used in Chapter 4 in terms of element type, element size, load
application, boundary conditions, the amplitude of initial geometric imperfection,
etc. The nonlinear finite element analyses of the stud were conducted under steady
state conditions. In the first step, temperatures distribution was applied with Riks off
method while in the second step the load was applied with Riks on method.
The yield strength and elastic modulus of steel decrease with increasing temperature
and therefore appropriately reduced mechanical properties were used to account for
the elevated temperature conditions. The reductions in yield strength and elastic
modulus at elevated temperatures were obtained using the equations developed by
Dolamune Kankanamge and Mahendran (2011). The nominal values were taken as
the yield strength at ambient temperature (for G500 grade steel fy is 500 MPa and for
G250 grade steel fy is 250 MPa). The value for elastic modulus was taken as 200 000
MPa for both high and low grade steels. The following equations represent the
reduction factors for yield strength and elastic modulus as developed by Dolamune
Kankanamge and Mahendran (2011).
Reduction Factors for Yield Strength
For G500 steels,

f y ,T  (T − 20 )4.56 
20 ≤ T < 300 o C = 1 −  (6.9a)
f y , 20  1x1010 T 
300 ≤ T < 600 o C

f y ,T 
= 0.95 −
(T − 300)1.45 
 (6.9b)
f y , 20  7.76T 
f y ,T
600 ≤ T ≤ 800 o C = −0.0004T + 0.35 (6.9c)
f y , 20
For G250 steels,

f y ,T
20 ≤ T ≤ 200 o C = −0.0005T + 1.01 (6.10a)
f y , 20
200 < T ≤ 800 o C

f y ,T
f y , 20
(
= 25 1.16 − T 0.022 ) (6.10b)
6-10
Parametric Studies
Reduction Factors for Elastic Modulus
For G500 and G250 steels,

ET
20 ≤ T ≤ 200 o C = −0.000835T + 1.0167 (6.11a)
E20
ET
200 < T ≤ 800 o C = −0.00135T + 1.1201 (6.11b)
E 20
The stress-strain relationship was established using Ranawaka and Mahendran’s
(2009a) predictive equations and appropriate modifications were made by Dolamune
Kankanamge and Mahendran (2011). The strain hardening material model was used
for steels with gradual yielding type stress-strain curve except for G250 steels at
100oC and 200oC, which have a stress-strain relationship with a well defined yield
point. For G250 steels at 100oC and 200oC the elastic-perfect plastic material model
was used. The following equations represent the stress-strain material model
proposed by Dolamune Kankanamge and Mahendran (2011).
Equation for Stress - Strain Curve

ηT
f  f y ,T  f T 
ε T = T + β    (6.12)
ET  
 T  f y ,T
E 
For G500 steels, 20 ≤ T ≤ 800 o C
β = 0.86
η T = −3.05 x10 −7 T 3 + 0.0005T 2 − 0.2615T + 62.653 (6.13a)
For G250 steels, 300 ≤ T ≤ 800o C

β = 1 .5
η T = 0.000138T 2 − 0.085468T + 19.212 (6.13b)
As used in Chapter 4, an amplitude of 0.006b was used for local buckling

imperfection where b is the web height. Residual stresses were not included in the
non-linear analysis since they rapidly decrease with increasing temperatures.
Thermal Expansion
The equations given in Eurocode 3 Part 1.2 (ECS, 2005) were used to determine the
thermal expansion of steel studs as was done in Chapter 4.
α = 1.2x10-5 + 0.4x10-8T – 2.416x10-4/T for 20ºC ≤ T < 750ºC (6.14a)
-2
α = 1.1x10 /T for 750ºC ≤ T ≤ 800ºC (6.14b)
6-11
Parametric Studies
Table 6.3 shows the parameters considered in the investigation of LSF wall panels.
In Chapters 4 and 5, the measured time-temperature profiles were used as the hot and
cold flange temperatures. The stud section of 90x40x15 was used in this study. The
steel stud with a thickness of 1.15 mm was analysed with the measured mechanical
properties of yield stress (569 MPa) and elastic modulus (213520 MPa). These
values were obtained from the experimental study conducted at QUT by Kolarkar
(2010). The screw spacing was used as 300 mm and it was assumed that the
plasterboard restraint was effective throughout the test.
In the current chapter, five different cases were selected to investigate the LSF wall
stud with varying screw spacing, plasterboard restraint, steel grade and steel
thickness (see Table 6.3). In all the cases the idealised time-temperature profiles
were used. In Case 1, the LSF wall stud was analysed with yield stress and elastic
modulus values of 500 MPa and 200000 MPa, respectively, at ambient temperature.
This set of results was obtained for comparison purposes with other cases. In Case 2,
the screw spacing was increased from 300 mm to 600 mm and 1200 mm. All other
parameters remained the same as for Case 1. This set of results was obtained to
investigate the behaviour of LSF wall studs with increased screw spacing (Cases 1
and 2). The hot side plasterboard restraint was removed for Case 3 and only the cold
side plasterboard restraint was considered. All other parameters were similar to Case
1. This set of results was obtained to investigate the effect of plasterboard fall off in
LSF walls under fire conditions (Cases 1 and 3).
In Case 4, G250 steel was used and the yield stress was assumed to be 250 MPa at
ambient temperature. All other parameters were similar to Case 1. This case was
used to investigate the behaviour of LSF wall studs with different steel grades (Cases
1 and 4). The steel thickness was increased to 1.95 mm for Case 5. All other
parameters were similar to Case 4. This set of results was used to investigate the
behaviour of LSF wall studs with varying steel thickness (Cases 4 and 5). Table 6.4
shows the overview of Sections 6.5 to 6.8 in this chapter.
6-12
Parametric Studies
Table 6.3: Parameters Considered in the Investigation of LSF Walls
Chapter 6
Parameters Chapter 5
Case 1 Case 2 Case 3 Case 4 Case 5
Yield
Stress 569 500 500 500 250 250
(MPa)
Elastic
Modulus 213520 200000 200000 200000 200000 200000
(MPa)
Thickness
1.15 1.15 1.15 1.15 1.15 1.95
(mm)
Screw
600 and
Spacing 300 300 300 300 300
1200
(mm)
Cold
Plasterboard Both Both Both Both Both
Flange
Restraint Flanges Flanges Flanges Flanges Flanges
Only
Time-
Temperature Measured Idealised Idealised Idealised Idealised Idealised
Profiles
Table 6.4: Overview of Sections 6.5 to 6.8
Section Parameters Considered Cases

6.5 Screw Spacing 1 and 2
6.6 Plasterboard Restraint 1 and 3
6.7 Steel Grade 1 and 4
6.8 Steel Thickness 4 and 5
6.4. Elastic Buckling Analyses at Ambient Temperature
Elastic buckling analyses of LSF wall studs at ambient temperature were conducted
on LSF wall studs to investigate their elastic buckling behaviour. An elastic buckling
load of 37.3 kN was obtained with a local buckling mode for Cases 1 to 4 as shown
in Figure 6.9. On the other hand an elastic buckling load of 159.5 kN was obtained
with flexural buckling mode about the major axis for Case 5 as shown in Figure 6.10.
In this case the local buckling is not the governing buckling mode because of the
increased thickness of 1.95 mm.
6-13
Parametric Studies
Figure 6.9: Local Buckling Mode for Cases 1 to 4
Figure 6.10: Flexural Buckling Mode about the Major Axis for Case 5
6-14
Parametric Studies
6.5. Influence of Screw Spacing
In this section the influence of screw spacing on the fire resistance of LSF wall studs
was investigated for 1.15 mm G500 steel studs. Three different screw spacings were
used (300 mm, 600 mm and 1200 mm). The elastic buckling modes for 300, 600 and
1200 mm screw spacings remained the same as shown in Figure 6.9. Figure 6.11 (a)
shows the ultimate failure mode of the LSF wall stud with 300 mm screw spacing. A
similar ultimate failure mode also was observed in the cases of 600 mm and 1200
mm screw spacings. However, the failure location changed when the screw spacing
was increased to 1200 mm. A slight bending about the minor axis was also observed
in this case as shown in Figure 6.11 (b).
Table 6.5 and Figure 6.12 show the variation of ultimate loads with time obtained
from FEA for 1.15 mm G500 steel studs lined on both sides by two layers of
plasterboard with glass fibre used as external insulation with different screw
spacings. The load carrying capacity did not reduce much when the screw spacing
was increased from 300 mm to 600 mm. However, the ultimate load reduced
considerably when the screw spacing was increased to 1200 mm. Tables 6.6 to 6.13
present the load ratios for the LSF wall studs with screw spacings of 300 mm and
600 mm for the chosen eight different wall configurations. The hot and cold flange
temperatures were obtained from the developed idealised time-temperature profiles
(see Section 6.2). It should be noted that in Table 6.6, the hot and cold flange
temperatures were the same for 60 and 70 minutes (see Figure 6.1). There is not
much difference in the load ratio when the screw spacing was increased even though
the ultimate load was reduced. This behaviour of LSF wall stud under fire conditions
can be clearly observed in Figures 6.13 to 6.20.
Figures 6.13 to 6.20 indicate that there is not much difference in the failure time
when a different screw spacing is used. Figure 6.18 includes the screw spacing of
1200 mm in addition to 300 mm and 600 mm. This figure also shows that the failure
time does not change when the screw spacing is varied. However, these FEA values
were obtained by assuming that the plasterboard integrity was not lost at elevated
temperatures. Hence the lateral support provided by the plasterboard was considered
to be always effective. However, practically it is not possible to maintain the
6-15
Parametric Studies
integrity of the plasterboard when larger screw spacings are used. The plasterboards
may fall off much earlier when a screw spacing of 1200 mm is used compared to 300
mm. Therefore the failure time will also be considerably less for 1200 mm screw
spacing compared to 300 mm screw spacing unlike what was predicted by FEA.
Local buckling
Thermal bowing
(a) Screw spacing of 300 mm
Local buckling
Bending about
the minor axis
(b) Screw spacing of 1200 mm
Figure 6.11: Ultimate Failure Modes Obtained for LSF Wall Studs
with Different Screw Spacings
6-16
Parametric Studies
Table 6.5: Ultimate Loads Obtained from FEA for 1.15 mm G500 Steel Studs
Ultimate Load (kN)

Hot Flange Cold Flange
Time Screw Screw Screw
Temperature Temperature
(min.) Spacing Spacing Spacing
(oC) (oC)
300 mm 600 mm 1200 mm
0 20 20 70.80 68.70 60.90
42 100 100 70.20 66.50 58.10
50 127 100 62.70 62.30 54.30
60 164 102 55.20 54.80 48.80
70 204 124 51.30 50.60 45.30
80 255 164 48.30 47.40 42.60
90 323 222 44.10 42.70 38.60
100 412 298 37.60 36.90 32.50
110 530 392 27.30 26.30 23.90
120 682 505 8.46 8.55 8.20
125 773 568 4.60 4.46 4.37
80.00
70.00
60.00
Ultimate Load (kN)
50.00
40.00
30.00
20.00
10.00
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130
Tim e (m in.)
300 mm 600 mm 1200 mm
Figure 6.12: Ultimate Load Variation with Time for 1.15 mm G500 Steel Studs
6-17
Parametric Studies
Table 6.6: FEA Results for 1.15 mm G500 Steel Studs

Lined on Both Sides by a Single Layer of Plasterboard
Screw Spacing Screw Spacing

Time 300 mm 600 mm
(min.) Ultimate Load Ultimate Load
(oC) (oC)
Load (kN) Ratio Load (kN) Ratio
0 20 20 70.80 1.00 68.70 1.00
14 100 100 70.20 0.99 66.50 0.97
20 196 100 50.00 0.71 49.30 0.72
30 344 184 38.80 0.55 37.80 0.55
40 471 287 31.40 0.44 30.30 0.44
50 577 390 18.80 0.27 18.70 0.27
60 661 661 7.19 0.10 7.19 0.10
70 725 725 4.93 0.07 4.92 0.07
Table 6.7: FEA Results for 1.15 mm G500 Steel Studs


Time 300 mm 600 mm
(oC) (oC)
0 20 20 70.80 1.00 68.70 1.00
40 100 100 70.20 0.99 66.50 0.97
50 158 100 56.20 0.79 55.60 0.81
60 221 134 49.80 0.70 49.00 0.71
70 285 195 47.10 0.67 46.10 0.67
80 348 256 43.10 0.61 41.90 0.61
90 412 316 38.30 0.54 37.20 0.54
100 475 377 32.40 0.46 31.70 0.46
110 539 438 25.50 0.36 25.00 0.36
120 663 498 9.34 0.13 9.34 0.14
130 784 559 4.21 0.06 4.19 0.06
Two Layers of Plasterboard with Glass Fibre Used as Cavity Insulation

Time 300 mm 600 mm
(oC) (oC)
0 20 20 70.80 1.00 68.70 1.00
52 100 100 70.20 0.99 66.50 0.97
60 180 100 52.40 0.74 51.40 0.75
70 292 119 40.70 0.57 39.20 0.57
80 404 169 34.00 0.48 32.70 0.48
90 515 218 26.20 0.37 25.60 0.37
100 627 289 13.30 0.19 13.30 0.19
110 739 409 7.04 0.10 7.04 0.10
6-18
Parametric Studies
Two Layers of Plasterboard with Rock Fibre Used as Cavity Insulation

Time 300 mm 600 mm
(oC) (oC)
0 20 20 70.80 1.00 68.70 1.00
52 100 100 70.20 0.99 66.50 0.97
60 177 100 52.80 0.75 51.90 0.76
70 279 119 41.70 0.59 40.60 0.59
80 381 160 35.50 0.50 34.40 0.50
90 483 200 28.40 0.40 27.50 0.40
100 585 241 19.00 0.27 18.80 0.27
110 687 281.6 10.50 0.15 10.50 0.15
120 789 322.2 5.44 0.08 5.48 0.08
Two Layers of Plasterboard with Cellulose Fibre Used as Cavity Insulation

Time 300 mm 600 mm
(oC) (oC)
0 20 20 70.80 1.00 68.70 1.00
52 100 100 70.20 0.99 66.50 0.97
60 176 100 53.00 0.75 52.40 0.76
70 266 118 43.20 0.61 42.00 0.61
80 355 156 36.70 0.52 35.50 0.52
90 445 195 31.40 0.44 30.30 0.44
100 534 233 24.90 0.35 24.50 0.36
110 651 320 11.80 0.17 11.70 0.17
115 750 405 6.63 0.09 6.63 0.10
Two Layers of Plasterboard with Glass Fibre Used as External Insulation

Time 300 mm 600 mm
(oC) (oC)
0 20 20 70.80 1.00 68.70 1.00
42 100 100 70.20 0.99 66.50 0.91
50 127 100 62.70 0.89 62.30 0.80
60 164 102 55.20 0.78 54.80 0.74
70 204 124 51.30 0.72 50.60 0.69
80 255 164 48.30 0.68 47.40 0.62
90 323 222 44.10 0.62 42.70 0.54
100 412 298 37.60 0.53 36.90 0.38
110 530 392 27.30 0.39 26.30 0.12
120 682 505 8.46 0.12 8.55 0.12
125 773 568 4.60 0.06 4.46 0.06
6-19
Parametric Studies
Two Layers of Plasterboard with Rock Fibre Used as External Insulation

Time 300 mm 600 mm
(oC) (oC)
0 20 20 70.80 1.00 68.70 1.00
70 100 100 70.20 0.99 66.50 0.97
80 150 100 57.60 0.81 57.20 0.83
90 207 133 51.90 0.73 51.30 0.75
100 272 177 47.10 0.67 46.30 0.67
110 343 233 42.30 0.60 40.90 0.60
120 418 301 37.30 0.53 36.40 0.53
130 497 381 30.40 0.43 29.70 0.43
140 577 472 17.50 0.25 17.40 0.25
150 659 575 8.70 0.12 8.53 0.12
160 740 690 5.09 0.07 5.06 0.07
Two Layers of Plasterboard with Cellulose Fibre Used as External Insulation

Time 300 mm 600 mm
(oC) (oC)
0 20 20 70.80 1.00 68.70 1.00
70 100 100 70.20 0.99 66.50 0.97
80 175 101 53.20 0.75 52.60 0.77
90 258 150 46.60 0.66 45.20 0.66
100 346 216 41.00 0.58 39.80 0.58
110 437 299 35.20 0.50 34.10 0.50
120 526 398 27.70 0.39 26.90 0.39
130 623 515 10.80 0.15 10.70 0.16
140 713 648 6.18 0.09 6.21 0.09
6-20
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70
Tim e (m in.)
300 mm 600 mm
Figure 6.13: FEA Results for 1.15 mm G500 Steel Studs Lined on Both Sides by
a Single Layer of Plasterboard with Different Screw Spacing
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Tim e (m in.)
300 mm 600 mm
Two Layers of Plasterboard with Different Screw Spacing
6-21
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110
Tim e (m in.)
300 mm 600 mm
Two Layers of Plasterboard with Glass Fibre Used as Cavity Insulation
with Different Screw Spacing
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
300 mm 600 mm
Two Layers of Plasterboard with Rock Fibre Used as Cavity Insulation
6-22
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
300 mm 600 mm
Two Layers of Plasterboard with Cellulose Fibre Used as Cavity Insulation
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130
Tim e (m in.)
300 mm 600 mm 1200 mm
Two Layers of Plasterboard with Glass Fibre Used as External Insulation
6-23
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160
Tim e (m in.)
300 mm 600 mm
Two Layers of Plasterboard with Rock Fibre Used as External Insulation
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Tim e (m in.)
300 mm 600 mm
Two Layers of Plasterboard with Cellulose Fibre Used as External Insulation
6-24
Parametric Studies
6.6. Validity of Plasterboard Restraint to Hot Flanges under Fire Conditions
The plasterboards attached to the LSF wall studs with screws provide lateral restraint
to the studs when it is subject to an axial compression load. However, with
increasing temperatures the plasterboards calcinate and lose their strength. Many
cracks will be developed in these calcinated plasterboards and eventually the fire side
plasterboards are likely to fall off under fire conditions. Therefore the lateral restraint
to the stud may not be available on the hot flange, and hence the axial compression
capacity of LSF wall stud will reduce considerably. The stud temperature at which
the fire side plasterboards fall off was found to vary in the previous studies on LSF
wall studs (Gerlich et al., 1996, Sultan, 1996, Buchanan and Gerlich, 1997 and
Kaitila, 2002).
According to Gerlich et al. (1996), the fire exposed side plasterboard will reduce its
ability to prevent buckling of the studs due to degradation when steel temperatures
reach critical levels (> 300-400oC). In the current study, these temperatures
recommended by Gerlich et al. (1996) are believed to be the average stud
temperatures. Gerlich et al. (1996) also recommended a 3 mm thickness of
undamaged gypsum to be retained to provide lateral restraint based on the
comparison of results from numerical modelling and tests. However, it is not
practical to measure this thickness in a full scale fire test. Sultan (1996) reported that
plasterboard fall-off occurs when the unexposed face of the board reaches about
600°C (Buchanan and Gerlich, 1997). In the current study it is believed that this
temperature is equal to the hot flange temperature of the stud. Kaitila (2002)
investigated the torsional-flexural buckling mode when the temperature during the
fire test was above the level of calcination of the plasterboards. He recommends a
value of 550oC as an approximate limit. In the study of Klippstein (1978) it was
assumed that the failure by weak axis flexural buckling or torsional buckling will be
prevented by the gypsum board cladding on the internal and external faces of the
wall. Alfawakhiri (2001) assumed that the flexural-torsional and weak axis buckling
failure modes are prevented by bridging and blocking. In the studies of Feng et al.
(2003b) and Zhao et al. (2005) the fire side plasterboards were assumed to provide
lateral restraint to the studs throughout the test. Hence the question of lateral restraint
6-25
Parametric Studies
provided by plasterboards was not raised in the studies of Klippstein (1978),

Alfawakhiri (2001), Feng et al. (2003b) and Zhao et al. (2005).
In the current experimental study the lateral or torsional buckling failure modes were
not observed in the studs of all three test specimens (Tests 1 to 3 in Table 6.1). This
may suggest that the plasterboards did not fully calcinate to lose its ability to provide
lateral restraint to the hot flange until failure. This is quite possible for the composite
panel where two plasterboards are used with insulation sandwiched between them. It
should also be noted that the hot flange temperature of the stud reached more than
600oC at the failure time for the load ratio of 0.2. Hence conservatively a hot flange
temperature value of 600oC is recommended in the current study as the limit beyond
which the plasterboard restraint is considered to be not effective.
Local buckling was observed as the ultimate failure mode of LSF wall stud when
both flanges are restrained by plasterboards (see Figure 6.21 (a)). Figure 6.21 (b)
shows the ultimate failure mode of LSF wall stud when only the cold flange is
restrained by plasterboards. In the latter case it was assumed that the fire side
plasterboards have fallen off and hence the lateral restraint to the hot flange is not
effective. Hence the flexural-torsional buckling failure mode was observed in this
case. However, the elastic buckling modes for these two different cases are the same
as shown in Figure 6.9.
Tables 6.14 to 6.21 show the FEA results of LSF wall studs with different conditions
of plasterboard restraints. The load ratio was calculated in the usual way when the
stud was restrained at both flanges. However, a different method was used when only
the cold flange was restrained. In this case the load ratio was equal to the case where
both flanges were restrained until the hot flange temperature was 600oC. When the
hot flange temperature was more than 600oC, the load ratio was calculated by finding
the ratio between the load at elevated temperature when only the cold flange was
restrained and the load at ambient temperature when both flanges were restrained.
Therefore the calculated load ratios show a sudden drop when the temperature
increases beyond 600oC, which represents the plasterboard fall off.
6-26
Parametric Studies
Figures 6.22 to 6.29 show the variation of load ratio with time when various hot
flange temperature limits (500oC, 550oC and 600oC) were incorporated. At higher
load ratios, the load ratio curves with time remained the same irrespective of the
temperature limits. However, the load ratio curves started to deviate when the load
ratio is about 0.2. This will be explained in detail later in this section.
Local buckling
Thermal bowing
(a) Both flanges are restrained
Flexural-torsional
buckling
(b) Only the cold flange is restrained
with Different Plasterboard Restraints
6-27
Parametric Studies
Table 6.14: FEA Results for 1.15 mm G500 Steel Studs Lined by a Single Layer
of Plasterboard at 300 mm Screw Spacing
Restrained Both Restrained only the

Time Flanges Cold Flange
(oC) (oC)
0 20 20 70.80 1.00 37.50 1.00
14 100 100 70.20 0.99 37.00 0.99
20 196 100 50.00 0.71 34.00 0.71
30 344 184 38.80 0.55 25.90 0.55
40 471 287 31.40 0.44 18.70 0.44
50 577 390 18.80 0.27 10.90 0.27
60 661 661 7.19 0.10 7.21 0.10
70 725 725 4.93 0.07 4.91 0.07
Plasterboard at 300 mm Screw Spacing

(oC) (oC)
0 20 20 70.80 1.00 37.50 1.00
40 100 100 70.20 0.99 37.00 0.99
50 158 100 56.20 0.79 37.20 0.79
60 221 134 49.80 0.70 33.50 0.70
70 285 195 47.10 0.67 30.70 0.67
80 348 256 43.10 0.61 27.40 0.61
90 412 316 38.30 0.54 23.60 0.54
100 475 377 32.40 0.46 19.60 0.46
110 539 438 25.50 0.36 14.80 0.36
120 663 498 9.34 0.13 6.15 0.09
130 784 559 4.21 0.06 2.89 0.04
Plasterboard at 300 mm Screw Spacing with Glass Fibre Used
as Cavity Insulation

(oC) (oC)
0 20 20 70.80 1.00 37.50 1.00
52 100 100 70.20 0.99 37.00 0.99
60 180 100 52.40 0.74 35.30 0.74
70 292 119 40.70 0.57 27.60 0.57
80 404 169 34.00 0.48 20.50 0.48
90 515 218 26.20 0.37 15.10 0.37
100 627 289 13.30 0.19 8.14 0.11
110 739 409 7.04 0.10 5.15 0.07
6-28
Parametric Studies
Plasterboard at 300 mm Screw Spacing with Rock Fibre Used
(oC) (oC)
0 20 20 70.80 1.00 37.50 1.00
52 100 100 70.20 0.99 37.00 0.99
60 177 100 52.80 0.75 35.50 0.75
70 279 119 41.70 0.59 28.40 0.59
80 381 160 35.50 0.50 21.60 0.50
90 483 200 28.40 0.40 17.20 0.40
100 585 241 19.00 0.27 10.00 0.27
110 687 281.6 10.50 0.15 7.10 0.10
120 789 322.2 5.44 0.08 4.69 0.07
Plasterboard at 300 mm Screw Spacing with Cellulose Fibre Used
(oC) (oC)
0 20 20 70.80 1.00 37.50 1.00
52 100 100 70.20 0.99 37.00 0.99
60 176 100 53.00 0.75 35.60 0.75
70 266 118 43.20 0.61 29.40 0.61
80 355 156 36.70 0.52 22.90 0.52
90 445 195 31.40 0.44 19.60 0.44
100 534 233 24.90 0.35 13.80 0.35
110 651 320 11.80 0.17 7.45 0.11
115 750 405 6.63 0.09 5.00 0.07
as External Insulation
(oC) (oC)
0 20 20 70.80 1.00 37.50 1.00
42 100 100 70.20 0.99 37.00 0.99
50 127 100 62.70 0.89 39.90 0.89
60 164 102 55.20 0.78 36.70 0.78
70 204 124 51.30 0.72 34.50 0.72
80 255 164 48.30 0.68 32.00 0.68
90 323 222 44.10 0.62 28.50 0.62
100 412 298 37.60 0.53 23.30 0.53
110 530 392 27.30 0.39 15.10 0.39
120 682 505 8.46 0.12 5.55 0.08
125 773 568 4.60 0.06 3.11 0.04
6-29
Parametric Studies

(oC) (oC)
0 20 20 70.80 1.00 37.50 1.00
70 100 100 70.20 0.99 37.00 0.99
80 150 100 57.60 0.81 37.80 0.81
90 207 133 51.90 0.73 34.60 0.73
100 272 177 47.10 0.67 31.10 0.67
110 343 233 42.30 0.60 27.20 0.60
120 418 301 37.30 0.53 22.90 0.53
130 497 381 30.40 0.43 17.80 0.43
140 577 472 17.50 0.25 11.30 0.25
150 659 575 8.70 0.12 6.44 0.09
160 740 690 5.09 0.07 4.09 0.06

(oC) (oC)
0 20 20 70.80 1.00 37.50 1.00
70 100 100 70.20 0.99 37.00 0.99
80 175 101 53.20 0.75 35.80 0.75
90 258 150 46.60 0.66 31.20 0.66
100 346 216 41.00 0.58 26.60 0.58
110 437 299 35.20 0.50 21.50 0.50
120 526 398 27.70 0.39 15.50 0.39
130 623 515 10.80 0.15 7.60 0.11
140 713 648 6.18 0.09 4.84 0.07
6-30
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70
Tim e (m in.)
Both flanges restrained Cold flange restrained; Hot flange > 600oC
Cold flange restrained; Hot flange > 550oC Cold flange restrained; Hot flange > 500oC
Figure 6.22: FEA Results for 1.15 mm G500 Steel Studs Lined by a Single Layer
of Plasterboard at 300 mm Screw Spacing
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Tim e (m in.)
Plasterboard at 300 mm Screw Spacing
6-31
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110
Tim e (m in.)
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
6-32
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130
Tim e (m in.)
Plasterboard at 300mm Screw Spacing with Glass Fibre Used
6-33
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160
Tim e (m in.)
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Tim e (m in.)
6-34
Parametric Studies
Table 6.22 shows the failure times of LSF wall panels when different hot flange
temperature limits were used. The failure time of LSF wall panel did not vary much
when the load ratio was more than 0.4. When the load ratio was about 0.2, again the
failure time did not change much when a hot flange temperature limit of 600oC was
used. A slight variation in failure time (≤ 6 mins.) was observed when the hot flange
temperature limit was used as 500oC. However, this variation is much less compared
to the complexity involved in the fire design of LSF walls and the variables within
LSF wall panel construction. Therefore it is concluded that the plasterboard restraint
can be assumed to be effective throughout the failure time in routine fire design
calculations.
Table 6.22: FEA Failure Times of LSF Wall Panels
with Different Plasterboard Restraints
Failure Time (min.)

Load CF Restrained;
Index Configuration Insulation Both
Ratio Flanges HF Temperature Limit
Restrained 600oC 550oC 500oC
0.7 20 20 20 20
1x1 None 0.4 42 42 42 42
0.2 54 54 48 48
0.7 61 61 61 61
2x2 None 0.4 106 106 106 102
0.2 117 116 116 111
0.7 62 62 62 62
Glass
CI-GF 0.4 87 87 87 83
Fibre
0.2 99 97 97 91
0.7 63 63 63 63
Rock
CI-RF 0.4 90 90 90 90
Fibre
0.2 106 104 98 98
0.7 64 64 64 64
Cellulose
CI-CF 0.4 95 95 95 92
Fibre
0.2 108 106 106 100
0.7 76 76 76 76
Glass
CP-GF 0.4 109 109 109 104
Fibre
0.2 117 116 116 111
0.7 95 95 95 95
Rock
CP-RF 0.4 132 132 131 131
Fibre
0.2 144 143 139 139
0.7 86 86 86 86
Cellulose
CP-CF 0.4 119 119 119 113
Fibre
0.2 128 127 127 122
Note: HF – Hot Flange; CF – Cold Flange
6-35
Parametric Studies
6.7. Influence of Steel Grade on the Fire Resistance of LSF Wall Studs
This section investigates the behaviour of LSF wall studs with different steel grades.
A high grade of G500 and a low grade of G250 were used in this study with yield
stresses of 500 MPa and 250 MPa, respectively, under ambient conditions. Local
buckling was observed as the elastic buckling mode for both cases as shown in
Figure 6.9. The ultimate failure modes of these studs were also local buckling
although the location changed from support to mid-height (see Figures 6.30 (a) and
(b)).
Tables 6.23 to 6.30 show the ultimate loads and the load ratios of LSF wall studs
with different steel grades. As expected the ultimate load was reduced considerably
when the low grade steel was used. However, the load ratio did not show such an
obvious pattern. Figures 6.31 to 6.38 show this variation of load ratios with time
when different steel grades are used. It was expected that the G500 load ratio curve
should stay above the G250 load ratio curve in these figures. However, a rapid
reduction was observed in G500 load ratio when the temperature started to increase
in the steel studs. During this time, the temperature difference across the stud
increased rapidly and hence thermal bowing was developed. The bending moment
generated by this thermal bowing was directly proportional to the load applied.
Therefore the G500 steel stud experienced a larger bending moment and the capacity
was reduced rapidly during this stage. Hence the load ratio curve stayed below the
G250 steel load ratio curve. This phase is longer for cavity insulated wall panels (see
Figures 6.33 to 6.35) and on the other hand shorter for externally insulated wall
panels (see Figures 6.36 to 6.38). This is due to the ever rising temperature difference
across the studs for the cavity insulated wall panels.
In the next phase the axial compressive capacity of the LSF wall stud was reduced
with time due to the effects of elevated temperatures. Therefore the above mentioned
rapid reduction was not possible this time for G500 steel with a small load. Further
the cold flange temperature was approaching the hot flange temperature and the
temperature difference across the stud was reduced, especially for externally
insulated wall panels. Hence G500 steel load ratio curve stayed above the G250 steel
load ratio curve during this phase.
6-36
Parametric Studies
In the final phase of all these figures, the load ratio curve of G500 steel is less than
the load ratio curve of G250. This is explained by the reduction factors
recommended by Dolamune Kankanamge and Mahendran (2011). Beyond about
540oC the reduction factor for G500 steel is less than the reduction factor for G250
steel according to their equations. This corresponds to this final phase of these load
ratio curves with different steel grades.
Local buckling
Thermal bowing
(a) Steel grade of G500
Local buckling
(b) Steel grade of G250
with Different Steel Grades
6-37
Parametric Studies
Table 6.23: FEA Results for 1.15 mm LSF Wall Studs Lined on Both Sides by a
Single Layer of Plasterboard at 300 mm Screw Spacing
G500 G250
Time Steel Steel
(oC) (oC)
0 20 20 70.80 1.00 43.60 1.00
14 100 100 70.20 0.99 40.30 0.92
20 196 100 50.00 0.71 34.10 0.78
30 344 184 38.80 0.55 26.50 0.61
40 471 287 31.40 0.44 16.30 0.37
50 577 390 18.80 0.27 11.80 0.27
60 661 661 7.19 0.10 6.51 0.15
70 725 725 4.93 0.07 4.19 0.10
Two Layers of Plasterboard at 300 mm Screw Spacing
G500 G250
Time Steel Steel
(oC) (oC)
0 20 20 70.80 1.00 43.60 1.00
40 100 100 70.20 0.99 40.30 0.92
50 158 100 56.20 0.79 37.70 0.86
60 221 134 49.80 0.70 33.80 0.78
70 285 195 47.10 0.67 31.40 0.72
80 348 256 43.10 0.61 24.10 0.55
90 412 316 38.30 0.54 18.30 0.42
100 475 377 32.40 0.46 15.60 0.36
110 539 438 25.50 0.36 13.00 0.30
120 663 498 12.20 0.17 8.61 0.20
130 784 559 8.63 0.12 3.35 0.08
Two Layers of Plasterboard at 300 mm Screw Spacing with Glass Fibre Used
G500 G250
Time Steel Steel
(oC) (oC)
0 20 20 70.80 1.00 43.60 1.00
52 100 100 70.20 0.99 40.30 0.92
60 180 100 52.40 0.74 35.50 0.81
70 292 119 40.70 0.57 27.30 0.63
80 404 169 34.00 0.48 21.90 0.50
90 515 218 26.20 0.37 16.00 0.37
100 627 289 13.30 0.19 10.50 0.24
110 739 409 7.04 0.10 6.07 0.14
6-38
Parametric Studies
Two Layers of Plasterboard at 300 mm Screw Spacing with Rock Fibre Used
G500 G250
Time Steel Steel
(oC) (oC)
0 20 20 70.80 1.00 43.60 1.00
52 100 100 70.20 0.99 40.30 0.92
60 177 100 52.80 0.75 35.80 0.82
70 279 119 41.70 0.59 28.20 0.65
80 381 160 35.50 0.50 23.10 0.53
90 483 200 28.40 0.40 18.00 0.41
100 585 241 19.00 0.27 12.70 0.29
110 687 281.6 10.50 0.15 8.34 0.19
120 789 322.2 5.44 0.08 4.28 0.10
Two Layers of Plasterboard at 300 mm Screw Spacing with Cellulose Fibre
Used as Cavity Insulation
G500 G250
Time Steel Steel
(oC) (oC)
0 20 20 70.80 1.00 43.60 1.00
52 100 100 70.20 0.99 40.30 0.92
60 176 100 53.00 0.75 35.80 0.82
70 266 118 43.20 0.61 29.20 0.67
80 355 156 36.70 0.52 24.60 0.56
90 445 195 31.40 0.44 19.90 0.46
100 534 233 24.90 0.35 14.90 0.34
110 651 320 11.80 0.17 9.56 0.22
115 750 405 6.63 0.09 5.63 0.13
G500 G250
Time Steel Steel
(oC) (oC)
0 20 20 70.80 1.00 43.60 1.00
42 100 100 70.20 0.99 40.30 0.92
50 127 100 62.70 0.89 40.10 0.92
60 164 102 55.20 0.78 37.10 0.85
70 204 124 51.30 0.72 34.80 0.80
80 255 164 48.30 0.68 32.80 0.75
90 323 222 44.10 0.62 27.70 0.64
100 412 298 37.60 0.53 18.60 0.43
110 530 392 27.30 0.39 13.30 0.31
120 682 505 8.46 0.12 7.78 0.18
125 773 568 4.6 0.06 3.76 0.09
6-39
Parametric Studies
G500 G250
Time Steel Steel
(oC) (oC)
0 20 20 70.80 1.00 43.60 1.00
70 100 100 70.20 0.99 40.30 0.92
80 150 100 57.60 0.81 38.50 0.88
90 207 133 51.90 0.73 35.00 0.80
100 272 177 47.10 0.67 32.00 0.73
110 343 233 42.30 0.60 25.70 0.59
120 418 301 37.30 0.53 18.20 0.42
130 497 381 30.40 0.43 14.70 0.34
140 577 472 17.50 0.25 11.60 0.27
150 659 575 8.70 0.12 8.16 0.19
160 740 690 5.09 0.07 4.38 0.10
Used as External Insulation
G500 G250
Time Steel Steel
(oC) (oC)
0 20 20 70.80 1.00 43.60 1.00
70 100 100 70.20 0.99 40.30 0.92
80 175 101 53.20 0.75 36.00 0.83
90 258 150 46.60 0.66 31.70 0.73
100 346 216 41.00 0.58 26.20 0.60
110 437 299 35.20 0.50 17.40 0.40
120 526 398 27.70 0.39 13.50 0.31
130 623 515 10.80 0.15 9.96 0.23
140 713 648 6.18 0.09 5.45 0.13
6-40
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70
Tim e (m in.)
G500 G250
Figure 6.31: FEA Results for 1.15 mm LSF Wall Studs Lined on Both Sides by a
Single Layer of Plasterboard at 300 mm Screw Spacing
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Tim e (m in.)
G500 G250
Two Layers of Plasterboard at 300 mm Screw Spacing
6-41
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110
Tim e (m in.)
G500 G250
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
G500 G250
6-42
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
G500 G250
Used as Cavity Insulation
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130
Tim e (m in.)
G500 G250
6-43
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160
Tim e (m in.)
G500 G250
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Tim e (m in.)
G500 G250
Used as External Insulation
6-44
Parametric Studies
6.8. Influence of Steel Thickness on the Fire Resistance of LSF Wall Studs
This section investigates the fire resistance of LSF wall studs with varying steel
thicknesses. The steel thickness was increased from 1.15 mm to 1.95 mm to
minimise any local buckling effects. The results obtained in Section 6.4 emphasize
this behaviour. Elastic buckling modes obtained were local buckling and flexural
buckling for steel studs with thicknesses of 1.15 mm and 1.95 mm, respectively (see
Figures 6.9 and 6.10).
Figures 6.39 (a) and (b) show the ultimate failure modes of 1.15 mm and 1.95 mm
G250 steel studs, respectively. Figure 6.39 (a) shows the LSF wall stud with local
buckling as the dominant failure mode. On the other hand Figure 6.39 (b) shows the
LSF wall stud with flexural buckling about the major axis as the dominant failure
mode. The flexural buckling failure was also confirmed with the rapid increase in
mid-height deflection.
Tables 6.31 to 6.38 show the axial compressive capacity and the load ratio of LSF
wall studs at elevated temperatures. As expected the axial compressive capacity of
1.95 mm LSF wall studs is much higher than that of 1.15 mm LSF wall studs.
However, the load ratio is reduced rapidly with time for 1.95 mm LSF wall studs
compared to 1.15 mm LSF wall studs. This is explained by the bending moment
developed by the higher load and thermal bowing deflection. The bending moment
generated by the magnified thermal bowing is directly proportional to the applied
load. Therefore 1.95 mm LSF wall studs experienced greater bending moment due to
the higher load and hence the capacity was reduced rapidly during this stage. The
variation of yield stress reduction factors did not influence the load ratio curves this
time since the same steel grade (G250) was used in this case. Therefore the load ratio
curve of 1.95 mm steel stud always stayed below the load ratio curve of 1.15 mm
steel stud unlike in the last case (see Section 6.7).
Figures 6.40 to 6.47 show the variation of load ratio with time with varying steel
thicknesses. The above mentioned behaviour can be seen explicitly in these figures.
Therefore it should be noted that the use of higher thickness leads to a reduced
failure time for a given load ratio compared to the use of thinner steel thickness. In
6-45
Parametric Studies
other words, the LSF wall studs with larger capacities will be subjected to higher
loads during a fire event and hence this will lead to reduced failure times.
Local buckling
Thermal bowing
(a) Steel thickness of 1.15 mm
Flexural buckling
(b) Steel thickness of 1.95 mm
with Different Thicknesses
6-46
Parametric Studies
Table 6.31: FEA Results for G250 Steel Studs Lined on Both Sides by a Single
Layer of Plasterboard at 300 mm Screw Spacing
Thickness Thickness
Time 1.15 mm 1.95 mm
(oC) (oC)
0 20 20 43.60 1.00 93.70 1.00
14 100 100 40.30 0.92 89.40 0.95
20 196 100 34.10 0.78 58.10 0.62
30 344 184 26.50 0.61 44.50 0.47
40 471 287 16.30 0.37 27.70 0.30
50 577 390 11.80 0.27 20.00 0.21
60 661 661 6.51 0.15 14.50 0.15
70 725 725 4.19 0.10 9.06 0.10
Layers of Plasterboard at 300 mm Screw Spacing
Thickness Thickness
(oC) (oC)
0 20 20 43.60 1.00 93.70 1.00
40 100 100 40.30 0.92 89.40 0.95
50 158 100 37.70 0.86 66.60 0.71
60 221 134 33.80 0.78 58.30 0.62
70 285 195 31.40 0.72 55.10 0.59
80 348 256 24.10 0.55 40.60 0.43
90 412 316 18.30 0.42 31.20 0.33
100 475 377 15.60 0.36 26.20 0.28
110 539 438 13.00 0.30 21.80 0.23
120 663 498 8.61 0.20 - -
130 784 559 3.35 0.08 - -
Layers of Plasterboard at 300 mm Screw Spacing with Glass Fibre Used
Thickness Thickness
(oC) (oC)
0 20 20 43.60 1.00 93.70 1.00
52 100 100 40.30 0.92 89.40 0.95
60 180 100 35.50 0.81 61.30 0.65
70 292 119 27.30 0.63 46.10 0.49
80 404 169 21.90 0.50 37.10 0.40
90 515 218 16.00 0.37 27.70 0.30
100 627 289 10.50 0.24 18.20 0.19
110 739 409 6.07 0.14 11.70 0.12
6-47
Parametric Studies
Layers of Plasterboard at 300 mm Screw Spacing with Rock Fibre Used
Thickness Thickness
(oC) (oC)
0 20 20 43.60 1.00 93.70 1.00
52 100 100 40.30 0.92 89.40 0.95
60 177 100 35.80 0.82 61.90 0.66
70 279 119 28.20 0.65 47.60 0.51
80 381 160 23.10 0.53 38.90 0.42
90 483 200 18.00 0.41 30.80 0.33
100 585 241 12.70 0.29 22.10 0.24
110 687 281.6 8.34 0.19 14.50 0.15
120 789 322.2 4.28 0.10 8.06 0.09
Layers of Plasterboard at 300 mm Screw Spacing with Cellulose Fibre Used
Thickness Thickness
(oC) (oC)
0 20 20 43.60 1.00 93.70 1.00
52 100 100 40.30 0.92 89.40 0.95
60 176 100 35.80 0.82 62.10 0.66
70 266 118 29.20 0.67 49.30 0.53
80 355 156 24.60 0.56 41.50 0.44
90 445 195 19.90 0.46 33.80 0.36
100 534 233 14.90 0.34 25.80 0.28
110 651 320 9.56 0.22 16.50 0.18
115 750 405 5.63 0.13 10.90 0.12
Layers of Plasterboard at 300 mm Screw Spacing with Glass Fibre Used
Thickness Thickness
(oC) (oC)
0 20 20 43.60 1.00 93.70 1.00
42 100 100 40.30 0.92 89.40 0.95
50 127 100 40.10 0.92 76.10 0.81
60 164 102 37.10 0.85 65.40 0.70
70 204 124 34.80 0.80 60.20 0.64
80 255 164 32.80 0.75 56.50 0.60
90 323 222 27.70 0.64 47.30 0.50
100 412 298 18.60 0.43 31.40 0.34
110 530 392 13.30 0.31 22.50 0.24
120 682 505 7.78 0.18 14.00 0.15
125 773 568 3.76 0.09 8.17 0.09
6-48
Parametric Studies
Layers of Plasterboard at 300 mm Screw Spacing with Rock Fibre Used
Thickness Thickness
(oC) (oC)
0 20 20 43.60 1.00 93.70 1.00
70 100 100 40.30 0.92 89.40 0.95
80 150 100 38.50 0.88 68.70 0.73
90 207 133 35.00 0.80 61.00 0.65
100 272 177 32.00 0.73 55.20 0.59
110 343 233 25.70 0.59 43.20 0.46
120 418 301 18.20 0.42 30.90 0.33
130 497 381 14.70 0.34 24.70 0.26
140 577 472 11.60 0.27 19.40 0.21
150 659 575 8.16 0.19 14.30 0.15
160 740 690 4.38 0.10 8.90 0.09
Layers of Plasterboard at 300 mm Screw Spacing with Cellulose Fibre Used
Thickness Thickness
(oC) (oC)
0 20 20 43.60 1.00 93.70 1.00
70 100 100 40.30 0.92 89.40 0.95
80 175 101 36.00 0.83 62.50 0.67
90 258 150 31.70 0.73 54.00 0.58
100 346 216 26.20 0.60 44.10 0.47
110 437 299 17.40 0.40 29.50 0.31
120 526 398 13.50 0.31 22.70 0.24
130 623 515 9.96 0.23 16.60 0.18
140 713 648 5.45 0.13 11.00 0.12
6-49
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70
Time (m in.)
t = 1. 15 mm t = 1. 95 mm
on Both Sides by a Single Layer of Plasterboard at 300 mm Screw Spacing
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Tim e (m in.)
t = 1. 15 mm t = 1. 95 mm
on Both Sides by Two Layers of Plasterboard at 300 mm Screw Spacing
6-50
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110
Tim e (m in.)
t = 1. 15 mm t = 1. 95 mm
on Both Sides by Two Layers of Plasterboard at 300 mm Screw Spacing with
Glass Fibre Used as Cavity Insulation
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
t = 1. 15 mm t = 1. 95 mm
with Rock Fibre Used as Cavity Insulation
6-51
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
t = 1. 15 mm t = 1. 95 mm
with Cellulose Fibre Used as Cavity Insulation
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130
Tim e (m in.)
t = 1. 15 mm t = 1. 95 mm
6-52
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160
Tim e (m in.)
t = 1. 15 mm t = 1. 95 mm
with Rock Fibre Used as External Insulation
1.00
0.90
0.80
0.70
Load Ratio
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Tim e (m in.)
t = 1. 15 mm t = 1. 95 mm
with Cellulose Fibre Used as External Insulation
6-53
Parametric Studies
6.9. Critical Hot Flange Temperature of LSF Wall Studs (Simplified Method)
Lawson (1993) adopted the limiting temperature method in BS 5950 Part 8 (BSI,
1990) to cold-formed thin-walled steel structures. The limiting temperature is a
function of the load ratio of the structural member. The load ratio is the ratio between
the load on the member at the fire limit state and the load carrying capacity of that
member under ambient conditions. Usually hot-rolled steel columns have a uniform
temperature distribution at failure due to fire attack from all four sides. However,
cold-formed thin-walled studs will be subjected to more severe non-uniform
temperature distributions. Hence the BS 5950 Part 8 (BSI, 1990) method is, strictly
speaking, not applicable to cold-formed thin-walled studs as Lawson (1993)
recommended (Feng et al., 2003b).
Kolarkar (2010) proposed simple design rules to find the failure times of LSF walls
by combining the yield stress reduction factors and idealised time-temperature
profiles. The load-bearing wall specimens tested in his study were subject to a load
of 15 kN per stud giving a load ratio of 0.2. The load ratio at the time of failure is
equal to the ratio of yield load at elevated temperature to the yield load at ambient
temperature, which is equivalent to the strength reduction factor (ratio of the yield
stress at elevated temperature to the yield stress at ambient temperature) considering
the cross-sectional area to remain constant during the test (assuming full yielding).
Hence the load ratio at failure was considered to be equivalent to a strength reduction
factor of 0.2. This gave a corresponding critical temperature of approximately 565oC
from the relationship between yield stress reduction factor and temperature. This
critical temperature corresponding to the load ratio was used with idealised time-
temperature profiles to obtain the approximate failure times of the hot flange of each
type of wall specimen.
A similar approach to Kolarkar (2010) was used in this section to determine the
failure times of LSF wall panels. However, instead of using the yield stress reduction
factors, the FEA results were used to find the critical temperature, ie. the maximum
hot flange temperature at failure. In the parametric study, finite element analyses
were conducted for five different cases as discussed in Section 6.3. Among these
FEA results, only three cases (Cases 1, 4 and 5) were considered in this section.
6-54
Parametric Studies
Similarly eight different wall configurations were considered in the parametric study
as discussed in Section 6.2. The load ratio curves with hot flange temperature at
failure agreed with each other for all three cavity insulated wall panels irrespective of
the insulations used. Similarly the load ratio curves agreed well for all externally
insulated wall panels irrespective of the insulation used. Therefore the eight wall
configurations used in the parametric study were reduced to four wall configurations
in this section. They are wall panels with single plasterboards, wall panels with
double plasterboards, cavity insulated wall panels with double plasterboards and
externally insulated wall panels with double plasterboards. Hence twelve different
sets (3 cases with 4 wall configurations) were considered in the current limiting
temperature method.
Figures 6.48 to 6.59 show the variation of load ratio with hot flange temperature at
failure for the 12 sets identified above. Figures 6.48 and 6.49 show the variation of
load ratios for LSF walls with single and double layers of plasterboard, respectively.
Figure 6.50 was plotted with FEA results for LSF wall panels with glass fibre, rock
fibre and cellulose fibre used as cavity insulation. It is interesting to note that all the
plots using different insulations merged together with the same trend. This clearly
indicates that the failure temperature of LSF walls do not depend on the type of
insulation. In other words the effect of using different types of insulation is simply to
delay the time to reach the same hot flange temperatures in the LSF wall studs.
Figure 6.51 was plotted with FEA results for LSF wall panels with glass fibre, rock
fibre and cellulose fibre used as external insulation. The close agreement of the plots
with different types of insulation in this figure also confirms the above finding.
Hence it was decided to plot the results of cavity and externally insulated wall panels
together. However, it was noted that the results of these two types of wall panels did
not merge together. This indicates that the arrangement of insulations and
plasterboards plays a major role in failure temperature although the type of insulation
does not affect it.
Figures 6.52 to 6.55 were obtained for Grade 250 steel with a thickness of 1.15 mm
and Figures 6.56 to 6.59 were obtained for Grade 250 steel with a thickness of 1.95
mm. They all resulted in same conclusions as discussed before.
6-55
Parametric Studies
The limiting temperature methods proposed by Kolarkar (2010), Lawson (1993) and
Eurocode 3 Part 1.2 (ECS, 2005) are also shown in these figures for comparison
purposes. According to Kolarkar (2010) the reduction in load ratio is based on the
material yield strength reduction factors with respect to temperature. The equations
proposed by Dolamune Kankanamge and Mahendran (2011) for high and low grade
steels were used in Figures 6.48 to 6.51 and 6.52 to 6.59, respectively. Kolarkar’s
(2010) method was found to be unsafe for 1.15 mm G500 steel studs when the load
ratio was more than 0.5. On the other hand it was found to be too conservative for
1.15 mm G500 steel studs when the load ratio was more than 0.4. Kolarkar’s (2010)
method agreed reasonably well with the current FEA results for 1.15 mm G250 steel
studs. However, the reason for this behaviour is unclear. On the other hand
Kolarkar’s (2010) method was unsafe for 1.95 mm G250 steel studs. These arbitrary
observations in relation to Kolarkar’s (2010) method indicates that this method
cannot be used in the fire design of LSF walls.
Lawson’s (1993) method was found to be unsafe for all the cases as shown in Figures
6.48 to 6.59. Eurocode 3 Part 1.2 (ECS, 2005) recommends a limiting temperature
value of 350oC irrespective of the load ratio which is shown by a vertical line in
these figures. This also did not agree with the FEA results. Therefore it was decided
to propose a new set of equations (Equations 6.15 to 6.26) to determine the critical
hot flange temperatures at failure for the above mentioned 12 sets (3 cases with 4
wall configurations). These equations were used with Equations 6.1 to 6.8 giving the
idealised time-temperature profiles to find the failure times of 24 sets (3 cases with 8
wall configurations).
6-56
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 100 200 300 400 500 600 700 800
o
Tem perature ( C)
FEA Kolarkar (2010) Lawson (1993) ECS Part 1.2 Poly. (FEA)
G500 Steel Studs Lined on Both Sides by a Single Layer of Plasterboard
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 100 200 300 400 500 600 700 800
o
Tem perature ( C)
G500 Steel Studs Lined on Both Sides by Two Layers of Plasterboard
6-57
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 100 200 300 400 500 600 700 800
Tem perature ( oC)
G500 Steel Studs Lined on Both Sides by Two Layers of Plasterboard with
Glass Fibre, Rock Fibre or Cellulose Fibre Used as Cavity Insulation
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 100 200 300 400 500 600 700 800
o
Tem perature ( C)
G500 Steel Studs on Both Sides by Two Layers of Plasterboard with Glass
Fibre, Rock Fibre or Cellulose Fibre Used as External Insulation
6-58
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 100 200 300 400 500 600 700 800
Tem perature ( oC)
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 100 200 300 400 500 600 700 800
Tem perature ( oC)
6-59
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 100 200 300 400 500 600 700 800
Tem perature ( oC)
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 100 200 300 400 500 600 700 800
o
Tem perature ( C)
Glass Fibre, Rock Fibre or Cellulose Fibre Used as External Insulation
6-60
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 100 200 300 400 500 600 700 800
Tem perature ( oC)
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 100 200 300 400 500 600 700 800
o
Tem perature ( C)
6-61
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 100 200 300 400 500 600 700 800
Tem perature ( oC)
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 100 200 300 400 500 600 700 800
o
Tem perature ( C)
Glass Fibre, Rock Fibre or Cellulose Fibre Used as External Insulation
6-62
Parametric Studies
Following equations represent the temperature values ranging from 100oC to 800oC
where T is the hot flange temperature in oC and LR is the load ratio. The load ratio
was more than 0.90 when the hot flange temperature was below 100oC. Similarly the
load ratio was less than 0.10 when the hot flange temperature was above 800oC.
Hence these phases were not included in the proposed equations.
1) 1.15 mm G500 steel studs lined on both sides by a single layer of plasterboard
LR = 6.4698x10-14 T5 – 9.7835x10-11 T4 + 3.7866x10-14 T3 + 4.406x10-6 T2
- 5.639x10-3 T + 1.483 (6.15)
2) 1.15 mm G500 steel studs lined on both sides by two layers of plasterboard
LR = 2.2434x10-16 T6 - 6.04755x10-13 T5 + 6.69788x10-10 T4 - 3.89424x10-7 T3
+ 1.2325x10-4 T2 - 2.06746x10-2 T + 2.1552 (6.16)
3) 1.15 mm G500 steel studs lined on both sides by two layers of plasterboard with
glass fibre, rock fibre or cellulose fibre used as cavity insulation
LR = -8.3x10-17 T6 + 2.1535x10-13 T5 - 1.9673x10-10 T4 + 6.8305x10-8 T3
- 7.48x10-7 T2 - 5.13x10-3 T + 1.46 (6.17)
glass fibre, rock fibre or cellulose fibre used as external insulation
LR = -3.077x10-17 T6 + 6.5564x10-14 T5 - 1.9045x10-11 T4 - 3.9x10-8 T3
+ 3.1557x10-5 T2 - 9.226x10-3 T + 1.632 (6.18)
LR = -5.6897x10-14 T5 + 1.2046x10-10 T4- 9.3096x10-8 T3 + 3.2051x10-5 T2
- 6.156x10-3 T + 1.3 (6.19)
6) 1.15 mm G250 steel stud lined on both sides by two layers of plasterboard
LR = -2.2723x10-14 T5 + 3.1615x10-11 T4 - 6.0199x10-9 T3 - 6.5211x10-6 T2
+ 1.1746x10-3 T + 0.85 (6.20)
6-63
Parametric Studies
LR = -1.4022x10-16 T6 + 3.8749 x10-13 T5 - 4.1991 x10-10 T4 + 2.25084 x10-7 T3
- 6.1175 x10-5 T2 + 6.31338 x10-3 T + 0.7189 (6.21)
LR = 3.5067 x10-16 T6 - 9.36134 x10-13 T5 + 9.6919 x10-10 T4 - 4.87467 x10-7 T3
+ 1.22537 x10-4 T2 - 1.5724 x10-2 T + 1.688 (6.22)
LR = -1.426 x10-13 T5 + 3.15954 x10-10 T4 - 2.649515 x10-7 T3 + 1.04696 x10-4 T2
- 2.05434 x10-2 T + 2.196 (6.23)
10) 1.95 mm G250 steel studs lined on both sides by two layers of plasterboard
LR = 8.4125x10-16 T6 - 2.21605x10-12 T5 + 2.30794x10-9 T4 - 1.203529x10-6 T3
+ 3.28476x10-4 T2 - 4.5579x10-2 T + 3.222 (6.24)
LR = 2.374 x10-17 T6 - 1.08515 x10-13 T5 + 1.83108 x10-10 T4 - 1.50213 x10-7 T3
+ 6.4183 x10-5 T2 - 1.4573 x10-2 T + 1.902 (6.25)
LR = 5.28664x10-16T6 - 1.464487 x10-12T5 + 1.601403 x10-9T4 - 8.75564 x10-7T3
+ 2.505 x10-4 T2 - 3.6733 x10-2 T + 2.854 (6.26)
6-64
Parametric Studies
6.10. Comparison of FEA Results with Fire Design Rules
The fire design rules of LSF wall panels were developed in Chapter 5 after
investigating the previous fire design rules proposed by other researchers (Klippstein,
1978, Gerlich et al., 1996, Ranby, 1999, Kaitila, 2002, Feng and Wang, 2005b and
Zhao et al., 2005). Two detailed fire design methods were developed based on
AS/NZS 4600 (SA, 2005) and Eurocode 3 Part 1.3 (ECS, 2006). Eurocode 3 Part 1.2
(ECS, 2005) was considered in Chapter 5, and it was identified to be not as accurate
as Eurocode 3 Part 1.3 (ECS, 2006) in predicting the fire resistance of LSF walls.
Similar outcome was obtained in the study of Feng and Wang (2005b). These fire
design rules were validated for the LSF wall studs with a yield stress of 569 MPa and
an elastic modulus of 213520 MPa. 1.15 mm LSF wall studs were considered with
the measured time-temperature profiles obtained from Chapter 3. A reasonable
agreement was obtained between the test, FEA results and fire design rules.
In this section the developed fire design rules based on AS/NZS 4600 and Eurocode
3 Part 1.3 will be validated for LSF wall studs with varying structural parameters
such as steel grade and thickness. In addition to this the simplified method developed
in Section 6.9 is also compared with the FEA results and fire design rules to validate
the accuracy of Equations 6.15 to 6.26 in predicting the critical hot flange
temperature as a function of load ratio.
Tables 6.39 to 6.46 and Figures 6.60 to 6.67 compare the predictions from the
proposed fire design rules and the simplified method with FEA results for 1.15 mm
G500 steel studs lined on both sides by plasterboard. According to FEA, AS/NZS
4600 (SA, 2005) and Eurocode 3 Part 1.3 (ECS, 2006) the axial compressive
capacities of LSF wall stud at ambient temperature were 70.8 kN, 52.54 kN and
47.92 kN, respectively. The codes were very conservative and hence resulted in
much lower ultimate loads compared to the FEA ultimate loads. Similar finding was
observed in the study of Bandula Heva (2009). However, the variation between the
ultimate loads obtained from FEA and design codes at ambient temperature is
beyond the scope of the current research. Although there is a considerable variation
between these actual ultimate load values, the load ratios of FEA and the fire design
rules in Chapter 5 agreed reasonably well with each other. The agreement between
6-65
Parametric Studies
AS/NZS 4600 (SA, 2005) and Eurocode 3 Part 1.3 (ECS, 2006) was very good
throughout the load ratio curves. The load ratio from the simplified method also
agreed well with the other three results (FEA, AS/NZS 4600 and Eurocode 3 Part
1.3.
Tables 6.47 to 6.54 and Figures 6.68 to 6.75 compare the predictions from the
proposed fire design rules and the simplified method with FEA results for 1.15 mm
G250 steel studs lined on both sides by plasterboard. According to FEA, AS/NZS
4600 (SA, 2005) and Eurocode 3 Part 1.3 (ECS, 2006) the axial compressive
capacities of LSF wall studs at ambient temperature were 43.6 kN, 37.26 kN and
35.44 kN, respectively. The load ratios agreed reasonably well with each other as
observed before for 1.15 mm G500 steel studs. In addition to this the load ratio from
the simplified method also agreed well with the other three values (FEA, AS/NZS
4600 (SA, 2005) and Eurocode 3 Part 1.3 (ECS, 2006)).
Tables 6.55 to 6.62 and Figures 6.76 to 6.83 compare the proposed fire design rules
and the simplified method with FEA results for 1.95 mm G250 steel studs lined on
both sides by plasterboards. According to FEA, AS/NZS 4600 (SA, 2005) and
Eurocode 3 Part 1.3 (ECS, 2006) the axial compressive capacities of LSF wall studs
at ambient temperature were 93.6 kN, 71.84 kN and 68.79 kN, respectively.
Considerable variation was observed between the ultimate load values. However, the
load ratios agreed reasonably well with each other as observed before for 1.15 mm
G500 steel studs and 1.15 mm G250 steel studs. In addition to this the load ratio
from the simplified method also agreed well with the other three values (FEA,
AS/NZS 4600 (SA, 2005) and Eurocode 3 Part 1.3 (ECS, 2006)).
6-66
Parametric Studies
Method for 1.15 mm G500 Steel Studs Lined on Both Sides
by a Single Layer of Plasterboard
FEA AS/NZS 4600 EC3 Part 1.3 Eqn.
HF CF
Time Ult. Ult. Ult.
Temp. Temp. Load Load Load Load
(min.) Load Load Load
(oC) (oC) Ratio Ratio Ratio Ratio
(kN) (kN) (kN)
0 20 20 70.80 1.00 52.54 1.00 47.92 1.00 1.00
14 100 100 70.20 0.99 50.18 0.96 45.76 0.95 0.99
20 196 100 50.00 0.71 40.70 0.77 37.20 0.78 0.71
30 344 184 38.80 0.55 33.14 0.63 30.14 0.63 0.55
40 471 287 31.40 0.44 25.40 0.48 20.41 0.43 0.45
50 577 390 18.80 0.27 13.94 0.27 11.24 0.23 0.26
60 661 661 7.19 0.10 6.97 0.13 6.79 0.14 0.10
70 725 725 4.93 0.07 4.73 0.09 4.57 0.10 0.07
by Two Layers of Plasterboard
HF CF
Time Ult. Ult. Ult.
(kN) (kN) (kN)
0 20 20 70.80 1.00 52.54 1.00 47.92 1.00 1.00
40 100 100 70.20 0.99 50.18 0.96 45.76 0.95 0.99
50 158 100 56.20 0.79 43.93 0.84 40.15 0.84 0.79
60 221 134 49.80 0.70 40.36 0.77 36.81 0.77 0.71
70 285 195 47.10 0.67 38.15 0.73 34.74 0.72 0.66
80 348 256 43.10 0.61 34.72 0.66 30.96 0.65 0.61
90 412 316 38.30 0.54 30.31 0.58 26.18 0.55 0.54
100 475 377 32.40 0.46 25.19 0.48 21.04 0.44 0.46
110 539 438 25.50 0.36 18.68 0.36 15.35 0.32 0.36
120 663 498 9.34 0.13 7.49 0.14 6.44 0.13 0.13
130 784 559 4.21 0.06 1.98 0.04 1.80 0.04 0.06
Method for 1.15 mm G500 Steel Studs Lined on Both Sides by Two Layers of
Plasterboard with Glass Fibre Used as Cavity Insulation
HF CF
Time Ult. Ult. Ult.
(kN) (kN) (kN)
0 20 20 70.80 1.00 52.54 1.00 47.92 1.00 1.00
52 100 100 70.20 0.99 50.18 0.96 45.76 0.95 0.99
60 180 100 52.40 0.74 41.98 0.80 38.35 0.80 0.74
70 292 119 40.70 0.57 34.73 0.66 32.05 0.67 0.57
80 404 169 34.00 0.48 28.62 0.54 26.00 0.54 0.49
90 515 218 26.20 0.37 20.40 0.39 16.68 0.35 0.36
100 627 289 13.30 0.19 8.20 0.16 8.23 0.17 0.21
110 739 409 7.04 0.10 3.44 0.07 3.30 0.07 0.10
6-67
Parametric Studies
Plasterboard with Rock Fibre Used as Cavity Insulation
HF CF
Time Ult. Ult. Ult.
(kN) (kN) (kN)
0 20 20 70.80 1.00 52.54 1.00 47.92 1.00 1.00
52 100 100 70.20 0.99 50.18 0.96 45.76 0.95 0.99
60 177 100 52.80 0.75 42.24 0.80 38.60 0.81 0.75
70 279 119 41.70 0.59 35.57 0.68 32.78 0.68 0.59
80 381 160 35.50 0.50 29.89 0.57 27.26 0.57 0.50
90 483 200 28.40 0.40 23.78 0.45 19.39 0.40 0.40
100 585 241 19.00 0.27 11.73 0.22 11.12 0.23 0.26
110 687 281.6 10.50 0.15 5.22 0.10 5.34 0.11 0.14
120 789 322.2 5.44 0.08 1.64 0.03 1.68 0.04 0.08
Plasterboard with Cellulose Fibre Used as Cavity Insulation
HF CF
Time Ult. Ult. Ult.
(kN) (kN) (kN)
0 20 20 70.80 1.00 52.54 1.00 47.92 1.00 1.00
52 100 100 70.20 0.99 50.18 0.96 45.76 0.95 0.99
60 176 100 53.00 0.75 42.31 0.81 38.65 0.81 0.75
70 266 118 43.20 0.61 36.42 0.69 33.49 0.70 0.60
80 355 156 36.70 0.52 31.52 0.60 28.85 0.60 0.52
90 445 195 31.40 0.44 26.50 0.50 22.72 0.47 0.45
100 534 233 24.90 0.35 18.29 0.35 15.12 0.32 0.34
110 651 320 11.80 0.17 7.12 0.14 7.14 0.15 0.18
115 750 405 6.63 0.09 2.97 0.06 2.88 0.06 0.09
Plasterboard with Glass Fibre Used as External Insulation
HF CF
Time Ult. Ult. Ult.
(kN) (kN) (kN)
0 20 20 70.80 1.00 52.54 1.00 47.92 1.00 1.00
42 100 100 70.20 0.99 50.18 0.96 45.76 0.95 0.98
50 127 100 62.70 0.89 46.97 0.89 42.88 0.89 0.89
60 164 102 55.20 0.78 43.37 0.83 39.60 0.83 0.79
70 204 124 51.30 0.72 41.18 0.78 37.55 0.78 0.72
80 255 164 48.30 0.68 39.09 0.74 35.64 0.74 0.67
90 323 222 44.10 0.62 35.90 0.68 32.44 0.68 0.62
100 412 298 37.60 0.53 30.07 0.57 26.05 0.54 0.54
110 530 392 27.30 0.39 19.55 0.37 15.66 0.33 0.36
120 682 505 8.46 0.12 6.44 0.12 5.50 0.11 0.10
125 773 568 4.60 0.06 2.41 0.05 2.16 0.05 0.07
6-68
Parametric Studies
Plasterboard with Rock Fibre Used as External Insulation

HF CF
Time Ult. Ult. Ult.
(kN) (kN) (kN)
0 20 20 70.80 1.00 52.54 1.00 47.92 1.00 1.00
70 100 100 70.20 0.99 50.18 0.96 45.76 0.95 0.98
80 150 100 57.60 0.81 44.66 0.85 40.77 0.85 0.82
90 207 133 51.90 0.73 41.38 0.79 37.75 0.79 0.72
100 272 177 47.10 0.67 38.35 0.73 34.98 0.73 0.65
110 343 233 42.30 0.60 34.56 0.66 30.99 0.65 0.60
120 418 301 37.30 0.53 29.63 0.56 25.57 0.53 0.53
130 497 381 30.40 0.43 23.00 0.44 18.78 0.39 0.42
140 577 472 17.50 0.25 14.37 0.27 11.54 0.24 0.27
150 659 575 8.70 0.12 7.42 0.14 6.83 0.14 0.13
160 740 690 5.09 0.07 4.02 0.08 3.87 0.08 0.07
Plasterboard with Cellulose Fibre Used as External Insulation

HF CF
Time Ult. Ult. Ult.
(kN) (kN) (kN)
0 20 20 70.80 1.00 52.54 1.00 47.92 1.00 1.00
70 100 100 70.20 0.99 50.18 0.96 45.76 0.95 0.98
80 175 101 53.20 0.75 42.42 0.81 38.75 0.81 0.77
90 258 150 46.60 0.66 38.27 0.73 34.96 0.73 0.66
100 346 216 41.00 0.58 33.92 0.65 30.56 0.64 0.60
110 437 299 35.20 0.50 28.12 0.54 23.64 0.49 0.51
120 526 398 27.70 0.39 19.99 0.38 16.10 0.34 0.36
130 623 515 10.80 0.15 10.00 0.19 8.12 0.17 0.19
140 713 648 6.18 0.09 4.97 0.09 4.81 0.10 0.08
6-69
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70
Tim e (m in.)
FEA AS/NZS 4600 EC3 Part 1.3 Equation
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Tim e (m in.)
6-70
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110
Tim e (m in.)
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
6-71
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130
Tim e (m in.)
6-72
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160
Tim e (m in.)
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Tim e (m in.)
6-73
Parametric Studies
HF CF
Time Ult. Ult. Ult.
(kN) (kN) (kN)
0 20 20 43.60 1.00 37.26 1.00 35.44 1.00 1.00
14 100 100 40.30 0.92 35.43 0.95 33.62 0.95 0.92
20 196 100 34.10 0.78 27.60 0.74 26.54 0.75 0.79
30 344 184 26.50 0.61 19.81 0.53 18.67 0.53 0.60
40 471 287 16.30 0.37 13.78 0.37 12.73 0.36 0.39
50 577 390 11.80 0.27 9.87 0.26 9.01 0.25 0.25
60 661 661 6.51 0.15 6.66 0.18 6.50 0.18 0.16
70 725 725 4.19 0.10 4.23 0.11 4.12 0.12 0.09
HF CF
Time Ult. Ult. Ult.
(kN) (kN) (kN)
0 20 20 43.60 1.00 37.26 1.00 35.44 1.00 1.00
40 100 100 40.30 0.92 35.43 0.95 33.62 0.95 0.90
50 158 100 37.70 0.86 30.34 0.81 29.07 0.82 0.87
60 221 134 33.80 0.78 27.14 0.73 25.93 0.73 0.79
70 285 195 31.40 0.72 24.23 0.65 23.09 0.65 0.68
80 348 256 24.10 0.55 20.33 0.55 19.55 0.55 0.56
90 412 316 18.30 0.42 16.95 0.45 16.39 0.46 0.45
100 475 377 15.60 0.36 14.02 0.38 13.65 0.38 0.35
110 539 438 13.00 0.30 11.44 0.31 11.20 0.32 0.28
120 663 498 8.61 0.20 6.58 0.18 5.91 0.17 0.21
130 784 559 3.35 0.08 1.59 0.04 1.42 0.04 0.08
HF CF
Time Ult. Ult. Ult.
(kN) (kN) (kN)
0 20 20 43.60 1.00 37.26 1.00 35.44 1.00 1.00
52 100 100 40.30 0.92 35.43 0.95 33.62 0.95 0.93
60 180 100 35.50 0.81 28.68 0.77 27.53 0.78 0.81
70 292 119 27.30 0.63 21.88 0.59 20.88 0.59 0.63
80 404 169 21.90 0.50 16.35 0.44 14.81 0.42 0.50
90 515 218 16.00 0.37 12.04 0.32 10.30 0.29 0.37
100 627 289 10.50 0.24 7.72 0.21 6.71 0.19 0.24
110 739 409 6.07 0.14 3.04 0.08 2.70 0.08 0.14
6-74
Parametric Studies
HF CF
Time Ult. Ult. Ult.
(kN) (kN) (kN)
0 20 20 43.60 1.00 37.26 1.00 35.44 1.00 1.00
52 100 100 40.30 0.92 35.43 0.95 33.62 0.95 0.93
60 177 100 35.80 0.82 28.90 0.78 27.74 0.78 0.82
70 279 119 28.20 0.65 22.75 0.61 21.67 0.61 0.65
80 381 160 23.10 0.53 17.35 0.47 15.92 0.45 0.53
90 483 200 18.00 0.41 13.29 0.36 11.45 0.32 0.41
100 585 241 12.70 0.29 9.66 0.26 7.99 0.23 0.29
110 687 282 8.34 0.19 4.91 0.13 4.31 0.12 0.19
120 789 322 4.28 0.10 1.27 0.03 1.20 0.03 0.10
HF CF
Time Ult. Ult. Ult.
(kN) (kN) (kN)
0 20 20 43.60 1.00 37.26 1.00 35.44 1.00 1.00
52 100 100 40.30 0.92 35.43 0.95 33.62 0.95 0.93
60 176 100 35.80 0.82 28.95 0.78 27.78 0.78 0.82
70 266 118 29.20 0.67 23.50 0.63 22.41 0.63 0.67
80 355 156 24.60 0.56 18.67 0.50 17.43 0.49 0.56
90 445 195 19.90 0.46 14.82 0.40 13.08 0.37 0.46
100 534 233 14.90 0.34 11.34 0.30 9.67 0.27 0.35
110 651 320 9.56 0.22 6.66 0.18 5.79 0.16 0.22
115 750 405 5.63 0.13 2.59 0.07 2.31 0.07 0.13
HF CF
Time Ult. Ult. Ult.
(kN) (kN) (kN)
0 20 20 43.60 1.00 37.26 1.00 35.44 1.00 1.00
42 100 100 40.30 0.92 35.43 0.95 33.62 0.95 0.94
50 127 100 40.10 0.92 32.87 0.88 31.25 0.88 0.89
60 164 102 37.10 0.85 29.86 0.80 28.60 0.81 0.85
70 204 124 34.80 0.80 28.02 0.75 26.84 0.76 0.81
80 255 164 32.80 0.75 25.68 0.69 24.41 0.69 0.75
90 323 222 27.70 0.64 21.72 0.58 20.81 0.59 0.62
100 412 298 18.60 0.43 16.76 0.45 16.22 0.46 0.45
110 530 392 13.30 0.31 11.63 0.31 11.12 0.31 0.30
120 682 505 7.78 0.18 5.75 0.15 5.09 0.14 0.17
125 773 568 3.76 0.09 2.00 0.05 1.78 0.05 0.08
6-75
Parametric Studies

HF CF
Time Ult. Ult. Ult.
(kN) (kN) (kN)
0 20 20 43.60 1.00 37.26 1.00 35.44 1.00 1.00
70 100 100 40.30 0.92 35.43 0.95 33.62 0.95 0.94
80 150 100 38.50 0.88 30.93 0.83 29.58 0.83 0.87
90 207 133 35.00 0.80 28.19 0.76 26.96 0.76 0.81
100 272 177 32.00 0.73 24.69 0.66 23.54 0.66 0.72
110 343 233 25.70 0.59 20.44 0.55 19.66 0.55 0.58
120 418 301 18.20 0.42 16.46 0.44 15.95 0.45 0.44
130 497 381 14.70 0.34 13.00 0.35 12.69 0.36 0.33
140 577 472 11.60 0.27 10.02 0.27 9.82 0.28 0.26
150 659 575 8.16 0.19 6.81 0.18 6.62 0.19 0.19
160 740 690 4.38 0.10 3.60 0.10 3.52 0.10 0.10

HF CF
Time Ult. Ult. Ult.
(kN) (kN) (kN)
0 20 20 43.60 1.00 37.26 1.00 35.44 1.00 1.00
70 100 100 40.30 0.92 35.43 0.95 33.62 0.95 0.94
80 175 101 36.00 0.83 29.04 0.78 27.87 0.79 0.84
90 258 150 31.70 0.73 25.04 0.67 23.81 0.67 0.74
100 346 216 26.20 0.60 20.12 0.54 19.26 0.54 0.58
110 437 299 17.40 0.40 15.44 0.41 14.79 0.42 0.40
120 526 398 13.50 0.31 11.81 0.32 11.39 0.32 0.30
130 623 515 9.96 0.23 8.33 0.22 7.99 0.23 0.23
140 713 648 5.45 0.13 4.63 0.12 4.49 0.13 0.13
6-76
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70
Tim e (m in.)
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Tim e (m in.)
6-77
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110
Tim e (m in.)
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
6-78
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130
Time (m in.)
6-79
Parametric Studies
1.00
0.90
0.80
0.70
Load Ratio
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160
Tim e (m in.)
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Tim e (m in.)
6-80
Parametric Studies
HF CF
Time Ult. Ult. Ult.
(kN) (kN) (kN)
0 20 20 93.60 1.00 71.84 1.00 68.79 1.00 1.00
14 100 100 89.40 0.96 68.29 0.95 65.35 0.95 0.95
20 196 100 58.10 0.62 50.49 0.70 49.56 0.72 0.62
30 344 184 44.50 0.48 35.34 0.49 34.62 0.50 0.47
40 471 287 27.70 0.30 24.29 0.34 23.37 0.34 0.31
50 577 390 20.00 0.21 17.26 0.24 16.36 0.24 0.20
60 661 661 14.50 0.15 12.67 0.18 12.42 0.18 0.16
70 725 725 9.06 0.10 8.06 0.11 7.87 0.11 0.10
HF CF
Time Ult. Ult. Ult.
(kN) (kN) (kN)
0 20 20 93.70 1.00 71.84 1.00 68.79 1.00 1.00
40 100 100 89.40 0.95 68.29 0.95 65.35 0.95 0.95
50 158 100 66.60 0.71 56.38 0.78 54.96 0.80 0.71
60 221 134 58.30 0.62 49.73 0.69 48.68 0.71 0.64
70 285 195 55.10 0.59 44.24 0.62 43.51 0.63 0.56
80 348 256 40.60 0.43 37.01 0.52 36.53 0.53 0.45
90 412 316 31.20 0.33 30.79 0.43 30.45 0.44 0.34
100 475 377 26.20 0.28 25.39 0.35 25.15 0.37 0.27
110 539 438 21.80 0.23 20.64 0.29 20.49 0.30 0.23
120 663 498 12.44 0.17 11.42 0.17 0.15
130 784 559 3.16 0.04 2.83 0.04
HF CF
Time Ult. Ult. Ult.
(kN) (kN) (kN)
0 20 20 93.70 1.00 71.84 1.00 68.79 1.00 1.00
52 100 100 89.40 0.95 68.29 0.95 65.35 0.95 0.95
60 180 100 61.30 0.65 52.71 0.73 51.63 0.75 0.65
70 292 119 46.10 0.49 39.04 0.54 38.54 0.56 0.49
80 404 169 37.10 0.40 28.65 0.40 27.28 0.40 0.40
90 515 218 27.70 0.30 20.70 0.29 18.90 0.27 0.29
100 627 289 18.20 0.19 14.08 0.20 12.50 0.18 0.20
110 739 409 11.70 0.12 6.12 0.09 5.31 0.08 0.12
6-81
Parametric Studies
HF CF
Time Ult. Ult. Ult.
(kN) (kN) (kN)
0 20 20 93.70 1.00 71.84 1.00 68.79 1.00 1.00
52 100 100 89.40 0.95 68.29 0.95 65.35 0.95 0.95
60 177 100 61.90 0.66 53.22 0.74 52.07 0.76 0.66
70 279 119 47.60 0.51 40.71 0.57 40.11 0.58 0.51
80 381 160 38.90 0.42 30.51 0.42 29.35 0.43 0.42
90 483 200 30.80 0.33 22.97 0.32 21.10 0.31 0.32
100 585 241 22.10 0.24 16.39 0.23 14.62 0.21 0.23
110 687 281.6 14.50 0.15 9.89 0.14 8.41 0.12 0.15
120 789 322.2 8.06 0.09 2.49 0.03 2.36 0.03 0.09
HF CF
Time Ult. Ult. Ult.
(kN) (kN) (kN)
0 20 20 93.70 1.00 71.84 1.00 68.79 1.00 1.00
52 100 100 89.40 0.95 68.29 0.95 65.35 0.95 0.95
60 176 100 62.10 0.66 53.31 0.74 52.15 0.76 0.66
70 266 118 49.30 0.53 42.23 0.59 41.55 0.60 0.52
80 355 156 41.50 0.44 33.03 0.46 32.18 0.47 0.44
90 445 195 33.80 0.36 25.82 0.36 24.12 0.35 0.36
100 534 233 25.80 0.28 19.46 0.27 17.72 0.26 0.28
110 651 320 16.50 0.18 12.85 0.18 11.24 0.16 0.18
115 750 405 10.90 0.12 5.22 0.07 4.56 0.07 0.11
HF CF
Time Ult. Ult. Ult.
(kN) (kN) (kN)
0 20 20 93.70 1.00 71.84 1.00 68.79 1.00 1.00
42 100 100 89.40 0.95 68.29 0.95 65.35 0.95 0.96
50 127 100 76.10 0.81 62.03 0.86 59.96 0.87 0.81
60 164 102 65.40 0.70 55.26 0.77 53.92 0.78 0.70
70 204 124 60.20 0.64 51.47 0.72 50.37 0.73 0.65
80 255 164 56.50 0.60 46.87 0.65 45.89 0.67 0.59
90 323 222 47.30 0.50 39.45 0.55 38.88 0.57 0.50
100 412 298 31.40 0.34 30.23 0.42 29.97 0.44 0.35
110 530 392 22.50 0.24 20.70 0.29 20.30 0.30 0.23
120 682 505 14.00 0.15 10.88 0.15 9.86 0.14 0.14
125 773 568 8.17 0.09 3.97 0.06 3.51 0.05 0.09
6-82
Parametric Studies

HF CF
Time Ult. Ult. Ult.
(kN) (kN) (kN)
0 20 20 93.70 1.00 71.84 1.00 68.79 1.00 1.00
70 100 100 89.40 0.95 68.29 0.95 65.35 0.95 0.96
80 150 100 68.70 0.73 57.63 0.80 56.10 0.82 0.73
90 207 133 61.00 0.65 51.87 0.72 50.69 0.74 0.64
100 272 177 55.20 0.59 44.98 0.63 44.17 0.64 0.58
110 343 233 43.20 0.46 37.00 0.52 36.53 0.53 0.47
120 418 301 30.90 0.33 29.64 0.41 29.38 0.43 0.34
130 497 381 24.70 0.26 23.36 0.33 23.24 0.34 0.25
140 577 472 19.40 0.21 17.98 0.25 17.87 0.26 0.21
150 659 575 14.30 0.15 12.86 0.18 12.62 0.18 0.16
160 740 690 8.90 0.09 6.81 0.09 6.71 0.10 0.09

HF CF
Time Ult. Ult. Ult.
(kN) (kN) (kN)
0 20 20 93.70 1.00 71.84 1.00 68.79 1.00 1.00
70 100 100 89.40 0.95 68.29 0.95 65.35 0.95 0.96
80 175 101 62.50 0.67 53.52 0.75 52.34 0.76 0.68
90 258 150 54.00 0.58 45.45 0.63 44.54 0.65 0.59
100 346 216 44.10 0.47 36.18 0.50 35.78 0.52 0.46
110 437 299 29.50 0.31 27.61 0.38 27.23 0.40 0.31
120 526 398 22.70 0.24 21.11 0.29 20.83 0.30 0.23
130 623 515 16.60 0.18 15.09 0.21 14.83 0.22 0.19
140 713 648 11.00 0.12 8.74 0.12 8.58 0.12 0.11
6-83
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70
Tim e (m in.)
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Tim e (m in.)
6-84
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110
Tim e (m in.)
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
6-85
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120
Tim e (m in.)
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130
Time (m in.)
6-86
Parametric Studies
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160
Tim e (m in.)
1.00
0.90
0.80
0.70
0.60
Load Ratio
0.50
0.40
0.30
0.20
0.10
0.00
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Tim e (m in.)
6-87
Parametric Studies
6.11. Failure Times Obtained from Test, FEA, Fire Design Rules and the
Simplified Method
Tables 6.63 to 6.65 show the failure times predicted for 1.15 mm G500 steel studs
lined on both sides by plasterboard for the load ratios of 0.2, 0.4 and 0.7,
respectively. These tables clearly indicate the superior performance of externally
insulated LSF wall panels. The adverse effect of the cavity insulation is that it will
reflect back the heat and hence the hot flange temperature of the stud will increase
rapidly. The other problem associated with cavity insulation is the thermal bowing
and magnification effects due to high temperature gradient across the stud. These
unfavourable effects will reduce the fire resistance of LSF wall panels considerably.
At a load ratio of 0.2 when rock fibre and cellulose fibre were used as cavity
insulation, the fire resistance reduced by 9% compared to the non-insulated wall
panels (see Table 6.63). However, the fire resistance is increased by 23% when rock
fibre was used as external insulation. Cellulose fibre was not as good as rock fibre
and still resulted in 9% increase in fire resistance when it was used as external
insulation compared to non-insulated wall panels. The studs withstand higher
temperatures when it is subjected to lower load ratios. However, at these higher
temperatures the glass fibre melts and hence it is ineffective. Therefore at a load ratio
of 0.2 when glass fibre was used as external insulation the failure time did not
change much compared to the non-insulated wall panels (see Tables 6.63, 6.66 and
6.69). Hence glass fibre is not recommended in LSF wall panels when it is subjected
to lower load ratios. It should be also noted that the fire resistance reduced by 8%
when glass fibre was used as cavity insulation. The reason for this is the glass fibre
was protected by two plasterboards when it is placed in the cavity. Therefore the
glass fibre will not melt fully and hence the adverse effects of cavity insulation will
be still valid.
At a load ratio of 0.4 the fire resistance reduced (on average) by 14% when
insulations (glass fibre, rock fibre and cellulose fibre) were used as cavity insulation
compared to the non-insulated wall panels (see Table 6.64). However, when rock
fibre and cellulose fibre were used as external insulation the fire resistance is
increased by 25% and 12%, respectively.
6-88
Parametric Studies
At a load ratio of 0.7 when glass fibre, rock fibre and cellulose fibre were used as
external insulation the fire resistance was increased by 25%, 56% and 41%,
respectively. This is a considerable improvement in fire resistance of LSF walls.
Tables 6.63 and 6.64 also include the full scale fire test results obtained from the
current experimental study and Kolarkar (2010). The agreement between the test and
FEA results was good and this indicates that the idealised time-temperature profiles
accurately represent the measured time-temperature profiles. It should be noted that
the slight variation in yield stress (from 569 MPa to 500 MPa) and elastic modulus
(from 213520 MPa to 200000 MPa) also did not have any significant effect on the
failure times of LSF walls. These tables also compare the FEA results with the
failure times predicted by the proposed fire design rules based on AS/NZS 4600 (SA,
2005) and Eurocode 3 Part 1.3 (ECS, 2006), and the simplified method in Chapter 5.
They all agreed well with FEA results for all the wall configurations.
Both Sides by Plasterboard for a Load Ratio of 0.2
Failure Time (min.)

Index Configuration Insulation AS/NZS Eurocode 3 Eqn.
Test FEA
4600 Part 1.3
1x1 None 53* 54 55 54 54
2x2 None 111* 117 117 116 117

Glass
CI-GF 101* 99 98 98 101
Fibre
Rock
CI-RF 107* 106 102 103 105
Fibre
Cellulose
CI-CF 110* 108 107 107 109
Fibre
Glass
CP-GF 118 117 117 116 116
Fibre
Rock
CP-RF 136*# 144 146 144 145
Fibre
Cellulose
CP-CF 124* 128 129 128 129
Fibre
( * ) - Tests conducted by Kolarkar (2010)

( # ) - Experimental error
6-89
Parametric Studies
Failure Time (min.)

Test FEA
4600 Part 1.3
1x1 None - 42 44 41 43
2x2 None - 106 106 103 106

Glass
CI-GF - 87 89 87 87
Fibre
Rock
CI-RF - 90 92 90 90
Fibre
Cellulose
CI-CF - 95 97 95 94
Fibre
Glass
CP-GF 108 109 109 107 108
Fibre
Rock
CP-RF 134 132 132 129 131
Fibre
Cellulose
CP-CF - 119 119 116 118
Fibre
Failure Time (min.)

FEA
4600 Part 1.3
1x1 None 20 25 25 20
2x2 None 61 74 73 62
Glass
CI-GF 62 67 68 62
Fibre
Rock
CI-RF 63 68 69 63
Fibre
Cellulose
CI-CF 64 69 70 63
Fibre
Glass
CP-GF 76 87 87 74
Fibre
Rock
CP-RF 95 104 104 93
Fibre
Cellulose
CP-CF 86 93 93 86
Fibre
6-90
Parametric Studies
lined on both sides by plasterboard for the load ratios of 0.2, 0.4 and 0.7,
respectively. These tables also clearly indicate the superior performance of externally
insulated LSF wall panels. As discussed before the externally insulated glass fibre
was ineffective at a load ratio of 0.2 and hence resulted in similar failure time as non-
insulated wall panels (see Table 6.66). The fire resistance reduced (on average) by
9% when insulations (glass fibre, rock fibre and cellulose fibre) were used as cavity
insulation compared to the non-insulated wall panels (see Table 6.66). However,
when rock fibre and cellulose fibre were used as external insulation the fire
resistance was increased by 23% and 11%, respectively. At a load ratio of 0.4 when
glass fibre, rock fibre and cellulose fibre were used as external insulation the fire
resistance was increased by 10%, 31% and 18%, respectively. At a load ratio of 0.7
when glass fibre, rock fibre and cellulose fibre were used as external insulation the
fire resistance was increased by 18%, 44% and 30%, respectively. This is a
considerable improvement in fire resistance of LSF walls. It should be noted that the
FEA results and the predicted failure times using the proposed fire design rules based
on AS/NZS 4600 (SA, 2005) and Eurocode 3 Part 1.3 (ECS, 2006), and the
simplified method agreed well with the FEA results.
Failure Time (min.)
Index Configuration Insulation AS/NZS Eurocode 3
FEA Eqn.
4600 Part 1.3
1x1 None 56 58 58 56
2x2 None 120 118 118 121

Glass
CI-GF 104 101 99 104
Fibre
Rock
CI-RF 109 105 102 109
Fibre
Cellulose
CI-CF 111 108 107 111
Fibre
Glass
CP-GF 118 117 117 118
Fibre
Rock
CP-RF 148 148 149 149
Fibre
Cellulose
CP-CF 133 132 133 133
Fibre
6-91
Parametric Studies
Failure Time (min.)

FEA
4600 Part 1.3
1x1 None 39 38 38 40
2x2 None 93 97 98 95
Glass
CI-GF 88 83 81 88
Fibre
Rock
CI-RF 91 86 84 91
Fibre
Cellulose
CI-CF 95 90 87 95
Fibre
Glass
CP-GF 102 104 104 103
Fibre
Rock
CP-RF 122 124 125 123
Fibre
Cellulose
CP-CF 110 111 112 110
Fibre
Failure Time (min.)

FEA
4600 Part 1.3
1x1 None 25 22 22 25
2x2 None 71 64 64 68
Glass
CI-GF 66 64 64 66
Fibre
Rock
CI-RF 67 65 65 67
Fibre
Cellulose
CI-CF 68 65 66 68
Fibre
Glass
CP-GF 84 78 78 84
Fibre
Rock
CP-RF 102 96 96 101
Fibre
Cellulose
CP-CF 92 87 88 93
Fibre
6-92
Parametric Studies
lined on both sides by plasterboards for the load ratios of 0.2, 0.4 and 0.7,
respectively. These tables also clearly indicate the superior performance of externally
insulated LSF wall panels. As discussed before the externally insulated glass fibre
was ineffective at a load ratio of 0.2 (see Table 6.69). The fire resistance reduced (on
average) by 9% when insulations (glass fibre, rock fibre and cellulose fibre) were
used as cavity insulation compared to the non-insulated wall panels (see Table 6.66).
However, when rock fibre and cellulose fibre were used as external insulation the
fire resistance was increased by 20% and 8%, respectively. At a load ratio of 0.4
when glass fibre, rock fibre and cellulose fibre were used as external insulation the
fire resistance was increased by 16%, 39% and 27%, respectively. At a load ratio of
0.7 when glass fibre, rock fibre and cellulose fibre were used as external insulation
the fire resistance was increased by 18%, 65% and 55%, respectively. This is a
considerable improvement in fire resistance of LSF walls. It should be noted that the
FEA results and the predicted failure times using the proposed fire design rules based
on AS/NZS 4600 (SA, 2005) and Eurocode 3 Part 1.3 (ECS, 2006), and the
simplified method agreed well with the FEA results.
Failure Time (min.)

FEA
4600 Part 1.3
1x1 None 52 56 57 51
2x2 None 117 118 117 114

Glass
CI-GF 99 100 98 100
Fibre
Rock
CI-RF 104 103 101 104
Fibre
Cellulose
CI-CF 108 108 106 108
Fibre
Glass
CP-GF 114 116 116 114
Fibre
Rock
CP-RF 141 147 148 142
Fibre
Cellulose
CP-CF 126 131 132 128
Fibre
6-93
Parametric Studies
Failure Time (min.)

FEA
4600 Part 1.3
1x1 None 34 36 36 34
2x2 None 83 94 96 85
Glass
CI-GF 80 80 80 80
Fibre
Rock
CI-RF 82 82 82 82
Fibre
Cellulose
CI-CF 85 86 86 85
Fibre
Glass
CP-GF 96 102 103 97
Fibre
Rock
CP-RF 115 121 123 115
Fibre
Cellulose
CP-CF 105 109 110 104
Fibre
Failure Time (min.)

FEA
4600 Part 1.3
1x1 None 19 20 21 19
2x2 None 51 59 61 51
Glass
CI-GF 59 62 63 59
Fibre
Rock
CI-RF 59 62 63 59
Fibre
Cellulose
CI-CF 59 63 64 59
Fibre
Glass
CP-GF 60 73 75 60
Fibre
Rock
CP-RF 84 92 94 84
Fibre
Cellulose
CP-CF 79 84 85 79
Fibre
6-94
Parametric Studies
In summary, the superior performance of externally insulated LSF wall panels is

verified in this section for different steel grades and thicknesses. It was concluded
that the proposed fire design rules based on AS/NZS 4600 (SA, 2005) and Eurocode
3 Part 1.3 (ECS, 2006) (discussed in Chapter 5) and the simplified method proposed
in this chapter (Equations 6.15 to 6.26) can be used to find the failure times of LSF
walls subject to standard fire conditions. The prediction was considered to be
reasonably good considering the uncertainties incorporated in LSF wall stud deign
under fire conditions.
6.12. Failure Temperatures Obtained from Test, FEA and Fire Design Rules
Tables 6.72 to 6.74 show the FEA failure temperatures of LSF wall studs with
different steel grades and thicknesses for the load ratios of 0.2, 0.4 and 0.7,
respectively. The failure temperature is the hot flange temperature of the steel stud at
the time of failure for a given load ratio. The failure temperatures were listed for
eight different wall configurations including non-insulated, cavity insulated and
externally insulated wall panels.
Higher values were obtained for the failure temperatures of LSF wall studs made of
1.15 mm G250 steel compared to those made of 1.95 mm G250 steel. This behaviour
was observed in all the wall configurations at load ratios of 0.2, 0.4 and 0.7 (see
Tables 6.72 to 6.74). The reason for this was due to the bending moment caused by
eccentric loading. The actual load the stud carries and the corresponding bending
moment due to thermal bowing and magnified thermal effects play a major role at
higher load ratios like 0.7. In other words the stud carries about 70% of its ultimate
capacity (at ambient temperature) while the thermal bowing and the magnified
effects are at the maximum level. Therefore the LSF wall stud with the highest
ultimate load at the ambient temperature will fail at the earliest with the lowest
failure temperature. The ultimate capacities of LSF wall studs according to FEA are
43.6 kN and 93.7 kN for 1.15 mm G250 steel studs and 1.95 mm G250 steel studs,
respectively. Hence 1.15 mm LSF wall studs withstood higher temperatures
compared to 1.95 mm LSF wall studs for a given load ratio.
6-95
Parametric Studies
The failure temperatures of 1.15 mm G500 and G250 steel studs did not follow a
similar pattern as explained before. This time, when lower load ratios are considered
the failure temperature is influenced mainly by yield stress reduction patterns. This is
because the loads are not large enough in this case to influence the failure
temperature compared to the yield stress reduction pattern. However, as discussed
before, the bending moment due to magnification effects is critical at higher load
ratios.
When the load ratio is 0.2, the corresponding steel temperatures should be used to
determine the yield stress. However, it should be noted that all the temperatures
listed in Table 6.72 are greater than 530oC. At temperatures beyond 530oC the
reduction in yield stress at elevated temperature for G500 grade steel is higher than
that for G250 grade steel (see Figure 6.84). Therefore Grade 500 steel studs
withstood only lower temperatures compared to G250 grade steel at a load ratio of
0.2.
When the load ratio is 0.4, all the temperatures listed in Table 6.73 are less than
530oC. At temperatures below 530oC, the reduction in yield stress is less for G500
grade steel compared to G250 grade steel (see Figure 6.84). Therefore Grade 500
steel studs withstood larger temperatures compared to G250 grade steel at a load
ratio of 0.4.
When higher load ratios are considered (for example 0.7), the failure temperature is
dominated by the bending moment due to eccentricity. Therefore the failure
temperatures of G500 steel studs were less than those of G250 steel studs as
explained before (see Table 6.74).
6-96
Parametric Studies
Table 6.72: FEA Failure Temperatures of LSF Wall Panels

for a Load Ratio of 0.2
FEA Failure Temperature (oC)

Index Configuration Insulation G500 G250 G250
t=1.15 mm t=1.15 mm t=1.95 mm
1x1 None 611 626 597
2x2 None 626 660 583

Glass
CI-GF 620 672 621
Fibre
Rock
CI-RF 643 678 630
Fibre
Cellulose
CI-CF 630 672 623
Fibre
Glass
CP-GF 636 656 597
Fibre
Rock
CP-RF 608 645 588
Fibre
Cellulose
CP-CF 603 648 589
Fibre


t=1.15 mm t=1.15 mm t=1.95 mm
1x1 None 497 457 398
2x2 None 513 432 369

Glass
CI-GF 485 488 399
Fibre
Rock
CI-RF 484 494 399
Fibre
Cellulose
CI-CF 487 489 402
Fibre
Glass
CP-GF 518 438 378
Fibre
Rock
CP-RF 510 435 378
Fibre
Cellulose
CP-CF 519 437 387
Fibre
6-97
Parametric Studies

t=1.15 mm t=1.15 mm t=1.95 mm
1x1 None 202 266 173
2x2 None 227 292 165

Glass
CI-GF 207 248 168
Fibre
Rock
CI-RF 207 248 167
Fibre
Cellulose
CI-CF 208 248 167
Fibre
Glass
CP-GF 234 285 163
Fibre
Rock
CP-RF 239 289 173
Fibre
Cellulose
CP-CF 221 277 167
Fibre
0.9
0.8
0.7
Reduction Factor
0.6
T = 530 oC
0.5
0.4
0.3
0.2
0.1
0
0 100 200 300 400 500 600 700 800
o
Tem perature ( C)
G500 G250
Figure 6.84: Reduction Factors for Yield Strength at Elevated Temperatures
6-98
Parametric Studies
6.13. Conclusions
This chapter has proposed suitable equations to predict the idealised time-
temperature profiles of eight different LSF wall panels subject to standard fire
conditions. The non-insulated LSF wall panels with single and double layers of
plasterboards were considered. Cavity and externally insulated wall panels were also
considered with glass fibre, rock fibre and cellulose fibre used as insulations.
This chapter has also described an extensive finite element analysis based parametric
study on the axial compression capacities and failure times of LSF wall studs subject
to non-uniform temperature distributions. The LSF wall panels with varying steel
grades, steel thicknesses, screw spacing and plasterboard restraint were considered in
this study. Although there was a slight reduction in the ultimate compression
capacity of LSF wall stud, it was found that there was not much difference in failure
time when a screw spacing other than the commonly used 300 mm spacing was used.
The failure times of LSF wall studs with and without plasterboard restraints were
investigated and it is concluded that the plasterboard restraint can be assumed to be
effective throughout the failure time in routine fire design calculations. The ultimate
load capacity at ambient temperature and the material yield stress reduction pattern at
elevated temperatures have a major influence on the fire resistance of LSF walls.
It was concluded that the failure temperatures of LSF walls do not depend on the
type of insulation. The effect of using different types of insulation is simply to delay
the time to reach the same hot flange temperatures in the LSF wall stud. The superior
performance of externally insulated LSF wall panels was verified in this section for
different steel grades and steel thicknesses. A new set of equations was proposed in
this chapter to find the failure temperature and the failure times of LSF wall studs
under non-uniform temperature distributions. The proposed load ratio equations can
be used to find the critical hot flange (failure) temperature of LSF wall studs with
different wall configurations, steel grades and thickness and load ratios. They can
then be used with the proposed idealised time-temperature profile equations to find
the failure times of LSF walls in each case. It was also found that the proposed fire
design rules based on AS/NZS 4600 (SA, 2005) and Eurocode 3 Part 1.3 (ECS,
2006) accurately predict the failure times of LSF wall panels under fire conditions
with different wall configurations, steel grades, thickness and load ratios.
6-99
Conclusions and Recommendations
CHAPTER 7
7. CONCLUSIONS AND RECOMMENDATIONS
This thesis has described a detailed investigation into the structural and thermal
behaviour of cold-formed light gauge steel frame (LSF) load-bearing wall panels
lined with gypsum plasterboards subject to standard fire conditions. It included both
the conventional steel wall systems with and without cavity insulation as well as a
new steel wall system based on a composite panel in which a layer of external
insulation was used between two plasterboards. Both full scale fire tests and
extensive numerical studies were conducted on a series of LSF load-bearing wall
panels under fire conditions. This research has thus developed comprehensive
experimental and numerical data for both the conventional and the new load-bearing
cold-formed steel wall systems under fire conditions including simple and accurate
methods to predict their fire resistance rating. It has considerably improved the
knowledge and understanding of the fire performance of LSF wall systems under fire
conditions, and has led to the development of both accurate design methods for fire
conditions and new LSF wall systems with increased fire resistance rating.
In the first phase of this research (see Chapter 3), full scale experiments were
conducted in the fire research laboratory of Queensland University of Technology to
evaluate the fire performance of LSF wall assemblies subject to different load ratios
under standard fire conditions. Three full scale fire tests were conducted with the
newly developed composite wall panel system developed at QUT. All three test LSF
walls were built using this composite panel system in which glass and rock fibre
insulations were sandwiched between the plasterboards on both sides of the steel
wall frame. Fire test results demonstrated the improved fire performance of LSF wall
assemblies when insulation was used externally between plasterboards instead of
using it in the cavity.
In the second phase (see Chapter 4), finite element models of LSF walls under fire
conditions were developed. A model was developed first to simulate the behaviour of
LSF wall studs at ambient temperature. It was then extended to simulate the
7-1
behaviour of LSF wall studs under non-uniform elevated temperature conditions

developed as a result of being exposed to fire. The finite strip analysis program
CUFSM was used for elastic buckling analyses of LSF wall studs at ambient
temperature while the finite element program ABAQUS was used for both elastic
buckling and nonlinear analyses under ambient and fire conditions. Elastic buckling
analysis results from ABAQUS and CUFSM agreed well while the nonlinear
analysis results compared well with fire test results, thus enabling a full validation of
the developed finite element models of LSF wall studs exposed to standard fire
conditions.
In the third phase of this research (see Chapter 5) a detailed investigation into the
prediction of axial compression strength and behaviour of LSF wall studs was
performed to increase the understanding of their behaviour under non-uniform
temperature conditions developed during fires. The measured time-temperature
profiles in the fire tests conducted at QUT were used in this investigation. The failure
times of 10 full scale LSF wall panels obtained from both experiments and finite
element analyses were used to investigate the accuracy of the currently available fire
design rules. In this investigation, the importance of considering thermal bowing and
its magnification effects, and the neutral axis shift in fire design was investigated.
New fire design rules were proposed within the guidelines of three cold-formed steel
structures codes, namely, AS/NZS 4600 (SA, 2005), NAS (AISI, 2007) and
Eurocode 3 Part 1.3 (ECS, 2006). The proposed design method is available as a
simple spread sheet based design tool to predict the fire resistance rating as a
function of load ratio given the details of LSF wall panel configuration (stud sizes,
steel grade and plasterboard and insulation details).
In the fourth phase (see Chapter 6), idealised time-temperature profiles for LSF wall
studs subject to standard fire conditions were developed based on 10 full scale fire
tests conducted in this research and Kolarkar (2010). These idealised time-
temperature profiles were used in further numerical investigations of LSF wall studs
under fire conditions. The validated finite element models developed in the second
phase were then used to conduct an extensive parametric study to investigate the
behaviour of LSF wall studs. This study included the effects of various parameters
such as steel grade, steel thickness, screw spacing, plasterboard restraint, various
7-2
insulation materials and load ratios. The nonlinear analysis results of LSF wall studs
obtained from this phase were used to compare and validate the fire design rules
developed in Phase 3 for varying steel grades and thicknesses. Finally, simple design
rules were also proposed to predict the failure hot flange temperatures and failure
times of LSF wall panels with varying wall configurations (single layer of
plasterboard, double layers of plasterboards, cavity insulted and externally insulated)
and structural parameters (steel grade and thickness) under different load ratios.
Main Research Outcomes
The most significant findings from this research are as follows:
Significantly improved the knowledge and understanding of the structural and

thermal performance of cold-formed LSF wall systems under fire conditions.
Load-bearing walls with varying arrangements of plasterboard and insulation
were included.
Demonstrated the superiority and benefits of a new cold-formed LSF wall

using a composite panel system in which a layer of insulation is placed
externally between the two plasterboards.
Developed comprehensive structural and thermal performance data from

experiments of the new LSF wall systems with different load ratios, which
can be used for accurate numerical modelling and design of LSF walls by fire
researchers and designers.
Developed suitable finite element models to simulate the structural and fire
performance of LSF wall panels under fire conditions. The use of accurate
numerical models allowed the inclusion of various complex thermal and
structural effects such as thermal bowing, local buckling and neutral axis shift
at non-uniform temperatures. The finite element analyses were conducted
under both steady and transient state conditions and validated with full scale
fire tests. The developed finite element models can be used to accurately
simulate the structural fire performance of LSF wall panels with varying
structural parameters without the need for further full scale fire tests.
Developed new fire design rules for LSF wall studs based on AS/NZS 4600
(SA, 2005), NAS (AISI, 2007) and Eurocode 3 Part 1.3 (ECS, 2006). These
7-3
fire design rules reduce the complexity involved in the existing fire design
rules developed by the previous researchers, without affecting the accuracy.
The new fire design rules are also based on the current cold-formed steel
design codes and include suitable interaction of compression and bending
actions.
Developed a design tool using a spread sheet based on AS/NZS 4600 (SA,
2005), NAS (AISI, 2007) and Eurocode 3 Part 1.3 (ECS, 2006). Using the
known structural parameters such as cross-sectional dimensions and
mechanical properties and the details of plasterboard and insulation
configuration used as input, Structural and Fire Engineers can obtain the
required load ratio versus failure time curves, failure loads and times and the
critical temperatures for different wall configurations.
Developed idealised time-temperature profiles of LSF wall studs under

standard fire conditions as a function of varying configurations of
plasterboards and insulation. Engineers, designers and researchers in this field
can use them without the need for further expensive and time consuming full
scale fire tests.
Considerably improved the understanding and knowledge of the influence of

steel grades, steel thicknesses, screw spacing and plasterboard restraint on the
fire resistance and performance of LSF wall panels.
Developed a simplified design method to predict the hot flange temperatures

at failure and failure times for varying load ratios, steel grades and
thicknesses, and wall configurations (plasterboard and insulation
arrangements). The proposed equations can be used with the idealised time-
temperature profiles to find the failure times of LSF wall panels.
Following conclusions and recommendations have been drawn based on the specific
topics investigated in this research.
7-4
7.1. Full Scale Fire Tests
The new cold-formed LSF wall systems with external insulation provided
considerably increased fire resistance rating with smaller lateral deformations
than the conventional cavity insulated LSF wall systems.
The temperature distributions of the studs were similar for thermally similar
wall panels independent of their load ratios.
The plasterboard fall-off occurs when its unexposed face reaches about
900°C.
Test results including temperature and deflection versus time are available
along with the stud failure modes, which can be used with finite element
modeling and validations.
7.2. Finite Element Modelling
The lateral restraint of the plasterboard increases the ultimate capacity of the
stud significantly. In FEA, restraining one node of the flange appears to be
adequate instead of restraining the whole flange.
Both transient and steady state methods could be used to analyse the LSF
wall studs under fire conditions.
FEA results confirmed the occurrence of local buckling in the stud even
before the failure.
When the stud was analysed under transient state conditions, the agreement
of the deflection versus time graphs with test results was very good compared
to the previous numerical studies of LSF walls under fire conditions due to
the use of accurate time-temperature profiles and mechanical properties in
finite element analyses.
In both steady and transient state analyses, the results agreed well with test
results when average temperatures along the stud were used. The developed
finite element model was able to predict the failure time within five minutes,
which is a considerable improvement compared to previous researches.
At higher load ratios, the net effect of thermal bowing and neutral axis shift
will act as an initial (global) geometric imperfection about the major axis.
The LSF wall stud is restrained about the minor axis and hence the failure
mode will be flexural buckling about the major axis.
7-5
At lower load ratios, the LSF wall stud withstand higher temperatures and the
yield stress of the steel reduces considerably particularly for high grade steel.
In this case the failure will be due to the local buckling of either hot flange or
cold flange depending on their temperature dependent yield stress and the
applied stress due to compression and bending effects.
In FEA, the failure location of LSF wall studs under fire conditions changes
from the cold flange at mid-height to the hot flange at support with
decreasing load ratios. The failure location under lower load ratios can also
be at the cold flange at mid-height when the cold flange temperature is as
high as the hot flange temperature. However, in full scale fire tests, the mid-
height will be close to the furnace due to thermal bowing. Hence the mid-
height temperature will be greater than the temperature at the supports.
Therefore at lower load ratios the hot flange failure will mostly occur at mid-
height than the supports.
In all the finite element analyses, the LSF test specimens under the load ratios
of 0.2 and 0.4 failed at about 600oC and 500oC, respectively. This shows that
the current limiting temperature method of using 350oC in the fire design of
cold-formed steel structures is too conservative.
A small reduction in load ratio was experienced even at lower temperatures
such as 100-300oC, where there is no reduction in material yield stress. This
is due to the reduction in elastic modulus and the induced bending moment
caused by thermal bowing.
Near failure, the hot flange temperatures increased rapidly for most of the
tests due to the fall off of some plasterboard pieces. Therefore a rapid fall in
load ratio was observed with time. This phase covers load ratios of 0.4 and
0.2. Hence the predicted failure times for load ratios of 0.2 and 0.4 were
closer to each other.
7.3. Fire Design Rules
Lateral deflections of the studs due to non-uniform temperature conditions

were predicted using the available equations proposed by previous
researches. These deflections agreed reasonably well with those obtained
from finite element analyses.
7-6
Applications of previously developed fire design rules based on AISI design

manual and Eurocode 3 Parts 1.2 and 1.3 to LSF wall studs were investigated
in detail and suitable modification were proposed where necessary.
Traditionally it was assumed that the entire stud cross-section has reached its
yield stress in the prediction of local buckling capacity. This resulted in an
over-estimation of the local buckling capacity of studs with non-uniform
temperature distributions due to the fact that the hot flange may fail before
the cold flange stress reaches its yield limit. Therefore it is recommended to
limit the yield stress of cold flange to 1.5 times the yield stress of hot flange
in the determination of local buckling capacity when a high non-uniform
temperature distribution exists.
It is recommended that the weighted average mechanical properties are used
instead of the mechanical property at web temperature to determine the
ultimate compression capacities of LSF wall studs under fire conditions.
A simplified method was proposed to determine the section moment capacity
that has minimal effects on the accuracy of the ultimate compression capacity
of LSF wall studs.
New fire design rules were developed for LSF wall studs based on AS/NZS
4600 (SA, 2005), NAS (AISI, 2007) and Eurocode 3 Part 1.3 (ECS, 2006).
The accuracy of the proposed fire design rules was verified with the available
test and FEA results. The agreement of failure times was reasonably good.
Developed a design tool using a spread sheet to find the failure load/load ratio
versus time curves, and temperatures for different steel stud sections, steel
grades and thicknesses and wall configurations.
7.4. Idealised Time - Temperature Profiles and Parametric Studies
Suitable equations were proposed to predict the idealised time-temperature

profiles of eight different LSF wall panels subject to standard fire conditions,
ie. non-insulated LSF wall panels with single and double layers of
plasterboards, cavity and externally insulated wall panels with glass fibre,
rock fibre and cellulose fibre used as insulations.
An extensive parametric study was conducted on the axial compression
capacities and failure times of LSF wall studs subject to non-uniform elevated
7-7
temperature distributions. The LSF wall panels with varying steel grades,
steel thicknesses, screw spacing and plasterboard restraint were considered.
Although there was a slight reduction in the ultimate compression capacity of
LSF wall stud, there was not much difference in failure time when a screw
spacing other than the commonly used 300 mm spacing was used. However,
practically it is not possible to maintain the integrity of the plasterboard when
larger screw spacings are used. The plasterboards may fall off much earlier
when a screw spacing of 600 mm is used compared to 300 mm. Therefore the
failure time will also be considerably less for 600 mm screw spacing
compared to 300 mm screw spacing.
The failure times of LSF wall studs with and without plasterboard restraints
were investigated and it is concluded that the plasterboard restraint can be
assumed to be effective throughout the failure time in routine fire design
calculations. This is particularly true for walls with two or more
plasterboards.
The ultimate load capacity at ambient temperature and the yield stress
reduction pattern at elevated temperatures play a major role in the fire
resistance of LSF walls.
The failure temperature of LSF walls did not depend on the type of
insulation. The effect of using different types of insulation is simply to delay
the time to reach the same hot flange temperatures in the LSF wall stud.
Glass fibre is not recommended in LSF wall panels when it is subjected to
lower load ratios. In this case the studs withstand higher temperatures and the
glass fibre melts and hence it is ineffective.
The superior performance of externally insulated LSF wall panels was
verified for different steel grades and thicknesses.
A new set of equations was proposed to find the failure temperature and the
failure times of LSF wall studs under non-uniform temperature distributions.
The proposed equations for load ratio can be used to find the critical hot
flange (failure) temperature of LSF wall studs with different wall
configurations, steel grades and thickness and load ratios.
The proposed equations can be used to find the failure times of LSF walls
using the proposed idealised time-temperature profile equations.
7-8
The proposed fire design rules based on AS/NZS 4600 (SA 2005), NAS
(AISI, 2007) and Eurocode 3 Part 1.3 (ECS, 2006) accurately predict the
failure times of LSF wall panels under fire conditions with different wall
configurations, steel grades, thickness and load ratios.
7.5. Future Research
This research has addressed the structural and thermal performance of cold-formed
LSF wall systems under fire conditions using an extensive experimental and
numerical study program. However, further research is needed in some areas as
shown next.
Experimental and numerical investigations on the fire performance of both

conventional and externally insulated (composite panel) load-bearing LSF
wall systems under real fire conditions.
Effect of varying stud geometry on the fire performance of both conventional
and externally insulated (composite panel) load-bearing LSF wall systems
using both experimental and numerical studies.
Effect of service holes in LSF wall studs on the fire performance of both
conventional and externally insulated (composite panel) load-bearing LSF
wall systems using both experimental and numerical studies.
Experimental investigations on the fire performance of both conventional and
externally insulated (composite panel) LSF wall systems at higher load ratios
(load ratio of 0.7) to verify the predictive equations developed in this thesis.
Experimental investigations on the fire performance of both conventional and
externally insulated (composite panel) load-bearing LSF wall systems using
different types of plasterboards used in the construction industry.
Investigation on the use of mechanical properties based on 0.5 and 1.5% total
strains for the fire design of cold-formed steel studs.
Development of accurate design rules to predict the compression capacities of
long columns at ambient temperature and suitable modifications to AS/NZS
4600 (SA, 2005) and Eurocode 3 Part 1.3 (ECS, 2006).
Investigation of the behavior and local buckling capacity of shorter columns
subjected to non-uniform elevated temperature distributions
7-9
References
REFERENCES
Alfawakhiri, F. (2001), Behaviour of Cold-formed-Steel-framed Walls and Floors in

Standard Fire Resistance Tests, PhD Thesis. Department of Civil and Environmental
Engineering, Carleton University, Ottawa, Ontario, Canada.
American Society for Testing and Materials (ASTM) (1995), Standard Test Methods
for Fire Tests of Building Construction and Materials, ASTM E119, West
Conshohocken, PA, USA.
American Iron and Steel Institute (AISI) (1991), Specification for the design of cold-
formed steel structural members, Washington, USA.
American Iron and Steel Institute (AISI) (2007), Specifications for the cold-formed
steel structural members, Cold-formed Steel Design Manual, Washington, USA.
Anderberg, Y. (1986), Modelling Steel Behaviour, International Conference on

Design of Structures against Fire, Birmingham, UK.
Belmiloudi, A. and Meur, G.L. (2005), Mathematical and Numerical analysis of

dehydration of gypsum plasterboards exposed to fire, Applied Mathematics and
Computation, Vol. 163, pp. 1023-1041.
British Standards Institute (BSI) (1990), Structural Use of Steelwork in Building, BS

5950, Part 8 Code of Practice for Fire Resistance Design, London, UK.
British Standards Institution (BSI) (1998), Structural Use of Steelwork in Building,

BS 5950, Part 5 Code of Practice for Design of Cold-formed Thin Gauge Sections,
London, UK.
Buchanan, A.H. (2001), Structural Design for Fire Safety, University of Canterbury,
New Zealand, John Wiley & Sons Ltd, England. ISBN 0 47188993 8.
R-1
References
Buchanan, A.H. and Gerlich, J.T. (1997), Fire performance of gypsum plasterboard,
Proceedings of the IPENZ Annual Conference, Wellington Institution of Professional
Engineers, New Zealand.
Canadian Standard Association (CSA) (1994), Cold formed Steel Structural

Members, S136, Etobicoke, ON, Canada.
Chen, J. and Young, B. (2007), Cold-formed steel lipped channel columns at

elevated temperatures, Engineering Structures, Vol. 29, pp. 2445-2456.
Cooke, G.M.E. (1987), Thermal Bowing and how it affects the Design of Fire
Separating Construction, Fire Research Station, Building Research Establishment,
Herts, UK.
Dolamune Kankanamge, N. (2010), Structural Behaviour and Design of Cold-formed

Steel Beams at Elevated Temperatures, PhD Thesis, QUT, Brisbane, Australia.
Dolamune Kankanamge, N. and Mahendran, M. (2011), Mechanical properties of

cold-formed steels at elevated temperatures, Thin-Walled Structures, Vol. 49, pp. 26-
44.
EN 1993-1-1 (2005), Eurocode 3: Design of steel structures. Part 1-1: General Rules,
European Committee for Standardization, Brussels.
EN 1993-1-2 (2005), Eurocode 3: Design of steel structures. Part 1-2: General Rules
- Structural Fire Design, European Committee for Standardization, Brussels.
EN 1993-1-3 (2006), Eurocode 3: Design of Steel Structures. Part 1-3: General Rules
- Supplementary Rules for Cold-formed Members and Sheeting, European
Committee for Standardization, Brussels.
EN 1993-1-5 (2006), Eurocode 3: Design of Steel Structures. Part 1-5: Plated

Structural Elements, European Committee for Standardization, Brussels.
R-2
References
Feng, M., Wang, Y.C. and Davies, J.M. (2001), Behaviour of cold-formed thin-
walled steel short columns under uniform high temperatures, Proceedings of the
International Seminar on Steel Structures in Fire, Shanghai, China.
Feng, M., Wang, Y.C. and Davies, J. M. (2003a), Thermal Performance of Cold-
formed Thin-walled Steel Panel Systems in Fire, Fire Safety Journal, Vol. 38, pp.
365-394.
Feng, M., Wang, Y.C. and Davies, J.M. (2003b), Axial strength of cold-formed thin-
walled steel channels under non-uniform temperatures in fire, Fire Safety Journal,
Vol. 38, pp. 679-707.
Feng, M. and Wang, Y.C. (2005a), An Experimental Study of Loaded Full-Scale

Cold-Formed Thin-Walled Steel Structural Panels Under Fire Conditions, Fire Safety
Journal, Vol. 40, pp. 43-63.
Feng, M. and Wang, Y.C. (2005b), An analysis of the structural behaviour of axially
loaded full-scale cold-formed thin-walled steel structural panels tested under fire
conditions, Thin-Walled Structures, Vol. 43, pp. 291-332.
Fleischmann, C. (1995), Analytical methods for determining fire resistance of

concrete members, SFPE handbook of fire protection engineering, 2nd Ed., P. J.
DiNenno, ed., National Fire Protection Association, Quincy, Mass, 4-202–4-216.
Gerlich, J.T. (1995), Design of Load-bearing Light Steel Frame Walls for Fire
Resistance, Fire Engineering Research Report 95/3, School of Engineering,
University of Canterbury, New Zealand.
Gerlich, J.T., Collier, P.C.R. and Buchanan, A.H. (1996), Design of Steel-framed
Walls for Fire Resistance, Fire and Materials, Vol. 20, No. 2, pp. 79-96.
R-3
References
Heva, Y.B. (2009), Behaviour and design of cold-formed steel compression members
at elevated temperatures, PhD Thesis, Queensland University of Technology,
Brisbane, Australia.
ISO 834 -1, (1999), Fire Resistance Tests – Elements of Building Construction. Part
1: General Requirements, International Organization for Standardization, Geneva,
Switzerland.
Kaitila, O. (2002), Finite Element Modelling of Cold-formed Steel Members at High

Temperatures, Licentiate Thesis, Helsinki University of Technology Laboratory of
Steel Structures, Espoo.
Kirby, B.R. and Preston, R.R. (1988), High temperature properties of hot-rolled,
structural steels for use in fire engineering design studies, Fire Safety Journal, Vol.
13(1), pp. 27-38.
Klippstein, K.H. (1980a), Behaviour of cold-formed steel studs in fire tests,

American Iron and Steel Institute, Washington, DC.
Klippstein, K.H. (1980b), Strength of Cold-Formed Studs Exposed to Fire, American

Iron and Steel Institute. Washington, DC.
Kodur, V.R. and Sultan, M.A. (2001), Factors governing fire resistance of
loadbearing steel stud walls, Proceeding of the Fifth AOSFST International
Conference, Newcastle, Australia, pp. 1-2.
Kolarkar, P. N. and Mahendran, M. (2008), Thermal Performance of Plasterboard

Lined Steel Stud Walls, Proceedings of the 19th International Specialty Conference
on Cold Formed Steel Structures 2008, St. Louis, Missouri, USA, pp. 517-530.
Kolarkar, P.N. (2010), Fire Performance of Plasterboard Lined Steel Stud Walls,
PhD Thesis, Queensland University of Technology, Brisbane, Australia.
R-4
References
Lawson, R.M. (1993), Building design using cold formed steel sections: fire
protection. The Steel Construction Institute, Publication P129.
Lawson, R.M. and Newman, G.M. (1990), Fire Resistant Design of Steel Structures
– A handbook to BS 5950: Part 8, SCI Publication 080, The Steel Construction
Institute, Berkshire, UK.
Lee, J.H., Mahendran, M. and Mäkeläinen, P. (2001), Buckling Behaviour of Thin-

Walled Compression Members at Elevated Temperatures, Proceedings of the 6th
Pacific Structural Steel Conference, Beijing, China, pp. 15-17.
Lee, J.H., Mahendran, M. and Mäkeläinen, P. (2003), Prediction of Mechanical

Properties of Light Gauge Steels at Elevated Temperatures, Journal of Construction
Steel Research, pp. 1517-1532.
Lee, J.H. (2004), Local Buckling Behaviour and Design of Cold-formed Steel
Compression Members at Elevated Temperatures, PhD Thesis, Queensland
University of Technology, Brisbane, Australia.
Lee, Y.K. (1999), Behaviour of Gypsum-sheathed cold-formed steel wall stud

panels, PhD Thesis, Oregon State University, Newport, USA.
Lie, T.T. (1972), Fire and buildings, Applied Science Publishers, London.
Lie, T.T. (1992), Structural Fire Protection, American Society of Civil Engineers,
New York, USA.
Luis, S.S., Rui, S. and Helana, G. (2010), Design of Steel Structures, European
Convention for Constructional Steelwork (ECCS), ISBN: 978-92-9147-098-3
Mecozzi, E. and Zhao, B. (2005), Development of Stress-strain Relationships of

Cold-formed Lightweight Steel at Elevated Temperatures, Proceedings of the
Eurosteel Conference, 5.1-41-5.1-49.
R-5
References
Milke, J. A. (1999), Analytical Methods to Evaluate Fire Resistance of Structural

Members, ASCE Journal of Structural Engineering, Vol. 125, No. 10, pp. 1179-1187.
Nylander, H. (1956), Torsion, Bending and Lateral Buckling of I-Beams,

Transactions of the Royal Institute of Technology, No. 102, Stockholm, Sweden.
Outinen, J. (1999), Mechanical Properties of Structural Steels at Elevated

Temperatures, Licentiate Thesis, Helsinki University of Technology, Finland.
Outinen. J., Kaitila, O. and Mäkeläinen, P. (2001), High-Temperature Testing of

Structural Steel and Modelling of Structures at Fire Temperatures, Research Report,
Helsinki University of Technology Laboratory of Steel Structures Publications TKK-
TER-23, Espoo, Finland.
Pettersson, O. (1986), Structural fire behaviour, Proceedings of the 1st International

Symposium, Fire Safety Science, International Association for Fire Safety Science,
Boston, pp. 228-247.
Pettersson, O., Magnusson, S.E. and Thor, J. (1976), Fire engineering design of steel
structures, Swedish Institute of Steel Construction, Stockholm.
Ramberg, W. and Osgood, W.R. (1943), Description of Stress-strain Curves by

Three Parameters, NACA Technical Note 902.
Ranby, A. (1999), Structural Fire Design of Thin Walled Steel Sections, Licentiate
Thesis, Department of Civil and Mining Engineering, Lulea University of
Technology, Stockholm, Sweden.
Ranawaka, T. (2006), Distortional Buckling Behaviour of Cold-Formed Steel

Compression Members at Elevated Temperatures, PhD Thesis, Queensland
University of Technology, Brisbane, Australia.
R-6
References
Ranawaka, T. and Mahendran, M. (2009), Experimental Study of the Mechanical

Properties of Light Guage Cold-formed Steels at Elevated Temperatures, Fire Safety
Journal, Vol. 44, pp. 219-229.
Schleich, J. B. (1993), International fire engineering design for steel structures: state
of the art, International Iron and Steel Institute, Brussels.
Schwartz, K.J. and Lie, T.T. (1985), Investigating the unexposed surface temperature
criteria of standard ASTM E119, Fire Technology, Vol. 21(3), pp. 169–180.
SCI Publication P129 (1993), Building Design using Cold-formed Steel Sections:
Fire Protection.
Schafer, B. and Pekoz, T. (1998), Direct Strength Prediction Of Cold-Formed Steel

Members Using Numerical Elastic Buckling Solutions, Proceedings, Second
International Conference on Thin-Walled Structures, Thin Walled Structures,
Research and Development, Singapore, pp. 137-144.
Standards Australia (SA) (1998), Steel structures, AS 4100, Sydney, Australia.
Standards Australia (SA) (1997), Gypsum linings in residential and light commercial
construction – application and finishing; Part 1: Gypsum Plasterboard, AS/NZS
2589.1, Sydney, Australia.
Standards Australia (SA) (1998), Gypsum Plasterboard, AS/NZS 2588, Sydney,

Australia.
Standards Australia (SA) (1988), Cold-formed Steel Structures, AS 1538, Sydney,

Australia.
Standards Australia (SA) (2005), Cold-formed steel structures, AS 4600, Sydney,

Australia.
R-7
References
Standards Australia (SA) (2005), Methods for fire tests on building materials,
components and structures, Part 4: Fire-resistance tests of elements of building
construction, AS 1530.4, Sydney, Australia.
StBK-K2 (1983), Plate, Column and Beam-Column Buckling, 2nd edition,

Stockholm, Sweden.
Sultan, M.A. (1996), A Model for Predicting Heat through Non-insulated Unloaded
Steel-Stud Gypsum Board Wall Assemblies Exposed to Fire, Fire Technology, Vol.
32, No. 3, pp. 239-259.
Telue, Y.K. (2001), Behaviour and Design of plasterboard lined cold-formed steel
stud wall systems under axial compression, PhD Thesis, Queensland University of
Technology, Australia.
Thomas, G.C., Buchanan, A.H., Carr, A.J., Fleishmann, C.M. and Moss, P.J. (1994),
Light timber framed walls exposed to compartment fires, Proceedings, Pacific
Timber Engineering Conference, Timber Research and Development Advisory
Council. Queensland, Australia, Vol. 2, pp. 531-538.
Tian, Y.S., Wang, J. and Lu, T.J. (2007), Axial load capacity of cold-formed steel
wall stud with sheeting, Thin-Walled Structures, Vol. 45, pp. 537-551.
Wang, Y.C. (2002), Behaviour and Design for Fire Safety, Spon Press, Taylor and
Francis Group, New York, NY 10001, USA.
Wang, Y.C. and Davies, J.M. (2000), Design of thin-walled steel channel columns in
fire using Eurocode 3 Part 1.3, Proceedings of the First International Workshop on
Structures in Fire, Copenhagen, pp. 181–93.
White, R.H. (1995), Analytical methods for determining fire resistance of timber
members, SFPE handbook of fire protection engineering, 2nd Ed., P. J. DiNenno,
ed., National Fire Protection Association, Quincy, Mass.
R-8
References
Zhao, J.C. (2000), Application of the direct iteration method for non-linear analysis
of steel frames in fire, Fire Safety Journal, Vol. 35, pp. 241-255.
Zhao, B., Kruppa, J., Renaud, C., O’Connor, M., Mecozzi, E., Apiazu,. W.,
Demarco, T., Karlstrom, P., Jumppanen, U., Kaitila, O., Oksanen, T. and Salmi, P.
(2005), Calculation rules of lightweight steel sections in fire situations, Technical
steel research, European Union.
R-9
Appendix A
Appendix A: Effective Area at Ambient Temperature based

on AS/NZS 4600 (SA, 2005)
b External Dimensions
d - 90 mm
c b - 40 mm
c - 15 mm
y
t - 1.15 mm
x
d
Length L = 2400 mm
Mechanical Properties at 20oC

c f y , 20 = 569 MPa
E 20 = 213520 MPa
Section properties of the full section

A = 224.71 mm 2
X cg = 13.46 mm Ycg = 44.425 mm
I x = 290097 mm 4 I y = 54325 mm 4 Using CUFSM

Computer Program
I w = 103958882 mm 6
J = 99.1 mm 4
xo = 33.73 mm yo = 0
Ix 290097 Ix 54333
rx = = = 35.93 mm ry = = = 15.55 mm
A 224.71 A 224.71
ro = rx2 + ry2 + xo2 + y o2 = 35.93 2 + 15.55 2 + 33.73 2 + 0 2 = 51.68 mm
A-1
Appendix A
Critical stress (fn) according to (Cl 3.4)
Sections not subject to torsional or flexural-torsional buckling (Cl 3.4.2)

Elastic flexural buckling stress about the major axis f ox
π 2E π 2 × 213520
f ox = 2
= 2
= 472.313 MPa
 l ex   2400 
   
 rx   35.93 
Elastic flexural buckling stress about the minor axis f oy
π 2E π 2 × 213520
f oy = 2
= 2
= 5661.827 MPa
 l ey   300 
   
r   15.55 
 y 
Elastic flexural buckling stress f oc
f oc = Lesser of f ox and f oy = 472.313 MPa
Non-dimensional slenderness λc (Cl 3.4.1)
fy 569
λc = = = 1.098 λc ≤ 1.5
f oc 472.313
(
Critical stress f n : f n = 0.658 λc f y for λc ≤ 1.5 and f n = 
2
)
 0.877 
 for λc > 1.5
 λ 2  fy
 c 
( ) (
f n = 0.658 λc f y = 0.6581.098 × 569 = 343.530 MPa
2 2
)
Effective widths of uniformly compressed stiffened elements (Cl 2.2.1)

Web
Plate elastic buckling stress f cr
 kπ 2 E  t   4 × π 2 × 213520  1.15 
2 2
f cr =  
2  
=    = 129.317 MPa
 12(1 − ν )  b   12(1 − 0.3 )  88.85 
2
The slenderness ratio λ
f* 343.530
λ= = = 1.630 > 0.673
f cr 129.317
A-2
Appendix A
Effective width factor ρ
 0.22   0.22 
1 −  1 − 
 λ   1.630  = 0.531
ρ= =
λ 1.630
d eff = 0.531 × 88.85 = 47.179 mm
d eff 1 = d eff 2 = 0.5d eff = 0.5 × 47.179 = 23.590 mm
Effective width of uniformly compressed unstiffened elements (Cl 2.3.1)

Lip
 kπ 2 E  t   0.43 × π 2 × 213520  1.15 

2 2
f cr =  2 
  =    = 527.410 MPa
 12(1 − ν )  b   12(1 − 0.3 2 )  14.425 
f* 343.530
λ= = = 0.807 > 0.673
f cr 527.410
 0.22 
1 − 
ρ=  0.807 
= 0.901
0.807
d se = 0.901 × 14.425 = 12.997 mm
Effective widths of uniformly compressed elements with an edge stiffener

(Cl 2.4)
Slenderness factor S
E 213520
S = 1.28 ∗
= 1.28 = 31.911
f 343.530
0.328S = 0.328 × 31.911 = 10.467

b 38.85
= = 33.783
t 1.15
b
> 0.328S Hence Case ( b ) of Cl 2.4.2
t
A-3
Appendix A
3
 (b / t )   (b / t ) 
I a = 399 t 4  − 0 . 328  ≤ t 4 115 + 5
 S   S 
3
 (38.85 / 1.15) 
399 × 1.15 4  − 0.328 = 272.204
 31.911 
 (38.85 / 1.15) 
1.15 4 115 + 5 = 221.678
 31.911 
I a = 221.678 mm 4
d 3 tSin θ
Is = ≤ Ia
12
14 . 425 3 × 1 . 15
= 287 . 650
12
I s = 221.678 mm 4
 (b / t )  1
n = 0.582 − ≥
 4 S  3
 (38.85 / 1.15) 
0.582 − 4 × 31.911  = 0.317
n= 1
3
dl 14 . 425 dl
= = 0 . 371 0 . 25 ≤ ≤ 0 .8
b 38 . 85 b
Plate buckling coefficient k (Table 2.4.2 of AS/NZS 4600)

n
 5d  I s 
k =  4.82 − l   + 0 .43 ≤ 4
 b  I a 
5 × 14 .425   221 .678 
0.333

=  4.82 − ×  + 0.43 ≤ 4
 38 .85   221 .678 
= 3.394

 kπ 2 E
2
 t 
=    
f cr
(
 12 1 − ν
2
)  b 
 3 . 394 × π 2 × 213520
2
  1 . 15 
=   ×   = 573 . 906 MPa
 (
12 × 1 − 0 . 3 2 )   38 . 85 
The Slenderness ratio λ
f∗ 343 . 530
λ = = = 0 . 774 > 0 . 673
f cr 573 . 906
A-4
Appendix A

0 . 22 0 . 22
1− 1−
ρ = λ = 0 . 774 = 0 . 925
λ 0 . 774
b eff = ρ b = 0 . 925 × 38 . 85 = 35 . 936 mm

beff  I s  35 .936  221 .678 
b eff 1 =   = ×  = 17 .968 mm
2  I a  2  221 .678 
b eff 2 = b eff − b eff 1 = 35 .936 − 17 . 968 = 17 .968 mm
I   221 .678 
C eff = d se  s  = 12 .997 ×   = 12 .997 mm
 Ia   221 .678 
beff1 beff2
t ceff
y deff1 23.590
deff1
deff2 23.590
beff1 17.968
x
beff2 17.968
ceff 12.997
deff2 t 1.15
t
ceff
beff1 beff2
Effective area of pure compression member at ambient temperature Aeff

Aeff = (d eff 1 + d eff 2 )t + 2beff 1t + 2beff 2 t + 2ceff t
= (23.590 + 23.590) × 1.15 + 2 × 17.968 × 1.15 + 2 × 17.968 × 1.15 + 2 × 12.997 × 1.15
= 166.803 mm 2
A-5
Appendix B
Appendix B: Effective Area at Elevated Temperature based

on AS/NZS 4600 (SA, 2005)
b
External Dimensions
d - 90 mm
c b - 40 mm
c - 15 mm
t - 1.15 mm
d Length L = 2400 mm
HF = 656°C
Web = 574°C
c CF = 492°C
f y , 656 = 49.844 MPa f y ,574 = 102.950 MPa f y , 492 = 235.710 MPa
f yt = ( f y , 656 (b + c)t + f y ,574 dt + f y , 492 (b + c)t ) / A
 49.844 × (38.85 + 14.425) × 1.15 + 102.950 × 88.85 × 1.15 

=   / 224.71 = 124.667 MPa
 + 235 . 710 × (38 . 85 + 14 . 425) × 1 . 15 
E 656 = 50070.440 MPa E 574 = 73707.104 MPa E 492 = 97343.768 MPa
Et = (E 656 (b + c)t + E574 dt + E 492 (b + c)t ) / A
 50070.440 × (38.85 + 14.425) × 1.15 + 73707.104 × 88.85 × 1.15 

=   / 224.71
 + 97343.768 × (38.85 + 14.425) × 1.15 
= 73707.104 MPa

A = 224.71 mm 2
X cg = 13.46 mm Ycg = 44.425 mm

Computer Program
I w = 103958882 mm 6
J = 99.1 mm 4
xo = 33.73 mm yo = 0
Ix 290097 Ix 54333
rx = = = 35.93 mm ry = = = 15.55 mm
A 224.71 A 224.71
B-1
Appendix B
Critical stress (fn) according to (Cl 3.4)
Sections not subject to torsional or flexural-torsional buckling (Cl 3.4.2)

Elastic flexural buckling stress about the major axis f ox
π 2 Et π 2 × 73707.104
f ox = 2
= 2
= 163.043 MPa
 l ex   2400 
   
 rx   35.93 
Elastic flexural buckling stress about the minor axis f oy
π 2 Et π 2 × 73707.104
f oy = 2
= 2
= 1954.463 MPa
 l ey   300 
   
r   15.55 
 y 
Elastic flexural buckling stress f oc
f oc = Lesser of f ox and f oy = 163.043 MPa
Non-dimensional slenderness λc (Cl 3.4.1)
f yt 124.667
λc = = = 0.874 λc ≤ 1.5
f oc 163.043
2
( )  0.877 
Critical stress f n : f n = 0.658 λc f y for λc ≤ 1.5 and f n = 
λ
 f for λc > 1.5
2  y
 c 
(
f n = 0.658 λc f yt = 0.658 0.874
2
) ( 2
)× 124.667 = 90.552 MPa
Effective widths of uniformly compressed stiffened elements (Cl 2.2.1)

Web
 kπ 2 E web  t   4 × π 2 × 73707.104  1.15 

2 2
f cr = 
 
2  
=    = 44.640 MPa
 12 (1 − ν )  b   12 × (1 − 0 . 3 2
)  88.85 
f* 90.552
λ= = = 1.424 > 0.673
f cr 44.640
B-2
Appendix B
 0.22   0.22 
1 −  1 − 
 λ   1.424  = 0.594
ρ= =
λ 1.424
d eff = 0.594 × 88.85 = 52.777 mm
d eff 1 = d eff 2 = 0.5d eff = 0.5 × 52.777 = 26.388 mm
Effective width of uniformly compressed unstiffened elements

Hot Lip (Cl 2.3.1)
 kπ 2 E656  t   0.43 × π 2 × 50070.440  1.15 

2 2
f cr =  2 
  =    = 123.678 MPa
 12(1 − ν )  b   12 × (1 − 0.3 2 )  14.425 
f* 90.552
λ= = = 0.856 > 0.673
f cr 123.678
 0.22 
1 − 
ρ=  0.856 
= 0.868
0.856
d se = 0.868 × 14.425 = 12.521 mm

Hot Flange (Cl 2.4)
E 50070.440
S = 1.28 ∗
= 1.28 = 30.099
f 90.552
0.328S = 0.328 × 30.099 = 9.872

b 38.85
= = 33.783
t 1.15
b
t
B-3
Appendix B
3
 (b / t )   (b / t ) 
I a = 399 t 4  − 0 . 328  ≤ t 4 115 + 5
 S   S 
3
 (38.85 / 1.15) 
399 × 1.15 4 ×  − 0.328 = 349.828
 30.099 
 (38.85 / 1.15) 
1.15 4 × 115 × + 5 = 234.496
 30.099 
I a = 234.496 mm 4
d 3 tSin θ
Is = ≤ Ia
12
14 . 425 3 × 1 . 15
= 287 . 650
12
I s = 234.496 mm 4
 (b / t )  1
n = 0.582 − ≥
 4 S  3
 (38.85 / 1.15) 
0.582 − 4 × 30.099  = 0.301
n = 0 .333
dl 14 . 425 dl
= = 0 . 371 0 . 25 ≤ ≤ 0 .8
b 38 . 85 b

n
 5d  I s 
k =  4.82 − l   + 0 .43 ≤ 4
 b  I a 
5 × 14 .425   234 .496 
0.333

=  4.82 − ×  + 0.43 ≤ 4
 38 .85   234 .496 
= 3.394

 kπ 2 E
2
 t 
=    
f cr
(
 12 1 − ν
2
)  b 
 3 . 394 × π 2 × 50070 . 440
2
  1 . 15 
=   ×   = 134 . 581 MPa
 (
12 × 1 − 0 . 3 2 )   38 . 85 
f∗ 90 . 552
λ = = = 0 . 820 > 0 . 673
f cr 134 . 581
B-4
Appendix B

0 . 22 0 . 22
1− 1−
ρ = λ = 0 . 820 = 0 . 892
λ 0 . 820
b eff = ρ b = 0 . 892 × 38 . 85 = 34 . 654 mm

beff  I s  34 .654  234 .496 
beff 1 =   = ×  = 17 .327 mm
2  I a  2  234 . 496 
b eff 2 = b eff − b eff 1 = 34 . 654 − 17 . 327 = 17 . 327 mm
I   234 .496 
c eff = d se  s  = 12 .521 ×   = 12 .521 mm
 Ia   234 .496 
Effective width of uniformly compressed unstiffened elements

Cold Lip (Cl 2.3.1)
 kπ 2 E 492  t   0.43 × π 2 × 97343.768  1.15 

2 2
f cr =  
2   =    = 240.446 MPa
 12(1 − ν )  b   12 × (1 − 0.3 2 )  14.425 
f* 90.552
λ= = = 0.614 < 0.673
f cr 240.446
ρ =1
d se = 1 × 14.425 = 14.425 mm

Cold Flange (Cl 2.4)
E 97343.768
S = 1.28 ∗
= 1.28 = 41.968
f 90.552
0.328S = 0.328 × 41.968 = 13.766

b 38.85
= = 33.783
t 1.15
b
t
B-5
Appendix B
3
 (b / t )   (b / t ) 
I a = 399 t 4  − 0 . 328  ≤ t 4 115 + 5
 S   S 
3
 (38.85 / 1.15) 
399 × 1.15 4 ×  − 0.328 = 75.720
 41.968 
 (38.85 / 1.15) 
1.15 4 × 115 × + 5 = 170.651
 41.968 
I a = 75.720 mm 4
d 3 tSin θ
Is = ≤ Ia
12
14 . 425 3 × 1 . 15
= 287 . 650
12
I s = 75.720 mm 4
 (b / t )  1
n = 0.582 − ≥
 4 S  3
 (38.85 / 1.15) 
0.582 − 4 × 41.968  = 0.381
n = 0 .381
dl 14 . 425 dl
= = 0 . 371 0 . 25 ≤ ≤ 0 .8
b 38 . 85 b

n
 5d  I s 
k =  4.82 − l   + 0 .43 ≤ 4
 b  I a 
5 × 14 .425   75 .720 
0.333

=  4.82 − ×  + 0.43 ≤ 4
 38 .85   75 .720 
= 3.394

 kπ 2 E
2
 t 
=    
f cr
(
 12 1 − ν
2
)  b 
 3 . 394 × π 2 × 97343 . 768
2
  1 . 15 
=   ×   = 261 . 644 MPa
 (
12 × 1 − 0 . 3 2 )   38 . 85 

f∗ 90 . 552
λ = = = 0 . 588 < 0 . 673
f cr 261 . 644
Effective width factor ρ = 1
B-6
Appendix B
b eff = ρ b = 1 × 38 . 85 = 38 . 85 mm
beff  I s  38 .85  75 .720 
beff 1 =   = ×  = 19 .425 mm
2  I a  2  75 .720 
b eff 2 = b eff − b eff 1 = 38 . 85 − 19 . 425 = 19 . 425 mm
I   75 .720 
c eff = d se  s  = 14 .425 ×   = 14 .425 mm
 Ia   75 .720 
beff1,cf beff2,cf
t ceff,cf
y
deff1
deff1 26.388 deff2 26.388
beff1,cf 19.425 beff1,hf 17.327

x
beff2,cf 19.425 beff2,hf 17.327
ceff,cf 14.425 ceff,hf 12.521
deff2
t
ceff,hf
beff1,hf beff2,hf
Effective area of pure compression member at elevated temperature Aeff,t

Aeff ,t = (d eff 1 + d eff 2 )t + beff 1,hf t + beff 2,hf t + ceff ,hf t + beff 1,cf t + beff 2,cf t + ceff ,cf t
= (26.388 + 26.388) × 1.15 + 17.327 × 1.15 + 17.327 × 1.15 + 12.521 × 1.15
+ 19.425 × 1.15 + 19.425 × 1.15 + 14.425 × 1.15
= 176.210 mm 2
B-7
Appendix C
Appendix C: Effective Area at Ambient Temperature based

on Eurocode 3 Part 1.3 (ECS, 2006)
b
External Dimensions
d - 90 mm
c
b - 40 mm
c - 15 mm
t - 1.15 mm
d
Length L = 2400 mm
c Mechanical properties at 20oC,

f y , 20 = 569 MPa
E 20 = 213520 MPa
E 20 213520
G20 = = = 82123.1 MPa
2(1 + ν ) 2 × (1 + 0.3)

A = 224.71 mm 2
X cg = 13.460 mm Ycg = 44.425 mm

Computer Program
I w = 103958882 mm 6
J = 99.1 mm 4
xo = 33.73 mm yo = 0
Ix 290097 Ix 54333
rx = = = 35.93 mm ry = = = 15.55 mm
A 224.71 A 224.71
ro = rx2 + ry2 + xo2 + y o2 = 35.93 2 + 15.55 2 + 33.73 2 + 0 2 = 51.68 mm
C-1
Appendix C
Calculation of effective area at ambient temperature

Eurocode 3 Part 1.3 (ECS, 2006) Local Buckling – (Clause 5.5)
Web - doubly supported at both edges.

(Cl. 4.4 & Annex A.1 (2) of Eurocode 3 Part 1.5)
Stress ratioψ = 1 for uniformly compressed internal element

Buckling factor kσ = 4 (Table 4.1 of Eurocode 3 Part 1.5)
(Annex A.1 (2) of Eurocode 3 Part 1.5)

kσ π 2 Et 2
Elastic critical buckling stress σ cr =
(
12 1 − ν 2 b 2 )
(Cl. 4.4 (2) of Eurocode 3 Part 1.5)
_ fy  bp (
 12 1 − ν f y 
2
) 0.5
Plate slenderness λ p = =    
σ cr  t   π Ekσ 
2
=  2
(
 88.85 12 × 1 − 0.3 × 569 
2
) 0.5
= 2.098 > 0.673


 1.15  π × 213520 × 4 
λ p − 0.055(3 + ψ ) 2.098 − 0.055 × (3 + 1)

The reduction factor ρ = 2
= = 0.427 ≤ 1.0
λ p
2.098 2
Effective depth of web d eff = ρd = 0.427 × 88.85 = 37.939 mm
d eff 1 = d eff 2 = 0.5 × d eff = 0.5 × 37.939 = 18.969 mm
C-2
Appendix C
Flange - doubly supported at both edges.

ψ = 1 and kσ = 4 for uniformly compressed internal element as for web
_
λp =
fy  bp
=   
(
 12 1 − ν f y 
2
) 0.5

σ cr  t   π Ekσ 
2
=  2
2
(
 38.85  12 × 1 − 0.3 × 569  ) 0.5
= 0.917 > 0.673


 1.15   π × 213520 × 4 
λ p − 0.055(3 + ψ ) 0.917 − 0.055 × (3 + 1)

ρ= 2
= = 0.829 ≤ 1.0
λ p
0.917 2
Effective width of flange beff = ρb = 0.829 × 38.85 = 32.207 mm
beff 1 = beff 2 = 0.5beff = 0.5 × 32.207 = 16.103 mm
Lip
(Cl. 5.5.3.2 of Eurocode 3 Part 1.3)
b pc 14.425
= = 0.371 > 0.35
bp 38.85
Buckling factor kσ
kσ = 0.5 + 0.83 × 3 (b pc b p − 0.35) = 0.5 + 0.83 × 3 (0.371 − 0.35) = 0.563

2 2
_
λp
 bp
=   
(
 12 1 − ν f y 
2
) 0.5
( )
 14.425   12 1 − 0.3 × 569 
=
2 0.5
= 0.908 > 0.748

  2 
  π Ekσ   1.15   π × 213520 × 0.563 
2
 t
λ p − 0.188 0.908 − 0.188

ρ= 2
= = 0.873 ≤ 1.0
λ p
0.908 2
ceff = ρc = 0.873 × 14.425 = 12.593 mm
C-3
Appendix C
Edge Stiffener
The effective cross-sectional area of the edge stiffener As
As = t (beff 2 + ceff ) = 1.15 × (16.103 + 12.593) = 33.000 mm 2
beff2
ceff × c eff / 2
2
ceff
y= = y
(b eff 2 + ceff ) 2(beff 2 + ceff ) ceff
The effective second moment of area of the edge stiffener Is
2 2
1 1  ceff
2
 c c eff
2

I s = beff 2 t + tceff + beff 2 t
3 3   + c t  eff − 
12 12  2(beff 2 + ceff )  eff  2 2(beff 2 + ceff ) 

2
1 1  12.593 2 
= × 16.103 × 1.15 + × 1.15 × 12.593 + 16.103 × 1.15 × 
3 3

12 12  2 × (16.103 + 12.593) 
2
 12.593 12.593 2 
+ 12.593 × 1.15 ×  −  = 515.613 mm 4
 2 2 × (16.103 + 12.593) 
The spring stiffness per unit length K
Et 3 1
K=
( 2 2 3
)
4 1 − ν b1 hw + b1 + 0.5b1b2 hw k f
where b1 and b2 are the distances from the web-flange junction to the gravity
centre of the effective area of the edge stiffener of both flanges
(b × t × (b − beff 2 / 2 ) + ceff × t × b )
b1 = b2 =
eff 2
As
=
(16.103 × 1.15 × (38.85 − 16.103 / 2) + 12.593 × 1.15 × 38.85) = 34.332 mm
33.000
213520 × 1.15 3 1
K= × = 0.452
(
4 × 1 − 0 .3 2
)
34.332 × 88.85 + 34.332 + 0.5 × 34.332 2 × 88.85 × 1
2 3
The elastic critical buckling stress σ cr , s
2 KEI s 2 × 0.452 × 213520 × 515.613

σ cr , s = = = 427.530 MPa
As 33.000
The relative slenderness λ d
f yb
λd = 0.65 < λ d < 1.38
σ cr , s = = 1.154
569
427.530
The reduction factor χ d
χ d = 1.47 − 0.723λ d = 1.47 − 0.723 × 1.154 = 0.636

Reduced strength χ d f yb = 0.636 × 569 = 361.884 MPa
C-4
Appendix C
Iteration 1
Flange
_
λp =
fy  bp
=   
(
 12 1 − ν f y 
2
) 0.5

σ cr  t   π Ekσ 
2
= 
2
(
 38.85  12 × 1 − 0.3 × 361.884  ) 0.5
= 0.731 > 0.673


 1.15   π × 213520 × 4
2

λ p − 0.055(3 + ψ ) 0.731 − 0.055 × (3 + 1)

ρ= 2
= = 0.956 ≤ 1.0
λ p
0.7312
beff = ρb = 0.956 × 38.85 = 37.141 mm
beff 2 = 0.5beff = 0.5 × 37.141 = 18.570 mm
Lip
_
λp
 bp
=   
(
 12 1 − ν f y 
2
) 0.5
( )
 14.425  12 1 − 0.3 × 361.884 
=
2 0.5
= 0.724 < 0.748

  2 
  π Ekσ   1.15   π × 213520 × 0.563 
2
 t
ρ = 1.0
ceff = ρc = 1 × 14.425 = 14.425 mm
Edge Stiffener
As = t (be 2 + ceff ) = 1.15 × (18.570 + 14.425) = 37.944 mm 2
2
1 1  14.425 2 
I s = × 18.570 × 1.15 3 + × 1.15 × 14.425 3 + 18.570 × 1.15 ×  
12 12  2 × (18.570 + 14.425) 
2
 14.425 14.425 2 
+ 14.425 × 1.15 ×  −  = 775.682 mm 4
 2 2 × (18 . 570 + 14 . 425 ) 
b1 = b2 =
(18.570 × 1.15 × (38.85 − 18.570 / 2) + 14.425 × 1.15 × 38.85) = 33.625 mm
37.944
213520 × 1.15 3 1
K= × = 0.473
(
4 × 1 − 0 .3 2
)
33.625 × 88.85 + 33.625 + 0.5 × 33.625 2 × 88.85 × 1
2 3
2 KEI s 2 × 0.473 × 213520 × 775.682

σ cr , s = = = 466.529 MPa
As 37.944
f yb
λd = 0.65 < λ d < 1.38
σ cr , s = = 1.104
569
466.529
χ d = 1.47 − 0.723λ d = 1.47 − 0.723 × 1.104 = 0.672

Reduced strength χ d f yb = 0.672 × 569 = 382.368 MPa
C-5
Appendix C
Initial
Element Parameters Iteration 1 Iteration 2 Iteration 3
Calculation
fy 569 361.884 382.368 382.937
λp 0.917 0.731 0.752 0.752
Flange
ρ 0.829 0.956 0.941 0.941
beff 2 16.103 18.570 18.279 18.279
λp 0.908 0.724 0.744 0.745
Lip ρ 0.873 1.000 1.000 1.000
ceff 12.593 14.425 14.425 14.425
As 33.000 37.944 37.610 37.610
Is 515.613 775.682 772.288 772.288
b1 = b2 34.332 33.625 33.741 33.741
K 0.452 0.473 0.469 0.469
Stiffener
σ cr , s 427.530 466.529 467.651 467.651
λd 1.154 1.104 1.103 1.103
χd 0.636 0.672 0.673 0.673
t red 0.774
t red = χ d × t = 0.673 × 1.15 = 0.774 mm

beff1 beff2
t tred ceff deff1 18.969

y
deff1 deff2 18.969
beff1 16.103
x beff2 18.279
ceff 14.425
t 1.15
deff2
t tred tred 0.774
ceff
beff1 beff2
Effective area of pure compression member at ambient temperature Aeff,20

Aeff , 20 = (d eff 1 + d eff 2 )t + 2beff 1t + 2beff 2 t red + 2ceff t red
= (18.969 + 18.969) × 1.15 + 2 × 16.103 × 1.15 + 2 × 18.279 × 0.774 + 2 × 14.425 × 0.774
= 131.291 mm 2
C-6
Appendix D
Appendix D: Effective Area at Elevated Temperature based

on Eurocode 3 Part 1.3 (ECS, 2006)
b External Dimensions
d - 90 mm
c b - 40 mm
c - 15 mm
t - 1.15 mm
d
Length L = 2400 mm
HF = 656°C
c
Web = 574°C
CF = 492°C
E 656 = 50070.440 MPa E 574 = 73707.104 MPa E 492 = 97343.768 MPa

A = 224.71 mm 2
X cg = 13.46 mm Ycg = 44.425 mm

Computer Program
I w = 103958882 mm 6
J = 99.1 mm 4
xo = 33.73 mm yo = 0
Ix 290097 Ix 54333
rx = = = 35.93 mm ry = = = 15.55 mm
A 224.71 A 224.71
ro = rx2 + ry2 + xo2 + y o2 = 35.93 2 + 15.55 2 + 33.73 2 + 0 2 = 51.68 mm
D-1
Appendix D
Calculation of effective area at elevated temperature

Eurocode 3 Part 1.3 (ECS, 2006) Local Buckling – (Clause 5.5)
Web - doubly supported at both edges.

Stress ratioψ = 1 for uniformly compressed internal element

Buckling factor kσ = 4 (Table 4.1 of Eurocode 3 Part 1.5)
(Annex A.1 (2) of Eurocode 3 Part 1.5)

kσ π 2 Et 2
Elastic critical buckling stress σ cr =
(
12 1 − ν 2 b 2 )
(Cl. 4.4 (2) of Eurocode 3 Part 1.5)
_ fy  bp (
 12 1 − ν f y 
2
) 0.5
Plate slenderness λ p = =    
σ cr  t   π Ekσ 
2
= 
(
 88.85 12 × 1 − 0.3 × 102.950 
2
) 0.5
= 1.519 > 0.673


 1.15  π × 73707.104 × 4 
2
λ p − 0.055(3 + ψ ) 1.519 − 0.055 × (3 + 1)

The reduction factor ρ = 2
= = 0.563 ≤ 1.0
λp 1.519 2
Effective depth of web d eff = ρd = 0.563 × 88.85 = 50.023 mm
d eff 1 = d eff 2 = 0.5 × d eff = 0.5 × 50.023 = 25.011 mm
D-2
Appendix D
Hot Flange - doubly supported at both edges.

_
λp =
fy  bp
=   
(
 12 1 − ν f y 
2
) 0.5

σ cr  t   π Ekσ 
2
=  2
(
 38.85  12 × 1 − 0.3 × 49.844 
2
) 0.5
= 0.561 < 0.673


 1.15   π × 50070.440 × 4 
ρ = 1.0
Effective width of flange beff = ρb = 1 × 38.85 = 38.85 mm
beff 1,hf = beff 2,hf = 0.5beff = 0.5 × 38.85 = 19.425 mm
Hot Lip
b pc 14.425
= = 0.371 > 0.35
bp 38.85
Buckling factor kσ
kσ = 0.5 + 0.83 × 3 (b pc b p − 0.35) = 0.5 + 0.83 × 3 (0.371 − 0.35) = 0.563

2 2
_
λp
 bp
=   
( 2
)
 12 1 − ν f y 
0.5
( )
 14.425   12 1 − 0.3 × 49.844 
=
2 0.5
= 0.555 < 0.748

  2 
  π Ekσ   1.15   π × 50070.440 × 0.563 
2
 t
ρ =1
ceff ,hf = ρc = 1 × 14.425 = 14.425 mm
D-3
Appendix D
Hot Edge Stiffener

As = t (beff 2 + ceff ) = 1.15 × (19.425 + 14.425) = 38.928 mm 2
beff2
ceff × c eff / 2
2
ceff
y= = y
2 2
1 1  ceff
2
 c c eff
2

I s = beff 2 t 3 + tceff + beff 2 t   + c t  eff − 
3


2
1 1  14.425 2 
= × 19.425 × 1.15 3 + × 1.15 × 14.425 3 + 19.425 × 1.15 ×  
12 12  2 × (19.425 + 14.425) 
2
 14.425 14.425 2 
+ 14.425 × 1.15 ×  −  = 785.320 mm 4
 2 2 × (19 . 425 + 14 . 425 ) 
Et 3 1
K=
( 2 2 3
)
4 1 − ν b1 hw + b1 + 0.5b1b2 hw k f
(b × t × (b − beff 2 / 2 ) + ceff × t × b )
b1 = b2 =
eff 2
As
=
(19.425 × 1.15 × (38.85 − 19.425 / 2) + 14.425 × 1.15 × 38.85) = 33.276 mm
38.928
50070.440 × 1.15 3 1
K= × = 0.113
(
4 × 1 − 0 .3 2
)
33.276 × 88.85 + 33.276 + 0.5 × 33.276 2 × 88.85 × 1
2 3
2 KEI s 2 × 0.113 × 50070.440 × 785.320

σ cr , s = = = 108.298 MPa
As 38.928
f yb
λd = 0.65 < λ d < 1.38
σ cr , s = = 0.678
49.844
108.298
χ d = 1.47 − 0.723λ d = 1.47 − 0.723 × 0.678 = 0.980

t red ,hf = λ d t = 0.980 × 1.15 = 1.127 mm
D-4
Appendix D
Cold Flange - doubly supported at both edges.

_
λp =
fy  bp
=   
(
 12 1 − ν f y 
2
) 0.5

σ cr  t   π Ekσ 
2
= 
2
(
 38.85  12 × 1 − 0.3 × 235.710  ) 0.5
= 0.874 > 0.673


 1.15   π × 97343.768 × 4 
2
λ p − 0.055(3 + ψ ) 0.874 − 0.055 × (3 + 1)

ρ= 2
= = 0.856 ≤ 1.0
λ p
0.874 2
Effective width of flange beff = ρb = 0.856 × 38.85 = 33.256 mm
beff 1,cf = beff 2,cf = 0.5beff = 0.5 × 32.256 = 16.628 mm
Cold Lip
b pc 14.425
= = 0.371 > 0.35
bp 38.85
Buckling factor kσ
kσ = 0.5 + 0.83 × 3 (b pc b p − 0.35) = 0.5 + 0.83 × 3 (0.371 − 0.35) = 0.563

2 2
_
λp
 bp
=   
(
 12 1 − ν f y 
2
) 0.5
( )
 14.425   12 1 − 0.3 × 235.710 
=
2 0.5
= 0.865 > 0.748

  2 
  π Ekσ   1.15   π × 97343.768 × 0.563 
2
 t
λ p − 0.188 0.865 − 0.188

ρ= 2
= = 0.905 ≤ 1.0
λ p
0.865 2
ceff ,cf = ρc = 0.905 × 14.425 = 13.055 mm
D-5
Appendix D
Cold Edge Stiffener

As = t (beff 2 + ceff ) = 1.15 × (16.628 + 13.055) = 34.135 mm 2
beff2
ceff × c eff / 2
2
ceff
y= = y
2 2
1 1  ceff
2
 c c eff
2

I s = beff 2 t + tceff + beff 2 t
3 3   + c t  eff − 

2
1 1  13.055 2 
= × 16.628 × 1.15 + × 1.15 × 13.055 + 16.628 × 1.15 × 
3 3

12 12  2 × (16.628 + 13.055) 
2
 13.055 13.055 2 
+ 13.055 × 1.15 ×  −  = 573.681 mm 4
 2 2 × (16.628 + 13.055) 
Et 3 1
K=
( 2 2 3
)
4 1 − ν b1 hw + b1 + 0.5b1b2 hw k f
(b × t × (b − beff 2 / 2 ) + ceff × t × b )
b1 = b2 =
eff 2
As
=
(16.628 × 1.15 × (38.85 − 16.628 / 2) + 13.055 × 1.15 × 38.85) = 34.193 mm
34.135
97343.768 × 1.15 3 1
K= × = 0.208
4 × 1 − 0 .3(2
)
34.193 × 88.85 + 34.193 + 0.5 × 34.193 2 × 88.85 × 1
2 3
2 KEI s 2 × 0.208 × 97343.768 × 573.681

σ cr , s = = = 199.688 MPa
As 34.135
f yb
λd = 0.65 < λ d < 1.38
σ cr , s = = 1.086
235.710
199.688
χ d = 1.47 − 0.723λ d = 1.47 − 0.723 × 1.086 = 0.685

Reduced strength χ d f yb = 0.685 × 235.710 = 161.461 MPa
D-6
Appendix D
Iteration 1
Flange
_
λp =
fy  bp
=   
(
 12 1 − ν f y 
2
) 0.5
=
( )
 38.85  12 × 1 − 0.3 × 161.461
2 0.5
= 0.724
  
σ cr  t   π Ekσ 
2
 1.15   π × 97343.768 × 4 
2
λ p − 0.055(3 + ψ ) 0.724 − 0.055 × (3 + 1)

ρ= 2
= = 0.962 ≤ 1.0
λ p
0.724 2
beff = ρb = 0.962 × 38.85 = 37.374 mm
beff 2,cf = 0.5beff = 0.5 × 37.374 = 18.687 mm
Lip
_
λp
 bp
=   
(
 12 1 − ν f y 
2
) 0.5
( )
 14.425   12 × 1 − 0.3 × 161.461 
=
2 0.5
= 0.716 < 0.748

  2 
  π Ekσ   1.15   π × 97343.768 × 0.563 
2
 t
ρ = 1.0
ceff ,cf = ρc = 1 × 14.425 = 14.425 mm
Edge Stiffener
As = t (be 2 + ceff ) = 1.15 × (18.687 + 14.425) = 38.079 mm 2

2
1 1  14.425 2 
I s = × 18.687 × 1.15 3 + × 1.15 × 14.425 3 + 18.687 × 1.15 ×  
12 12  2 × (18.687 + 14.425) 
2
 14.425 14.425 2 
+ 14.425 × 1.15 ×  −  = 777.030 mm 4
 2 2 × (18.687 + 14.425) 
b1 = b2 =
(18.687 × 1.15 × (38.85 − 18.687 / 2) + 14.425 × 1.15 × 38.85) = 33.577 mm
38.079
97343.768 × 1.15 3 1
K= × = 0.216
4 × (1 − 0.3 )
2
33.577 × 88.85 + 33.577 + 0.5 × 33.577 2 × 88.85 × 1
2 3
2 KEI s 2 × 0.216 × 97343.768 × 777.030

σ cr , s = = = 212.297 MPa
As 38.079
f yb
λd = 0.65 < λ d < 1.38
σ cr , s = = 1.054
235.710
212.297
χ d = 1.47 − 0.723λ d = 1.47 − 0.723 × 1.054 = 0.708

Reduced strength χ d f yb = 0.708 × 235.710 = 166.883 MPa
D-7
Appendix D
Initial
Element Parameters Iteration 1 Iteration 2 Iteration 3
Calculation
fy 235.710 161.461 166.883 167.118
λp 0.874 0.724 0.736 0.736
Flange
ρ 0.856 0.962 0.953 0.953
beff 2,cf 16.628 18.687 18.512 18.512
λp 0.865 0.716 0.728 0.729
Lip ρ 0.905 1.000 1.000 1.000
ceff ,cf 13.055 14.425 14.425 14.425
As 34.135 38.079 37.878 37.878
Is 573.681 777.030 775.010 775.010
b1 = b2 34.193 33.577 33.647 33.647
K 0.208 0.216 0.215 0.215
Stiffener
σ cr , s 199.688 212.297 212.652 212.652
λd 1.086 1.054 1.053 1.053
χd 0.685 0.708 0.709 0.709
t red ,cf 0.815
t red ,cf = χ d × t = 0.709 × 1.15 = 0.815 mm
beff1,cf beff2,cf
t tred,cf ceff,cf
y deff1 25.011 deff2 25.011
deff1
beff1,hf 19.425 beff2,hf 19.425
x ceff,hf 14.425 tred,hf 1.127
beff1,cf 16.628 beff2,cf 18.512
ceff,cf 14.425 tred,cf 0.815

deff2
t tred,hf ceff,hf
beff1,hf beff2,hf
Effective area of pure compression member at elevated temperature Aeff,t

Aeff ,t = (d eff 1 + d eff 2 )t + beff 1,hf t + beff 2, hf t red , hf + ceff ,hf t red ,hf + beff 1,cf t + beff 2,cf t red ,cf + ceff ,cf t red ,cf
= (25.011 + 25.011) × 1.15 + 19.425 × 1.15 + 19.425 × 1.127 + 14.425 × 1.127
+ 16.628 × 1.15 + 18.512 × 0.815 + 14.425 × 0.815
= 163.979 mm 2
D-8
Appendix E
Appendix E: Klippstein (1980)

b
External Dimensions
d - 90 mm
c b - 40 mm
c - 15 mm
t - 1.15 mm
HF = 656°C
Web = 574°C
c CF = 492°C
 49.844 × (38.85 + 14.425) × 1.15 + 102.950 × 88.85 × 1.15 

=   / 224.71 = 124.667 MPa
 + 235.710 × (38.85 + 14.425) × 1.15 
E 656 = 50070.440 MPa E 574 = 73707.104 MPa E 492 = 97343.768 MPa
Et = (E 656 (b + c)t + E574 dt + E 492 (b + c)t ) / A
 50070.440 × (38.85 + 14.425) × 1.15 + 73707.104 × 88.85 × 1.15 

=   / 224.71
 + 97343.768 × (38.85 + 14.425) × 1.15 
= 73707.104 MPa

A = 224.71 mm 2
X cg = 13.46 mm Ycg = 44.425 mm

Computer Program
I w = 103958882 mm 6
J = 99.1 mm 4
xo = 33.73 mm yo = 0
Ix 290097 Ix 54333
rx = = = 35.93 mm ry = = = 15.55 mm
A 224.71 A 224.71
E-1
Appendix E
Effective area at elevated temperature for pure compression (see Appendix B)
beff1,cf beff2,cf
t ceff,cf
deff1 26.388 deff2 26.388
y
deff1 beff1,cf 19.425 beff1,hf 17.327
beff2,cf 19.425 beff2,hf 17.327

x
deff2 Aeff 176.210 fn 90.552

t
ceff,hf
beff1,hf beff2,hf
A eff , t 176 . 210

QT = = = 0 . 784
A 224 . 71
F alt =
12
Q T f yt −
3 Q T f yt ( )
2
 kL 
 
2
23 23 π 2 E t  r 
3 (0.784 × 124.667 )  1 × 2400 
2 2
12
= × 0.784 × 124.667 − × × 
23 23 π 2 × 73707.104  35.930 
= 43.352 MPa
Ix 290097
Sx = = = 6530 mm 3
y max 88 . 85
2
( )
Assuming a lateral deflection of 20 mm
1
PT =
1 e
+
 23  S x f yt
A F alt 
 12 
1
=
1 20
+
 23  6530 × 124 .667
224 .71 ×  × 43 .352 
 12 
= 12800 N
Ultimate Capacity of LSF Stud under Fire Conditions = 12.8 kN
E-2
Appendix F
Appendix F: Gerlich et al. (1996)

b
External Dimensions
d - 90 mm
c b - 40 mm
c - 15 mm
t - 1.15 mm
HF = 656°C
Web = 574°C
c CF = 492°C
 49.844 × (38.85 + 14.425) × 1.15 + 102.950 × 88.85 × 1.15 

=   / 224.71 = 124.667 MPa
 + 235.710 × (38.85 + 14.425) × 1.15 
E 656 = 50070.440 MPa E 574 = 73707.104 MPa E 492 = 97343.768 MPa
Et = (E 656 (b + c)t + E574 dt + E 492 (b + c)t ) / A
 50070.440 × (38.85 + 14.425) × 1.15 + 73707.104 × 88.85 × 1.15 

=   / 224.71
 + 97343.768 × (38.85 + 14.425) × 1.15 
= 73707.104 MPa

A = 224.71 mm 2
X cg = 13.46 mm Ycg = 44.425 mm

Computer Program
I w = 103958882 mm 6
J = 99.1 mm 4
xo = 33.73 mm yo = 0
Ix 290097 Ix 54333
rx = = = 35.93 mm ry = = = 15.55 mm
A 224.71 A 224.71
F-1
Appendix F
Effective area at ambient temperature for pure compression (see Appendix A)
beff1 beff2
deff1 23.590
t ceff
y deff2 23.590
deff1
beff1 17.968
beff2 17.968
x
ceff 12.997
t 1.15
deff2 Aeff 166.803

t
ceff
beff1 beff2
Thermal expansion coefficient α T

α T = 1.2 × 10 −5 + 0.4 × 10 −8 × T − (2.416 × 10 −4 / T )
= 1.2 × 10 −5 + 0.4 × 10 −8 × 574 − (2.416 × 10 −4 / 574) = 1.388 × 10 −5
Thermal bowing e∆T

α T (HF − CF )L2 1.388 × 10 −5 × (656 − 492 ) × 2400 2
e ∆T = = = 18.446 mm
8d 8 × 88.85
Ix 290097
Sx = = = 6530 mm 3
y max 88.85
2
( )
f oc = 163.043 MPa (see Appendix B)
f y ,CF 235.710
λc = = = 1.202
f oc 163.043
( ) (
f n = 0.658 λc f y = 0.6581.202 × 235.710 = 128.751 MPa
2 2
)
Pa Ma
+ =1
Pn M nx
Pa Pa e
+ =1
f n A eff , 20 S x f y ,CF
1
Pa =
1 e
+
f n A eff , 20 S x f y , CF
F-2
Appendix F
Initial Step
1
Pa = = 17080 N
1 18.446
+
128.751 × 166.803 6530 × 235.710
P 17080
µ= = = 8.938 × 10 − 4
ET I x 73707.104 × 290097
   
   
 
+ e∆T  − 1 = 18.446 + 18.446 × 
1 1
e = e ∆T − 1
  µL   −
 cos 8.938 × 10 × 2400  
4
 cos     
  2    2
   
= 38.600 mm
Iteration 1
1
Pa = = 13958 N
1 38 . 600
+
128 . 751 × 166 . 803 6530 × 235 . 710
Initial Iteration 1 Iteration 2 Iteration 3 Iteration 4 Iteration 5

Calculations
Pa 17080 13958 14760 14569 14615 14604
µ 8.938x10 -4
8.080x10 -4
8.308x10 -4
8.254x10 -4
8.267x10 -4
8.264x10-4
e 38.6 32.611 33.979 33.643 33.723 33.705
F-3
Appendix G
Appendix G: Kaitila (2002)
External Dimensions
b d - 90 mm
b - 40 mm
c c - 15 mm
t - 1.15 mm
Length L = 2400 mm
d
HF = 656°C
Web = 574°C
c CF = 492°C
E 656 = 50070.440 MPa E 574 = 73707.104 MPa E 492 = 97343.768 MPa

A = 224.71 mm 2
X cg = 13.46 mm Ycg = 44.425 mm

Computer Program
I w = 103958882 mm 6
J = 99.1 mm 4
xo = 33.73 mm yo = 0
Ix 290097 Ix 54333
rx = = = 35.93 mm ry = = = 15.55 mm
A 224.71 A 224.71
ro = rx2 + ry2 + xo2 + y o2 = 35.93 2 + 15.55 2 + 33.73 2 + 0 2 = 51.68 mm
G-1
Appendix G
Effective area at ambient temperature for pure compression (see Appendix C)

beff1 beff2
deff1 18.969
t tred ceff
deff2 18.969
y
deff2 beff1 16.103
beff2 18.279
x
ceff 14.425
t 1.15
deff1 tred 0.774

t tred ceff Aeff 131.291
beff1 beff2

α T = 1.2 × 10 −5 + 0.4 × 10 −8 × T − (2.416 × 10 −4 / T )
= 1.2 × 10 −5 + 0.4 × 10 −8 × 574 − (2.416 × 10 −4 / 574) = 1.388 × 10 −5
α T (HF − CF )L2 1.388 × 10 −5 × (656 − 492 ) × 2400 2

e ∆T = = = 18.446 mm
8d 8 × 88.85
Neutral Axis Shift e∆E

Distance from New Neutral axis to cold flange yCF
d2  t
ECF (b + c ) + E web + E HF (b + c ) d − 
t
y CF =
2 2  2
ECF (b + c ) + E web d + E HF (b + c )
97343.768 × (38.85 + 14.425) × 1.15 / 2 + 73707.104 × 88.85 2 / 2
+ 50070.44 × (38.85 + 14.425) × (88.85 − 1.15 / 2)
=
97343.768 × (38.85 + 14.425) + 73707.104 × 88.85 + 50070.44 × (38.85 + 14.425)
= 36.757 mm
d 88.85
e ∆E = − y CF = − 36.757 = 7.668 mm
2 2
Deflection due to thermal bowing and neutral axis shift

e∆T − e∆E = 18.446 − 7.668 = 10.778 mm
G-2
Appendix G
The ultimate load for local buckling N eff
N eff = Aeff f y , web = 131.291 × 102.950 = 13516.408 N
The ultimate failure load for flexural buckling N b , fi ,t , Rd at time t
l x 2400 ly 300
λx = = = 66.797 λy = = = 19.293
rx 35.93 ry 15.55
λ = maximum of λ x and λ y = 66.797
0.5
E   73707.104 
0.5
λ 1= π ×  web  = π ×  = 84.079
f   102.905 
 yweb 
0.5
λ  Aeff  0.5
66.797  131.291 
λ = ×  = ×  = 0.607
λ1  Ag 
 84.060  224.71 
( ( ) )
φ = 0.5 × 1 + α × λ − 0.2 + λ = 0.5 × (1 + 0.34 × (0.607 − 0.2) + 0.607 2 ) = 0.753
2
1 1
χ= = = 0.834
φ + φ −λ 2 2
0.753 + 0.753 − 0.6072 2
N b , fi ,t , Rd = χN eff = 0.834 × 13516.408 = 11272.684 N
Bending Moment Capacity M x ,eff
I eff , 20 = 2 × (1 / 12)t red ceff

3
+ 2 × (1 / 12)beff 2 t red
3
(
+ 2 × (1 / 12)beff 1t 3 + (1 / 12)t d eff
3
1 + d eff 2
3
)
+ 2 × C eff t red (d / 2 − ceff / 2 ) + 2 × beff 2 t red (d / 2 ) + 2 × beff 1t (d / 2 )
2 2 2
+ 2 × d eff 1t (d / 2 − d eff 1 / 2 )
2
= 2 × (1 / 12) × 0.774 × 14.425 3 + 2 × (1 / 12) × 18.279 × 0.774 3

+ 2 × (1 / 12) × 16.103 × 1.15 3 + (1 / 12) × 1.15 × 18.969 3 + 18.969 3 ( )
+ 2 × 14.425 × 0.774 × (88.85 / 2 − 14.425 / 2 )
2
+ 2 × 18.279 × 0.774 × (88.85 / 2 ) + 2 × 16.103 × 1.15 × (88.85 / 2 )

2 2
+ 2 × 18.969 × 1.15 × (88.85 / 2 − 18.969 / 2 ) = 214826 mm 4

2
I eff , 20 f y , web 214826 × 102.950

M x ,eff = = = 424554.867 Nmm
(d − y CF ) (88.85 − 36.757)
G-3
Appendix G
Euler Buckling Load N cr ,T , y
∑ ET I = (1 / 12)td 3 E web + (1 / 12)tc 3 (E HF + E CF ) + (1 / 12)bt 3 (E HF + E CF )

( ) 2
( )
2
(
+ dt d / 2 − y CF Eweb + ct y CF − c / 2 E CF + ct d − y CF − c / 2 E HF )
2
( )E
2 2
+ bt y CF E CF + bt d − y CF HF
= (1 / 12) × 1.15 × 88.85 3 × 73707.104 + (1 / 12) × 1.15 × 14.425 3 (50070.440 + 97343.768)

+ (1 / 12) × 38.85 × 1.15 3 × (50070.440 + 97343.768)
+ 88.85 ×1.15 × (88.85 / 2 − 36.757 ) × 73707.104
2
+ 14.425 × 1.15(36.757 − 14.425 / 2 ) × 97343.768

2
+ 14.425 × 1.15 × (88.85 − 36.757 − 14.425 / 2 ) × 50070.440

2
+ 38.85 × 1.15 × 36.757 2 × 97343.768 + 38.85 × 1.15 × (88.85 − 36.757 ) × 50070.440

2
= 2.047 × 1010 Nmm 2
π 2 ∑ ET I π 2 × 2.047 × 1010
N cr ,T , y = = = 35074.792 N
L2 2400 2
G-4
Appendix G
Interaction of Compression and Bending
Assuming N ∗ = 0 for initial step

(e∆T − e∆E ) 1
= 10.778 ×
1
= 10.778 mm
 N ∗   0 
1 −  1 − 
 N   35074.792 
 cr ,T , y 
µ x = λθ (2 β M , x − 4) = 0.607 × (2 × 1.3 − 4) = −0.850

µx N * (−0.850) × 0
kx = 1− = 1− =1
χN eff 0.834 × 13516.408
N* k N *e
+ x =1
χN eff M x ,eff
N* 1 × N * × 10.778
+ =1
0.834 × 13516.408 424554.867
N * = 8765 N
Iteration 1
e = (e∆T − e∆E )
1 1
= 10.778 × = 14.369 mm
 
∗  8765 
1 − N  1 − 
 N   35074.792 
 cr ,T , y 
µ N* (−0.850) × 8765
kx = 1− x = 1− = 1.661 > 1.5 k x = 1 .5
χN eff 0.834 × 13516.408
N* k x N *e
+ =1
χN eff M x ,eff , p
N* 1.5 × N * × 14.369
+ =1
0.834 × 13516.408 424554.867
N * = 7170 N
Initial
Iteration 1 Iteration 2 Iteration 3 Iteration 4 Iteration 5
Calculations
e 10.778 14.369 13.547 13.622 13.615 13.615
kx 1 1.5 1.5 1.5 1.5 1.5
N* 8765 7170 7322 7308 7309 7309
G-5
Appendix G
The ultimate failure load for flexural buckling with Uniform Temperature
N b , fi ,t , Rd at time t
T = 656 o C
f y , 656 = 49.844 MPa
E 656 = 50070.440 MPa
Aeff , 20 = 131.291 mm 2
l x 2400 ly 300
λx = = = 66.797 λy = = = 19.293
rx 35.93 ry 15.55

0.5
E   50070.440 
0.5
λ 1= π ×  web  = π ×  = 99.571
f   49.844 
 yweb 
0.5
λ  Aeff  0.5
66.797  131.291 
λ = ×  = ×  = 0.513
λ1  Ag 
 99.571  224.7 
( ( ) )
φ = 0.5 × 1 + α × λ − 0.2 + λ = 0.5 × (1 + 0.34 × (0.513 − 0.2) + 0.513 2 ) = 0.685
2
1 1
χ= = = 0.878
φ + φ2 −λ
2
0.685 + 0.685 2 − 0.513 2
N b , fi ,t , Rd = χf y , 656 Aeff , 20 = 0.878 × 49.844 × 131.291 = 5746 N
Here, the LSF wall stud is assumed to be under uniform temperature distributions
and hence the bending effects are not considered.
G-6
Appendix H
Appendix H: Ranby (1999)

All the parameters are obtained from Appendix C
I eff , 20 f y , web 214826 × 102.950

M x ,eff = = = 601690.473 Nmm
y CF 36.757

e = 10.778 mm
µ x = −0.850
kx = 1
N* k x N *e N* 1 × N * × 10.778
+ = + =1
χN eff M x ,eff , p 0.834 × 13516.408 601690.473
N * = 9379 N
Iteration 1
e = (e∆T − e∆E )
1 1
= 10.778 × = 14.712 mm
 ∗   9379 
1 − N  1 − 
 N   35074.792 
 cr ,T , y 
µ N* (−0.850) × 9379
kx = 1− x = 1− = 1.707 > 1.5 k x = 1 .5
χN eff 0.834 × 13516.408
N* k x N *e N* 1.5 × N * × 14.712
+ = + =1
χN eff M x ,eff , p 0.834 × 13516.408 601690.473
N * = 7975 N
Initial
Calculations
e 10.778 14.712 13.950 14.013 14.008 14.009
kx 1 1.5 1.5 1.5 1.5 1.5
N* 9379 7975 8098 8088 8089 8088
H-1
Appendix I
Appendix I: Feng and Wang (2005) - EC3 Part 1.2 Method

b
External Dimensions
d - 90 mm
c b - 40 mm
c - 15 mm
t - 1.15 mm
HF = 656°C
Web = 574°C
c CF = 492°C
 49.844 × (38.85 + 14.425) × 1.15 + 102.950 × 88.85 × 1.15 

=   / 224.71 = 124.667 MPa
 + 235.710 × (38.85 + 14.425) × 1.15 
E 656 = 50070.440 MPa E 574 = 73707.104 MPa E 492 = 97343.768 MPa
Et = (E 656 (b + c)t + E574 dt + E 492 (b + c)t ) / A
 50070.440 × (38.85 + 14.425) × 1.15 + 73707.104 × 88.85 × 1.15 

=   / 224.71
 + 97343.768 × (38.85 + 14.425) × 1.15 
= 73707.104 MPa

A = 224.71 mm 2
X cg = 13.46 mm Ycg = 44.425 mm

Computer Program
I w = 103958882 mm 6
J = 99.1 mm 4
xo = 33.73 mm yo = 0
Ix 290097 Ix 54333
rx = = = 35.93 mm ry = = = 15.55 mm
A 224.71 A 224.71
I-1
Appendix I
Effective area at ambient temperature for pure compression (see Appendix C)
beff1 beff2
deff1 18.969
t tred ceff
y deff2 18.969
deff2
beff1 16.103
beff2 18.279
x
ceff 14.425
t 1.15
deff1 tred 0.774
t tred ceff
Aeff,20 131.291
beff1 beff2

α T = 1.2 × 10 −5 + 0.4 × 10 −8 × T − (2.416 × 10 −4 / T )
= 1.2 × 10 −5 + 0.4 × 10 −8 × 574 − (2.416 × 10 −4 / 574) = 1.388 × 10 −5
α T (HF − CF )L2 1.388 × 10 −5 × (656 − 492 ) × 2400 2

e ∆T = = = 18.446 mm
8d 8 × 88.85
Neutral axis shift about the major axis e∆E
Aeff = {( E 656 + E 492 )(beff 1t + beff 2 t red + ceff t red ) + E574 t (d eff 1 + d eff 2 )} / Et
= {(50070.440 + 97343.768) × (16.103 × 1.15 + 18.279 × 0.774 + 14.425 × 0.774)
+ 73707.104 × 1.15 × (18.969 + 18.969)} / 73707.104
= 131.291 mm 2
Yeff = {E 492 ceff t red (d − ceff / 2) + E 492 beff 2 t red d + E 492 beff 1td + E574 d eff 2 t (d − d eff 2 / 2)
+ E574 d eff 1td eff 1 / 2) + E656ceff tred ceff / 2} /{ Aeff × Et }
= {97343.768 × 14.425 × 0.774 × (88.85 − 14.425 / 2) + 97343.768 × 18.279 × 0.774 × 88.85

+ 97343.768 × 16.103 × 1.15 × 88.85 + 73707.104 × 18.969 × 1.15 × (88.85 − 18.969 / 2)
+ 73707.104 × 18.969 × 1.15 × 18.969 / 2 + 50070.440 × 14.425 × 0.774 × 14.425 / 2}
/{73707.104 × 131.291}
= 53.544 mm
e∆E = Yeff − d / 2 = 53.544 − 88.85 / 2 = 9.119 mm
I-2
Appendix I
N eff = Aeff , 20 f yt = 131.291 × 124.667 = 16367.655 N
l x 2400
λx = = = 66.797
rx 35.93
ly 300
λy = = = 19.293
ry 15.55
0.5
E  213520 
0.5
λ1 = π   = π ×  = 60.857
f   569 
 y
0.5
λ  Aeff 
0.5
66.797  131.291 
λ =   = ×  = 0.839
λ1  A  60.857  224.71 
The non-dimensional slenderness λθ for the temperature θ (Cl 4.2.3.2)

0.5
 k y ,θ   124.667 / 569 
0.5

λ θ = λ 
 = 0.839 ×  73707.104 / 213520  = 0.668
 k E ,θ 
235 235
α = 0.65 = 0.65 × = 0.418
fy 569
( )
φθ = 0.5 1 + α λ θ + λ θ 2 = 0.5 × (1 + 0.418 × 0.668 + 0.668 2 ) = 0.863
The reduction factor for flexural buckling in the fire situation χ fi

1 1
χ fi = = = 0.710
φθ + φθ 2 − λθ
2
0.863 + 0.863 2 − 0.668 2
N b, fi ,t , Rd = χ fi N eff = 0.710 × 16367.655 = 11621.035 N
I-3
Appendix I
Bending Moment Capacity M x ,eff ,1

Case 1: At Mid-height :- Hot Flange in Tension
Effective width of web

ψ = −1 kσ = 23.9
_
λp =
fy  bp
=   
( 2
)
 12 1 − ν f y 
0.5
( )
 88.85  12 × 1 − 0.3 × 569 
2 0.5
 = ×  2  = 0.858
σ cr  t   π Ekσ 
2
 1.15   π × 213520 × 23.9 
λ p − 0.055(3 + ψ ) 0.858 − 0.055 × (3 − 1)

ρ= 2
= = 1.016 > 1.0
λ p
0.858 2
• Web is fully effective in this case

• Effective flange and lip widths on the cold side are the same as under uniform
compression (see Appendix C)
• The flange and the lip on the hot side will not buckle because they are on the
tension side
beff1 beff2
Cold Flange
t tred ceff
y
d 88.85 b 38.85
x c 14.425 t 1.15
d
beff1 16.103 beff2 18.279
ceff 14.425 tred 0.774
t t
c
b Hot Flange
Aeff = {E 656 t (b + c) + E574 td + E 492 (tbeff 1 + t red beff 2 + t red ceff )} / Et

= {50070.440 × 1.15 × (38.85 + 14.425) + 73707.104 × 1.15 × 88.85
+ 97343.768 × (1.15 × 16.103 + 0.774 × 18.279 + 0.774 × 14.425)} / 73707.104
= 201.684 mm 2
Yeff = {E 492 ceff t red (d − ceff / 2) + E 492 beff 2 t red d + E 492 beff 1td + E 574 dtd / 2)
+ E 656 ctc / 2} /{ Aeff Et }
= {97343.768 × 14.425 × 0.774 × (88.85 − 14.425 / 2) + 97343.768 × 18.279 × 0.774 × 88.85

+ 97343.768 × 16.103 × 1.15 × 88.85 + 73707.104 × 88.85 × 1.15 × 88.85 / 2
+ 50070.440 × 14.425 × 1.15 × 14.425 / 2} /{73707.104 × 201.684}
= 47.884 mm
I-4
Appendix I
f y , 492 f y , 492
f y , 492 beff1 beff2 f y , 492
t tred ceff
y f y , 492 (d − Yeff − c)
E 519
E 492 (d − Yeff )
x’
d x
f yp
f y , 629
Y eff
t t
c
f y , 656 b f y , 656
f y , 656 f y , 656
At Point P the distance from Hot flange Y p = 31.1 mm ,

(656 − 492)
T p = 656 − × 31.1 = 599 o C
88.85
f yp = 64.616 MPa E p = 66500.804 MPa
f y , 492 (Y − Yp )
Stress at P = E p eff
E492 (d − Yeff )
235.710 (47.884 − 31.1)
= × 66500.804 × = 65.973MPa
97343.768 (88.85 − 47.884)
≈ f yp
M x ,eff ,1 = ct × f 656 (Yeff − c / 2) + ct × 0.5( f 629 − f 656 )(Yeff − 2c / 3)

+ bt × f 656 Yeff + Y p t × f 656 (Yeff − Y p / 2)
+ Ypt × 0.5( f yp − f 656 )(Yeff − 2Yp / 3) + (Yeff − Yp )t × 0.5 f yp × (Yeff − Yp )2 / 3
+ (d − Yeff )t × 0.5 f 492 × (d − Yeff )2 / 3 + (beff 1t + beff 2 t red ) f 492 × (d − Yeff )
f 492 (d − Yeff − c)
+ ceff tred { E519 }( d − Yeff − ceff / 2)
E492 (d − Yeff )
f 492 (d − Yeff − ceff )
+ ceff tred 0.5{ f 492 − E519 }(d − Yeff − ceff / 3)
E492 (d − Yeff )
I-5
Appendix I
= 14.425 × 1.15 × 49.844 × (47.884 − 14.425 / 2)

+ 14.425 × 1.15 × 0.5 × (55.990 − 49.844)(47.884 − 2 × 14.425 / 3)
+ 38.85 × 1.15 × 49.844 × 47.884 + 31.1 × 1.15 × 49.844 × (47.884 − 31.1 / 2)
+ 31.1 × 1.15 × 0.5 × (64.616 − 49.844)(47.884 − 2 × 31.1 / 3)
+ (47.884 − 31.1) × 1.15 × 0.5 × 64.616 × (47.884 − 31.1)2 / 3
+ (88.85 − 47.884) × 1.15 × 0.5 × 235.710 × (88.85 − 47.884)2 / 3
+ (16.103 × 1.15 + 18.279 × 0.774) × 235.710 × (88.85 − 47.884)
235.710 (88.85 − 47.884 − 14.425)
+ 14.425 × 0.774 × { × 89560.964}
97343.768 (88.85 − 47.884)
× (88.85 − 47.884 − 14.425 / 2)
235.710 (88.85 − 47.884 − 14.425)
+ 14.425 × 0.774 × 0.5 × {235.710 − × 89560.964}
97343.768 (88.85 − 47.884)
× (88.85 − 47.884 − 14.425 / 3)
= 753237 Nmm
Bending Moment Capacity M x ,eff , 2

Case 2: At Support :- Cold Flange in Tension
• Effective width of web is similar to Case 1 (fully effective in this case)

• Effective flange and lip widths on the hot side are the same as under uniform
compression
• The flange and the lip on the cold side will not buckle because they are on the
tension side
b
Cold Flange
t t c
y
d 88.85 b 38.85
c 14.425 t 1.15
d x
beff1 16.103 beff2 18.279
ceff 14.425 tred 0.774
t tred
ceff
beff1 beff2 Hot Flange
I-6
Appendix I
= {97343.768 × 1.15 × (38.85 + 14.425) + 73707.104 × 1.15 × 88.85

+ 50070.440 × (1.15 × 16.103 + 0.774 × 18.279 + 0.774 × 14.425)} / 73707.104
= 212.866 mm 2
Yeff = {E 492 ct (d − c / 2) + E 492 btd + E 574 dtd / 2 + E 656 ceff t red ceff / 2} /{ Aeff E t }
= {97343.768 × 14.425 × 1.15 × (88.85 − 14.425 / 2)

+ 97343.768 × 38.85 × 1.15 × 88.85 + 73707.104 × 88.85 × 1.15 × 88.85 / 2
+ 50070.440 × 14.425 × 0.774 × 14.425 / 2} /{73707.104 × 212.866}
= 54.612 mm
f y , 656 (d − Yeff ) f y , 656 (d − Yeff )

E 492 E 492
E 656 Yeff E 656 Yeff
t t c
y f y , 656 (d − Yeff − c)
E 519
E 656 Yeff
x’
d x
f y , 656 c
E 629
E656 Yeff
Y eff
t tred
ceff
f y , 656 beff1 beff2 f y , 656
f y , 656 f y , 656
f y , 656 ceff
M x ,eff , 2 = ceff t red × 0.5( f y ,656 − E 629 )(Yeff − ceff / 3)
E 656 Y eff
f y , 656 ceff
+ ceff t red × ( E 629 )(Yeff − ceff / 2) + (beff 1t + beff 2 t red ) f y , 656 Yeff
E 656 Y eff
f y , 656 (d − Y eff )
+ Yeff t × 0.5 f 656 (2Yeff / 3) + (d − Yeff )t × 0.5 × E 492 × 2(d − Yeff ) / 3
E 656 Y eff
f y , 656 (d − Yeff ) f y , 656 (d − Yeff − c)
+ bt E 492 (d − Yeff ) + ct E 519 (d − Yeff − c / 2)
f y , 656 (d − Yeff ) f y ,656 (d − Yeff − c)
+ ct × 0.5{ E 492 − E519 }(d − Yeff − c / 3)
I-7
Appendix I
49.844 14.425
= 14.425 × 0.774 × 0.5(49.844 − × × 57853.244)(54.612 − 14.425 / 3)
50070.44 54.612
49.844 14.425
+ 14.425 × 0.774 × ( × × 57853.244)(54.612 − 14.425 / 2)
50070.44 54.612
+ (16.103 × 1.15 + 18.279 × 0.774) × 49.844 × 54.612 + 54.612 × 1.15 × 0.5 × 49.844 × (2 × 54.612 / 3)
49.844 (88.85 − 54.612)
+ (88.85 − 54.612) × 1.15 × 0.5 × × × 97343.768 × 2 × (88.85 − 54.612) / 3
50070.44 54.612
49.844 (88.85 − 54.612)
+ 38.85 × 1.15 × 97343.768 × (88.85 − 54.612)
50070.44 54.612
49.844 (88.85 − 54.612 − 14.425)
+ 14.425 × 1.15 × × × 89560.964 × (88.85 − 54.612 − 14.425 / 2)
50070.44 54.612
49.844 (88.85 − 54.612)
+ 14.425 × 1.15 × 0.5{ × × 97343.768
50070.44 54.612
49.844 (88.85 − 54.612 − 14.425)
− × × 89560.964} × (88.85 − 54.612 − 14.425 / 3)
50070.44 54.612
= 305250 Nmm
Y = {E 492 ct (d − c / 2) + E 492 btd + E w1 (d / 4)t (7 d / 8) + E w 2 (d / 4)t (5d / 8) + E w3 (d / 4)t (3d / 8)

+ E w 4 (d / 4)t (d / 8) + E 656 ctc / 2} /{ A Et }
= {97343.768 × 14.425 × 1.15 × (88.85 − 14.425 / 2) + 97343.768 × 38.85 × 1.15 × 88.85
+ 91434.602 × (88.85 / 4) × 1.15 × (7 × 88.85 / 8) + 79616.270 × (88.85 / 4) × 1.15 × (5 × 88.85 / 8)
+ 67797.938 × (88.85 / 4) × 1.15 × (3 × 88.85 / 8) + 55979.606 × (88.85 / 4) × 1.15 × (88.85 / 8)
+ 50070.440 × 14.425 × 1.15 × 14.425 / 2} /{224.71 × 73707.104}
= 53.876 mm
∑E T (
I = ECF (1 / 12)tc 3 + ct (d − Y − c / 2) 2 + (1 / 12)bt 3 + bt (d − Y ) 2 )
( ) (
+ E w1 (1 / 12)t (d / 4) 3 + (d / 4)t (d − Y − d / 8) 2 + E w 2 (1 / 12)t (d / 4) 3 + (d / 4)t (d − Y − 3d / 8) 2 )
+ E w3 ((1 / 12)t (d / 4) + (d / 4)t (Y − 3d / 8) ) + E ((1 / 12)t (d / 4)
3 2
w4
3
+ (d / 4)t (Y − d / 8) 2
)
+ E HF ((1 / 12)tc + ct (Y − c / 2) + (1 / 12)bt + bt (Y ) )
3 2 3 2
 (1 / 12) × 1.15 × 14.425 3 + 14.425 × 1.15 × (88.85 − 53.876 − 14.425 / 2) 2 

= 97343.768 ×  

 + (1 / 12) × 38.85 × 1.15 + 38.85 × 1.15 × (88.85 − 53.876)
3 2

(
+ 91434.602 × (1 / 12) × 1.15 × (88.85 / 4) + (88.85 / 4) × 1.15 × (88.85 − 53.876 − 88.85 / 8) 2
3
)
+ 79616.270 × ((1 / 12) × 1.15 × (88.85 / 4) 3
+ (88.85 / 4) × 1.15 × (88.85 − 53.876 − 3 × 88.85 / 8) 2 )
+ 67797.938 × ((1 / 12) × 1.15 × (88.85 / 4) 3
+ (88.85 / 4) × 1.15 × (53.876 − 3 × 88.85 / 8) 2 )
+ 55979.606 × ((1 / 12) × 1.15 × (88.85 / 4) 3
+ (88.85 / 4) × 1.15 × (53.876 − 88.85 / 8) 2 )
 (1 / 12) × 1.15 × 14.425 + 14.425 × 1.15 × (53.876 − 14.425 / 2)
3 2

+ 50070.440 ×  

 + (1 / 12) × 38.85 × 1.15 + 38.85 × 1.15 × 53.876
3 2

= 1.990 × 1010 Nmm 2
I-8
Appendix I

At Mid-height

2α∆T
PL
e= 3d + e∆T − e∆E
64 * 16
EI − P
L3 3L
2 × 1.388 × 10 −5 × (656 − 492)
× 0 × 2400
= 3 × 88 . 85 + 18.446 − 9.119 = 9.327 mm
64 16
× 1.990 × 10 − 10
×0
2400 3 3 × 2400
µ x = λθ (2β M , x − 4) = 0.668 × (2 × 1.3 − 4) = −0.935
µx N * (−0.935) × 0
kx = 1− = 1− =1
χN eff 0.710 × 16367.655
N* k x N *e
+ =1
N* 1 × N * × 9.327
+ =1
0.710 × 16367.655 753237
N * = 10159 N
Iteration 1
2 × 1.388 × 10 −5 × (656 − 492)

× 10159 × 2400
e= 3 × 88 . 85 + 18.446 − 9.119 = 15.314 mm
64 16
× 1.990 × 10 −10
× 10159
2400 3 3 × 2400
µ N* (−0.935) × 10159
kx = 1− x = 1− = 1.817 > 1.5 k x = 1 .5
χN eff 0.710 × 16367.655
N* k N *e N* 1.5 × N * × 15.314
+ x = + =1
χN eff M x,eff , p 0.710 × 16367.655 753237
N * = 8580 N
Initial
Calculations
e 9.327 15.314 14.141 14.266 14.253 14.254
kx 1 1.5 1.5 1.5 1.5 1.5
N* 10159 8580 8756 8737 8739 8738
Ultimate Capacity of LSF Stud at mid-height = 8.738 kN
I-9
Appendix I
At Support
e = e∆E = 9.119 mm
µ x = λθ (2β M , x − 4) = 0.668 × (2 × 1.3 − 4) = −0.935

µx N * (−0.935) × 0
kx = 1− = 1− =1
χN eff 0.710 × 16367.655
N* k x N *e
+ =1
N* 1 × N * × 9.119
+ =1
0.710 × 16367.655 305250
N * = 8626 N
Iteration 1
µx N * (−0.935) × 8626
kx = 1− = 1− = 1.694 > 1.5 k x = 1 .5
χN eff 0.710 × 16367.655
N* k N *e
+ x =1
N* 1.5 × N * × 9.119
+ =1
0.710 × 16367.655 305250
N * = 7642 N
Initial
Iteration 1 Iteration 2
Calculations
kx 1 1.5 1.5
N* 8626 7642 7642
Ultimate Capacity of LSF Stud at Support = 7.642 kN
Therefore the Ultimate Capacity of LSF Stud under Fire Conditions = 7.642 kN
I-10
Appendix J
Feng and Wang (2005) - Modified EC3 Part 1.3 Method

b
External Dimensions
d - 90 mm
c b - 40 mm
c - 15 mm
t - 1.15 mm
HF = 656°C
Web = 574°C
c CF = 492°C
 49.844 × (38.85 + 14.425) × 1.15 + 102.950 × 88.85 × 1.15 

=   / 224.71 = 124.667 MPa
 + 235.710 × (38.85 + 14.425) × 1.15 
E 656 = 50070.440 MPa E 574 = 73707.104 MPa E 492 = 97343.768 MPa
Et = (E 656 (b + c)t + E574 dt + E 492 (b + c)t ) / A
 50070.440 × (38.85 + 14.425) × 1.15 + 73707.104 × 88.85 × 1.15 

=   / 224.71
 + 97343.768 × (38.85 + 14.425) × 1.15 
= 73707.104 MPa

A = 224.71 mm 2
X cg = 13.46 mm Ycg = 44.425 mm

Computer Program
I w = 103958882 mm 6
J = 99.1 mm 4
xo = 33.73 mm yo = 0
Ix 290097 Ix 54333
rx = = = 35.93 mm ry = = = 15.55 mm
A 224.71 A 224.71
J-1
Appendix J
Effective area at elevated temperature for pure compression (see Appendix D)
beff1,cf beff2,cf
t tred,cf ceff,cf
y
deff1
deff1 25.011 deff2 25.011
beff1,hf 19.425 beff2,hf 19.425

x
ceff,hf 14.425 tred,hf 1.127
beff1,cf 16.628 beff2,cf 18.512
deff2 ceff,cf 14.425 tred,cf 0.815

t tred,hf ceff,hf
beff1,hf beff2,hf
α T = 1.2 × 10 −5 + 0.4 × 10 −8 × T − (2.416 × 10 −4 / T )
= 1.2 × 10 −5 + 0.4 × 10 −8 × 574 − (2.416 × 10 −4 / 574) = 1.388 × 10 −5
α T (HF − CF )L2 1.388 × 10 −5 × (656 − 492 ) × 2400 2

e ∆T = = = 18.446 mm
8d 8 × 88.85
Aeff = {E656 (beff 1, hf t + beff 2,hf t red ,hf + ceff , hf t red , hf ) + E 492 (beff 1,cf t + beff 2,cf t red ,cf + ceff ,cf t red ,cf )
+ E574t (d eff 1 + d eff 2 )} / Et
= {50070.440 × (19.425 × 1.15 + 19.425 × 1.127 + 14.425 × 1.127)
+ 97343.768 × (16.628 × 1.15 + 18.512 × 0.815 + 14.425 × 0.815)
+ 73707.104 × 1.15 × (25.011 + 25.011)} / 73707.104
= 159.322 mm 2
Yeff = {E 492 ceff ,cf t red ,cf (d − ceff ,cf / 2) + E 492 beff 2,cf t red ,cf d + E 492 beff 1,cf td + E 574 d eff 2 t (d − d eff 2 / 2)
+ E 574 d eff 1td eff 1 / 2) + E 656 ceff ,hf t red ,hf ceff ,hf / 2} /{ Aeff × E t }
= {97343.768 × 14.425 × 0.815 × (88.85 − 14.425 / 2) + 97343.768 × 18.512 × 0.815 × 88.85

+ 97343.768 × 16.628 × 1.15 × 88.85 + 73707.104 × 25.011 × 1.15 × (88.85 − 25.011 / 2)
+ 73707.104 × 25.011 × 1.15 × 25.011 / 2 + 50070.440 × 14.425 × 1.127 × 14.425 / 2}
/{73707.104 × 159.322}
= 49.692 mm
e∆E = Yeff − d / 2 = 49.692 − 88.85 / 2 = 5.267 mm
J-2
Appendix J
N eff = f y , 656 (beff 1,hf t + beff 2,hf t red , hf + c eff ,hf t red ,hf ) + f y , 492 (beff 1,cf t + beff 2,cf t red ,cf + c eff ,cf t red ,cf )
+ f y ,574 t (d eff 1 + d eff 2 )
= 49.844 × (19.425 × 1.15 + 19.425 × 1.127 + 14.425 × 1.127)
+ 235.710 × (16.628 × 1.15 + 18.512 × 0.815 + 14.425 × 0.815)
+ 102.950 × 1.15 × (25.011 + 25.011)
= 19771.790 N
l x 2400
λx = = = 66.797
rx 35.93
ly 300
λy = = = 19.293
ry 15.55
0.5
E   73707.104 
0.5
λ1 = π  t  = π ×  = 76.389
f   124.667 
 y
0.5
λ  Aeff 
0.5
66.797  159.322 
λ =   = ×  = 0.736
λ1  A  76.389  224.71 
( 2
)
φ = 0.5 1 + α (λ − 0.2) + λ = 0.5 × (1 + 0.34 × (0.736 − 0.2) + 0.736 2 ) = 0.862

1 1
χ fi = = = 0.763
φ + φ −λ 2 2
0.862 + 0.862 2 − 0.736 2
J-3
Appendix J

Effective area of web

ψ = −1 kσ = 23.9
_
λp =
fy  bp
=   
(
 12 1 − ν f y 
2
) 0.5
= ×
( 2
)
 88.85  12 × 1 − 0.3 × 102.950 
0.5
= 0.621
   2 
σ cr  t   π Ekσ 
2
 1.15   π × 73707.104 × 23.9 
λ p < 0.673 ρ = 1 .0

compression (see Appendix D)
tension side
beff1,cf beff2,cf Cold Flange
t tred,cf ceff,cf
y
d 88.85 b 38.85
x c 14.425 t 1.15
d
beff1,cf 16.628 beff2,cf 18.512
t t
c
Hot Flange
b
Aeff = {E 656 t (b + c) + E574 td + E 492 (tbeff 1,cf + t red ,cf beff 2,cf + t red ,cf ceff ,cf )} / Et
= {50070.440 × 1.15 × (38.85 + 14.425) + 73707.104 × 1.15 × 88.85
+ 97343.768 × (1.15 × 16.628 + 0.815 × 18.512 + 0.815 × 14.425)} / 73707.104
= 204.503 mm 2
Yeff = {E 492 ceff ,cf t red ,cf (d − ceff ,cf / 2) + E 492 beff 2,cf t red ,cf d + E 492 beff 1,cf td + E 574 dtd / 2)
+ E 656 ctc / 2} /{ Aeff E t }
= {97343.768 × 14.425 × 0.815 × (88.85 − 14.425 / 2) + 97343.768 × 18.512 × 0.815 × 88.85
+ 97343.768 × 16.628 × 1.15 × 88.85 + 73707.104 × 88.85 × 1.15 × 88.85 / 2
+ 50070.440 × 14.425 × 1.15 × 14.425 / 2} /{73707.104 × 204.503}
= 48.421 mm
J-4
Appendix J
f y , 492 f y , 492
f y , 492 beff1,cf beff2,cf f y , 492
t tred,cf ceff,cf
y f y , 492 (d − Yeff − c)
E 519
E 492 (d − Yeff )
x’
d x
f yp
f y , 629
Y eff
t t
c
f y , 656 b f y , 656
f y , 656 f y , 656
At Point P the distance from Hot flange Y p = 31.6 mm ,

(656 − 492)
T p = 656 − × 31.6 = 598 o C
88.85
f yp = 66.130 MPa E p = 66789.056 MPa
Stress at P
f y , 492 (Yeff − Y p )
P= Ep
E 492 (d − Yeff )
235.710 (47.884 − 31.6)
= × 66789.056 × = 64.285 MPa
97343.768 (88.85 − 47.884)
≈ f yp
M x ,eff , p = ct × f 656 (Yeff − c / 2) + ct × 0.5( f 629 − f 656 )(Yeff − 2c / 3)

+ bt × f 656 Yeff + Y p t × f 656 (Yeff − Y p / 2)
+ Ypt × 0.5( f yp − f 656 )(Yeff − 2Yp / 3) + (Yeff − Yp )t × 0.5 f yp × (Yeff − Yp )2 / 3
+ (d − Yeff )t × 0.5 f 492 × (d − Yeff )2 / 3 + (beff 1t + beff 2 t red ) f 492 × (d − Yeff )
f 492 (d − Yeff − c)
+ ceff tred { E519 }( d − Yeff − ceff / 2)
E492 (d − Yeff )
f 492 (d − Yeff − ceff )
+ ceff tred 0.5{ f 492 − E519 }(d − Yeff − ceff / 3)
E492 (d − Yeff )
J-5
Appendix J
= 14.425 × 1.15 × 49.844 × (48.421 − 14.425 / 2)

+ 14.425 × 1.15 × 0.5 × (55.990 − 49.844)(48.421 − 2 × 14.425 / 3)
+ 38.85 × 1.15 × 49.844 × 48.421 + 31.6 × 1.15 × 49.844 × (48.421 − 31.6 / 2)
+ 31.6 × 1.15 × 0.5 × (66.130 − 49.844)(48.421 − 2 × 31.6 / 3)
+ (48.421 − 31.6) × 1.15 × 0.5 × 66.130 × (48.421 − 31.6)2 / 3
+ (88.85 − 48.421) × 1.15 × 0.5 × 235.710 × (88.85 − 48.421)2 / 3
+ (16.628 × 1.15 + 18.512 × 0.815) × 235.710 × (88.85 − 48.421)
235.710 (88.85 − 48.421 − 14.425)
+ 14.425 × 0.815 × { × 89560.964}
97343.768 (88.85 − 48.421)
× (88.85 − 48.421 − 14.425 / 2)
235.710 (88.85 − 48.421 − 14.425)
+ 14.425 × 0.815 × 0.5 × {235.710 − × 89560.964}
97343.768 (88.85 − 48.421)
× (88.85 − 48.421 − 14.425 / 3)
= 766540 Nmm

• Effective width of web is similar to Case 1 (fully effective in this case)

• Effective flange and lip widths on the hot side are the same as under uniform
compression
• The flange and the lip on the cold side will not buckle because they are on the
tension side
b
Cold Flange
t t c
y
d 88.85 b 38.85
c 14.425 t 1.15
d x
beff1,hf 19.425 beff2,hf 19.425
t tred,hf
ceff,hf
beff1,hf beff2,hf Hot Flange
J-6
Appendix J
= {97343.768 × 1.15 × (38.85 + 14.425) + 73707.104 × 1.15 × 88.85

+ 50070.440 × (1.15 × 19.425 + 1.127 × 19.425 + 1.127 × 14.425)} / 73707.104
= 224.181 mm 2
Yeff = {E 492 ct (d − c / 2) + E 492 btd + E 574 dtd / 2 + E 656 ceff t red ceff / 2} /{ Aeff E t }
= {97343.768 × 14.425 × 1.15 × (88.85 − 14.425 / 2)

+ 97343.768 × 38.85 × 1.15 × 88.85 + 73707.104 × 88.85 × 1.15 × 88.85 / 2
+ 50070.440 × 14.425 × 1.127 × 14.425 / 2} /{73707.104 × 224.181}
= 51.967 mm
f y , 656 (d − Yeff ) f y , 656 (d − Yeff )

E 492 E 492
t t c
y f y , 656 (d − Yeff − c)
E 519
E 656 Yeff
x’
d x
f y , 656 c
E 629
E656 Yeff
Y eff
t tred
ceff
f y , 656 beff1 beff2 f y , 656
f y , 656 f y , 656
f y , 656 ceff
M x ,eff , 2 = ceff t red × 0.5( f y ,656 − E 629 )(Yeff − ceff / 3)
E 656 Y eff
f y , 656 ceff
+ ceff t red × ( E 629 )(Yeff − ceff / 2) + (beff 1t + beff 2 t red ) f y , 656 Yeff
E 656 Y eff
f y , 656 (d − Y eff )
+ Yeff t × 0.5 f 656 (2Yeff / 3) + (d − Yeff )t × 0.5 × E 492 × 2(d − Yeff ) / 3
E 656 Y eff
f y , 656 (d − Yeff ) f y , 656 (d − Yeff − c)
+ bt E 492 (d − Yeff ) + ct E 519 (d − Yeff − c / 2)
f y , 656 (d − Yeff ) f y ,656 (d − Yeff − c)
+ ct × 0.5{ E 492 − E519 }(d − Yeff − c / 3)
J-7
Appendix J
49.844 14.425
= 14.425 × 1.127 × 0.5 × (49.844 − × × 57853.244)(51.967 − 14.425 / 3)
50070.44 51.967
49.844 14.425
+ 14.425 × 1.127 × ( × × 57853.244) × (51.967 − 14.425 / 2)
50070.44 51.967
+ (19.425 × 1.15 + 19.425 × 1.127) × 49.844 × 51.967 + 51.967 × 1.15 × 0.5 × 49.844 × (2 × 51.967 / 3)
49.844 (88.85 − 51.967)
+ (88.85 − 51.967) × 1.15 × 0.5 × × × 97343.768 × 2 × (88.85 − 51.967) / 3
50070.44 51.967
49.844 (88.85 − 51.967)
+ 38.85 × 1.15 × 97343.768 × (88.85 − 51.967)
50070.44 51.967
49.844 (88.85 − 51.967 − 14.425)
+ 14.425 × 1.15 × × × 89560.964 × (88.85 − 51.967 − 14.425 / 2)
50070.44 51.967
49.844 (88.85 − 51.967)
+ 14.425 × 1.15 × 0.5{ × × 97343.768
50070.44 51.967
49.844 (88.85 − 51.967 − 14.425)
− × × 89560.964} × (88.85 − 51.967 − 14.425 / 3)
50070.44 51.967
= 366985 Nmm
Y = {E 492 ct (d − c / 2) + E 492 btd + E w1 (d / 4)t (7 d / 8) + E w 2 (d / 4)t (5d / 8) + E w3 (d / 4)t (3d / 8)

+ E w 4 (d / 4)t (d / 8) + E 656 ctc / 2} /{ A Et }
= {97343.768 × 14.425 × 1.15 × (88.85 − 14.425 / 2) + 97343.768 × 38.85 × 1.15 × 88.85
+ 91434.602 × (88.85 / 4) × 1.15 × (7 × 88.85 / 8) + 79616.270 × (88.85 / 4) × 1.15 × (5 × 88.85 / 8)
+ 67797.938 × (88.85 / 4) × 1.15 × (3 × 88.85 / 8) + 55979.606 × (88.85 / 4) × 1.15 × (88.85 / 8)
+ 50070.440 × 14.425 × 1.15 × 14.425 / 2} /{224.71 × 73707.104}
= 53.876 mm
∑E T (
( ) (
+ E w1 (1 / 12)t (d / 4) 3 + (d / 4)t (d − Y − d / 8) 2 + E w 2 (1 / 12)t (d / 4) 3 + (d / 4)t (d − Y − 3d / 8) 2 )
+ E w3 ((1 / 12)t (d / 4) + (d / 4)t (Y − 3d / 8) ) + E ((1 / 12)t (d / 4)
3 2
w4
3
+ (d / 4)t (Y − d / 8) 2 )
+ E HF ((1 / 12)tc + ct (Y − c / 2) + (1 / 12)bt + bt (Y ) )
3 2 3 2
 (1 / 12) × 1.15 × 14.425 3 + 14.425 × 1.15 × (88.85 − 53.876 − 14.425 / 2) 2 

= 97343.768 ×  

 + ( 1 / 12 ) × 38 . 85 × 1 . 15 3
+ 38 . 85 × 1 . 15 × (88 . 85 − 53 . 876 ) 2

(
+ 91434.602 × (1 / 12) × 1.15 × (88.85 / 4) + (88.85 / 4) × 1.15 × (88.85 − 53.876 − 88.85 / 8) 2
3
)
+ 79616.270 × ((1 / 12) × 1.15 × (88.85 / 4) 3
+ (88.85 / 4) × 1.15 × (88.85 − 53.876 − 3 × 88.85 / 8) 2 )
+ 67797.938 × ((1 / 12) × 1.15 × (88.85 / 4) 3
+ (88.85 / 4) × 1.15 × (53.876 − 3 × 88.85 / 8) 2 )
+ 55979.606 × ((1 / 12) × 1.15 × (88.85 / 4) 3
+ (88.85 / 4) × 1.15 × (53.876 − 88.85 / 8) 2
)
 (1 / 12) × 1.15 × 14.425 + 14.425 × 1.15 × (53.876 − 14.425 / 2)
3 2

+ 50070.440 ×  

 + (1 / 12) × 38.85 × 1.15 + 38.85 × 1.15 × 53.876
3 2

= 1.990 × 1010 Nmm 2
J-8
Appendix J

At Mid-height
2α∆T
PL
e= 3d + e∆T − e∆E
64 * 16
EI − P
L3 3L
2 × 1.388 × 10 −5 × (656 − 492)
× 0 × 2400
= 3 × 88.85 + 18.446 − 5.267 = 13.179 mm
64 16
× 1.990 × 10 −10
×0
2400 3 3 × 2400
µ x = λθ (2β M , x − 4) = 0.736 × (2 × 1.3 − 4) = −1.030
µx N * (−1.030) × 0
kx = 1− = 1− =1
χN eff 0.763 × 19771.790
N* k N *e
+ x =1
N* 1 × N * × 13.179
+ =1
0.763 × 19771.790 766540
N * = 11979 N
Iteration 1
2 × 1.388 × 10 −5 × (656 − 492)

× 11979 × 2400
e= 3 × 88.85 + 18.446 − 5.267 = 20.675 mm
64 16
× 1.990 × 10 −
10
× 11979
2400 3 3 × 2400
µx N * (−1.030) × 11979
kx = 1− = 1− = 1.818 > 1.5 k x = 1 .5
χN eff 0.763 × 19771.790
N* k x N *e N* 1.5 × N * × 20.675
+ = + =1
χN eff M x ,eff , p 0.763 × 19771.790 766540
N * = 9368 N
Initial
Calculations
e 13.179 20.675 18.563 18.847 18.807 18.812
kx 1 1.5 1.5 1.5 1.5 1.5
N* 11979 9368 9745 9693 9700 9699
Ultimate Capacity of LSF Stud at Mid-height = 9.699 kN
J-9
Appendix J
At Support
e = e∆E = 5.267 mm
µ x = λθ (2β M , x − 4) = 0.736 × (2 × 1.3 − 4) = −1.030

µx N * (−1.030) × 0
kx = 1− = 1− =1
χN eff 0.763 × 19771.790
N* k x N *e
+ =1
N* 1 × N * × 5.267
+ =1
0.763 × 19771.790 366985
N * = 12401 N
Iteration 1
µx N * (−1.030) × 12401
kx = 1− = 1− = 1.847 > 1.5 k x = 1 .5
χN eff 0.763 × 19771.790
N* k N *e
+ x =1
N* 1.5 × N * × 5.267
+ =1
0.763 × 19771.790 366985
N * = 11388 N
Initial
Iteration 1 Iteration 2
Calculations
kx 1 1.5 1.5
N* 12401 11388 11388
J-10
Appendix K
Appendix K: Zhao et al. (2005)

b
External Dimensions
d - 90 mm
c b - 40 mm
c - 15 mm
t - 1.15 mm
HF = 656°C
Web = 574°C
c CF = 492°C
 49.844 × (38.85 + 14.425) × 1.15 + 102.950 × 88.85 × 1.15 

=   / 224.71 = 124.667 MPa
 + 235.710 × (38.85 + 14.425) × 1.15 
E 656 = 50070.440 MPa E 574 = 73707.104 MPa E 492 = 97343.768 MPa
Et = (E 656 (b + c)t + E574 dt + E 492 (b + c)t ) / A
 50070.440 × (38.85 + 14.425) × 1.15 + 73707.104 × 88.85 × 1.15 

=   / 224.71
 + 97343.768 × (38.85 + 14.425) × 1.15 
= 73707.104 MPa

A = 224.71 mm 2
X cg = 13.46 mm Ycg = 44.425 mm

Computer Program
I w = 103958882 mm 6
J = 99.1 mm 4
xo = 33.73 mm yo = 0
Ix 290097 Ix 54333
rx = = = 35.93 mm ry = = = 15.55 mm
A 224.71 A 224.71
K-1
Appendix K
beff1,cf beff2,cf
t tred,cf ceff,cf
y
deff1
deff1 25.011 deff2 25.011
beff1,hf 19.425 beff2,hf 19.425

x
beff1,cf 16.628 beff2,cf 18.512

t tred,hf ceff,hf
beff1,hf beff2,hf
α T = 1.2 × 10 −5 + 0.4 × 10 −8 × T − (2.416 × 10 −4 / T )
= 1.2 × 10 −5 + 0.4 × 10 −8 × 574 − (2.416 × 10 −4 / 574) = 1.388 × 10 −5
α T (HF − CF )L2 1.388 × 10 −5 × (656 − 492 ) × 2400 2

e ∆T = = = 18.446 mm
8d 8 × 88.85
+ E574t (d eff 1 + d eff 2 )} / Et
= {50070.440 × (19.425 × 1.15 + 19.425 × 1.127 + 14.425 × 1.127)
+ 97343.768 × (16.628 × 1.15 + 18.512 × 0.815 + 14.425 × 0.815)
+ 73707.104 × 1.15 × (25.011 + 25.011)} / 73707.104
= 159.322 mm 2
Yeff = {E 492 ceff ,cf t red ,cf (d − ceff ,cf / 2) + E 492 beff 2,cf t red ,cf d + E 492 beff 1,cf td + E 574 d eff 2 t (d − d eff 2 / 2)
+ E 574 d eff 1td eff 1 / 2) + E 656 ceff ,hf t red ,hf ceff ,hf / 2} /{ Aeff × E t }
= {97343.768 × 14.425 × 0.815 × (88.85 − 14.425 / 2) + 97343.768 × 18.512 × 0.815 × 88.85

+ 97343.768 × 16.628 × 1.15 × 88.85 + 73707.104 × 25.011 × 1.15 × (88.85 − 25.011 / 2)
+ 73707.104 × 25.011 × 1.15 × 25.011 / 2 + 50070.440 × 14.425 × 1.127 × 14.425 / 2}
/{73707.104 × 159.322}
= 49.692 mm
e∆E = Yeff − d / 2 = 49.692 − 88.85 / 2 = 5.267 mm
K-2
Appendix K
The ultimate load for local buckling N fi ,c , Rd
N fi ,c , Rd = ∑ Aeff ,i f y ,θ ,i
N fi ,c , Rd = f y ,656 (beff 1,hf t + beff 2,hf t red ,hf + ceff ,hf t red ,hf ) + f y , 492 (beff 1,cf t + beff 2,cf t red ,cf + c eff ,cf t red ,cf )
+ f y ,574 t (d eff 1 + d eff 2 )
= 49.844 × (19.425 × 1.15 + 19.425 × 1.127 + 14.425 × 1.127)
+ 235.710 × (16.628 × 1.15 + 18.512 × 0.815 + 14.425 × 0.815)
+ 102.950 × 1.15 × (25.011 + 25.011)
= 19771.790 N
Euler Buckling Load N fi ,cr
ECF (b + c )t / 2 + E web d 2 / 2 + E HF (b + c )(d − t / 2 )

y CF =
ECF (b + c ) + E web d + E HF (b + c )
97343.768 × (38.85 + 14.425) × 1.15 / 2 + 73707.104 × 88.85 2 / 2

+ 50070.44 × (38.85 + 14.425) × (88.85 − 1.15 / 2)
=
97343.768 × (38.85 + 14.425) + 73707.104 × 88.85 + 50070.44 × (38.85 + 14.425)
= 36.757 mm
∑ Ei ,θ I iθ = (1 / 12)td 3 E web + (1 / 12)tc 3 (E HF + ECF ) + (1 / 12)bt 3 (E HF + ECF )

( ) 2
( )
2
+ dt d / 2 − y CF Eweb + ct y CF − c / 2 E CF + ct d − y CF − c / 2 E HF ( ) 2
( )E
2 2
+ bt y CF E CF + bt d − y CF HF
= (1 / 12) × 1.15 × 88.85 3 × 73707.104 + (1 / 12) × 1.15 × 14.425 3 (50070.440 + 97343.768)

+ (1 / 12) × 38.85 × 1.15 3 × (50070.440 + 97343.768)
+ 88.85 ×1.15 × (88.85 / 2 − 36.757 ) × 73707.104
2
+ 14.425 × 1.15(36.757 − 14.425 / 2 ) × 97343.768

2
+ 14.425 × 1.15 × (88.85 − 36.757 − 14.425 / 2 ) × 50070.440

2
+ 38.85 × 1.15 × 36.757 2 × 97343.768 + 38.85 × 1.15 × (88.85 − 36.757 ) × 50070.440

2
= 2.047 × 1010 Nmm 2
π 2 ∑ Ei ,θ I i ,θ π 2 × 2.047 × 1010
N fi ,cr = = = 35074.792 N
L2 2400 2
K-3
Appendix K
The ultimate failure load for flexural buckling N fi , Rd
Relative slenderness at elevated temperature λ θ
λθ = ∑A f
i
i y ,θ ,i / N fi ,cr
(b + c)t ( f y ,656 + f y , 492 ) + dtf y ,574

=
N fi ,cr
(38.85 + 14.425) × 1.15 × (49.844 + 235.710) + 88.85 × 1.15 × 102.950
=
35074.792
= 0.894
( )
φθ = 0.5 1 + α (λ θ − 0.2) + λ θ 2 = 0.5 × (1 + 0.34 × (0.894 − 0.2) + 0.894 2 ) = 1.018
The reduction factor for buckling resistance χ

1 1
χ= = = 0.664
φθ + φθ − λθ
2 2
1 .018 + 1 .018 2
− 0.894 2
N fi , Rd = χ λ θ N fi ,c , Rd = 0.664 × 0.894 × 19771.790 = 11736.851 N

M x ,eff ,1 = 766540 Nmm

see Appendix J
M x ,eff , 2 = 366985 Nmm
K-4
Appendix K
e = e∆T − e∆E
= 18.446 − 5.267 = 13.179 mm
kx = 1
N* k N *e
+ x =1
N fi , Rd M fi , Rd , x
N* 1 × N * × 13.179
+ =1
11736.851 766540
N * = 9766 N
Iteration 1
1 1
e = e ∆T − e∆E = 18.446 × − 5.267
 *   9766 
1 − N  1 − 
 N   35074.792 
 fi ,cr 
= 20.297 mm
kx = 1
N* k x N *e
+ =1
N fi , Rd M fi , Rd , x
N* 1 × N * × 20.297
+ =1
11736.851 766540
N * = 8954 N
Initial
Iteration 1 Iteration 2 Iteration 3 Iteration 4
Calculations
e 13.179 20.297 19.502 19.582 19.574
N* 9766 8954 9038 9030 9030
K-5
Appendix K
At Support
e = e∆E = 5.267 mm
kx = 1
N* k N *e
+ x =1
N* 1 × N * × 5.267
+ =1
11736.851 366985
N * = 10045 N
K-6
Appendix L
Appendix L: Proposed Fire Design Rules - EC3 Part 1.3

b
External Dimensions
d - 90 mm
c b - 40 mm
c - 15 mm
t - 1.15 mm
HF = 656°C
Web = 574°C
c CF = 492°C
 49.844 × (38.85 + 14.425) × 1.15 + 102.950 × 88.85 × 1.15 

=   / 224.71 = 124.667 MPa
 + 235.710 × (38.85 + 14.425) × 1.15 
E 656 = 50070.440 MPa E 574 = 73707.104 MPa E 492 = 97343.768 MPa
Et = (E 656 (b + c)t + E574 dt + E 492 (b + c)t ) / A
 50070.440 × (38.85 + 14.425) × 1.15 + 73707.104 × 88.85 × 1.15 

=   / 224.71
 + 97343.768 × (38.85 + 14.425) × 1.15 
= 73707.104 MPa

A = 224.71 mm 2
X cg = 13.46 mm Ycg = 44.425 mm

Computer Program
I w = 103958882 mm 6
J = 99.1 mm 4
xo = 33.73 mm yo = 0
Ix 290097 Ix 54333
rx = = = 35.93 mm ry = = = 15.55 mm
A 224.71 A 224.71
L-1
Appendix L
beff1,cf beff2,cf
t tred,cf ceff,cf
y
deff1
deff1 25.011 deff2 25.011
beff1,hf 19.425 beff2,hf 19.425

x
beff1,cf 16.628 beff2,cf 18.512

t tred,hf ceff,hf
beff1,hf beff2,hf

α T = 1.2 × 10 −5 + 0.4 × 10 −8 × T − (2.416 × 10 −4 / T )
= 1.2 × 10 −5 + 0.4 × 10 −8 × 574 − (2.416 × 10 −4 / 574) = 1.388 × 10 −5

α T (HF − CF )L2 1.388 × 10 −5 × (656 − 492 ) × 2400 2
e ∆T = = = 18.446 mm
8d 8 × 88.85

+ E574t (d eff 1 + d eff 2 )} / Et
= {50070.440 × (19.425 × 1.15 + 19.425 × 1.127 + 14.425 × 1.127)
+ 97343.768 × (16.628 × 1.15 + 18.512 × 0.815 + 14.425 × 0.815)
+ 73707.104 × 1.15 × (25.011 + 25.011)} / 73707.104
= 159.322 mm 2
Yeff = {E 492 ceff ,cf t red ,cf (d − ceff ,cf / 2) + E 492 beff 2,cf t red ,cf d + E 492 beff 1,cf td + E 574 d eff 1t (d − d eff 1 / 2)
+ E 574 d eff 2 td eff 2 / 2) + E 656 ceff ,hf t red , hf c eff ,hf / 2} /{ Aeff × E t }
= {97343.768 × 14.425 × 0.815 × (88.85 − 14.425 / 2) + 97343.768 × 18.512 × 0.815 × 88.85

+ 97343.768 × 16.628 × 1.15 × 88.85 + 73707.104 × 25.011 × 1.15 × (88.85 − 25.011 / 2)
+ 73707.104 × 25.011 × 1.15 × 25.011 / 2 + 50070.440 × 14.425 × 1.127 × 14.425 / 2}
/{73707.104 × 159.322}
= 49.692 mm
e∆E = Yeff − d / 2 = 49.692 − 88.85 / 2 = 5.267 mm
L-2
Appendix L
f y , 656 = 49.844 MPa f y , 492 = 235.710 MPa
f y ,CF > 1.5 f y , HF Hence f y ,CF is limited to 1.5 f y , HF as discussed in Chapter 5
f y ,CF = 1.5 × 49.844 = 74.766 MPa
N eff = f y , 656 (beff 1,hf t + beff 2,hf t red , hf + c eff ,hf t red ,hf ) + f y , 492 (beff 1,cf t + beff 2,cf t red ,cf + c eff ,cf t red ,cf )
+ f y ,574 t (d eff 1 + d eff 2 )
= 49.844 × (19.425 × 1.15 + 19.425 × 1.127 + 14.425 × 1.127)
+ 74.766 × (16.628 × 1.15 + 18.512 × 0.815 + 14.425 × 0.815)
+ 102.950 × 1.15 × (25.011 + 25.011)
12373.862 N
l x 2400
λx = = = 66.797
rx 35.93
ly 300
λy = = = 19.293
ry 15.55
0.5
E   73707.104 
0.5
λ1 = π  t  = π ×  = 76.389
f   124.667 
 y
0.5
λ  Aeff 
0.5
66.797  159.322 
λ =   = ×  = 0.736
λ1  A  76.389  224.71 
( 2
)
φ = 0.5 1 + α (λ − 0.2) + λ = 0.5 × (1 + 0.34 × (0.736 − 0.2) + 0.736 2 ) = 0.862

1 1
χ fi = = = 0.763
φ + φ −λ 2 2
0.862 + 0.862 2 − 0.736 2
L-3
Appendix L
Effective area at elevated temperature for pure bending
Effective area of web

ψ = −1 kσ = 23.9
_
λp =
fy  bp
= 
(
 12 1 − ν f y 
2
) 0.5
( )
 88.85  12 × 1 − 0.3 × 102.950 
=
2 0.5
   ×  2  = 0.621
σ cr  t   π Ekσ 
2
 1.15   π × 73707.104 × 23.9 
λ p < 0.673 ρ = 1 .0

compression
tension side
beff1,cf beff2,cf
t tred,cf ceff,cf
y
d 88.85 b 38.85
x c 14.425 t 1.15
d
beff1,cf 16.628 beff2,cf 18.512
t t
c
b
Aeff = {E 656 t (b + c) + E574 td + E 492 (tbeff 1,cf + t red ,cf beff 2,cf + t red ,cf ceff ,cf )} / Et
= {50070.440 × 1.15 × (38.85 + 14.425) + 73707.104 × 1.15 × 88.85
+ 97343.768 × (1.15 × 16.628 + 0.815 × 18.512 + 0.815 × 14.425)} / 73707.104
= 204.503 mm 2
Yeff = {E 492 ceff ,cf t red ,cf (d − ceff ,cf / 2) + E 492 beff 2,cf t red ,cf d + E 492 beff 1,cf td
+ E 574 dtd / 2) + E 656 ctc / 2} /{ Aeff E t }
= {97343.768 × 14.425 × 0.815 × (88.85 − 14.425 / 2) + 97343.768 × 18.512 × 0.815 × 88.85
+ 97343.768 × 16.628 × 1.15 × 88.85 + 73707.104 × 88.85 × 1.15 × 88.85 / 2
+ 50070.440 × 14.425 × 1.15 × 14.425 / 2} /{73707.104 × 204.503}
= 48.421 mm
L-4
Appendix L
,cf + (1 / 12)beff 2 ,cf t red ,cf + (1 / 12)beff 1,cf t

 E 492 {(1 / 12)t red ,cf ceff
3 3 3

 
+ c ( 2
) (
 eff ,cf t red ,cf d − Yeff − ceff ,cf / 2 + beff 2,cf t red ,cf d − Yeff
2
) 

=
( ) ( )
I eff ,T 2 2 
 + beff 1,cf t d − Yeff } + E574 {(1 / 12)td + dt d / 2 − Yeff } 
3
 2
 + E656 {(1 / 12)bt + (1 / 12)tc + btYeff + ct Yeff − c / 2 } 
3 3
( 2
) 
Et
 97343.768 × {(1 / 12) × 0.815 × 14.425 3 + (1 / 12) × 18.512 × 0.815 3 
 
 + (1 / 12) × 16.628 × 1.15 + 14.425 × 0.815 × (88.85 − 48.421 − 14.425 / 2 ) 
3 2
 
 + 18.512 × 0.815 × (88.85 − 48.421) + 16.628 × 1.15 × (88.85 − 48.421) } 
2 2
= 2 
 + 73707.104 × {(1 / 12) × 1.15 × 88.85 + 88.85 × 1.15 × (88.85 / 2 − 48.421) }
3
 + 50070.440 × {(1 / 12) × 38.85 × 1.15 3 + (1 / 12) × 1.15 × 14.425 3 

 
 + 38.85 × 1.15 × 48.4212 + 14.425 × 1.15 × (48.421 − 14.425 / 2 )2 } 
 
73707.104
= 250595 mm 4
y max = Maximum of Y eff and (d − Y eff )

= Maximum of 48.421 mm and (88.85 - 48.421) mm
= 48.421 mm
At Mid-height
I eff ,T f y 250595 × 124.667
M x ,eff = = = 645194 Nmm
y max 48.421
At Support
I eff ,T f y , HF 250595 × 49.844
M x ,eff = = = 257960 Nmm
y max 48.421
L-5
Appendix L
d 7d d 5d d 3d d d
Y = {E 492 ct (d − c / 2) + E 492 btd + E w1 t + E w2 t + E w3 t + E w4 t
4 8 4 8 4 8 4 8
+ E656 ctc / 2} /{ AEt }
= {97343.768 × 14.425 × 1.15 × (88.85 − 14.425 / 2) + 97343.768 × 38.85 × 1.15 × 88.85
+ 91434.602 × (88.85 / 4) × 1.15 × (7 × 88.85 / 8) + 79616.270 × (88.85 / 4) × 1.15 × (5 × 88.85 / 8)
+ 67797.938 × (88.85 / 4) × 1.15 × (3 × 88.85 / 8) + 55979.606 × (88.85 / 4) × 1.15 × (88.85 / 8)
+ 50070.440 × 14.425 × 1.15 × 14.425 / 2} /{224.71 × 73707.104}
= 53.876 mm
∑E T (
(
+ E w1 (1 / 12)t (d / 4) 3 + (d / 4)t (d − Y − d / 8) 2 )
+ Ew2 ((1 / 12)t (d / 4) + (d / 4)t (d − Y − 3d / 8) )
3 2
+ E w3 ((1 / 12)t (d / 4) + (d / 4)t (Y − 3d / 8) )

3 2
+ Ew4 ((1 / 12)t (d / 4) + (d / 4)t (Y − d / 8) )

3 2
+ E HF ((1 / 12)tc + ct (Y − c / 2) + (1 / 12)bt + bt (Y ) )

3 2 3 2
 (1 / 12) × 1.15 × 14.425 3 + 14.425 × 1.15 × (88.85 − 53.876 − 14.425 / 2) 2 

= 97343.768 ×  

 + ( 1 / 12 ) × 38 . 85 × 1 . 15 3
+ 38 . 85 × 1 . 15 × (88 . 85 − 53 . 876 ) 2

(
+ 91434.602 × (1 / 12) × 1.15 × (88.85 / 4) + (88.85 / 4) × 1.15 × (88.85 − 53.876 − 88.85 / 8) 2
3
)
+ 79616.270 × ((1 / 12) × 1.15 × (88.85 / 4) 3
+ (88.85 / 4) × 1.15 × (88.85 − 53.876 − 3 × 88.85 / 8) 2 )
+ 67797.938 × ((1 / 12) × 1.15 × (88.85 / 4) 3
+ (88.85 / 4) × 1.15 × (53.876 − 3 × 88.85 / 8) 2
)
+ 55979.606 × ((1 / 12) × 1.15 × (88.85 / 4) 3
+ (88.85 / 4) × 1.15 × (53.876 − 88.85 / 8) 2
)
 (1 / 12) × 1.15 × 14.425 3 + 14.425 × 1.15 × (53.876 − 14.425 / 2) 2 
+ 50070.440 ×  

 + ( 1 / 12 ) × 38 . 85 × 1 . 15 3
+ 38 . 85 × 1 . 15 × 53 . 876 2

= 1.990 × 1010 Nmm 2
π 2 ∑ ET I π 2 × 1.990 × 1010
N cr ,T , y = = = 34098.112 N
L2 2400 2
L-6
Appendix L
At Mid-height

N*
C m = 0.79 + 0.21ψ + 0.36(ψ − 0.33)
N cr
0
= 0.79 + 0.21 × 1 + 0.36 × (1 − 0.33) = 1 .0
34098.112
1 Cm
e1 − e2 = e∆T − e ∆E
 ∗   ∗ 
1 − N  1 − χ N 
 N   N cr ,T , y 
 cr ,T , y  
1 1
= × 18.446 − × 5.267 = 13.179 mm
 0   0 
1 −  1 − 0.763 × 
 34098.112   34098.112 
N* N *k x e N* N * × 13.179
+ = + =1
χN eff M x,eff 0.763 × 12373.862 645194
N * = 7915 N
Iteration 1
N*
C m = 0.79 + 0.21ψ + 0.36(ψ − 0.33)
N cr
7915
= 0.79 + 0.21 × 1 + 0.36 × (1 − 0.33) = 1.056
34098.112
1 1.056
e1 − e2 = × 18.446 − × 5.267 = 17.263 mm
 7915   7915 
1 −  1 − 0.763 × 
 34098.112   34098.112 
N* N *k x e N* N * × 17.263
+ = + =1
χN eff M x,eff 0.763 × 12373.862 645194
N * = 7537 N
Initial
Calculations
Cm 1.000 1.056 1.053 1.053 1.053
kxe 13.179 17.263 17.009 17.025 17.024
N* 7915 7537 7560 7558 7558
L-7
Appendix L
At Support

N*
C m = 0.79 + 0.21ψ + 0.36(ψ − 0.33)
N cr
0
= 0.79 + 0.21 × 1 + 0.36 × (1 − 0.33) = 1 .0
34098.112
Cm 1
e1 − e2 = e ∆E = × 5.267 = 5.267 mm
 N ∗   0 
1 − χ  1 − 0.763 × 
 N   34098.112 
 cr ,T , y 
N * *
N kxe N* N * × 5.267
+ = + =1
χN eff M x,eff 0.763 × 12373.862 257960
N * = 7915 N
Iteration 1
N*
C m = 0.79 + 0.21ψ + 0.36(ψ − 0.33)
N cr
7915
= 0.79 + 0.21 × 1 + 0.36 × (1 − 0.33) = 1.056
34098.112
1.056
e1 − e2 = × 5.267 = 6.759 mm
 7915 
1 − 0.763 × 
 34098.112 
N* N *k x e N* N * × 6.759
+ = + =1
χN eff M x,eff 0.763 × 12373.862 257960
N * = 7569 N
Initial
Calculations
Cm 1.000 1.056 1.054 1.054 1.054
kxe 5.267 6.759 6.683 6.686 6.686
N* 7915 7569 7586 7585 7585
L-8
Appendix M
Appendix M: Proposed Fire Design Rules - AS/NZS 4600

b
External Dimensions
d - 90 mm
c b - 40 mm
c - 15 mm
t - 1.15 mm
HF = 656°C
Web = 574°C
c CF = 492°C
 49.844 × (38.85 + 14.425) × 1.15 + 102.950 × 88.85 × 1.15 

=   / 224.71 = 124.667 MPa
 + 235.710 × (38.85 + 14.425) × 1.15 
E 656 = 50070.440 MPa E 574 = 73707.104 MPa E 492 = 97343.768 MPa
Et = (E 656 (b + c)t + E574 dt + E 492 (b + c)t ) / A
 50070.440 × (38.85 + 14.425) × 1.15 + 73707.104 × 88.85 × 1.15 

=   / 224.71
 + 97343.768 × (38.85 + 14.425) × 1.15 
= 73707.104 MPa

A = 224.71 mm 2
X cg = 13.46 mm Ycg = 44.425 mm

Computer Program
I w = 103958882 mm 6
J = 99.1 mm 4
xo = 33.73 mm yo = 0
Ix 290097 Ix 54333
rx = = = 35.93 mm ry = = = 15.55 mm
A 224.71 A 224.71
M-1
Appendix M
Effective area at elevated temperature for pure compression (see Appendix B)
beff1,cf beff2,cf
t ceff,cf
deff1 26.388 deff2 26.388
y
deff1
beff1,cf 19.425 beff1,hf 17.327
beff2,cf 19.425 beff2,hf 17.327

x
deff2 Aeff 176.210 fn 90.552

t
ceff,hf
beff1,hf beff2,hf

α T = 1.2 × 10 −5 + 0.4 × 10 −8 × T − (2.416 × 10 −4 / T )
= 1.2 × 10 −5 + 0.4 × 10 −8 × 574 − (2.416 × 10 −4 / 574) = 1.388 × 10 −5

α T (HF − CF )L2 1.388 × 10 −5 × (656 − 492 ) × 2400 2
e ∆T = = = 18.446 mm
8d 8 × 88.85

Aeff = {E656 (beff 1, hf t + beff 2,hf t + ceff ,hf t ) + E 492 (beff 1,cf t + beff 2,cf t + ceff ,cf t ) + E574 t (d eff 1 + d eff 2 )} / Et
= {50070.440 × (2 × 17.327 × 1.15 + 12.521 × 1.15) + 97343.768 × (2 × 19.425 × 1.15 + 14.425 × 1.15)
+ 73707.104 × 1.15 × 2 × 26.388} / 73707.104
= 178.459 mm 2
Yeff = {E 492 ceff ,cf t (d − ceff ,cf / 2) + E 492 beff 2,cf td + E 492 beff 1,cf td + E 574 d eff 1t (d − d eff 1 / 2)
+ E 574 d eff 2 td eff 2 / 2) + E 656 ceff ,hf tceff ,hf / 2} /{ Aeff × E t }
= {97343.768 × 14.425 × 1.15 × (88.85 − 14.425 / 2) + 2 × 97343.768 × 19.425 × 1.15 × 88.85

+ 73707.104 × 26.388 × 1.15 × (88.85 − 26.388 / 2) + 73707.104 × 26.388 × 1.15 × 26.388 / 2
+ 50070.440 × 12.521 × 1.15 × 12.521 / 2} /{73707.104 × 178.459}
= 54.851 mm
e ∆E = Yeff − d / 2 = 54.851 − 88.85 / 2 = 10.426 mm
Member capacity N c
N c = Ae f n = 176.210 × 90.552 = 15956.168 N
M-2
Appendix M
Effective area at elevated temperature for pure bending
Effective width of web

 kπ 2 E web  t   24 × π 2 × 73707.104  1.15 
2 2
f cr =  
2   =    = 267.842 MPa
 12(1 − ν )  b   12 × (1 − 0.3 )
2
 88.85 
f* 90.552
λ= = = 0.581 < 0.673
f cr 267.842
Effective width factor ρ = 1

compression (fully effective in this case)
tension side
b
t c
y
d 88.85
x b 38.85
d
c 14.425
t 1.15
t
c
b
Aeff = {E 656 t (b + c) + E 574 td + E 492 t (b + c)} / E t

= {50070.440 × 1.15 × (38.85 + 14.425) + 73707.104 × 1.15 × 88.85
+ 97343.768 × 1.15 × (38.85 × 14.425)} / 73707.104
= 224.71 mm 2
Yeff = {E 492 ct (d − c / 2) + E 492 btd + E574 dtd / 2) + E 656 ctc / 2} /{ Aeff E t }

= {97343.768 × 14.425 × 1.15 × (88.85 − 14.425 / 2) + 97343.768 × 38.85 × 1.15 × 88.85
+ 73707.104 × 88.85 × 1.15 × 88.85 / 2 + 50070.440 × 14.425 × 1.15 × 14.425 / 2}
/{73707.104 × 224.71}
= 51.852 mm
M-3
Appendix M
( ) ( )
 E {(1 / 12)tc 3 + (1 / 12)bt 3 + ct d − Y − c / 2 2 + bt d − Y 2 }
 492 eff eff 
I eff ,T

(
=  + E574 {(1 / 12)td 3 + dt d / 2 − Yeff }
2
) 

 
2
( ) 2
 + E656 {(1 / 12)bt 3 + (1 / 12)tc 3 + btYeff + ct Yeff − c / 2 }



Et
 97343.768 × {(1 / 12) × 1.15 × 14.425 3 + (1 / 12) × 38.85 × 1.15 3 
 
 + 14.425 × 1.15 × (88.85 − 51.852 − 14.425 / 2 )2 + 38.85 × 1.15 × (88.85 − 51.852 )2 }
 
=  + 73707.104 × {(1 / 12) × 1.15 × 88.85 3 + 88.85 × 1.15 × (88.85 / 2 − 51.852 ) }
2

 
 + 50070.440 × {(1 / 12) × 38.85 × 1.15 + (1 / 12) × 1.15 × 14.425
3 3

 + 38.85 × 1.15 × 51.852 2 + 14.425 × 1.15 × (51.852 − 14.425 / 2 )2 } 
 
73707.104
= 277701 mm 4
y max = Maximum of Y eff and (d − Y eff )

= Maximum of 51.852 mm and (88.85 – 51.852) mm
= 51.852 mm
At Mid-height
I eff ,T f y 277701 × 124.667
M x ,eff = = = 667672 Nmm
y max 51.852
At Support
I eff ,T f y , HF 277701 × 49.844
M x ,eff = = = 266947 Nmm
y max 51.852
M-4
Appendix M
Euler Buckling Load N e
d 7d d 5d d 3d d d
Y = {E 492 ct (d − c / 2) + E 492 btd + E w1 t + E w2 t + E w3 t + E w4 t
4 8 4 8 4 8 4 8
+ E656 ctc / 2} /{ AEt }
= {97343.768 × 14.425 × 1.15 × (88.85 − 14.425 / 2) + 97343.768 × 38.85 × 1.15 × 88.85
+ 91434.602 × (88.85 / 4) × 1.15 × (7 × 88.85 / 8) + 79616.270 × (88.85 / 4) × 1.15 × (5 × 88.85 / 8)
+ 67797.938 × (88.85 / 4) × 1.15 × (3 × 88.85 / 8) + 55979.606 × (88.85 / 4) × 1.15 × (88.85 / 8)
+ 50070.440 × 14.425 × 1.15 × 14.425 / 2} /{224.71 × 73707.104}
= 53.876 mm
∑E T (
(
+ E w1 (1 / 12)t (d / 4) 3 + (d / 4)t (d − Y − d / 8) 2)
+ Ew2 ((1 / 12)t (d / 4) + (d / 4)t (d − Y − 3d / 8) )
3 2
+ E w3 ((1 / 12)t (d / 4) + (d / 4)t (Y − 3d / 8) )

3 2
+ Ew4 ((1 / 12)t (d / 4) + (d / 4)t (Y − d / 8) )

3 2
+ E HF ((1 / 12)tc + ct (Y − c / 2) + (1 / 12)bt + bt (Y ) )

3 2 3 2
 (1 / 12) × 1.15 × 14.425 3 + 14.425 × 1.15 × (88.85 − 53.876 − 14.425 / 2) 2 

= 97343.768 ×  

 + ( 1 / 12 ) × 38 . 85 × 1 . 15 3
+ 38 . 85 × 1 . 15 × (88 . 85 − 53 . 876 ) 2

(
+ 91434.602 × (1 / 12) × 1.15 × (88.85 / 4) + (88.85 / 4) × 1.15 × (88.85 − 53.876 − 88.85 / 8) 2
3
)
+ 79616.270 × ((1 / 12) × 1.15 × (88.85 / 4) 3
+ (88.85 / 4) × 1.15 × (88.85 − 53.876 − 3 × 88.85 / 8) 2 )
+ 67797.938 × ((1 / 12) × 1.15 × (88.85 / 4) 3
+ (88.85 / 4) × 1.15 × (53.876 − 3 × 88.85 / 8) 2 )
+ 55979.606 × ((1 / 12) × 1.15 × (88.85 / 4) 3
+ (88.85 / 4) × 1.15 × (53.876 − 88.85 / 8) 2
)
 (1 / 12) × 1.15 × 14.425 + 14.425 × 1.15 × (53.876 − 14.425 / 2)
3 2

+ 50070.440 ×  

 + (1 / 12) × 38.85 × 1.15 + 38.85 × 1.15 × 53.876
3 2

= 1.990 × 1010 Nmm 2
π 2 ∑ ET I π 2 × 1.990 × 1010
Ne = = = 34098.112 N
L2 2400 2
M-5
Appendix M

At Mid-height
e = e∆T − e∆E = 18.446 − 10.426 = 8.020 mm

C mx = 1
 N∗   0 
α nx = 1 −  = 1 −  =1
 N e   34098.112 
N* C N *e N* 1 × N * × 8.020
+ m = + =1
N c α nx M x ,eff 15956.168 1 × 667672
N * = 13390 N
Iteration 1
 N∗   13390 
α nx = 1 −  = 1 −  = 0.607
 N e   34098.112 
N* C N *e N* 1 × N * × 8.020
+ m = + =1
N c α nx M x ,eff 15956.168 0.607 × 667672
N * = 12127 N
Initial
Calculations
α 1.000 0.607 0.644 0.639 0.640 0.640
N* 13390 12127 12297 12275 12279 12279
M-6
Appendix M
At Support
e = e∆E = 10.426 mm
C mx = 1
 N∗   0 
α nx = 1 −  = 1 −  =1
 N e   34098.112 
N* C N *e N* 1 × N * × 10.426
+ m = + =1
N c α nx M x ,eff 15956.168 1 × 266947
N * = 9830 N
Iteration 1
 N∗   9830 
α nx = 1 −  = 1 −  = 0.712
 N e   34098.112 
N* C N *e N* 1 × N * × 10.426
+ m = + =1
N c α nx M x ,eff 15956.168 0.712 × 266947
N * = 8509 N
Initial
Calculations
α 1.000 0.712 0.750 0.744 0.745 0.745
N* 9830 8509 8715 8683 8688 8688
Therefore Ultimate Capacity of LSF Stud under Fire Conditions = 8.688 kN
M-7

Shanmuganathan Gunalan Thesis PDF

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Shanmuganathan Gunalan Thesis PDF

Caricato da

Copyright:

Formati disponibili

Structural Behaviour and Design of

Cold-formed Steel Wall Systems

under Fire Conditions

School of Urban Development

A Thesis Submitted to the School of Urban Development,

Idealised time-temperature profiles were developed based on the measured

In summary, this research has developed comprehensive experimental and numerical

1. Gunalan, S. and Mahendran, M. (2010), Structural and Fire Behaviour of a

QUT Conference Papers

Proposed International Journal Papers (to be submitted by November 2011)

1. Gunalan, S., Kolarkar, P. and Mahendran, M. (2011a), Experimental Study

Keywords …………………………………………………………………………. iii

2. LITERATURE REVIEW ........................................................................... 2-1

3. EXPERIMENTAL STUDY ON THE STRUCTURAL AND THERMAL

3.3.3. Displacement Measurement .................................................................. 3-16

4. DEVELOPMENT AND VALIDATION OF FINITE ELEMENT

4.4.2. Loading and Boundary Conditions ....................................................... 4-29

5. REVIEW AND DEVELOPMENT OF FIRE DESIGN RULES FOR

5.2.5. Feng and Wang (2005b) .......................................................................... 5-6

6. IDEALISED TIME-TEMPERATURE PROFILES AND

7. CONCLUSIONS AND RECOMMENDATIONS .................................... 7-1

Figure 1.1: Light Gauge Steel Framing (LSF) Systems............................................ 1-2

Figure 3.27: Temperature Difference across Central Studs ................................... 3-47

Figure 4.1: Cross-sectional Dimensions and Boundary Conditions of the LSF

Figure 4.5: Stress – Strain Models ............................................................................ 4-8

A gross cross-sectional area of the stud

fyT, fyo yield stress at temperature T and room temperature, respectively

Nb,fi,t,Rd design resistance at time t of a compression member with a non-uniform

Zx elastic section modulus about major axis

α imperfection factor based on the relevant buckling curve

∆Mx,Ed, ∆My,Ed additional moments due to shift of centroidal axis

Author would like to thank Queensland University of Technology (QUT) and

Under fire conditions, cold-formed thin-walled steel sections heat up quickly

Figure 1.1: Light Gauge Steel Framing (LSF) Systems

1.2. Cold-formed Steel Members

The method of manufacturing is important as it differentiates these products from

Image from http://www.globalspec.com [Accessed July 22, 2011]

Figure 1.2: Cold-formed Steel Sections

1.3. Light Gauge Steel Frame (LSF) Walls

1.3.1. Elements Used

(a) LSF wall (b) LSF clad with Plasterboards

Figure 1.3: LSF Wall System

1.3.3. Behaviour of LSF Walls in Fire

1.3.4. Applications of LSF

1.4. Fire Safety

1.4.3. Structural Adequacy

1.4.4. Fire Resistance of LSF Walls

Calculation of the structural behaviour of steel structures is more difficult than

1.5. Composite Wall Panels

Insulation Studs Cavity Studs Plasterboards

Figure 1.4: Composite Wall Panels Developed by

1.6. Research Problem

1.7. Research Objectives and Scope

The specific tasks of this study are to,

1.8. Thesis Contents

2.2. Cold-formed Steel Members

2.2.2. Cold-formed Steel Design Standards

2.3. Light Gauge Steel Frame (LSF) Wall Assemblies

Studs Centroid with

Ambient side Smaller

Figure 2.1: LSF Wall and the Steel Stud Cross-section

2.3.1. Fire Resistance of LSF Members

According to Milke (1999), fire resistance of structural assemblies depends on fire